拆分独立的enpoint

HTTP_API_README.md

@@ -27,9 +27,9 @@
 %% {"result": Data}
 %%--------------------------------------------------------------------
 json_data(Data) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"result">> => Data
-    }, [force_utf8]).
+    })).
 
 %%--------------------------------------------------------------------
 %% @doc
@@ -42,12 +42,12 @@ json_data(Data) ->
 %% }
 %%--------------------------------------------------------------------
 json_error(ErrCode, ErrMessage) when is_integer(ErrCode), is_binary(ErrMessage) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"error">> => #{
             <<"code">> => ErrCode,
             <<"message">> => ErrMessage
         }
-    }, [force_utf8]).
+    })).
 ```
 
 ### 📘 返回格式说明
@@ -157,7 +157,7 @@ json_error(ErrCode, ErrMessage) when is_integer(ErrCode), is_binary(ErrMessage)
 | 参数名 | 类型 | 必填 | 说明 |
 |--------|------|------|------|
 | uuid | binary (string) | ✅ | 主机唯一标识符 |
-| task_id | integer | ✅ | 任务 ID |
+| task_id | integer | ✅ | 任务 ID，和 uuid 一起作为部署任务唯一标识 |
 | config | map | ✅ | 部署配置内容 |
 
 #### 响应参数
@@ -255,13 +255,16 @@ json_error(ErrCode, ErrMessage) when is_integer(ErrCode), is_binary(ErrMessage)
 #### 示例响应
 ```json
 {
-  "result": {
+  "data": "ok"
-    "task_id": 1001,
-    "status": "deployed"
-  }
 }
 ```
 
+部署过程实时日志通过 SSE 获取：
+
+```http
+GET /event_stream?uuid=qbxmjyzrkpntfgswaevodhluicqzxplkm&task_id=1
+```
+
 #### 错误响应
 ```json
 {
@@ -475,9 +478,9 @@ json_error(ErrCode, ErrMessage) when is_integer(ErrCode), is_binary(ErrMessage)
 %% {"result": Data}
 %%--------------------------------------------------------------------
 json_data(Data) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"result">> => Data
-    }, [force_utf8]).
+    })).
 
 %%--------------------------------------------------------------------
 %% @doc
@@ -490,12 +493,12 @@ json_data(Data) ->
 %% }
 %%--------------------------------------------------------------------
 json_error(ErrCode, ErrMessage) when is_integer(ErrCode), is_binary(ErrMessage) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"error">> => #{
             <<"code">> => ErrCode,
             <<"message">> => ErrMessage
         }
-    }, [force_utf8]).
+    })).
 ```
 
 ### 📘 返回格式说明
@@ -909,9 +912,9 @@ Content-Type: application/json
 %% {"result": Data}
 %%--------------------------------------------------------------------
 json_data(Data) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"result">> => Data
-    }, [force_utf8]).
+    })).
 
 %%--------------------------------------------------------------------
 %% @doc
@@ -924,12 +927,12 @@ json_data(Data) ->
 %% }
 %%--------------------------------------------------------------------
 json_error(ErrCode, ErrMessage) when is_integer(ErrCode), is_binary(ErrMessage) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"error">> => #{
             <<"code">> => ErrCode,
             <<"message">> => ErrMessage
         }
-    }, [force_utf8]).
+    })).
 ```
 
 ### 📘 返回格式说明
@@ -1079,7 +1082,7 @@ json_error(ErrCode, ErrMessage) when is_integer(ErrCode), is_binary(ErrMessage)
 | 参数名 | 类型 | 必填 | 说明 |
 |--------|------|------|------|
 | uuid | binary (string) | ✅ | 主机唯一标识符 |
-| auth | boolean | ✅ | `true` 激活, `false` 取消激活 |
+| auth | boolean | ✅ | `true` 激活, `false` 取消激活。该操作只修改 iot 本地和持久化授权状态，不通知 efka；efka 连接可继续保持在线，数据是否处理由 iot_host 状态决定。 |
 
 #### 响应参数
 | 字段 | 类型 | 说明 |

apps/endpoint/src/endpoint.app.src

@@ -0,0 +1,24 @@
+{application, endpoint,
+ [{description, "Endpoint OTP application"},
+  {vsn, "0.1.0"},
+  {registered, [
+    endpoint_sup,
+    endpoint_adapter_sup,
+    endpoint_subscription,
+    endpoint_log
+  ]},
+  {mod, {endpoint_app, []}},
+  {applications, [
+    emqtt,
+    brod,
+    hackney,
+    gproc,
+    crypto,
+    kernel,
+    stdlib
+  ]},
+  {env, []},
+  {modules, []},
+  {licenses, ["Apache 2.0"]},
+  {links, []}
+ ]}.

apps/endpoint/src/endpoint.erl

@@ -31,32 +31,32 @@ start_link(Endpoint = #endpoint{id = Id, config = #kafka_endpoint{}}) ->
     LocalName = get_name(Id),
     endpoint_kafka:start_link(LocalName, Endpoint).
 
--spec get_name(Id :: integer()) -> atom().
+-spec get_name(Id :: integer()) -> term().
 get_name(Id) when is_integer(Id) ->
-    list_to_atom("endpoint:" ++ integer_to_list(Id)).
+    {endpoint, Id}.
 
 -spec get_pid(Id :: integer()) -> undefined | pid().
 get_pid(Id) when is_integer(Id) ->
-    whereis(get_name(Id)).
+    gproc:whereis_name({n, l, get_name(Id)}).
 
--spec get_alias_name(Name :: binary()) -> atom().
+-spec get_alias_name(Name :: binary()) -> term().
 get_alias_name(Name) when is_binary(Name) ->
-    list_to_atom("endpoint:" ++ binary_to_list(Name)).
+    {endpoint_alias, Name}.
 
 -spec get_alias_pid(Name :: binary()) -> undefined | pid().
 get_alias_pid(Name) when is_binary(Name) ->
     gproc:whereis_name({n, l, get_alias_name(Name)}).
 
--spec forward(Pid :: pid(), Metric :: binary()) -> no_return().
+-spec forward(Pid :: pid(), Metric :: binary()) -> ok.
 forward(Pid, Metric) when is_pid(Pid), is_binary(Metric) ->
     gen_server:cast(Pid, {forward, Metric}).
 
 reload(Pid, NEndpoint = #endpoint{}) when is_pid(Pid) ->
-    gen_statem:cast(Pid, {reload, NEndpoint}).
+    gen_server:cast(Pid, {reload, NEndpoint}).
 
 -spec clean_up(Pid :: pid()) -> ok.
 clean_up(Pid) when is_pid(Pid) ->
-    gen_server:call(Pid, clean_up, 5000).
+    gen_server:cast(Pid, cleanup).
 
 -spec get_protocol(Endpoint :: #endpoint{}) -> atom().
 get_protocol(#endpoint{config = #http_endpoint{}}) ->
@@ -68,7 +68,7 @@ get_protocol(#endpoint{config = #kafka_endpoint{}}) ->
 
 -spec is_support(Protocol :: atom()) -> boolean().
 is_support(Protocol) when is_atom(Protocol) ->
-    {ok, Props} = application:get_env(iot, endpoints),
+    {ok, Props} = application:get_env(endpoint, endpoints),
     SupportProtocols = proplists:get_value(support_protocols, Props, []),
     lists:member(Protocol, SupportProtocols).
 
@@ -92,16 +92,27 @@ endpoint_record(#{<<"id">> := Id, <<"matcher">> := Matcher, <<"title">> := Title
 
 -spec parse_config(Protocol :: binary(), Config :: map()) -> {ok, #mqtt_endpoint{} | #kafka_endpoint{} | #http_endpoint{}} | {error, Errors :: [Error :: binary()]}.
 parse_config(<<"mqtt">>, #{<<"host">> := Host, <<"port">> := Port0, <<"client_id">> := ClientId, <<"username">> := Username, <<"password">> := Password, <<"topic">> := Topic, <<"qos">> := Qos}) ->
-    Port = if is_binary(Port0) -> binary_to_integer(Port0); is_integer(Port0) -> Port0 end,
+    {Port, PortErrors} = case parse_integer(Port0) of
+        {ok, ParsedPort} ->
+            {ParsedPort, []};
+        error ->
+            {undefined, [<<"port invalid">>]}
+    end,
 
-    Errors = lists:filtermap(fun(Term) ->
+    CheckTerms = case Port of
+        undefined ->
+            [{host, Host}, {username, Username}, {password, Password}, {topic, Topic}, {qos, Qos}];
+        _ ->
+            [{host, Host}, {port, Port}, {username, Username}, {password, Password}, {topic, Topic}, {qos, Qos}]
+    end,
+    Errors = PortErrors ++ lists:filtermap(fun(Term) ->
         case check_mqtt_argument(Term) of
             ok ->
                 false;
             {error, Error} ->
                 {true, Error}
         end
-                             end, [{host, Host}, {port, Port}, {username, Username}, {password, Password}, {topic, Topic}, {qos, Qos}]),
+                             end, CheckTerms),
     case Errors =:= [] of
         true ->
             {ok, #mqtt_endpoint{
@@ -118,11 +129,24 @@ parse_config(<<"mqtt">>, #{<<"host">> := Host, <<"port">> := Port0, <<"client_id
     end;
 parse_config(<<"http">>, C = #{<<"url">> := Url, <<"pool_size">> := PoolSize}) ->
     Token = maps:get(<<"token">>, C, <<>>),
-    {ok, #http_endpoint{
+    Errors = lists:filtermap(fun(Term) ->
-        url = Url,
+        case check_http_argument(Term) of
-        token = Token,
+            ok ->
-        pool_size = PoolSize
+                false;
-    }};
+            {error, Error} ->
+                {true, Error}
+        end
+                             end, [{url, Url}, {token, Token}, {pool_size, PoolSize}]),
+    case Errors =:= [] of
+        true ->
+            {ok, #http_endpoint{
+                url = Url,
+                token = Token,
+                pool_size = PoolSize
+            }};
+        false ->
+            {error, Errors}
+    end;
 parse_config(<<"kafka">>, #{<<"sasl_config">> := #{<<"username">> := Username, <<"password">> := Password, <<"mechanism">> := Mechanism0}, <<"bootstrap_servers">> := BootstrapServers, <<"topic">> := Topic}) ->
     Errors = lists:filtermap(fun(Term) ->
         case check_kafka_argument(Term) of
@@ -168,8 +192,8 @@ parse_config(_, _) ->
 -spec parse_kafka_bootstrap_servers(BootstrapServers :: [binary()]) -> Servers :: [{Host :: string(), Port :: integer()}].
 parse_kafka_bootstrap_servers(BootstrapServers) when is_list(BootstrapServers) ->
     lists:map(fun(S) ->
-        [Host0, Port0] = binary:split(S, <<":">>),
+        {ok, Host, Port} = parse_kafka_bootstrap_server(S),
-        {binary_to_list(Host0), binary_to_integer(Port0)}
+        {Host, Port}
               end, BootstrapServers).
 
 -spec parse_kafka_mechanism(Mechanism0 :: binary()) -> atom().
@@ -218,17 +242,10 @@ check_kafka_argument({bootstrap_servers, BootstrapServers}) ->
     case is_list(BootstrapServers) andalso length(BootstrapServers) > 0 of
         true ->
             InvalidServers = lists:filtermap(fun(S) ->
-                case binary:split(S, <<":">>) of
+                case parse_kafka_bootstrap_server(S) of
-                    [Host0, Port0] ->
+                    {ok, _Host, _Port} ->
-                        Host = binary_to_list(Host0),
+                        false;
-                        Port = binary_to_integer(Port0),
+                    error ->
-                        case not string:is_empty(Host) andalso (is_integer(Port) andalso Port > 0) of
-                            true ->
-                                false;
-                            false ->
-                                {true, S}
-                        end;
-                    _ ->
                         {true, S}
                 end
                                              end, BootstrapServers),
@@ -243,6 +260,29 @@ check_kafka_argument({bootstrap_servers, BootstrapServers}) ->
             {error, <<"bootstrap_servers is empty">>}
     end.
 
+-spec check_http_argument(tuple()) -> ok | {error, Reason :: binary()}.
+check_http_argument({url, Url}) ->
+    case is_binary(Url) andalso Url /= <<>> of
+        true ->
+            ok;
+        false ->
+            {error, <<"url is empty">>}
+    end;
+check_http_argument({token, Token}) ->
+    case is_binary(Token) of
+        true ->
+            ok;
+        false ->
+            {error, <<"token invalid">>}
+    end;
+check_http_argument({pool_size, PoolSize}) ->
+    case is_integer(PoolSize) andalso PoolSize > 0 of
+        true ->
+            ok;
+        false ->
+            {error, <<"pool_size invalid">>}
+    end.
+
 -spec check_mqtt_argument(tuple()) -> ok | {error, Reason :: binary()}.
 check_mqtt_argument({host, Host}) ->
     case Host /= <<>> of
@@ -286,3 +326,34 @@ check_mqtt_argument({qos, Qos}) ->
         false ->
             {error, <<"qos invalid">>}
     end.
+
+-spec parse_integer(term()) -> {ok, integer()} | error.
+parse_integer(Value) when is_integer(Value) ->
+    {ok, Value};
+parse_integer(Value) when is_binary(Value) ->
+    try binary_to_integer(Value) of
+        Int ->
+            {ok, Int}
+    catch
+        _:_ ->
+            error
+    end;
+parse_integer(_) ->
+    error.
+
+-spec parse_kafka_bootstrap_server(term()) -> {ok, string(), integer()} | error.
+parse_kafka_bootstrap_server(Server) when is_binary(Server) ->
+    case binary:split(Server, <<":">>) of
+        [Host0, Port0] ->
+            Host = binary_to_list(Host0),
+            case parse_integer(Port0) of
+                {ok, Port} when Host =/= [], Port > 0 ->
+                    {ok, Host, Port};
+                _ ->
+                    error
+            end;
+        _ ->
+            error
+    end;
+parse_kafka_bootstrap_server(_) ->
+    error.

apps/endpoint/src/endpoint_adapter_sup.erl

@@ -3,13 +3,14 @@
 %% @end
 %%%-------------------------------------------------------------------
 
--module(endpoint_sup).
+-module(endpoint_adapter_sup).
 
 -behaviour(supervisor).
 -include("endpoint.hrl").
 
 -export([start_link/0]).
 -export([ensured_endpoint_started/1, delete_endpoint/1]).
+-export([start_endpoints/1]).
 
 -export([init/1]).
 
@@ -29,21 +30,15 @@ start_link() ->
 %%                  modules => modules()}   % optional
 init([]) ->
     SupFlags = #{strategy => one_for_one, intensity => 1000, period => 3600},
-    Endpoints = iot_api:get_all_endpoints(),
+    {ok, {SupFlags, []}}.
-    ChildSpecs = lists:filtermap(fun(EndpointInfo) ->
+
-        case endpoint:endpoint_record(EndpointInfo) of
+-spec start_endpoints(Endpoints :: list()) -> ok.
-            error ->
+start_endpoints(Endpoints) when is_list(Endpoints) ->
-                false;
+    lists:foreach(fun(Endpoint) ->
-            {ok, Endpoint} ->
+        Spec = child_spec(Endpoint),
-                case endpoint:is_support(endpoint:get_protocol(Endpoint)) of
+        {ok, _} = supervisor:start_child(?MODULE, Spec)
-                    true ->
+                  end, Endpoints),
-                        {true, child_spec(Endpoint)};
+    ok.
-                    false ->
-                        false
-                end
-        end
-                                 end, Endpoints),
-    {ok, {SupFlags, ChildSpecs}}.
 
 -spec ensured_endpoint_started(Endpoint :: #endpoint{}) -> {ok, Pid :: pid()} | {error, Reason :: any()}.
 ensured_endpoint_started(Endpoint = #endpoint{}) ->
@@ -59,8 +54,24 @@ ensured_endpoint_started(Endpoint = #endpoint{}) ->
 -spec delete_endpoint(Id :: integer()) -> ok | {error, Reason :: any()}.
 delete_endpoint(Id) when is_integer(Id) ->
     Name = endpoint:get_name(Id),
-    supervisor:terminate_child(?MODULE, Name),
+    case supervisor:terminate_child(?MODULE, Name) of
-    supervisor:delete_child(?MODULE, Name).
+        ok ->
+            delete_endpoint_child(Name);
+        {error, not_found} ->
+            delete_endpoint_child(Name);
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+delete_endpoint_child(Name) ->
+    case supervisor:delete_child(?MODULE, Name) of
+        ok ->
+            ok;
+        {error, not_found} ->
+            ok;
+        {error, Reason} ->
+            {error, Reason}
+    end.
 
 child_spec(Endpoint = #endpoint{id = Id}) ->
     Name = endpoint:get_name(Id),

apps/endpoint/src/endpoint_app.erl

@@ -0,0 +1,15 @@
+%%%-------------------------------------------------------------------
+%% @doc endpoint application entrypoint.
+%% @end
+%%%-------------------------------------------------------------------
+-module(endpoint_app).
+
+-behaviour(application).
+
+-export([start/2, stop/1]).
+
+start(_StartType, _StartArgs) ->
+    endpoint_sup:start_link().
+
+stop(_State) ->
+    ok.

apps/endpoint/src/endpoint_buffer.erl

@@ -0,0 +1,304 @@
+
+%%%-------------------------------------------------------------------
+%%% @author aresei
+%%% @copyright (C) 2023, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 06. 7月 2023 12:02
+%%%-------------------------------------------------------------------
+-module(endpoint_buffer).
+
+-include("endpoint.hrl").
+
+%% 消息重发间隔
+-define(RETRY_INTERVAL, 5000).
+%% 最大重试次数，不包含首次发送
+-define(MAX_RETRY_TIMES, 3).
+%% 与 endpoint_outbox 的单条记录限制保持一致。
+-define(MAX_PAYLOAD_BYTES, 32 * 1024).
+
+-export([new/2, append/2, append_only/2, trigger_next/1, trigger_n/1, handle_timeout/4, cleanup/1, recover_inflight/1, ack/2, stat/1, resize/2]).
+-export_type([buffer/0]).
+
+-type flight_source() :: outbox | memory.
+-type timer_entry() :: {reference(), binary(), non_neg_integer(), flight_source()}.
+
+-record(buffer, {
+    endpoint :: #endpoint{},
+    outbox :: endpoint_outbox:outbox(),
+    %% 当前待 ack 的数据及其重试定时器 #{Id => {TimerRef, Payload, RetryTimes, Source}}
+    timer_map = #{} :: #{integer() => timer_entry()},
+    %% 窗口大小，允许最大的未确认消息数
+    window_size = 10,
+    %% 未确认的消息数
+    flight_num = 0,
+    %% 记录成功处理的消息数
+    acc_num = 0
+}).
+
+-type buffer() :: #buffer{}.
+
+-spec new(Endpoint :: #endpoint{}, WindowSize :: integer()) -> Buffer :: #buffer{}.
+new(Endpoint = #endpoint{id = Id}, WindowSize) when is_integer(WindowSize), WindowSize > 0 ->
+    %% 读取配置文件
+    {ok, Endpoints} = application:get_env(endpoint, endpoints),
+    RootDir = proplists:get_value(root_dir, Endpoints),
+    OutboxDir = filename:join(RootDir, integer_to_list(Id)),
+
+    {ok, Outbox} = endpoint_outbox:open(OutboxDir, #{}),
+    #buffer{outbox = Outbox, endpoint = Endpoint, window_size = WindowSize}.
+
+-spec append(Payload :: binary(), Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
+append(Payload, Buffer = #buffer{}) when is_binary(Payload) ->
+    case validate_payload_size(Payload, Buffer) of
+        ok ->
+            %% Persist before dispatch so a VM crash does not lose in-flight data.
+            case append_to_outbox(Payload, Buffer) of
+                {ok, NBuffer} ->
+                    trigger_next(NBuffer);
+                {dropped, NBuffer} ->
+                    trigger_next(NBuffer);
+                {error, NBuffer} ->
+                    NBuffer
+            end;
+        error ->
+            Buffer
+    end.
+
+-spec append_only(Payload :: binary(), Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
+append_only(Payload, Buffer = #buffer{}) when is_binary(Payload) ->
+    case validate_payload_size(Payload, Buffer) of
+        ok ->
+            case append_to_outbox(Payload, Buffer) of
+                {ok, NBuffer} ->
+                    NBuffer;
+                {dropped, NBuffer} ->
+                    NBuffer;
+                {error, NBuffer} ->
+                    NBuffer
+            end;
+        error ->
+            Buffer
+    end.
+
+-spec trigger_n(Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
+trigger_n(Buffer = #buffer{window_size = WindowSize}) ->
+    %% 最多允许window_size
+    lists:foldl(fun(_, Buffer0) -> trigger_next(Buffer0) end, Buffer, lists:seq(1, WindowSize)).
+
+%% 触发读取下一条数据
+-spec trigger_next(Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
+trigger_next(Buffer = #buffer{outbox = Outbox, flight_num = FlightNum, window_size = WindowSize}) ->
+    case FlightNum < WindowSize of
+        false ->
+            Buffer;
+        true ->
+            case endpoint_outbox:next(Outbox) of
+                eof ->
+                    Buffer;
+                {ok, Id, Payload, NOutbox} ->
+                    ReceiverPid = self(),
+                    ReceiverPid ! {next_data, Id, Payload},
+                    schedule_retry(Id, Payload, 0, outbox, Buffer#buffer{outbox = NOutbox, flight_num = FlightNum + 1});
+                {error, Reason} ->
+                    logger:warning("[endpoint_buffer] read next outbox failed, endpoint_id: ~p, reason: ~p",
+                        [buffer_endpoint_id(Buffer), Reason]),
+                    Buffer
+            end
+    end.
+
+-spec handle_timeout(reference(), integer(), binary(), #buffer{}) -> #buffer{}.
+handle_timeout(TimerRef, Id, _Payload, Buffer = #buffer{timer_map = TimerMap})
+        when is_reference(TimerRef), is_integer(Id) ->
+    case maps:take(Id, TimerMap) of
+        {{TimerRef, Payload, RetryTimes, Source}, NTimerMap} when RetryTimes < ?MAX_RETRY_TIMES ->
+            logger:warning("[endpoint_buffer] retry message, endpoint_id: ~p, id: ~p, retry: ~p/~p",
+                [buffer_endpoint_id(Buffer), Id, RetryTimes + 1, ?MAX_RETRY_TIMES]),
+            ReceiverPid = self(),
+            ReceiverPid ! {next_data, Id, Payload},
+            schedule_retry(Id, Payload, RetryTimes + 1, Source, Buffer#buffer{timer_map = NTimerMap});
+        {{TimerRef, Payload, RetryTimes, outbox}, NTimerMap} ->
+            logger:warning("[endpoint_buffer] drop message after retries exhausted, endpoint_id: ~p, id: ~p, retries: ~p",
+                [buffer_endpoint_id(Buffer), Id, RetryTimes]),
+            drop_outbox_message(Id, Payload, RetryTimes, Buffer#buffer{timer_map = NTimerMap});
+        {{TimerRef, _, RetryTimes, memory}, NTimerMap} ->
+            logger:warning("[endpoint_buffer] drop memory message after retries exhausted, endpoint_id: ~p, id: ~p, retries: ~p",
+                [buffer_endpoint_id(Buffer), Id, RetryTimes]),
+            drop_memory_message(Buffer#buffer{timer_map = NTimerMap});
+        {{_OtherTimerRef, _OtherPayload, _RetryTimes, _Source}, _NTimerMap} ->
+            Buffer;
+        error ->
+            Buffer
+    end.
+
+-spec ack(Id :: integer(), Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
+ack(Id, Buffer = #buffer{timer_map = TimerMap, outbox = Outbox, acc_num = AccNum, flight_num = FlightNum}) when is_integer(Id) ->
+    case maps:take(Id, TimerMap) of
+        {{TimerRef, _Payload, _RetryTimes, memory}, NTimerMap} ->
+            _ = erlang:cancel_timer(TimerRef),
+            NBuffer = Buffer#buffer{
+                timer_map = NTimerMap,
+                acc_num = AccNum + 1,
+                flight_num = max(FlightNum - 1, 0)
+            },
+            trigger_next(NBuffer);
+        {{TimerRef, Payload, RetryTimes, outbox}, NTimerMap} ->
+            _ = erlang:cancel_timer(TimerRef),
+            case endpoint_outbox:ack(Id, Outbox) of
+                {ok, NOutbox} ->
+                    NBuffer = Buffer#buffer{
+                        timer_map = NTimerMap,
+                        outbox = NOutbox,
+                        acc_num = AccNum + 1,
+                        flight_num = max(FlightNum - 1, 0)
+                    },
+                    trigger_next(NBuffer);
+                {error, Reason} ->
+                    logger:warning("[endpoint_buffer] ack outbox failed, endpoint_id: ~p, id: ~p, reason: ~p",
+                        [buffer_endpoint_id(Buffer), Id, Reason]),
+                    schedule_retry(Id, Payload, RetryTimes, outbox, Buffer#buffer{timer_map = NTimerMap})
+            end;
+        error ->
+            Buffer
+    end.
+
+%% 获取当前统计信息
+-spec stat(Buffer :: #buffer{}) -> map().
+stat(#buffer{acc_num = AccNum, outbox = Outbox, flight_num = FlightNum, timer_map = TimerMap}) ->
+    OutboxStat = endpoint_outbox:stat(Outbox),
+    WriteSeq = maps:get(write_seq, OutboxStat, 0),
+    AckedSeq = maps:get(acked_seq, OutboxStat, 0),
+    OutboxFlightNum = count_inflight(outbox, TimerMap),
+    MemoryFlightNum = count_inflight(memory, TimerMap),
+    QueueNum = max(WriteSeq - AckedSeq - OutboxFlightNum, 0),
+    OutboxStat#{
+        <<"acc_num">> => AccNum,
+        <<"queue_num">> => QueueNum,
+        <<"inflight_num">> => FlightNum,
+        <<"outbox_inflight_num">> => OutboxFlightNum,
+        <<"memory_inflight_num">> => MemoryFlightNum
+    }.
+
+-spec cleanup(Buffer :: #buffer{}) -> #buffer{}.
+cleanup(Buffer = #buffer{timer_map = TimerMap}) ->
+    cancel_timers(TimerMap),
+    NBuffer0 = persist_memory_inflight(Buffer),
+    reset_reader_after_recover(NBuffer0#buffer{timer_map = #{}, flight_num = 0}).
+
+-spec recover_inflight(Buffer :: #buffer{}) -> #buffer{}.
+recover_inflight(Buffer = #buffer{timer_map = TimerMap}) ->
+    cancel_timers(TimerMap),
+    NBuffer0 = persist_memory_inflight(Buffer),
+    case endpoint_outbox:reset_reader(NBuffer0#buffer.outbox) of
+        {ok, NOutbox} ->
+            NBuffer0#buffer{outbox = NOutbox, timer_map = #{}, flight_num = 0}
+    end.
+
+-spec resize(Buffer :: #buffer{}, WindowSize :: integer()) -> #buffer{}.
+resize(Buffer = #buffer{}, WindowSize) when is_integer(WindowSize), WindowSize > 0 ->
+    trigger_n(Buffer#buffer{window_size = WindowSize}).
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec buffer_endpoint_id(buffer()) -> integer().
+buffer_endpoint_id(#buffer{endpoint = #endpoint{id = Id}}) ->
+    Id.
+
+-spec validate_payload_size(binary(), buffer()) -> ok | error.
+validate_payload_size(Payload, Buffer) ->
+    case byte_size(Payload) =< ?MAX_PAYLOAD_BYTES of
+        true ->
+            ok;
+        false ->
+            logger:warning("[endpoint_buffer] payload too large, endpoint_id: ~p, size: ~p, max_size: ~p",
+                [buffer_endpoint_id(Buffer), byte_size(Payload), ?MAX_PAYLOAD_BYTES]),
+            error
+    end.
+
+-spec append_to_outbox(binary(), buffer()) -> {ok | dropped | error, buffer()}.
+append_to_outbox(Payload, Buffer = #buffer{outbox = Outbox}) ->
+    case endpoint_outbox:append(Payload, Outbox) of
+        {ok, _Seq, NOutbox} ->
+            {ok, Buffer#buffer{outbox = NOutbox}};
+        {dropped, capacity_reached, NOutbox} ->
+            logger:warning("[endpoint_buffer] outbox capacity reached, endpoint_id: ~p", [buffer_endpoint_id(Buffer)]),
+            {dropped, Buffer#buffer{outbox = NOutbox}};
+        {error, Reason} ->
+            logger:warning("[endpoint_buffer] append outbox failed, endpoint_id: ~p, reason: ~p",
+                [buffer_endpoint_id(Buffer), Reason]),
+            {error, Buffer}
+    end.
+
+-spec schedule_retry(integer(), binary(), non_neg_integer(), flight_source(), buffer()) -> buffer().
+schedule_retry(Id, Payload, RetryTimes, Source, Buffer = #buffer{timer_map = TimerMap})
+        when is_integer(Id), is_binary(Payload), is_integer(RetryTimes), RetryTimes >= 0 ->
+    TimerRef = erlang:start_timer(?RETRY_INTERVAL, self(), {endpoint_buffer_retry, Id, Payload}),
+    Buffer#buffer{timer_map = maps:put(Id, {TimerRef, Payload, RetryTimes, Source}, TimerMap)}.
+
+-spec cancel_timers(#{integer() => timer_entry()}) -> ok.
+cancel_timers(TimerMap) ->
+    lists:foreach(fun({_Id, {TimerRef, _Payload, _RetryTimes, _Source}}) ->
+        _ = erlang:cancel_timer(TimerRef),
+        ok
+    end, maps:to_list(TimerMap)),
+    ok.
+
+-spec drop_outbox_message(integer(), binary(), non_neg_integer(), buffer()) -> buffer().
+drop_outbox_message(Id, Payload, RetryTimes, Buffer = #buffer{outbox = Outbox, acc_num = AccNum, flight_num = FlightNum}) ->
+    case endpoint_outbox:ack(Id, Outbox) of
+        {ok, NOutbox} ->
+            trigger_next(Buffer#buffer{
+                outbox = NOutbox,
+                acc_num = AccNum + 1,
+                flight_num = max(FlightNum - 1, 0)
+            });
+        {error, Reason} ->
+            logger:warning("[endpoint_buffer] ack dropped message failed, endpoint_id: ~p, id: ~p, reason: ~p",
+                [buffer_endpoint_id(Buffer), Id, Reason]),
+            schedule_retry(Id, Payload, RetryTimes, outbox, Buffer)
+    end.
+
+-spec drop_memory_message(buffer()) -> buffer().
+drop_memory_message(Buffer = #buffer{acc_num = AccNum, flight_num = FlightNum}) ->
+    trigger_next(Buffer#buffer{
+        acc_num = AccNum + 1,
+        flight_num = max(FlightNum - 1, 0)
+    }).
+
+-spec persist_memory_inflight(buffer()) -> buffer().
+persist_memory_inflight(Buffer = #buffer{timer_map = TimerMap}) ->
+    MemoryInflight = lists:sort(
+        fun({IdA, _PayloadA}, {IdB, _PayloadB}) -> IdA > IdB end,
+        [{Id, Payload} || {Id, {_TimerRef, Payload, _RetryTimes, memory}} <- maps:to_list(TimerMap)]
+    ),
+    lists:foldl(fun({_Id, Payload}, AccBuffer) ->
+        case append_to_outbox(Payload, AccBuffer) of
+            {ok, NBuffer} ->
+                NBuffer;
+            {dropped, NBuffer} ->
+                NBuffer;
+            {error, NBuffer} ->
+                NBuffer
+        end
+    end, Buffer, MemoryInflight).
+
+-spec reset_reader_after_recover(buffer()) -> buffer().
+reset_reader_after_recover(Buffer = #buffer{outbox = Outbox}) ->
+    case endpoint_outbox:reset_reader(Outbox) of
+        {ok, NOutbox} ->
+            Buffer#buffer{outbox = NOutbox}
+    end.
+
+-spec count_inflight(flight_source(), #{integer() => timer_entry()}) -> non_neg_integer().
+count_inflight(Source, TimerMap) ->
+    maps:fold(fun(_Id, {_TimerRef, _Payload, _RetryTimes, EntrySource}, Acc) ->
+        case EntrySource =:= Source of
+            true ->
+                Acc + 1;
+            false ->
+                Acc
+        end
+    end, 0, TimerMap).

apps/endpoint/src/endpoint_http.erl

@@ -30,10 +30,10 @@
 %%%===================================================================
 
 %% @doc Spawns the server and registers the local name (unique)
--spec(start_link(LocalName :: atom(), Endpoint :: #endpoint{}) ->
+-spec(start_link(LocalName :: term(), Endpoint :: #endpoint{}) ->
     {ok, Pid :: pid()} | ignore | {error, Reason :: term()}).
-start_link(LocalName, Endpoint = #endpoint{config = #http_endpoint{}}) when is_atom(LocalName) ->
+start_link(LocalName, Endpoint = #endpoint{config = #http_endpoint{}}) ->
-    gen_server:start_link({local, LocalName}, ?MODULE, [Endpoint], []).
+    gen_server:start_link({via, gproc, {n, l, LocalName}}, ?MODULE, [Endpoint], []).
 
 %%%===================================================================
 %%% gen_server callbacks
@@ -44,9 +44,10 @@ start_link(LocalName, Endpoint = #endpoint{config = #http_endpoint{}}) when is_a
 -spec(init(Args :: term()) ->
     {ok, State :: #state{}} | {ok, State :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term()} | ignore).
-init([Endpoint = #endpoint{matcher = Matcher}]) ->
+init([Endpoint = #endpoint{matcher = Matcher, config = #http_endpoint{pool_size = PoolSize}}]) ->
+    ok = endpoint_util:set_metadata(),
     endpoint_subscription:subscribe(Matcher, self()),
-    Buffer = endpoint_buffer:new(Endpoint, 10),
+    Buffer = endpoint_buffer:new(Endpoint, PoolSize),
     {ok, #state{endpoint = Endpoint, buffer = Buffer}}.
 
 %% @private
@@ -69,20 +70,17 @@ handle_call(get_stat, _From, State = #state{buffer = Buffer}) ->
     {noreply, NewState :: #state{}} |
     {noreply, NewState :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term(), NewState :: #state{}}).
-handle_cast({forward, Metric}, State = #state{buffer = Buffer, endpoint = #endpoint{config = #http_endpoint{token = Token}}}) ->
+handle_cast({forward, Metric}, State = #state{buffer = Buffer}) ->
-    Tuple = case is_binary(Token) andalso Token /= <<>> of
+    NBuffer = endpoint_buffer:append(Metric, Buffer),
-        true ->
-            Sign = iot_util:sha256(erlang:iolist_to_binary([Token, Metric, Token])),
-            {Metric, Sign};
-        false ->
-            {Metric, <<>>}
-    end,
-    NBuffer = endpoint_buffer:append(Tuple, Buffer),
     {noreply, State#state{buffer = NBuffer}};
+handle_cast({reload, NEndpoint = #endpoint{matcher = NMatcher, config = #http_endpoint{pool_size = PoolSize}}}, State = #state{endpoint = #endpoint{matcher = Matcher}, buffer = Buffer}) ->
+    ensure_subscription(Matcher, NMatcher),
+    NBuffer = endpoint_buffer:resize(Buffer, PoolSize),
+    {noreply, State#state{endpoint = NEndpoint, buffer = NBuffer}};
 
 handle_cast(cleanup, State = #state{buffer = Buffer}) ->
-    endpoint_buffer:cleanup(Buffer),
+    NBuffer = endpoint_buffer:cleanup(Buffer),
-    {noreply, State}.
+    {noreply, State#state{buffer = NBuffer}}.
 
 %% @private
 %% @doc Handling all non call/cast messages
@@ -90,32 +88,25 @@ handle_cast(cleanup, State = #state{buffer = Buffer}) ->
     {noreply, NewState :: #state{}} |
     {noreply, NewState :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term(), NewState :: #state{}}).
-handle_info({next_data, Id, {Metric, Sign}}, State = #state{buffer = Buffer, endpoint = #endpoint{config = #http_endpoint{url = Url}}}) ->
+handle_info({next_data, Id, Metric}, State = #state{buffer = Buffer, endpoint = #endpoint{config = #http_endpoint{url = Url, token = Token}}}) ->
-    BaseHeaders = [{<<"Content-Type">>, <<"application/json">>}],
+    Headers = [{<<"Content-Type">>, <<"application/json">>}] ++ patch_headers(Metric, Token),
-    ExtraHeaders = if
-                       Sign =:= <<>> -> [];
-                       true -> [{<<"X-Signature">>, Sign}]
-                   end,
-    Headers = BaseHeaders ++ ExtraHeaders,
-
     case hackney:request(post, Url, Headers, Metric) of
-        {ok, 200, _, ClientRef} ->
+        {ok, HttpCode, _, ClientRef} when HttpCode >= 200, HttpCode < 300 ->
-            {ok, RespBody} = hackney:body(ClientRef),
+            RespBody = read_response_body(ClientRef),
-            hackney:close(ClientRef),
+            logger:debug("[endpoint_http] url: ~p, response is: ~p", [Url, RespBody]),
-            lager:debug("[endpoint_http] url: ~p, response is: ~p", [Url, RespBody]),
             NBuffer = endpoint_buffer:ack(Id, Buffer),
-
             {noreply, State#state{buffer = NBuffer}};
         {ok, HttpCode, _, ClientRef} ->
-            {ok, RespBody} = hackney:body(ClientRef),
+            RespBody = read_response_body(ClientRef),
-            hackney:close(ClientRef),
+            logger:warning("[endpoint_http] url: ~p, http_code: ~p, response is: ~p", [Url, HttpCode, RespBody]),
-            lager:debug("[endpoint_http] url: ~p, http_code: ~p, response is: ~p", [Url, HttpCode, RespBody]),
+            {noreply, State};
-            NBuffer = endpoint_buffer:ack(Id, Buffer),
-            {noreply, State#state{buffer = NBuffer}};
         {error, Reason} ->
-            lager:warning("[endpoint_http] url: ~p, get error: ~p", [Url, Reason]),
+            logger:warning("[endpoint_http] url: ~p, get error: ~p", [Url, Reason]),
             {noreply, State}
-    end.
+    end;
+handle_info({timeout, TimerRef, {endpoint_buffer_retry, Id, Payload}}, State = #state{buffer = Buffer}) ->
+    NBuffer = endpoint_buffer:handle_timeout(TimerRef, Id, Payload, Buffer),
+    {noreply, State#state{buffer = NBuffer}}.
 
 %% @private
 %% @doc This function is called by a gen_server when it is about to
@@ -124,7 +115,8 @@ handle_info({next_data, Id, {Metric, Sign}}, State = #state{buffer = Buffer, end
 %% with Reason. The return value is ignored.
 -spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
     State :: #state{}) -> term()).
-terminate(_Reason, _State = #state{}) ->
+terminate(_Reason, #state{buffer = Buffer}) ->
+    _ = endpoint_buffer:cleanup(Buffer),
     ok.
 
 %% @private
@@ -138,3 +130,28 @@ code_change(_OldVsn, State = #state{}, _Extra) ->
 %%%===================================================================
 %%% Internal functions
 %%%===================================================================
+
+-spec ensure_subscription(binary(), binary()) -> ok.
+ensure_subscription(Matcher, Matcher) ->
+    ok;
+ensure_subscription(Matcher, NMatcher) ->
+    ok = endpoint_subscription:unsubscribe(Matcher, self()),
+    endpoint_subscription:subscribe(NMatcher, self()).
+
+-spec read_response_body(reference()) -> binary() | term().
+read_response_body(ClientRef) ->
+    case hackney:body(ClientRef) of
+        {ok, RespBody} ->
+            hackney:close(ClientRef),
+            RespBody;
+        {error, Reason} ->
+            hackney:close(ClientRef),
+            Reason
+    end.
+
+-spec patch_headers(Metric :: binary(), any()) -> list().
+patch_headers(Metric, Token) when is_binary(Token), Token /= <<>> ->
+    Sign = endpoint_util:sha256(erlang:iolist_to_binary([Token, Metric, Token])),
+    [{<<"X-Signature">>, Sign}];
+patch_headers(_, _) ->
+    [].

apps/endpoint/src/endpoint_kafka.erl

@@ -0,0 +1,261 @@
+%%%-------------------------------------------------------------------
+%%% @author aresei
+%%% @copyright (C) 2023, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 06. 7月 2023 12:02
+%%%-------------------------------------------------------------------
+-module(endpoint_kafka).
+
+-include("endpoint.hrl").
+-behaviour(gen_statem).
+
+%% API
+-export([start_link/2]).
+
+%% gen_statem callbacks
+-export([callback_mode/0, init/1, terminate/3, code_change/4]).
+-export([disconnected/3, connected/3]).
+
+%% 消息重发间隔
+-define(RETRY_INTERVAL, 5000).
+
+-record(state, {
+    endpoint :: #endpoint{},
+    buffer :: endpoint_buffer:buffer(),
+    client_id :: atom(),
+    client_pid :: undefined | pid()
+}).
+
+-type kafka_state() :: disconnected | connected.
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+
+-spec start_link(LocalName :: term(), Endpoint :: #endpoint{}) ->
+    {ok, pid()} | ignore | {error, term()}.
+start_link(LocalName, Endpoint = #endpoint{}) ->
+    gen_statem:start_link({via, gproc, {n, l, LocalName}}, ?MODULE, [Endpoint], []).
+
+%%%===================================================================
+%%% gen_statem callbacks
+%%%===================================================================
+
+-spec callback_mode() -> state_functions.
+callback_mode() ->
+    state_functions.
+
+-spec init(term()) -> gen_statem:init_result(kafka_state(), #state{}).
+init([Endpoint = #endpoint{id = Id, matcher = Matcher}]) ->
+    ok = endpoint_util:set_metadata(),
+    erlang:process_flag(trap_exit, true),
+    ok = endpoint_subscription:subscribe(Matcher, self()),
+    Buffer = endpoint_buffer:new(Endpoint, 10),
+    ClientId = list_to_atom("brod_client:" ++ integer_to_list(Id)),
+    {ok, disconnected, #state{endpoint = Endpoint, buffer = Buffer, client_id = ClientId},
+     [{state_timeout, 0, connect}]}.
+
+-spec disconnected(gen_statem:event_type(), term(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+disconnected({call, From}, get_stat, State = #state{buffer = Buffer}) ->
+    reply_stat(From, Buffer, State);
+disconnected(cast, {forward, Metric}, State = #state{buffer = Buffer}) ->
+    store_metric(Metric, Buffer, State);
+disconnected(cast, cleanup, State = #state{buffer = Buffer}) ->
+    cleanup_buffer(Buffer, State);
+disconnected(cast, {reload, NEndpoint = #endpoint{matcher = NMatcher}},
+             State = #state{endpoint = #endpoint{matcher = Matcher}, client_id = ClientId}) ->
+    reload_endpoint(Matcher, NMatcher, ClientId, NEndpoint, State);
+disconnected(state_timeout, connect, State) ->
+    try_connect(State);
+disconnected(info, {timeout, TimerRef, {endpoint_buffer_retry, Id, Payload}}, State = #state{buffer = Buffer}) ->
+    NBuffer = endpoint_buffer:handle_timeout(TimerRef, Id, Payload, Buffer),
+    {keep_state, State#state{buffer = NBuffer}};
+disconnected(info, {next_data, _Id, _Tuple}, State) ->
+    {keep_state, State};
+disconnected(info, {ack, Id}, State = #state{buffer = Buffer}) ->
+    ack_buffer(Id, Buffer, State);
+disconnected(info, Info, State) ->
+    unknown_info(Info, disconnected, State);
+disconnected(EventType, EventContent, State) ->
+    unknown_event(EventType, EventContent, disconnected, State).
+
+-spec connected(gen_statem:event_type(), term(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+connected({call, From}, get_stat, State = #state{buffer = Buffer}) ->
+    reply_stat(From, Buffer, State);
+connected(cast, {forward, Metric}, State = #state{buffer = Buffer}) ->
+    forward_metric(Metric, Buffer, State);
+connected(cast, cleanup, State = #state{buffer = Buffer}) ->
+    cleanup_buffer(Buffer, State);
+connected(cast, {reload, NEndpoint = #endpoint{matcher = NMatcher}},
+          State = #state{endpoint = #endpoint{matcher = Matcher}, client_id = ClientId}) ->
+    reload_endpoint(Matcher, NMatcher, ClientId, NEndpoint, State);
+connected(info, {next_data, Id, Metric},
+          State = #state{
+              buffer = Buffer,
+              client_pid = ClientPid,
+              client_id = ClientId,
+              endpoint = #endpoint{config = #kafka_endpoint{topic = Topic}}
+          }) ->
+    ReceiverPid = self(),
+    AckCb = fun(Partition, BaseOffset) ->
+        logger:debug("[endpoint_kafka] ack partion: ~p, offset: ~p", [Partition, BaseOffset]),
+        ReceiverPid ! {ack, Id}
+    end,
+    case catch brod:produce_cb(ClientPid, Topic, random, <<>>, Metric, AckCb) of
+        {ok, _CallRef} ->
+            {keep_state, State};
+        {ok, _CallRef, _ProducerPid} ->
+            {keep_state, State};
+        {error, Reason} ->
+            logger:warning("[endpoint_kafka] produce topic: ~p, get error: ~p", [Topic, Reason]),
+            stop_kafka_client(ClientId),
+            NBuffer = endpoint_buffer:recover_inflight(Buffer),
+            {next_state, disconnected, State#state{client_pid = undefined, buffer = NBuffer},
+             [{state_timeout, ?RETRY_INTERVAL, connect}]};
+        {'EXIT', Reason} ->
+            logger:warning("[endpoint_kafka] produce topic: ~p, exit with reason: ~p", [Topic, Reason]),
+            stop_kafka_client(ClientId),
+            NBuffer = endpoint_buffer:recover_inflight(Buffer),
+            {next_state, disconnected, State#state{client_pid = undefined, buffer = NBuffer},
+             [{state_timeout, ?RETRY_INTERVAL, connect}]}
+    end;
+connected(info, {timeout, TimerRef, {endpoint_buffer_retry, Id, Payload}}, State = #state{buffer = Buffer}) ->
+    NBuffer = endpoint_buffer:handle_timeout(TimerRef, Id, Payload, Buffer),
+    {keep_state, State#state{buffer = NBuffer}};
+connected(info, {ack, Id}, State = #state{buffer = Buffer}) ->
+    ack_buffer(Id, Buffer, State);
+connected(info, {'EXIT', ClientPid, Reason},
+          State = #state{client_pid = ClientPid, endpoint = #endpoint{title = Title}, buffer = Buffer}) ->
+    logger:warning("[endpoint_kafka] endpoint: ~p, conn pid exit with reason: ~p", [Title, Reason]),
+    NBuffer = endpoint_buffer:recover_inflight(Buffer),
+    {next_state, disconnected, State#state{client_pid = undefined, buffer = NBuffer},
+     [{state_timeout, ?RETRY_INTERVAL, connect}]};
+connected(info, Info, State) ->
+    unknown_info(Info, connected, State);
+connected(EventType, EventContent, State) ->
+    unknown_event(EventType, EventContent, connected, State).
+
+-spec terminate(term(), kafka_state(), #state{}) -> term().
+terminate(Reason, _StateName, #state{endpoint = #endpoint{title = Title}, buffer = Buffer, client_id = ClientId}) ->
+    logger:debug("[endpoint_kafka] endpoint: ~p, terminate with reason: ~p", [Title, Reason]),
+    stop_kafka_client(ClientId),
+    endpoint_buffer:cleanup(Buffer),
+    ok.
+
+-spec code_change(term() | {down, term()}, kafka_state(), #state{}, term()) ->
+    {ok, kafka_state(), #state{}} | {error, term()}.
+code_change(_OldVsn, StateName, State = #state{}, _Extra) ->
+    {ok, StateName, State}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec reply_stat(gen_statem:from(), endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+reply_stat(From, Buffer, State) ->
+    Stat = endpoint_buffer:stat(Buffer),
+    {keep_state, State, [{reply, From, {ok, Stat}}]}.
+
+-spec forward_metric(binary(), endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+forward_metric(Metric, Buffer, State) ->
+    NBuffer = endpoint_buffer:append(Metric, Buffer),
+    {keep_state, State#state{buffer = NBuffer}}.
+
+-spec store_metric(binary(), endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+store_metric(Metric, Buffer, State) ->
+    NBuffer = endpoint_buffer:append_only(Metric, Buffer),
+    {keep_state, State#state{buffer = NBuffer}}.
+
+-spec cleanup_buffer(endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+cleanup_buffer(Buffer, _State) ->
+    NBuffer = endpoint_buffer:cleanup(Buffer),
+    {keep_state, _State#state{buffer = NBuffer}}.
+
+-spec ack_buffer(integer(), endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+ack_buffer(Id, Buffer, State) ->
+    NBuffer = endpoint_buffer:ack(Id, Buffer),
+    {keep_state, State#state{buffer = NBuffer}}.
+
+-spec reload_endpoint(binary(), binary(), atom(), #endpoint{}, #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+reload_endpoint(Matcher, NMatcher, ClientId, NEndpoint, State = #state{}) ->
+    ensure_subscription(Matcher, NMatcher),
+    stop_kafka_client(ClientId),
+    NBuffer = endpoint_buffer:recover_inflight(State#state.buffer),
+    {next_state, disconnected, State#state{endpoint = NEndpoint, client_pid = undefined, buffer = NBuffer},
+     [{state_timeout, 0, connect}]}.
+
+-spec try_connect(#state{}) -> gen_statem:event_handler_result(kafka_state(), #state{}).
+try_connect(State = #state{
+                buffer = Buffer,
+                client_id = ClientId,
+                endpoint = #endpoint{
+                    title = Title,
+                    config = #kafka_endpoint{
+                        sasl_config = SaslConfig,
+                        bootstrap_servers = BootstrapServers,
+                        topic = Topic
+                    }
+                }
+            }) ->
+    logger:debug("[endpoint_kafka] endpoint: ~p, create postman", [Title]),
+    BaseConfig = [
+        {reconnect_cool_down_seconds, 5},
+        {socket_options, [{keepalive, true}]}
+    ],
+    ClientConfig = case SaslConfig of
+                       {Mechanism, Username, Password} ->
+                           [{sasl, {Mechanism, Username, Password}} | BaseConfig];
+                       undefined ->
+                           BaseConfig
+                   end,
+    case catch brod:start_link_client(BootstrapServers, ClientId, ClientConfig) of
+        {ok, ClientPid} ->
+            case brod:start_producer(ClientId, Topic, []) of
+                ok ->
+                    NBuffer = endpoint_buffer:trigger_n(Buffer),
+                    {next_state, connected, State#state{buffer = NBuffer, client_pid = ClientPid}};
+                {error, Reason} ->
+                    logger:debug("[endpoint_kafka] start_producer: ~p, get error: ~p", [ClientId, Reason]),
+                    stop_kafka_client(ClientId),
+                    {keep_state, State#state{client_pid = undefined},
+                     [{state_timeout, ?RETRY_INTERVAL, connect}]}
+            end;
+        Error ->
+            logger:debug("[endpoint_kafka] start_client: ~p, get error: ~p", [ClientId, Error]),
+            {keep_state, State#state{client_pid = undefined},
+             [{state_timeout, ?RETRY_INTERVAL, connect}]}
+    end.
+
+-spec unknown_info(term(), kafka_state(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+unknown_info(Info, StateName, State) ->
+    logger:warning("[endpoint_kafka] unknown message: ~p, status: ~p", [Info, StateName]),
+    {keep_state, State}.
+
+-spec unknown_event(gen_statem:event_type(), term(), kafka_state(), #state{}) ->
+    gen_statem:event_handler_result(kafka_state(), #state{}).
+unknown_event(EventType, EventContent, StateName, State) ->
+    logger:warning("[endpoint_kafka] unknown event: ~p, content: ~p, status: ~p", [EventType, EventContent, StateName]),
+    {keep_state, State}.
+
+-spec ensure_subscription(binary(), binary()) -> ok.
+ensure_subscription(Matcher, Matcher) ->
+    ok;
+ensure_subscription(Matcher, NMatcher) ->
+    ok = endpoint_subscription:unsubscribe(Matcher, self()),
+    endpoint_subscription:subscribe(NMatcher, self()).
+
+-spec stop_kafka_client(atom()) -> ok.
+stop_kafka_client(ClientId) when is_atom(ClientId) ->
+    _ = catch brod:stop_client(ClientId),
+    ok.

apps/endpoint/src/endpoint_log.erl

@@ -0,0 +1,227 @@
+%%%-------------------------------------------------------------------
+%%% @doc Endpoint diagnostic log.
+%%%
+%%% 当前用于记录 publish 时没有匹配到任何订阅者的数据，避免这类信息进入
+%%% endpoint_subscription 的热路径同步磁盘写入。
+%%%-------------------------------------------------------------------
+-module(endpoint_log).
+
+-behaviour(gen_server).
+
+%% API
+-export([start_link/0, unmatched_publish/2]).
+
+%% gen_server callbacks
+-export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
+
+-define(SERVER, ?MODULE).
+-define(LOG_NAME, endpoint_unmatched_publish_log).
+-define(DEFAULT_PATH_TEMPLATE, "${endpoint_root}/endpoint_log/unmatched_publish.log").
+-define(DEFAULT_ENDPOINT_ROOT_DIR, "log").
+-define(DEFAULT_MAX_BYTES, 10485760).
+-define(DEFAULT_MAX_FILES, 10).
+-define(MAX_LOG_CONTENT_BYTES, 32 * 1024).
+
+-record(state, {
+    enabled = false :: boolean(),
+    log_name = ?LOG_NAME :: atom()
+}).
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+
+-spec start_link() -> {ok, pid()} | ignore | {error, term()}.
+start_link() ->
+    gen_server:start_link({local, ?SERVER}, ?MODULE, [], []).
+
+-spec unmatched_publish(RouteKey :: binary(), Content :: binary()) -> ok.
+unmatched_publish(RouteKey, Content) when is_binary(RouteKey), is_binary(Content) ->
+    case erlang:whereis(?SERVER) of
+        undefined ->
+            ok;
+        Pid when is_pid(Pid) ->
+            gen_server:cast(Pid, {unmatched_publish, RouteKey, Content}),
+            ok
+    end.
+
+%%%===================================================================
+%%% gen_server callbacks
+%%%===================================================================
+
+-spec init(term()) -> {ok, #state{}}.
+init([]) ->
+    ok = endpoint_util:set_metadata(),
+    case open_log() of
+        ok ->
+            {ok, #state{enabled = true}};
+        {error, Reason} ->
+            logger:warning("[endpoint_log] open disk_log failed, reason: ~p", [Reason]),
+            {ok, #state{enabled = false}}
+    end.
+
+-spec handle_call(term(), gen_server:from(), #state{}) ->
+    {reply, ok, #state{}}.
+handle_call(_Request, _From, State) ->
+    {reply, ok, State}.
+
+-spec handle_cast(term(), #state{}) -> {noreply, #state{}}.
+handle_cast({unmatched_publish, RouteKey, Content}, State = #state{enabled = true, log_name = LogName}) ->
+    Entry = encode_unmatched_publish(RouteKey, Content),
+    case disk_log:blog(LogName, Entry) of
+        ok ->
+            {noreply, State};
+        {error, Reason} ->
+            logger:warning("[endpoint_log] write unmatched publish log failed, reason: ~p", [Reason]),
+            {noreply, State#state{enabled = false}}
+    end;
+handle_cast({unmatched_publish, _RouteKey, _Content}, State) ->
+    {noreply, State};
+handle_cast(_Request, State) ->
+    {noreply, State}.
+
+-spec handle_info(term(), #state{}) -> {noreply, #state{}}.
+handle_info(Info, State) ->
+    logger:debug("[endpoint_log] unknown info: ~p", [Info]),
+    {noreply, State}.
+
+-spec terminate(term(), #state{}) -> ok.
+terminate(_Reason, #state{enabled = true, log_name = LogName}) ->
+    _ = disk_log:close(LogName),
+    ok;
+terminate(_Reason, _State) ->
+    ok.
+
+-spec code_change(term() | {down, term()}, #state{}, term()) ->
+    {ok, #state{}} | {error, term()}.
+code_change(_OldVsn, State = #state{}, _Extra) ->
+    {ok, State}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec open_log() -> ok | {error, term()}.
+open_log() ->
+    Config = log_config(),
+    Path = log_path(Config),
+    ok = filelib:ensure_dir(Path),
+    case disk_log:open([
+        {name, ?LOG_NAME},
+        {file, Path},
+        {type, wrap},
+        {format, external},
+        {size, {config_pos_integer(max_bytes, Config, ?DEFAULT_MAX_BYTES),
+                config_pos_integer(max_files, Config, ?DEFAULT_MAX_FILES)}},
+        {linkto, self()},
+        {repair, true}
+    ]) of
+        {ok, ?LOG_NAME} ->
+            ok;
+        {repaired, ?LOG_NAME, _Recovered, _BadBytes} ->
+            ok;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec log_config() -> proplists:proplist().
+log_config() ->
+    case application:get_env(endpoint, endpoint_log) of
+        {ok, Config} when is_list(Config) ->
+            Config;
+        _ ->
+            []
+    end.
+
+-spec log_path(proplists:proplist()) -> file:filename_all().
+log_path(Config) ->
+    Template = proplists:get_value(path, Config, ?DEFAULT_PATH_TEMPLATE),
+    expand_path_template(Template).
+
+-spec endpoint_root_dir() -> file:filename_all().
+endpoint_root_dir() ->
+    case application:get_env(endpoint, endpoints) of
+        {ok, Endpoints} ->
+            proplists:get_value(root_dir, Endpoints, ?DEFAULT_ENDPOINT_ROOT_DIR);
+        undefined ->
+            ?DEFAULT_ENDPOINT_ROOT_DIR
+    end.
+
+-spec config_pos_integer(atom(), proplists:proplist(), pos_integer()) -> pos_integer().
+config_pos_integer(Key, Config, Default) ->
+    case proplists:get_value(Key, Config, Default) of
+        Value when is_integer(Value), Value > 0 ->
+            Value;
+        _ ->
+            Default
+    end.
+
+-spec expand_path_template(file:filename_all()) -> file:filename_all().
+expand_path_template(Template) ->
+    TemplateBin = iolist_to_binary(Template),
+    Vars = path_vars(),
+    Expanded = lists:foldl(fun({Name, Value}, Acc) ->
+        replace_path_var(Name, Value, Acc)
+    end, TemplateBin, Vars),
+    binary_to_list(Expanded).
+
+-spec path_vars() -> [{binary(), binary()}].
+path_vars() ->
+    {{Year, Month, Day}, {Hour, _Minute, _Second}} = calendar:local_time(),
+    [
+        {<<"endpoint_root">>, iolist_to_binary(endpoint_root_dir())},
+        {<<"date">>, format_date(Year, Month, Day)},
+        {<<"year">>, integer_to_binary(Year)},
+        {<<"month">>, two_digits(Month)},
+        {<<"day">>, two_digits(Day)},
+        {<<"hour">>, two_digits(Hour)},
+        {<<"node">>, atom_to_binary(node(), utf8)}
+    ].
+
+-spec replace_path_var(binary(), binary(), binary()) -> binary().
+replace_path_var(Name, Value, Template) ->
+    Template0 = binary:replace(Template, <<"${", Name/binary, "}">>, Value, [global]),
+    binary:replace(Template0, <<"{", Name/binary, "}">>, Value, [global]).
+
+-spec encode_unmatched_publish(binary(), binary()) -> binary().
+encode_unmatched_publish(RouteKey, Content) ->
+    {LoggedContent, Truncated} = maybe_truncate(Content),
+    iolist_to_binary([
+        format_local_time(), <<"\t">>,
+        <<"unmatched_publish">>, <<"\t">>,
+        RouteKey, <<"\t">>,
+        integer_to_binary(byte_size(Content)), <<"\t">>,
+        boolean_to_binary(Truncated), <<"\t">>,
+        base64:encode(LoggedContent), <<"\n">>
+    ]).
+
+-spec boolean_to_binary(boolean()) -> binary().
+boolean_to_binary(true) ->
+    <<"true">>;
+boolean_to_binary(false) ->
+    <<"false">>.
+
+-spec format_local_time() -> binary().
+format_local_time() ->
+    MilliSeconds = erlang:system_time(millisecond) rem 1000,
+    {{Year, Month, Day}, {Hour, Minute, Second}} = calendar:local_time(),
+    iolist_to_binary(io_lib:format(
+        "~4..0B-~2..0B-~2..0B ~2..0B:~2..0B:~2..0B.~3..0B",
+        [Year, Month, Day, Hour, Minute, Second, MilliSeconds]
+    )).
+
+-spec format_date(integer(), integer(), integer()) -> binary().
+format_date(Year, Month, Day) ->
+    iolist_to_binary(io_lib:format("~4..0B-~2..0B-~2..0B", [Year, Month, Day])).
+
+-spec two_digits(integer()) -> binary().
+two_digits(Value) ->
+    iolist_to_binary(io_lib:format("~2..0B", [Value])).
+
+-spec maybe_truncate(binary()) -> {binary(), boolean()}.
+maybe_truncate(Content) when byte_size(Content) =< ?MAX_LOG_CONTENT_BYTES ->
+    {Content, false};
+maybe_truncate(Content) ->
+    MaxBytes = ?MAX_LOG_CONTENT_BYTES,
+    <<LoggedContent:MaxBytes/binary, _/binary>> = Content,
+    {LoggedContent, true}.

apps/endpoint/src/endpoint_mqtt.erl

@@ -0,0 +1,285 @@
+%%%-------------------------------------------------------------------
+%%% @author aresei
+%%% @copyright (C) 2023, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 06. 7月 2023 12:02
+%%%-------------------------------------------------------------------
+-module(endpoint_mqtt).
+
+-include("endpoint.hrl").
+-behaviour(gen_statem).
+
+%% API
+-export([start_link/2]).
+
+%% gen_statem callbacks
+-export([callback_mode/0, init/1, terminate/3, code_change/4]).
+-export([disconnected/3, connected/3]).
+
+%% 消息重发间隔
+-define(RETRY_INTERVAL, 15000).
+
+-record(state, {
+    endpoint :: #endpoint{},
+    buffer :: endpoint_buffer:buffer(),
+    conn_pid :: undefined | pid(),
+    %% 待确认的数据, #{PacketId :: integer() => Id :: integer()}
+    inflight = #{}
+}).
+
+-type mqtt_state() :: disconnected | connected.
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+
+-spec start_link(LocalName :: term(), Endpoint :: #endpoint{}) ->
+    {ok, pid()} | ignore | {error, term()}.
+start_link(LocalName, Endpoint = #endpoint{}) ->
+    gen_statem:start_link({via, gproc, {n, l, LocalName}}, ?MODULE, [Endpoint], []).
+
+%%%===================================================================
+%%% gen_statem callbacks
+%%%===================================================================
+
+-spec callback_mode() -> state_functions.
+callback_mode() ->
+    state_functions.
+
+-spec init(term()) -> gen_statem:init_result(mqtt_state(), #state{}).
+init([Endpoint = #endpoint{matcher = Matcher}]) ->
+    ok = endpoint_util:set_metadata(),
+    erlang:process_flag(trap_exit, true),
+    ok = endpoint_subscription:subscribe(Matcher, self()),
+    Buffer = endpoint_buffer:new(Endpoint, 10),
+    {ok, disconnected, #state{endpoint = Endpoint, buffer = Buffer}, [{next_event, internal, connect}]}.
+
+-spec disconnected(gen_statem:event_type(), term(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+disconnected({call, From}, get_stat, State = #state{buffer = Buffer}) ->
+    reply_stat(From, Buffer, State);
+disconnected(cast, {forward, Metric}, State = #state{buffer = Buffer}) ->
+    store_metric(Metric, Buffer, State);
+disconnected(cast, cleanup, State = #state{buffer = Buffer}) ->
+    cleanup_buffer(Buffer, State);
+disconnected(cast, {reload, NEndpoint = #endpoint{matcher = NMatcher}},
+             State = #state{endpoint = #endpoint{matcher = Matcher}, conn_pid = ConnPid}) ->
+    reload_endpoint(Matcher, NMatcher, ConnPid, NEndpoint, State);
+disconnected(internal, connect, State) ->
+    {keep_state, State, [{next_event, internal, do_connect}]};
+disconnected(state_timeout, connect, State) ->
+    {keep_state, State, [{next_event, internal, do_connect}]};
+disconnected(internal, do_connect,
+             State = #state{
+                 buffer = Buffer,
+                 endpoint = #endpoint{
+                     title = Title,
+                     config = #mqtt_endpoint{
+                         host = Host,
+                         port = Port,
+                         username = Username,
+                         password = Password,
+                         client_id = ClientId
+                     }
+                 }
+             }) ->
+    logger:debug("[endpoint_mqtt] endpoint: ~ts, create postman", [Title]),
+    Opts = [
+        {owner, self()},
+        {clientid, ClientId},
+        {host, binary_to_list(Host)},
+        {port, Port},
+        {tcp_opts, []},
+        {username, binary_to_list(Username)},
+        {password, binary_to_list(Password)},
+        {keepalive, 86400},
+        {auto_ack, true},
+        {connect_timeout, 5000},
+        {proto_ver, v5},
+        {retry_interval, 5000}
+    ],
+    case connect_mqtt(Opts) of
+        {ok, ConnPid} ->
+            logger:debug("[endpoint_mqtt] connect success, pid: ~p", [ConnPid]),
+            NBuffer = endpoint_buffer:trigger_n(Buffer),
+            {next_state, connected, State#state{conn_pid = ConnPid, buffer = NBuffer}};
+        {error, Reason} ->
+            logger:warning("[endpoint_mqtt] connect get error: ~p", [Reason]),
+            {keep_state, State#state{conn_pid = undefined, inflight = #{}},
+             [{state_timeout, ?RETRY_INTERVAL, connect}]}
+    end;
+disconnected(info, {timeout, TimerRef, {endpoint_buffer_retry, Id, Payload}}, State = #state{buffer = Buffer}) ->
+    NBuffer = endpoint_buffer:handle_timeout(TimerRef, Id, Payload, Buffer),
+    {keep_state, State#state{buffer = NBuffer}};
+disconnected(info, {next_data, _Id, _Tuple}, State) ->
+    {keep_state, State};
+disconnected(info, {'EXIT', ConnPid, Reason},
+             State = #state{endpoint = #endpoint{title = Title}, conn_pid = ConnPid}) ->
+    logger:warning("[endpoint_mqtt] endpoint: ~p, conn pid exit with reason: ~p", [Title, Reason]),
+    {keep_state, State#state{conn_pid = undefined, inflight = #{}},
+     [{state_timeout, ?RETRY_INTERVAL, connect}]};
+disconnected(info, Info, State) ->
+    unknown_info(Info, disconnected, State);
+disconnected(EventType, EventContent, State) ->
+    unknown_event(EventType, EventContent, disconnected, State).
+
+-spec connected(gen_statem:event_type(), term(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+connected({call, From}, get_stat, State = #state{buffer = Buffer}) ->
+    reply_stat(From, Buffer, State);
+connected(cast, {forward, Metric}, State = #state{buffer = Buffer}) ->
+    forward_metric(Metric, Buffer, State);
+connected(cast, cleanup, State = #state{buffer = Buffer}) ->
+    cleanup_buffer(Buffer, State);
+connected(cast, {reload, NEndpoint = #endpoint{matcher = NMatcher}},
+          State = #state{endpoint = #endpoint{matcher = Matcher}, conn_pid = ConnPid}) ->
+    reload_endpoint(Matcher, NMatcher, ConnPid, NEndpoint, State);
+connected(internal, connect, State) ->
+    {keep_state, State};
+connected(info, {next_data, Id, Metric},
+          State = #state{
+              conn_pid = ConnPid,
+              buffer = Buffer,
+              inflight = InFlight,
+              endpoint = #endpoint{config = #mqtt_endpoint{topic = Topic, qos = Qos}}
+          }) ->
+    logger:debug("[endpoint_mqtt] will publish topic: ~p, metric: ~p, qos: ~p", [Topic, Metric, Qos]),
+    case emqtt:publish(ConnPid, Topic, #{}, Metric, [{qos, Qos}, {retain, true}]) of
+        ok ->
+            NBuffer = endpoint_buffer:ack(Id, Buffer),
+            {keep_state, State#state{buffer = NBuffer}};
+        {ok, PacketId} ->
+            {keep_state, State#state{inflight = maps:put(PacketId, Id, InFlight)}};
+        {error, Reason} ->
+            logger:warning("[endpoint_mqtt] send message to topic: ~p, get error: ~p", [Topic, Reason]),
+            stop_mqtt_conn(ConnPid),
+            NBuffer = endpoint_buffer:recover_inflight(Buffer),
+            {next_state, disconnected, State#state{conn_pid = undefined, inflight = #{}, buffer = NBuffer},
+             [{state_timeout, ?RETRY_INTERVAL, connect}]}
+    end;
+connected(info, {timeout, TimerRef, {endpoint_buffer_retry, Id, Payload}}, State = #state{buffer = Buffer}) ->
+    NBuffer = endpoint_buffer:handle_timeout(TimerRef, Id, Payload, Buffer),
+    {keep_state, State#state{buffer = NBuffer}};
+connected(info, {disconnected, ReasonCode, Properties},
+          State = #state{conn_pid = ConnPid, buffer = Buffer}) ->
+    logger:debug("[endpoint_mqtt] Recv a DISONNECT packet - ReasonCode: ~p, Properties: ~p", [ReasonCode, Properties]),
+    stop_mqtt_conn(ConnPid),
+    NBuffer = endpoint_buffer:recover_inflight(Buffer),
+    {next_state, disconnected, State#state{conn_pid = undefined, inflight = #{}, buffer = NBuffer},
+     [{state_timeout, ?RETRY_INTERVAL, connect}]};
+connected(info, {publish, Message = #{packet_id := _PacketId, payload := Payload}}, State) ->
+    logger:debug("[endpoint_mqtt] Recv a publish packet: ~p, payload: ~p", [Message, Payload]),
+    {keep_state, State};
+connected(info, {puback, #{packet_id := PacketId}},
+          State = #state{inflight = Inflight, buffer = Buffer}) ->
+    case maps:take(PacketId, Inflight) of
+        {Id, RestInflight} ->
+            NBuffer = endpoint_buffer:ack(Id, Buffer),
+            {keep_state, State#state{buffer = NBuffer, inflight = RestInflight}};
+        error ->
+            {keep_state, State}
+    end;
+connected(info, {'EXIT', ConnPid, Reason},
+          State = #state{endpoint = #endpoint{title = Title}, conn_pid = ConnPid, buffer = Buffer}) ->
+    logger:warning("[endpoint_mqtt] endpoint: ~p, conn pid exit with reason: ~p", [Title, Reason]),
+    NBuffer = endpoint_buffer:recover_inflight(Buffer),
+    {next_state, disconnected, State#state{conn_pid = undefined, inflight = #{}, buffer = NBuffer},
+     [{state_timeout, ?RETRY_INTERVAL, connect}]};
+connected(info, Info, State) ->
+    unknown_info(Info, connected, State);
+connected(EventType, EventContent, State) ->
+    unknown_event(EventType, EventContent, connected, State).
+
+-spec terminate(term(), mqtt_state(), #state{}) -> term().
+terminate(Reason, _StateName, #state{endpoint = #endpoint{title = Title}, buffer = Buffer, conn_pid = ConnPid}) ->
+    logger:debug("[endpoint_mqtt] endpoint: ~p, terminate with reason: ~p", [Title, Reason]),
+    stop_mqtt_conn(ConnPid),
+    endpoint_buffer:cleanup(Buffer),
+    ok.
+
+-spec code_change(term() | {down, term()}, mqtt_state(), #state{}, term()) ->
+    {ok, mqtt_state(), #state{}} | {error, term()}.
+code_change(_OldVsn, StateName, State = #state{}, _Extra) ->
+    {ok, StateName, State}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec reply_stat(gen_statem:from(), endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+reply_stat(From, Buffer, State) ->
+    Stat = endpoint_buffer:stat(Buffer),
+    {keep_state, State, [{reply, From, {ok, Stat}}]}.
+
+-spec forward_metric(binary(), endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+forward_metric(Metric, Buffer, State) ->
+    NBuffer = endpoint_buffer:append(Metric, Buffer),
+    {keep_state, State#state{buffer = NBuffer}}.
+
+-spec store_metric(binary(), endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+store_metric(Metric, Buffer, State) ->
+    NBuffer = endpoint_buffer:append_only(Metric, Buffer),
+    {keep_state, State#state{buffer = NBuffer}}.
+
+-spec cleanup_buffer(endpoint_buffer:buffer(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+cleanup_buffer(Buffer, State) ->
+    NBuffer = endpoint_buffer:cleanup(Buffer),
+    {keep_state, State#state{buffer = NBuffer}}.
+
+-spec reload_endpoint(binary(), binary(), undefined | pid(), #endpoint{}, #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+reload_endpoint(Matcher, NMatcher, ConnPid, NEndpoint, State = #state{}) ->
+    ensure_subscription(Matcher, NMatcher),
+    stop_mqtt_conn(ConnPid),
+    NBuffer = endpoint_buffer:recover_inflight(State#state.buffer),
+    {next_state, disconnected, State#state{endpoint = NEndpoint, conn_pid = undefined, inflight = #{}, buffer = NBuffer},
+     [{next_event, internal, do_connect}]}.
+
+-spec unknown_info(term(), mqtt_state(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+unknown_info(Info, StateName, State) ->
+    logger:warning("[endpoint_mqtt] unknown message: ~p, status: ~p", [Info, StateName]),
+    {keep_state, State}.
+
+-spec unknown_event(gen_statem:event_type(), term(), mqtt_state(), #state{}) ->
+    gen_statem:event_handler_result(mqtt_state(), #state{}).
+unknown_event(EventType, EventContent, StateName, State) ->
+    logger:warning("[endpoint_mqtt] unknown event: ~p, content: ~p, status: ~p", [EventType, EventContent, StateName]),
+    {keep_state, State}.
+
+-spec ensure_subscription(binary(), binary()) -> ok.
+ensure_subscription(Matcher, Matcher) ->
+    ok;
+ensure_subscription(Matcher, NMatcher) ->
+    ok = endpoint_subscription:unsubscribe(Matcher, self()),
+    endpoint_subscription:subscribe(NMatcher, self()).
+
+-spec connect_mqtt(list()) -> {ok, pid()} | {error, term()}.
+connect_mqtt(Opts) ->
+    try
+        {ok, ConnPid} = emqtt:start_link(Opts),
+        logger:debug("[endpoint_mqtt] start connect, options: ~p", [Opts]),
+        case emqtt:connect(ConnPid, 5000) of
+            {ok, _} ->
+                {ok, ConnPid};
+            {error, Reason} ->
+                stop_mqtt_conn(ConnPid),
+                {error, Reason}
+        end
+    catch
+        _:Error ->
+            {error, Error}
+    end.
+
+-spec stop_mqtt_conn(undefined | pid()) -> ok.
+stop_mqtt_conn(undefined) ->
+    ok;
+stop_mqtt_conn(ConnPid) when is_pid(ConnPid) ->
+    _ = catch emqtt:stop(ConnPid),
+    ok.

apps/endpoint/src/endpoint_outbox.erl

@@ -0,0 +1,953 @@
+%%%-------------------------------------------------------------------
+%%% @author Codex
+%%% @doc
+%%% endpoint 持久化 outbox 的第一阶段实现：
+%%% - 一个 endpoint 一个目录
+%%% - segment 按递增编号命名
+%%% - metadata 只记录连续 ack 指针
+%%% - inflight 不落盘，崩溃后从 acked_seq + 1 重放
+%%%-------------------------------------------------------------------
+-module(endpoint_outbox).
+
+-export([open/2, close/1, append/2, ack/2, next/1, reset_reader/1, stat/1]).
+-export_type([outbox/0]).
+
+-record(segment, {
+    id :: pos_integer(),
+    path :: file:filename_all(),
+    start_seq = 0 :: non_neg_integer(),
+    end_seq = 0 :: non_neg_integer(),
+    records = 0 :: non_neg_integer(),
+    bytes = 0 :: non_neg_integer()
+}).
+
+-record(outbox_writer, {
+    segment = 1 :: pos_integer(),
+    next_seq = 1 :: pos_integer(),
+    fd = undefined :: undefined | file:fd()
+}).
+
+-record(outbox_reader, {
+    segment :: undefined | pos_integer(),
+    offset = 0 :: non_neg_integer(),
+    fd = undefined :: undefined | file:fd()
+}).
+
+-record(outbox_checkpoint, {
+    acked_segment = 1 :: pos_integer(),
+    acked_seq = 0 :: non_neg_integer(),
+    pending_acks = [] :: [pos_integer()]
+}).
+
+-record(outbox, {
+    dir :: file:filename_all(),
+    metadata_path :: file:filename_all(),
+    max_records = 500000 :: pos_integer(),
+    max_bytes = 134217728 :: pos_integer(),
+    max_segments = 10 :: pos_integer(),
+    segments = [] :: [#segment{}],
+    writer = #outbox_writer{} :: #outbox_writer{},
+    reader = #outbox_reader{} :: #outbox_reader{},
+    checkpoint = #outbox_checkpoint{} :: #outbox_checkpoint{}
+}).
+
+-type outbox() :: #outbox{}.
+
+-define(METADATA_FILE, "metadata.json").
+-define(SEGMENT_EXT, ".log").
+-define(MAX_PAYLOAD_BYTES, 32 * 1024).
+-define(RECORD_MAGIC, <<"EPOBOX01">>).
+-define(RECORD_VERSION, 1).
+-define(RECORD_HEADER_BYTES, 21).
+-define(RECORD_FOOTER_BYTES, 4).
+-define(RECORD_OVERHEAD_BYTES, ?RECORD_HEADER_BYTES + ?RECORD_FOOTER_BYTES).
+-define(RESYNC_READ_BYTES, 4096).
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+
+%% Opts 支持的参数：
+%% - max_records => pos_integer()
+%%     单个 segment 允许的最大记录数，默认 500000
+%% - max_bytes => pos_integer()
+%%     单个 segment 允许的最大字节数，默认 134217728
+%% - max_segments => pos_integer()
+%%     未确认 segment 的最大数量，默认 10
+-spec open(file:filename_all(), map()) -> {ok, outbox()} | {error, term()}.
+open(Dir, Opts) when is_map(Opts) ->
+    ok = ensure_dir(Dir),
+    MetadataPath = filename:join(Dir, ?METADATA_FILE),
+    Segments = load_segments(Dir),
+    Meta = read_metadata(MetadataPath),
+    LastSegmentId = last_segment_id(Segments, maps:get(write_segment, Meta, 1)),
+    LastSeq = last_seq(Segments, maps:get(write_seq, Meta, 0)),
+    AckedSeq = min(maps:get(acked_seq, Meta, 0), LastSeq),
+    AckedSegment = segment_id_for_seq(Segments, AckedSeq, maps:get(acked_segment, Meta, 1)),
+    Outbox0 = #outbox{
+        dir = Dir,
+        metadata_path = MetadataPath,
+        max_records = maps:get(max_records, Opts, 500000),
+        max_bytes = maps:get(max_bytes, Opts, 134217728),
+        max_segments = maps:get(max_segments, Opts, 10),
+        writer = #outbox_writer{
+            segment = LastSegmentId,
+            next_seq = LastSeq + 1
+        },
+        checkpoint = #outbox_checkpoint{
+            acked_segment = AckedSegment,
+            acked_seq = AckedSeq
+        },
+        segments = Segments
+    },
+    {ReadSegment, ReadOffset} = locate_reader(Outbox0),
+    Outbox1 = Outbox0#outbox{
+        reader = #outbox_reader{
+            segment = ReadSegment,
+            offset = ReadOffset
+        }
+    },
+    maybe
+        {ok, Outbox} ?= open_fds(Outbox1),
+        ok ?= persist_metadata(Outbox),
+        {ok, Outbox}
+    else
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec close(outbox()) -> ok.
+close(Outbox = #outbox{}) ->
+    _ = close_open_fds(Outbox),
+    ok.
+
+-spec append(binary(), outbox()) ->
+    {ok, Seq :: pos_integer(), outbox()} | {dropped, capacity_reached, outbox()} | {error, term()}.
+append(Payload, Outbox = #outbox{}) when is_binary(Payload) ->
+    case validate_payload_size(Payload) of
+        ok ->
+            RecordBytes = encoded_record_size(Payload),
+            case ensure_writable_segment(RecordBytes, Outbox) of
+                {drop, NOutbox} ->
+                    {dropped, capacity_reached, NOutbox};
+                {ok, SegmentId, NOutbox0} ->
+                    maybe
+                        {ok, NOutbox1} ?= ensure_write_fd(SegmentId, NOutbox0),
+                        Writer1 = NOutbox1#outbox.writer,
+                        Seq = Writer1#outbox_writer.next_seq,
+                        SegmentPath = segment_path(NOutbox1#outbox.dir, SegmentId),
+                        ok ?= append_record(Writer1#outbox_writer.fd, Seq, Payload),
+                        Segments = update_written_segment(
+                            NOutbox1#outbox.segments, SegmentId, SegmentPath, Seq, RecordBytes),
+                        NOutbox2 = NOutbox1#outbox{
+                            writer = Writer1#outbox_writer{
+                                segment = SegmentId,
+                                next_seq = Seq + 1
+                            },
+                            segments = Segments
+                        },
+                        {ok, NOutbox} ?= maybe_reset_reader_after_append(Outbox, NOutbox2),
+                        ok ?= persist_metadata(NOutbox),
+                        {ok, Seq, NOutbox}
+                    else
+                        {error, Reason} ->
+                            {error, Reason}
+                    end
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec ack(pos_integer(), outbox()) -> {ok, outbox()} | {error, term()}.
+ack(Seq, Outbox = #outbox{checkpoint = Checkpoint}) when is_integer(Seq), Seq > 0 ->
+    AckedSeq = Checkpoint#outbox_checkpoint.acked_seq,
+    case Seq =< AckedSeq of
+        true ->
+            {ok, Outbox};
+        false ->
+            PendingAcks0 = ordsets:add_element(Seq, Checkpoint#outbox_checkpoint.pending_acks),
+            {NackedSeq, PendingAcks} = advance_acked_seq(AckedSeq, PendingAcks0),
+            NackedSegment = segment_id_for_seq(
+                Outbox#outbox.segments,
+                NackedSeq,
+                Checkpoint#outbox_checkpoint.acked_segment),
+            Outbox0 = Outbox#outbox{
+                checkpoint = Checkpoint#outbox_checkpoint{
+                    acked_seq = NackedSeq,
+                    acked_segment = NackedSegment,
+                    pending_acks = PendingAcks
+                }
+            },
+            maybe
+                {ok, Outbox1} ?= prune_acked_segments(Outbox0),
+                ok ?= maybe_persist_metadata(NackedSeq =/= AckedSeq, Outbox1),
+                {ok, Outbox1}
+            else
+                {error, Reason} ->
+                    {error, Reason}
+            end
+    end.
+
+-spec next(outbox()) ->
+    eof | {ok, Seq :: pos_integer(), Payload :: binary(), outbox()} | {error, term()}.
+next(#outbox{reader = #outbox_reader{segment = undefined}}) ->
+    eof;
+next(Outbox = #outbox{
+        reader = Reader,
+        segments = Segments
+    }) ->
+    SegmentId = Reader#outbox_reader.segment,
+    Offset = Reader#outbox_reader.offset,
+    case ensure_read_fd(Outbox) of
+        {ok, NOutbox0} ->
+            Reader0 = NOutbox0#outbox.reader,
+            case read_next_record(Reader0#outbox_reader.fd, Offset) of
+                eof ->
+                    case next_segment_id(Segments, SegmentId) of
+                        undefined ->
+                            eof;
+                        NextSegmentId ->
+                            case switch_read_segment(NextSegmentId, 0, NOutbox0) of
+                                {ok, NOutbox1} ->
+                                    next(NOutbox1);
+                                {error, Reason} ->
+                                    {error, Reason}
+                            end
+                    end;
+                {ok, _RecordOffset, Seq, Payload, NextOffset} ->
+                    {ok, Seq, Payload, NOutbox0#outbox{
+                        reader = Reader0#outbox_reader{offset = NextOffset}
+                    }};
+                {error, Reason} ->
+                    {error, Reason}
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec reset_reader(outbox()) -> {ok, outbox()}.
+reset_reader(Outbox = #outbox{}) ->
+    {ReadSegment, ReadOffset} = locate_reader(Outbox),
+    switch_read_segment(ReadSegment, ReadOffset, Outbox).
+
+-spec stat(outbox()) -> map().
+stat(#outbox{
+        dir = Dir,
+        writer = #outbox_writer{
+            segment = WriteSegment,
+            next_seq = NextSeq
+        },
+        checkpoint = #outbox_checkpoint{
+            acked_segment = AckedSegment,
+            acked_seq = AckedSeq
+        },
+        segments = Segments
+    }) ->
+    #{
+        dir => Dir,
+        write_segment => WriteSegment,
+        write_seq => max(NextSeq - 1, 0),
+        acked_segment => AckedSegment,
+        acked_seq => AckedSeq,
+        segment_num => length(Segments),
+        pending_segment_num => pending_segment_count(Segments, AckedSeq)
+    }.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec ensure_dir(file:filename_all()) -> ok.
+ensure_dir(Dir) ->
+    ok = filelib:ensure_dir(filename:join(Dir, "dummy")),
+    ok.
+
+-spec read_metadata(file:filename_all()) -> map().
+read_metadata(Path) ->
+    case file:read_file(Path) of
+        {ok, Bin} ->
+            case catch json:decode(Bin) of
+                #{<<"write_segment">> := WriteSegment,
+                  <<"write_seq">> := WriteSeq,
+                  <<"acked_segment">> := AckedSegment,
+                  <<"acked_seq">> := AckedSeq}
+                        when is_integer(WriteSegment), is_integer(WriteSeq),
+                             is_integer(AckedSegment), is_integer(AckedSeq) ->
+                    #{
+                        write_segment => WriteSegment,
+                        write_seq => WriteSeq,
+                        acked_segment => AckedSegment,
+                        acked_seq => AckedSeq
+                    };
+                _ ->
+                    default_metadata()
+            end;
+        {error, enoent} ->
+            default_metadata();
+        {error, _} ->
+            default_metadata()
+    end.
+
+-spec default_metadata() -> map().
+default_metadata() ->
+    #{
+        write_segment => 1,
+        write_seq => 0,
+        acked_segment => 1,
+        acked_seq => 0
+    }.
+
+-spec persist_metadata(outbox()) -> ok | {error, term()}.
+persist_metadata(#outbox{
+        metadata_path = MetadataPath,
+        writer = #outbox_writer{
+            segment = WriteSegment,
+            next_seq = NextSeq
+        },
+        checkpoint = #outbox_checkpoint{
+            acked_segment = AckedSegment,
+            acked_seq = AckedSeq
+        }
+    }) ->
+    TmpPath = MetadataPath ++ ".tmp",
+    Metadata = #{
+        <<"write_segment">> => WriteSegment,
+        <<"write_seq">> => max(NextSeq - 1, 0),
+        <<"acked_segment">> => AckedSegment,
+        <<"acked_seq">> => AckedSeq
+    },
+    Bin = iolist_to_binary(json:encode(Metadata)),
+    ok = file:write_file(TmpPath, Bin),
+    ok = file:rename(TmpPath, MetadataPath),
+    ok.
+
+-spec load_segments(file:filename_all()) -> [#segment{}].
+load_segments(Dir) ->
+    Pattern = filename:join(Dir, "*" ++ ?SEGMENT_EXT),
+    SegmentFiles = lists:sort(filelib:wildcard(Pattern)),
+    lists:filtermap(fun(Path) ->
+        case parse_segment_id(filename:basename(Path)) of
+            {ok, Id} ->
+                case scan_segment(Path, Id) of
+                    {ok, Segment = #segment{records = Records}} when Records > 0 ->
+                        {true, Segment};
+                    _ ->
+                        false
+                end;
+            error ->
+                false
+        end
+    end, SegmentFiles).
+
+-spec parse_segment_id(string()) -> {ok, pos_integer()} | error.
+parse_segment_id(Filename) ->
+    case filename:extension(Filename) of
+        ?SEGMENT_EXT ->
+            Base = filename:rootname(Filename, ?SEGMENT_EXT),
+            try
+                Id = list_to_integer(Base),
+                case Id > 0 of
+                    true ->
+                        {ok, Id};
+                    false ->
+                        error
+                end
+            catch
+                _:_ ->
+                    error
+            end;
+        _ ->
+            error
+    end.
+
+-spec scan_segment(file:filename_all(), pos_integer()) -> {ok, #segment{}} | {error, term()}.
+scan_segment(Path, Id) ->
+    case file:open(Path, [read, raw, binary]) of
+        {ok, Fd} ->
+            try
+                scan_segment_records(Fd, 0, #segment{id = Id, path = Path})
+            after
+                ok = file:close(Fd)
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec scan_segment_records(file:fd(), non_neg_integer(), #segment{}) -> {ok, #segment{}} | {error, term()}.
+scan_segment_records(Fd, Offset, Segment = #segment{}) ->
+    case read_next_record(Fd, Offset) of
+        eof ->
+            {ok, Segment};
+        {ok, RecordOffset, Seq, _Payload, NextOffset} ->
+            RecordBytes = NextOffset - RecordOffset,
+            NextSegment = Segment#segment{
+                start_seq = case Segment#segment.records of
+                    0 -> Seq;
+                    _ -> Segment#segment.start_seq
+                end,
+                end_seq = Seq,
+                records = Segment#segment.records + 1,
+                bytes = Segment#segment.bytes + RecordBytes
+            },
+            scan_segment_records(Fd, NextOffset, NextSegment);
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec encoded_record_size(binary()) -> pos_integer().
+encoded_record_size(Payload) ->
+    ?RECORD_OVERHEAD_BYTES + byte_size(Payload).
+
+-spec validate_payload_size(binary()) -> ok | {error, payload_too_large}.
+validate_payload_size(Payload) when byte_size(Payload) =< ?MAX_PAYLOAD_BYTES ->
+    ok;
+validate_payload_size(_Payload) ->
+    {error, payload_too_large}.
+
+-spec ensure_writable_segment(pos_integer(), outbox()) ->
+    {ok, pos_integer(), outbox()} | {drop, outbox()}.
+ensure_writable_segment(RecordBytes, Outbox = #outbox{
+        max_records = MaxRecords,
+        max_bytes = MaxBytes,
+        max_segments = MaxSegments,
+        segments = Segments,
+        writer = #outbox_writer{segment = WriteSegment},
+        checkpoint = #outbox_checkpoint{acked_seq = AckedSeq}
+    }) ->
+    case find_segment(Segments, WriteSegment) of
+        false ->
+            {ok, WriteSegment, Outbox};
+        #segment{records = Records, bytes = Bytes} ->
+            NeedRoll = Records > 0 andalso (Records >= MaxRecords orelse Bytes + RecordBytes > MaxBytes),
+            case NeedRoll of
+                false ->
+                    {ok, WriteSegment, Outbox};
+                true ->
+                    case pending_segment_count(Segments, AckedSeq) >= MaxSegments of
+                        true ->
+                            {drop, Outbox};
+                        false ->
+                            {ok, WriteSegment + 1, Outbox#outbox{
+                                writer = ((Outbox#outbox.writer)#outbox_writer{
+                                    segment = WriteSegment + 1
+                                })
+                            }}
+                    end
+            end
+    end.
+
+-spec append_record(file:fd(), pos_integer(), binary()) -> ok | {error, term()}.
+append_record(Fd, Seq, Payload) when Fd =/= undefined ->
+    PayloadSize = byte_size(Payload),
+    Crc = record_crc(?RECORD_VERSION, Seq, PayloadSize, Payload),
+    Bin = <<
+        ?RECORD_MAGIC/binary,
+        ?RECORD_VERSION:8,
+        Seq:64/unsigned-big-integer,
+        PayloadSize:32/unsigned-big-integer,
+        Payload/binary,
+        Crc:32/unsigned-big-integer
+    >>,
+    case file:write(Fd, Bin) of
+        ok ->
+            ok;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec update_written_segment([#segment{}], pos_integer(), file:filename_all(), pos_integer(), pos_integer()) -> [#segment{}].
+update_written_segment(Segments, SegmentId, Path, Seq, RecordBytes) ->
+    case take_segment(SegmentId, Segments) of
+        {undefined, Rest} ->
+            lists:sort(fun compare_segment/2, [
+                #segment{
+                    id = SegmentId,
+                    path = Path,
+                    start_seq = Seq,
+                    end_seq = Seq,
+                    records = 1,
+                    bytes = RecordBytes
+                } | Rest
+            ]);
+        {Segment, Rest} ->
+            lists:sort(fun compare_segment/2, [
+                Segment#segment{
+                    end_seq = Seq,
+                    records = Segment#segment.records + 1,
+                    bytes = Segment#segment.bytes + RecordBytes
+                } | Rest
+            ])
+    end.
+
+-spec compare_segment(#segment{}, #segment{}) -> boolean().
+compare_segment(#segment{id = Id0}, #segment{id = Id1}) ->
+    Id0 =< Id1.
+
+-spec find_segment([#segment{}], pos_integer()) -> false | #segment{}.
+find_segment(Segments, SegmentId) ->
+    lists:keyfind(SegmentId, #segment.id, Segments).
+
+-spec take_segment(pos_integer(), [#segment{}]) -> {undefined | #segment{}, [#segment{}]}.
+take_segment(SegmentId, Segments) ->
+    case lists:partition(fun(#segment{id = Id}) -> Id =:= SegmentId end, Segments) of
+        {[Segment], Rest} ->
+            {Segment, Rest};
+        {[], Rest} ->
+            {undefined, Rest}
+    end.
+
+-spec last_segment_id([#segment{}], pos_integer()) -> pos_integer().
+last_segment_id([], Default) ->
+    Default;
+last_segment_id(Segments, _Default) ->
+    (lists:last(Segments))#segment.id.
+
+-spec last_seq([#segment{}], non_neg_integer()) -> non_neg_integer().
+last_seq([], Default) ->
+    Default;
+last_seq(Segments, _Default) ->
+    (lists:last(Segments))#segment.end_seq.
+
+-spec segment_id_for_seq([#segment{}], non_neg_integer(), pos_integer()) -> pos_integer().
+segment_id_for_seq(_Segments, 0, Default) ->
+    Default;
+segment_id_for_seq(Segments, Seq, Default) ->
+    case lists:dropwhile(fun(#segment{end_seq = EndSeq}) -> EndSeq < Seq end, Segments) of
+        [#segment{id = SegmentId, start_seq = StartSeq, end_seq = EndSeq} | _]
+                when Seq >= StartSeq, Seq =< EndSeq ->
+            SegmentId;
+        _ ->
+            Default
+    end.
+
+-spec pending_segment_count([#segment{}], non_neg_integer()) -> non_neg_integer().
+pending_segment_count(Segments, AckedSeq) ->
+    length([Segment || Segment = #segment{end_seq = EndSeq} <- Segments, EndSeq > AckedSeq]).
+
+-spec advance_acked_seq(non_neg_integer(), [pos_integer()]) -> {non_neg_integer(), [pos_integer()]}.
+advance_acked_seq(AckedSeq, PendingAcks) ->
+    NextSeq = AckedSeq + 1,
+    case ordsets:is_element(NextSeq, PendingAcks) of
+        true ->
+            advance_acked_seq(NextSeq, ordsets:del_element(NextSeq, PendingAcks));
+        false ->
+            {AckedSeq, PendingAcks}
+    end.
+
+-spec prune_acked_segments(outbox()) -> {ok, outbox()} | {error, term()}.
+prune_acked_segments(Outbox = #outbox{
+        segments = Segments,
+        checkpoint = #outbox_checkpoint{acked_seq = AckedSeq}
+    }) ->
+    {DeleteSegments, KeepSegments} = lists:partition(fun(#segment{end_seq = EndSeq}) ->
+        EndSeq =< AckedSeq
+    end, Segments),
+    Outbox0 = detach_deleted_fds(DeleteSegments, Outbox#outbox{segments = KeepSegments}),
+    ok = lists:foldl(fun(#segment{path = Path}, ok) ->
+        case file:delete(Path) of
+            ok ->
+                ok;
+            {error, enoent} ->
+                ok;
+            {error, Reason} ->
+                throw({delete_segment_failed, Path, Reason})
+        end
+    end, ok, DeleteSegments),
+    {ok, reset_reader_after_prune(Outbox0)}.
+
+-spec reset_reader_after_prune(outbox()) -> outbox().
+reset_reader_after_prune(Outbox = #outbox{
+        reader = #outbox_reader{segment = ReadSegment},
+        segments = Segments
+    }) ->
+    case ReadSegment =:= undefined orelse lists:keymember(ReadSegment, #segment.id, Segments) of
+        true ->
+            Outbox;
+        false ->
+            case reset_reader(Outbox) of
+                {ok, NOutbox} ->
+                    NOutbox
+            end
+    end.
+
+-spec locate_reader(outbox()) -> {undefined | pos_integer(), non_neg_integer()}.
+locate_reader(#outbox{
+        segments = Segments,
+        checkpoint = #outbox_checkpoint{acked_seq = AckedSeq}
+    }) ->
+    NextSeq = AckedSeq + 1,
+    case lists:dropwhile(fun(#segment{end_seq = EndSeq}) -> EndSeq < NextSeq end, Segments) of
+        [] ->
+            {undefined, 0};
+        [#segment{id = SegmentId, start_seq = StartSeq} = Segment | _] ->
+            case NextSeq =< StartSeq of
+                true ->
+                    {SegmentId, 0};
+                false ->
+                    case locate_offset_for_seq(Segment, NextSeq) of
+                        {ok, Offset} ->
+                            {SegmentId, Offset};
+                        {error, _} ->
+                            {SegmentId, 0}
+                    end
+            end
+    end.
+
+-spec locate_offset_for_seq(#segment{}, pos_integer()) -> {ok, non_neg_integer()} | {error, term()}.
+locate_offset_for_seq(#segment{path = Path}, TargetSeq) ->
+    case file:open(Path, [read, raw, binary]) of
+        {ok, Fd} ->
+            try
+                locate_offset_loop(Fd, TargetSeq, 0)
+            after
+                ok = file:close(Fd)
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec locate_offset_loop(file:fd(), pos_integer(), non_neg_integer()) ->
+    {ok, non_neg_integer()} | {error, term()}.
+locate_offset_loop(Fd, TargetSeq, Offset) ->
+    case read_next_record(Fd, Offset) of
+        eof ->
+            {ok, Offset};
+        {ok, RecordOffset, Seq, _Payload, _NextOffset} when Seq >= TargetSeq ->
+            {ok, RecordOffset};
+        {ok, _RecordOffset, _Seq, _Payload, NextOffset} ->
+            locate_offset_loop(Fd, TargetSeq, NextOffset);
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec read_next_record(file:fd(), non_neg_integer()) ->
+    eof | {ok, non_neg_integer(), pos_integer(), binary(), non_neg_integer()} | {error, term()}.
+read_next_record(Fd, Offset) ->
+    case read_record_at(Fd, Offset) of
+        eof ->
+            eof;
+        {ok, Seq, Payload, NextOffset} ->
+            {ok, Offset, Seq, Payload, NextOffset};
+        {error, _Reason} ->
+            resync_next_record(Fd, Offset + 1)
+    end.
+
+-spec read_record_at(file:fd(), non_neg_integer()) ->
+    eof | {ok, pos_integer(), binary(), non_neg_integer()} | {error, term()}.
+read_record_at(Fd, Offset) ->
+    maybe
+        {ok, _} ?= file:position(Fd, Offset),
+        read_record(Fd, Offset)
+    else
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec read_record(file:fd(), non_neg_integer()) ->
+    eof | {ok, pos_integer(), binary(), non_neg_integer()} | {error, term()}.
+read_record(Fd, Offset) ->
+    case file:read(Fd, ?RECORD_HEADER_BYTES) of
+        eof ->
+            eof;
+        {ok, <<
+            Magic:8/binary,
+            Version:8,
+            Seq:64/unsigned-big-integer,
+            PayloadSize:32/unsigned-big-integer
+        >>} when Magic =:= ?RECORD_MAGIC, Version =:= ?RECORD_VERSION ->
+            maybe
+                ok ?= validate_read_payload_size(PayloadSize),
+                read_record_body(Fd, Offset, Seq, PayloadSize)
+            else
+                {error, Reason} ->
+                    {error, Reason}
+            end;
+        {ok, Header} when byte_size(Header) < ?RECORD_HEADER_BYTES ->
+            {error, truncated_header};
+        {ok, _} ->
+            {error, invalid_record_header};
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec validate_read_payload_size(non_neg_integer()) -> ok | {error, payload_too_large}.
+validate_read_payload_size(PayloadSize) when PayloadSize =< ?MAX_PAYLOAD_BYTES ->
+    ok;
+validate_read_payload_size(_PayloadSize) ->
+    {error, payload_too_large}.
+
+-spec read_record_body(file:fd(), non_neg_integer(), pos_integer(), non_neg_integer()) ->
+    {ok, pos_integer(), binary(), non_neg_integer()} | {error, term()}.
+read_record_body(Fd, Offset, Seq, PayloadSize) ->
+    case file:read(Fd, PayloadSize + ?RECORD_FOOTER_BYTES) of
+        {ok, <<Payload:PayloadSize/binary, StoredCrc:32/unsigned-big-integer>>} ->
+            ExpectedCrc = record_crc(?RECORD_VERSION, Seq, PayloadSize, Payload),
+            case StoredCrc =:= ExpectedCrc of
+                true ->
+                    {ok, Seq, Payload, Offset + ?RECORD_OVERHEAD_BYTES + PayloadSize};
+                false ->
+                    {error, invalid_record_crc}
+            end;
+        eof ->
+            {error, truncated_payload};
+        {ok, _} ->
+            {error, truncated_payload};
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec record_crc(byte(), pos_integer(), non_neg_integer(), binary()) -> non_neg_integer().
+record_crc(Version, Seq, PayloadSize, Payload) ->
+    erlang:crc32(<<
+        Version:8,
+        Seq:64/unsigned-big-integer,
+        PayloadSize:32/unsigned-big-integer,
+        Payload/binary
+    >>).
+
+-spec resync_next_record(file:fd(), non_neg_integer()) ->
+    eof | {ok, non_neg_integer(), pos_integer(), binary(), non_neg_integer()} | {error, term()}.
+resync_next_record(Fd, SearchOffset) ->
+    case find_next_magic(Fd, SearchOffset) of
+        eof ->
+            eof;
+        {ok, RecordOffset} ->
+            case read_record_at(Fd, RecordOffset) of
+                eof ->
+                    eof;
+                {ok, Seq, Payload, NextOffset} ->
+                    {ok, RecordOffset, Seq, Payload, NextOffset};
+                {error, _Reason} ->
+                    resync_next_record(Fd, RecordOffset + 1)
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec find_next_magic(file:fd(), non_neg_integer()) ->
+    eof | {ok, non_neg_integer()} | {error, term()}.
+find_next_magic(Fd, Offset) ->
+    maybe
+        {ok, _} ?= file:position(Fd, Offset),
+        find_next_magic_loop(Fd, Offset, <<>>)
+    else
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec find_next_magic_loop(file:fd(), non_neg_integer(), binary()) ->
+    eof | {ok, non_neg_integer()} | {error, term()}.
+find_next_magic_loop(Fd, Offset, Tail) ->
+    case file:read(Fd, ?RESYNC_READ_BYTES) of
+        eof ->
+            eof;
+        {ok, Bin} ->
+            SearchBin = <<Tail/binary, Bin/binary>>,
+            BaseOffset = Offset - byte_size(Tail),
+            case binary:match(SearchBin, ?RECORD_MAGIC) of
+                {Pos, _Len} ->
+                    {ok, BaseOffset + Pos};
+                nomatch ->
+                    NextTail = trailing_bytes(SearchBin, byte_size(?RECORD_MAGIC) - 1),
+                    find_next_magic_loop(Fd, Offset + byte_size(Bin), NextTail)
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec trailing_bytes(binary(), non_neg_integer()) -> binary().
+trailing_bytes(Bin, KeepBytes) ->
+    Size = byte_size(Bin),
+    case Size =< KeepBytes of
+        true ->
+            Bin;
+        false ->
+            binary:part(Bin, Size - KeepBytes, KeepBytes)
+    end.
+
+-spec next_segment_id([#segment{}], pos_integer()) -> undefined | pos_integer().
+next_segment_id(Segments, SegmentId) ->
+    case lists:dropwhile(fun(#segment{id = Id}) -> Id =< SegmentId end, Segments) of
+        [#segment{id = NextSegmentId} | _] ->
+            NextSegmentId;
+        [] ->
+            undefined
+    end.
+
+-spec segment_path(file:filename_all(), pos_integer()) -> file:filename_all().
+segment_path(Dir, SegmentId) ->
+    filename:join(Dir, io_lib:format("~6..0B~s", [SegmentId, ?SEGMENT_EXT])).
+
+-spec open_fds(outbox()) -> {ok, outbox()} | {error, term()}.
+open_fds(Outbox) ->
+    case maybe_open_write_fd(Outbox) of
+        {ok, Outbox0} ->
+            maybe
+                {ok, NOutbox} ?= maybe_open_read_fd(Outbox0),
+                {ok, NOutbox}
+            else
+                {error, Reason} ->
+                    _ = close_open_fds(Outbox0),
+                    {error, Reason}
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec maybe_open_write_fd(outbox()) -> {ok, outbox()} | {error, term()}.
+maybe_open_write_fd(Outbox = #outbox{
+        segments = Segments,
+        writer = #outbox_writer{
+            segment = WriteSegment,
+            fd = undefined
+        }
+    }) ->
+    case find_segment(Segments, WriteSegment) of
+        false ->
+            {ok, Outbox};
+        _ ->
+            ensure_write_fd(WriteSegment, Outbox)
+    end.
+
+-spec maybe_open_read_fd(outbox()) -> {ok, outbox()} | {error, term()}.
+maybe_open_read_fd(Outbox = #outbox{reader = #outbox_reader{segment = undefined}}) ->
+    {ok, Outbox};
+maybe_open_read_fd(Outbox = #outbox{reader = #outbox_reader{fd = undefined}}) ->
+    ensure_read_fd(Outbox);
+maybe_open_read_fd(Outbox = #outbox{}) ->
+    {ok, Outbox}.
+
+-spec ensure_write_fd(pos_integer(), outbox()) -> {ok, outbox()} | {error, term()}.
+ensure_write_fd(SegmentId, Outbox = #outbox{
+        writer = #outbox_writer{
+            segment = SegmentId,
+            fd = Fd
+        }
+    })
+        when Fd =/= undefined ->
+    {ok, Outbox};
+ensure_write_fd(SegmentId, Outbox = #outbox{dir = Dir}) ->
+    Writer0 = (Outbox#outbox.writer)#outbox_writer{segment = SegmentId},
+    Outbox0 = close_write_fd(Outbox#outbox{writer = Writer0}),
+    Path = segment_path(Dir, SegmentId),
+    case file:open(Path, [append, raw, binary]) of
+        {ok, Fd} ->
+            Writer1 = (Outbox0#outbox.writer)#outbox_writer{fd = Fd},
+            {ok, Outbox0#outbox{writer = Writer1}};
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec ensure_read_fd(outbox()) -> {ok, outbox()} | {error, term()}.
+ensure_read_fd(Outbox = #outbox{reader = #outbox_reader{segment = undefined}}) ->
+    {ok, Outbox};
+ensure_read_fd(Outbox = #outbox{reader = #outbox_reader{fd = Fd}}) when Fd =/= undefined ->
+    {ok, Outbox};
+ensure_read_fd(Outbox = #outbox{
+        reader = Reader,
+        dir = Dir
+    }) ->
+    SegmentId = Reader#outbox_reader.segment,
+    Offset = Reader#outbox_reader.offset,
+    Path = segment_path(Dir, SegmentId),
+    case open_read_fd(Path, Offset) of
+        {ok, Fd} ->
+            {ok, Outbox#outbox{
+                reader = Reader#outbox_reader{fd = Fd}
+            }};
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec switch_read_segment(undefined | pos_integer(), non_neg_integer(), outbox()) ->
+    {ok, outbox()} | {error, term()}.
+switch_read_segment(undefined, _Offset, Outbox) ->
+    Reader0 = #outbox_reader{segment = undefined, offset = 0},
+    {ok, close_read_fd(Outbox#outbox{reader = Reader0})};
+switch_read_segment(SegmentId, Offset, Outbox = #outbox{reader = Reader}) ->
+    ensure_read_fd(close_read_fd(Outbox#outbox{
+        reader = Reader#outbox_reader{
+            segment = SegmentId,
+            offset = Offset
+        }
+    })).
+
+-spec maybe_reset_reader_after_append(outbox(), outbox()) -> {ok, outbox()} | {error, term()}.
+maybe_reset_reader_after_append(#outbox{reader = #outbox_reader{segment = undefined}}, Outbox) ->
+    reset_reader(Outbox);
+maybe_reset_reader_after_append(_PrevOutbox, Outbox) ->
+    {ok, Outbox}.
+
+-spec detach_deleted_fds([#segment{}], outbox()) -> outbox().
+detach_deleted_fds(DeleteSegments, Outbox = #outbox{
+        reader = #outbox_reader{segment = ReadSegment},
+        writer = #outbox_writer{segment = WriteSegment}
+    }) ->
+    DeleteIds = [SegmentId || #segment{id = SegmentId} <- DeleteSegments],
+    Outbox0 = case lists:member(ReadSegment, DeleteIds) of
+        true ->
+            close_read_fd(Outbox);
+        false ->
+            Outbox
+    end,
+    case lists:member(WriteSegment, DeleteIds) of
+        true ->
+            close_write_fd(Outbox0);
+        false ->
+            Outbox0
+    end.
+
+-spec close_open_fds(outbox()) -> outbox().
+close_open_fds(Outbox) ->
+    close_write_fd(close_read_fd(Outbox)).
+
+-spec close_read_fd(outbox()) -> outbox().
+close_read_fd(Outbox = #outbox{reader = #outbox_reader{fd = undefined}}) ->
+    Outbox;
+close_read_fd(Outbox = #outbox{reader = Reader}) ->
+    Fd = Reader#outbox_reader.fd,
+    _ = maybe_close_fd(Fd),
+    Outbox#outbox{reader = Reader#outbox_reader{fd = undefined}}.
+
+-spec close_write_fd(outbox()) -> outbox().
+close_write_fd(Outbox = #outbox{writer = #outbox_writer{fd = undefined}}) ->
+    Outbox;
+close_write_fd(Outbox = #outbox{writer = Writer}) ->
+    Fd = Writer#outbox_writer.fd,
+    _ = maybe_close_fd(Fd),
+    Outbox#outbox{writer = Writer#outbox_writer{fd = undefined}}.
+
+-spec maybe_close_fd(file:fd()) -> ok.
+maybe_close_fd(Fd) ->
+    case file:close(Fd) of
+        ok ->
+            ok;
+        {error, badarg} ->
+            ok;
+        {error, terminated} ->
+            ok;
+        {error, _Reason} ->
+            ok
+    end.
+
+-spec maybe_persist_metadata(boolean(), outbox()) -> ok | {error, term()}.
+maybe_persist_metadata(true, Outbox) ->
+    persist_metadata(Outbox);
+maybe_persist_metadata(false, _Outbox) ->
+    ok.
+
+-spec open_read_fd(file:filename_all(), non_neg_integer()) -> {ok, file:fd()} | {error, term()}.
+open_read_fd(Path, Offset) ->
+    case file:open(Path, [read, raw, binary]) of
+        {ok, Fd} ->
+            maybe
+                {ok, _} ?= file:position(Fd, Offset),
+                {ok, Fd}
+            else
+                {error, Reason} ->
+                    ok = file:close(Fd),
+                    {error, Reason}
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.

apps/endpoint/src/endpoint_subscription.erl

@@ -0,0 +1,457 @@
+%%%-------------------------------------------------------------------
+%%% @author anlicheng
+%%% @copyright (C) 2025, <COMPANY>
+%%% @doc
+%%% Endpoint 数据分发订阅索引。
+%%%
+%%% 当前实现面向较大订阅量场景，publish 热路径不再全量遍历订阅表，而是把
+%%% 订阅拆成 exact 和 wildcard 两类：
+%%%
+%%% <ul>
+%%%   <li>精确订阅写入 endpoint_subscription_exact，publish 时按 RouteKey
+%%%       直接 ets:lookup/2。</li>
+%%%   <li>包含通配符的订阅写入 ETS trie。trie 边存放在
+%%%       endpoint_subscription_trie_edge，节点订阅存放在
+%%%       endpoint_subscription_trie_sub。</li>
+%%%   <li>endpoint_subscription_reverse 保存 SubscriberPid 到订阅对象的反向索引，
+%%%       用于 unsubscribe 和进程 DOWN 清理。</li>
+%%%   <li>endpoint_subscription_pid 记录每个 SubscriberPid 的 monitor 和订阅计数，
+%%%       同一个 pid 只 monitor 一次。</li>
+%%% </ul>
+%%%
+%%% 所有 matcher 共用同一个 root node，root node id 固定为 0。新 trie 节点使用
+%%% 单调递增的 integer node id。subscribe/unsubscribe 通过本 gen_server 串行写 ETS；
+%%% publish/2 直接并发读取 ETS，避免中心 gen_server 成为数据分发瓶颈。
+%%%
+%%% 匹配规则：
+%%%
+%%% <ul>
+%%%   <li>普通 segment 必须完全相等，例如 <<"device/a/temp">> 只精确匹配
+%%%       同名订阅。</li>
+%%%   <li><<"*">> 是单级通配，只匹配一个 segment。例如
+%%%       <<"device/*/temp">> 匹配 <<"device/a/temp">>，不匹配
+%%%       <<"device/a/b/temp">>。</li>
+%%%   <li><<"+">> 是末尾多级通配，只允许出现在 matcher 最后一段，并且至少
+%%%       匹配一个剩余 segment。例如 <<"device/+">> 匹配 <<"device/a">>
+%%%       和 <<"device/a/temp">>，不匹配 <<"device">>。</li>
+%%% </ul>
+%%%
+%%% publish/2 的匹配流程：
+%%%
+%%% <ol>
+%%%   <li>从 exact 表直接查找 RouteKey 对应的订阅。</li>
+%%%   <li>把 RouteKey 按 "/" 拆分后，从 trie root 开始逐层匹配。每层最多查找
+%%%       当前 segment 分支和 <<"*">> 分支。</li>
+%%%   <li>每消费一层前，收集当前候选节点上的 plus 订阅；全部 segment 消费完后，
+%%%       收集候选节点上的 exact 订阅。</li>
+%%%   <li>合并 exact 和 trie 命中结果，按原有优先级排序，再按 SubscriberPid 去重。
+%%%       同一个 pid 同时被多个 matcher 命中时，保留优先级最高的订阅。</li>
+%%%   <li>对最终订阅者执行 endpoint:forward/2。</li>
+%%% </ol>
+%%% @end
+%%% Created : 07. 11月 2025 16:27
+%%%-------------------------------------------------------------------
+-module(endpoint_subscription).
+-author("anlicheng").
+
+-behaviour(gen_server).
+
+%% API
+-export([start_link/0]).
+-export([subscribe/2, unsubscribe/2, publish/2, get_subscribers/0]).
+-export([is_valid_components/1, of_components/1]).
+
+%% gen_server callbacks
+-export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
+
+-define(SERVER, ?MODULE).
+-define(EXACT_TAB, endpoint_subscription_exact).
+-define(TRIE_EDGE_TAB, endpoint_subscription_trie_edge).
+-define(TRIE_SUB_TAB, endpoint_subscription_trie_sub).
+-define(REVERSE_TAB, endpoint_subscription_reverse).
+-define(PID_TAB, endpoint_subscription_pid).
+-define(ROOT_NODE, 0).
+
+%% 定义订阅者
+-record(subscriber, {
+    topic :: binary(),
+    subscriber_pid :: pid(),
+    components = [],
+    monitor_ref :: undefined | reference(),
+    %% 优先级
+    %% 1. 完全匹配的topic优先级别最高
+    %% 2. 带 * 的订阅
+    %% 3. 带 + 的订阅
+    order :: integer()
+}).
+
+-record(state, {
+    exact_tid :: ets:tid(),
+    edge_tid :: ets:tid(),
+    trie_sub_tid :: ets:tid(),
+    reverse_tid :: ets:tid(),
+    pid_tid :: ets:tid(),
+    next_node_id = 1 :: pos_integer()
+}).
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+
+-spec subscribe(Topic :: binary(), SubscriberPid :: pid()) -> ok | {error, Reason :: binary()}.
+subscribe(Topic, SubscriberPid) when is_binary(Topic), is_pid(SubscriberPid) ->
+    gen_server:call(?SERVER, {subscribe, Topic, SubscriberPid}).
+
+-spec unsubscribe(Topic :: binary(), SubscriberPid :: pid()) -> ok.
+unsubscribe(Topic, SubscriberPid) when is_binary(Topic), is_pid(SubscriberPid) ->
+    gen_server:call(?SERVER, {unsubscribe, Topic, SubscriberPid}).
+
+-spec get_subscribers() -> {ok, Subscribers :: list()}.
+get_subscribers() ->
+    gen_server:call(?SERVER, get_subscribers).
+
+-spec publish(RouteKey :: binary(), Content :: binary()) -> ok.
+publish(RouteKey, Content) when is_binary(RouteKey), is_binary(Content) ->
+    case ets:info(?EXACT_TAB) of
+        undefined ->
+            ok;
+        _ ->
+            MatchedSubscribers = match_route_key(RouteKey),
+            lists:foreach(fun(#subscriber{subscriber_pid = SubscriberPid}) ->
+                endpoint:forward(SubscriberPid, Content)
+            end, MatchedSubscribers),
+            maybe_log_unmatched_publish(RouteKey, Content, MatchedSubscribers),
+            logger:debug("[endpoint_subscription] route_key: ~p, match_count: ~p", [RouteKey, length(MatchedSubscribers)]),
+            ok
+    end.
+
+%% @doc Spawns the server and registers the local name (unique)
+-spec(start_link() ->
+    {ok, Pid :: pid()} | ignore | {error, Reason :: term()}).
+start_link() ->
+    gen_server:start_link({local, ?SERVER}, ?MODULE, [], []).
+
+%%%===================================================================
+%%% gen_server callbacks
+%%%===================================================================
+
+%% @private
+%% @doc Initializes the server
+-spec(init(Args :: term()) ->
+    {ok, State :: #state{}} | {ok, State :: #state{}, timeout() | hibernate} |
+    {stop, Reason :: term()} | ignore).
+init([]) ->
+    ok = endpoint_util:set_metadata(),
+    ExactTid = ets:new(?EXACT_TAB, [named_table, protected, bag, {read_concurrency, true}]),
+    EdgeTid = ets:new(?TRIE_EDGE_TAB, [named_table, protected, set, {read_concurrency, true}]),
+    TrieSubTid = ets:new(?TRIE_SUB_TAB, [named_table, protected, bag, {read_concurrency, true}]),
+    ReverseTid = ets:new(?REVERSE_TAB, [named_table, protected, bag]),
+    PidTid = ets:new(?PID_TAB, [named_table, protected, set]),
+    {ok, #state{
+        exact_tid = ExactTid,
+        edge_tid = EdgeTid,
+        trie_sub_tid = TrieSubTid,
+        reverse_tid = ReverseTid,
+        pid_tid = PidTid
+    }}.
+
+%% @private
+%% @doc Handling call messages
+-spec(handle_call(Request :: term(), From :: {pid(), Tag :: term()},
+    State :: #state{}) ->
+    {reply, Reply :: term(), NewState :: #state{}} |
+    {reply, Reply :: term(), NewState :: #state{}, timeout() | hibernate} |
+    {noreply, NewState :: #state{}} |
+    {noreply, NewState :: #state{}, timeout() | hibernate} |
+    {stop, Reason :: term(), Reply :: term(), NewState :: #state{}} |
+    {stop, Reason :: term(), NewState :: #state{}}).
+%%  同一个SubscriberPid只能订阅同一个topic一次
+handle_call(get_subscribers, _From, State = #state{exact_tid = ExactTid, trie_sub_tid = TrieSubTid}) ->
+    Subscribers = exact_subscribers(ExactTid) ++ trie_subscribers(TrieSubTid),
+    {reply, {ok, Subscribers}, State};
+
+handle_call({subscribe, Topic, SubscriberPid}, _From, State = #state{reverse_tid = ReverseTid}) ->
+    Components = of_components(Topic),
+    case is_valid_components(Components) of
+        true ->
+            case has_subscription(ReverseTid, Topic, SubscriberPid) of
+                true ->
+                    {reply, ok, State};
+                false ->
+                    {MonitorRef, State1} = ensure_pid_monitor(SubscriberPid, State),
+                    Sub = #subscriber{
+                        topic = Topic,
+                        subscriber_pid = SubscriberPid,
+                        components = Components,
+                        monitor_ref = MonitorRef,
+                        order = order_num(Components)
+                    },
+                    State2 = insert_subscription(Topic, Components, SubscriberPid, Sub, State1),
+                    {reply, ok, State2}
+            end;
+        false ->
+            {reply, {error, <<"invalid topic name">>}, State}
+    end;
+
+handle_call({unsubscribe, Topic, SubscriberPid}, _From, State = #state{reverse_tid = ReverseTid}) ->
+    Removed = [Reverse || Reverse = {SubscriberPid0, Topic0, _Kind, _NodeId, _Sub} <- ets:lookup(ReverseTid, SubscriberPid),
+        SubscriberPid =:= SubscriberPid0, Topic =:= Topic0],
+    lists:foreach(fun(Reverse) -> delete_subscription(Reverse, State) end, Removed),
+    State1 = release_pid_monitor(SubscriberPid, length(Removed), State),
+    {reply, ok, State1}.
+
+%% @private
+%% @doc Handling cast messages
+-spec(handle_cast(Request :: term(), State :: #state{}) ->
+    {noreply, NewState :: #state{}} |
+    {noreply, NewState :: #state{}, timeout() | hibernate} |
+    {stop, Reason :: term(), NewState :: #state{}}).
+handle_cast(_Request, State = #state{}) ->
+    {noreply, State}.
+
+%% @private
+%% @doc Handling all non call/cast messages
+-spec(handle_info(Info :: timeout() | term(), State :: #state{}) ->
+    {noreply, NewState :: #state{}} |
+    {noreply, NewState :: #state{}, timeout() | hibernate} |
+    {stop, Reason :: term(), NewState :: #state{}}).
+handle_info({'DOWN', MonitorRef, process, SubscriberPid, Reason}, State = #state{reverse_tid = ReverseTid, pid_tid = PidTid}) ->
+    logger:debug("[endpoint_subscription] subscriber: ~p, down with reason: ~p", [SubscriberPid, Reason]),
+    Removed = ets:lookup(ReverseTid, SubscriberPid),
+    lists:foreach(fun(Reverse) -> delete_subscription(Reverse, State) end, Removed),
+    case ets:lookup(PidTid, SubscriberPid) of
+        [{SubscriberPid, MonitorRef, _Count}] ->
+            ets:delete(PidTid, SubscriberPid);
+        _ ->
+            ok
+    end,
+    {noreply, State};
+
+handle_info(Info, State = #state{}) ->
+    logger:debug("[endpoint_subscription] get unknown info: ~p", [Info]),
+    {noreply, State}.
+
+%% @private
+%% @doc This function is called by a gen_server when it is about to
+%% terminate. It should be the opposite of Module:init/1 and do any
+%% necessary cleaning up. When it returns, the gen_server terminates
+%% with Reason. The return value is ignored.
+-spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
+    State :: #state{}) -> term()).
+terminate(_Reason, _State = #state{}) ->
+    ok.
+
+%% @private
+%% @doc Convert process state when code is changed
+-spec(code_change(OldVsn :: term() | {down, term()}, State :: #state{},
+    Extra :: term()) ->
+    {ok, NewState :: #state{}} | {error, Reason :: term()}).
+code_change(_OldVsn, State = #state{}, _Extra) ->
+    {ok, State}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec match_route_key(binary()) -> [#subscriber{}].
+match_route_key(RouteKey) when is_binary(RouteKey) ->
+    ExactSubs = [Sub || {_Topic, Sub} <- ets:lookup(?EXACT_TAB, RouteKey)],
+    TrieSubs = match_trie(of_components(RouteKey)),
+    Sorted = lists:sort(fun compare_subscriber/2, ExactSubs ++ TrieSubs),
+    dedupe_subscribers(Sorted).
+
+-spec maybe_log_unmatched_publish(binary(), binary(), [#subscriber{}]) -> ok.
+maybe_log_unmatched_publish(RouteKey, Content, []) ->
+    endpoint_log:unmatched_publish(RouteKey, Content);
+maybe_log_unmatched_publish(_RouteKey, _Content, _MatchedSubscribers) ->
+    ok.
+
+-spec of_components(Topic :: binary()) -> [binary()].
+of_components(Topic) when is_binary(Topic) ->
+    binary:split(Topic, <<$/>>, [global]).
+
+is_valid_components([]) ->
+    true;
+is_valid_components([<<$+>>|T]) ->
+    length(T) =:= 0;
+is_valid_components([<<$*>>|T]) ->
+    is_valid_components(T);
+is_valid_components([_|T]) ->
+    is_valid_components(T).
+
+-spec order_num(Components :: list()) -> integer().
+order_num([]) ->
+    1;
+order_num([<<$*>>|_]) ->
+    2;
+order_num([<<$+>>|_]) ->
+    3;
+order_num([_|Tail]) ->
+    order_num(Tail).
+
+-spec has_subscription(ets:tid(), binary(), pid()) -> boolean().
+has_subscription(ReverseTid, Topic, SubscriberPid) ->
+    Subscriptions = ets:lookup(ReverseTid, SubscriberPid),
+    lists:any(fun({SubscriberPid0, Topic0, _Kind, _NodeId, _Sub}) ->
+        Topic =:= Topic0 andalso SubscriberPid =:= SubscriberPid0
+    end, Subscriptions).
+
+-spec compare_subscriber(#subscriber{}, #subscriber{}) -> boolean().
+compare_subscriber(#subscriber{order = Order0, topic = Topic0}, #subscriber{order = Order1, topic = Topic1}) ->
+    case Order0 =:= Order1 of
+        true ->
+            Topic0 =< Topic1;
+        false ->
+            Order0 < Order1
+    end.
+
+-spec dedupe_subscribers([#subscriber{}]) -> [#subscriber{}].
+dedupe_subscribers(Subscribers) ->
+    {_, Result} = lists:foldl(fun(S = #subscriber{subscriber_pid = SubscriberPid}, {Seen, Acc}) ->
+        case sets:is_element(SubscriberPid, Seen) of
+            true ->
+                {Seen, Acc};
+            false ->
+                {sets:add_element(SubscriberPid, Seen), [S | Acc]}
+        end
+    end, {sets:new(), []}, Subscribers),
+    lists:reverse(Result).
+
+-spec insert_subscription(binary(), [binary()], pid(), #subscriber{}, #state{}) -> #state{}.
+insert_subscription(Topic, Components, SubscriberPid, Sub, State = #state{exact_tid = ExactTid, trie_sub_tid = TrieSubTid, reverse_tid = ReverseTid}) ->
+    case has_wildcard(Components) of
+        false ->
+            true = ets:insert(ExactTid, {Topic, Sub}),
+            true = ets:insert(ReverseTid, {SubscriberPid, Topic, exact, undefined, Sub}),
+            State;
+        true ->
+            {Kind, NodeId, State1} = insert_trie_subscription(Components, State),
+            true = ets:insert(TrieSubTid, {{NodeId, subscription_match_type(Kind)}, Sub}),
+            true = ets:insert(ReverseTid, {SubscriberPid, Topic, Kind, NodeId, Sub}),
+            State1
+    end.
+
+-spec insert_trie_subscription([binary()], #state{}) -> {trie_exact | trie_plus, non_neg_integer(), #state{}}.
+insert_trie_subscription(Components, State) ->
+    case Components of
+        [] ->
+            {trie_exact, ?ROOT_NODE, State};
+        _ ->
+            case lists:last(Components) of
+                <<"+">> ->
+                    Prefix = lists:sublist(Components, length(Components) - 1),
+                    {NodeId, State1} = ensure_path(Prefix, State),
+                    {trie_plus, NodeId, State1};
+                _ ->
+                    {NodeId, State1} = ensure_path(Components, State),
+                    {trie_exact, NodeId, State1}
+            end
+    end.
+
+-spec subscription_match_type(trie_exact | trie_plus) -> exact | plus.
+subscription_match_type(trie_exact) ->
+    exact;
+subscription_match_type(trie_plus) ->
+    plus.
+
+-spec ensure_path([binary()], #state{}) -> {non_neg_integer(), #state{}}.
+ensure_path(Components, State) ->
+    ensure_path(?ROOT_NODE, Components, State).
+
+-spec ensure_path(non_neg_integer(), [binary()], #state{}) -> {non_neg_integer(), #state{}}.
+ensure_path(NodeId, [], State) ->
+    {NodeId, State};
+ensure_path(NodeId, [Segment | Rest], State = #state{edge_tid = EdgeTid, next_node_id = NextNodeId}) ->
+    EdgeKey = {NodeId, Segment},
+    case ets:lookup(EdgeTid, EdgeKey) of
+        [{EdgeKey, ChildNodeId}] ->
+            ensure_path(ChildNodeId, Rest, State);
+        [] ->
+            true = ets:insert(EdgeTid, {EdgeKey, NextNodeId}),
+            ensure_path(NextNodeId, Rest, State#state{next_node_id = NextNodeId + 1})
+    end.
+
+-spec delete_subscription(tuple(), #state{}) -> ok.
+delete_subscription(Reverse = {_SubscriberPid, Topic, exact, undefined, Sub}, #state{exact_tid = ExactTid, reverse_tid = ReverseTid}) ->
+    true = ets:delete_object(ExactTid, {Topic, Sub}),
+    true = ets:delete_object(ReverseTid, Reverse),
+    ok;
+delete_subscription(Reverse = {_SubscriberPid, _Topic, Kind, NodeId, Sub}, #state{trie_sub_tid = TrieSubTid, reverse_tid = ReverseTid}) ->
+    true = ets:delete_object(TrieSubTid, {{NodeId, subscription_match_type(Kind)}, Sub}),
+    true = ets:delete_object(ReverseTid, Reverse),
+    ok.
+
+-spec ensure_pid_monitor(pid(), #state{}) -> {reference(), #state{}}.
+ensure_pid_monitor(SubscriberPid, State = #state{pid_tid = PidTid}) ->
+    case ets:lookup(PidTid, SubscriberPid) of
+        [{SubscriberPid, MonitorRef, Count}] ->
+            true = ets:insert(PidTid, {SubscriberPid, MonitorRef, Count + 1}),
+            {MonitorRef, State};
+        [] ->
+            MonitorRef = erlang:monitor(process, SubscriberPid),
+            true = ets:insert(PidTid, {SubscriberPid, MonitorRef, 1}),
+            {MonitorRef, State}
+    end.
+
+-spec release_pid_monitor(pid(), non_neg_integer(), #state{}) -> #state{}.
+release_pid_monitor(_SubscriberPid, 0, State) ->
+    State;
+release_pid_monitor(SubscriberPid, RemovedCount, State = #state{pid_tid = PidTid}) ->
+    case ets:lookup(PidTid, SubscriberPid) of
+        [{SubscriberPid, MonitorRef, Count}] when Count =< RemovedCount ->
+            erlang:demonitor(MonitorRef, [flush]),
+            ets:delete(PidTid, SubscriberPid),
+            State;
+        [{SubscriberPid, MonitorRef, Count}] ->
+            true = ets:insert(PidTid, {SubscriberPid, MonitorRef, Count - RemovedCount}),
+            State;
+        [] ->
+            State
+    end.
+
+-spec match_trie([binary()]) -> [#subscriber{}].
+match_trie(Components) ->
+    match_trie([?ROOT_NODE], Components, []).
+
+-spec match_trie([non_neg_integer()], [binary()], [#subscriber{}]) -> [#subscriber{}].
+match_trie(NodeIds, [], Acc) ->
+    collect_trie_subscribers(NodeIds, exact) ++ Acc;
+match_trie(NodeIds, [Segment | Rest], Acc) ->
+    PlusSubs = collect_trie_subscribers(NodeIds, plus),
+    NextNodeIds = next_trie_nodes(NodeIds, Segment),
+    match_trie(NextNodeIds, Rest, PlusSubs ++ Acc).
+
+-spec collect_trie_subscribers([non_neg_integer()], exact | plus) -> [#subscriber{}].
+collect_trie_subscribers(NodeIds, MatchType) ->
+    lists:flatmap(fun(NodeId) ->
+        [Sub || {{NodeId0, MatchType0}, Sub} <- ets:lookup(?TRIE_SUB_TAB, {NodeId, MatchType}),
+            NodeId0 =:= NodeId, MatchType0 =:= MatchType]
+    end, NodeIds).
+
+-spec next_trie_nodes([non_neg_integer()], binary()) -> [non_neg_integer()].
+next_trie_nodes(NodeIds, Segment) ->
+    lists:usort(lists:flatmap(fun(NodeId) ->
+        lookup_child(NodeId, Segment) ++ lookup_child(NodeId, <<"*">>)
+    end, NodeIds)).
+
+-spec lookup_child(non_neg_integer(), binary()) -> [non_neg_integer()].
+lookup_child(NodeId, Segment) ->
+    EdgeKey = {NodeId, Segment},
+    case ets:lookup(?TRIE_EDGE_TAB, EdgeKey) of
+        [{EdgeKey, ChildNodeId}] ->
+            [ChildNodeId];
+        [] ->
+            []
+    end.
+
+-spec has_wildcard([binary()]) -> boolean().
+has_wildcard(Components) ->
+    lists:any(fun(Component) -> Component =:= <<"*">> orelse Component =:= <<"+">> end, Components).
+
+-spec exact_subscribers(ets:tid()) -> [#subscriber{}].
+exact_subscribers(ExactTid) ->
+    [Sub || {_Topic, Sub} <- ets:tab2list(ExactTid)].
+
+-spec trie_subscribers(ets:tid()) -> [#subscriber{}].
+trie_subscribers(TrieSubTid) ->
+    [Sub || {_Key, Sub} <- ets:tab2list(TrieSubTid)].

apps/endpoint/src/endpoint_sup.erl

@@ -3,7 +3,7 @@
 %% @end
 %%%-------------------------------------------------------------------
 
--module(endpoint_sup_sup).
+-module(endpoint_sup).
 
 -behaviour(supervisor).
 -include("endpoint.hrl").
@@ -28,6 +28,15 @@ start_link() ->
 init([]) ->
     SupFlags = #{strategy => one_for_all, intensity => 1000, period => 3600},
     ChildSpecs = [
+        #{
+            id => endpoint_log,
+            start => {'endpoint_log', start_link, []},
+            restart => permanent,
+            shutdown => 2000,
+            type => worker,
+            modules => ['endpoint_log']
+        },
+
         #{
             id => endpoint_subscription,
             start => {'endpoint_subscription', start_link, []},
@@ -38,12 +47,12 @@ init([]) ->
         },
 
         #{
-            id => 'endpoint_sup',
+            id => 'endpoint_adapter_sup',
-            start => {'endpoint_sup', start_link, []},
+            start => {'endpoint_adapter_sup', start_link, []},
             restart => permanent,
             shutdown => 2000,
             type => supervisor,
-            modules => ['endpoint_sup']
+            modules => ['endpoint_adapter_sup']
         }
     ],
     {ok, {SupFlags, ChildSpecs}}.

apps/endpoint/src/endpoint_tester.erl

@@ -0,0 +1,93 @@
+%%%-------------------------------------------------------------------
+%%% @author anlicheng
+%%% @copyright (C) 2026, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 11. 5月 2026 20:16
+%%%-------------------------------------------------------------------
+-module(endpoint_tester).
+-author("anlicheng").
+-include("endpoint.hrl").
+
+%% API
+-export([test/1]).
+
+%% http接口的测试
+test(#http_endpoint{url = Url, pool_size = PoolSize}) when is_integer(PoolSize), PoolSize > 0 ->
+    Body = <<"">>,
+    ContentType = "application/json",
+    case httpc:request(post, {Url, [], ContentType, Body}, [], []) of
+        {ok, _} ->
+            ok;
+        {error, Reason} ->
+            {error, Reason}
+    end;
+
+%% 测试mqtt
+test(#mqtt_endpoint{host = Host, port = Port, username = Username, password = Password}) ->
+    %% 用户测试的client
+    ClientId = "mqtt_client_test:" ++ iot_util:rand_bytes(16),
+    Opts = [
+        {owner, self()},
+        {clientid, ClientId},
+        {host, binary_to_list(Host)},
+        {port, Port},
+        {tcp_opts, []},
+        {username, binary_to_list(Username)},
+        {password, binary_to_list(Password)},
+        {keepalive, 86400},
+        {auto_ack, true},
+        {connect_timeout, 5000},
+        {proto_ver, v5},
+        {retry_interval, 5000}
+    ],
+
+    case emqtt:start_link(Opts) of
+        {ok, ConnPid} ->
+            case catch emqtt:connect(ConnPid, 5000) of
+                {ok, _} ->
+                    emqtt:stop(ConnPid),
+                    ok;
+                {error, _Reason} ->
+                    {error, <<"connect mqtt server failed">>};
+                _Error ->
+                    emqtt:stop(ConnPid),
+                    {error, <<"connect mqtt server failed">>}
+            end;
+        Other ->
+            logger:warning("[endpint_handler] test connect mqtt with options: ~p, get error: ~p", [Opts, Other]),
+            {error, <<"connect mqtt server failed">>}
+    end;
+
+%% 测试kafka
+test(#kafka_endpoint{sasl_config = SaslConfig, bootstrap_servers = BootstrapServers, topic = Topic}) ->
+    BaseConfig = [
+        {reconnect_cool_down_seconds, 5},
+        {socket_options, [{keepalive, true}]}
+    ],
+
+    ClientConfig = case SaslConfig of
+                       {Mechanism, Username, Password} ->
+                           [{sasl, {Mechanism, Username, Password}}|BaseConfig];
+                       undefined ->
+                           BaseConfig
+                   end,
+    ClientId = brod_client_test,
+    _ = catch brod:stop_client(ClientId),
+
+    case catch brod:start_link_client(BootstrapServers, ClientId, ClientConfig) of
+        {ok, _ClientPid} ->
+            case brod:start_producer(ClientId, Topic, _ProducerConfig = []) of
+                ok ->
+                    ok = brod:stop_client(ClientId),
+                    ok;
+                {error, Reason} ->
+                    logger:debug("[endpint_handler] start_producer: ~p, get error: ~p", [ClientId, Reason]),
+                    _ = catch brod:stop_client(ClientId),
+                    {error, <<"config kafka server failed">>}
+            end;
+        Error ->
+            logger:debug("[endpint_handler] start_client: ~p, get error: ~p", [ClientId, Error]),
+            {error, <<"config kafka server failed">>}
+    end.

apps/endpoint/src/endpoint_util.erl

@@ -0,0 +1,18 @@
+%%%-------------------------------------------------------------------
+%%% @doc Endpoint-local utility functions.
+%%%-------------------------------------------------------------------
+-module(endpoint_util).
+
+-export([set_metadata/0, sha256/1]).
+
+-spec set_metadata() -> ok.
+set_metadata() ->
+    logger:set_process_metadata(#{domain => [endpoint]}).
+
+-spec sha256(string() | binary()) -> binary().
+sha256(Str) when is_list(Str) ->
+    sha256(unicode:characters_to_binary(Str));
+sha256(Bin) when is_binary(Bin) ->
+    HashBin = crypto:hash(sha256, Bin),
+    HexStr = lists:flatten([io_lib:format("~2.16.0B", [B]) || B <- binary:bin_to_list(HashBin)]),
+    list_to_binary(string:lowercase(HexStr)).

apps/iot/include/domain_model.hrl

@@ -0,0 +1,20 @@
+%%%-------------------------------------------------------------------
+%%% @author Codex
+%%% @doc
+%%% 领域模型定义
+%%% @end
+%%%-------------------------------------------------------------------
+
+-record(host_info, {
+    id :: integer(),
+    uuid :: binary(),
+    authorize_status :: integer(),
+    status :: integer()
+}).
+
+-record(device_info, {
+    id :: integer(),
+    host_id :: integer(),
+    device_uuid :: binary(),
+    status :: integer()
+}).

apps/iot/include/iot.hrl

@@ -13,6 +13,10 @@
 -define(HOST_ONLINE, 1).
 -define(HOST_NOT_JOINED, -1).
 
+%% 主机是否授权处理上报数据
+-define(HOST_DENIED, 0).
+-define(HOST_AUTHORIZED, 1).
+
 %% 设备是否在线状态
 -define(DEVICE_OFFLINE, 0).
 -define(DEVICE_ONLINE, 1).
@@ -22,52 +26,3 @@
 -define(TASK_STATUS_INIT,   -1).  %% 未接入
 -define(TASK_STATUS_FAILED, 0).    %% 离线
 -define(TASK_STATUS_OK,     1).     %% 在线
-
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%%%% 二级分类定义
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
-%% 缓存数据库表
--record(kv, {
-    key :: binary(),
-    val :: binary() | list() | map() | sets:set(),
-    expire_at = 0 :: integer(),
-    type :: atom()
-}).
-
-%% id生成器
--record(id_generator, {
-    tab :: atom(),
-    increment_id = 0 :: integer()
-}).
-
-%% 统计项
--record(option, {
-    success_num = 0,
-    fail_num = 0
-}).
-
-%% 北向数据
--record(north_data, {
-    id = 0 :: integer(),
-    location_code :: binary(),
-    %% 数据库类型的endpoint， 可以返回list: [{K, V}, {K1, V1}]
-    fields :: [{K :: binary(), V :: any()}],
-    timestamp = 0 :: integer()
-}).
-
-%% 事件相关的数据
--record(event_data, {
-    id = 0 :: integer(),
-    location_code :: binary(),
-    event_type :: integer(),
-    params :: map()
-}).
-
-%% 发送数据
--record(post_data, {
-    id = 0 :: integer(),
-    location_code :: binary(),
-    %% 数据库类型的endpoint， 可以返回list: [{K, V}, {K1, V1}]
-    body :: binary() | list()
-}).

apps/iot/include/iot_tables.hrl

@@ -1,23 +1,17 @@
 %%%-------------------------------------------------------------------
 %%% @author anlicheng
-%%% @copyright (C) 2025, <COMPANY>
+%%% @copyright (C) 2026, <COMPANY>
 %%% @doc
 %%%
 %%% @end
-%%% Created : 08. 5月 2025 12:08
+%%% Created : 24. 4月 2026 16:54
 %%%-------------------------------------------------------------------
 -author("anlicheng").
 
-%% 用来保存微服务的配置
+-record(efka_client, {
--record(service_config, {
+    uuid :: binary(),
-    service_id :: binary(),
+    token_hash :: binary(),
-    host_uuid :: binary(),
+    salt :: binary(),
-    config_json = <<>> :: binary(),
+    heartbeat_secret :: binary(),
-    %% 保持上一个版本的内容，错误时回滚
+    timestamp = 0 :: integer()
-    last_config_json = <<>> :: binary(),
-    %% 最后一次修改的用户id
-    last_edit_user :: integer(),
-    %% 状态： 0: 停止, 1: 运行中
-    update_ts = 0,
-    create_ts = 0
 }).

apps/iot/include/message.hrl

@@ -1,86 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author anlicheng
-%%% @copyright (C) 2025, <COMPANY>
-%%% @doc
-%%% 扩展部分, 1: 支持基于topic的pub/sub机制; 2. 基于target的单点通讯和广播
-%%% @end
-%%% Created : 21. 4月 2025 17:28
-%%%-------------------------------------------------------------------
--author("anlicheng").
-
-%% efka主动发起的消息体类型, 消息大类
--define(PACKET_REQUEST, 16#01).
--define(PACKET_RESPONSE, 16#02).
-
-%% efka主动发起不需要返回的数据
--define(PACKET_CAST, 16#03).
-
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%%%% 二级分类定义
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
-%% 主机端上报数据类型标识
--define(MESSAGE_AUTH_REQUEST, 16#01).
--define(MESSAGE_AUTH_REPLY, 16#02).
-
--define(MESSAGE_COMMAND, 16#03).
--define(MESSAGE_DEPLOY, 16#04).
--define(MESSAGE_PUB, 16#05).
-
--define(MESSAGE_DATA, 16#06).
--define(MESSAGE_EVENT, 16#07).
-
-%% efka主动上报的event-stream流, 单向消息，主要是: docker-create的实时处理逻辑上报
--define(MESSAGE_EVENT_STREAM, 16#08).
-
--define(MESSAGE_JSONRPC_REQUEST, 16#F0).
--define(MESSAGE_JSONRPC_REPLY, 16#F1).
-
-%%%% 命令类型子分类, 不需要返回值
-%% 授权
--define(COMMAND_AUTH, 16#08).
-
--record(auth_request, {
-    uuid :: binary(),
-    username :: binary(),
-    salt :: binary(),
-    token :: binary(),
-    timestamp :: integer()
-}).
-
--record(auth_reply, {
-    code :: integer(),
-    payload :: binary()
-}).
-
--record(pub, {
-    topic :: binary(),
-    qos = 0 :: integer(),
-    content :: binary()
-}).
-
--record(command, {
-    command_type :: integer(),
-    command :: binary()
-}).
-
--record(jsonrpc_request, {
-    method :: binary(),
-    params = <<>> :: any()
-}).
-
--record(jsonrpc_reply, {
-    result :: any() | undefined,
-    error :: any() | undefined
-}).
-
--record(data, {
-    route_key :: binary(),
-    metric :: binary()
-}).
-
--record(task_event_stream, {
-    task_id :: integer(),
-    type :: binary(),
-    stream :: binary()
-}).

apps/iot/include/protocol.hrl

@@ -0,0 +1,16 @@
+%%%-------------------------------------------------------------------
+%%% @author anlicheng
+%%% @copyright (C) 2025, <COMPANY>
+%%% @doc
+%%% 传输层协议常量定义
+%%% @end
+%%%-------------------------------------------------------------------
+-author("anlicheng").
+
+%% 一级帧类型
+%% REQUEST: 需要响应的请求帧
+%% REPLY: 对 REQUEST 的响应帧
+%% CAST: 单向消息帧，不需要响应
+-define(FRAME_REQUEST, 16#01).
+-define(FRAME_REPLY, 16#02).
+-define(FRAME_CAST, 16#03).

apps/iot/priv/2024-01-18-log.txt

@@ -1,20 +0,0 @@
-{"device_uuid":"20118448129232486417014256831677", "key":"epi", "label":"kWh", "name":"正向总有功电能", "timestamp":1705576081, "type":"AI", "unit":5, "value":314498}
-{"device_uuid":"20118448129232486417014256831677", "key":"a_voltage", "label":"V", "name":"A相电压", "timestamp":1705576081, "type":"AI", "unit":0, "value":224.1}
-{"device_uuid":"20118448129232486417014256831677", "key":"a_current", "label":"A", "name":"A相电流", "timestamp":1705576081, "type":"AI", "unit":3, "value":0.394}
-{"device_uuid":"20118448129232486417014256831677", "key":"b_voltage", "label":"V", "name":"B相电压", "timestamp":1705576081, "type":"AI", "unit":0, "value":225.3}
-{"device_uuid":"20118448129232486417014256831677", "key":"b_current", "label":"A", "name":"B相电流", "timestamp":1705576081, "type":"AI", "unit":3, "value":0.323}
-{"device_uuid":"20118448129232486417014256831677", "key":"c_voltage", "label":"V", "name":"C相电压", "timestamp":1705576081, "type":"AI", "unit":0, "value":225}
-{"device_uuid":"20118448129232486417014256831677", "key":"c_current", "label":"A", "name":"C相电流", "timestamp":1705576081, "type":"AI", "unit":3, "value":0.269}
-{"device_uuid":"20118448129232486417014256831677", "key":"active_power", "label":"kW", "name":"瞬时总有功功率", "timestamp":1705576081, "type":"AI", "unit":23, "value":20.91}
-{"device_uuid":"20118448129232486417014256831677", "key":"power_factor", "label":"", "name":"总功率因数", "timestamp":1705576081, "type":"AI", "unit":16, "value":0.994}
-{"device_uuid":"20118448549921177617014256841701", "key":"epi", "label":"kWh", "name":"正向总有功电能", "timestamp":1705576038, "type":"AI", "unit":5, "value":299493.6}
-{"device_uuid":"20118448549921177617014256841701", "key":"a_voltage", "label":"V", "name":"A相电压", "timestamp":1705576038, "type":"AI", "unit":0, "value":225.9}
-{"device_uuid":"20118448549921177617014256841701", "key":"a_current", "label":"A", "name":"A相电流", "timestamp":1705576038, "type":"AI", "unit":3, "value":0.318}
-{"device_uuid":"20118448549921177617014256841701", "key":"b_voltage", "label":"V", "name":"B相电压", "timestamp":1705576038, "type":"AI", "unit":0, "value":225.7}
-{"device_uuid":"20118448549921177617014256841701", "key":"b_current", "label":"A", "name":"B相电流", "timestamp":1705576038, "type":"AI", "unit":3, "value":0.305}
-{"device_uuid":"20118448549921177617014256841701", "key":"c_voltage", "label":"V", "name":"C相电压", "timestamp":1705576038, "type":"AI", "unit":0, "value":225}
-{"device_uuid":"20118448549921177617014256841701", "key":"c_current", "label":"A", "name":"C相电流", "timestamp":1705576038, "type":"AI", "unit":3, "value":0.3}
-{"device_uuid":"20118448549921177617014256841701", "key":"active_power", "label":"kW", "name":"瞬时总有功功率", "timestamp":1705576038, "type":"AI", "unit":23, "value":15.056}
-{"device_uuid":"20118448549921177617014256841701", "key":"power_factor", "label":"", "name":"总功率因数", "timestamp":1705576038, "type":"AI", "unit":16, "value":1}
-{"device_uuid":"20118448970190438417014256851724", "key":"epi", "label":"kWh", "name":"正向总有功电能", "timestamp":1705576117, "type":"AI", "unit":5, "value":285011}
-{"device_uuid":"20118448970190438417014256851724", "key":"a_voltage", "label":"V", "name":"A相电压", "timestamp":1705576117, "type":"AI", "unit":0, "value":225.1}

apps/iot/priv/ssl/server.crt

@@ -0,0 +1,23 @@
+-----BEGIN CERTIFICATE-----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+-----END CERTIFICATE-----

apps/iot/priv/ssl/server.csr

@@ -0,0 +1,18 @@
+-----BEGIN CERTIFICATE REQUEST-----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+-----END CERTIFICATE REQUEST-----

apps/iot/priv/ssl/server.key

@@ -0,0 +1,28 @@
+-----BEGIN PRIVATE KEY-----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+-----END PRIVATE KEY-----

apps/iot/src/adapters/control_api/iot_api_client.erl

@@ -6,8 +6,9 @@
 %%% @end
 %%% Created : 24. 12月 2023 15:42
 %%%-------------------------------------------------------------------
--module(iot_api).
+-module(iot_api_client).
 -author("anlicheng").
+-include("domain_model.hrl").
 
 %% API
 -export([ai_event/1]).
@@ -31,20 +32,30 @@ get_all_hosts() ->
             []
     end.
 
--spec get_host_by_uuid(UUID :: binary()) -> undefined | {ok, HostInfo :: map()}.
+-spec get_host_by_uuid(UUID :: binary()) -> undefined | {ok, HostInfo :: #host_info{}}.
 get_host_by_uuid(UUID) when is_binary(UUID) ->
     case do_get("/get_host_by_uuid", [{<<"uuid">>, UUID}]) of
         {ok, HostInfo} ->
-            {ok, HostInfo};
+            case host_info_record(HostInfo) of
+                {ok, Record} ->
+                    {ok, Record};
+                error ->
+                    undefined
+            end;
         _ ->
             undefined
     end.
 
--spec get_host_by_id(HostId :: integer()) -> undefined | {ok, HostInfo :: map()}.
+-spec get_host_by_id(HostId :: integer()) -> undefined | {ok, HostInfo :: #host_info{}}.
 get_host_by_id(HostId) when is_integer(HostId) ->
     case do_get("/get_host_by_id", [{<<"host_id">>, integer_to_binary(HostId)}]) of
         {ok, HostInfo} ->
-            {ok, HostInfo};
+            case host_info_record(HostInfo) of
+                {ok, Record} ->
+                    {ok, Record};
+                error ->
+                    undefined
+            end;
         _ ->
             undefined
     end.
@@ -54,15 +65,27 @@ get_host_by_id(HostId) when is_integer(HostId) ->
 change_host_status(UUID, NStatus) when is_binary(UUID), is_integer(NStatus) ->
     do_post("/change_host_status", #{<<"uuid">> => UUID, <<"new_status">> => NStatus}).
 
--spec get_host_devices(HostId :: integer()) -> {ok, Devices :: [map()]} | {error, Reason::any()}.
+-spec get_host_devices(HostId :: integer()) -> {ok, Devices :: [#device_info{}]} | {error, Reason::any()}.
 get_host_devices(HostId) when is_integer(HostId) ->
-    do_get("/get_host_devices", [{<<"host_id">>, integer_to_binary(HostId)}]).
+    case do_get("/get_host_devices", [{<<"host_id">>, integer_to_binary(HostId)}]) of
+        {ok, DeviceInfos} when is_list(DeviceInfos) ->
+            device_info_records(DeviceInfos);
+        {ok, _Other} ->
+            {error, invalid_device_infos};
+        Error ->
+            Error
+    end.
 
--spec get_device_by_uuid(DeviceUUID :: binary()) -> {ok, DeviceInfo :: map()} | undefined.
+-spec get_device_by_uuid(DeviceUUID :: binary()) -> {ok, DeviceInfo :: #device_info{}} | undefined.
 get_device_by_uuid(DeviceUUID) when is_binary(DeviceUUID) ->
     case do_get("/get_device_by_uuid", [{<<"device_uuid">>, DeviceUUID}]) of
         {ok, DeviceInfo} ->
-            {ok, DeviceInfo};
+            case device_info_record(DeviceInfo) of
+                {ok, Record} ->
+                    {ok, Record};
+                error ->
+                    undefined
+            end;
         _ ->
             undefined
     end.
@@ -106,18 +129,18 @@ ai_event(Id) when is_integer(Id) ->
         <<"token">> => Token,
         <<"id">> => Id
     },
-    Body = iolist_to_binary(jiffy:encode(ReqData, [force_utf8])),
+    Body = iolist_to_binary(json:encode(ReqData)),
     case hackney:request(post, Url, Headers, Body, [{pool, false}]) of
         {ok, 200, _, ClientRef} ->
             {ok, RespBody} = hackney:body(ClientRef),
-            lager:debug("[iot_api] send body: ~p, get error is: ~p", [Body, RespBody]),
+            logger:debug("[iot_api_client] send body: ~p, get error is: ~p", [Body, RespBody]),
             hackney:close(ClientRef);
         {ok, HttpCode, _, ClientRef} ->
             {ok, RespBody} = hackney:body(ClientRef),
             hackney:close(ClientRef),
-            lager:warning("[iot_api] send body: ~p, get error is: ~p", [Body, {HttpCode, RespBody}]);
+            logger:warning("[iot_api_client] send body: ~p, get error is: ~p", [Body, {HttpCode, RespBody}]);
         {error, Reason} ->
-            lager:warning("[iot_api] send body: ~p, get error is: ~p", [Body, Reason])
+            logger:warning("[iot_api_client] send body: ~p, get error is: ~p", [Body, Reason])
     end.
 
 %%%-------------------------------------------------------------------
@@ -132,13 +155,13 @@ do_post(Path, Params) when is_list(Path), is_map(Params) ->
         {<<"Accept">>, <<"application/json">>}
     ],
     Url = BaseUrl ++ Path,
-    Body = iolist_to_binary(jiffy:encode(Params, [force_utf8])),
+    Body = iolist_to_binary(json:encode(Params)),
     case hackney:request(post, Url, Headers, Body, [{pool, false}]) of
         {ok, 200, _, ClientRef} ->
             {ok, RespBody} = hackney:body(ClientRef),
-            lager:debug("[iot_api] request url: ~p, send body: ~p, get response is: ~p", [Url, Body, RespBody]),
+            logger:debug("[iot_api_client] request url: ~p, send body: ~p, get response is: ~p", [Url, Body, RespBody]),
             hackney:close(ClientRef),
-            case catch jiffy:decode(RespBody, [return_maps]) of
+            case catch json:decode(RespBody) of
                 #{<<"result">> := Result} ->
                     {ok, Result};
                 #{<<"error">> := #{<<"code">> := Code, <<"message">> := Message}} ->
@@ -151,10 +174,10 @@ do_post(Path, Params) when is_list(Path), is_map(Params) ->
         {ok, HttpCode, _, ClientRef} ->
             {ok, RespBody} = hackney:body(ClientRef),
             hackney:close(ClientRef),
-            lager:warning("[iot_api] request url: ~p, send body: ~p, get error is: ~p", [Url, Body, {HttpCode, RespBody}]),
+            logger:warning("[iot_api_client] request url: ~p, send body: ~p, get error is: ~p", [Url, Body, {HttpCode, RespBody}]),
             {error, {HttpCode, RespBody}};
         {error, Reason} ->
-            lager:warning("[iot_api] request url: ~p, send body: ~p, get error is: ~p", [Url, Body, Reason]),
+            logger:warning("[iot_api_client] request url: ~p, send body: ~p, get error is: ~p", [Url, Body, Reason]),
             {error, Reason}
     end.
 
@@ -177,8 +200,8 @@ do_get(Path, Params) when is_list(Path), is_list(Params) ->
         {ok, 200, _, ClientRef} ->
             {ok, RespBody} = hackney:body(ClientRef),
             hackney:close(ClientRef),
-            lager:debug("[iot_api] url: ~p, get response is: ~p", [Url, RespBody]),
+            logger:debug("[iot_api_client] url: ~p, get response is: ~p", [Url, RespBody]),
-            case catch jiffy:decode(RespBody, [return_maps]) of
+            case catch json:decode(RespBody) of
                 #{<<"result">> := Result} ->
                     {ok, Result};
                 #{<<"error">> := #{<<"code">> := Code, <<"message">> := Message}} ->
@@ -191,9 +214,46 @@ do_get(Path, Params) when is_list(Path), is_list(Params) ->
         {ok, HttpCode, _, ClientRef} ->
             {ok, RespBody} = hackney:body(ClientRef),
             hackney:close(ClientRef),
-            lager:warning("[iot_api] request url: ~p, get error is: ~p", [Url, {HttpCode, RespBody}]),
+            logger:warning("[iot_api_client] request url: ~p, get error is: ~p", [Url, {HttpCode, RespBody}]),
             {error, {HttpCode, RespBody}};
         {error, Reason} ->
-            lager:warning("[iot_api] request url: ~p, get error is: ~p", [Url, Reason]),
+            logger:warning("[iot_api_client] request url: ~p, get error is: ~p", [Url, Reason]),
             {error, Reason}
     end.
+
+-spec host_info_record(map()) -> {ok, #host_info{}} | error.
+host_info_record(#{<<"id">> := Id, <<"uuid">> := UUID, <<"authorize_status">> := AuthorizeStatus, <<"status">> := Status})
+        when is_integer(Id), is_binary(UUID), is_integer(AuthorizeStatus), is_integer(Status) ->
+    {ok, #host_info{
+        id = Id,
+        uuid = UUID,
+        authorize_status = AuthorizeStatus,
+        status = Status
+    }};
+host_info_record(_) ->
+    error.
+
+-spec device_info_record(map()) -> {ok, #device_info{}} | error.
+device_info_record(#{<<"id">> := Id, <<"host_id">> := HostId, <<"device_uuid">> := DeviceUUID, <<"status">> := Status})
+        when is_integer(Id), is_integer(HostId), is_binary(DeviceUUID), is_integer(Status) ->
+    {ok, #device_info{
+        id = Id,
+        host_id = HostId,
+        device_uuid = DeviceUUID,
+        status = Status
+    }};
+device_info_record(_) ->
+    error.
+
+-spec device_info_records([map()]) -> {ok, [#device_info{}]} | {error, invalid_device_info}.
+device_info_records(DeviceInfos) ->
+    lists:foldr(fun(DeviceInfo, Acc) ->
+        case {device_info_record(DeviceInfo), Acc} of
+            {{ok, Record}, {ok, Records}} ->
+                {ok, [Record | Records]};
+            {error, _} ->
+                {error, invalid_device_info};
+            {_, Error = {error, _}} ->
+                Error
+        end
+    end, {ok, []}, DeviceInfos).

apps/iot/src/adapters/influxdb/influx_client.erl

@@ -57,7 +57,7 @@ get_precision(Timestamp) when is_integer(Timestamp) ->
             <<"ms">>
     end.
 
--spec write_data(Measurement :: binary(), Tags :: map(), FieldsList :: list(), Timestamp :: integer()) -> no_return().
+-spec write_data(Measurement :: binary(), Tags :: map(), FieldsList :: list(), Timestamp :: integer()) -> term().
 write_data(Measurement, Tags, FieldsList, Timestamp) when is_binary(Measurement), is_map(Tags), is_list(FieldsList), is_integer(Timestamp) ->
     %% 过来掉没有key的选项
     NFieldsList = lists:filter(fun data_filter/1, FieldsList),
@@ -76,12 +76,12 @@ write_data(Measurement, Tags, FieldsList, Timestamp) when is_binary(Measurement)
             ok
     end.
 
--spec write(Pid :: pid(), Bucket :: binary(), Org :: binary(), Points :: list()) -> no_return().
+-spec write(Pid :: pid(), Bucket :: binary(), Org :: binary(), Points :: list()) -> ok.
 write(Pid, Bucket, Org, Points) when is_pid(Pid), is_binary(Bucket), is_binary(Org), is_list(Points) ->
     write(Pid, Bucket, Org, <<"ms">>, Points).
 
 %% Precision的值为: ms|ns|s; 表示时间的精度，默认为毫秒(ms)
--spec write(Pid :: pid(), Bucket :: binary(), Org :: binary(), Precision :: binary(), Points :: list()) -> no_return().
+-spec write(Pid :: pid(), Bucket :: binary(), Org :: binary(), Precision :: binary(), Points :: list()) -> ok.
 write(Pid, Bucket, Org, Precision, Points) when is_pid(Pid), is_binary(Bucket), is_binary(Org), is_binary(Precision), is_list(Points) ->
     gen_server:cast(Pid, {write, Bucket, Org, Precision, Points}).
 
@@ -101,6 +101,7 @@ start_link(Opts) when is_list(Opts) ->
     {ok, State :: #state{}} | {ok, State :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term()} | ignore).
 init([InfluxProps]) ->
+    ok = iot_log:set_metadata(),
     Token = proplists:get_value(token, InfluxProps),
     Host = proplists:get_value(host, InfluxProps),
     Port = proplists:get_value(port, InfluxProps),
@@ -151,18 +152,18 @@ handle_cast({write, Bucket, Org, Precision, Points}, State = #state{host = Host,
         query => Query
     }),
 
-    lager:debug("[influx_client] url is: ~p, headers: ~p, body: ~ts", [Url, Headers, Body]),
+    logger:debug("[influx_client] url is: ~p, headers: ~p, body: ~ts", [Url, Headers, Body]),
     case hackney:request(post, Url, Headers, GZipBody, [{pool, false}]) of
         {ok, StatusCode, _RespHeaders, ClientRef} ->
             case hackney:body(ClientRef) of
                 {ok, RespBody} ->
-                    lager:debug("[influx_client] status_code: ~p, response body is: ~p", [StatusCode, RespBody]);
+                    logger:debug("[influx_client] status_code: ~p, response body is: ~p", [StatusCode, RespBody]);
                 {error, Error} ->
-                    lager:warning("[influx_client] status_code: ~p, error is: ~p", [Error])
+                    logger:warning("[influx_client] status_code: ~p, error is: ~p", [Error])
             end,
             hackney:close(ClientRef);
         {error, Reason} ->
-            lager:warning("[influx_client] request result is: ~p", [Reason])
+            logger:warning("[influx_client] request result is: ~p", [Reason])
     end,
 
     {noreply, State}.

apps/iot/src/adapters/influxdb/influx_point.erl

@@ -45,7 +45,7 @@ field_val(V) when is_float(V) ->
 field_val(V) when is_binary(V) ->
     <<$", V/binary, $">>;
 field_val(V) when is_list(V); is_map(V) ->
-    S = base64:encode(iolist_to_binary(jiffy:encode(V, [force_utf8]))),
+    S = base64:encode(iolist_to_binary(json:encode(V))),
     <<$", "base64:", S/binary, $">>;
 field_val(true) ->
     <<"true">>;
@@ -56,4 +56,3 @@ as_list(L) when is_list(L) ->
     L;
 as_list(L) when is_map(L) ->
     maps:to_list(L).
-

apps/iot/src/devtools/endpoint_kafka_test.erl

@@ -8,12 +8,12 @@
 %%%-------------------------------------------------------------------
 -module(endpoint_kafka_test).
 -author("anlicheng").
--include("endpoint.hrl").
+-include_lib("endpoint/include/endpoint.hrl").
 
 %% API
--export([start_test/0, test_consumer/0]).
+-export([start_manual/0, test_consumer/0]).
 
-start_test() ->
+start_manual() ->
     Name = endpoint:get_name(100),
     {ok, Pid} = endpoint_kafka:start_link(Name, #endpoint{
         id = 100,
@@ -29,10 +29,10 @@ start_test() ->
         }
     }),
 
-    Json = jiffy:encode(#{
+    Json = iolist_to_binary(json:encode(#{
         <<"name">> => <<"anlicheng">>,
         <<"age">> => 30
-    }, [force_utf8]),
+    })),
 
     endpoint:forward(Pid, Json),
     ok.
@@ -49,7 +49,7 @@ test_consumer() ->
 
     ok = brod:start_client(KafkaBootstrapEndpoints, client1, ClientConfig),
     SubscriberCallbackFun = fun(_Partition, Msg, ShellPid = CallbackState) ->
-        lager:debug("call here msg: ~p", [Msg]),
+        logger:debug("call here msg: ~p", [Msg]),
         ShellPid ! Msg, {ok, ack, CallbackState}
                             end,
     Res = brod_topic_subscriber:start_link(client1, Topic, all,
@@ -57,4 +57,4 @@ test_consumer() ->
         _CommittedOffsets=[], message, SubscriberCallbackFun,
         _CallbackState=self()),
 
-    lager:debug("start subscriber res: ~p", [Res]).
+    logger:debug("start subscriber res: ~p", [Res]).

apps/iot/src/devtools/endpoint_mqtt_subscriber.erl

@@ -67,23 +67,23 @@ init([]) ->
         {retry_interval, 5}
     ],
 
-    lager:debug("[opts] is: ~p", [Opts]),
+    logger:debug("[opts] is: ~p", [Opts]),
     case emqtt:start_link(Opts) of
         {ok, ConnPid} ->
             %% 监听和host相关的全部事件
-            lager:debug("[iot_mqtt_subscriber] start conntecting, pid: ~p", [ConnPid]),
+            logger:debug("[iot_mqtt_subscriber] start conntecting, pid: ~p", [ConnPid]),
             {ok, _} = emqtt:connect(ConnPid),
-            lager:debug("[iot_mqtt_subscriber] connect success, pid: ~p", [ConnPid]),
+            logger:debug("[iot_mqtt_subscriber] connect success, pid: ~p", [ConnPid]),
             SubscribeResult = emqtt:subscribe(ConnPid, ?Topics),
 
-            lager:debug("[iot_mqtt_subscriber] subscribe topics: ~p, result is: ~p", [?Topics, SubscribeResult]),
+            logger:debug("[iot_mqtt_subscriber] subscribe topics: ~p, result is: ~p", [?Topics, SubscribeResult]),
 
             {ok, #state{conn_pid = ConnPid}};
         ignore ->
-            lager:debug("[iot_mqtt_subscriber] connect emqx get ignore"),
+            logger:debug("[iot_mqtt_subscriber] connect emqx get ignore"),
             {stop, ignore};
         {error, Reason} ->
-            lager:debug("[iot_mqtt_subscriber] connect emqx get error: ~p", [Reason]),
+            logger:debug("[iot_mqtt_subscriber] connect emqx get error: ~p", [Reason]),
             {stop, Reason}
     end.
 
@@ -116,19 +116,19 @@ handle_cast(_Request, State = #state{}) ->
     {noreply, NewState :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term(), NewState :: #state{}}).
 handle_info({disconnect, ReasonCode, Properties}, State = #state{}) ->
-    lager:debug("[iot_mqtt_subscriber] Recv a DISONNECT packet - ReasonCode: ~p, Properties: ~p", [ReasonCode, Properties]),
+    logger:debug("[iot_mqtt_subscriber] Recv a DISONNECT packet - ReasonCode: ~p, Properties: ~p", [ReasonCode, Properties]),
     {stop, disconnected, State};
 %% 必须要做到消息的快速分发，数据的json反序列需要在host进程进行
 handle_info({publish, #{packet_id := _PacketId, payload := Payload, qos := Qos, topic := Topic}}, State = #state{conn_pid = _ConnPid}) ->
-    lager:debug("[iot_mqtt_subscriber] Recv a topic: ~p, publish packet: ~p, qos: ~p", [Topic, Payload, Qos]),
+    logger:debug("[iot_mqtt_subscriber] Recv a topic: ~p, publish packet: ~p, qos: ~p", [Topic, Payload, Qos]),
     %% 将消息分发到对应的host进程去处理
     {noreply, State};
 handle_info({puback, Packet = #{packet_id := _PacketId}}, State = #state{}) ->
-    lager:debug("[iot_mqtt_subscriber] receive puback packet: ~p", [Packet]),
+    logger:debug("[iot_mqtt_subscriber] receive puback packet: ~p", [Packet]),
     {noreply, State};
 
 handle_info(Info, State = #state{}) ->
-    lager:debug("[iot_mqtt_subscriber] get info: ~p", [Info]),
+    logger:debug("[iot_mqtt_subscriber] get info: ~p", [Info]),
     {noreply, State}.
 
 %% @private
@@ -144,10 +144,10 @@ terminate(Reason, _State = #state{conn_pid = ConnPid}) when is_pid(ConnPid) ->
     {ok, _Props, _ReasonCode} = emqtt:unsubscribe(ConnPid, #{}, TopicNames),
 
     ok = emqtt:disconnect(ConnPid),
-    lager:debug("[iot_mqtt_subscriber] terminate with reason: ~p", [Reason]),
+    logger:debug("[iot_mqtt_subscriber] terminate with reason: ~p", [Reason]),
     ok;
 terminate(Reason, _State) ->
-    lager:debug("[iot_mqtt_subscriber] terminate with reason: ~p", [Reason]),
+    logger:debug("[iot_mqtt_subscriber] terminate with reason: ~p", [Reason]),
     ok.
 
 %% @private

apps/iot/src/devtools/eval_test.erl

@@ -13,13 +13,5 @@
 -export([test/0]).
 
 test() ->
-
+    {error, disabled}.
-    {ok, Content} = file:read_file("/tmp/test.erl"),
-
-    {ok, Tokens, _} = erl_scan:string(binary_to_list(Content)),
-    {ok, ExprList} = erl_parse:parse_exprs(Tokens),
-
-    {value, F, _NewBindings} = erl_eval:exprs(ExprList, []),
-    F(#{name => <<"test">>}).
-
 

apps/iot/src/devtools/iot_mock.erl

@@ -12,7 +12,6 @@
 
 %% API
 -export([rsa_encode/1]).
--export([insert_services/1]).
 -export([test_influxdb/0]).
 
 test_influxdb() ->
@@ -29,34 +28,20 @@ test_influxdb() ->
                                          end)
                   end, lists:seq(1, 100)).
 
-insert_services(Num) ->
-    lists:foreach(fun(Id) ->
-        Res = mysql_pool:insert(mysql_iot, <<"micro_service">>,
-            #{
-                <<"name">> => <<"微服务"/utf8, (integer_to_binary(Id))/binary>>,
-                <<"code">> => <<"1223423423423423"/utf8>>,
-                <<"type">> => 1,
-                <<"version">> => <<"v1.0">>,
-                <<"url">> => <<"https://www.baidu.com">>,
-                <<"detail">> => <<"这是一个关于测试的微服务"/utf8>>
-            }, false),
-        lager:debug("insert service result is: ~p", [Res])
-                  end, lists:seq(1, Num)).
-
 rsa_encode(Data) when is_binary(Data) ->
     %% 读取相关配置
     PublicPemFile = "/tmp/keys/public.pem",
 
     %% 私钥保存解析后的
     {ok, PubBin} = file:read_file(PublicPemFile),
-    lager:debug("pub bin is: ~p", [PubBin]),
+    logger:debug("pub bin is: ~p", [PubBin]),
     [Pub] = public_key:pem_decode(PubBin),
-    lager:debug("pub pem bin is: ~p", [Pub]),
+    logger:debug("pub pem bin is: ~p", [Pub]),
     PubKey = public_key:pem_entry_decode(Pub),
-    lager:debug("the public key is: ~p", [PubKey]),
+    logger:debug("the public key is: ~p", [PubKey]),
 
     EncData = public_key:encrypt_public(Data, PubKey),
-    lager:debug("enc data is: ~p", [EncData]),
+    logger:debug("enc data is: ~p", [EncData]),
 
     rsa_decode(EncData),
 
@@ -68,13 +53,13 @@ rsa_decode(EncData) when is_binary(EncData) ->
 
     %% 私钥保存解析后的
     {ok, PubBin} = file:read_file(PublicPemFile),
-    lager:debug("pub bin is: ~p", [PubBin]),
+    logger:debug("pub bin is: ~p", [PubBin]),
     [Pub] = public_key:pem_decode(PubBin),
-    lager:debug("pub pem bin is: ~p", [Pub]),
+    logger:debug("pub pem bin is: ~p", [Pub]),
     PubKey = public_key:pem_entry_decode(Pub),
-    lager:debug("the public key is: ~p", [PubKey]),
+    logger:debug("the public key is: ~p", [PubKey]),
 
     PlainData = public_key:decrypt_private(EncData, PubKey),
-    lager:debug("plain data is: ~p", [PlainData]),
+    logger:debug("plain data is: ~p", [PlainData]),
 
     ok.

apps/iot/src/docker/docker_container_builder.erl

@@ -0,0 +1,588 @@
+%%%-------------------------------------------------------------------
+%%% @author
+%%% @copyright (C) 2026, <COMPANY>
+%%% @doc
+%%% ContainerRequest wire builder helpers.
+%%% @end
+%%%-------------------------------------------------------------------
+-module(docker_container_builder).
+
+-export([list_request/0, config_request/2, deploy_request/2, start_request/1, stop_request/1, kill_request/1, remove_request/1]).
+
+-spec list_request() -> map().
+list_request() ->
+    #{action => list, all => true}.
+
+-spec config_request(binary(), binary()) -> map().
+config_request(ContainerName, ConfigJson) when is_binary(ContainerName), is_binary(ConfigJson) ->
+    #{
+        action => config,
+        target => container_ref(ContainerName),
+        config => ConfigJson
+    }.
+
+-spec deploy_request(integer(), map()) -> {ok, map()} | {error, binary()}.
+deploy_request(TaskId, Config) when is_integer(TaskId), is_map(Config), TaskId >= 0 ->
+    try
+        ensure_supported_deploy_config(Config),
+        validate_deploy_config(Config),
+        Params = build_container_deploy_params(Config),
+        {ok, #{action => deploy, task_id => TaskId, params => Params}}
+    catch
+        throw:{error, Reason} ->
+            {error, Reason}
+    end;
+deploy_request(TaskId, _Config) when is_integer(TaskId), TaskId < 0 ->
+    {error, <<"task_id must be non-negative">>};
+deploy_request(_TaskId, _Config) ->
+    {error, <<"invalid deploy request">>}.
+
+-spec start_request(binary()) -> map().
+start_request(ContainerName) when is_binary(ContainerName) ->
+    #{action => start, target => container_ref(ContainerName)}.
+
+-spec stop_request(binary()) -> map().
+stop_request(ContainerName) when is_binary(ContainerName) ->
+    #{
+        action => stop,
+        target => container_ref(ContainerName),
+        timeout_seconds => 0
+    }.
+
+-spec kill_request(binary()) -> map().
+kill_request(ContainerName) when is_binary(ContainerName) ->
+    #{
+        action => kill,
+        target => container_ref(ContainerName),
+        signal => <<>>
+    }.
+
+-spec remove_request(binary()) -> map().
+remove_request(ContainerName) when is_binary(ContainerName) ->
+    #{
+        action => remove,
+        target => container_ref(ContainerName),
+        force => false,
+        remove_volumes => false
+    }.
+
+-spec container_ref(binary()) -> map().
+container_ref(ContainerName) when is_binary(ContainerName) ->
+    #{name => ContainerName, id => <<>>}.
+
+-spec ensure_supported_deploy_config(map()) -> ok.
+ensure_supported_deploy_config(Config) when is_map(Config) ->
+    UnsupportedKeys = [Key || Key <- [<<"container_dir">>], maps:is_key(Key, Config)],
+    case UnsupportedKeys of
+        [] ->
+            ok;
+        _ ->
+            Unsupported = iolist_to_binary(lists:join(<<", ">>, UnsupportedKeys)),
+            throw({error, <<"unsupported container config keys: ", Unsupported/binary>>})
+    end.
+
+-spec validate_deploy_config(map()) -> ok.
+validate_deploy_config(Config) when is_map(Config) ->
+    Required = [
+        {<<"image">>, binary},
+        {<<"container_name">>, binary},
+        {<<"command">>, {list, binary}},
+        {<<"restart">>, binary}
+    ],
+    Optional = [
+        {<<"privileged">>, boolean},
+        {<<"entrypoint">>, {list, binary}},
+        {<<"envs">>, {list, binary}},
+        {<<"ports">>, {list, binary}},
+        {<<"expose">>, {list, binary}},
+        {<<"volumes">>, {list, binary}},
+        {<<"networks">>, {list, binary}},
+        {<<"labels">>, {map, {binary, binary}}},
+        {<<"user">>, binary},
+        {<<"working_dir">>, binary},
+        {<<"hostname">>, binary},
+        {<<"network_mode">>, binary},
+        {<<"cap_add">>, {list, binary}},
+        {<<"cap_drop">>, {list, binary}},
+        {<<"devices">>, {list, binary}},
+        {<<"mem_limit">>, binary},
+        {<<"mem_reservation">>, binary},
+        {<<"cpu_shares">>, non_neg_integer},
+        {<<"cpus">>, non_neg_number},
+        {<<"ulimits">>, {map, {binary, binary}}},
+        {<<"sysctls">>, {map, {binary, binary}}},
+        {<<"tmpfs">>, {list, binary}},
+        {<<"extra_hosts">>, {list, binary}},
+        {<<"healthcheck">>, map}
+    ],
+    Errors = check_required(Config, Required) ++ check_optional(Config, Optional) ++ check_healthcheck(Config),
+    case Errors of
+        [] ->
+            ok;
+        _ ->
+            throw({error, iolist_to_binary(lists:join(<<"|||">>, Errors))})
+    end.
+
+-spec check_required(map(), list()) -> [binary()].
+check_required(Config, Fields) ->
+    lists:foldl(
+        fun({Key, Type}, ErrAcc) ->
+            case maps:get(Key, Config, undefined) of
+                undefined ->
+                    [iolist_to_binary(io_lib:format("miss requied parameter: ~p", [Key])) | ErrAcc];
+                Value ->
+                    case check_type(Value, Type) of
+                        true ->
+                            ErrAcc;
+                        false ->
+                            [iolist_to_binary(io_lib:format("required parameter: ~p, type must be: ~ts", [Key, type_name(Type)])) | ErrAcc]
+                    end
+            end
+        end,
+        [], Fields).
+
+-spec check_optional(map(), list()) -> [binary()].
+check_optional(Config, Fields) ->
+    lists:foldl(
+        fun({Key, Type}, ErrAcc) ->
+            case maps:get(Key, Config, undefined) of
+                undefined ->
+                    ErrAcc;
+                Value ->
+                    case check_type(Value, Type) of
+                        true ->
+                            ErrAcc;
+                        false ->
+                            [iolist_to_binary(io_lib:format("optional parameter: ~p, type must be: ~ts", [Key, type_name(Type)])) | ErrAcc]
+                    end
+            end
+        end,
+        [], Fields).
+
+-spec type_name(tuple() | atom()) -> binary().
+type_name(binary) ->
+    <<"string">>;
+type_name(integer) ->
+    <<"integer">>;
+type_name(non_neg_integer) ->
+    <<"non-negative integer">>;
+type_name(number) ->
+    <<"number">>;
+type_name(non_neg_number) ->
+    <<"non-negative number">>;
+type_name(duration) ->
+    <<"duration string or non-negative integer">>;
+type_name(list) ->
+    <<"list">>;
+type_name({list, binary}) ->
+    <<"list of string">>;
+type_name({list, number}) ->
+    <<"list of number">>;
+type_name({list, integer}) ->
+    <<"list of integer">>;
+type_name(map) ->
+    <<"map">>;
+type_name({map, {binary, binary}}) ->
+    <<"map of string:string">>;
+type_name({map, {binary, any}}) ->
+    <<"map of string:any">>;
+type_name(boolean) ->
+    <<"boolean">>.
+
+-spec check_type(any(), any()) -> boolean().
+check_type(Value, binary) ->
+    is_binary(Value);
+check_type(Value, integer) ->
+    is_integer(Value);
+check_type(Value, non_neg_integer) ->
+    is_integer(Value) andalso Value >= 0;
+check_type(Value, number) ->
+    is_number(Value);
+check_type(Value, non_neg_number) ->
+    is_number(Value) andalso Value >= 0;
+check_type(Value, duration) ->
+    is_binary(Value) orelse (is_integer(Value) andalso Value >= 0);
+check_type(Value, list) when is_list(Value) ->
+    true;
+check_type(Value, {list, binary}) when is_list(Value) ->
+    lists:all(fun(E) -> is_binary(E) end, Value);
+check_type(Value, {list, number}) when is_list(Value) ->
+    lists:all(fun(E) -> is_number(E) end, Value);
+check_type(Value, {list, integer}) when is_list(Value) ->
+    lists:all(fun(E) -> is_integer(E) end, Value);
+check_type(Value, map) when is_map(Value) ->
+    true;
+check_type(Value, {map, {binary, binary}}) when is_map(Value) ->
+    lists:all(fun({K, V}) -> is_binary(K) andalso is_binary(V) end, maps:to_list(Value));
+check_type(Value, {map, {binary, any}}) when is_map(Value) ->
+    lists:all(fun({K, _}) -> is_binary(K) end, maps:to_list(Value));
+check_type(Value, boolean) ->
+    is_boolean(Value);
+check_type(_, _) ->
+    false.
+
+-spec check_healthcheck(map()) -> [binary()].
+check_healthcheck(Config) ->
+    case maps:get(<<"healthcheck">>, Config, undefined) of
+        undefined ->
+            [];
+        Healthcheck when is_map(Healthcheck) ->
+            Fields = [
+                {<<"test">>, {list, binary}},
+                {<<"interval">>, duration},
+                {<<"timeout">>, duration},
+                {<<"retries">>, non_neg_integer}
+            ],
+            KeyErrors = [<<"optional parameter: <<\"healthcheck\">>, type must be: map of string:any">> ||
+                {Key, _Value} <- maps:to_list(Healthcheck), not is_binary(Key)],
+            KeyErrors ++ check_optional(Healthcheck, Fields);
+        _ ->
+            []
+    end.
+
+-spec build_container_deploy_params(map()) -> map().
+build_container_deploy_params(Config) when is_map(Config) ->
+    #{
+        container_name => maps:get(<<"container_name">>, Config),
+        create => build_docker_create_options(Config)
+    }.
+
+-spec build_docker_create_options(map()) -> map().
+build_docker_create_options(Config) when is_map(Config) ->
+    #{
+        config => build_docker_container_config(Config),
+        host_config => build_docker_host_config(Config),
+        networking_config => build_docker_networking_config(Config)
+    }.
+
+-spec build_docker_container_config(map()) -> map().
+build_docker_container_config(Config) when is_map(Config) ->
+    #{
+        image => maps:get(<<"image">>, Config),
+        cmd => maps:get(<<"command">>, Config),
+        entrypoint => maps:get(<<"entrypoint">>, Config, []),
+        env => maps:get(<<"envs">>, Config, []),
+        labels => maps:get(<<"labels">>, Config, #{}),
+        volumes => build_container_volumes(maps:get(<<"volumes">>, Config, [])),
+        user => maps:get(<<"user">>, Config, <<>>),
+        working_dir => maps:get(<<"working_dir">>, Config, <<>>),
+        hostname => maps:get(<<"hostname">>, Config, <<>>),
+        exposed_ports => build_exposed_ports(Config),
+        healthcheck => build_healthcheck(maps:get(<<"healthcheck">>, Config, undefined))
+    }.
+
+-spec build_docker_host_config(map()) -> map().
+build_docker_host_config(Config) when is_map(Config) ->
+    #{
+        binds => build_host_binds(maps:get(<<"volumes">>, Config, [])),
+        network_mode => maps:get(<<"network_mode">>, Config, <<>>),
+        restart_policy => build_restart_policy(maps:get(<<"restart">>, Config)),
+        privileged => maps:get(<<"privileged">>, Config, false),
+        cap_add => maps:get(<<"cap_add">>, Config, []),
+        cap_drop => maps:get(<<"cap_drop">>, Config, []),
+        devices => build_device_mappings(maps:get(<<"devices">>, Config, [])),
+        memory => default_uint64(parse_optional_size_bytes(maps:get(<<"mem_limit">>, Config, undefined), <<"mem_limit">>)),
+        memory_reservation => default_uint64(parse_optional_size_bytes(maps:get(<<"mem_reservation">>, Config, undefined), <<"mem_reservation">>)),
+        nano_cpus => default_uint64(parse_optional_nano_cpus(maps:get(<<"cpus">>, Config, undefined))),
+        cpu_shares => default_uint64(maps:get(<<"cpu_shares">>, Config, undefined)),
+        port_bindings => build_port_bindings(maps:get(<<"ports">>, Config, [])),
+        ulimits => build_ulimits(maps:get(<<"ulimits">>, Config, #{})),
+        tmpfs => maps:from_list(build_tmpfs_options(maps:get(<<"tmpfs">>, Config, []))),
+        sysctls => maps:get(<<"sysctls">>, Config, #{}),
+        extra_hosts => maps:get(<<"extra_hosts">>, Config, [])
+    }.
+
+-spec build_docker_networking_config(map()) -> map().
+build_docker_networking_config(Config) when is_map(Config) ->
+    Networks = maps:get(<<"networks">>, Config, []),
+    #{endpoints => [#{name => Network} || Network <- Networks]}.
+
+-spec build_restart_policy(binary()) -> map().
+build_restart_policy(Restart0) when is_binary(Restart0) ->
+    case binary:split(Restart0, <<":">>) of
+        [Name, RetryCountBin] ->
+            #{name => Name, maximum_retry_count => parse_uint32(RetryCountBin, <<"restart">>)};
+        [Name] ->
+            #{name => Name, maximum_retry_count => 0}
+    end.
+
+-spec build_healthcheck(undefined | map()) -> undefined | map().
+build_healthcheck(undefined) ->
+    undefined;
+build_healthcheck(Healthcheck) when is_map(Healthcheck) ->
+    #{
+        test => maps:get(<<"test">>, Healthcheck, []),
+        interval_ns => parse_duration_ns(maps:get(<<"interval">>, Healthcheck, <<"0s">>), <<"healthcheck.interval">>),
+        timeout_ns => parse_duration_ns(maps:get(<<"timeout">>, Healthcheck, <<"0s">>), <<"healthcheck.timeout">>),
+        retries => maps:get(<<"retries">>, Healthcheck, 0)
+    }.
+
+-spec default_uint64(undefined | non_neg_integer()) -> non_neg_integer().
+default_uint64(undefined) ->
+    0;
+default_uint64(Value) when is_integer(Value), Value >= 0 ->
+    Value.
+
+-spec parse_optional_nano_cpus(undefined | number()) -> undefined | non_neg_integer().
+parse_optional_nano_cpus(undefined) ->
+    undefined;
+parse_optional_nano_cpus(Cpus) when is_integer(Cpus), Cpus >= 0 ->
+    Cpus * 1000000000;
+parse_optional_nano_cpus(Cpus) when is_float(Cpus), Cpus >= 0 ->
+    trunc(Cpus * 1000000000).
+
+-spec build_container_volumes([binary()]) -> [binary()].
+build_container_volumes(VolumeSpecs) when is_list(VolumeSpecs) ->
+    [ContainerPath || VolumeSpec <- VolumeSpecs, {_HostPath, ContainerPath, _ReadOnly} <- [parse_volume_spec(VolumeSpec)]].
+
+-spec build_host_binds([binary()]) -> [binary()].
+build_host_binds(VolumeSpecs) when is_list(VolumeSpecs) ->
+    [volume_bind(HostPath, ContainerPath, ReadOnly) ||
+        VolumeSpec <- VolumeSpecs,
+        {HostPath, ContainerPath, ReadOnly} <- [parse_volume_spec(VolumeSpec)]].
+
+-spec parse_volume_spec(binary()) -> {binary(), binary(), boolean()}.
+parse_volume_spec(VolumeSpec) when is_binary(VolumeSpec) ->
+    case binary:split(VolumeSpec, <<":">>, [global]) of
+        [HostPath, ContainerPath] when HostPath =/= <<>>, ContainerPath =/= <<>> ->
+            {HostPath, ContainerPath, false};
+        [HostPath, ContainerPath | Modes] when HostPath =/= <<>>, ContainerPath =/= <<>> ->
+            {HostPath, ContainerPath, lists:member(<<"ro">>, Modes)};
+        _ ->
+            throw({error, <<"invalid volume binding">>})
+    end.
+
+-spec volume_bind(binary(), binary(), boolean()) -> binary().
+volume_bind(HostPath, ContainerPath, true) when is_binary(HostPath), is_binary(ContainerPath) ->
+    <<HostPath/binary, ":", ContainerPath/binary, ":ro">>;
+volume_bind(HostPath, ContainerPath, false) when is_binary(HostPath), is_binary(ContainerPath) ->
+    <<HostPath/binary, ":", ContainerPath/binary>>.
+
+-spec build_exposed_ports(map()) -> [map()].
+build_exposed_ports(Config) when is_map(Config) ->
+    ExposePorts = [build_exposed_port(ExposeSpec) || ExposeSpec <- maps:get(<<"expose">>, Config, [])],
+    BoundPorts = [
+        #{container_port => ContainerPort, protocol => Protocol} ||
+        #{container_port := ContainerPort, protocol := Protocol} <- build_port_bindings(maps:get(<<"ports">>, Config, []))
+    ],
+    unique_ports(ExposePorts ++ BoundPorts).
+
+-spec build_exposed_port(binary()) -> map().
+build_exposed_port(ExposeSpec) when is_binary(ExposeSpec) ->
+    case binary:split(ExposeSpec, <<"/">>) of
+        [PortBin] ->
+            #{container_port => parse_uint32(PortBin, <<"expose">>), protocol => <<"tcp">>};
+        [PortBin, Protocol] ->
+            #{container_port => parse_uint32(PortBin, <<"expose">>), protocol => Protocol}
+    end.
+
+-spec build_port_bindings([binary()]) -> [map()].
+build_port_bindings(PortSpecs) when is_list(PortSpecs) ->
+    [build_port_binding(PortSpec) || PortSpec <- PortSpecs].
+
+-spec build_port_binding(binary()) -> map().
+build_port_binding(PortSpec) when is_binary(PortSpec) ->
+    case binary:split(PortSpec, <<":">>, [global]) of
+        [HostPortBin, ContainerPortSpec] when HostPortBin =/= <<>>, ContainerPortSpec =/= <<>> ->
+            {ContainerPort, Protocol} = parse_container_port_spec(ContainerPortSpec, <<"ports">>),
+            #{
+                host_ip => <<>>,
+                host_port => parse_tcp_port(HostPortBin, <<"ports.host_port">>),
+                container_port => ContainerPort,
+                protocol => Protocol
+            };
+        _ ->
+            throw({error, <<"invalid port binding">>})
+    end.
+
+-spec parse_container_port_spec(binary(), binary()) -> {non_neg_integer(), binary()}.
+parse_container_port_spec(PortSpec, Field) when is_binary(PortSpec), is_binary(Field) ->
+    case binary:split(PortSpec, <<"/">>) of
+        [PortBin] ->
+            {parse_tcp_port(PortBin, Field), <<"tcp">>};
+        [PortBin, Protocol] when Protocol =/= <<>> ->
+            {parse_tcp_port(PortBin, Field), Protocol};
+        _ ->
+            throw({error, <<"invalid port binding">>})
+    end.
+
+-spec unique_ports([map()]) -> [map()].
+unique_ports(Ports) when is_list(Ports) ->
+    {_Seen, Result} = lists:foldl(
+        fun(Port = #{container_port := ContainerPort, protocol := Protocol}, {Seen, Acc}) ->
+            Key = {ContainerPort, Protocol},
+            case maps:is_key(Key, Seen) of
+                true ->
+                    {Seen, Acc};
+                false ->
+                    {Seen#{Key => true}, [Port | Acc]}
+            end
+        end,
+        {#{}, []},
+        Ports),
+    lists:reverse(Result).
+
+-spec build_device_mappings([binary()]) -> [map()].
+build_device_mappings(DeviceSpecs) when is_list(DeviceSpecs) ->
+    [build_device_mapping(DeviceSpec) || DeviceSpec <- DeviceSpecs].
+
+-spec build_device_mapping(binary()) -> map().
+build_device_mapping(DeviceSpec) when is_binary(DeviceSpec) ->
+    case binary:split(DeviceSpec, <<":">>, [global]) of
+        [HostPath, ContainerPath] when HostPath =/= <<>>, ContainerPath =/= <<>> ->
+            #{path_on_host => HostPath, path_in_container => ContainerPath, cgroup_permissions => <<"rwm">>};
+        [HostPath, ContainerPath, Permissions] when HostPath =/= <<>>, ContainerPath =/= <<>>, Permissions =/= <<>> ->
+            #{path_on_host => HostPath, path_in_container => ContainerPath, cgroup_permissions => Permissions};
+        _ ->
+            throw({error, <<"invalid device mapping">>})
+    end.
+
+-spec build_ulimits(map()) -> [map()].
+build_ulimits(Ulimits) when is_map(Ulimits) ->
+    [build_ulimit(Name, Value) || {Name, Value} <- maps:to_list(Ulimits)].
+
+-spec build_ulimit(binary(), binary()) -> map().
+build_ulimit(Name, Value) when is_binary(Name), is_binary(Value) ->
+    case binary:split(Value, <<":">>) of
+        [SoftBin, HardBin] ->
+            #{name => Name, soft => parse_uint64(SoftBin, <<"ulimits.soft">>), hard => parse_uint64(HardBin, <<"ulimits.hard">>)};
+        [LimitBin] ->
+            Limit = parse_uint64(LimitBin, <<"ulimits.limit">>),
+            #{name => Name, soft => Limit, hard => Limit}
+    end.
+
+-spec build_tmpfs_options([binary()]) -> [{binary(), binary()}].
+build_tmpfs_options(TmpfsSpecs) when is_list(TmpfsSpecs) ->
+    [build_tmpfs_option(TmpfsSpec) || TmpfsSpec <- TmpfsSpecs].
+
+-spec build_tmpfs_option(binary()) -> {binary(), binary()}.
+build_tmpfs_option(TmpfsSpec) when is_binary(TmpfsSpec) ->
+    case binary:split(TmpfsSpec, <<":">>) of
+        [Path] when Path =/= <<>> ->
+            {Path, <<>>};
+        [Path, Options] when Path =/= <<>> ->
+            {Path, Options};
+        _ ->
+            throw({error, <<"invalid tmpfs mount">>})
+    end.
+
+-spec parse_optional_size_bytes(undefined | binary(), binary()) -> undefined | non_neg_integer().
+parse_optional_size_bytes(undefined, _Field) ->
+    undefined;
+parse_optional_size_bytes(Value, Field) when is_binary(Value) ->
+    parse_size_bytes(Value, Field).
+
+-spec parse_duration_ns(binary() | integer(), binary()) -> non_neg_integer().
+parse_duration_ns(Value, _Field) when is_integer(Value), Value >= 0 ->
+    Value;
+parse_duration_ns(Value, Field) when is_binary(Value) ->
+    parse_scaled_uint64(Value, Field, #{
+        <<"ns">> => 1,
+        <<"us">> => 1000,
+        <<"ms">> => 1000000,
+        <<"s">> => 1000000000,
+        <<"m">> => 60000000000,
+        <<"h">> => 3600000000000,
+        <<>> => 1000000000
+    }).
+
+-spec parse_size_bytes(binary(), binary()) -> non_neg_integer().
+parse_size_bytes(Value, Field) when is_binary(Value) ->
+    parse_scaled_uint64(Value, Field, #{
+        <<"b">> => 1,
+        <<"k">> => 1024,
+        <<"kb">> => 1024,
+        <<"ki">> => 1024,
+        <<"kib">> => 1024,
+        <<"m">> => 1048576,
+        <<"mb">> => 1048576,
+        <<"mi">> => 1048576,
+        <<"mib">> => 1048576,
+        <<"g">> => 1073741824,
+        <<"gb">> => 1073741824,
+        <<"gi">> => 1073741824,
+        <<"gib">> => 1073741824,
+        <<"t">> => 1099511627776,
+        <<"tb">> => 1099511627776,
+        <<"ti">> => 1099511627776,
+        <<"tib">> => 1099511627776,
+        <<>> => 1
+    }).
+
+-spec parse_scaled_uint64(binary(), binary(), map()) -> non_neg_integer().
+parse_scaled_uint64(Value0, Field, Multipliers) when is_binary(Value0), is_binary(Field), is_map(Multipliers) ->
+    Value = trim_binary(Value0),
+    LowerValue = lower_binary(Value),
+    {NumberBin, Unit} = split_numeric_suffix(LowerValue),
+    case maps:get(Unit, Multipliers, undefined) of
+        undefined ->
+            throw({error, <<"invalid value for ", Field/binary, ": ", Value0/binary>>});
+        Multiplier ->
+            trunc(parse_decimal(NumberBin, Field) * Multiplier)
+    end.
+
+-spec parse_uint32(binary(), binary()) -> non_neg_integer().
+parse_uint32(Value, Field) when is_binary(Value), is_binary(Field) ->
+    Parsed = parse_uint64(Value, Field),
+    case Parsed =< 16#FFFFFFFF of
+        true ->
+            Parsed;
+        false ->
+            throw({error, <<"value overflow for ", Field/binary>>})
+    end.
+
+-spec parse_tcp_port(binary(), binary()) -> non_neg_integer().
+parse_tcp_port(Value, Field) when is_binary(Value), is_binary(Field) ->
+    Parsed = parse_uint32(Value, Field),
+    case Parsed =< 65535 of
+        true ->
+            Parsed;
+        false ->
+            throw({error, <<"port out of range for ", Field/binary>>})
+    end.
+
+-spec parse_uint64(binary(), binary()) -> non_neg_integer().
+parse_uint64(Value0, Field) when is_binary(Value0), is_binary(Field) ->
+    Value = trim_binary(Value0),
+    case catch binary_to_integer(Value) of
+        Parsed when is_integer(Parsed), Parsed >= 0 ->
+            Parsed;
+        _ ->
+            throw({error, <<"invalid unsigned integer for ", Field/binary, ": ", Value0/binary>>})
+    end.
+
+-spec parse_decimal(binary(), binary()) -> float().
+parse_decimal(Value, Field) when is_binary(Value), is_binary(Field) ->
+    case catch binary_to_integer(Value) of
+        ParsedInt when is_integer(ParsedInt), ParsedInt >= 0 ->
+            float(ParsedInt);
+        _ ->
+            case catch binary_to_float(Value) of
+                ParsedFloat when is_float(ParsedFloat), ParsedFloat >= 0 ->
+                    ParsedFloat;
+                _ ->
+                    throw({error, <<"invalid number for ", Field/binary, ": ", Value/binary>>})
+            end
+    end.
+
+-spec split_numeric_suffix(binary()) -> {binary(), binary()}.
+split_numeric_suffix(Value) when is_binary(Value) ->
+    split_numeric_suffix(Value, <<>>).
+
+-spec split_numeric_suffix(binary(), binary()) -> {binary(), binary()}.
+split_numeric_suffix(<<Char, Rest/binary>>, Acc)
+  when (Char >= $0 andalso Char =< $9) orelse Char =:= $. ->
+    split_numeric_suffix(Rest, <<Acc/binary, Char>>);
+split_numeric_suffix(Rest, <<>>) ->
+    throw({error, <<"invalid numeric value: ", Rest/binary>>});
+split_numeric_suffix(Rest, Acc) ->
+    {Acc, Rest}.
+
+-spec trim_binary(binary()) -> binary().
+trim_binary(Value) when is_binary(Value) ->
+    Trimmed = string:trim(binary_to_list(Value)),
+    list_to_binary(Trimmed).
+
+-spec lower_binary(binary()) -> binary().
+lower_binary(Value) when is_binary(Value) ->
+    list_to_binary(string:lowercase(binary_to_list(Value))).

apps/iot/src/endpoint/endpoint_buffer.erl

@@ -1,105 +0,0 @@
-
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 06. 7月 2023 12:02
-%%%-------------------------------------------------------------------
--module(endpoint_buffer).
-
--include("endpoint.hrl").
-
-%% 消息重发间隔
--define(RETRY_INTERVAL, 5000).
-
--export([new/2, append/2, trigger_next/1, trigger_n/1, cleanup/1, ack/2, stat/1]).
--export_type([buffer/0]).
-
--record(buffer, {
-    endpoint :: #endpoint{},
-    next_id = 1 :: integer(),
-    %% 当前数据所在的游标
-    cursor = 0 :: integer(),
-    %% ets存储的引用
-    tid :: ets:tid(),
-    %% 定时器
-    timer_pid :: pid(),
-    %% 窗口大小，允许最大的未确认消息数
-    window_size = 10,
-    %% 未确认的消息数
-    flight_num = 0,
-    %% 记录成功处理的消息数
-    acc_num = 0
-}).
-
--record(north_data, {
-    id :: integer(),
-    tuple :: any()
-}).
-
--type buffer() :: #buffer{}.
-
--spec new(Endpoint :: #endpoint{}, WindowSize :: integer()) -> Buffer :: #buffer{}.
-new(Endpoint = #endpoint{id = Id}, WindowSize) when is_integer(WindowSize), WindowSize > 0 ->
-    %% 初始化存储
-    EtsName = list_to_atom("endpoint_buffer_ets:" ++ integer_to_list(Id)),
-    Tid = ets:new(EtsName, [ordered_set, private, {keypos, 2}]),
-    %% 定义重发器
-    {ok, TimerPid} = endpoint_timer:start_link(?RETRY_INTERVAL),
-
-    #buffer{cursor = 0, tid = Tid, timer_pid = TimerPid, endpoint = Endpoint, window_size = WindowSize}.
-
--spec append(tuple(), Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
-append(Tuple, Buffer = #buffer{tid = Tid, next_id = NextId, window_size = WindowSize, flight_num = FlightNum}) ->
-    NorthData = #north_data{id = NextId, tuple = Tuple},
-    true = ets:insert(Tid, NorthData),
-    NBuffer = Buffer#buffer{next_id = NextId + 1},
-    case FlightNum < WindowSize of
-        true ->
-            trigger_next(NBuffer);
-        false ->
-            NBuffer
-    end.
-
--spec trigger_n(Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
-trigger_n(Buffer = #buffer{window_size = WindowSize}) ->
-    %% 最多允许window_size
-    lists:foldl(fun(_, Buffer0) -> trigger_next(Buffer0) end, Buffer, lists:seq(1, WindowSize)).
-
-%% 触发读取下一条数据
--spec trigger_next(Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
-trigger_next(Buffer = #buffer{tid = Tid, cursor = Cursor, timer_pid = TimerPid, flight_num = FlightNum}) ->
-    case ets:first(Tid) of
-        '$end_of_table' ->
-            Buffer;
-        Key ->
-            [#north_data{id = Id, tuple = Tuple}|_] = ets:take(Tid, Key),
-            ReceiverPid = self(),
-            ReceiverPid ! {next_data, Id, Tuple},
-            endpoint_timer:task(TimerPid, Id, fun() -> ReceiverPid ! {next_data, Id, Tuple} end),
-            Buffer#buffer{flight_num = FlightNum + 1, cursor = Cursor + 1}
-    end.
-
--spec ack(Id :: integer(), Buffer :: #buffer{}) -> NBuffer :: #buffer{}.
-ack(Id, Buffer = #buffer{timer_pid = TimerPid, acc_num = AccNum, flight_num = FlightNum}) when is_integer(Id) ->
-    endpoint_timer:ack(TimerPid, Id),
-    trigger_next(Buffer#buffer{acc_num = AccNum + 1, flight_num = FlightNum - 1}).
-
-%% 获取当前统计信息
--spec stat(Buffer :: #buffer{}) -> map().
-stat(#buffer{acc_num = AccNum, tid = Tid}) ->
-    #{
-        <<"acc_num">> => AccNum,
-        <<"queue_num">> => ets:info(Tid, size)
-    }.
-
--spec cleanup(Buffer :: #buffer{}) -> ok.
-cleanup(#buffer{timer_pid = TimerPid}) ->
-    endpoint_timer:cleanup(TimerPid),
-    ok.
-
-%%%===================================================================
-%%% Internal functions
-%%%===================================================================

apps/iot/src/endpoint/endpoint_kafka.erl

@@ -1,184 +0,0 @@
-
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 06. 7月 2023 12:02
-%%%-------------------------------------------------------------------
--module(endpoint_kafka).
-
--include("endpoint.hrl").
--behaviour(gen_server).
-
-%% API
--export([start_link/2]).
-
-%% gen_server callbacks
--export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
-
-%% 消息重发间隔
--define(RETRY_INTERVAL, 5000).
-
--define(DISCONNECTED, disconnected).
--define(CONNECTED, connected).
-
--record(state, {
-    endpoint :: #endpoint{},
-    buffer :: endpoint_buffer:buffer(),
-    client_id :: atom(),
-    client_pid :: undefined | pid(),
-    status = ?DISCONNECTED
-}).
-
-%%%===================================================================
-%%% API
-%%%===================================================================
-
-%% @doc Creates a gen_statem process which calls Module:init/1 to
-%% initialize. To ensure a synchronized start-up procedure, this
-%% function does not return until Module:init/1 has returned.
-start_link(LocalName, Endpoint = #endpoint{}) when is_atom(LocalName) ->
-    gen_server:start_link({local, LocalName}, ?MODULE, [Endpoint], []).
-
-%%%===================================================================
-%%% gen_statem callbacks
-%%%===================================================================
-
-%% @private
-%% @doc Whenever a gen_statem is started using gen_statem:start/[3,4] or
-%% gen_statem:start_link/[3,4], this function is called by the new
-%% process to initialize.
-init([Endpoint = #endpoint{id = Id, matcher = Matcher}]) ->
-    endpoint_subscription:subscribe(Matcher, self()),
-
-    erlang:process_flag(trap_exit, true),
-    %% 创建转发器, 避免阻塞当前进程的创建，因此采用了延时初始化的机制
-    erlang:start_timer(0, self(), connect),
-    %% 初始化存储
-    Buffer = endpoint_buffer:new(Endpoint, 10),
-    ClientId = list_to_atom("brod_client:" ++ integer_to_list(Id)),
-
-    {ok, #state{endpoint = Endpoint, buffer = Buffer, status = ?DISCONNECTED, client_id = ClientId}}.
-
-%% @private
-%% @doc Handling call messages
--spec(handle_call(Request :: term(), From :: {pid(), Tag :: term()},
-    State :: #state{}) ->
-    {reply, Reply :: term(), NewState :: #state{}} |
-    {reply, Reply :: term(), NewState :: #state{}, timeout() | hibernate} |
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), Reply :: term(), NewState :: #state{}} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_call(get_stat, _From, State = #state{buffer = Buffer}) ->
-    Stat = endpoint_buffer:stat(Buffer),
-    {reply, {ok, Stat}, State}.
-
-%% @private
-%% @doc Handling cast messages
--spec(handle_cast(Request :: term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_cast({forward, Metric}, State = #state{buffer = Buffer}) ->
-    NBuffer = endpoint_buffer:append(Metric, Buffer),
-    {noreply, State#state{buffer = NBuffer}}.
-
-%% @private
-%% @doc Handling all non call/cast messages
--spec(handle_info(Info :: timeout() | term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_info({timeout, _, connect}, State = #state{buffer = Buffer, status = ?DISCONNECTED, client_id = ClientId,
-    endpoint = #endpoint{title = Title, config = #kafka_endpoint{sasl_config = SaslConfig, bootstrap_servers = BootstrapServers, topic = Topic}}}) ->
-    lager:debug("[endpoint_kafka] endpoint: ~p, create postman", [Title]),
-
-    BaseConfig = [
-        {reconnect_cool_down_seconds, 5},
-        {socket_options, [{keepalive, true}]}
-    ],
-
-    ClientConfig = case SaslConfig of
-                     {Mechanism, Username, Password} ->
-                         [{sasl, {Mechanism, Username, Password}}|BaseConfig];
-                     undefined ->
-                         BaseConfig
-                 end,
-
-    case catch brod:start_link_client(BootstrapServers, ClientId, ClientConfig) of
-        {ok, ClientPid} ->
-            case brod:start_producer(ClientId, Topic, _ProducerConfig = []) of
-                ok ->
-                    NBuffer = endpoint_buffer:trigger_next(Buffer),
-                    {noreply, State#state{buffer = NBuffer, client_pid = ClientPid, status = ?CONNECTED}};
-                {error, Reason} ->
-                    lager:debug("[endpoint_kafka] start_producer: ~p, get error: ~p", [ClientId, Reason]),
-                    brod:stop_client(ClientId),
-                    retry_connect(),
-                    {noreply, State#state{status = ?DISCONNECTED, client_pid = undefined}}
-            end;
-        Error ->
-            lager:debug("[endpoint_kafka] start_client: ~p, get error: ~p", [ClientId, Error]),
-            retry_connect(),
-            {noreply, State#state{status = ?DISCONNECTED, client_pid = undefined}}
-    end;
-
-%% 离线时，忽略数据发送逻辑
-handle_info({next_data, _Id, _Tuple}, State = #state{status = ?DISCONNECTED}) ->
-    {noreply, State};
-%% 发送数据到mqtt服务器
-handle_info({next_data, Id, Metric}, State = #state{status = ?CONNECTED, client_pid = ClientPid,
-    endpoint = #endpoint{config = #kafka_endpoint{topic = Topic}}}) ->
-
-    ReceiverPid = self(),
-    AckCb = fun(Partition, BaseOffset) ->
-        lager:debug("[endpoint_kafka] ack partion: ~p, offset: ~p", [Partition, BaseOffset]),
-        ReceiverPid ! {ack, Id}
-            end,
-    _ = brod:produce_cb(ClientPid, Topic, random, <<>>, Metric, AckCb),
-
-    {noreply, State};
-
-handle_info({ack, Id}, State = #state{buffer = Buffer}) ->
-    NBuffer = endpoint_buffer:ack(Id, Buffer),
-    {noreply, State#state{buffer = NBuffer}};
-
-%% postman进程挂掉时，重新建立新的
-handle_info({'EXIT', ClientPid, Reason}, State = #state{client_pid = ClientPid, endpoint = #endpoint{title = Title}}) ->
-    lager:warning("[endpoint_kafka] endpoint: ~p, conn pid exit with reason: ~p", [Title, Reason]),
-    retry_connect(),
-    {noreply, State#state{client_pid = undefined, status = ?DISCONNECTED}};
-
-handle_info(Info, State = #state{status = Status}) ->
-    lager:warning("[endpoint_kafka] unknown message: ~p, status: ~p", [Info, Status]),
-    {noreply, State}.
-
-%% @private
-%% @doc This function is called by a gen_server when it is about to
-%% terminate. It should be the opposite of Module:init/1 and do any
-%% necessary cleaning up. When it returns, the gen_server terminates
-%% with Reason. The return value is ignored.
--spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
-    State :: #state{}) -> term()).
-terminate(Reason, #state{endpoint = #endpoint{title = Title}, buffer = Buffer}) ->
-    lager:debug("[endpoint_kafka] endpoint: ~p, terminate with reason: ~p", [Title, Reason]),
-    endpoint_buffer:cleanup(Buffer),
-    ok.
-
-%% @private
-%% @doc Convert process state when code is changed
--spec(code_change(OldVsn :: term() | {down, term()}, State :: #state{},
-    Extra :: term()) ->
-    {ok, NewState :: #state{}} | {error, Reason :: term()}).
-code_change(_OldVsn, State = #state{}, _Extra) ->
-    {ok, State}.
-
-%%%===================================================================
-%%% Internal functions
-%%%===================================================================
-
-retry_connect() ->
-    erlang:start_timer(?RETRY_INTERVAL, self(), connect).

apps/iot/src/endpoint/endpoint_mqtt.erl

@@ -1,203 +0,0 @@
-
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 06. 7月 2023 12:02
-%%%-------------------------------------------------------------------
--module(endpoint_mqtt).
-
--include("endpoint.hrl").
--behaviour(gen_server).
-
-%% API
--export([start_link/2]).
-
-%% gen_server callbacks
--export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
-
-%% 消息重发间隔
--define(RETRY_INTERVAL, 15000).
-
--define(DISCONNECTED, disconnected).
--define(CONNECTED, connected).
-
--record(state, {
-    endpoint :: #endpoint{},
-    buffer :: endpoint_buffer:buffer(),
-    conn_pid :: undefined | pid(),
-    %% 待确认的数据, #{PacketId :: integer() => Id :: integer()}
-    inflight = #{},
-
-    status = disconnected
-}).
-
-%%%===================================================================
-%%% API
-%%%===================================================================
-
-%% @doc Creates a gen_statem process which calls Module:init/1 to
-%% initialize. To ensure a synchronized start-up procedure, this
-%% function does not return until Module:init/1 has returned.
-start_link(LocalName, Endpoint = #endpoint{}) when is_atom(LocalName) ->
-    gen_server:start_link({local, LocalName}, ?MODULE, [Endpoint], []).
-
-%%%===================================================================
-%%% gen_statem callbacks
-%%%===================================================================
-
-%% @private
-%% @doc Whenever a gen_statem is started using gen_statem:start/[3,4] or
-%% gen_statem:start_link/[3,4], this function is called by the new
-%% process to initialize.
-init([Endpoint = #endpoint{matcher = Matcher}]) ->
-    % erlang:process_flag(trap_exit, true),
-    endpoint_subscription:subscribe(Matcher, self()),
-
-    %% 创建转发器, 避免阻塞当前进程的创建，因此采用了延时初始化的机制
-    erlang:start_timer(0, self(), create_postman),
-    %% 初始化存储
-    Buffer = endpoint_buffer:new(Endpoint, 10),
-
-    {ok, #state{endpoint = Endpoint, buffer = Buffer, status = ?DISCONNECTED}}.
-
-%% @private
-%% @doc Handling call messages
--spec(handle_call(Request :: term(), From :: {pid(), Tag :: term()},
-    State :: #state{}) ->
-    {reply, Reply :: term(), NewState :: #state{}} |
-    {reply, Reply :: term(), NewState :: #state{}, timeout() | hibernate} |
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), Reply :: term(), NewState :: #state{}} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_call(get_stat, _From, State = #state{buffer = Buffer}) ->
-    Stat = endpoint_buffer:stat(Buffer),
-    {reply, {ok, Stat}, State}.
-
-%% @private
-%% @doc Handling cast messages
--spec(handle_cast(Request :: term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_cast({forward, Metric}, State = #state{buffer = Buffer}) ->
-    NBuffer = endpoint_buffer:append(Metric, Buffer),
-    {noreply, State#state{buffer = NBuffer}}.
-
-%% @private
-%% @doc Handling all non call/cast messages
--spec(handle_info(Info :: timeout() | term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_info({timeout, _, create_postman}, State = #state{buffer = Buffer, status = ?DISCONNECTED,
-    endpoint = #endpoint{title = Title, config = #mqtt_endpoint{host = Host, port = Port, username = Username, password = Password, client_id = ClientId}}}) ->
-    lager:debug("[endpoint_mqtt] endpoint: ~ts, create postman", [Title]),
-    Opts = [
-        {owner, self()},
-        {clientid, ClientId},
-        {host, binary_to_list(Host)},
-        {port, Port},
-        {tcp_opts, []},
-        {username, binary_to_list(Username)},
-        {password, binary_to_list(Password)},
-        {keepalive, 86400},
-        {auto_ack, true},
-        {connect_timeout, 5000},
-        {proto_ver, v5},
-        {retry_interval, 5000}
-    ],
-
-    try
-        {ok, ConnPid} = emqtt:start_link(Opts),
-        lager:debug("[endpoint_mqtt] start connect, options: ~p", [Opts]),
-        case emqtt:connect(ConnPid, 5000) of
-            {ok, _} ->
-                lager:debug("[endpoint_mqtt] connect success, pid: ~p", [ConnPid]),
-                NBuffer = endpoint_buffer:trigger_n(Buffer),
-                {noreply, State#state{conn_pid = ConnPid, buffer = NBuffer, status = ?CONNECTED}};
-            {error, Reason} ->
-                lager:warning("[endpoint_mqtt] connect get error: ~p", [Reason]),
-                erlang:start_timer(?RETRY_INTERVAL, self(), create_postman),
-                {noreply, State}
-        end
-    catch _:Error->
-        lager:warning("[endpoint_mqtt] connect get error: ~p", [Error]),
-        erlang:start_timer(?RETRY_INTERVAL, self(), create_postman),
-        {noreply, State}
-    end;
-
-%% 离线时，忽略数据发送逻辑
-handle_info({next_data, _Id, _Tuple}, State = #state{status = ?DISCONNECTED}) ->
-    {keep_state, State};
-%% 发送数据到mqtt服务器
-handle_info({next_data, Id, Metric}, State = #state{status = ?CONNECTED, conn_pid = ConnPid, buffer = Buffer, inflight = InFlight,
-    endpoint = #endpoint{config = #mqtt_endpoint{topic = Topic, qos = Qos}}}) ->
-
-    lager:debug("[endpoint_mqtt] will publish topic: ~p, metric: ~p, qos: ~p", [Topic, Metric, Qos]),
-    case emqtt:publish(ConnPid, Topic, #{}, Metric, [{qos, Qos}, {retain, true}]) of
-        ok ->
-            NBuffer = endpoint_buffer:ack(Id, Buffer),
-            {noreply, State#state{buffer = NBuffer}};
-        {ok, PacketId} ->
-            {noreply, State#state{inflight = maps:put(PacketId, Id, InFlight)}};
-        {error, Reason} ->
-            lager:warning("[endpoint_mqtt] send message to topic: ~p, get error: ~p", [Topic, Reason]),
-            {stop, Reason, State}
-    end;
-
-handle_info({disconnected, ReasonCode, Properties}, State = #state{status = ?CONNECTED}) ->
-    lager:debug("[endpoint_mqtt] Recv a DISONNECT packet - ReasonCode: ~p, Properties: ~p", [ReasonCode, Properties]),
-    erlang:start_timer(?RETRY_INTERVAL, self(), create_postman),
-    {noreply, State#state{conn_pid = undefined, status = ?DISCONNECTED}};
-
-handle_info({publish, Message = #{packet_id := _PacketId, payload := Payload}}, State = #state{status = ?CONNECTED}) ->
-    lager:debug("[endpoint_mqtt] Recv a publish packet: ~p, payload: ~p", [Message, Payload]),
-    {noreply, State};
-
-%% 收到确认的消息
-handle_info({puback, #{packet_id := PacketId}}, State = #state{status = ?CONNECTED, inflight = Inflight, buffer = Buffer}) ->
-    case maps:take(PacketId, Inflight) of
-        {Id, RestInflight} ->
-            NBuffer = endpoint_buffer:ack(Id, Buffer),
-            {noreply, State#state{buffer = NBuffer, inflight = RestInflight}};
-        error ->
-            {noreply, State}
-    end;
-
-%% postman进程挂掉时，重新建立新的
-handle_info({'EXIT', ConnPid, Reason}, State = #state{endpoint = #endpoint{title = Title}, conn_pid = ConnPid}) ->
-    lager:warning("[endpoint_mqtt] endpoint: ~p, conn pid exit with reason: ~p", [Title, Reason]),
-    erlang:start_timer(?RETRY_INTERVAL, self(), create_postman),
-    {noreply, State#state{conn_pid = undefined, status = ?DISCONNECTED}};
-
-handle_info(Info, State = #state{status = Status}) ->
-    lager:warning("[endpoint_mqtt] unknown message: ~p, status: ~p", [Info, Status]),
-    {noreply, State}.
-
-%% @private
-%% @doc This function is called by a gen_server when it is about to
-%% terminate. It should be the opposite of Module:init/1 and do any
-%% necessary cleaning up. When it returns, the gen_server terminates
-%% with Reason. The return value is ignored.
--spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
-    State :: #state{}) -> term()).
-terminate(Reason, #state{endpoint = #endpoint{title = Title}, buffer = Buffer}) ->
-    lager:debug("[endpoint_mqtt] endpoint: ~p, terminate with reason: ~p", [Title, Reason]),
-    endpoint_buffer:cleanup(Buffer),
-    ok.
-
-%% @private
-%% @doc Convert process state when code is changed
--spec(code_change(OldVsn :: term() | {down, term()}, State :: #state{},
-    Extra :: term()) ->
-    {ok, NewState :: #state{}} | {error, Reason :: term()}).
-code_change(_OldVsn, State = #state{}, _Extra) ->
-    {ok, State}.
-
-%%%===================================================================
-%%% Internal functions
-%%%===================================================================

apps/iot/src/endpoint/endpoint_subscription.erl

@@ -1,206 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author anlicheng
-%%% @copyright (C) 2025, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 07. 11月 2025 16:27
-%%%-------------------------------------------------------------------
--module(endpoint_subscription).
--author("anlicheng").
-
--behaviour(gen_server).
-
-%% API
--export([start_link/0]).
--export([subscribe/2, publish/2, get_subscribers/0]).
--export([match_components/2, is_valid_components/1, of_components/1]).
-
-%% gen_server callbacks
--export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
-
--define(SERVER, ?MODULE).
-
-%% 定义订阅者
--record(subscriber, {
-    topic :: binary(),
-    subscriber_pid :: pid(),
-    components = [],
-    %% 优先级
-    %% 1. 完全匹配的topic优先级别最高
-    %% 2. 带 * 的订阅
-    %% 3. 带 + 的订阅
-    order :: integer()
-}).
-
--record(state, {
-    subscribers = []
-}).
-
-%%%===================================================================
-%%% API
-%%%===================================================================
-
--spec subscribe(Topic :: binary(), SubscriberPid :: pid()) -> ok | {error, Reason :: binary()}.
-subscribe(Topic, SubscriberPid) when is_binary(Topic), is_pid(SubscriberPid) ->
-    gen_server:call(?SERVER, {subscribe, Topic, SubscriberPid}).
-
--spec get_subscribers() -> {ok, Subscribers :: map()}.
-get_subscribers() ->
-    gen_server:call(?SERVER, get_subscribers).
-
--spec publish(RouteKey :: binary(), Content :: binary()) -> no_return().
-publish(RouteKey, Content) when is_binary(RouteKey), is_binary(Content) ->
-    gen_server:cast(?SERVER, {publish, RouteKey, Content}).
-
-%% @doc Spawns the server and registers the local name (unique)
--spec(start_link() ->
-    {ok, Pid :: pid()} | ignore | {error, Reason :: term()}).
-start_link() ->
-    gen_server:start_link({local, ?SERVER}, ?MODULE, [], []).
-
-%%%===================================================================
-%%% gen_server callbacks
-%%%===================================================================
-
-%% @private
-%% @doc Initializes the server
--spec(init(Args :: term()) ->
-    {ok, State :: #state{}} | {ok, State :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term()} | ignore).
-init([]) ->
-    {ok, #state{}}.
-
-%% @private
-%% @doc Handling call messages
--spec(handle_call(Request :: term(), From :: {pid(), Tag :: term()},
-    State :: #state{}) ->
-    {reply, Reply :: term(), NewState :: #state{}} |
-    {reply, Reply :: term(), NewState :: #state{}, timeout() | hibernate} |
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), Reply :: term(), NewState :: #state{}} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-%%  同一个SubscriberPid只能订阅同一个topic一次
-handle_call(get_subscribers, _From, State = #state{subscribers = Subscribers}) ->
-    {reply, {ok, Subscribers}, State};
-handle_call({subscribe, Topic, SubscriberPid}, _From, State = #state{subscribers = Subscribers}) ->
-    Components = of_components(Topic),
-    case is_valid_components(Components) of
-        true ->
-            Sub = #subscriber{topic = Topic, subscriber_pid = SubscriberPid, components = Components, order = order_num(Components)},
-            %% 建立到SubscriberPid的monitor，进程退出需要清理订阅
-            erlang:monitor(process, SubscriberPid),
-
-            {reply, ok, State#state{subscribers = Subscribers ++ [Sub]}};
-        false ->
-            {reply, {error, <<"invalid topic name">>}, State}
-    end.
-
-%% @private
-%% @doc Handling cast messages
--spec(handle_cast(Request :: term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-%% 发布消息
-handle_cast({publish, RouteKey, Metric}, State = #state{subscribers = Subscribers}) ->
-    MatchedSubscribers = match_subscribers(Subscribers, RouteKey),
-    lists:foreach(fun(#subscriber{subscriber_pid = SubscriberPid}) ->
-        endpoint:forward(SubscriberPid, Metric)
-                  end, MatchedSubscribers),
-    lager:debug("[efka_subscription] route_key: ~p, metric: ~p, match subscribers: ~p", [RouteKey, Metric, MatchedSubscribers]),
-    {noreply, State}.
-
-%% @private
-%% @doc Handling all non call/cast messages
--spec(handle_info(Info :: timeout() | term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_info({'DOWN', _Ref, process, SubscriberPid, Reason}, State = #state{subscribers = Subscribers}) ->
-    lager:debug("[efka_subscription] subscriber: ~p, down with reason: ~p", [SubscriberPid, Reason]),
-    NSubscribers = lists:filter(fun(#subscriber{subscriber_pid = Pid0}) -> SubscriberPid /= Pid0 end, Subscribers),
-    {noreply, State#state{subscribers = NSubscribers}};
-
-handle_info(Info, State = #state{}) ->
-    lager:debug("[efka_subscription] get unknown info: ~p", [Info]),
-    {noreply, State}.
-
-%% @private
-%% @doc This function is called by a gen_server when it is about to
-%% terminate. It should be the opposite of Module:init/1 and do any
-%% necessary cleaning up. When it returns, the gen_server terminates
-%% with Reason. The return value is ignored.
--spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
-    State :: #state{}) -> term()).
-terminate(_Reason, _State = #state{}) ->
-    ok.
-
-%% @private
-%% @doc Convert process state when code is changed
--spec(code_change(OldVsn :: term() | {down, term()}, State :: #state{},
-    Extra :: term()) ->
-    {ok, NewState :: #state{}} | {error, Reason :: term()}).
-code_change(_OldVsn, State = #state{}, _Extra) ->
-    {ok, State}.
-
-%%%===================================================================
-%%% Internal functions
-%%%===================================================================
-
-%% 查找满足条件订阅者
--spec match_subscribers(Subscribers :: [#subscriber{}], Topic :: binary()) -> [#subscriber{}].
-match_subscribers(Subscribers, Topic) when is_list(Subscribers), is_binary(Topic) ->
-    Components = of_components(Topic),
-    lists:foldl(fun(S = #subscriber{components = Components0, subscriber_pid = Pid0}, Acc) ->
-        case match_components(Components0, Components) andalso not contain_channel(Pid0, Acc) of
-            true ->
-                [S|Acc];
-            false ->
-                Acc
-        end
-                end, [], Subscribers).
-
--spec contain_channel(Pid :: pid(), Subscribers :: list()) -> boolean().
-contain_channel(Pid, Subscribers) when is_pid(Pid), is_list(Subscribers) ->
-    lists:search(fun(#subscriber{subscriber_pid = Pid0}) -> Pid == Pid0 end, Subscribers) /= false.
-
-%% 开始对比订阅的topic和发布的topic的Components信息
-%% *表示单级匹配，+表示多级匹配；+只能出现一次，并且只能在末尾
--spec match_components(list(), list()) -> boolean().
-match_components(A, B) when is_list(A), is_list(B) ->
-    match_components(A, B, false).
-match_components([<<"+">>], [_|_], _) ->
-    true;
-match_components([], [], _) ->
-    true;
-match_components([<<"*">>|T0], [_|T1], _) ->
-    match_components(T0, T1, false);
-match_components([C0|T0], [C0|T1], _) ->
-    match_components(T0, T1, false);
-match_components(_, _, _) ->
-    false.
-
--spec of_components(Topic :: binary()) -> [binary()].
-of_components(Topic) when is_binary(Topic) ->
-    binary:split(Topic, <<$/>>, [global]).
-
-is_valid_components([]) ->
-    true;
-is_valid_components([<<$+>>|T]) ->
-    length(T) =:= 0;
-is_valid_components([<<$*>>|T]) ->
-    is_valid_components(T);
-is_valid_components([_|T]) ->
-    is_valid_components(T).
-
--spec order_num(Components :: list()) -> integer().
-order_num([]) ->
-    1;
-order_num([<<$*>>|_]) ->
-    2;
-order_num([<<$+>>|_]) ->
-    3;
-order_num([_|Tail]) ->
-    order_num(Tail).

apps/iot/src/endpoint/endpoint_timer.erl

@@ -1,128 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author anlicheng
-%%% @copyright (C) 2024, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 07. 5月 2024 10:30
-%%%-------------------------------------------------------------------
--module(endpoint_timer).
--author("anlicheng").
--include("endpoint.hrl").
-
--behaviour(gen_server).
-
-%% API
--export([start_link/1]).
--export([task/3, ack/2, cleanup/1]).
-
-%% gen_server callbacks
--export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
-
--record(state, {
-    retry_interval = 0,
-    %% 定时器
-    timer_map = #{}
-}).
-
-%%%===================================================================
-%%% API
-%%%===================================================================
-
-task(Pid, Id, Task) when is_pid(Pid), is_integer(Id), is_function(Task, 0) ->
-    gen_server:cast(Pid, {task, Id, Task}).
-
-ack(Pid, Id) when is_pid(Pid), is_integer(Id) ->
-    gen_server:cast(Pid, {ack, Id}).
-
-cleanup(Pid) when is_pid(Pid) ->
-    gen_server:cast(Pid, cleanup).
-
-%% @doc Spawns the server and registers the local name (unique)
--spec(start_link(RetryInterval :: integer()) ->
-    {ok, Pid :: pid()} | ignore | {error, Reason :: term()}).
-start_link(RetryInterval) when is_integer(RetryInterval) ->
-    gen_server:start_link(?MODULE, [RetryInterval], []).
-
-%%%===================================================================
-%%% gen_server callbacks
-%%%===================================================================
-
-%% @private
-%% @doc Initializes the server
--spec(init(Args :: term()) ->
-    {ok, State :: #state{}} | {ok, State :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term()} | ignore).
-init([RetryInterval]) ->
-    {ok, #state{retry_interval = RetryInterval}}.
-
-%% @private
-%% @doc Handling call messages
--spec(handle_call(Request :: term(), From :: {pid(), Tag :: term()},
-    State :: #state{}) ->
-    {reply, Reply :: term(), NewState :: #state{}} |
-    {reply, Reply :: term(), NewState :: #state{}, timeout() | hibernate} |
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), Reply :: term(), NewState :: #state{}} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_call(_Request, _From, State = #state{}) ->
-    {reply, ok, State}.
-
-%% @private
-%% @doc Handling cast messages
--spec(handle_cast(Request :: term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_cast({task, Id, Task}, State = #state{retry_interval = RetryInterval, timer_map = TimerMap}) ->
-    TimerRef = erlang:start_timer(RetryInterval, self(), {repost_ticker, {Id, Task}}),
-    {noreply, State#state{timer_map = maps:put(Id, TimerRef, TimerMap)}};
-
-%% 取消
-handle_cast({ack, Id}, State = #state{timer_map = TimerMap}) ->
-    case maps:take(Id, TimerMap) of
-        error ->
-            {noreply, State};
-        {TimerRef, NTimerMap} ->
-            is_reference(TimerRef) andalso erlang:cancel_timer(TimerRef),
-            {noreply, State#state{timer_map = NTimerMap}}
-    end;
-
-handle_cast(cleanup, State = #state{timer_map = TimerMap}) ->
-    lists:foreach(fun({_, TimerRef}) -> catch erlang:cancel_timer(TimerRef) end, maps:to_list(TimerMap)),
-    {noreply, State#state{timer_map = #{}}}.
-
-%% @private
-%% @doc Handling all non call/cast messages
--spec(handle_info(Info :: timeout() | term(), State :: #state{}) ->
-    {noreply, NewState :: #state{}} |
-    {noreply, NewState :: #state{}, timeout() | hibernate} |
-    {stop, Reason :: term(), NewState :: #state{}}).
-handle_info({timeout, _, {repost_ticker, {Id, Task}}}, State = #state{retry_interval = RetryInterval, timer_map = TimerMap}) ->
-    Task(),
-    TimerRef = erlang:start_timer(RetryInterval, self(), {repost_ticker, {Id, Task}}),
-
-    {noreply, State#state{timer_map = maps:put(Id, TimerRef, TimerMap)}}.
-
-%% @private
-%% @doc This function is called by a gen_server when it is about to
-%% terminate. It should be the opposite of Module:init/1 and do any
-%% necessary cleaning up. When it returns, the gen_server terminates
-%% with Reason. The return value is ignored.
--spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
-    State :: #state{}) -> term()).
-terminate(_Reason, _State = #state{}) ->
-    ok.
-
-%% @private
-%% @doc Convert process state when code is changed
--spec(code_change(OldVsn :: term() | {down, term()}, State :: #state{},
-    Extra :: term()) ->
-    {ok, NewState :: #state{}} | {error, Reason :: term()}).
-code_change(_OldVsn, State = #state{}, _Extra) ->
-    {ok, State}.
-
-%%%===================================================================
-%%% Internal functions
-%%%===================================================================

apps/iot/src/host/iot_container_task.erl

@@ -0,0 +1,147 @@
+%%%-------------------------------------------------------------------
+%%% @doc Owns one container deployment task stream.
+%%% @end
+%%%-------------------------------------------------------------------
+-module(iot_container_task).
+
+-behaviour(gen_server).
+
+%% API
+-export([start_link/2]).
+-export([get_pid/2, subscribe/3, stream/3, close/2]).
+
+%% gen_server callbacks
+-export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
+
+-define(MAX_EVENTS, 200).
+-define(STOP_AFTER_CLOSE, 300000).
+
+-record(state, {
+    uuid :: binary(),
+    task_id :: integer(),
+    status = pending :: pending | running | success | fail,
+    events = [] :: [{binary(), binary()}],
+    listeners = #{} :: #{pid() => reference()},
+    close_reason = undefined :: undefined | binary(),
+    stop_ref = undefined :: undefined | reference()
+}).
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+
+-spec start_link(binary(), integer()) -> {ok, pid()} | ignore | {error, term()}.
+start_link(UUID, TaskId) when is_binary(UUID), is_integer(TaskId) ->
+    gen_server:start_link({via, gproc, {n, l, name(UUID, TaskId)}}, ?MODULE, [UUID, TaskId], []).
+
+-spec get_pid(binary(), integer()) -> undefined | pid().
+get_pid(UUID, TaskId) when is_binary(UUID), is_integer(TaskId) ->
+    gproc:whereis_name({n, l, name(UUID, TaskId)}).
+
+-spec subscribe(binary(), integer(), pid()) -> ok | {error, term()}.
+subscribe(UUID, TaskId, ListenerPid) when is_binary(UUID), is_integer(TaskId), is_pid(ListenerPid) ->
+    case iot_container_task_sup:ensure_started(UUID, TaskId) of
+        {ok, Pid} ->
+            gen_server:call(Pid, {subscribe, ListenerPid});
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec stream(pid(), binary(), binary()) -> ok.
+stream(Pid, Type, Stream) when is_pid(Pid), is_binary(Type), is_binary(Stream) ->
+    gen_server:cast(Pid, {stream, Type, Stream}).
+
+-spec close(pid(), binary()) -> ok.
+close(Pid, Reason) when is_pid(Pid), is_binary(Reason) ->
+    gen_server:cast(Pid, {close, Reason}).
+
+%%%===================================================================
+%%% gen_server callbacks
+%%%===================================================================
+
+-spec init([binary() | integer()]) -> {ok, #state{}}.
+init([UUID, TaskId]) ->
+    ok = iot_log:set_metadata(),
+    {ok, #state{uuid = UUID, task_id = TaskId}}.
+
+-spec handle_call(term(), {pid(), term()}, #state{}) ->
+    {reply, term(), #state{}}.
+handle_call({subscribe, ListenerPid}, _From, State0 = #state{task_id = TaskId, events = Events, listeners = Listeners, close_reason = CloseReason}) ->
+    MonRef = erlang:monitor(process, ListenerPid),
+    maps:get(ListenerPid, Listeners, undefined) =/= undefined andalso erlang:demonitor(maps:get(ListenerPid, Listeners), [flush]),
+    lists:foreach(fun({Type, Stream}) ->
+        ListenerPid ! {stream_data, TaskId, Type, Stream}
+    end, Events),
+    case CloseReason of
+        undefined ->
+            ok;
+        Reason ->
+            ListenerPid ! {stream_close, TaskId, Reason}
+    end,
+    {reply, ok, State0#state{listeners = maps:put(ListenerPid, MonRef, Listeners)}};
+handle_call(_Request, _From, State) ->
+    {reply, ok, State}.
+
+-spec handle_cast(term(), #state{}) -> {noreply, #state{}}.
+handle_cast({stream, Type, Stream}, State0 = #state{task_id = TaskId, events = Events0, listeners = Listeners}) ->
+    Event = {Type, Stream},
+    Events = trim_events(Events0 ++ [Event]),
+    maps:foreach(fun(ListenerPid, _MonRef) ->
+        ListenerPid ! {stream_data, TaskId, Type, Stream}
+    end, Listeners),
+    {noreply, State0#state{status = running, events = Events}};
+handle_cast({close, Reason}, State0 = #state{task_id = TaskId, listeners = Listeners}) ->
+    maps:foreach(fun(ListenerPid, _MonRef) ->
+        ListenerPid ! {stream_close, TaskId, Reason}
+    end, Listeners),
+    Status = close_status(Reason),
+    {noreply, ensure_stop_timer(State0#state{status = Status, close_reason = Reason})};
+handle_cast(_Request, State) ->
+    {noreply, State}.
+
+-spec handle_info(term(), #state{}) -> {noreply, #state{}} | {stop, normal, #state{}}.
+handle_info({'DOWN', _Ref, process, ListenerPid, _Reason}, State = #state{listeners = Listeners}) ->
+    {noreply, State#state{listeners = maps:remove(ListenerPid, Listeners)}};
+handle_info(stop_after_close, State) ->
+    {stop, normal, State};
+handle_info(_Info, State) ->
+    {noreply, State}.
+
+-spec terminate(term(), #state{}) -> ok.
+terminate(_Reason, _State) ->
+    ok.
+
+-spec code_change(term(), #state{}, term()) -> {ok, #state{}}.
+code_change(_OldVsn, State, _Extra) ->
+    {ok, State}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec name(binary(), integer()) -> term().
+name(UUID, TaskId) ->
+    {iot_container_task, UUID, TaskId}.
+
+-spec trim_events([{binary(), binary()}]) -> [{binary(), binary()}].
+trim_events(Events) ->
+    Len = length(Events),
+    case Len > ?MAX_EVENTS of
+        true ->
+            lists:nthtail(Len - ?MAX_EVENTS, Events);
+        false ->
+            Events
+    end.
+
+-spec close_status(binary()) -> success | fail.
+close_status(<<"success">>) ->
+    success;
+close_status(_) ->
+    fail.
+
+-spec ensure_stop_timer(#state{}) -> #state{}.
+ensure_stop_timer(State = #state{stop_ref = undefined}) ->
+    Ref = erlang:send_after(?STOP_AFTER_CLOSE, self(), stop_after_close),
+    State#state{stop_ref = Ref};
+ensure_stop_timer(State) ->
+    State.

apps/iot/src/host/iot_container_task_sup.erl

@@ -0,0 +1,100 @@
+%%%-------------------------------------------------------------------
+%%% @doc Dynamic supervisor for container deployment tasks.
+%%% @end
+%%%-------------------------------------------------------------------
+-module(iot_container_task_sup).
+
+-behaviour(supervisor).
+
+%% API
+-export([start_link/0]).
+-export([ensure_started/2, stream/4, close/3, fail/3]).
+
+%% supervisor callbacks
+-export([init/1]).
+
+-define(SERVER, ?MODULE).
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+
+-spec start_link() -> {ok, pid()} | ignore | {error, term()}.
+start_link() ->
+    supervisor:start_link({local, ?SERVER}, ?MODULE, []).
+
+-spec ensure_started(binary(), integer()) -> {ok, pid()} | {error, term()}.
+ensure_started(UUID, TaskId) when is_binary(UUID), is_integer(TaskId) ->
+    case iot_container_task:get_pid(UUID, TaskId) of
+        undefined ->
+            case supervisor:start_child(?SERVER, [UUID, TaskId]) of
+                {ok, Pid} ->
+                    {ok, Pid};
+                {ok, Pid, _Info} ->
+                    {ok, Pid};
+                {error, {already_started, Pid}} ->
+                    {ok, Pid};
+                {error, {shutdown, {failed_to_start_child, _Id, {already_started, Pid}}}} ->
+                    {ok, Pid};
+                {error, Reason} ->
+                    {error, Reason}
+            end;
+        Pid when is_pid(Pid) ->
+            {ok, Pid}
+    end.
+
+-spec stream(binary(), integer(), binary(), binary()) -> ok.
+stream(UUID, TaskId, Type, Stream)
+  when is_binary(UUID), is_integer(TaskId), is_binary(Type), is_binary(Stream) ->
+    case ensure_started(UUID, TaskId) of
+        {ok, Pid} ->
+            iot_container_task:stream(Pid, Type, Stream);
+        {error, Reason} ->
+            logger:warning("[iot_container_task_sup] start task failed, uuid: ~p, task_id: ~p, reason: ~p", [UUID, TaskId, Reason]),
+            ok
+    end.
+
+-spec close(binary(), integer(), binary()) -> ok.
+close(UUID, TaskId, Reason) when is_binary(UUID), is_integer(TaskId), is_binary(Reason) ->
+    case ensure_started(UUID, TaskId) of
+        {ok, Pid} ->
+            iot_container_task:close(Pid, Reason);
+        {error, StartReason} ->
+            logger:warning("[iot_container_task_sup] start task before close failed, uuid: ~p, task_id: ~p, reason: ~p", [UUID, TaskId, StartReason]),
+            ok
+    end.
+
+-spec fail(binary(), integer(), term()) -> ok.
+fail(UUID, TaskId, Reason) when is_binary(UUID), is_integer(TaskId) ->
+    ReasonBin = reason_to_binary(Reason),
+    ok = stream(UUID, TaskId, <<"error">>, ReasonBin),
+    close(UUID, TaskId, <<"fail">>).
+
+%%%===================================================================
+%%% supervisor callbacks
+%%%===================================================================
+
+-spec init([]) -> {ok, {{simple_one_for_one, non_neg_integer(), pos_integer()}, [supervisor:child_spec()]}}.
+init([]) ->
+    SupFlags = {simple_one_for_one, 10, 60},
+    ChildSpec = #{
+        id => iot_container_task,
+        start => {iot_container_task, start_link, []},
+        restart => temporary,
+        shutdown => 5000,
+        type => worker,
+        modules => [iot_container_task]
+    },
+    {ok, {SupFlags, [ChildSpec]}}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec reason_to_binary(term()) -> binary().
+reason_to_binary(Reason) when is_binary(Reason) ->
+    Reason;
+reason_to_binary(timeout) ->
+    <<"timeout">>;
+reason_to_binary(Reason) ->
+    unicode:characters_to_binary(io_lib:format("~p", [Reason])).

apps/iot/src/host/iot_host.erl

@@ -0,0 +1,458 @@
+%%%-------------------------------------------------------------------
+%%% @author
+%%% @copyright (C) 2023, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 22. 9月 2023 16:38
+%%%-------------------------------------------------------------------
+-module(iot_host).
+-author("aresei").
+-include("iot.hrl").
+-include("domain_model.hrl").
+
+-behaviour(gen_statem).
+
+%% 心跳包检测时间间隔, 15分钟检测一次
+-define(HEARTBEAT_INTERVAL, 900 * 1000).
+
+%% 状态
+-define(STATE_DENIED, denied).
+-define(STATE_ACTIVATED, activated).
+
+%% API
+-export([start_link/2, get_name/1, get_alias_name/1, get_pid/1, handle/2, activate/2]).
+-export([lookup_pid/1]).
+-export([get_metric/1, get_status/1, kill/1]).
+%% 通讯相关
+-export([pub/4, attach_channel/2]).
+-export([deploy_container/3, start_container/2, stop_container/2, remove_container/2, kill_container/2, config_container/3, get_containers/1, await_reply/3]).
+-export([heartbeat/1]).
+
+%% gen_statem callbacks
+-export([init/1, handle_event/4, terminate/3, code_change/4, callback_mode/0]).
+
+-record(state, {
+    host_id :: integer(),
+    %% 从数据库里面读取到的数据
+    uuid :: binary(),
+    %% 主机状态
+    host_status :: integer(),
+
+    has_session = false :: boolean(),
+    %% 心跳计数器
+    heartbeat_counter = 0 :: integer(),
+    %% websocket相关
+    channel_pid :: undefined | pid(),
+    %% 主机的相关信息
+    metrics = #{} :: map()
+}).
+
+%%%===================================================================
+%%% API
+%%%===================================================================
+-spec get_pid(UUID :: binary()) -> undefined | pid().
+get_pid(UUID) when is_binary(UUID) ->
+    Name = get_name(UUID),
+    gproc:whereis_name({n, l, Name}).
+
+-spec lookup_pid(UUID :: binary()) -> {ok, pid()} | {error, Reason :: any()}.
+lookup_pid(UUID) when is_binary(UUID) ->
+    case get_pid(UUID) of
+        undefined ->
+            {error, <<"host not found">>};
+        HostPid ->
+            {ok, HostPid}
+    end.
+
+-spec get_name(UUID :: binary()) -> term().
+get_name(UUID) when is_binary(UUID) ->
+    {iot_host, UUID}.
+
+-spec get_alias_name(HostId :: integer()) -> term().
+get_alias_name(HostId0) when is_integer(HostId0) ->
+    {iot_host_id, HostId0}.
+
+-spec kill(UUID :: binary()) -> ok | true.
+kill(UUID) when is_binary(UUID) ->
+    case get_pid(UUID) of
+        undefined ->
+            ok;
+        Pid ->
+            exit(Pid, kill)
+    end.
+
+%% 处理消息
+-spec handle(Pid :: pid(), Packet :: tuple()) -> ok.
+handle(Pid, Packet) when is_pid(Pid) ->
+    gen_statem:cast(Pid, {handle, Packet}).
+
+-spec get_status(Pid :: pid()) -> {ok, Status :: map()}.
+get_status(Pid) when is_pid(Pid) ->
+    gen_statem:call(Pid, get_status).
+
+%% 激活主机, true 表示激活； false表示关闭激活
+-spec activate(Pid :: pid(), Auth :: boolean()) -> ok | {error, term()}.
+activate(Pid, Auth) when is_pid(Pid), is_boolean(Auth) ->
+    gen_statem:call(Pid, {activate, Auth}).
+
+-spec get_metric(Pid :: pid()) -> {ok, MetricInfo :: map()}.
+get_metric(Pid) when is_pid(Pid) ->
+    gen_statem:call(Pid, get_metric).
+
+-spec attach_channel(pid(), pid()) -> ok | {error, Reason :: binary()}.
+attach_channel(Pid, ChannelPid) when is_pid(Pid), is_pid(ChannelPid) ->
+    gen_statem:call(Pid, {attach_channel, ChannelPid}).
+
+-type request_ref() :: binary().
+
+-spec get_containers(Pid :: pid()) -> {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+get_containers(Pid) when is_pid(Pid) ->
+    container_call(Pid, docker_container_builder:list_request()).
+
+-spec config_container(Pid :: pid(), ContainerName :: binary(), ConfigJson :: binary()) -> {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+config_container(Pid, ContainerName, ConfigJson) when is_pid(Pid), is_binary(ContainerName), is_binary(ConfigJson) ->
+    container_call(Pid, docker_container_builder:config_request(ContainerName, ConfigJson)).
+
+-spec deploy_container(Pid :: pid(), TaskId :: integer(), Config :: map()) -> {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+deploy_container(Pid, TaskId, Config) when is_pid(Pid), is_integer(TaskId), is_map(Config) ->
+    case docker_container_builder:deploy_request(TaskId, Config) of
+        {ok, Request} ->
+            container_call(Pid, Request);
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec start_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+start_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
+    container_call(Pid, docker_container_builder:start_request(ContainerName)).
+
+-spec stop_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+stop_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
+    container_call(Pid, docker_container_builder:stop_request(ContainerName)).
+
+-spec kill_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+kill_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
+    container_call(Pid, docker_container_builder:kill_request(ContainerName)).
+
+-spec remove_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+remove_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
+    container_call(Pid, docker_container_builder:remove_request(ContainerName)).
+
+-spec await_reply(Pid :: pid(), Ref :: request_ref(), Timeout :: integer()) ->
+    ok | {ok, Result :: term()} | {error, Reason :: term()}.
+await_reply(Pid, Ref, Timeout) when is_pid(Pid), is_binary(Ref), is_integer(Timeout) ->
+    receive
+        {command_reply, Ref, ok} ->
+            ok;
+        {command_reply, Ref, {ok, Result}} ->
+            {ok, Result};
+        {command_reply, Ref, {error, Reason}} ->
+            {error, Reason}
+    after Timeout ->
+        ok = gen_statem:call(Pid, {cancel_command_call, Ref}),
+        flush_reply(Ref),
+        {error, timeout}
+    end.
+
+-spec pub(Pid :: pid(), Topic :: binary(), Qos :: integer(),  Content :: binary()) -> ok | {error, Reason :: any()}.
+pub(Pid, Topic, Qos, Content) when is_pid(Pid), is_binary(Topic), is_integer(Qos), is_binary(Content) ->
+    gen_statem:call(Pid, {pub, Topic, Qos, Content}).
+
+-spec heartbeat(Pid :: undefined | pid()) -> ok.
+heartbeat(undefined) ->
+    ok;
+heartbeat(Pid) when is_pid(Pid) ->
+    gen_statem:cast(Pid, heartbeat).
+
+%% @doc Creates a gen_statem process which calls Module:init/1 to
+%% initialize. To ensure a synchronized start-up procedure, this
+%% function does not return until Module:init/1 has returned.
+start_link(Name, UUID) when is_binary(UUID) ->
+    gen_statem:start_link({via, gproc, {n, l, Name}}, ?MODULE, [UUID], []).
+
+%%%===================================================================
+%%% gen_statem callbacks
+%%%===================================================================
+
+%% @private
+%% @doc Whenever a gen_statem is started using gen_statem:start/[3,4] or
+%% gen_statem:start_link/[3,4], this function is called by the new
+%% process to initialize.
+init([UUID]) ->
+    ok = iot_log:set_metadata(),
+    case iot_api_client:get_host_by_uuid(UUID) of
+        {ok, #host_info{id = HostId, authorize_status = AuthorizeStatus, status = HostStatus}} ->
+            %% 通过host_id注册别名, 可以避免通过查询数据库获取HostPid
+            AliasName = get_alias_name(HostId),
+            global:register_name(AliasName, self()),
+
+            %% 心跳检测机制
+            erlang:start_timer(?HEARTBEAT_INTERVAL, self(), heartbeat_ticker),
+
+            StateName = case AuthorizeStatus =:= ?HOST_AUTHORIZED of
+                            true  -> ?STATE_ACTIVATED;
+                            false -> ?STATE_DENIED
+                        end,
+            {ok, StateName, #state{host_id = HostId, uuid = UUID, host_status = HostStatus, has_session = false}};
+        undefined ->
+            logger:warning("[iot_host] host uuid: ~p, load host info failed", [UUID]),
+            ignore
+    end.
+
+%% @private
+%% @doc This function is called by a gen_statem when it needs to find out
+%% the callback mode of the callback module.
+callback_mode() ->
+    handle_event_function.
+
+%% @private
+%% @doc If callback_mode is handle_event_function, then whenever a
+%% gen_statem receives an event from call/2, cast/2, or as a normal
+%% process message, this function is called.
+
+handle_event({call, From}, get_metric, _, State = #state{metrics = Metrics}) ->
+    {keep_state, State, [{reply, From, {ok, Metrics}}]};
+
+%% 获取主机的状态
+handle_event({call, From}, get_status, _, State = #state{channel_pid = ChannelPid, heartbeat_counter = HeartbeatCounter, metrics = Metrics, has_session = HasSession, host_status = HostStatus}) ->
+    HasChannel = (ChannelPid /= undefined),
+    Reply = #{
+        <<"has_channel">> => HasChannel,
+        <<"has_session">> => HasSession,
+        <<"host_status">> => HostStatus,
+        <<"heartbeat_counter">> => HeartbeatCounter,
+        <<"metrics">> => Metrics
+    },
+    {keep_state, State, [{reply, From, {ok, Reply}}]};
+
+handle_event({call, From}, {container_call, ReceiverPid, Request}, _, State = #state{uuid = UUID, channel_pid = ChannelPid, has_session = HasSession}) ->
+    case HasSession andalso is_pid(ChannelPid) of
+        true ->
+            Ref = request_ref(),
+            ok = ssl_channel:container_call(ChannelPid, ReceiverPid, Ref, Request),
+            {keep_state, State, [{reply, From, {ok, Ref}}]};
+        false ->
+            logger:debug("[iot_host] uuid: ~p, invalid state: ~p", [UUID, state_map(State)]),
+            {keep_state, State, [{reply, From, {error, <<"主机离线，发送命令失败"/utf8>>}}]}
+    end;
+
+handle_event({call, From}, {cancel_command_call, Ref}, _, State = #state{channel_pid = ChannelPid}) ->
+    case is_pid(ChannelPid) of
+        true ->
+            ok = ssl_channel:cancel_command_call(ChannelPid, Ref),
+            {keep_state, State, [{reply, From, ok}]};
+        false ->
+            {keep_state, State, [{reply, From, ok}]}
+    end;
+
+%% 发送指令时, pub/sub
+handle_event({call, From}, {pub, Topic, Qos, Content}, StateName, State = #state{uuid = UUID, channel_pid = ChannelPid, has_session = HasSession}) ->
+    case StateName of
+        ?STATE_ACTIVATED ->
+            case HasSession andalso is_pid(ChannelPid) of
+                true ->
+                    logger:debug("[iot_host] host: ~p, publish to topic: ~p, content: ~p", [UUID, Topic, Content]),
+                    %% 通过websocket发送消息
+                    ssl_channel:pub(ChannelPid, Topic, Qos, Content),
+
+                    {keep_state, State, [{reply, From, ok}]};
+                false ->
+                    logger:debug("[iot_host] uuid: ~p, publish to topic: ~p, content: ~p, invalid state: ~p", [UUID, Topic, Content, state_map(State)]),
+                    {keep_state, State, [{reply, From, {error, <<"主机离线，发送失败"/utf8>>}}]}
+            end;
+        ?STATE_DENIED ->
+            {keep_state, State, [{reply, From, {error, <<"主机未激活，发送失败"/utf8>>}}]}
+    end;
+
+%% 激活/关闭授权只修改 iot 本地授权状态；efka 连接保持在线，数据是否处理由 host 状态决定。
+handle_event({call, From}, {activate, Auth}, _, State = #state{uuid = UUID}) ->
+    NStateName = case Auth of
+                     true ->
+                         ?STATE_ACTIVATED;
+                     false ->
+                         ?STATE_DENIED
+                 end,
+    logger:debug("[iot_host] uuid: ~p, state_name change to ~p", [UUID, NStateName]),
+    {next_state, NStateName, State, [{reply, From, ok}]};
+
+%% 绑定channel
+handle_event({call, From}, {attach_channel, ChannelPid}, StateName, State = #state{uuid = UUID, channel_pid = OldChannelPid, host_status = HostStatus}) ->
+    case OldChannelPid == undefined orelse OldChannelPid =:= ChannelPid of
+        true ->
+            case StateName of
+                ?STATE_ACTIVATED ->
+                    erlang:monitor(process, ChannelPid),
+                    %% 更新主机为在线状态
+                    NState = mark_host_online(UUID, HostStatus, State#state{channel_pid = ChannelPid, has_session = true}),
+                    {keep_state, NState, [{reply, From, ok}]};
+                %% 主机未激活
+                ?STATE_DENIED ->
+                    logger:notice("[iot_host] attach_channel host_id uuid: ~p, channel: ~p, host denied locally", [UUID, ChannelPid]),
+                    erlang:monitor(process, ChannelPid),
+                    NState = mark_host_online(UUID, HostStatus, State#state{channel_pid = ChannelPid, has_session = true}),
+                    {keep_state, NState, [{reply, From, ok}]}
+            end;
+        false ->
+            logger:notice("[iot_host] attach_channel host_id uuid: ~p, old channel exists: ~p", [UUID, OldChannelPid]),
+            {keep_state, State, [{reply, From, {error, <<"channel existed">>}}]}
+    end;
+
+%% 数据分发
+handle_event(cast, {handle, {data, RouteKey, MetricBin}}, ?STATE_ACTIVATED,
+    State = #state{uuid = UUID, has_session = true}) ->
+    logger:debug("[iot_host] metric_data host: ~p, route_key: ~p, metric: ~p", [UUID, RouteKey, MetricBin]),
+    endpoint_subscription:publish(get_route_key(RouteKey), MetricBin),
+    {keep_state, State};
+
+handle_event(cast, {handle, {data, RouteKey, MetricBin}}, StateName,
+    State = #state{uuid = UUID, has_session = true}) ->
+    logger:notice("[iot_host] host_uuid: ~p, state_name: ~p, metric_data route_key: ~p, metric: ~p, discard", [UUID, StateName, RouteKey, MetricBin]),
+    {keep_state, State};
+
+%% ping的数据是通过aes加密后的，因此需要在有会话的情况下才行
+handle_event(cast, {handle, {ping, Metrics}}, ?STATE_ACTIVATED, State = #state{uuid = UUID, has_session = true}) ->
+    logger:debug("[iot_host] ping host_id uuid: ~p, get ping: ~p", [UUID, Metrics]),
+    {keep_state, State#state{metrics = Metrics}};
+
+%% 心跳机制
+handle_event(cast, heartbeat, _, State = #state{uuid = UUID, heartbeat_counter = HeartbeatCounter, host_status = HostStatus}) ->
+    NState = mark_host_online(UUID, HostStatus, State#state{heartbeat_counter = HeartbeatCounter + 1}),
+    {keep_state, NState};
+
+%% UDP 心跳丢失但 SSL channel 仍在时，不能把 host 标记为离线。
+handle_event(info, {timeout, _, heartbeat_ticker}, _,
+    State = #state{uuid = UUID, heartbeat_counter = 0, host_status = HostStatus, channel_pid = ChannelPid}) when is_pid(ChannelPid) ->
+
+    erlang:start_timer(?HEARTBEAT_INTERVAL, self(), heartbeat_ticker),
+    logger:warning("[iot_host] uuid: ~p, udp heartbeat lost but ssl channel is alive: ~p", [UUID, ChannelPid]),
+    NState = mark_host_online(UUID, HostStatus, State#state{heartbeat_counter = 0}),
+    {keep_state, NState};
+
+%% 没有收到 UDP 心跳且没有 SSL channel，主机下线, 设备状态不变
+handle_event(info, {timeout, _, heartbeat_ticker}, _, State = #state{uuid = UUID, heartbeat_counter = 0, host_status = HostStatus}) ->
+    logger:warning("[iot_host] uuid: ~p, heartbeat lost, devices will unknown", [UUID]),
+    erlang:start_timer(?HEARTBEAT_INTERVAL, self(), heartbeat_ticker),
+    NState = mark_host_offline(UUID, HostStatus, State#state{channel_pid = undefined, has_session = false, heartbeat_counter = 0}),
+    {keep_state, NState};
+
+%% 其他情况下需要重置系统计数器
+handle_event(info, {timeout, _, heartbeat_ticker}, _, State = #state{}) ->
+    erlang:start_timer(?HEARTBEAT_INTERVAL, self(), heartbeat_ticker),
+    {keep_state, State#state{heartbeat_counter = 0}};
+
+%% 当websocket断开的时候，主机的状态不一定改变；主机的状态改变通过心跳机制，会话状态需要改变
+handle_event(info, {'DOWN', _Ref, process, ChannelPid, Reason}, _, State = #state{uuid = UUID, channel_pid = ChannelPid, has_session = HasSession}) ->
+    logger:warning("[iot_host] uuid: ~p, channel: ~p, down with reason: ~p, has_session: ~p, state: ~p", [UUID, ChannelPid, Reason, HasSession, State]),
+    {keep_state, State#state{channel_pid = undefined, has_session = false}};
+
+handle_event(info, {'DOWN', _Ref, process, Pid, Reason}, _, State = #state{uuid = UUID}) ->
+    logger:debug("[iot_host] uuid: ~p, process_pid: ~p, down with reason: ~p, state: ~p", [UUID, Pid, Reason, State]),
+    {keep_state, State};
+
+handle_event(Event, Info, StateName, State = #state{uuid = UUID}) ->
+    logger:warning("[iot_host] host: ~p, event: ~p, unknown message: ~p, state_name: ~p, state: ~p", [UUID, Event, Info, StateName, state_map(State)]),
+    {keep_state, State}.
+
+%% @private
+%% @doc This function is called by a gen_statem when it is about to
+%% terminate. It should be the opposite of Module:init/1 and do any
+%% necessary cleaning up. When it returns, the gen_statem terminates with
+%% Reason. The return value is ignored.
+terminate(Reason, _StateName, _State = #state{uuid = UUID, has_session = HasSession}) ->
+    logger:debug("[iot_host] host: ~p, terminate with reason: ~p, has_session: ~p", [UUID, Reason, HasSession]),
+    ok.
+
+%% @private
+%% @doc Convert process state when code is changed
+code_change(_OldVsn, StateName, State = #state{}, _Extra) ->
+    {ok, StateName, State}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec container_call(Pid :: pid(), Request :: term()) ->
+    {ok, Ref :: request_ref()} | {error, Reason :: any()}.
+container_call(Pid, Request) when is_pid(Pid) ->
+    gen_statem:call(Pid, {container_call, self(), Request}).
+
+-spec get_route_key(binary()) -> binary().
+get_route_key(<<"">>) ->
+    <<"/">>;
+get_route_key(RouteKey) when is_binary(RouteKey) ->
+    RouteKey.
+
+%% 将当前的state转换成map
+state_map(#state{host_id = HostId, uuid = UUID, host_status = HostStatus, has_session = HasSession, heartbeat_counter = HeartbeatCounter, channel_pid = ChannelPid, metrics = Metrics}) ->
+    #{
+        host_id => HostId,
+        uuid => UUID,
+        host_status => HostStatus,
+        has_session => HasSession,
+        heartbeat_counter => HeartbeatCounter,
+        channel_pid => ChannelPid,
+        metrics => Metrics
+    }.
+
+flush_reply(Ref) ->
+    receive
+        {command_reply, Ref, _Reply} ->
+            ok
+    after 0 ->
+        ok
+    end.
+
+-spec request_ref() -> request_ref().
+request_ref() ->
+    crypto:strong_rand_bytes(16).
+
+-spec mark_host_online(binary(), integer(), #state{}) -> #state{}.
+mark_host_online(UUID, HostStatus, State) ->
+    case maybe_mark_host_online(UUID, HostStatus) of
+        keep_status ->
+            State;
+        {next_status, NStatus} ->
+            State#state{host_status = NStatus}
+    end.
+
+-spec mark_host_offline(binary(), integer(), #state{}) -> #state{}.
+mark_host_offline(UUID, HostStatus, State) ->
+    case maybe_mark_host_offline(UUID, HostStatus) of
+        keep_status ->
+            State;
+        {next_status, NStatus} ->
+            State#state{host_status = NStatus}
+    end.
+
+-spec maybe_mark_host_online(binary(), integer()) -> {next_status, NStatus :: integer()} | keep_status.
+maybe_mark_host_online(UUID, HostStatus) ->
+    case HostStatus of
+        ?HOST_OFFLINE ->
+            case iot_api_client:change_host_status(UUID, ?HOST_ONLINE) of
+                {ok, Result} ->
+                    logger:debug("[iot_host] uuid: ~p, change host status online success: ~p", [UUID, Result]),
+                    {next_status, ?HOST_ONLINE};
+                {error, Reason} ->
+                    logger:warning("[iot_host] uuid: ~p, change host status online failed: ~p", [UUID, Reason]),
+                    keep_status
+            end;
+        _ ->
+            keep_status
+    end.
+
+-spec maybe_mark_host_offline(binary(), integer()) -> {next_status, NStatus :: integer()} | keep_status.
+maybe_mark_host_offline(UUID, HostStatus) ->
+    case HostStatus of
+        ?HOST_ONLINE ->
+            case iot_api_client:change_host_status(UUID, ?HOST_OFFLINE) of
+                {ok, Result} ->
+                    logger:debug("[iot_host] uuid: ~p, change host status offline success: ~p", [UUID, Result]),
+                    {next_status, ?HOST_OFFLINE};
+                {error, Reason} ->
+                    logger:warning("[iot_host] uuid: ~p, change host status offline failed: ~p", [UUID, Reason]),
+                    keep_status
+            end;
+        _ ->
+            keep_status
+    end.

apps/iot/src/host/iot_host_sup.erl

@@ -0,0 +1,85 @@
+%%%-------------------------------------------------------------------
+%%% @author aresei
+%%% @copyright (C) 2023, <COMPANY>
+%%% @doc
+%%% @end
+%%%-------------------------------------------------------------------
+-module(iot_host_sup).
+-include("iot.hrl").
+
+-behaviour(supervisor).
+
+-export([start_link/0, init/1, delete_host/1, ensured_host_started/1]).
+
+start_link() ->
+    supervisor:start_link({local, ?MODULE}, ?MODULE, []).
+
+init([]) ->
+    Specs = lists:map(fun child_spec/1, iot_api_client:get_all_hosts()),
+
+    {ok, {#{strategy => one_for_one, intensity => 1000, period => 3600}, Specs}}.
+
+-spec ensured_host_started(UUID :: binary()) -> {ok, Pid :: pid()} | {error, Reason :: any()}.
+ensured_host_started(UUID) when is_binary(UUID) ->
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            %% 尝试删除下host对应的信息
+            case delete_host(UUID) of
+                ok ->
+                    case supervisor:start_child(?MODULE, child_spec(UUID)) of
+                        {ok, Pid} when is_pid(Pid) ->
+                            {ok, Pid};
+                        {error, {'already_started', Pid}} when is_pid(Pid) ->
+                            {ok, Pid};
+                        {error, Error} ->
+                            {error, Error}
+                    end;
+                {error, Reason} ->
+                    {error, Reason}
+            end;
+        Pid when is_pid(Pid) ->
+            {ok, Pid}
+    end.
+
+delete_host(UUID) when is_binary(UUID) ->
+    Id = iot_host:get_name(UUID),
+    case supervisor:terminate_child(?MODULE, Id) of
+        ok ->
+            delete_host_child(Id, UUID);
+        {error, not_found} ->
+            delete_host_child(Id, UUID);
+        {error, Reason} ->
+            logger:warning("[iot_host_sup] terminate host: ~p failed: ~p", [UUID, Reason]),
+            {error, Reason}
+    end.
+
+delete_host_child(Id, UUID) ->
+    case supervisor:delete_child(?MODULE, Id) of
+        ok ->
+            ok;
+        {error, not_found} ->
+            ok;
+        {error, running} ->
+            %% ensure killed then delete again
+            iot_host:kill(UUID),
+            case supervisor:delete_child(?MODULE, Id) of
+                ok ->
+                    ok;
+                {error, not_found} ->
+                    ok;
+                {error, Reason} ->
+                    {error, Reason}
+            end;
+        {error, Reason} ->
+            {error, Reason}
+    end.
+
+-spec child_spec(UUID :: binary()) -> map().
+child_spec(UUID) when is_binary(UUID) ->
+    Id = iot_host:get_name(UUID),
+    #{id => Id,
+        start => {iot_host, start_link, [Id, UUID]},
+        restart => permanent,
+        shutdown => 2000,
+        type => worker,
+        modules => ['iot_host']}.

apps/iot/src/http/container_handler.erl

@@ -0,0 +1,239 @@
+%%%-------------------------------------------------------------------
+%%% @author licheng5
+%%% @copyright (C) 2020, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 26. 4月 2020 3:36 下午
+%%%-------------------------------------------------------------------
+-module(container_handler).
+-author("licheng5").
+-include("iot.hrl").
+
+-define(REQ_TIMEOUT, 10000).
+
+%% API
+-export([handle_request/4]).
+
+handle_request("GET", "/container/get_all", #{<<"uuid">> := UUID}, _) when is_binary(UUID) ->
+    %% 检查ConfigJson是否是合法的json字符串
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            {ok, 200, iot_util:json_error(-1, <<"host not found">>)};
+        Pid when is_pid(Pid) ->
+                    case iot_host:get_containers(Pid) of
+                {ok, Ref} ->
+                    case iot_host:await_reply(Pid, Ref, ?REQ_TIMEOUT) of
+                        ok ->
+                            {ok, 200, request_success_response(<<"ok">>)};
+                        {ok, Result} ->
+                            {ok, 200, request_success_response(Result)};
+                        {error, Reason} ->
+                            request_error_http_response(Reason)
+                    end;
+                {error, Reason} when is_binary(Reason) ->
+                    {ok, 200, iot_util:json_error(-1, Reason)}
+            end
+    end;
+
+%% 下发config.json, 微服务接受后，保存服务配置
+handle_request("POST", "/container/push_config", _,
+    #{<<"uuid">> := UUID, <<"container_name">> := ContainerName, <<"config">> := Config, <<"timeout">> := Timeout0})
+    when is_binary(UUID), is_binary(ContainerName), is_binary(Config), is_integer(Timeout0) ->
+
+    %% 检查ConfigJson是否是合法的json字符串
+    true = iot_util:is_json(Config),
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            {ok, 200, iot_util:json_error(-1, <<"host not found">>)};
+        Pid when is_pid(Pid) ->
+            Timeout = Timeout0 * 1000,
+            case iot_host:config_container(Pid, ContainerName, Config) of
+                {ok, Ref} ->
+                    case iot_host:await_reply(Pid, Ref, Timeout) of
+                        ok ->
+                            {ok, 200, request_success_response(<<"ok">>)};
+                        {ok, Result} ->
+                            {ok, 200, request_success_response(Result)};
+                        {error, Reason} ->
+                            request_error_http_response(Reason)
+                    end;
+                {error, Reason} when is_binary(Reason) ->
+                    {ok, 200, iot_util:json_error(-1, Reason)}
+            end
+    end;
+
+%% 部署微服务
+handle_request("POST", "/container/deploy", _, #{<<"uuid">> := UUID, <<"task_id">> := TaskId, <<"config">> := Config})
+    when is_binary(UUID), is_integer(TaskId), is_map(Config) ->
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
+        Pid when is_pid(Pid) ->
+            case iot_container_task_sup:ensure_started(UUID, TaskId) of
+                {ok, _TaskPid} ->
+                    handle_deploy_container(Pid, UUID, TaskId, Config);
+                {error, Reason} ->
+                    {ok, 500, iot_util:json_error(500, reason_to_binary(Reason))}
+            end
+    end;
+
+%% 启动服务
+handle_request("POST", "/container/start", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
+        Pid when is_pid(Pid) ->
+            case iot_host:start_container(Pid, ContainerName) of
+                {ok, Ref} ->
+                    case iot_host:await_reply(Pid, Ref, ?REQ_TIMEOUT) of
+                        ok ->
+                            {ok, 200, request_success_response(<<"ok">>)};
+                        {ok, Result} ->
+                            {ok, 200, request_success_response(Result)};
+                        {error, Reason} ->
+                            request_error_http_response(Reason)
+                    end;
+                {error, Reason} when is_binary(Reason) ->
+                    {ok, 200, iot_util:json_error(400, Reason)}
+            end
+    end;
+
+%% 停止服务
+handle_request("POST", "/container/stop", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
+        Pid when is_pid(Pid) ->
+            case iot_host:stop_container(Pid, ContainerName) of
+                {ok, Ref} ->
+                    case iot_host:await_reply(Pid, Ref, ?REQ_TIMEOUT) of
+                        ok ->
+                            {ok, 200, request_success_response(<<"ok">>)};
+                        {ok, Result} ->
+                            {ok, 200, request_success_response(Result)};
+                        {error, Reason} ->
+                            request_error_http_response(Reason)
+                    end;
+                {error, Reason} when is_binary(Reason) ->
+                    {ok, 200, iot_util:json_error(400, Reason)}
+            end
+    end;
+
+handle_request("POST", "/container/kill", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
+        Pid when is_pid(Pid) ->
+            case iot_host:kill_container(Pid, ContainerName) of
+                {ok, Ref} ->
+                    case iot_host:await_reply(Pid, Ref, ?REQ_TIMEOUT) of
+                        ok ->
+                            {ok, 200, request_success_response(<<"ok">>)};
+                        {ok, Result} ->
+                            {ok, 200, request_success_response(Result)};
+                        {error, Reason} ->
+                            request_error_http_response(Reason)
+                    end;
+                {error, Reason} when is_binary(Reason) ->
+                    {ok, 200, iot_util:json_error(400, Reason)}
+            end
+    end;
+
+%% 删除容器
+handle_request("POST", "/container/remove", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
+    case iot_host:get_pid(UUID) of
+        undefined ->
+            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
+        Pid when is_pid(Pid) ->
+            case iot_host:remove_container(Pid, ContainerName) of
+                {ok, Ref} ->
+                    case iot_host:await_reply(Pid, Ref, ?REQ_TIMEOUT) of
+                        ok ->
+                            {ok, 200, request_success_response(<<"ok">>)};
+                        {ok, Result} ->
+                            {ok, 200, request_success_response(Result)};
+                        {error, Reason} ->
+                            request_error_http_response(Reason)
+                    end;
+                {error, Reason} when is_binary(Reason) ->
+                    {ok, 200, iot_util:json_error(400, Reason)}
+            end
+    end;
+
+handle_request(_, Path, _, _) ->
+    Path1 = list_to_binary(Path),
+    {ok, 200, iot_util:json_error(-1, <<"url: ", Path1/binary, " not found">>)}.
+
+request_success_response(Result) when is_binary(Result) ->
+    case decode_json_bytes(Result) of
+        {ok, Data} ->
+            iot_util:json_data(Data);
+        error ->
+            iot_util:json_data(Result)
+    end;
+request_success_response(Result) ->
+    iot_util:json_data(Result).
+
+request_error_response(Code, Reason) when is_integer(Code), is_binary(Reason) ->
+    case decode_json_bytes(Reason) of
+        {ok, #{<<"message">> := Message}} when is_binary(Message) ->
+            iot_util:json_error(Code, Message);
+        {ok, Message} when is_binary(Message) ->
+            iot_util:json_error(Code, Message);
+        _ ->
+            iot_util:json_error(Code, Reason)
+    end.
+
+-spec request_error_http_response(Reason :: term()) ->
+    {ok, HttpStatus :: 400 | 504, Body :: binary()}.
+request_error_http_response(Reason) ->
+    HttpStatus = request_error_status(Reason),
+    {ok, HttpStatus, request_error_response(HttpStatus, reason_to_binary(Reason))}.
+
+-spec request_error_status(Reason :: term()) -> integer().
+request_error_status(timeout) ->
+    504;
+request_error_status(<<"timeout">>) ->
+    504;
+request_error_status(_) ->
+    400.
+
+-spec reason_to_binary(term()) -> binary().
+reason_to_binary(Reason) when is_binary(Reason) ->
+    Reason;
+reason_to_binary(timeout) ->
+    <<"timeout">>;
+reason_to_binary(invalid_response) ->
+    <<"invalid response">>;
+reason_to_binary(Reason) ->
+    unicode:characters_to_binary(io_lib:format("~p", [Reason])).
+
+decode_json_bytes(Data) when is_binary(Data) ->
+    case catch json:decode(Data) of
+        {'EXIT', _} ->
+            error;
+        {error, _} ->
+            error;
+        Decoded ->
+            {ok, Decoded}
+    end.
+
+-spec handle_deploy_container(pid(), binary(), integer(), map()) ->
+    {ok, 200 | 400 | 504, binary()}.
+handle_deploy_container(Pid, UUID, TaskId, Config) ->
+    case iot_host:deploy_container(Pid, TaskId, Config) of
+        {ok, Ref} ->
+            case iot_host:await_reply(Pid, Ref, ?REQ_TIMEOUT) of
+                ok ->
+                    {ok, 200, request_success_response(<<"ok">>)};
+                {ok, Result} ->
+                    {ok, 200, request_success_response(Result)};
+                {error, Reason} ->
+                    ok = iot_container_task_sup:fail(UUID, TaskId, Reason),
+                    request_error_http_response(Reason)
+            end;
+        {error, Reason} when is_binary(Reason) ->
+            ok = iot_container_task_sup:fail(UUID, TaskId, Reason),
+            {ok, 200, iot_util:json_error(400, Reason)}
+    end.

apps/iot/src/http/endpoint_handler.erl

@@ -8,7 +8,7 @@
 %%%-------------------------------------------------------------------
 -module(endpoint_handler).
 -author("licheng5").
--include("endpoint.hrl").
+-include_lib("endpoint/include/endpoint.hrl").
 
 %% API
 -export([handle_request/4]).
@@ -30,17 +30,17 @@ handle_request("POST", "/endpoint/run_statuses", _, Ids) when is_list(Ids) ->
     {ok, 200, iot_util:json_data(Statuses)};
 
 handle_request("POST", "/endpoint/start", _, #{<<"id">> := Id}) when is_integer(Id) ->
-    case iot_api:get_endpoint(Id) of
+    case iot_api_client:get_endpoint(Id) of
         undefined ->
             {ok, 200, iot_util:json_error(404, <<"endpoint not found">>)};
         {ok, EndpointInfo} ->
             case endpoint:endpoint_record(EndpointInfo) of
                 {ok, Endpoint = #endpoint{title = Title}} ->
-                    case endpoint_sup:ensured_endpoint_started(Endpoint) of
+                    case endpoint_adapter_sup:ensured_endpoint_started(Endpoint) of
                         {ok, Pid} when is_pid(Pid) ->
                             {ok, 200, iot_util:json_data(<<"success">>)};
                         {error, Reason} ->
-                            lager:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
+                            logger:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
                             {ok, 200, iot_util:json_error(404, <<"start endpoint error">>)}
                     end;
                 error ->
@@ -49,21 +49,21 @@ handle_request("POST", "/endpoint/start", _, #{<<"id">> := Id}) when is_integer(
     end;
 
 handle_request("POST", "/endpoint/stop", _, #{<<"id">> := Id}) when is_integer(Id) ->
-    case iot_api:get_endpoint(Id) of
+    case iot_api_client:get_endpoint(Id) of
         undefined ->
             {ok, 200, iot_util:json_error(404, <<"endpoint not found">>)};
         {ok, _} ->
-            case endpoint_sup:delete_endpoint(Id) of
+            case endpoint_adapter_sup:delete_endpoint(Id) of
                 ok ->
                     {ok, 200, iot_util:json_data(<<"success">>)};
                 {error, Reason} ->
-                    lager:warning("[endpoint_handler] stop endpoint id: ~p, get error: ~p", [Id, Reason]),
+                    logger:warning("[endpoint_handler] stop endpoint id: ~p, get error: ~p", [Id, Reason]),
                     {ok, 200, iot_util:json_error(404, <<"stop endpoint error">>)}
             end
     end;
 
 handle_request("POST", "/endpoint/restart", _, #{<<"id">> := Id}) when is_integer(Id) ->
-    case iot_api:get_endpoint(Id) of
+    case iot_api_client:get_endpoint(Id) of
         undefined ->
             {ok, 200, iot_util:json_error(404, <<"endpoint not found">>)};
         {ok, EndpointInfo} ->
@@ -71,25 +71,25 @@ handle_request("POST", "/endpoint/restart", _, #{<<"id">> := Id}) when is_intege
                 {ok, Endpoint = #endpoint{title = Title}} ->
                     case endpoint:get_pid(Id) of
                         undefined ->
-                            case endpoint_sup:ensured_endpoint_started(Endpoint) of
+                            case endpoint_adapter_sup:ensured_endpoint_started(Endpoint) of
                                 {ok, Pid} when is_pid(Pid) ->
                                     {ok, 200, iot_util:json_data(<<"success">>)};
                                 {error, Reason} ->
-                                    lager:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
+                                    logger:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
                                     {ok, 200, iot_util:json_error(404, <<"restart endpoint error">>)}
                             end;
                         Pid when is_pid(Pid) ->
-                            case endpoint_sup:delete_endpoint(Id) of
+                            case endpoint_adapter_sup:delete_endpoint(Id) of
                                 ok ->
-                                    case endpoint_sup:ensured_endpoint_started(Endpoint) of
+                                    case endpoint_adapter_sup:ensured_endpoint_started(Endpoint) of
                                         {ok, Pid0} when is_pid(Pid0) ->
                                             {ok, 200, iot_util:json_data(<<"success">>)};
                                         {error, Reason} ->
-                                            lager:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
+                                            logger:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
                                             {ok, 200, iot_util:json_error(404, <<"restart endpoint error">>)}
                                     end;
                                 {error, Reason} ->
-                                    lager:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
+                                    logger:warning("[endpoint_handler] start endpoint: ~p, get error: ~p", [Title, Reason]),
                                     {ok, 200, iot_util:json_error(404, <<"stop endpoint error">>)}
                             end
                     end;
@@ -100,57 +100,23 @@ handle_request("POST", "/endpoint/restart", _, #{<<"id">> := Id}) when is_intege
 
 %% http接口的测试
 handle_request("POST", "/endpoint/test", _, #{<<"protocol">> := <<"http">>, <<"config">> := #{<<"url">> := Url, <<"pool_size">> := PoolSize}}) when is_integer(PoolSize), PoolSize > 0 ->
-    Body = <<"">>,
+    case endpoint_tester:test(#http_endpoint{url = Url, pool_size = PoolSize}) of
-    ContentType = "application/json",
+        ok ->
-    case httpc:request(post, {Url, [], ContentType, Body}, [], []) of
-        {ok, _} ->
             {ok, 200, iot_util:json_data(<<"ok">>)};
         {error, Reason} ->
-            lager:debug("[endpint_handler] test http: ~p, error: ~p", [Url, Reason]),
+            logger:debug("[endpint_handler] test http: ~p, error: ~p", [Url, Reason]),
             {ok, 200, iot_util:json_error(-1, <<"url failed">>)}
     end;
 
 %% 测试mqtt
 handle_request("POST", "/endpoint/test", _, #{<<"protocol">> := <<"mqtt">>, <<"config">> := Config}) ->
     case endpoint:parse_config(<<"mqtt">>, Config) of
-        {ok, #mqtt_endpoint{host = Host, port = Port, username = Username, password = Password}} ->
+        {ok, MqttEndpoint = #mqtt_endpoint{}} ->
-            %% 用户测试的client
+            case endpoint_tester:test(MqttEndpoint) of
-            ClientId = "mqtt_client_test:" ++ iot_util:rand_bytes(16),
+                ok ->
-            Opts = [
+                    {ok, 200, iot_util:json_data(<<"ok">>)};
-                {owner, self()},
+                {error, Reason} ->
-                {clientid, ClientId},
+                    {ok, 200, iot_util:json_error(-1, Reason)}
-                {host, binary_to_list(Host)},
-                {port, Port},
-                {tcp_opts, []},
-                {username, binary_to_list(Username)},
-                {password, binary_to_list(Password)},
-                {keepalive, 86400},
-                {auto_ack, true},
-                {connect_timeout, 5000},
-                {proto_ver, v5},
-                {retry_interval, 5000}
-            ],
-
-            case emqtt:start_link(Opts) of
-                {ok, ConnPid} ->
-                    lager:debug("[endpint_handler] start connect, options: ~p", [Opts]),
-                    case catch emqtt:connect(ConnPid, 5000) of
-                        {ok, _} ->
-                            lager:debug("[endpint_handler] connect success, pid: ~p", [ConnPid]),
-                            emqtt:stop(ConnPid),
-                            {ok, 200, iot_util:json_data(<<"ok">>)};
-                        {error, Reason} ->
-                            lager:warning("[endpint_handler] connect get error: ~p", [Reason]),
-                            emqtt:stop(ConnPid),
-                            {ok, 200, iot_util:json_error(-1, <<"connect mqtt server failed">>)};
-                        Error ->
-                            lager:warning("[endpint_handler] connect get error: ~p", [Error]),
-                            emqtt:stop(ConnPid),
-                            {ok, 200, iot_util:json_error(-1, <<"connect mqtt server failed">>)}
-                    end;
-                Other ->
-                    lager:warning("[endpint_handler] test connect mqtt with options: ~p, get error: ~p", [Opts, Other]),
-                    {ok, 200, iot_util:json_error(-1, <<"connect mqtt server failed">>)}
             end;
         {error, Errors} ->
             {ok, 200, iot_util:json_error(-1, Errors)}
@@ -159,33 +125,12 @@ handle_request("POST", "/endpoint/test", _, #{<<"protocol">> := <<"mqtt">>, <<"c
 %% 测试kafka
 handle_request("POST", "/endpoint/test", _, #{<<"protocol">> := <<"kafka">>, <<"config">> := Config}) ->
     case endpoint:parse_config(<<"kafka">>, Config) of
-        {ok, #kafka_endpoint{sasl_config = SaslConfig, bootstrap_servers = BootstrapServers, topic = Topic}} ->
+        {ok, KafkaEndpoint = #kafka_endpoint{}} ->
-            BaseConfig = [
+            case endpoint_tester:test(KafkaEndpoint) of
-                {reconnect_cool_down_seconds, 5},
+                ok ->
-                {socket_options, [{keepalive, true}]}
+                    {ok, 200, iot_util:json_data(<<"ok">>)};
-            ],
+                {error, Reason} ->
-
+                    {ok, 200, iot_util:json_error(-1, Reason)}
-            ClientConfig = case SaslConfig of
-                               {Mechanism, Username, Password} ->
-                                   [{sasl, {Mechanism, Username, Password}}|BaseConfig];
-                               undefined ->
-                                   BaseConfig
-                           end,
-            ClientId = list_to_atom("brod_client_test:" ++ iot_util:rand_bytes(16)),
-
-            case catch brod:start_link_client(BootstrapServers, ClientId, ClientConfig) of
-                {ok, _ClientPid} ->
-                    case brod:start_producer(ClientId, Topic, _ProducerConfig = []) of
-                        ok ->
-                            ok = brod:stop_client(ClientId),
-                            {ok, 200, iot_util:json_data(<<"ok">>)};
-                        {error, Reason} ->
-                            lager:debug("[endpint_handler] start_producer: ~p, get error: ~p", [ClientId, Reason]),
-                            {ok, 200, iot_util:json_error(-1, <<"config kafka server failed">>)}
-                    end;
-                Error ->
-                    lager:debug("[endpint_handler] start_client: ~p, get error: ~p", [ClientId, Error]),
-                    {ok, 200, iot_util:json_error(-1, <<"config kafka server failed">>)}
             end;
         {error, Errors} ->
             {ok, 200, iot_util:json_error(-1, Errors)}
@@ -195,7 +140,7 @@ handle_request("POST", "/endpoint/test", _, #{<<"protocol">> := <<"kafka">>, <<"
 handle_request("POST", "/endpoint/publish_metric", _, #{<<"route_key">> := RouteKey, <<"metric">> := Metric0}) when is_binary(RouteKey) ->
     if
         is_map(Metric0) orelse is_list(Metric0) ->
-            Metric = jiffy:encode(Metric0, [force_utf8]),
+            Metric = iolist_to_binary(json:encode(Metric0)),
             endpoint_subscription:publish(RouteKey, Metric),
             {ok, 200, iot_util:json_data(<<"ok">>)};
         is_binary(Metric0) ->

apps/iot/src/http/event_stream_handler.erl

@@ -0,0 +1,73 @@
+%%%-------------------------------------------------------------------
+%%% @author anlicheng
+%%% @copyright (C) 2025, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 08. 5月 2025 13:00
+%%%-------------------------------------------------------------------
+-module(event_stream_handler).
+-author("anlicheng").
+
+%% API
+-export([init/2]).
+
+init(Req0, Opts) ->
+    ok = iot_log:set_metadata(),
+    Method = binary_to_list(cowboy_req:method(Req0)),
+    Path = binary_to_list(cowboy_req:path(Req0)),
+    GetParams0 = cowboy_req:parse_qs(Req0),
+    GetParams = maps:from_list(GetParams0),
+
+    logger:debug("method: ~p, path: ~p,  get: ~p", [Method, Path, GetParams]),
+    case parse_stream_params(GetParams) of
+        {ok, UUID, TaskId} ->
+            Req1 = cowboy_req:stream_reply(200, #{
+                <<"Content-Type">> => <<"text/event-stream">>,
+                <<"Cache-Control">> => <<"no-cache">>,
+                <<"Connection">> => <<"keep-alive">>
+            }, Req0),
+
+            ok = iot_container_task:subscribe(UUID, TaskId, self()),
+            receiver_events(TaskId, Req1),
+
+            {ok, Req1, Opts};
+        {error, Reason} ->
+            Req1 = cowboy_req:reply(400, #{
+                <<"Content-Type">> => <<"application/json">>
+            }, iot_util:json_error(400, Reason), Req0),
+
+            {ok, Req1, Opts}
+    end.
+
+receiver_events(TaskId, Req) ->
+    receive
+        {stream_data, TaskId, Type, Stream} ->
+            Data = iolist_to_binary(json:encode(#{<<"type">> => Type, <<"stream">> => Stream})),
+            Body = iolist_to_binary([<<"event: message\n">>, <<"data: ", Data/binary, "\n">>, <<"\n">>]),
+            ok = cowboy_req:stream_body(Body, nofin, Req),
+
+            receiver_events(TaskId, Req);
+        {stream_close, TaskId, Reason} ->
+            CloseFrame = iolist_to_binary([<<"event: close\n">>, <<"data: ", Reason/binary, "\n">>, <<"\n">>]),
+            ok = cowboy_req:stream_body(CloseFrame, fin, Req)
+    after 30000 ->
+            ok = cowboy_req:stream_body(<<": heartbeat\n\n">>, nofin, Req),
+            receiver_events(TaskId, Req)
+    end.
+
+-spec parse_stream_params(map()) -> {ok, binary(), integer()} | {error, binary()}.
+parse_stream_params(#{<<"uuid">> := UUID, <<"task_id">> := TaskId0}) when is_binary(UUID), is_binary(TaskId0) ->
+    try binary_to_integer(TaskId0) of
+        TaskId when TaskId >= 0 ->
+            {ok, UUID, TaskId};
+        _ ->
+            {error, <<"task_id must be non-negative">>}
+    catch
+        error:badarg ->
+            {error, <<"task_id must be integer">>}
+    end;
+parse_stream_params(#{<<"task_id">> := _TaskId0}) ->
+    {error, <<"uuid required">>};
+parse_stream_params(_) ->
+    {error, <<"uuid and task_id required">>}.

apps/iot/src/http/host_handler.erl

@@ -18,7 +18,7 @@
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 
 handle_request("GET", "/host/metric", #{<<"uuid">> := UUID}, _) ->
-    lager:debug("[host_handler] get host metric uuid is: ~p", [UUID]),
+    logger:debug("[host_handler] get host metric uuid is: ~p", [UUID]),
     case iot_host:get_pid(UUID) of
         undefined ->
             {ok, 200, iot_util:json_error(404, <<"host not found">>)};
@@ -56,26 +56,22 @@ handle_request("POST", "/host/delete", _, #{<<"uuid">> := UUID}) when is_binary(
 handle_request("POST", "/host/activate", _, #{<<"uuid">> := UUID, <<"auth">> := true}) when is_binary(UUID) ->
     case iot_host_sup:ensured_host_started(UUID) of
         {error, Reason} ->
-            lager:debug("[host_handler] activate host_id: ~p, failed with reason: ~p", [UUID, Reason]),
+            logger:debug("[host_handler] activate host_id: ~p, failed with reason: ~p", [UUID, Reason]),
             {ok, 200, iot_util:json_error(400, <<"host not found">>)};
         {ok, Pid} when is_pid(Pid) ->
-            lager:debug("[host_handler] activate host_id: ~p, start", [UUID]),
+            logger:debug("[host_handler] activate host_id: ~p, start", [UUID]),
-            ok = iot_host:activate(Pid, true),
+            auth_response(iot_host:activate(Pid, true))
-
-            {ok, 200, iot_util:json_data(<<"success">>)}
     end;
 
 %% 处理主机的授权的 取消激活
 handle_request("POST", "/host/activate", _, #{<<"uuid">> := UUID, <<"auth">> := false}) when is_binary(UUID) ->
     case iot_host_sup:ensured_host_started(UUID) of
         {error, Reason} ->
-            lager:debug("[host_handler] activate host_id: ~p, failed with reason: ~p", [UUID, Reason]),
+            logger:debug("[host_handler] activate host_id: ~p, failed with reason: ~p", [UUID, Reason]),
             {ok, 200, iot_util:json_error(400, <<"host not found">>)};
         {ok, Pid} when is_pid(Pid) ->
-            lager:debug("[host_handler] activate host_id: ~p, start", [UUID]),
+            logger:debug("[host_handler] activate host_id: ~p, start", [UUID]),
-            ok = iot_host:activate(Pid, false),
+            auth_response(iot_host:activate(Pid, false))
-
-            {ok, 200, iot_util:json_data(<<"success">>)}
     end;
 
 %% 主机事件发布
@@ -85,11 +81,15 @@ handle_request("POST", "/host/pub", _, #{<<"uuid">> := UUID, <<"topic">> := Topi
     Qos = case Qos0 > 0 of true -> 1; false -> 0 end,
     case iot_host_sup:ensured_host_started(UUID) of
         {error, Reason} ->
-            lager:debug("[host_handler] pub host_id: ~p, topic: ~p, failed with reason: ~p", [UUID, Topic, Reason]),
+            logger:debug("[host_handler] pub host_id: ~p, topic: ~p, failed with reason: ~p", [UUID, Topic, Reason]),
             {ok, 200, iot_util:json_error(400, <<"host not found">>)};
         {ok, Pid} when is_pid(Pid) ->
-            ok = iot_host:pub(Pid, Topic, Qos, Content),
+            case iot_host:pub(Pid, Topic, Qos, Content) of
-            {ok, 200, iot_util:json_data(<<"success">>)}
+                ok ->
+                    {ok, 200, iot_util:json_data(<<"success">>)};
+                {error, Reason} ->
+                    {ok, 200, iot_util:json_error(400, Reason)}
+            end
     end;
 
 handle_request(_, Path, _, _) ->
@@ -99,3 +99,28 @@ handle_request(_, Path, _, _) ->
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 %% helper methods
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+auth_response(ok) ->
+    {ok, 200, iot_util:json_data(<<"success">>)};
+auth_response({error, Reason}) ->
+    {ok, auth_error_status(Reason), iot_util:json_error(auth_error_status(Reason), reason_to_binary(Reason))}.
+
+auth_error_status(timeout) ->
+    504;
+auth_error_status(<<"timeout">>) ->
+    504;
+auth_error_status(_) ->
+    400.
+
+reason_to_binary(Reason) when is_binary(Reason) ->
+    Reason;
+reason_to_binary(timeout) ->
+    <<"timeout">>;
+reason_to_binary(invalid_response) ->
+    <<"invalid response">>;
+reason_to_binary({failed, Reason}) ->
+    reason_to_binary(Reason);
+reason_to_binary({channel_closed, Reason}) ->
+    <<"channel closed: ", (reason_to_binary(Reason))/binary>>;
+reason_to_binary(Reason) ->
+    unicode:characters_to_binary(io_lib:format("~p", [Reason])).

apps/iot/src/http/http_protocol.erl

@@ -12,16 +12,28 @@
 %% API
 -export([init/2]).
 
+-define(MAX_BODY_BYTES, 10 * 1024 * 1024).
+
 init(Req0, Opts = [Mod|_]) ->
+    ok = iot_log:set_metadata(),
     Method = binary_to_list(cowboy_req:method(Req0)),
     Path = binary_to_list(cowboy_req:path(Req0)),
     GetParams0 = cowboy_req:parse_qs(Req0),
     GetParams = maps:from_list(GetParams0),
-    {ok, PostParams, Req1} = parse_body(Req0),
+    case parse_body(Req0) of
+        {ok, PostParams, Req1} ->
+            handle_request(Mod, Method, Path, GetParams, PostParams, Req1, Opts);
+        {error, StatusCode, Resp, Req1} ->
+            Req2 = cowboy_req:reply(StatusCode, #{
+                <<"Content-Type">> => <<"application/json;charset=utf-8">>
+            }, Resp, Req1),
+            {ok, Req2, Opts}
+    end.
 
+handle_request(Mod, Method, Path, GetParams, PostParams, Req1, Opts) ->
     try Mod:handle_request(Method, Path, GetParams, PostParams) of
         {ok, StatusCode, Resp} ->
-            lager:debug("[http_protocol] request path: ~p, get_params: ~p, post_params: ~p, response: ~ts",
+            logger:debug("[http_protocol] request path: ~p, get_params: ~p, post_params: ~p, response: ~ts",
                 [Path, GetParams, PostParams, Resp]),
             AcceptEncoding = cowboy_req:header(<<"accept-encoding">>, Req1, <<>>),
             Req2 = case iolist_size(Resp) >= 1024 andalso supported_gzip(AcceptEncoding) of
@@ -45,7 +57,7 @@ init(Req0, Opts = [Mod|_]) ->
             }, ErrResp, Req1),
             {ok, Req2, Opts};
         _:Error:Stack ->
-            lager:warning("[http_handler] get error: ~p, stack: ~p", [Error, Stack]),
+            logger:warning("[http_handler] get error: ~p, stack: ~p", [Error, Stack]),
             Req2 = cowboy_req:reply(500, #{
                 <<"Content-Type">> => <<"text/html;charset=utf-8">>
             }, <<"Internal Server Error">>, Req1),
@@ -60,17 +72,18 @@ parse_body(Req0) ->
     ContentType = cowboy_req:header(<<"content-type">>, Req0),
     case ContentType of
         <<"application/json", _/binary>> ->
-            {ok, Body, Req1} = read_body(Req0),
+            case read_body(Req0) of
-            case Body =/= <<"">> of
+                {ok, Body, Req1} ->
-                 true ->
+                    decode_json_body(Body, Req1);
-                    {ok, catch jiffy:decode(Body, [return_maps]), Req1};
+                {error, payload_too_large, Req1} ->
-                false ->
+                    {error, 413, iot_util:json_error(413, <<"payload too large">>), Req1}
-                    {ok, #{}, Req1}
             end;
         <<"application/x-www-form-urlencoded">> ->
-            {ok, PostParams0, Req1} = cowboy_req:read_urlencoded_body(Req0),
+            case cowboy_req:read_urlencoded_body(Req0) of
-            PostParams = maps:from_list(PostParams0),
+                {ok, PostParams0, Req1} ->
-            {ok, PostParams, Req1};
+                    PostParams = maps:from_list(PostParams0),
+                    {ok, PostParams, Req1}
+            end;
         _ ->
             {ok, #{}, Req0}
     end.
@@ -81,7 +94,33 @@ read_body(Req) ->
 read_body(Req, AccData) ->
     case cowboy_req:read_body(Req) of
         {ok, Data, Req1} ->
-            {ok, <<AccData/binary, Data/binary>>, Req1};
+            NAccData = <<AccData/binary, Data/binary>>,
+            case byte_size(NAccData) =< ?MAX_BODY_BYTES of
+                true ->
+                    {ok, NAccData, Req1};
+                false ->
+                    {error, payload_too_large, Req1}
+            end;
         {more, Data, Req1} ->
-            read_body(Req1, <<AccData/binary, Data/binary>>)
+            NAccData = <<AccData/binary, Data/binary>>,
+            case byte_size(NAccData) =< ?MAX_BODY_BYTES of
+                true ->
+                    read_body(Req1, NAccData);
+                false ->
+                    {error, payload_too_large, Req1}
+            end
+    end.
+
+decode_json_body(<<"">>, Req) ->
+    {ok, #{}, Req};
+decode_json_body(Body, Req) ->
+    case catch json:decode(Body) of
+        {'EXIT', _} ->
+            {error, 400, iot_util:json_error(400, <<"invalid json">>), Req};
+        {error, _} ->
+            {error, 400, iot_util:json_error(400, <<"invalid json">>), Req};
+        Decoded when is_map(Decoded); is_list(Decoded) ->
+            {ok, Decoded, Req};
+        _ ->
+            {error, 400, iot_util:json_error(400, <<"invalid json">>), Req}
     end.

apps/iot/src/http_handlers/container_handler.erl

@@ -1,309 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author licheng5
-%%% @copyright (C) 2020, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 26. 4月 2020 3:36 下午
-%%%-------------------------------------------------------------------
--module(container_handler).
--author("licheng5").
--include("iot.hrl").
-
--define(REQ_TIMEOUT, 10000).
-
-%% API
--export([handle_request/4]).
-
-handle_request("GET", "/container/get_all", #{<<"uuid">> := UUID}, _) when is_binary(UUID) ->
-    %% 检查ConfigJson是否是合法的json字符串
-    case iot_host:get_pid(UUID) of
-        undefined ->
-            {ok, 200, iot_util:json_error(-1, <<"host not found">>)};
-        Pid when is_pid(Pid) ->
-            case iot_host:get_containers(Pid) of
-                {ok, Ref} ->
-                    case iot_host:await_reply(Ref, ?REQ_TIMEOUT) of
-                        {ok, Result} ->
-                            {ok, 200, iot_util:json_data(Result)};
-                        {error, Reason} ->
-                            {ok, 200, iot_util:json_error(-1, Reason)}
-                    end;
-                {error, Reason} when is_binary(Reason) ->
-                    {ok, 200, iot_util:json_error(-1, Reason)}
-            end
-    end;
-
-%% 下发config.json, 微服务接受后，保存服务配置
-handle_request("POST", "/container/push_config", _,
-    #{<<"uuid">> := UUID, <<"container_name">> := ContainerName, <<"config">> := Config, <<"timeout">> := Timeout0})
-    when is_binary(UUID), is_binary(ContainerName), is_binary(Config), is_integer(Timeout0) ->
-
-    %% 检查ConfigJson是否是合法的json字符串
-    true = iot_util:is_json(Config),
-    case iot_host:get_pid(UUID) of
-        undefined ->
-            {ok, 200, iot_util:json_error(-1, <<"host not found">>)};
-        Pid when is_pid(Pid) ->
-            Timeout = Timeout0 * 1000,
-            case iot_host:config_container(Pid, ContainerName, Config) of
-                {ok, Ref} ->
-                    case iot_host:await_reply(Ref, Timeout) of
-                        {ok, Result} ->
-                            {ok, 200, iot_util:json_data(Result)};
-                        {error, Reason} ->
-                            {ok, 200, iot_util:json_error(-1, Reason)}
-                    end;
-                {error, Reason} when is_binary(Reason) ->
-                    {ok, 200, iot_util:json_error(-1, Reason)}
-            end
-    end;
-
-%% 部署微服务
-handle_request("POST", "/container/deploy", _, #{<<"uuid">> := UUID, <<"task_id">> := TaskId, <<"config">> := Config})
-    when is_binary(UUID), is_integer(TaskId), is_map(Config) ->
-
-    case validate_config(Config) of
-        ok ->
-            case iot_host:get_pid(UUID) of
-                undefined ->
-                    {ok, 200, iot_util:json_error(404, <<"host not found">>)};
-                Pid when is_pid(Pid) ->
-                    case iot_host:deploy_container(Pid, TaskId, Config) of
-                        {ok, Ref} ->
-                            case iot_host:await_reply(Ref, ?REQ_TIMEOUT) of
-                                {ok, Result} ->
-                                    {ok, 200, iot_util:json_data(Result)};
-                                {error, Reason} ->
-                                    {ok, 200, iot_util:json_error(400, Reason)}
-                            end;
-                        {error, Reason} when is_binary(Reason) ->
-                            {ok, 200, iot_util:json_error(400, Reason)}
-                    end
-            end;
-        {error, Errors} ->
-            Reason = iolist_to_binary(lists:join(<<"|||">>, Errors)),
-            {ok, 200, iot_util:json_error(400, Reason)}
-    end;
-
-%% 启动服务
-handle_request("POST", "/container/start", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
-    case iot_host:get_pid(UUID) of
-        undefined ->
-            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
-        Pid when is_pid(Pid) ->
-            case iot_host:start_container(Pid, ContainerName) of
-                {ok, Ref} ->
-                    case iot_host:await_reply(Ref, ?REQ_TIMEOUT) of
-                        {ok, Result} ->
-                            {ok, 200, iot_util:json_data(Result)};
-                        {error, Reason} ->
-                            {ok, 200, iot_util:json_error(400, Reason)}
-                    end;
-                {error, Reason} when is_binary(Reason) ->
-                    {ok, 200, iot_util:json_error(400, Reason)}
-            end
-    end;
-
-%% 停止服务
-handle_request("POST", "/container/stop", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
-    case iot_host:get_pid(UUID) of
-        undefined ->
-            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
-        Pid when is_pid(Pid) ->
-            case iot_host:stop_container(Pid, ContainerName) of
-                {ok, Ref} ->
-                    case iot_host:await_reply(Ref, ?REQ_TIMEOUT) of
-                        {ok, Result} ->
-                            {ok, 200, iot_util:json_data(Result)};
-                        {error, Reason} ->
-                            {ok, 200, iot_util:json_error(400, Reason)}
-                    end;
-                {error, Reason} when is_binary(Reason) ->
-                    {ok, 200, iot_util:json_error(400, Reason)}
-            end
-    end;
-
-handle_request("POST", "/container/kill", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
-    case iot_host:get_pid(UUID) of
-        undefined ->
-            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
-        Pid when is_pid(Pid) ->
-            case iot_host:kill_container(Pid, ContainerName) of
-                {ok, Ref} ->
-                    case iot_host:await_reply(Ref, ?REQ_TIMEOUT) of
-                        {ok, Result} ->
-                            {ok, 200, iot_util:json_data(Result)};
-                        {error, Reason} ->
-                            {ok, 200, iot_util:json_error(400, Reason)}
-                    end;
-                {error, Reason} when is_binary(Reason) ->
-                    {ok, 200, iot_util:json_error(400, Reason)}
-            end
-    end;
-
-%% 删除容器
-handle_request("POST", "/container/remove", _, #{<<"uuid">> := UUID, <<"container_name">> := ContainerName}) when is_binary(UUID), is_binary(ContainerName) ->
-    case iot_host:get_pid(UUID) of
-        undefined ->
-            {ok, 200, iot_util:json_error(404, <<"host not found">>)};
-        Pid when is_pid(Pid) ->
-            case iot_host:remove_container(Pid, ContainerName) of
-                {ok, Ref} ->
-                    case iot_host:await_reply(Ref, ?REQ_TIMEOUT) of
-                        {ok, Result} ->
-                            {ok, 200, iot_util:json_data(Result)};
-                        {error, Reason} ->
-                            {ok, 200, iot_util:json_error(400, Reason)}
-                    end;
-                {error, Reason} when is_binary(Reason) ->
-                    {ok, 200, iot_util:json_error(400, Reason)}
-            end
-    end;
-
-handle_request(_, Path, _, _) ->
-    Path1 = list_to_binary(Path),
-    {ok, 200, iot_util:json_error(-1, <<"url: ", Path1/binary, " not found">>)}.
-
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%% helper methods
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
-validate_config(Config) when is_map(Config) ->
-    %% 必选参数
-    Required = [
-        {<<"image">>, binary},
-        {<<"container_name">>, binary},
-        {<<"command">>, {list, binary}},
-        {<<"restart">>, binary}
-    ],
-
-    %% 可选参数（附带默认值）
-    Optional = [
-        {<<"privileged">>, boolean},
-        {<<"envs">>, {list, binary}},
-        {<<"ports">>, {list, binary}},
-        {<<"expose">>, {list, binary}},
-        {<<"volumes">>, {list, binary}},
-        {<<"networks">>, {list, binary}},
-        {<<"labels">>, {map, {binary, binary}}},
-        {<<"user">>, binary},
-        {<<"working_dir">>, binary},
-        {<<"hostname">>, binary},
-        {<<"cap_add">>, {list, binary}},
-        {<<"cap_drop">>, {list, binary}},
-        {<<"devices">>, {list, binary}},
-        {<<"mem_limit">>, binary},
-        {<<"mem_reservation">>, binary},
-        {<<"cpu_shares">>, integer},
-        {<<"cpus">>, number},
-        {<<"ulimits">>, {map, {binary, binary}}},
-        {<<"sysctls">>, {map, {binary, binary}}},
-        {<<"tmpfs">>, {list, binary}},
-        {<<"extra_hosts">>, {list, binary}},
-        {<<"healthcheck">>, {map, {binary, any}}}
-    ],
-
-    Errors1 = check_required(Config, Required),
-    Errors2 = check_optional(Config, Optional),
-
-    Errors = Errors1 ++ Errors2,
-    case Errors of
-        [] ->
-            ok;
-        _  ->
-            {error, lists:map(fun erlang:iolist_to_binary/1, Errors)}
-    end.
-
-%%------------------------------------------------------------------------------
-%% 校验必选项
-%%------------------------------------------------------------------------------
-check_required(Config, Fields) ->
-    lists:foldl(
-        fun({Key, Type}, ErrAcc) ->
-            case maps:get(Key, Config, undefined) of
-                undefined ->
-                    [io_lib:format("miss requied parameter: ~p", [Key]) | ErrAcc];
-                Value ->
-                    case check_type(Value, Type) of
-                        true ->
-                            ErrAcc;
-                        false ->
-                            [io_lib:format("required parameter: ~p, type must be: ~p", [Key, type_name(Type)]) | ErrAcc]
-                    end
-            end
-        end,
-        [], Fields).
-
-%%------------------------------------------------------------------------------
-%% 校验可选项（支持默认值填充）
-%%------------------------------------------------------------------------------
-check_optional(Config, Fields) ->
-    lists:foldl(
-        fun({Key, Type}, ErrAcc) ->
-            case maps:get(Key, Config, undefined) of
-                undefined ->
-                    ErrAcc;
-                Value ->
-                    case check_type(Value, Type) of
-                        true ->
-                            ErrAcc;
-                        false ->
-                            [io_lib:format("optional parameter: ~p, type must be: ~p", [Key, type_name(Type)]) | ErrAcc]
-                    end
-            end
-        end,
-        [], Fields).
-
-%%------------------------------------------------------------------------------
-%% 类型检查辅助函数（binary版）
-%%------------------------------------------------------------------------------
--spec type_name(tuple() | atom()) -> binary().
-type_name(binary) ->
-    <<"string">>;
-type_name(integer) ->
-    <<"integer">>;
-type_name(number) ->
-    <<"number">>;
-type_name(list) ->
-    <<"list">>;
-type_name({list, binary}) ->
-    <<"list of string">>;
-type_name({list, number}) ->
-    <<"list of number">>;
-type_name({list, integer}) ->
-    <<"list of integer">>;
-type_name(map) ->
-    <<"map">>;
-type_name({map, {binary, binary}}) ->
-    <<"map of string:string">>;
-type_name({map, {binary, any}}) ->
-    <<"map of string:any">>;
-type_name(boolean) ->
-    <<"boolean">>.
-
--spec check_type(Value :: any(), any()) -> boolean().
-check_type(Value, binary) ->
-    is_binary(Value);
-check_type(Value, integer) ->
-    is_integer(Value);
-check_type(Value, number) ->
-    is_number(Value);
-check_type(Value, list) when is_list(Value) ->
-    true;
-check_type(Value, {list, binary}) when is_list(Value) ->
-    lists:all(fun(E) -> is_binary(E) end, Value);
-check_type(Value, {list, number}) when is_list(Value) ->
-    lists:all(fun(E) -> is_number(E) end, Value);
-check_type(Value, {list, integer}) when is_list(Value) ->
-    lists:all(fun(E) -> is_integer(E) end, Value);
-check_type(Value, map) when is_map(Value) ->
-    true;
-check_type(Value, {map, {binary, binary}}) when is_map(Value) ->
-    lists:all(fun({K, V}) -> is_binary(K) andalso is_binary(V) end, maps:to_list(Value));
-check_type(Value, {map, {binary, any}}) when is_map(Value) ->
-    lists:all(fun({K, _}) -> is_binary(K) end, maps:to_list(Value));
-check_type(Value, boolean) ->
-    is_boolean(Value);
-check_type(_, _) ->
-    false.

apps/iot/src/http_handlers/event_stream_handler.erl

@@ -1,47 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author anlicheng
-%%% @copyright (C) 2025, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 08. 5月 2025 13:00
-%%%-------------------------------------------------------------------
--module(event_stream_handler).
--author("anlicheng").
-
-%% API
--export([init/2]).
-
-init(Req0, Opts) ->
-    Method = binary_to_list(cowboy_req:method(Req0)),
-    Path = binary_to_list(cowboy_req:path(Req0)),
-    GetParams0 = cowboy_req:parse_qs(Req0),
-    GetParams = maps:from_list(GetParams0),
-
-    #{<<"task_id">> := TaskId0} = GetParams,
-    TaskId = binary_to_integer(TaskId0),
-
-    lager:debug("method: ~p, path: ~p,  get: ~p", [Method, Path, GetParams]),
-    Req1 = cowboy_req:stream_reply(200, #{
-        <<"Content-Type">> => <<"text/event-stream">>,
-        <<"Cache-Control">> => <<"no-cache">>,
-        <<"Connection">> => <<"keep-alive">>
-    }, Req0),
-
-    ok = iot_event_stream_observer:add_listener(self(), TaskId),
-    receiver_events(TaskId, Req1),
-
-    {ok, Req1, Opts}.
-
-receiver_events(TaskId, Req) ->
-    receive
-        {stream_data, TaskId, Type, Stream} ->
-            Data = jiffy:encode(#{<<"type">> => Type, <<"stream">> => Stream}, [force_utf8]),
-            Body = iolist_to_binary([<<"event: message\n">>, <<"data: ", Data/binary, "\n">>, <<"\n">>]),
-            ok = cowboy_req:stream_body(Body, nofin, Req),
-
-            receiver_events(TaskId, Req);
-        {stream_close, TaskId, Reason} ->
-            CloseFrame = iolist_to_binary([<<"event: close\n">>, <<"data: ", Reason/binary, "\n">>, <<"\n">>]),
-            ok = cowboy_req:stream_body(CloseFrame, fin, Req)
-    end.

apps/iot/src/iot.app.src

@@ -6,17 +6,14 @@
   {applications,
    [
     sync,
+    endpoint,
     emqtt,
     eredis,
     ranch,
     cowboy,
-    lager,
-    jiffy,
     brod,
-    parse_trans,
     hackney,
     poolboy,
-    mysql,
     gproc,
     % gpb,
     mnesia,

apps/iot/src/iot_app.erl

@@ -9,34 +9,41 @@
 -export([start/2, stop/1]).
 
 start(_StartType, _StartArgs) ->
+    ok = iot_log:set_metadata(),
     io:setopts([{encoding, unicode}]),
     %% 加速内存的回收
     erlang:system_flag(fullsweep_after, 16),
-    %% 启动mnesia数据库
-    start_mnesia(),
 
-    %% 启动http服务
+    iot_mnesia:safe_start(),
+
+    %% 启动http服务。当前 supervisor 初始化依赖本机 simulator API，因此保留此顺序。
     start_http_server(),
 
-    %% 启动tcp服务
+    %% 启动endpoints的相关依赖
-    start_tcp_server(),
+    start_endpoints(),
+
+    %% 启动ssl服务
+    start_ssl_server(),
 
     iot_sup:start_link().
 
 stop(_State) ->
+    stop_started_services(),
     ok.
 
 %% internal functions
 
-%% 启动内存数据库
+start_endpoints() ->
-start_mnesia() ->
+    EndpointInfos = iot_api_client:get_all_endpoints(),
-    ok = ensure_mnesia_schema(),
+    Endpoints = lists:filtermap(fun(EndpointInfo) ->
-    %% 启动数据库
+        case endpoint:endpoint_record(EndpointInfo) of
-    ok = mnesia:start(),
+            error ->
-    Tables = mnesia:system_info(tables),
+                false;
-    lager:debug("[iot_app] tables: ~p", [Tables]),
+            {ok, Endpoint} ->
-    %% 创建数据库表
+                {true, Endpoint}
-    ok.
+        end
+                                end, EndpointInfos),
+    ok = endpoint_adapter_sup:start_endpoints(Endpoints).
 
 start_http_server() ->
     {ok, Props} = application:get_env(iot, http_server),
@@ -50,6 +57,7 @@ start_http_server() ->
             {"/host/[...]", http_protocol, [host_handler]},
             {"/container/[...]", http_protocol, [container_handler]},
             {"/endpoint/[...]", http_protocol, [endpoint_handler]},
+            {"/simulator/[...]", http_protocol, [simulator_api_handler]},
             {"/event_stream", event_stream_handler, []}
         ]}
     ]),
@@ -65,41 +73,36 @@ start_http_server() ->
     },
     {ok, Pid} = cowboy:start_clear(http_listener, TransOpts, #{env => #{dispatch => Dispatcher}}),
 
-    lager:debug("[http_server] the http server start at: ~p, pid is: ~p", [Port, Pid]).
+    logger:debug("[http_server] the http server start at: ~p, pid is: ~p", [Port, Pid]).
 
-%% 启动tcp服务
+%% 启动ssl服务
-start_tcp_server() ->
+start_ssl_server() ->
-    {ok, Props} = application:get_env(iot, tcp_server),
+    {ok, Props} = application:get_env(iot, ssl_server),
     Acceptors = proplists:get_value(acceptors, Props, 50),
     MaxConnections = proplists:get_value(max_connections, Props, 10240),
     Backlog = proplists:get_value(backlog, Props, 1024),
     Port = proplists:get_value(port, Props),
+    PrivDir = code:priv_dir(iot),
+    CertFile = filename:join([PrivDir, "ssl", "server.crt"]),
+    KeyFile = filename:join([PrivDir, "ssl", "server.key"]),
 
     TransOpts = #{
         max_connections => MaxConnections,
         num_acceptors => Acceptors,
         shutdown => brutal_kill,
         socket_opts => [
-            {nodelay, false},
+            {nodelay, true},
             {backlog, Backlog},
-            {port, Port}
+            {port, Port},
+            {certfile, CertFile},
+            {keyfile, KeyFile}
         ]
     },
-    {ok, _} = ranch:start_listener(tcp_server, ranch_tcp, TransOpts, tcp_channel, []),
+    {ok, _} = ranch:start_listener(ssl_server, ranch_ssl, TransOpts, ssl_channel, []),
-    lager:debug("[iot_app] the tcp server start at: ~p", [Port]).
+    logger:debug("[iot_app] the ssl server start at: ~p", [Port]).
 
--spec ensure_mnesia_schema() -> any().
+stop_started_services() ->
-ensure_mnesia_schema() ->
+    _ = cowboy:stop_listener(http_listener),
-    case mnesia:system_info(use_dir) of
+    _ = ranch:stop_listener(ssl_server),
-        true ->
+    _ = iot_mnesia:stop(),
-            ok;
+    ok.
-        false ->
-            mnesia:stop(),
-            case mnesia:create_schema([node()]) of
-                ok -> ok;
-                {error, {_, {already_exists, _}}} -> ok;
-                Error ->
-                    lager:debug("[iot_app] create mnesia schema failed with error: ~p", [Error]),
-                    throw({init_schema, Error})
-            end
-    end.

apps/iot/src/iot_auth.erl

@@ -1,33 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%% 授权校验模块
-%%% @end
-%%% Created : 27. 6月 2023 09:48
-%%%-------------------------------------------------------------------
--module(iot_auth).
--author("aresei").
-
-%% API
--export([check/5]).
-
-%% 检测token是否是合法值
--spec check(Username :: binary(), Token :: binary(), UUID :: binary(), Salt :: binary(), Timestamp :: integer()) -> boolean().
-check(Username, Token, UUID, Salt, Timestamp) when is_binary(Username), is_binary(Token), is_binary(UUID), is_binary(Salt), is_integer(Timestamp) ->
-    true;
-check(Username, Token, UUID, Salt, Timestamp) when is_binary(Username), is_binary(Token), is_binary(UUID), is_binary(Salt), is_integer(Timestamp) ->
-    BinTimestamp = integer_to_binary(Timestamp),
-    %% 1分钟内有效
-    case iot_util:current_time() - Timestamp =< 60 of
-        true ->
-            {ok, PreTokens} = application:get_env(iot, pre_tokens),
-            case proplists:get_value(Username, PreTokens) of
-                undefined ->
-                    false;
-                PreToken when is_binary(PreToken) ->
-                    iot_util:md5(<<Salt/binary, "!", PreToken/binary, "!", UUID/binary, "!", BinTimestamp/binary>>) =:= Token
-            end;
-        false ->
-            false
-    end.

apps/iot/src/iot_host.erl

@@ -1,390 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 22. 9月 2023 16:38
-%%%-------------------------------------------------------------------
--module(iot_host).
--author("aresei").
--include("iot.hrl").
--include("message.hrl").
-
--behaviour(gen_statem).
-
-%% 心跳包检测时间间隔, 15分钟检测一次
--define(HEARTBEAT_INTERVAL, 900 * 1000).
-
-%% 状态
--define(STATE_DENIED, denied).
--define(STATE_ACTIVATED, activated).
-
-%% API
--export([start_link/2, get_name/1, get_alias_name/1, get_pid/1, handle/2, activate/2]).
--export([get_metric/1, get_status/1, kill/1]).
-%% 通讯相关
--export([pub/4, attach_channel/2, command/3]).
--export([deploy_container/3, start_container/2, stop_container/2, remove_container/2, kill_container/2, config_container/3, get_containers/1, await_reply/2]).
--export([heartbeat/1]).
-
-%% gen_statem callbacks
--export([init/1, handle_event/4, terminate/3, code_change/4, callback_mode/0]).
-
--record(state, {
-    host_id :: integer(),
-    %% 从数据库里面读取到的数据
-    uuid :: binary(),
-    has_session = false :: boolean(),
-    %% 心跳计数器
-    heartbeat_counter = 0 :: integer(),
-    %% websocket相关
-    channel_pid :: undefined | pid(),
-    %% 主机的相关信息
-    metrics = #{} :: map()
-}).
-
-%%%===================================================================
-%%% API
-%%%===================================================================
--spec get_pid(UUID :: binary()) -> undefined | pid().
-get_pid(UUID) when is_binary(UUID) ->
-    Name = get_name(UUID),
-    whereis(Name).
-
--spec get_name(UUID :: binary()) -> atom().
-get_name(UUID) when is_binary(UUID) ->
-    binary_to_atom(<<"iot_host:", UUID/binary>>).
-
--spec get_alias_name(HostId :: integer()) -> atom().
-get_alias_name(HostId0) when is_integer(HostId0) ->
-    HostId = integer_to_binary(HostId0),
-    binary_to_atom(<<"iot_host_id:", HostId/binary>>).
-
--spec kill(UUID :: binary()) -> no_return().
-kill(UUID) when is_binary(UUID) ->
-    case whereis(get_name(UUID)) of
-        undefined ->
-            ok;
-        Pid ->
-            exit(Pid, kill)
-    end.
-
-%% 处理消息
--spec handle(Pid :: pid(), Packet :: {atom(), any()}) -> no_return().
-handle(Pid, Packet) when is_pid(Pid) ->
-    gen_statem:cast(Pid, {handle, Packet}).
-
--spec get_status(Pid :: pid()) -> {ok, Status :: map()}.
-get_status(Pid) when is_pid(Pid) ->
-    gen_statem:call(Pid, get_status).
-
-%% 激活主机, true 表示激活； false表示关闭激活
--spec activate(Pid :: pid(), Auth :: boolean()) -> ok.
-activate(Pid, Auth) when is_pid(Pid), is_boolean(Auth) ->
-    gen_statem:call(Pid, {activate, Auth}).
-
--spec get_metric(Pid :: pid()) -> {ok, MetricInfo :: map()}.
-get_metric(Pid) when is_pid(Pid) ->
-    gen_statem:call(Pid, get_metric).
-
--spec attach_channel(pid(), pid()) -> ok | {error, Reason :: binary()} | {denied, Reason :: binary()}.
-attach_channel(Pid, ChannelPid) when is_pid(Pid), is_pid(ChannelPid) ->
-    gen_statem:call(Pid, {attach_channel, ChannelPid}).
-
--spec get_containers(Pid :: pid()) -> {ok, Ref :: reference()} | {error, Reason :: any()}.
-get_containers(Pid) when is_pid(Pid) ->
-    Request = #jsonrpc_request{method = <<"get_containers">>, params = #{}},
-    EncConfigBin = message_codec:encode(?MESSAGE_JSONRPC_REQUEST, Request),
-    gen_statem:call(Pid, {jsonrpc_call, self(), EncConfigBin}).
-
--spec config_container(Pid :: pid(), ContainerName :: binary(), ConfigJson :: binary()) -> {ok, Ref :: reference()} | {error, Reason :: any()}.
-config_container(Pid, ContainerName, ConfigJson) when is_pid(Pid), is_binary(ContainerName), is_binary(ConfigJson) ->
-    Request = #jsonrpc_request{method = <<"config_container">>, params = #{<<"container_name">> => ContainerName, <<"config">> => ConfigJson}},
-    EncConfigBin = message_codec:encode(?MESSAGE_JSONRPC_REQUEST, Request),
-    gen_statem:call(Pid, {jsonrpc_call, self(), EncConfigBin}).
-
--spec deploy_container(Pid :: pid(), TaskId :: integer(), Config :: map()) -> {ok, Ref :: reference()} | {error, Reason :: any()}.
-deploy_container(Pid, TaskId, Config) when is_pid(Pid), is_integer(TaskId), is_map(Config) ->
-    Request = #jsonrpc_request{method = <<"deploy">>, params = #{<<"task_id">> => TaskId, <<"config">> => Config}},
-    EncDeployBin = message_codec:encode(?MESSAGE_JSONRPC_REQUEST, Request),
-    gen_statem:call(Pid, {jsonrpc_call, self(), EncDeployBin}).
-
--spec start_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: reference()} | {error, Reason :: any()}.
-start_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
-    Request = #jsonrpc_request{method = <<"start_container">>, params = #{<<"container_name">> => ContainerName}},
-    EncCallBin = message_codec:encode(?MESSAGE_JSONRPC_REQUEST, Request),
-    gen_statem:call(Pid, {jsonrpc_call, self(), EncCallBin}).
-
--spec stop_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: reference()} | {error, Reason :: any()}.
-stop_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
-    Request = #jsonrpc_request{method = <<"stop_container">>, params = #{<<"container_name">> => ContainerName}},
-    EncCallBin = message_codec:encode(?MESSAGE_JSONRPC_REQUEST, Request),
-    gen_statem:call(Pid, {jsonrpc_call, self(), EncCallBin}).
-
--spec kill_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: reference()} | {error, Reason :: any()}.
-kill_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
-    Request = #jsonrpc_request{method = <<"kill_container">>, params = #{<<"container_name">> => ContainerName}},
-    EncCallBin = message_codec:encode(?MESSAGE_JSONRPC_REQUEST, Request),
-    gen_statem:call(Pid, {jsonrpc_call, self(), EncCallBin}).
-
--spec remove_container(Pid :: pid(), ContainerName :: binary()) -> {ok, Ref :: reference()} | {error, Reason :: any()}.
-remove_container(Pid, ContainerName) when is_pid(Pid), is_binary(ContainerName) ->
-    Request = #jsonrpc_request{method = <<"remove_container">>, params = #{<<"container_name">> => ContainerName}},
-    EncCallBin = message_codec:encode(?MESSAGE_JSONRPC_REQUEST, Request),
-    gen_statem:call(Pid, {jsonrpc_call, self(), EncCallBin}).
-
--spec await_reply(Ref :: reference(), Timeout :: integer()) -> {ok, Result :: binary()} | {error, Reason :: binary()}.
-await_reply(Ref, Timeout) when is_reference(Ref), is_integer(Timeout) ->
-    receive
-        {jsonrpc_reply, Ref, #jsonrpc_reply{result = Result, error = undefined}} ->
-            {ok, Result};
-        {jsonrpc_reply, Ref, #jsonrpc_reply{result = undefined, error = #{<<"message">> := Message}}} ->
-            {error, Message}
-    after Timeout ->
-        {error, <<"timeout">>}
-    end.
-
--spec pub(Pid :: pid(), Topic :: binary(), Qos :: integer(),  Content :: binary()) -> ok | {error, Reason :: any()}.
-pub(Pid, Topic, Qos, Content) when is_pid(Pid), is_binary(Topic), is_integer(Qos), is_binary(Content) ->
-    gen_statem:call(Pid, {pub, Topic, Qos, Content}).
-
--spec command(Pid :: pid(), CommandType :: integer(),  Command :: binary()) -> ok | {error, Reason :: any()}.
-command(Pid, CommandType, Command) when is_pid(Pid), is_integer(CommandType), is_binary(Command) ->
-    gen_statem:call(Pid, {command, CommandType, Command}).
-
--spec heartbeat(Pid :: pid()) -> no_return().
-heartbeat(undefined) ->
-    ok;
-heartbeat(Pid) when is_pid(Pid) ->
-    gen_statem:cast(Pid, heartbeat).
-
-%% @doc Creates a gen_statem process which calls Module:init/1 to
-%% initialize. To ensure a synchronized start-up procedure, this
-%% function does not return until Module:init/1 has returned.
-start_link(Name, UUID) when is_atom(Name), is_binary(UUID) ->
-    gen_statem:start_link({local, Name}, ?MODULE, [UUID], []).
-
-%%%===================================================================
-%%% gen_statem callbacks
-%%%===================================================================
-
-%% @private
-%% @doc Whenever a gen_statem is started using gen_statem:start/[3,4] or
-%% gen_statem:start_link/[3,4], this function is called by the new
-%% process to initialize.
-init([UUID]) ->
-    case iot_api:get_host_by_uuid(UUID) of
-        {ok, #{<<"id">> := HostId, <<"authorize_status">> := AuthorizeStatus}} ->
-            %% 通过host_id注册别名, 可以避免通过查询数据库获取HostPid
-            AliasName = get_alias_name(HostId),
-            global:register_name(AliasName, self()),
-
-            %% 心跳检测机制
-            erlang:start_timer(?HEARTBEAT_INTERVAL, self(), heartbeat_ticker),
-
-            StateName = case AuthorizeStatus =:= 1 of
-                            true  -> ?STATE_ACTIVATED;
-                            false -> ?STATE_DENIED
-                        end,
-            {ok, StateName, #state{host_id = HostId, uuid = UUID, has_session = false}};
-        undefined ->
-            lager:warning("[iot_host] host uuid: ~p, loaded from mysql failed", [UUID]),
-            ignore
-    end.
-
-%% @private
-%% @doc This function is called by a gen_statem when it needs to find out
-%% the callback mode of the callback module.
-callback_mode() ->
-    handle_event_function.
-
-%% @private
-%% @doc If callback_mode is handle_event_function, then whenever a
-%% gen_statem receives an event from call/2, cast/2, or as a normal
-%% process message, this function is called.
-
-handle_event({call, From}, get_metric, _, State = #state{metrics = Metrics}) ->
-    {keep_state, State, [{reply, From, {ok, Metrics}}]};
-
-%% 获取主机的状态
-handle_event({call, From}, get_status, _, State = #state{channel_pid = ChannelPid, heartbeat_counter = HeartbeatCounter, metrics = Metrics, has_session = HasSession}) ->
-    HasChannel = (ChannelPid /= undefined),
-    Reply = #{
-        <<"has_channel">> => HasChannel,
-        <<"has_session">> => HasSession,
-        <<"heartbeat_counter">> => HeartbeatCounter,
-        <<"metrics">> => Metrics
-    },
-    {keep_state, State, [{reply, From, {ok, Reply}}]};
-
-%% 只要channel存在，就负责将消息推送到边缘端主机
-handle_event({call, From}, {jsonrpc_call, ReceiverPid, RpcCall}, _, State = #state{uuid = UUID, channel_pid = ChannelPid, has_session = HasSession}) ->
-    case HasSession andalso is_pid(ChannelPid) of
-        true ->
-            %% 通过websocket发送请求
-            Ref = tcp_channel:jsonrpc_call(ChannelPid, ReceiverPid, RpcCall),
-            {keep_state, State, [{reply, From, {ok, Ref}}]};
-        false ->
-            lager:debug("[iot_host] uuid: ~p, invalid state: ~p", [UUID, state_map(State)]),
-            {keep_state, State, [{reply, From, {error, <<"主机离线，发送请求失败"/utf8>>}}]}
-    end;
-
-%% 发送指令时, pub/sub
-handle_event({call, From}, {pub, Topic, Qos, Content}, ?STATE_ACTIVATED, State = #state{uuid = UUID, channel_pid = ChannelPid, has_session = HasSession}) ->
-    case HasSession andalso is_pid(ChannelPid) of
-        true ->
-            lager:debug("[iot_host] host: ~p, publish to topic: ~p, content: ~p", [UUID, Topic, Content]),
-            %% 通过websocket发送请求
-            tcp_channel:pub(ChannelPid, Topic, Qos, Content),
-
-            {keep_state, State, [{reply, From, ok}]};
-        false ->
-            lager:debug("[iot_host] uuid: ~p, publish to topic: ~p, content: ~p, invalid state: ~p", [UUID, Topic, Content, state_map(State)]),
-            {keep_state, State, [{reply, From, {error, <<"主机离线，发送失败"/utf8>>}}]}
-    end;
-
-%% 发送指令时
-handle_event({call, From}, {command, CommandType, Command}, ?STATE_ACTIVATED, State = #state{uuid = UUID, channel_pid = ChannelPid, has_session = HasSession}) ->
-    case HasSession andalso is_pid(ChannelPid) of
-        true ->
-            lager:debug("[iot_host] host: ~p, command_type: ~p, command: ~p", [UUID, CommandType, Command]),
-            %% 通过websocket发送请求
-            tcp_channel:command(ChannelPid, CommandType, Command),
-            {keep_state, State, [{reply, From, ok}]};
-        false ->
-            lager:debug("[iot_host] host: ~p, command_type: ~p, command: ~p, invalid state: ~p", [UUID, CommandType, Command, state_map(State)]),
-            {keep_state, State, [{reply, From, {error, <<"主机离线，发送指令失败"/utf8>>}}]}
-    end;
-
-%% 激活主机
-handle_event({call, From}, {activate, true}, _, State = #state{uuid = UUID, channel_pid = ChannelPid}) ->
-    case is_pid(ChannelPid) of
-        true ->
-            lager:debug("[iot_host] uuid: ~p, activate: true", [UUID]),
-            tcp_channel:command(ChannelPid, ?COMMAND_AUTH, <<1:8>>);
-        false ->
-            lager:debug("[iot_host] uuid: ~p, activate: true, no channel", [UUID])
-    end,
-    {next_state, ?STATE_ACTIVATED, State, [{reply, From, ok}]};
-
-%% 关闭授权
-handle_event({call, From}, {activate, false}, _, State = #state{uuid = UUID, channel_pid = ChannelPid}) ->
-    case is_pid(ChannelPid) of
-        true ->
-            tcp_channel:command(ChannelPid, ?COMMAND_AUTH, <<0:8>>),
-            lager:debug("[iot_host] uuid: ~p, activate: false", [UUID]),
-            tcp_channel:stop(ChannelPid, closed);
-        false ->
-            lager:debug("[iot_host] uuid: ~p, activate: false, no channel", [UUID])
-    end,
-    {next_state, ?STATE_DENIED, State#state{channel_pid = undefined, has_session = false}, [{reply, From, ok}]};
-
-%% 绑定channel
-handle_event({call, From}, {attach_channel, ChannelPid}, StateName, State = #state{uuid = UUID, channel_pid = undefined}) ->
-    case StateName of
-        ?STATE_ACTIVATED ->
-            erlang:monitor(process, ChannelPid),
-            %% 更新主机为在线状态
-            ChangeResult = iot_api:change_host_status(UUID, ?HOST_ONLINE),
-            lager:debug("[iot_host] host_id(attach_channel) uuid: ~p, will change status, result: ~p", [UUID, ChangeResult]),
-            {keep_state, State#state{channel_pid = ChannelPid, has_session = true}, [{reply, From, ok}]};
-        %% 主机未激活
-        ?STATE_DENIED ->
-            lager:notice("[iot_host] attach_channel host_id uuid: ~p, channel: ~p, host inactivated", [UUID, ChannelPid]),
-            erlang:monitor(process, ChannelPid),
-            {keep_state, State#state{channel_pid = ChannelPid}, [{reply, From, {denied, <<"host inactivated">>}}]}
-    end;
-%% 已经绑定了channel
-handle_event({call, From}, {attach_channel, _}, _, State = #state{uuid = UUID, channel_pid = OldChannelPid}) ->
-    lager:notice("[iot_host] attach_channel host_id uuid: ~p, old channel exists: ~p", [UUID, OldChannelPid]),
-    {keep_state, State, [{reply, From, {error, <<"channel existed">>}}]};
-
-%% 数据分发
-handle_event(cast, {handle, {data, #data{route_key = RouteKey0, metric = Metric}}}, ?STATE_ACTIVATED,
-    State = #state{uuid = UUID, has_session = true}) ->
-    lager:debug("[iot_host] metric_data host: ~p, route_key: ~p, metric: ~p", [UUID, RouteKey0, Metric]),
-    RouteKey = get_route_key(RouteKey0),
-    endpoint_subscription:publish(RouteKey, Metric),
-    {keep_state, State};
-
-%% ping的数据是通过aes加密后的，因此需要在有会话的情况下才行
-handle_event(cast, {handle, {ping, Metrics}}, ?STATE_ACTIVATED, State = #state{uuid = UUID, has_session = true}) ->
-    lager:debug("[iot_host] ping host_id uuid: ~p, get ping: ~p", [UUID, Metrics]),
-    {keep_state, State#state{metrics = Metrics}};
-
-%% 心跳机制
-handle_event(cast, heartbeat, _, State = #state{heartbeat_counter = HeartbeatCounter}) ->
-    {keep_state, State#state{heartbeat_counter = HeartbeatCounter + 1}};
-
-%% 没有收到心跳包，主机下线, 设备状态不变
-handle_event(info, {timeout, _, heartbeat_ticker}, _, State = #state{uuid = UUID, heartbeat_counter = 0, channel_pid = ChannelPid}) ->
-    lager:warning("[iot_host] uuid: ~p, heartbeat lost, devices will unknown", [UUID]),
-    {ok, #{<<"status">> := Status}} = iot_api:get_host_by_uuid(UUID),
-    case Status of
-        ?HOST_NOT_JOINED ->
-            lager:debug("[iot_host] host: ~p, host_maybe_offline, host not joined, can not change to offline", [UUID]);
-        ?HOST_OFFLINE ->
-            lager:debug("[iot_host] host: ~p, host_maybe_offline, host now is offline, do nothing", [UUID]);
-        ?HOST_ONLINE ->
-            iot_api:change_host_status(UUID, ?HOST_OFFLINE)
-    end,
-
-    %% 关闭channel，主机需要重新连接，才能保存状态的一致
-    is_pid(ChannelPid) andalso tcp_channel:stop(ChannelPid, closed),
-    erlang:start_timer(?HEARTBEAT_INTERVAL, self(), heartbeat_ticker),
-
-    {keep_state, State#state{channel_pid = undefined, has_session = false, heartbeat_counter = 0}};
-
-%% 其他情况下需要重置系统计数器
-handle_event(info, {timeout, _, heartbeat_ticker}, _, State = #state{}) ->
-    erlang:start_timer(?HEARTBEAT_INTERVAL, self(), heartbeat_ticker),
-    {keep_state, State#state{heartbeat_counter = 0}};
-
-%% 当websocket断开的时候，主机的状态不一定改变；主机的状态改变通过心跳机制，会话状态需要改变
-handle_event(info, {'DOWN', _Ref, process, ChannelPid, Reason}, _, State = #state{uuid = UUID, channel_pid = ChannelPid, has_session = HasSession}) ->
-    lager:warning("[iot_host] uuid: ~p, channel: ~p, down with reason: ~p, has_session: ~p, state: ~p", [UUID, ChannelPid, Reason, HasSession, State]),
-    {keep_state, State#state{channel_pid = undefined, has_session = false}};
-
-handle_event(info, {'DOWN', _Ref, process, Pid, Reason}, _, State = #state{uuid = UUID}) ->
-    lager:debug("[iot_host] uuid: ~p, process_pid: ~p, down with reason: ~p, state: ~p", [UUID, Pid, Reason, State]),
-    {keep_state, State};
-
-handle_event(Event, Info, StateName, State = #state{uuid = UUID}) ->
-    lager:warning("[iot_host] host: ~p, event: ~p, unknown message: ~p, state_name: ~p, state: ~p", [UUID, Event, Info, StateName, state_map(State)]),
-    {keep_state, State}.
-
-%% @private
-%% @doc This function is called by a gen_statem when it is about to
-%% terminate. It should be the opposite of Module:init/1 and do any
-%% necessary cleaning up. When it returns, the gen_statem terminates with
-%% Reason. The return value is ignored.
-terminate(Reason, _StateName, _State = #state{uuid = UUID, has_session = HasSession}) ->
-    lager:debug("[iot_host] host: ~p, terminate with reason: ~p, has_session: ~p", [UUID, Reason, HasSession]),
-    ok.
-
-%% @private
-%% @doc Convert process state when code is changed
-code_change(_OldVsn, StateName, State = #state{}, _Extra) ->
-    {ok, StateName, State}.
-
-%%%===================================================================
-%%% Internal functions
-%%%===================================================================
-
--spec get_route_key(binary()) -> binary().
-get_route_key(<<"">>) ->
-    <<"/">>;
-get_route_key(RouteKey) when is_binary(RouteKey) ->
-    RouteKey.
-
-%% 将当前的state转换成map
-state_map(#state{host_id = HostId, uuid = UUID, has_session = HasSession, heartbeat_counter = HeartbeatCounter, channel_pid = ChannelPid, metrics = Metrics}) ->
-    #{
-        host_id => HostId,
-        uuid => UUID,
-        has_session => HasSession,
-        heartbeat_counter => HeartbeatCounter,
-        channel_pid => ChannelPid,
-        metrics => Metrics
-    }.

apps/iot/src/iot_host_sup.erl

@@ -1,59 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%% @end
-%%%-------------------------------------------------------------------
--module(iot_host_sup).
--include("iot.hrl").
-
--behaviour(supervisor).
-
--export([start_link/0, init/1, delete_host/1, ensured_host_started/1]).
-
-start_link() ->
-    supervisor:start_link({local, ?MODULE}, ?MODULE, []).
-
-init([]) ->
-    Specs = lists:map(fun child_spec/1, iot_api:get_all_hosts()),
-
-    {ok, {#{strategy => one_for_one, intensity => 1000, period => 3600}, Specs}}.
-
--spec ensured_host_started(UUID :: binary()) -> {ok, Pid :: pid()} | {error, Reason :: any()}.
-ensured_host_started(UUID) when is_binary(UUID) ->
-    case iot_host:get_pid(UUID) of
-        undefined ->
-            %% 尝试删除下host对应的信息
-            delete_host(UUID),
-            case supervisor:start_child(?MODULE, child_spec(UUID)) of
-                {ok, Pid} when is_pid(Pid) ->
-                    {ok, Pid};
-                {error, {'already_started', Pid}} when is_pid(Pid) ->
-                    {ok, Pid};
-                {error, Error} ->
-                    {error, Error}
-            end;
-        Pid when is_pid(Pid) ->
-            {ok, Pid}
-    end.
-
-delete_host(UUID) ->
-    Id = iot_host:get_name(UUID),
-    ok = supervisor:terminate_child(?MODULE, Id),
-    case supervisor:delete_child(?MODULE, Id) of
-        {error, running} ->
-            %% ensure killed then delete again
-            iot_host:kill(UUID),
-            supervisor:delete_child(?MODULE, Id);
-        _ ->
-            ok
-    end.
-
-child_spec(UUID) ->
-    Id = iot_host:get_name(UUID),
-    #{id => Id,
-        start => {iot_host, start_link, [Id, UUID]},
-        restart => permanent,
-        shutdown => 2000,
-        type => worker,
-        modules => ['iot_host']}.

apps/iot/src/iot_http_client.erl

@@ -1,40 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author licheng5
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 03. 3月 2023 11:48
-%%%-------------------------------------------------------------------
--module(iot_http_client).
--author("licheng5").
-
-%% API
--export([post/2]).
-
-post(Url, Body) when is_list(Url), is_binary(Body) ->
-    case hackney:request(post, Url, [], Body) of
-        {ok, 200, _, ClientRef} ->
-            case hackney:body(ClientRef) of
-                {ok, RespBody} ->
-                    lager:debug("[iot_http_client] url: ~p, response is: ~p", [Url, RespBody]),
-                    ok;
-                {error, Reason} ->
-                    lager:warning("[iot_http_client] url: ~p, get error: ~p", [Url, Reason]),
-                    {error, Reason}
-            end;
-
-        {ok, HttpCode, _, ClientRef} ->
-            case hackney:body(ClientRef) of
-                {ok, RespBody} ->
-                    lager:debug("[iot_http_client] url: ~p, http_code: ~p, response is: ~p", [Url, HttpCode, RespBody]),
-                    ok;
-                {error, Reason} ->
-                    lager:warning("[iot_http_client] url: ~p, http_code: ~p, get error: ~p", [Url, HttpCode, Reason]),
-                    {error, Reason}
-            end;
-
-        {error, Reason} ->
-            lager:warning("[iot_http_client] url: ~p, get error: ~p", [Url, Reason]),
-            {error, Reason}
-    end.

apps/iot/src/iot_log.erl

@@ -0,0 +1,10 @@
+%%%-------------------------------------------------------------------
+%%% @doc Logger helper for project-scoped metadata.
+%%%-------------------------------------------------------------------
+-module(iot_log).
+
+-export([set_metadata/0]).
+
+-spec set_metadata() -> ok.
+set_metadata() ->
+    logger:set_process_metadata(#{domain => [iot]}).

apps/iot/src/iot_sup.erl

@@ -29,21 +29,21 @@ init([]) ->
 
     Specs = [
         #{
-            id => 'iot_event_stream_observer',
+            id => 'iot_container_task_sup',
-            start => {'iot_event_stream_observer', start_link, []},
+            start => {'iot_container_task_sup', start_link, []},
-            restart => permanent,
-            shutdown => 2000,
-            type => worker,
-            modules => ['iot_event_stream_observer']
-        },
-
-        #{
-            id => endpoint_sup_sup,
-            start => {'endpoint_sup_sup', start_link, []},
             restart => permanent,
             shutdown => 2000,
             type => supervisor,
-            modules => ['endpoint_sup_sup']
+            modules => ['iot_container_task_sup']
+        },
+
+        #{
+            id => 'udp_server',
+            start => {'udp_server', start_link, []},
+            restart => permanent,
+            shutdown => 2000,
+            type => worker,
+            modules => ['udp_server']
         },
 
         #{

apps/iot/src/iot_udp_handler.erl

@@ -1,26 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2023, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 31. 8月 2023 13:48
-%%%-------------------------------------------------------------------
--module(iot_udp_handler).
--author("aresei").
-
-%% API
--export([start_link/2, loop/2]).
-
-start_link(Transport, Peer) ->
-    {ok, spawn_link(?MODULE, loop, [Transport, Peer])}.
-
-loop(Transport = {udp, Server, _Sock}, Peer) ->
-    receive
-        {datagram, Server, <<Len:16, HostUUID:Len/binary>>} ->
-            Pid = iot_host:get_pid(HostUUID),
-            iot_host:heartbeat(Pid),
-            loop(Transport, Peer);
-        {datagram, Server, _} ->
-            exit(normal)
-    end.

apps/iot/src/iot_util.erl

@@ -10,15 +10,15 @@
 -author("licheng5").
 
 %% API
--export([timestamp/0, number_format/2, current_time/0, timestamp_of_seconds/0, float_to_binary/2, int_format/2, file_uri/1]).
+-export([timestamp/0, number_format/2, current_time/0, timestamp_of_seconds/0, float_to_binary/2, int_format/2]).
--export([step/3, chunks/2, rand_bytes/1, uuid/0, md5/1, parse_mapper/1]).
+-export([step/3, rand_bytes/1, uuid/0, md5/1, parse_mapper/1]).
 -export([json_data/1, json_error/2, is_json/1]).
 -export([queue_limited_in/3, assert_call/2, assert/2]).
 -export([sha256/1]).
 
 -spec is_json(Json :: term()) -> boolean().
 is_json(Json) when is_binary(Json) ->
-    case catch jiffy:decode(Json, [return_maps]) of
+    case catch json:decode(Json) of
         Result when is_list(Result) orelse is_map(Result) ->
             true;
         _ ->
@@ -61,33 +61,18 @@ step(Start, End, Step, Acc) when Start < End ->
 step(_, _, _, Acc) ->
     lists:reverse(Acc).
 
-%% 将数据分组
--spec chunks(list(), integer()) -> [list()].
-chunks(List, Size) when is_list(List), is_integer(Size), Size > 0, length(List) =< Size ->
-    [List];
-chunks(List, Size) when is_list(List), is_integer(Size), Size > 0 ->
-    chunks0(List, Size, Size, [], []).
-chunks0([], _, _, [], AccTarget) ->
-    lists:reverse(AccTarget);
-chunks0([], _, _, Target, AccTarget) ->
-    lists:reverse([lists:reverse(Target) | AccTarget]);
-chunks0(List, Size, 0, Target, AccTarget) ->
-    chunks0(List, Size, Size, [], [lists:reverse(Target) | AccTarget]);
-chunks0([Hd | Tail], Size, Num, Target, AccTarget) ->
-    chunks0(Tail, Size, Num - 1, [Hd | Target], AccTarget).
-
 json_data(Data) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"result">> => Data
-    }, [force_utf8]).
+    })).
 
 json_error(ErrCode, ErrMessage) when is_integer(ErrCode) ->
-    jiffy:encode(#{
+    iolist_to_binary(json:encode(#{
         <<"error">> => #{
             <<"code">> => ErrCode,
             <<"message">> => ErrMessage
         }
-    }, [force_utf8]).
+    })).
 
 uuid() ->
     rand_bytes(16).
@@ -132,18 +117,9 @@ hex(N) ->
 
 %% 转换映射器
 -spec parse_mapper(Mapper :: binary() | string()) -> error | {ok, F :: fun((binary(), any()) -> any())}.
-parse_mapper(Mapper) when is_binary(Mapper) ->
+parse_mapper(_Mapper) ->
-    parse_mapper(binary_to_list(Mapper));
+    %% Dynamic Erlang eval is intentionally disabled in runtime code.
-parse_mapper(Mapper) when is_list(Mapper) ->
+    error.
-    {ok, Tokens, _} = erl_scan:string(Mapper),
-    {ok, ExprList} = erl_parse:parse_exprs(Tokens),
-    {value, F, _} = erl_eval:exprs(ExprList, []),
-    case is_function(F, 2) orelse is_function(F, 3) of
-        true ->
-            {ok, F};
-        false ->
-            error
-    end.
 
 -spec float_to_binary(Num :: number(), integer()) -> binary().
 float_to_binary(V, _) when is_integer(V) ->
@@ -158,13 +134,3 @@ assert(false, F) when is_function(F) ->
     F();
 assert(false, Msg) ->
     throw(Msg).
-
-
--spec file_uri(Filename :: binary()) -> error | {ok, FileUri :: binary()}.
-file_uri(Filename) when is_binary(Filename) ->
-    case binary:split(Filename, <<"-">>, [global]) of
-        [Year, Month, Day | _] ->
-            {ok, <<"https://lgsiot.njau.edu.cn/upload/", Year/binary, $/, Month/binary, $/, Day/binary, $/, Filename/binary>>};
-        _ ->
-            error
-    end.

apps/iot/src/message/message_codec.erl

@@ -1,141 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author anlicheng
-%%% @copyright (C) 2025, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 17. 9月 2025 16:05
-%%%-------------------------------------------------------------------
--module(message_codec).
--author("anlicheng").
--include("message.hrl").
-
--define(I8, 1).
--define(I16, 2).
--define(I32, 3).
--define(Bytes, 4).
-
-%% API
--export([encode/2, decode/1]).
-
--spec encode(MessageType :: integer(), Message :: any()) -> binary().
-encode(MessageType, Message) when is_integer(MessageType) ->
-    Bin = encode0(Message),
-    <<MessageType, Bin/binary>>.
-encode0(#auth_request{uuid = UUID, username = Username, salt = Salt, token = Token, timestamp = Timestamp}) ->
-    iolist_to_binary([
-        marshal(?Bytes, UUID),
-        marshal(?Bytes, Username),
-        marshal(?Bytes, Salt),
-        marshal(?Bytes, Token),
-        marshal(?I32, Timestamp)
-    ]);
-encode0(#auth_reply{code = Code, payload = Payload}) ->
-    iolist_to_binary([
-        marshal(?I32, Code),
-        marshal(?Bytes, Payload)
-    ]);
-encode0(#jsonrpc_reply{result = Result, error = undefined}) ->
-    ResultBin = erlang:term_to_binary(#{<<"result">> => Result}),
-    iolist_to_binary([
-        marshal(?Bytes, ResultBin)
-    ]);
-encode0(#jsonrpc_reply{result = undefined, error = Error}) ->
-    ResultBin = erlang:term_to_binary(#{<<"error">> => Error}),
-    iolist_to_binary([
-        marshal(?Bytes, ResultBin)
-    ]);
-encode0(#pub{topic = Topic, qos = Qos, content = Content}) ->
-    iolist_to_binary([
-        marshal(?Bytes, Topic),
-        marshal(?I8, Qos),
-        marshal(?Bytes, Content)
-    ]);
-encode0(#command{command_type = CommandType, command = Command}) ->
-    iolist_to_binary([
-        marshal(?I32, CommandType),
-        marshal(?Bytes, Command)
-    ]);
-
-encode0(#jsonrpc_request{method = Method, params = Params}) ->
-    ReqBody = erlang:term_to_binary(#{<<"method">> => Method, <<"params">> => Params}),
-    iolist_to_binary([
-        marshal(?Bytes, ReqBody)
-    ]);
-encode0(#data{route_key = RouteKey, metric = Metric}) ->
-    iolist_to_binary([
-        marshal(?Bytes, RouteKey),
-        marshal(?Bytes, Metric)
-    ]);
-encode0(#task_event_stream{task_id = TaskId, type = Type, stream = Stream}) ->
-    iolist_to_binary([
-        marshal(?I32, TaskId),
-        marshal(?Bytes, Type),
-        marshal(?Bytes, Stream)
-    ]).
-
--spec decode(Bin :: binary()) -> {ok, Message :: any()} | error.
-decode(<<PacketType:8, Packet/binary>>) ->
-    case unmarshal(Packet) of
-        {ok, Fields} ->
-            decode0(PacketType, Fields);
-        error ->
-            error
-    end.
-decode0(?MESSAGE_AUTH_REQUEST, [UUID, Username, Salt, Token, Timestamp]) ->
-    {ok, #auth_request{uuid = UUID, username = Username, salt = Salt, token = Token, timestamp = Timestamp}};
-decode0(?MESSAGE_JSONRPC_REPLY, [ReplyBin]) ->
-    case erlang:binary_to_term(ReplyBin) of
-        #{<<"result">> := Result} ->
-            {ok, #jsonrpc_reply{result = Result}};
-        #{<<"error">> := Error} ->
-            {ok, #jsonrpc_reply{error = Error}};
-        _ ->
-            error
-    end;
-decode0(?MESSAGE_PUB, [Topic, Qos, Content]) ->
-    {ok, #pub{topic = Topic, qos = Qos, content = Content}};
-decode0(?MESSAGE_COMMAND, [CommandType, Command]) ->
-    {ok, #command{command_type = CommandType, command = Command}};
-decode0(?MESSAGE_AUTH_REPLY, [Code, Payload]) ->
-    {ok, #auth_reply{code = Code, payload = Payload}};
-decode0(?MESSAGE_JSONRPC_REQUEST, [ReqBody]) ->
-    #{<<"method">> := Method, <<"params">> := Params} = erlang:binary_to_term(ReqBody),
-    {ok, #jsonrpc_request{method = Method, params = Params}};
-decode0(?MESSAGE_DATA, [RouteKey, Metric]) ->
-    {ok, #data{route_key = RouteKey, metric = Metric}};
-decode0(?MESSAGE_EVENT_STREAM, [TaskId, Type, Stream]) ->
-    {ok, #task_event_stream{task_id = TaskId, type = Type, stream = Stream}};
-decode0(_, _) ->
-    error.
-
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%%% helper methods
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
--spec marshal(Type :: ?I8 | ?I16 | ?I32 | ?Bytes, Field :: integer() | binary()) -> binary().
-marshal(?I8, Field) when is_integer(Field) ->
-    <<?I8, Field:8>>;
-marshal(?I16, Field) when is_integer(Field) ->
-    <<?I16, Field:16>>;
-marshal(?I32, Field) when is_integer(Field) ->
-    <<?I32, Field:32>>;
-marshal(?Bytes, Field) when is_binary(Field) ->
-    Len = byte_size(Field),
-    <<?Bytes, Len:16, Field/binary>>.
-
--spec unmarshal(Bin :: binary()) -> {ok, Components :: [any()]} | error.
-unmarshal(Bin) when is_binary(Bin) ->
-    unmarshal(Bin, []).
-unmarshal(<<>>, Acc) ->
-    {ok, lists:reverse(Acc)};
-unmarshal(<<?I8, F:8, Rest/binary>>, Acc) ->
-    unmarshal(Rest, [F|Acc]);
-unmarshal(<<?I16, F:16, Rest/binary>>, Acc) ->
-    unmarshal(Rest, [F|Acc]);
-unmarshal(<<?I32, F:32, Rest/binary>>, Acc) ->
-    unmarshal(Rest, [F|Acc]);
-unmarshal(<<?Bytes, Len:16, F:Len/binary, Rest/binary>>, Acc) ->
-    unmarshal(Rest, [F|Acc]);
-unmarshal(_, _) ->
-    error.

apps/iot/src/mnesia/efka_client_store.erl

@@ -0,0 +1,154 @@
+%%%-------------------------------------------------------------------
+%%% @doc efka_client table store based on mnesia.
+%%%-------------------------------------------------------------------
+-module(efka_client_store).
+
+-include("iot_tables.hrl").
+
+-export([ensure_table/0, register/2, get/1, update/1, delete/1, list/0]).
+-export([auth/2, verify_heartbeat/4]).
+
+-define(TOKEN_HASH_ITERATIONS, 32).
+-define(TOKEN_HASH_BYTES, 32).
+-define(HEARTBEAT_TIMESTAMP_WINDOW, 120).
+
+-spec ensure_table() -> ok | {error, term()}.
+ensure_table() ->
+    case mnesia:create_table(efka_client, [
+        {attributes, record_info(fields, efka_client)},
+        {type, set},
+        {record_name, efka_client},
+        {disc_copies, [node()]}
+    ]) of
+        {atomic, ok} ->
+            ok;
+        {aborted, {already_exists, efka_client}} ->
+            ok;
+        {aborted, Reason} ->
+            {error, Reason}
+    end.
+
+-spec auth(UUID :: binary(), Token :: binary()) -> boolean().
+auth(UUID, Token) when is_binary(UUID), UUID =/= <<>>, is_binary(Token) ->
+    case mnesia:dirty_read(efka_client, UUID) of
+        [] ->
+            false;
+        [#efka_client{token_hash = TokenHash, salt = Salt}] ->
+            hash_token(Token, Salt) =:= TokenHash
+    end.
+
+-spec verify_heartbeat(UUID :: binary(), Timestamp :: integer(), Payload :: binary(), Mac :: binary()) -> boolean().
+verify_heartbeat(UUID, Timestamp, Payload, Mac)
+  when is_binary(UUID), UUID =/= <<>>, is_integer(Timestamp), is_binary(Payload), is_binary(Mac) ->
+    case valid_heartbeat_timestamp(Timestamp) of
+        true ->
+            case mnesia:dirty_read(efka_client, UUID) of
+                [] ->
+                    false;
+                [#efka_client{heartbeat_secret = HeartbeatSecret}] ->
+                    ExpectedMac = crypto:mac(hmac, sha256, HeartbeatSecret, Payload),
+                    ExpectedMac =:= Mac
+            end;
+        false ->
+            false
+    end.
+
+-spec register(UUID :: binary(), Token :: binary()) -> ok | {error, term()}.
+register(UUID, Token) when is_binary(UUID), UUID =/= <<>>, is_binary(Token), byte_size(Token) >= 32 ->
+    case mnesia:transaction(fun() ->
+        case mnesia:read(efka_client, UUID, write) of
+            [] ->
+                %% 1. 生成 16 字节随机盐
+                Salt = crypto:strong_rand_bytes(16),
+                TokenHash = hash_token(Token, Salt),
+                HeartbeatSecret = heartbeat_secret(Token),
+                ok = mnesia:write(#efka_client{
+                    uuid = UUID,
+                    token_hash = TokenHash,
+                    salt = Salt,
+                    heartbeat_secret = HeartbeatSecret,
+                    timestamp = iot_util:timestamp()
+                }),
+                ok;
+            [_] ->
+                {error, already_exists}
+        end
+                            end) of
+        {atomic, ok} ->
+            ok;
+        {atomic, {error, Reason}} ->
+            {error, Reason};
+        {aborted, Reason} ->
+            {error, Reason}
+    end.
+
+-spec get(binary()) -> {ok, #efka_client{}} | not_found | {error, term()}.
+get(UUID) when is_binary(UUID), UUID =/= <<>> ->
+    case mnesia:transaction(fun() ->
+        case mnesia:read(efka_client, UUID, read) of
+            [Client = #efka_client{}] ->
+                {ok, Client};
+            [] ->
+                not_found
+        end
+    end) of
+        {atomic, Result} ->
+            Result;
+        {aborted, Reason} ->
+            {error, Reason}
+    end.
+
+-spec update(#efka_client{}) -> ok | {error, term()}.
+update(Client = #efka_client{uuid = UUID}) when is_binary(UUID), UUID =/= <<>> ->
+    case mnesia:transaction(fun() ->
+        case mnesia:read(efka_client, UUID, write) of
+            [_] ->
+                ok = mnesia:write(Client),
+                ok;
+            [] ->
+                {error, not_found}
+        end
+    end) of
+        {atomic, ok} ->
+            ok;
+        {atomic, {error, Reason}} ->
+            {error, Reason};
+        {aborted, Reason} ->
+            {error, Reason}
+    end.
+
+-spec delete(binary()) -> ok | {error, term()}.
+delete(UUID) when is_binary(UUID), UUID =/= <<>> ->
+    case mnesia:transaction(fun() ->
+        ok = mnesia:delete({efka_client, UUID}),
+        ok
+    end) of
+        {atomic, ok} ->
+            ok;
+        {aborted, Reason} ->
+            {error, Reason}
+    end.
+
+-spec list() -> {ok, [#efka_client{}]} | {error, term()}.
+list() ->
+    case mnesia:transaction(fun() ->
+        mnesia:foldl(fun(Client = #efka_client{}, Acc) -> [Client | Acc] end, [], efka_client)
+    end) of
+        {atomic, Clients} ->
+            {ok, lists:reverse(Clients)};
+        {aborted, Reason} ->
+            {error, Reason}
+    end.
+
+-spec hash_token(binary(), binary()) -> binary().
+hash_token(Token, Salt) when is_binary(Token), is_binary(Salt) ->
+    crypto:pbkdf2_hmac(sha256, Token, Salt, ?TOKEN_HASH_ITERATIONS, ?TOKEN_HASH_BYTES).
+
+-spec heartbeat_secret(binary()) -> binary().
+heartbeat_secret(Token) when is_binary(Token) ->
+    crypto:hash(sha256, Token).
+
+-spec valid_heartbeat_timestamp(integer()) -> boolean().
+valid_heartbeat_timestamp(Timestamp) ->
+    Now = iot_util:current_time(),
+    Timestamp =< Now andalso Now - Timestamp =< ?HEARTBEAT_TIMESTAMP_WINDOW.

apps/iot/src/mnesia/iot_mnesia.erl

@@ -0,0 +1,74 @@
+%%%-------------------------------------------------------------------
+%%% @doc Mnesia bootstrap and runtime startup helpers.
+%%%
+%%% `init/0` is a manual, one-time bootstrap operation. Application
+%%% startup must call `start/0`, which only starts an existing schema and
+%%% verifies required tables.
+%%%-------------------------------------------------------------------
+-module(iot_mnesia).
+
+-include("iot_tables.hrl").
+
+-export([safe_start/0, stop/0]).
+
+-define(TABLES, [efka_client]).
+-define(WAIT_TIMEOUT, 5000).
+
+safe_start() ->
+    case ensure_schema() of
+        create_and_start ->
+            ok = mnesia:start(),
+            ok = create_tables(),
+            ok = wait_for_tables();
+        only_start ->
+            ok = mnesia:start(),
+            ok = wait_for_tables()
+    end,
+    log_runtime_info().
+
+ensure_schema() ->
+    stopped = mnesia:stop(),
+    case mnesia:create_schema([node()]) of
+        ok ->
+            create_and_start;
+        {error, {_, {already_exists, _}}} ->
+            only_start;
+        {error, {_, already_exists}} ->
+            only_start;
+        {error, _Reason} ->
+            only_start
+    end.
+
+-spec stop() -> ok.
+stop() ->
+    _ = mnesia:stop(),
+    ok.
+
+-spec create_tables() -> ok.
+create_tables() ->
+    case efka_client_store:ensure_table() of
+        ok ->
+            ok;
+        {error, Reason} ->
+            {error, {create_table_failed, {efka_client, Reason}}}
+    end.
+
+-spec wait_for_tables() -> ok.
+wait_for_tables() ->
+    case mnesia:wait_for_tables(?TABLES, ?WAIT_TIMEOUT) of
+        ok ->
+            ok;
+        {timeout, Tables} ->
+            {error, {mnesia_tables_timeout, Tables}};
+        {error, Reason} ->
+            {error, {mnesia_tables_error, Reason}}
+    end.
+
+-spec log_runtime_info() -> ok.
+log_runtime_info() ->
+    logger:debug("[iot_mnesia] node: ~p, dir: ~p, tables: ~p", [
+        node(),
+        mnesia:system_info(directory),
+        mnesia:system_info(tables)
+    ]),
+    ok.

apps/iot/src/mysql/mysql_pool.erl

@@ -1,48 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2018, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 29. 九月 2018 17:01
-%%%-------------------------------------------------------------------
--module(mysql_pool).
--author("aresei").
-
-%% API
--export([get_row/2, get_row/3, get_all/2, get_all/3]).
--export([update/4, update_by/2, update_by/3, insert/4]).
-
-%% 从数据库中查找一行记录
--spec get_row(Pool :: atom(), Sql::binary()) -> {ok, Record::map()} | undefined.
-get_row(Pool, Sql) when is_atom(Pool), is_binary(Sql) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:get_row(ConnPid, Sql) end).
-
--spec get_row(Pool :: atom(), Sql::binary(), Params::list()) -> {ok, Record::map()} | undefined.
-get_row(Pool, Sql, Params) when is_atom(Pool), is_binary(Sql), is_list(Params) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:get_row(ConnPid, Sql, Params) end).
-
--spec get_all(Pool :: atom(), Sql::binary()) -> {ok, Rows::list()} | {error, Reason :: any()}.
-get_all(Pool, Sql) when is_atom(Pool), is_binary(Sql) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:get_all(ConnPid, Sql) end).
-
--spec get_all(Pool :: atom(), Sql::binary(), Params::list()) -> {ok, Rows::list()} | {error, Reason::any()}.
-get_all(Pool, Sql, Params) when is_atom(Pool), is_binary(Sql), is_list(Params) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:get_all(ConnPid, Sql, Params) end).
-
--spec insert(Pool :: atom(), Table :: binary(), Fields :: map() | list(), boolean()) ->
-    ok | {ok, InsertId :: integer()} | {error, Reason :: any()}.
-insert(Pool, Table, Fields, FetchInsertId) when is_atom(Pool), is_binary(Table), is_list(Fields); is_map(Fields), is_boolean(FetchInsertId) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:insert(ConnPid, Table, Fields, FetchInsertId) end).
-
--spec update_by(Pool :: atom(), UpdateSql :: binary()) -> {ok, AffectedRows :: integer()} | {error, Reason :: any()}.
-update_by(Pool, UpdateSql) when is_atom(Pool), is_binary(UpdateSql) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:update_by(ConnPid, UpdateSql) end).
-
--spec update_by(Pool :: atom(), UpdateSql :: binary(), Params :: list()) -> {ok, AffectedRows :: integer()} | {error, Reason :: any()}.
-update_by(Pool, UpdateSql, Params) when is_atom(Pool), is_binary(UpdateSql) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:update_by(ConnPid, UpdateSql, Params) end).
-
--spec update(Pool :: atom(), Table :: binary(), Fields :: map(), WhereFields :: map()) -> {ok, AffectedRows::integer()} | {error, Reason::any()}.
-update(Pool, Table, Fields, WhereFields) when is_atom(Pool), is_binary(Table), is_map(Fields), is_map(WhereFields) ->
-    poolboy:transaction(Pool, fun(ConnPid) -> mysql_provider:update(ConnPid, Table, Fields, WhereFields) end).

apps/iot/src/mysql/mysql_provider.erl

@@ -1,144 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author aresei
-%%% @copyright (C) 2018, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 29. 九月 2018 17:01
-%%%-------------------------------------------------------------------
--module(mysql_provider).
--author("aresei").
-
-%% API
--export([get_row/2, get_row/3, get_all/2, get_all/3]).
--export([update/4, update_by/2, update_by/3, insert/4]).
-
-%% 从数据库中查找一行记录
--spec get_row(ConnPid :: pid(), Sql::binary()) -> {ok, Record::map()} | undefined.
-get_row(ConnPid, Sql) when is_pid(ConnPid), is_binary(Sql) ->
-    lager:debug("[mysql_client] get_row sql is: ~p", [Sql]),
-    case mysql:query(ConnPid, Sql) of
-        {ok, Names, [Row | _]} ->
-            {ok, maps:from_list(lists:zip(Names, Row))};
-        {ok, _, []} ->
-            undefined;
-        Error ->
-            lager:warning("[mysql_client] get error: ~p", [Error]),
-            undefined
-    end.
-
--spec get_row(ConnPid :: pid(), Sql::binary(), Params::list()) -> {ok, Record::map()} | undefined.
-get_row(ConnPid, Sql, Params) when is_pid(ConnPid), is_binary(Sql), is_list(Params) ->
-    lager:debug("[mysql_client] get_row sql is: ~p, params: ~p", [Sql, Params]),
-    case mysql:query(ConnPid, Sql, Params) of
-        {ok, Names, [Row | _]} ->
-            {ok, maps:from_list(lists:zip(Names, Row))};
-        {ok, _, []} ->
-            undefined;
-        Error ->
-            lager:warning("[mysql_client] get error: ~p", [Error]),
-            undefined
-    end.
-
--spec get_all(ConnPid :: pid(), Sql::binary()) -> {ok, Rows::list()} | {error, Reason :: any()}.
-get_all(ConnPid, Sql) when is_pid(ConnPid), is_binary(Sql) ->
-    lager:debug("[mysql_client] get_all sql is: ~p", [Sql]),
-    case mysql:query(ConnPid, Sql) of
-        {ok, Names, Rows} ->
-            {ok, lists:map(fun(Row) -> maps:from_list(lists:zip(Names, Row)) end, Rows)};
-        {error, Reason} ->
-            lager:warning("[mysql_client] get error: ~p", [Reason]),
-            {error, Reason}
-    end.
-
--spec get_all(ConnPid :: pid(), Sql::binary(), Params::list()) -> {ok, Rows::list()} | {error, Reason::any()}.
-get_all(ConnPid, Sql, Params) when is_pid(ConnPid), is_binary(Sql), is_list(Params) ->
-    lager:debug("[mysql_client] get_all sql is: ~p, params: ~p", [Sql, Params]),
-    case mysql:query(ConnPid, Sql, Params) of
-        {ok, Names, Rows} ->
-            {ok, lists:map(fun(Row) -> maps:from_list(lists:zip(Names, Row)) end, Rows)};
-        {error, Reason} ->
-            lager:warning("[mysql_client] get error: ~p", [Reason]),
-            {error, Reason}
-    end.
-
--spec insert(ConnPid :: pid(), Table :: binary(), Fields :: map() | list(), boolean()) ->
-    ok | {ok, InsertId :: integer()} | {error, Reason :: any()}.
-insert(ConnPid, Table, Fields, FetchInsertId) when is_pid(ConnPid), is_binary(Table), is_map(Fields), is_boolean(FetchInsertId) ->
-    insert(ConnPid, Table, maps:to_list(Fields), FetchInsertId);
-insert(ConnPid, Table, Fields, FetchInsertId) when is_pid(ConnPid), is_binary(Table), is_list(Fields), is_boolean(FetchInsertId) ->
-    {Keys, Values} = kvs(Fields),
-
-    FieldSql = iolist_to_binary(lists:join(<<", ">>, Keys)),
-    Placeholders = lists:duplicate(length(Keys), <<"?">>),
-    ValuesPlaceholder = iolist_to_binary(lists:join(<<", ">>, Placeholders)),
-
-    Sql = <<"INSERT INTO ", Table/binary, "(", FieldSql/binary, ") VALUES(", ValuesPlaceholder/binary, ")">>,
-    lager:debug("[mysql_client] insert sql is: ~p, params: ~p", [Sql, Values]),
-    case mysql:query(ConnPid, Sql, Values) of
-        ok ->
-            case FetchInsertId of
-                true ->
-                    InsertId = mysql:insert_id(ConnPid),
-                    {ok, InsertId};
-                false ->
-                    ok
-            end;
-        Error ->
-            Error
-    end.
-
--spec update_by(ConnPid :: pid(), UpdateSql :: binary()) -> {ok, AffectedRows :: integer()} | {error, Reason :: any()}.
-update_by(ConnPid, UpdateSql) when is_pid(ConnPid), is_binary(UpdateSql) ->
-    lager:debug("[mysql_client] updateBySql sql: ~p", [UpdateSql]),
-    case mysql:query(ConnPid, UpdateSql) of
-        ok ->
-            AffectedRows = mysql:affected_rows(ConnPid),
-            {ok, AffectedRows};
-        Error ->
-            Error
-    end.
-
--spec update_by(ConnPid :: pid(), UpdateSql :: binary(), Params :: list()) -> {ok, AffectedRows :: integer()} | {error, Reason :: any()}.
-update_by(ConnPid, UpdateSql, Params) when is_pid(ConnPid), is_binary(UpdateSql) ->
-    lager:debug("[mysql_client] updateBySql sql: ~p, params: ~p", [UpdateSql, Params]),
-    case mysql:query(ConnPid, UpdateSql, Params) of
-        ok ->
-            AffectedRows = mysql:affected_rows(ConnPid),
-            {ok, AffectedRows};
-        Error ->
-            Error
-    end.
-
--spec update(ConnPid :: pid(), Sql :: binary(), Fields :: map(), WhereFields :: map()) ->
-    {ok, AffectedRows::integer()} | {error, Reason::any()}.
-update(ConnPid, Table, Fields, WhereFields) when is_pid(ConnPid), is_binary(Table), is_map(Fields), is_map(WhereFields) ->
-    %% 拼接set
-    {SetKeys, SetVals} = kvs(Fields),
-    SetKeys1 = lists:map(fun(K) when is_binary(K) -> <<"`", K/binary, "` = ?">> end, SetKeys),
-    SetSql = iolist_to_binary(lists:join(<<", ">>, SetKeys1)),
-
-    %% 拼接where
-    {WhereKeys, WhereVals} = kvs(WhereFields),
-    WhereKeys1 = lists:map(fun(K) when is_binary(K) -> <<"`", K/binary, "` = ?">> end, WhereKeys),
-    WhereSql = iolist_to_binary(lists:join(<<" AND ">>, WhereKeys1)),
-
-    Params = SetVals ++ WhereVals,
-
-    Sql = <<"UPDATE ", Table/binary, " SET ",  SetSql/binary, " WHERE ", WhereSql/binary>>,
-    lager:debug("[mysql_client] update sql is: ~p, params: ~p", [Sql, Params]),
-    case mysql:query(ConnPid, Sql, Params) of
-        ok ->
-            AffectedRows = mysql:affected_rows(ConnPid),
-            {ok, AffectedRows};
-        Error ->
-            lager:error("[mysql_client] update sql: ~p, params: ~p, get a error: ~p", [Sql, Params, Error]),
-            Error
-    end.
-
--spec kvs(Fields :: map() | list()) -> {Keys :: list(), Values :: list()}.
-kvs(Fields) when is_map(Fields) ->
-    kvs(maps:to_list(Fields));
-kvs(Fields) when is_list(Fields) ->
-    {Keys0, Values0} = lists:foldl(fun({K, V}, {Acc0, Acc1}) -> {[K|Acc0], [V|Acc1]} end, {[], []}, Fields),
-    {lists:reverse(Keys0), lists:reverse(Values0)}.

apps/iot/src/simulator/simulator_api_handler.erl

@@ -0,0 +1,156 @@
+%%%-------------------------------------------------------------------
+%%% @author Codex
+%%% @doc
+%%% 本地模拟 control api，用于替代 iot_api_client 依赖的远端接口。
+%%% @end
+%%%-------------------------------------------------------------------
+-module(simulator_api_handler).
+
+-include_lib("endpoint/include/endpoint.hrl").
+
+-export([handle_request/4]).
+
+-define(HOST_UUID, <<"qbxmjyzrkpntfgswaevodhluicqzxplkm">>).
+-define(HOST_ID, 1).
+-define(DEVICE_UUID, <<"sim-device-001">>).
+-define(DEVICE_ID, 1).
+-define(ENDPOINT_ID, 1).
+-define(TABLE, ?MODULE).
+
+-spec handle_request(string(), string(), map(), map()) ->
+    {ok, non_neg_integer(), iodata()}.
+handle_request("GET", "/simulator/get_all_hosts", _, _) ->
+    {ok, 200, iot_util:json_data([?HOST_UUID])};
+handle_request("GET", "/simulator/get_host_by_uuid", #{<<"uuid">> := UUID}, _) when is_binary(UUID) ->
+    reply_host_by_uuid(UUID);
+handle_request("GET", "/simulator/get_host_by_id", #{<<"host_id">> := HostIdBin}, _) when is_binary(HostIdBin) ->
+    case binary_to_integer(HostIdBin) of
+        ?HOST_ID ->
+            {ok, 200, iot_util:json_data(host_info())};
+        _ ->
+            {ok, 200, iot_util:json_error(404, <<"host not found">>)}
+    end;
+handle_request("POST", "/simulator/change_host_status", _, #{<<"uuid">> := UUID, <<"new_status">> := Status})
+        when is_binary(UUID), is_integer(Status) ->
+    reply_change_host_status(UUID, Status);
+handle_request("GET", "/simulator/get_host_devices", #{<<"host_id">> := HostIdBin}, _) when is_binary(HostIdBin) ->
+    case binary_to_integer(HostIdBin) of
+        ?HOST_ID ->
+            {ok, 200, iot_util:json_data([device_info()])};
+        _ ->
+            {ok, 200, iot_util:json_error(404, <<"host not found">>)}
+    end;
+handle_request("GET", "/simulator/get_device_by_uuid", #{<<"device_uuid">> := DeviceUUID}, _) when is_binary(DeviceUUID) ->
+    case DeviceUUID of
+        ?DEVICE_UUID ->
+            {ok, 200, iot_util:json_data(device_info())};
+        _ ->
+            {ok, 200, iot_util:json_error(404, <<"device not found">>)}
+    end;
+handle_request("POST", "/simulator/change_device_status", _, #{<<"device_uuid">> := DeviceUUID, <<"new_status">> := Status})
+        when is_binary(DeviceUUID), is_integer(Status) ->
+    reply_change_device_status(DeviceUUID, Status);
+handle_request("GET", "/simulator/get_all_endpoints", _, _) ->
+    {ok, 200, iot_util:json_data([endpoint_info()])};
+handle_request("GET", "/simulator/get_endpoint", #{<<"id">> := IdBin}, _) when is_binary(IdBin) ->
+    case binary_to_integer(IdBin) of
+        ?ENDPOINT_ID ->
+            {ok, 200, iot_util:json_data(endpoint_info())};
+        _ ->
+            {ok, 200, iot_util:json_error(404, <<"endpoint not found">>)}
+    end;
+handle_request("POST", "/simulator/endpoint_sink", _, Body) ->
+    logger:debug("[simulator_api_handler] receive endpoint sink body: ~p", [Body]),
+    {ok, 200, iot_util:json_data(<<"ok">>)};
+handle_request(_, Path, _, _) ->
+    PathBin = list_to_binary(Path),
+    {ok, 200, iot_util:json_error(-1, <<"url: ", PathBin/binary, " not found">>)}.
+
+%%%===================================================================
+%%% Internal functions
+%%%===================================================================
+
+-spec reply_host_by_uuid(binary()) -> {ok, non_neg_integer(), iodata()}.
+reply_host_by_uuid(?HOST_UUID) ->
+    {ok, 200, iot_util:json_data(host_info())};
+reply_host_by_uuid(_) ->
+    {ok, 200, iot_util:json_error(404, <<"host not found">>)}.
+
+-spec reply_change_host_status(binary(), integer()) -> {ok, non_neg_integer(), iodata()}.
+reply_change_host_status(?HOST_UUID, Status) ->
+    ok = put_state({host_status, ?HOST_UUID}, Status),
+    {ok, 200, iot_util:json_data(1)};
+reply_change_host_status(_, _) ->
+    {ok, 200, iot_util:json_error(404, <<"host not found">>)}.
+
+-spec reply_change_device_status(binary(), integer()) -> {ok, non_neg_integer(), iodata()}.
+reply_change_device_status(?DEVICE_UUID, Status) ->
+    ok = put_state({device_status, ?DEVICE_UUID}, Status),
+    {ok, 200, iot_util:json_data(1)};
+reply_change_device_status(_, _) ->
+    {ok, 200, iot_util:json_error(404, <<"device not found">>)}.
+
+-spec host_info() -> map().
+host_info() ->
+    #{
+        <<"id">> => ?HOST_ID,
+        <<"uuid">> => ?HOST_UUID,
+        <<"authorize_status">> => 1,
+        <<"status">> => get_state({host_status, ?HOST_UUID}, 1)
+    }.
+
+-spec device_info() -> map().
+device_info() ->
+    #{
+        <<"id">> => ?DEVICE_ID,
+        <<"host_id">> => ?HOST_ID,
+        <<"device_uuid">> => ?DEVICE_UUID,
+        <<"status">> => get_state({device_status, ?DEVICE_UUID}, 1)
+    }.
+
+-spec endpoint_info() -> map().
+endpoint_info() ->
+    #{
+        <<"id">> => ?ENDPOINT_ID,
+        <<"matcher">> => <<"simulator.metric">>,
+        <<"title">> => <<"Simulator HTTP Endpoint">>,
+        <<"type">> => <<"http">>,
+        <<"config">> => #{
+            <<"url">> => <<"http://127.0.0.1:18090/simulator/endpoint_sink">>,
+            <<"pool_size">> => 4
+        },
+        <<"status">> => 1,
+        <<"updated_at">> => iot_util:timestamp(),
+        <<"created_at">> => iot_util:timestamp()
+    }.
+
+-spec ensure_table() -> ets:tid().
+ensure_table() ->
+    case ets:info(?TABLE) of
+        undefined ->
+            try ets:new(?TABLE, [named_table, public, set]) of
+                Tid ->
+                    Tid
+            catch
+                error:badarg ->
+                    ?TABLE
+            end;
+        _ ->
+            ?TABLE
+    end.
+
+-spec get_state(term(), term()) -> term().
+get_state(Key, Default) ->
+    _ = ensure_table(),
+    case ets:lookup(?TABLE, Key) of
+        [{_, Value}] ->
+            Value;
+        [] ->
+            Default
+    end.
+
+-spec put_state(term(), term()) -> ok.
+put_state(Key, Value) ->
+    _ = ensure_table(),
+    true = ets:insert(?TABLE, {Key, Value}),
+    ok.

apps/iot/src/tcp/tcp_channel.erl

@@ -1,195 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author licheng5
-%%% @copyright (C) 2021, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 11. 1月 2021 上午12:17
-%%%-------------------------------------------------------------------
--module(tcp_channel).
--author("licheng5").
--include("message.hrl").
--behaviour(ranch_protocol).
-
-%% API
--export([pub/4, jsonrpc_call/3, command/3]).
-
--export([start_link/3, stop/2]).
-%% gen_server callbacks
--export([init/3, handle_call/3, handle_cast/2, handle_info/2, code_change/3, terminate/2]).
-
--record(state, {
-    transport,
-    socket,
-
-    uuid :: undefined | binary(),
-    %% 用户进程id
-    host_pid = undefined,
-    %% 发送消息对应的id
-    packet_id = 1 :: integer(),
-
-    %% 请求响应的对应关系
-    inflight = #{}
-}).
-
-%% 向通道中写入消息
--spec pub(Pid :: pid(), Topic :: binary(), Qos :: integer(), Content :: binary()) -> no_return().
-pub(Pid, Topic, Qos, Content) when is_pid(Pid), is_binary(Topic), is_integer(Qos), is_binary(Content) ->
-    gen_server:cast(Pid, {pub, Topic, Qos, Content}).
-
-%% 向通道中写入消息
--spec command(Pid :: pid(), CommandType :: integer(), Command :: binary()) -> no_return().
-command(Pid, CommandType, Command) when is_pid(Pid), is_integer(CommandType), is_binary(Command) ->
-    gen_server:cast(Pid, {command, CommandType, Command}).
-
-%% 向通道中写入消息
--spec jsonrpc_call(Pid :: pid(), ReceiverPid :: pid(), CallBin :: binary()) -> Ref :: reference().
-jsonrpc_call(Pid, ReceiverPid, CallBin) when is_pid(Pid), is_pid(ReceiverPid), is_binary(CallBin) ->
-    Ref = make_ref(),
-    gen_server:cast(Pid, {jsonrpc_call, ReceiverPid, Ref, CallBin}),
-    Ref.
-
-%% 关闭方法
--spec stop(Pid :: pid(), Reason :: any()) -> no_return().
-stop(undefined, _Reason) ->
-    ok;
-stop(Pid, Reason) when is_pid(Pid) ->
-    gen_server:stop(Pid, Reason, 5000).
-
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%% 逻辑处理方法
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
-start_link(Ref, Transport, Opts) ->
-    {ok, proc_lib:spawn_link(?MODULE, init, [Ref, Transport, Opts])}.
-
-init(Ref, Transport, _Opts = []) ->
-    {ok, Socket} = ranch:handshake(Ref),
-    lager:debug("[sdlan_channel] get a new connection: ~p", [Socket]),
-    Transport:setopts(Socket, [binary, {active, true}, {packet, 4}]),
-    % erlang:start_timer(?PING_TICKER, self(), ping_ticker),
-    gen_server:enter_loop(?MODULE, [], #state{transport = Transport, socket = Socket}).
-
-handle_call(_Request, _From, State) ->
-    {reply, ok, State}.
-
-%% 发送消息, 基于pub/sub机制
-handle_cast({pub, Topic, Qos, Content}, State = #state{transport = Transport, socket = Socket}) ->
-    EncPub = message_codec:encode(?MESSAGE_PUB, #pub{topic = Topic, qos = Qos, content = Content}),
-    Transport:send(Socket, <<?PACKET_CAST, EncPub/binary>>),
-    {noreply, State};
-
-%% 发送Command消息
-handle_cast({command, CommandType, Command}, State = #state{transport = Transport, socket = Socket}) ->
-    EncCommand = message_codec:encode(?MESSAGE_COMMAND, #command{command_type = CommandType, command = Command}),
-    Transport:send(Socket, <<?PACKET_CAST, EncCommand/binary>>),
-    {noreply, State};
-
-%% 推送消息
-handle_cast({jsonrpc_call, ReceiverPid, Ref, CallBin}, State = #state{transport = Transport, socket = Socket, packet_id = PacketId, inflight = Inflight}) ->
-    Transport:send(Socket, <<?PACKET_REQUEST, PacketId:32, CallBin/binary>>),
-    {noreply, State#state{packet_id = PacketId + 1, inflight = maps:put(PacketId, {ReceiverPid, Ref}, Inflight)}}.
-
-%% auth验证
-handle_info({tcp, Socket, <<?PACKET_REQUEST, PacketId:32, RequestBin/binary>>}, State = #state{transport = Transport, socket = Socket}) ->
-    {ok, #auth_request{uuid = UUID, username = Username, token = Token, salt = Salt, timestamp = Timestamp}} = message_codec:decode(RequestBin),
-    lager:debug("[ws_channel] auth uuid: ~p", [UUID]),
-    case iot_auth:check(Username, Token, UUID, Salt, Timestamp) of
-        true ->
-            case iot_api:get_host_by_uuid(UUID) of
-                undefined ->
-                    lager:warning("[ws_channel] uuid: ~p, user: ~p, host not found", [UUID, Username]),
-                    {stop, State};
-                {ok, _} ->
-                    %% 尝试启动主机的服务进程
-                    {ok, HostPid} = iot_host_sup:ensured_host_started(UUID),
-                    case iot_host:attach_channel(HostPid, self()) of
-                        ok ->
-                            %% 建立到host的monitor
-                            erlang:monitor(process, HostPid),
-                            AuthReplyBin = message_codec:encode(?MESSAGE_AUTH_REPLY, #auth_reply{code = 0, payload = <<"ok">>}),
-                            Transport:send(Socket, <<?PACKET_RESPONSE, PacketId:32, AuthReplyBin/binary>>),
-
-                            {noreply, State#state{uuid = UUID, host_pid = HostPid}};
-                        {denied, Reason} when is_binary(Reason) ->
-                            erlang:monitor(process, HostPid),
-                            AuthReplyBin = message_codec:encode(?MESSAGE_AUTH_REPLY, #auth_reply{code = 1, payload = Reason}),
-                            Transport:send(Socket, <<?PACKET_RESPONSE, PacketId:32, AuthReplyBin/binary>>),
-                            lager:debug("[ws_channel] uuid: ~p, attach channel get error: ~p, stop channel", [UUID, Reason]),
-
-                            {noreply, State#state{uuid = UUID, host_pid = HostPid}};
-
-                        {error, Reason} when is_binary(Reason) ->
-                            AuthReplyBin = message_codec:encode(?MESSAGE_AUTH_REPLY, #auth_reply{code = 2, payload = Reason}),
-                            Transport:send(Socket, <<?PACKET_RESPONSE, PacketId:32, AuthReplyBin/binary>>),
-                            lager:debug("[ws_channel] uuid: ~p, attach channel get error: ~p, stop channel", [UUID, Reason]),
-
-                            {stop, State}
-                    end
-            end;
-        false ->
-            lager:warning("[ws_channel] uuid: ~p, user: ~p, auth failed", [UUID, Username]),
-            {stop, State}
-    end;
-
-handle_info({tcp, Socket, <<?PACKET_CAST, CastBin/binary>>}, State = #state{socket = Socket, host_pid = HostPid}) when is_pid(HostPid) ->
-    {ok, CastMessage} = message_codec:decode(CastBin),
-    case CastMessage of
-        #data{} = Data ->
-            iot_host:handle(HostPid, {data, Data});
-        #task_event_stream{task_id = TaskId, type = <<"close">>, stream = Reason} ->
-            iot_event_stream_observer:stream_close(TaskId, Reason);
-        #task_event_stream{task_id = TaskId, type = Type, stream = Stream} ->
-            lager:debug("[tcp_channel] get task_id: ~p, type: ~ts, stream: ~ts", [TaskId, Type, Stream]),
-            iot_event_stream_observer:stream_data(TaskId, Type, Stream)
-    end,
-    {noreply, State};
-
-%handle_info({tcp, Socket, <<?PACKET_PING, PingData/binary>>}, State = #state{socket = Socket, host_pid = HostPid}) when is_pid(HostPid) ->
-%    Ping = message_pb:decode_msg(PingData, ping),
-%    iot_host:handle(HostPid, {ping, Ping}),
-%    {noreply, State};
-
-%% 主机端的消息响应
-handle_info({tcp, Socket, <<?PACKET_RESPONSE, PacketId:32, ResponseBin/binary>>}, State = #state{socket = Socket, inflight = Inflight}) when PacketId > 0 ->
-    {ok, RpcReply} = message_codec:decode(ResponseBin),
-    case maps:take(PacketId, Inflight) of
-        error ->
-            {noreply, State};
-        {{ReceiverPid, Ref}, NInflight} ->
-            case is_pid(ReceiverPid) andalso is_process_alive(ReceiverPid) of
-                true ->
-                    ReceiverPid ! {jsonrpc_reply, Ref, RpcReply};
-                false ->
-                    lager:warning("[ws_channel] get async_call_reply message: ~p, packet_id: ~p, but receiver_pid is deaded", [RpcReply, PacketId])
-            end,
-            {noreply, State#state{inflight = NInflight}}
-    end;
-
-handle_info({tcp_error, Sock, Reason}, State = #state{socket = Sock}) ->
-    lager:notice("[sdlan_channel] tcp_error: ~p", [Reason]),
-    {stop, normal, State};
-
-handle_info({tcp_closed, Sock}, State = #state{socket = Sock}) ->
-    lager:notice("[sdlan_channel] tcp_closed"),
-    {stop, normal, State};
-
-%% 关闭当前通道
-handle_info({stop, Reason}, State) ->
-    {stop, Reason, State};
-
-%% 主机进程挂掉的时候，需要关闭掉链接
-handle_info({'DOWN', _, process, HostPid, Reason}, State = #state{uuid = UUID, host_pid = HostPid}) ->
-    lager:debug("[ws_channel] uuid: ~p, channel will close because host exited with reason: ~p", [UUID, Reason]),
-    {stop, State};
-
-handle_info(Info, State) ->
-    lager:warning("[sdlan_channel] get a unknown message: ~p, channel will closed, state: ~p", [Info, State]),
-    {noreply, State}.
-
-terminate(Reason, #state{}) ->
-    lager:warning("[sdlan_channel] stop with reason: ~p", [Reason]),
-    ok.
-
-code_change(_OldVsn, State, _Extra) ->
-    {ok, State}.

apps/iot/src/transport/ssl_channel.erl

@@ -0,0 +1,369 @@
+%%%-------------------------------------------------------------------
+%%% @author licheng5
+%%% @copyright (C) 2021, <COMPANY>
+%%% @doc
+%%%
+%%% @end
+%%% Created : 11. 1月 2021 上午12:17
+%%%-------------------------------------------------------------------
+-module(ssl_channel).
+-author("licheng5").
+-behaviour(ranch_protocol).
+
+-define(INFLIGHT_TIMEOUT, 60000).
+-define(SSL_IDLE_TIMEOUT, 120000).
+
+%% API
+-export([pub/4, container_call/4, cancel_command_call/2]).
+
+-export([start_link/3, stop/2]).
+%% gen_server callbacks
+-export([init/3, handle_call/3, handle_cast/2, handle_info/2, code_change/3, terminate/2]).
+
+-record(state, {
+    transport,
+    socket,
+
+    uuid :: undefined | binary(),
+    is_authed = false :: boolean(),
+    %% 用户进程id
+    host_pid = undefined,
+
+    %% iot 发起的 command 与 command_response 的对应关系
+    inflight = #{},
+
+    idle_timer_ref :: undefined | reference()
+}).
+
+-record(inflight_command, {
+    receiver_pid :: undefined | pid(),
+    timer_ref :: reference()
+}).
+
+-type request_ref() :: binary().
+
+%% 向通道中写入消息
+-spec pub(Pid :: pid(), Topic :: binary(), Qos :: integer(), Content :: binary()) -> ok.
+pub(Pid, Topic, Qos, Content) when is_pid(Pid), is_binary(Topic), is_integer(Qos), is_binary(Content) ->
+    gen_server:cast(Pid, {pub, Topic, Qos, Content}).
+
+-spec container_call(Pid :: pid(), ReceiverPid :: pid(), Ref :: request_ref(), Request :: map()) -> ok.
+container_call(Pid, ReceiverPid, Ref, Request) when is_pid(Pid), is_pid(ReceiverPid), is_binary(Ref), is_map(Request) ->
+    gen_server:cast(Pid, {command_call, ReceiverPid, Ref,  {container, Request}}),
+    ok.
+
+-spec cancel_command_call(Pid :: pid(), Ref :: request_ref()) -> ok.
+cancel_command_call(Pid, Ref) when is_pid(Pid), is_binary(Ref) ->
+    gen_server:call(Pid, {cancel_command_call, Ref}).
+
+%% 关闭方法
+-spec stop(Pid :: undefined | pid(), Reason :: any()) -> ok.
+stop(undefined, _Reason) ->
+    ok;
+stop(Pid, Reason) when is_pid(Pid) ->
+    gen_server:stop(Pid, Reason, 5000).
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% 逻辑处理方法
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+start_link(Ref, Transport, Opts) ->
+    {ok, proc_lib:spawn_link(?MODULE, init, [Ref, Transport, Opts])}.
+
+init(Ref, Transport, _Opts = []) ->
+    ok = iot_log:set_metadata(),
+    {ok, Socket} = ranch:handshake(Ref),
+    logger:debug("[ssl_channel] get a new connection: ~p", [Socket]),
+    Transport:setopts(Socket, [binary, {active, true}, {packet, 4}]),
+    IdleTimerRef = start_idle_timer(),
+    gen_server:enter_loop(?MODULE, [], #state{transport = Transport, socket = Socket, idle_timer_ref = IdleTimerRef}).
+
+handle_call({cancel_command_call, Ref}, _From, State = #state{inflight = Inflight}) ->
+    case maps:take(Ref, Inflight) of
+        {#inflight_command{timer_ref = TimerRef}, NInflight} ->
+            erlang:cancel_timer(TimerRef),
+            {reply, ok, State#state{inflight = NInflight}};
+        error ->
+            {reply, ok, State}
+    end;
+handle_call(_Request, _From, State) ->
+    {reply, ok, State}.
+
+%% 发送消息, 基于pub/sub机制
+handle_cast({pub, Topic, Qos, Content}, State = #state{transport = Transport, socket = Socket}) ->
+    Packet = term_to_binary({<<"message">>, {<<"pub">>, #{<<"topic">> => Topic, <<"qos">> => Qos, <<"content">> => Content}}}),
+    case Transport:send(Socket, Packet) of
+        ok ->
+            {noreply, State};
+        {error, Reason} ->
+            logger:warning("[ssl_channel] send pub failed, reason: ~p", [Reason]),
+            {stop, {send_failed, Reason}, State}
+    end;
+
+%% iot 请求 efka 时使用 command/command_response 语义。
+handle_cast({command_call, ReceiverPid, Ref, Body}, State = #state{transport = Transport, socket = Socket, inflight = Inflight}) ->
+    Packet = term_to_binary({<<"command">>, Ref, encode_command_body(Body)}),
+    case Transport:send(Socket, Packet) of
+        ok ->
+            TimerRef = erlang:start_timer(?INFLIGHT_TIMEOUT, self(), {command_timeout, Ref}),
+            CommandInfo = #inflight_command{receiver_pid = ReceiverPid, timer_ref = TimerRef},
+            {noreply, State#state{inflight = maps:put(Ref, CommandInfo, Inflight)}};
+        {error, Reason} ->
+            logger:warning("[ssl_channel] send command failed, ref: ~p, reason: ~p", [Ref, Reason]),
+            deliver_command_error(ReceiverPid, Ref, {send_failed, Reason}),
+            {stop, {send_failed, Reason}, State}
+    end.
+
+handle_info({timeout, TimerRef, {command_timeout, Ref}}, State = #state{inflight = Inflight}) ->
+    case maps:get(Ref, Inflight, undefined) of
+        #inflight_command{receiver_pid = ReceiverPid, timer_ref = TimerRef} ->
+            logger:warning("[ws_channel] command timeout, ref: ~p", [Ref]),
+            reply_command_timeout(ReceiverPid, Ref),
+            {noreply, State#state{inflight = maps:remove(Ref, Inflight)}};
+        _ ->
+            {noreply, State}
+    end;
+
+handle_info({timeout, TimerRef, ssl_idle_timeout}, State = #state{idle_timer_ref = TimerRef}) ->
+    logger:notice("[ssl_channel] ssl channel idle timeout"),
+    {stop, ssl_idle_timeout, State#state{idle_timer_ref = undefined}};
+handle_info({timeout, _TimerRef, ssl_idle_timeout}, State) ->
+    {noreply, State};
+
+%% 关闭当前通道
+handle_info({stop, Reason}, State) ->
+    {stop, Reason, State};
+
+%% 主机进程挂掉的时候，需要关闭掉链接
+handle_info({'DOWN', _, process, HostPid, Reason}, State = #state{uuid = UUID, host_pid = HostPid}) ->
+    logger:debug("[ws_channel] uuid: ~p, channel will close because host exited with reason: ~p", [UUID, Reason]),
+    {stop, Reason, State};
+
+handle_info({ssl, Socket, PacketBin}, State = #state{socket = Socket}) when is_binary(PacketBin) ->
+    State1 = reset_idle_timer(State),
+    try binary_to_term(PacketBin, [safe]) of
+        {<<"request">>, Ref, Body} ->
+            handle_request_frame(Ref, Body, State1);
+        {<<"message">>, Body} ->
+            handle_message_frame(Body, State1);
+        {<<"command_response">>, Ref, Response} ->
+            handle_command_response_frame(Ref, Response, State1);
+        Other ->
+            logger:warning("[ssl_channel] unsupported packet: ~p", [Other]),
+            {stop, bad_packet, State1}
+    catch error:Error ->
+        logger:warning("[ssl_channel] binary_to_term get error: ~p, packet_size: ~p", [Error, byte_size(PacketBin)]),
+        {stop, bad_packet, State1}
+    end;
+
+handle_info({ssl_closed, Socket}, State = #state{socket = Socket}) ->
+    logger:notice("[ssl_channel] ssl socket closed"),
+    {stop, normal, State};
+
+handle_info({ssl_error, Socket, Reason}, State = #state{socket = Socket}) ->
+    logger:notice("[ssl_channel] ssl socket error: ~p", [Reason]),
+    {stop, normal, State};
+
+handle_info({ssl_passive, Socket}, State = #state{transport = Transport, socket = Socket}) ->
+    ok = Transport:setopts(Socket, [{active, true}]),
+    {noreply, State};
+
+handle_info(Info, State) ->
+    logger:warning("[ssl_channel] get a unknown message: ~p, state: ~p", [Info, State]),
+    {noreply, State}.
+
+terminate(Reason, #state{inflight = Inflight, transport = Transport, socket = Socket, idle_timer_ref = IdleTimerRef}) ->
+    cancel_timer(IdleTimerRef),
+    maps:foreach(fun(Ref, CommandInfo) -> reply_command_closed(Ref, CommandInfo, Reason) end, Inflight),
+    Transport:close(Socket),
+    logger:warning("[ssl_channel] stop with reason: ~p", [Reason]),
+    ok.
+
+code_change(_OldVsn, State, _Extra) ->
+    {ok, State}.
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%%%% helper methods
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+-spec handle_request_frame(request_ref(), tuple(), #state{}) -> {noreply, #state{}} | {stop, term(), #state{}}.
+handle_request_frame(Ref, {<<"auth_request">>, #{<<"uuid">> := UUID}}, State = #state{is_authed = true}) ->
+    logger:warning("[ws_channel] repeated auth request, ref: ~p, uuid: ~p, close channel", [Ref, UUID]),
+    {stop, repeated_auth, State};
+handle_request_frame(Ref, {<<"auth_request">>, #{<<"uuid">> := UUID, <<"token">> := Token, <<"timestamp">> := Timestamp}}, State = #state{transport = Transport, socket = Socket}) ->
+    maybe
+        ok ?= auth(Token, UUID, Timestamp),
+        {ok, HostPid} ?= iot_host:lookup_pid(UUID),
+        ok ?= iot_host:attach_channel(HostPid, self()),
+        erlang:monitor(process, HostPid),
+        ok = send_reply_frame(Transport, Socket, Ref, {<<"auth_response">>, <<"ok">>}),
+        logger:debug("[ws_channel] auth uuid: ~p", [UUID]),
+        {noreply, State#state{uuid = UUID, is_authed = true, host_pid = HostPid}}
+    else
+        {error, Reason} ->
+            logger:warning("[ws_channel] uuid: ~p, auth failed with reason: ~p", [UUID, Reason]),
+            case send_reply_frame(Transport, Socket, Ref, {<<"auth_response">>, {<<"error">>, {<<"failed">>, Reason}}}) of
+                ok ->
+                    {stop, Reason, State};
+                {error, SendReason} ->
+                    logger:warning("[ws_channel] uuid: ~p, send auth failed response failed: ~p", [UUID, SendReason]),
+                    {stop, {send_failed, SendReason}, State}
+            end
+    end;
+handle_request_frame(Ref, {<<"container">>, ContainerCommand}, State) ->
+    logger:warning("[ws_channel] unsupported request message type: container, ref: ~p, command: ~p", [Ref, ContainerCommand]),
+    {stop, normal, State};
+handle_request_frame(Ref, Body, State) ->
+    logger:warning("[ws_channel] unsupported request body, ref: ~p, body: ~p", [Ref, Body]),
+    {stop, normal, State}.
+
+-spec handle_message_frame(term(), #state{}) ->
+    {noreply, #state{}} | {stop, term(), #state{}}.
+handle_message_frame(<<"ping">>, State = #state{transport = Transport, socket = Socket}) ->
+    Packet = term_to_binary({<<"message">>, <<"pong">>}),
+    case Transport:send(Socket, Packet) of
+        ok ->
+            {noreply, State};
+        {error, Reason} ->
+            logger:warning("[ssl_channel] send pong failed, reason: ~p", [Reason]),
+            {stop, {send_failed, Reason}, State}
+    end;
+handle_message_frame({<<"data">>, #{<<"route_key">> := RouteKey, <<"metric">> := Metric}}, State = #state{host_pid = HostPid}) when is_pid(HostPid) ->
+    iot_host:handle(HostPid, {data, RouteKey, Metric}),
+    {noreply, State};
+handle_message_frame({<<"task_event">>, Event}, State = #state{uuid = UUID, host_pid = HostPid}) when is_binary(UUID), is_pid(HostPid) ->
+    handle_event_stream_frame(UUID, Event),
+    {noreply, State};
+handle_message_frame(Body, State) ->
+    logger:warning("[ssl_channel] unsupported message body: ~p", [Body]),
+    {noreply, State}.
+
+-spec handle_event_stream_frame(binary(), map()) -> ok.
+handle_event_stream_frame(UUID, #{<<"task_id">> := TaskId, <<"type">> := <<"close">>, <<"stream">> := Reason}) ->
+    logger:debug("[ssl_channel] get uuid: ~p, task_id: ~p, close with reason: ~p", [UUID, TaskId, Reason]),
+    iot_container_task_sup:close(UUID, TaskId, Reason);
+handle_event_stream_frame(UUID, #{<<"task_id">> := TaskId, <<"type">> := Type, <<"stream">> := Stream}) ->
+    logger:debug("[ssl_channel] get uuid: ~p, task_id: ~p, type: ~ts, stream: ~ts", [UUID, TaskId, Type, Stream]),
+    iot_container_task_sup:stream(UUID, TaskId, Type, Stream);
+handle_event_stream_frame(UUID, Event) ->
+    logger:warning("[ssl_channel] invalid task_event, uuid: ~p, event: ~p", [UUID, Event]),
+    ok.
+
+-spec handle_command_response_frame(request_ref(), tuple(), #state{}) ->
+    {noreply, #state{}}.
+handle_command_response_frame(Ref, Reply, State = #state{inflight = Inflight}) when is_binary(Ref) ->
+    case maps:take(Ref, Inflight) of
+        error ->
+            {noreply, State};
+        {#inflight_command{receiver_pid = ReceiverPid, timer_ref = TimerRef}, NInflight} ->
+            erlang:cancel_timer(TimerRef),
+            deliver_command_response(ReceiverPid, Ref, Reply),
+            {noreply, State#state{inflight = NInflight}}
+    end;
+handle_command_response_frame(Ref, Reply, State) ->
+    logger:warning("[ws_channel] unexpected command_response frame, ref: ~p, reply: ~p", [Ref, Reply]),
+    {noreply, State}.
+
+-spec send_reply_frame(module(), any(), request_ref(), tuple()) -> ok | {error, term()}.
+send_reply_frame(Transport, Socket, Ref, Reply) ->
+    Packet = term_to_binary({<<"response">>, Ref, Reply}),
+    Transport:send(Socket, Packet).
+
+-spec decode_command_response({binary(), term()} | tuple()) ->
+    ok | {ok, term()} | {error, term()}.
+decode_command_response({<<"container">>, <<"ok">>}) ->
+    ok;
+decode_command_response({<<"container">>, {<<"ok">>, Result}}) ->
+    {ok, Result};
+decode_command_response({<<"container">>, {<<"error">>, Reason}}) ->
+    {error, Reason};
+decode_command_response(_Reply) ->
+    {error, invalid_response}.
+
+-spec deliver_command_response(undefined | pid(), request_ref(), tuple()) -> ok.
+deliver_command_response(undefined, _Ref, _Reply) ->
+    ok;
+deliver_command_response(ReceiverPid, Ref, Reply) when is_pid(ReceiverPid) ->
+    case is_process_alive(ReceiverPid) of
+        true ->
+            ReceiverPid ! {command_reply, Ref, decode_command_response(Reply)};
+        false ->
+            logger:warning("[ws_channel] get command_response: ~p, ref: ~p, but receiver_pid is deaded", [Reply, Ref])
+    end.
+
+-spec deliver_command_error(undefined | pid(), request_ref(), term()) -> ok.
+deliver_command_error(ReceiverPid, Ref, Reason) when is_pid(ReceiverPid) ->
+    ReceiverPid ! {command_reply, Ref, {error, Reason}},
+    ok;
+deliver_command_error(_ReceiverPid, _Ref, _Reason) ->
+    ok.
+
+-spec reply_command_timeout(undefined | pid(), request_ref()) -> ok.
+reply_command_timeout(ReceiverPid, Ref) when is_pid(ReceiverPid) ->
+    ReceiverPid ! {command_reply, Ref, {error, timeout}},
+    ok;
+reply_command_timeout(_ReceiverPid, _Ref) ->
+    ok.
+
+-spec reply_command_closed(request_ref(), #inflight_command{}, term()) -> ok.
+reply_command_closed(Ref, #inflight_command{receiver_pid = ReceiverPid}, Reason) when is_pid(ReceiverPid) ->
+    ReceiverPid ! {command_reply, Ref, {error, {channel_closed, Reason}}},
+    ok;
+reply_command_closed(_Ref, _CommandInfo, _Reason) ->
+    ok.
+
+%% 检测token是否是合法值
+-spec auth(Token :: binary(), UUID :: binary(), Timestamp :: integer()) ->
+    ok | {error, Reason :: binary()}.
+auth(Token, UUID, Timestamp) when is_binary(Token), is_binary(UUID), is_integer(Timestamp) ->
+    %% 1分钟内有效
+    Now = iot_util:current_time(),
+    case Timestamp =< Now andalso Now - Timestamp =< 60 of
+        true ->
+            case efka_client_store:auth(UUID, Token) of
+                true ->
+                    ok;
+                false ->
+                    {error, <<"invalid token">>}
+            end;
+        false ->
+            {error, <<"invalid timestamp">>}
+    end.
+
+-spec encode_command_body({container, map()}) -> {binary(), map()}.
+encode_command_body({container, Request}) ->
+    {<<"container">>, safe_term(Request)}.
+
+-spec safe_term(term()) -> term().
+safe_term(true) ->
+    true;
+safe_term(false) ->
+    false;
+safe_term(undefined) ->
+    undefined;
+safe_term(Value) when is_atom(Value) ->
+    atom_to_binary(Value, utf8);
+safe_term(Value) when is_map(Value) ->
+    maps:from_list([{safe_term(K), safe_term(V)} || {K, V} <- maps:to_list(Value)]);
+safe_term(Value) when is_list(Value) ->
+    [safe_term(Item) || Item <- Value];
+safe_term(Value) when is_tuple(Value) ->
+    list_to_tuple([safe_term(Item) || Item <- tuple_to_list(Value)]);
+safe_term(Value) ->
+    Value.
+
+-spec start_idle_timer() -> reference().
+start_idle_timer() ->
+    erlang:start_timer(?SSL_IDLE_TIMEOUT, self(), ssl_idle_timeout).
+
+-spec reset_idle_timer(#state{}) -> #state{}.
+reset_idle_timer(State = #state{idle_timer_ref = TimerRef}) ->
+    cancel_timer(TimerRef),
+    State#state{idle_timer_ref = start_idle_timer()}.
+
+-spec cancel_timer(undefined | reference()) -> ok.
+cancel_timer(undefined) ->
+    ok;
+cancel_timer(TimerRef) ->
+    _ = erlang:cancel_timer(TimerRef),
+    ok.

apps/iot/src/transport/udp_server.erl

@@ -1,19 +1,22 @@
 %%%-------------------------------------------------------------------
 %%% @author anlicheng
-%%% @copyright (C) 2025, <COMPANY>
+%%% @copyright (C) 2026, <COMPANY>
 %%% @doc
 %%%
 %%% @end
-%%% Created : 26. 9月 2025 12:19
+%%% Created : 09. 5月 2026 18:02
 %%%-------------------------------------------------------------------
--module(iot_event_stream_observer).
+-module(udp_server).
 -author("anlicheng").
 
 -behaviour(gen_server).
 
+-define(HEARTBEAT_VERSION, 1).
+-define(HEARTBEAT_NONCE_BYTES, 16).
+-define(HEARTBEAT_MAC_BYTES, 32).
+
 %% API
 -export([start_link/0]).
--export([add_listener/2, stream_data/3, stream_close/2]).
 
 %% gen_server callbacks
 -export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
@@ -21,25 +24,13 @@
 -define(SERVER, ?MODULE).
 
 -record(state, {
-    listeners = #{}
+    socket :: gen_udp:socket()
 }).
 
 %%%===================================================================
 %%% API
 %%%===================================================================
 
--spec add_listener(ListenerPid :: pid(), TaskId :: integer()) -> ok.
-add_listener(ListenerPid, TaskId) when is_pid(ListenerPid), is_integer(TaskId) ->
-    gen_server:call(?SERVER, {add_listener, ListenerPid, TaskId}).
-
--spec stream_data(TaskId :: integer(), Type :: binary(), Stream :: binary()) -> no_return().
-stream_data(TaskId, Type, Stream) when is_integer(TaskId), is_binary(Type), is_binary(Stream) ->
-    gen_server:cast(?SERVER, {stream_data, TaskId, Type, Stream}).
-
--spec stream_close(TaskId :: integer(), Reason :: binary()) -> no_return().
-stream_close(TaskId, Reason) when is_integer(TaskId), is_binary(Reason) ->
-    gen_server:cast(?SERVER, {stream_close, TaskId, Reason}).
-
 %% @doc Spawns the server and registers the local name (unique)
 -spec(start_link() ->
     {ok, Pid :: pid()} | ignore | {error, Reason :: term()}).
@@ -56,7 +47,12 @@ start_link() ->
     {ok, State :: #state{}} | {ok, State :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term()} | ignore).
 init([]) ->
-    {ok, #state{}}.
+    ok = iot_log:set_metadata(),
+    {ok, UdpServerProps} = application:get_env(iot, udp_server),
+    Port = proplists:get_value(port, UdpServerProps),
+    logger:debug("[udp_server] start at port: ~p", [Port]),
+    {ok, Socket} = gen_udp:open(Port, [binary, {active, true}]),
+    {ok, #state{socket = Socket}}.
 
 %% @private
 %% @doc Handling call messages
@@ -68,9 +64,8 @@ init([]) ->
     {noreply, NewState :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term(), Reply :: term(), NewState :: #state{}} |
     {stop, Reason :: term(), NewState :: #state{}}).
-handle_call({add_listener, ListenerPid, TaskId}, _From, State = #state{listeners = Listeners}) ->
+handle_call(_Request, _From, State = #state{}) ->
-    erlang:monitor(process, ListenerPid),
+    {reply, ok, State}.
-    {reply, ok, State#state{listeners = maps:put(TaskId, ListenerPid, Listeners)}}.
 
 %% @private
 %% @doc Handling cast messages
@@ -78,21 +73,7 @@ handle_call({add_listener, ListenerPid, TaskId}, _From, State = #state{listeners
     {noreply, NewState :: #state{}} |
     {noreply, NewState :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term(), NewState :: #state{}}).
-handle_cast({stream_data, TaskId, Type, Stream}, State = #state{listeners = Listeners}) ->
+handle_cast(_Request, State = #state{}) ->
-    case maps:find(TaskId, Listeners) of
-        error ->
-            ok;
-        {ok, ListenerPid} ->
-            is_process_alive(ListenerPid) andalso ListenerPid ! {stream_data, TaskId, Type, Stream}
-    end,
-    {noreply, State};
-handle_cast({stream_close, TaskId, Reason}, State = #state{listeners = Listeners}) ->
-    case maps:find(TaskId, Listeners) of
-        error ->
-            ok;
-        {ok, ListenerPid} ->
-            is_process_alive(ListenerPid) andalso ListenerPid ! {stream_close, TaskId, Reason}
-    end,
     {noreply, State}.
 
 %% @private
@@ -101,9 +82,12 @@ handle_cast({stream_close, TaskId, Reason}, State = #state{listeners = Listeners
     {noreply, NewState :: #state{}} |
     {noreply, NewState :: #state{}, timeout() | hibernate} |
     {stop, Reason :: term(), NewState :: #state{}}).
-handle_info({'DOWN', _Ref, process, Pid, _Reason}, State = #state{listeners = Listeners}) ->
+handle_info({udp, Socket, Ip, Port, Packet}, State = #state{socket = Socket}) ->
-    NListeners = maps:filter(fun(_, ListenerPid) -> ListenerPid /= Pid end, Listeners),
+    handle_heartbeat_packet(Packet, {Ip, Port}),
-    {noreply, State#state{listeners = NListeners}}.
+    {noreply, State};
+handle_info(Info, State = #state{}) ->
+    logger:warning("[udp_server] ignore unknown info: ~p", [Info]),
+    {noreply, State}.
 
 %% @private
 %% @doc This function is called by a gen_server when it is about to
@@ -112,7 +96,8 @@ handle_info({'DOWN', _Ref, process, Pid, _Reason}, State = #state{listeners = Li
 %% with Reason. The return value is ignored.
 -spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
     State :: #state{}) -> term()).
-terminate(_Reason, _State = #state{}) ->
+terminate(_Reason, _State = #state{socket = Socket}) ->
+    gen_udp:close(Socket),
     ok.
 
 %% @private
@@ -126,3 +111,31 @@ code_change(_OldVsn, State = #state{}, _Extra) ->
 %%%===================================================================
 %%% Internal functions
 %%%===================================================================
+
+-spec handle_heartbeat_packet(binary(), term()) -> ok.
+handle_heartbeat_packet(Packet, Peer) when is_binary(Packet) ->
+    case decode_heartbeat_packet(Packet) of
+        {ok, UUID, Timestamp, Payload, Mac} ->
+            case efka_client_store:verify_heartbeat(UUID, Timestamp, Payload, Mac) of
+                true ->
+                    Pid = iot_host:get_pid(UUID),
+                    iot_host:heartbeat(Pid);
+                false ->
+                    logger:warning("[udp_server] ignore invalid heartbeat auth from peer: ~p, uuid: ~p", [Peer, UUID]),
+                    ok
+            end;
+        error ->
+            logger:warning("[udp_server] ignore invalid heartbeat packet from peer: ~p, packet: ~p", [Peer, Packet]),
+            ok
+    end.
+
+-spec decode_heartbeat_packet(binary()) ->
+    {ok, UUID :: binary(), Timestamp :: integer(), Payload :: binary(), Mac :: binary()} | error.
+decode_heartbeat_packet(Packet = <<?HEARTBEAT_VERSION:8, UUIDLen:16, UUID:UUIDLen/binary,
+    Timestamp:64/unsigned-big, _Nonce:?HEARTBEAT_NONCE_BYTES/binary, _Mac:?HEARTBEAT_MAC_BYTES/binary>>)
+    when UUIDLen > 0 ->
+    PayloadSize = byte_size(Packet) - ?HEARTBEAT_MAC_BYTES,
+    <<Payload:PayloadSize/binary, Mac:?HEARTBEAT_MAC_BYTES/binary>> = Packet,
+    {ok, UUID, Timestamp, Payload, Mac};
+decode_heartbeat_packet(_Packet) ->
+    error.

apps/iot/src/util/http_client.erl

@@ -1,34 +0,0 @@
-%%%-------------------------------------------------------------------
-%%% @author anlicheng
-%%% @copyright (C) 2025, <COMPANY>
-%%% @doc
-%%%
-%%% @end
-%%% Created : 18. 8月 2025 18:40
-%%%-------------------------------------------------------------------
--module(http_client).
--author("anlicheng").
-
-%% API
--export([post/3]).
-
-%% Headers = [
-%%   {<<"content-type">>, <<"application/json">>}
-%% ]
--spec post(Url :: string(), Headers :: list(), Body :: binary()) -> {ok, RespBody :: binary()} | {error, Reason :: any()}.
-post(Url, Headers, Body) when is_list(Url), is_list(Headers), is_binary(Body) ->
-    case hackney:request(post, Url, Headers, Body, [{pool, false}]) of
-        {ok, 200, _, ClientRef} ->
-            {ok, RespBody} = hackney:body(ClientRef),
-            lager:debug("[iot_api] send body: ~p, get error is: ~p", [Body, RespBody]),
-            hackney:close(ClientRef),
-            {ok, RespBody};
-        {ok, HttpCode, _, ClientRef} ->
-            {ok, RespBody} = hackney:body(ClientRef),
-            hackney:close(ClientRef),
-            lager:warning("[iot_api] send body: ~p, get error is: ~p", [Body, {HttpCode, RespBody}]),
-            {error, {HttpCode, RespBody}};
-        {error, Reason} ->
-            lager:warning("[iot_api] send body: ~p, get error is: ~p", [Body, Reason]),
-            {error, Reason}
-    end.

config/sys-dev.config

@@ -2,114 +2,75 @@
   {iot, [
     {http_server, [
       {port, 18090},
-      {acceptors, 500},
+      {acceptors, 5},
-      {max_connections, 10240},
+      {max_connections, 1024},
-      {backlog, 10240}
+      {backlog, 1024}
     ]},
 
-    {tcp_server, [
+    {ssl_server, [
-      {port, 18092},
+      {port, 1443},
-      {acceptors, 500},
+      {acceptors, 5},
-      {max_connections, 10240},
+      {max_connections, 1024},
-      {backlog, 10240}
+      {backlog, 1024}
-    ]},
-
-    {redis_server, [
-      {port, 16379},
-      {acceptors, 500},
-      {max_connections, 10240},
-      {backlog, 10240}
     ]},
 
     {udp_server, [
       {port, 18080}
     ]},
 
-    {api_url, "http://100.123.0.4/api/v1"},
+    {api_url, "http://127.0.0.1:18090/simulator"}
+  ]},
+
+  {endpoint, [
     %% 支持的协议
     {endpoints, [
+      {root_dir, "/usr/local/code/database/"},
       {support_protocols, [
         http
       ]}
     ]},
 
-    %% 目标服务器地址
+    {endpoint_log, [
-    {emqx_server, [
+      {path, "${endpoint_root}/endpoint_log/unmatched_publish.log"},
-      {host, {39, 98, 184, 67}},
+      {max_bytes, 10485760},
-      {port, 1883},
+      {max_files, 10}
-      {tcp_opts, []},
-      {username, "test"},
-      {password, "test1234"},
-      {keepalive, 86400},
-      {retry_interval, 5}
-    ]},
-
-    %% 权限检验时的预埋token
-    {pre_tokens, [
-      {<<"test">>, <<"iot2023">>}
     ]}
-
-    %{pools, [
-    %  %% mysql连接池配置
-    %  {mysql_iot,
-    %    [{size, 10}, {max_overflow, 20}, {worker_module, mysql}],
-    %    [
-    %      {host, "47.111.101.3"},
-    %      {port, 3306},
-    %      {user, "root"},
-    %      {connect_mode, synchronous},
-    %      {keep_alive, true},
-    %      {password, "r3a-7Qrh#3Q"},
-    %      {database, "nannong_demo"}
-    %    ]
-    %  },
-
-    %  %% redis连接池
-    %  {redis_pool,
-    %    [{size, 10}, {max_overflow, 20}, {worker_module, eredis}],
-    %    [
-    %      {host, "127.0.0.1"},
-    %      {port, 6379},
-    %      {database, 1}
-    %    ]
-    %  }
-    %]}
-
   ]},
 
+  %% 系统日志配置，使用 OTP logger
+  {kernel, [
+    %% 设置 Logger 的 primary log level
+    {logger_level, debug},
+    {logger, [
+      {handler, default, logger_std_h,
+        #{
+          level => debug,
+          filter_default => stop,
+          filters => [
+            {iot_domain, {fun logger_filters:domain/2, {log, sub, [iot]}}},
+            {endpoint_domain, {fun logger_filters:domain/2, {log, sub, [endpoint]}}}
+          ],
+          formatter => {logger_formatter, #{template => [time, " [", level, "] ", msg, "\n"]}}
+        }
+      },
 
-  %% 系统日志配置，系统日志为lager, 支持日志按日期自动分割
+      {handler, disk, logger_disk_log_h,
-  {lager, [
+        #{
-    {colored, true},
+          level => debug,
-    %% Whether to write a crash log, and where. Undefined means no crash logger.
+          filter_default => stop,
-    {crash_log, "trade_hub.crash.log"},
+          filters => [
-    %% Maximum size in bytes of events in the crash log - defaults to 65536
+            {iot_domain, {fun logger_filters:domain/2, {log, sub, [iot]}}},
-    {crash_log_msg_size, 65536},
+            {endpoint_domain, {fun logger_filters:domain/2, {log, sub, [endpoint]}}}
-    %% Maximum size of the crash log in bytes, before its rotated, set
+          ],
-    %% to 0 to disable rotation - default is 0
+          config => #{
-    {crash_log_size, 10485760},
+          file => "log/debug.log",
-    %% What time to rotate the crash log - default is no time
+          max_no_files => 10,
-    %% rotation. See the README for a description of this format.
+          max_no_bytes => 524288000
-    {crash_log_date, "$D0"},
+        },
-    %% Number of rotated crash logs to keep, 0 means keep only the
+          formatter => {logger_formatter, #{template => [time, " [", level, "] ", msg, "\n"]}}
-    %% current one - default is 0
+        }
-    {crash_log_count, 5},
+      }
-    %% Whether to redirect error_logger messages into lager - defaults to true
-    {error_logger_redirect, true},
 
-    %% How big the gen_event mailbox can get before it is switched into sync mode
-    {async_threshold, 20},
-    %% Switch back to async mode, when gen_event mailbox size decrease from `async_threshold'
-    %% to async_threshold - async_threshold_window
-    {async_threshold_window, 5},
-
-    {handlers, [
-      %% debug | info | warning | error, 日志级别
-      {lager_console_backend, debug},
-      {lager_file_backend, [{file, "debug.log"}, {level, debug}, {size, 314572800}]},
-      {lager_file_backend, [{file, "notice.log"}, {level, notice}, {size, 314572800}]},
-      {lager_file_backend, [{file, "error.log"}, {level, error}, {size, 314572800}]},
-      {lager_file_backend, [{file, "info.log"}, {level, info}, {size, 314572800}]}
     ]}
   ]}
 

config/sys-prod.config

@@ -2,83 +2,74 @@
   {iot, [
     {http_server, [
       {port, 18080},
-      {acceptors, 500},
+      {acceptors, 5},
-      {max_connections, 10240},
+      {max_connections, 1024},
-      {backlog, 10240}
+      {backlog, 1024}
     ]},
 
-    {redis_server, [
+    {ssl_server, [
-      {port, 16379},
+      {port, 1443},
-      {acceptors, 500},
+      {acceptors, 5},
-      {max_connections, 10240},
+      {max_connections, 1024},
-      {backlog, 10240}
+      {backlog, 1024}
     ]},
 
     {udp_server, [
       {port, 18080}
     ]},
 
-    %% 权限检验时的预埋token
+    {api_url, "https://lgsiot.njau.edu.cn/api/v1/taskLog"}
-    {pre_tokens, [
-      {<<"test">>, <<"iot2023">>}
-    ]},
-
-    {api_url, "https://lgsiot.njau.edu.cn/api/v1/taskLog"},
-
-    {influxdb, [
-      {host, "172.19.0.4"},
-      {port, 8086},
-      {token, <<"A-ZRjqMK_7NR45lXXEiR7AEtYCd1ETzq9Z61FTMQLb5O4-1hSf8sCrjdPB84e__xsrItKHL3qjJALgbYN-H_VQ==">>}
-    ]}
-
-    %{pools, [
-    %  %% redis连接池
-    %  {redis_pool,
-    %    [{size, 10}, {max_overflow, 20}, {worker_module, eredis}],
-    %    [
-    %      {host, "172.30.6.175"},
-    %      {port, 26379},
-    %      {database, 1}
-    %    ]
-    %  }
-
-    %]}
-
   ]},
 
+  {endpoint, [
+    {endpoints, [
+      {root_dir, "/var/lib/endpoint/database/"},
+      {support_protocols, [
+        http
+      ]}
+    ]},
 
-  %% 系统日志配置，系统日志为lager, 支持日志按日期自动分割
+    {endpoint_log, [
-  {lager, [
+      {path, "/var/lib/endpoint/endpoint_log/unmatched_publish.log"},
-    {colored, true},
+      {max_bytes, 10485760},
-    %% Whether to write a crash log, and where. Undefined means no crash logger.
+      {max_files, 10}
-    {crash_log, "trade_hub.crash.log"},
+    ]}
-    %% Maximum size in bytes of events in the crash log - defaults to 65536
+  ]},
-    {crash_log_msg_size, 65536},
-    %% Maximum size of the crash log in bytes, before its rotated, set
-    %% to 0 to disable rotation - default is 0
-    {crash_log_size, 10485760},
-    %% What time to rotate the crash log - default is no time
-    %% rotation. See the README for a description of this format.
-    {crash_log_date, "$D0"},
-    %% Number of rotated crash logs to keep, 0 means keep only the
-    %% current one - default is 0
-    {crash_log_count, 5},
-    %% Whether to redirect error_logger messages into lager - defaults to true
-    {error_logger_redirect, true},
 
-    %% How big the gen_event mailbox can get before it is switched into sync mode
+  %% 系统日志配置，使用 OTP logger
-    {async_threshold, 20},
+  {kernel, [
-    %% Switch back to async mode, when gen_event mailbox size decrease from `async_threshold'
+    %% 设置 Logger 的 primary log level
-    %% to async_threshold - async_threshold_window
+    {logger_level, debug},
-    {async_threshold_window, 5},
+    {logger, [
+      {handler, default, logger_std_h,
+        #{
+          level => debug,
+          filter_default => stop,
+          filters => [
+            {iot_domain, {fun logger_filters:domain/2, {log, sub, [iot]}}},
+            {endpoint_domain, {fun logger_filters:domain/2, {log, sub, [endpoint]}}}
+          ],
+          formatter => {logger_formatter, #{template => [time, " [", level, "] ", msg, "\n"]}}
+        }
+      },
+
+      {handler, disk, logger_disk_log_h,
+        #{
+          level => debug,
+          filter_default => stop,
+          filters => [
+            {iot_domain, {fun logger_filters:domain/2, {log, sub, [iot]}}},
+            {endpoint_domain, {fun logger_filters:domain/2, {log, sub, [endpoint]}}}
+          ],
+          config => #{
+          file => "log/debug.log",
+          max_no_files => 10,
+          max_no_bytes => 524288000
+        },
+          formatter => {logger_formatter, #{template => [time, " [", level, "] ", msg, "\n"]}}
+        }
+      }
 
-    {handlers, [
-      %% debug | info | warning | error, 日志级别
-      {lager_console_backend, debug},
-      {lager_file_backend, [{file, "debug.log"}, {level, debug}, {size, 314572800}]},
-      {lager_file_backend, [{file, "notice.log"}, {level, notice}, {size, 314572800}]},
-      {lager_file_backend, [{file, "error.log"}, {level, error}, {size, 314572800}]},
-      {lager_file_backend, [{file, "info.log"}, {level, info}, {size, 314572800}]}
     ]}
   ]}
 

config/sys-test.config

@@ -0,0 +1,77 @@
+[
+  {iot, [
+    {http_server, [
+      {port, 18090},
+      {acceptors, 5},
+      {max_connections, 1024},
+      {backlog, 1024}
+    ]},
+
+    {ssl_server, [
+      {port, 1443},
+      {acceptors, 5},
+      {max_connections, 1024},
+      {backlog, 1024}
+    ]},
+
+    {udp_server, [
+      {port, 18080}
+    ]},
+
+    {api_url, "http://127.0.0.1/api/v1"}
+  ]},
+
+  {endpoint, [
+    %% 支持的协议
+    {endpoints, [
+      {root_dir, "/usr/local/code/database/"},
+      {support_protocols, [
+        http
+      ]}
+    ]},
+
+    {endpoint_log, [
+      {path, "${endpoint_root}/endpoint_log/unmatched_publish.log"},
+      {max_bytes, 10485760},
+      {max_files, 10}
+    ]}
+  ]},
+
+  %% 系统日志配置，使用 OTP logger
+  {kernel, [
+    %% 设置 Logger 的 primary log level
+    {logger_level, debug},
+    {logger, [
+      {handler, default, logger_std_h,
+        #{
+          level => debug,
+          filter_default => stop,
+          filters => [
+            {iot_domain, {fun logger_filters:domain/2, {log, sub, [iot]}}},
+            {endpoint_domain, {fun logger_filters:domain/2, {log, sub, [endpoint]}}}
+          ],
+          formatter => {logger_formatter, #{template => [time, " [", level, "] ", msg, "\n"]}}
+        }
+      },
+
+      {handler, disk, logger_disk_log_h,
+        #{
+          level => debug,
+          filter_default => stop,
+          filters => [
+            {iot_domain, {fun logger_filters:domain/2, {log, sub, [iot]}}},
+            {endpoint_domain, {fun logger_filters:domain/2, {log, sub, [endpoint]}}}
+          ],
+          config => #{
+          file => "log/debug.log",
+          max_no_files => 10,
+          max_no_bytes => 524288000
+        },
+          formatter => {logger_formatter, #{template => [time, " [", level, "] ", msg, "\n"]}}
+        }
+      }
+
+    ]}
+  ]}
+
+].

config/vm.args

@@ -5,7 +5,7 @@
 +K true
 +A30
 
--mnesia dir '"/usr/local/var/mnesia/iot"'
+-mnesia dir '"/var/lib/iot/mnesia/"'
 -mnesia dump_log_write_threshold 50000
 -mnesia dc_dump_limit 40
 

docs/container_deploy_request.md

@@ -0,0 +1,670 @@
+# /container/deploy 容器创建请求参数说明
+
+本文档说明 `iot` HTTP 接口 `POST /container/deploy` 当前支持的 JSON 参数、校验规则，以及服务端把 JSON 解码成 Erlang map 后如何转换成下发给 `efka` 的容器部署命令。
+
+对应代码：
+
+- HTTP 入口：[src/transport/http/container_handler.erl](/usr/local/code/cloudkit/iot/src/transport/http/container_handler.erl:63)
+- HTTP JSON 解析：[src/transport/http/http_protocol.erl](/usr/local/code/cloudkit/iot/src/transport/http/http_protocol.erl:68)
+- 参数校验与内部 map 构造：[src/docker/docker_container_builder.erl](/usr/local/code/cloudkit/iot/src/docker/docker_container_builder.erl:25)
+- 主机命令下发：[src/host/iot_host.erl](/usr/local/code/cloudkit/iot/src/host/iot_host.erl:99)
+
+## 1. 接口
+
+```http
+POST /container/deploy
+Content-Type: application/json
+```
+
+HTTP body 必须是 JSON object。`http_protocol` 会使用 `json:decode/1` 解析请求体：
+
+- JSON object -> Erlang map
+- JSON array -> Erlang list
+- JSON string -> Erlang binary
+- JSON integer -> Erlang integer
+- JSON float -> Erlang float
+- JSON boolean -> Erlang `true | false`
+
+`/container/deploy` handler 只接受顶层 map，且必须匹配：
+
+```erlang
+#{<<"uuid">> := UUID, <<"task_id">> := TaskId, <<"config">> := Config}
+```
+
+其中：
+
+- `UUID` 必须是 binary，也就是 JSON string。
+- `TaskId` 必须是 integer。
+- `Config` 必须是 map，也就是 JSON object。
+
+## 2. 完整 JSON 格式
+
+下面示例包含当前支持的所有字段。实际请求可以只传必填字段和需要的可选字段。
+
+```json
+{
+  "uuid": "qbxmjyzrkpntfgswaevodhluicqzxplkm",
+  "task_id": 1001,
+  "config": {
+    "image": "docker.io/library/nginx:latest",
+    "container_name": "my_nginx",
+    "command": ["nginx", "-g", "daemon off;"],
+    "restart": "always",
+    "entrypoint": ["/docker-entrypoint.sh"],
+    "envs": ["ENV=prod", "TZ=Asia/Shanghai"],
+    "expose": ["80", "443/tcp", "53/udp"],
+    "volumes": ["/host/data:/data", "/host/log:/var/log:ro"],
+    "networks": ["bridge"],
+    "network_mode": "bridge",
+    "labels": {
+      "app": "nginx",
+      "env": "prod"
+    },
+    "user": "www-data",
+    "working_dir": "/app",
+    "hostname": "myhost",
+    "privileged": false,
+    "cap_add": ["NET_ADMIN"],
+    "cap_drop": ["MKNOD"],
+    "devices": ["/dev/ttyUSB0:/dev/ttyUSB0:rwm"],
+    "mem_limit": "512m",
+    "mem_reservation": "256m",
+    "cpu_shares": 512,
+    "cpus": 1.5,
+    "ulimits": {
+      "nofile": "1024:2048"
+    },
+    "sysctls": {
+      "net.ipv4.ip_forward": "1"
+    },
+    "tmpfs": ["/tmp", "/run:rw,size=64m"],
+    "extra_hosts": ["host.docker.internal:host-gateway"],
+    "healthcheck": {
+      "test": ["CMD-SHELL", "curl -f http://localhost || exit 1"],
+      "interval": "30s",
+      "timeout": "10s",
+      "retries": 3
+    }
+  }
+}
+```
+
+## 3. 顶层参数
+
+| 字段 | JSON 类型 | 必填 | 说明 |
+| --- | --- | --- | --- |
+| `uuid` | string | 是 | 目标 efka 所属主机 UUID。服务端用它查找 `iot_host` 进程，不会下发到 efka。 |
+| `task_id` | non-negative integer | 是 | 部署任务 ID。会进入内部命令的 `task_id` 字段，用于关联部署结果和部署日志流。`iot` 内部使用 `{uuid, task_id}` 作为任务唯一标识。 |
+| `config` | object | 是 | 容器创建配置。会被校验并转换成内部部署参数 map。 |
+
+顶层没有 `timeout` 字段。当前 HTTP handler 等待 efka command response 的超时时间固定为 10 秒。
+
+部署过程的实时反馈通过 SSE 读取：
+
+```http
+GET /event_stream?uuid=<host_uuid>&task_id=<task_id>
+```
+
+`/container/deploy` 收到合法请求后会先在 `iot` 内部创建或复用 `{uuid, task_id}` 对应的任务进程，然后再向 `efka` 下发部署命令。SSE handler 订阅同一个任务进程；该进程会缓存最近的部署事件，支持多个页面同时订阅，并在收到 efka 的 close 事件后关闭 SSE。
+
+## 4. config 参数
+
+### 必填字段
+
+| 字段 | JSON 类型 | 内部类型 | 说明 |
+| --- | --- | --- | --- |
+| `image` | string | binary | 镜像名。没有 tag 时 efka 部署逻辑会补 `:latest`。 |
+| `container_name` | string | binary | 容器名称。 |
+| `command` | string[] | binary list | 容器启动命令，对应 Docker create config 的 `cmd`。 |
+| `restart` | string | binary | 重启策略，例如 `no`、`always`、`unless-stopped`、`on-failure`、`on-failure:3`。 |
+
+### 可选字段
+
+| 字段 | JSON 类型 | 默认值 | 说明 |
+| --- | --- | --- | --- |
+| `entrypoint` | string[] | `[]` | Docker create config 的 `entrypoint`。 |
+| `envs` | string[] | `[]` | 环境变量列表，例如 `["A=1"]`，对应 Docker create config 的 `env`。 |
+| `expose` | string[] | `[]` | 容器暴露端口，只表示容器端口，不绑定宿主机端口。 |
+| `ports` | string[] | `[]` | 宿主机到容器的端口映射，格式为 `host_port:container_port`，例如 `["8080:80", "443:443"]`。 |
+| `volumes` | string[] | `[]` | volume bind 列表，格式见后文。 |
+| `networks` | string[] | `[]` | Docker network 名称列表，用于 networking config。 |
+| `network_mode` | string | `""` | Docker host config 的 `network_mode`。 |
+| `labels` | object string:string | `{}` | 容器 labels。key/value 都必须是 string。 |
+| `user` | string | `""` | 容器运行用户。 |
+| `working_dir` | string | `""` | 容器工作目录。 |
+| `hostname` | string | `""` | 容器 hostname。 |
+| `privileged` | boolean | `false` | 是否 privileged。 |
+| `cap_add` | string[] | `[]` | 追加 Linux capability。 |
+| `cap_drop` | string[] | `[]` | 删除 Linux capability。 |
+| `devices` | string[] | `[]` | 设备映射列表，格式见后文。 |
+| `mem_limit` | string | `0` | 内存上限，解析成字节数。 |
+| `mem_reservation` | string | `0` | 内存软限制，解析成字节数。 |
+| `cpu_shares` | non-negative integer | `0` | Docker CPU shares。 |
+| `cpus` | non-negative number | `0` | CPU 数量，转换成 nano cpus。 |
+| `ulimits` | object string:string | `[]` | ulimit 配置，格式见后文。 |
+| `sysctls` | object string:string | `{}` | sysctl 配置。key/value 都必须是 string。 |
+| `tmpfs` | string[] | `{}` | tmpfs mount 配置，格式见后文。 |
+| `extra_hosts` | string[] | `[]` | 额外 hosts，例如 `["host.docker.internal:host-gateway"]`。 |
+| `healthcheck` | object | `undefined` | 健康检查配置，格式见后文。 |
+
+### 不支持字段
+
+| 字段 | 当前行为 | 说明 |
+| --- | --- | --- |
+| `container_dir` | 明确拒绝 | 容器目录由 efka 按系统默认规则管理，HTTP 调用方不能指定 efka 主机上的部署目录。 |
+| `env_file` | 忽略 | 当前校验不会识别该字段，后续构造 Docker create options 时也不会使用。 |
+| 其他未知字段 | 忽略 | 除 `container_dir` 外，未知字段不会报错，也不会进入内部部署参数。 |
+
+## 5. 校验规则
+
+校验分为两层。
+
+第一层在 `container_handler`：
+
+- `uuid` 必须是 binary。
+- `task_id` 必须是非负 integer。
+- `config` 必须是 map。
+
+第二层在 `docker_container_builder:deploy_request/2`：
+
+- 先拒绝不支持的 `container_dir` 字段。
+- 检查必填字段是否存在。
+- 检查已知字段类型。
+- 检查 `healthcheck` 内部字段类型，避免无效嵌套字段进入构造阶段。
+- 构造内部部署 map。
+- 对需要解析的字段做格式解析，例如端口、volume、size、duration、ulimit。
+
+类型错误会返回类似：
+
+```text
+required parameter: <<"command">>, type must be: list of string
+optional parameter: <<"labels">>, type must be: map of string:string
+```
+
+缺少必填字段会返回类似：
+
+```text
+miss requied parameter: <<"image">>
+```
+
+注意：错误文本中的 `requied` 是当前代码里的原始拼写。
+
+## 6. 内部转换结果
+
+`config` 校验通过后，`docker_container_builder:deploy_request(TaskId, Config)` 生成：
+
+```erlang
+#{
+    <<"action">> => <<"deploy">>,
+    <<"task_id">> => TaskId,
+    <<"params">> => #{
+        <<"container_name">> => ContainerName,
+        <<"create">> => #{
+            <<"config">> => ContainerConfig,
+            <<"host_config">> => HostConfig,
+            <<"networking_config">> => NetworkingConfig
+        }
+    }
+}
+```
+
+该 map 会通过 efka/iot 长连接协议下发：
+
+```erlang
+{<<"command">>, Ref, {<<"container">>, #{
+    <<"action">> => <<"deploy">>,
+    <<"task_id">> => TaskId,
+    <<"params">> => Params
+}}}
+```
+
+efka 返回：
+
+```erlang
+{<<"command_response">>, Ref, {<<"container">>, Reply}}
+```
+
+其中 `Ref` 是 `crypto:strong_rand_bytes(16)` 生成的 16 字节 binary。网络帧只使用 safe term，协议 label、command map key 和 action 使用 binary；`efka` 收到后直接按 binary key/action 处理。
+
+HTTP handler 最多等待 10 秒。超时返回 HTTP 504，其他参数或执行错误通常返回 HTTP 400；找不到 host 返回业务错误 code 404。
+
+## 7. Docker create config 映射
+
+`create.config` 由 `build_docker_container_config/1` 生成：
+
+| 内部字段 | 来源 JSON 字段 | 转换规则 |
+| --- | --- | --- |
+| `image` | `image` | 原值。 |
+| `cmd` | `command` | 原值。 |
+| `entrypoint` | `entrypoint` | 默认 `[]`。 |
+| `env` | `envs` | 默认 `[]`。 |
+| `labels` | `labels` | 默认 `{}`。 |
+| `volumes` | `volumes` | 只保留 container path 列表。 |
+| `user` | `user` | 默认 `""`。 |
+| `working_dir` | `working_dir` | 默认 `""`。 |
+| `hostname` | `hostname` | 默认 `""`。 |
+| `exposed_ports` | `expose` + `ports` | 转成 `#{container_port, protocol}` map 列表；`ports` 中的容器端口会自动补进 `exposed_ports`。 |
+| `healthcheck` | `healthcheck` | 未传时为 `undefined`。 |
+
+## 8. Docker host config 映射
+
+`create.host_config` 由 `build_docker_host_config/1` 生成：
+
+| 内部字段 | 来源 JSON 字段 | 转换规则 |
+| --- | --- | --- |
+| `binds` | `volumes` | 转成 Docker bind 字符串列表。 |
+| `network_mode` | `network_mode` | 默认 `""`。 |
+| `restart_policy` | `restart` | 转成 `#{name, maximum_retry_count}`。 |
+| `privileged` | `privileged` | 默认 `false`。 |
+| `cap_add` | `cap_add` | 默认 `[]`。 |
+| `cap_drop` | `cap_drop` | 默认 `[]`。 |
+| `devices` | `devices` | 转成设备映射 map 列表。 |
+| `memory` | `mem_limit` | 解析成字节数，未传为 `0`。 |
+| `memory_reservation` | `mem_reservation` | 解析成字节数，未传为 `0`。 |
+| `nano_cpus` | `cpus` | `cpus * 1000000000`，未传为 `0`。 |
+| `cpu_shares` | `cpu_shares` | 未传为 `0`。 |
+| `port_bindings` | `ports` | 转成 `#{host_ip, host_port, container_port, protocol}` map 列表。 |
+| `ulimits` | `ulimits` | 转成 ulimit map 列表。 |
+| `tmpfs` | `tmpfs` | 转成 map。 |
+| `sysctls` | `sysctls` | 默认 `{}`。 |
+| `extra_hosts` | `extra_hosts` | 默认 `[]`。 |
+
+## 9. Docker networking config 映射
+
+`create.networking_config` 由 `build_docker_networking_config/1` 生成：
+
+```erlang
+#{<<"endpoints">> => [#{<<"name">> => Network} || Network <- Networks]}
+```
+
+来源字段：
+
+```json
+"networks": ["bridge", "mynet"]
+```
+
+转换结果：
+
+```erlang
+#{<<"endpoints">> => [
+    #{<<"name">> => <<"bridge">>},
+    #{<<"name">> => <<"mynet">>}
+]}
+```
+
+## 10. 复杂字段转换规则
+
+### restart
+
+输入：
+
+```json
+"restart": "always"
+```
+
+转换：
+
+```erlang
+#{<<"name">> => <<"always">>, <<"maximum_retry_count">> => 0}
+```
+
+输入：
+
+```json
+"restart": "on-failure:3"
+```
+
+转换：
+
+```erlang
+#{<<"name">> => <<"on-failure">>, <<"maximum_retry_count">> => 3}
+```
+
+### expose
+
+输入：
+
+```json
+["80", "443/tcp", "53/udp"]
+```
+
+转换：
+
+```erlang
+[
+    #{<<"container_port">> => 80, <<"protocol">> => <<"tcp">>},
+    #{<<"container_port">> => 443, <<"protocol">> => <<"tcp">>},
+    #{<<"container_port">> => 53, <<"protocol">> => <<"udp">>}
+]
+```
+
+端口必须是无符号整数，且不能超过 `4294967295`。
+
+### ports
+
+输入：
+
+```json
+["8080:80", "443:443", "8053:53/udp"]
+```
+
+转换为 `create.host_config.port_bindings`：
+
+```erlang
+[
+    #{<<"host_ip">> => <<>>, <<"host_port">> => 8080, <<"container_port">> => 80, <<"protocol">> => <<"tcp">>},
+    #{<<"host_ip">> => <<>>, <<"host_port">> => 443, <<"container_port">> => 443, <<"protocol">> => <<"tcp">>},
+    #{<<"host_ip">> => <<>>, <<"host_port">> => 8053, <<"container_port">> => 53, <<"protocol">> => <<"udp">>}
+]
+```
+
+规则：
+
+- 端口映射之间使用冒号分隔：`host_port:container_port`。
+- 容器端口可以带协议：`host_port:container_port/protocol`。
+- 未指定协议时默认为 `tcp`。
+- host port 和 container port 都不能为空，且必须是 `0..65535` 范围内的无符号整数。
+- `ports` 中出现的容器端口会自动补进 `create.config.exposed_ports`，因此不需要在 `expose` 里重复声明。
+
+### volumes
+
+输入：
+
+```json
+["/host/data:/data", "/host/log:/var/log:ro", "/host/cache:/cache:rw"]
+```
+
+转换为 `create.config.volumes`：
+
+```erlang
+[<<"/data">>, <<"/var/log">>, <<"/cache">>]
+```
+
+转换为 `create.host_config.binds`：
+
+```erlang
+[
+    <<"/host/data:/data">>,
+    <<"/host/log:/var/log:ro">>,
+    <<"/host/cache:/cache">>
+]
+```
+
+规则：
+
+- `host_path:container_path` -> 读写挂载。
+- `host_path:container_path:ro` -> 只读挂载。
+- `host_path:container_path:rw` -> 当前实现视为读写挂载，输出时不会保留 `:rw`。
+- host path 和 container path 都不能为空。
+
+### devices
+
+输入：
+
+```json
+["/dev/ttyUSB0:/dev/ttyUSB0", "/dev/snd:/dev/snd:rwm"]
+```
+
+转换：
+
+```erlang
+[
+    #{
+        <<"path_on_host">> => <<"/dev/ttyUSB0">>,
+        <<"path_in_container">> => <<"/dev/ttyUSB0">>,
+        <<"cgroup_permissions">> => <<"rwm">>
+    },
+    #{
+        <<"path_on_host">> => <<"/dev/snd">>,
+        <<"path_in_container">> => <<"/dev/snd">>,
+        <<"cgroup_permissions">> => <<"rwm">>
+    }
+]
+```
+
+规则：
+
+- `host_path:container_path` -> 权限默认为 `rwm`。
+- `host_path:container_path:permissions` -> 使用第三段作为权限。
+- 任意路径或 permissions 为空时返回 `invalid device mapping`。
+
+### ulimits
+
+输入：
+
+```json
+{
+  "nofile": "1024:2048",
+  "nproc": "4096"
+}
+```
+
+转换：
+
+```erlang
+[
+    #{<<"name">> => <<"nofile">>, <<"soft">> => 1024, <<"hard">> => 2048},
+    #{<<"name">> => <<"nproc">>, <<"soft">> => 4096, <<"hard">> => 4096}
+]
+```
+
+规则：
+
+- `"soft:hard"` -> 分别设置 soft 和 hard。
+- `"limit"` -> soft 和 hard 都等于 limit。
+- 数值必须是无符号整数。
+
+### tmpfs
+
+输入：
+
+```json
+["/tmp", "/run:rw,size=64m"]
+```
+
+转换：
+
+```erlang
+#{
+    <<"/tmp">> => <<>>,
+    <<"/run">> => <<"rw,size=64m">>
+}
+```
+
+规则：
+
+- `"path"` -> options 为 `""`。
+- `"path:options"` -> options 为第二段。
+- path 不能为空。
+
+### healthcheck
+
+输入：
+
+```json
+{
+  "test": ["CMD-SHELL", "curl -f http://localhost || exit 1"],
+  "interval": "30s",
+  "timeout": "10s",
+  "retries": 3
+}
+```
+
+转换：
+
+```erlang
+#{
+    <<"test">> => [<<"CMD-SHELL">>, <<"curl -f http://localhost || exit 1">>],
+    <<"interval_ns">> => 30000000000,
+    <<"timeout_ns">> => 10000000000,
+    <<"retries">> => 3
+}
+```
+
+字段规则：
+
+| 字段 | JSON 类型 | 默认值 | 说明 |
+| --- | --- | --- | --- |
+| `test` | string[] | `[]` | Docker healthcheck test 参数列表。 |
+| `interval` | string 或 non-negative integer | `"0s"` | string 会解析时间单位；integer 直接视为纳秒。 |
+| `timeout` | string 或 non-negative integer | `"0s"` | string 会解析时间单位；integer 直接视为纳秒。 |
+| `retries` | non-negative integer | `0` | Docker healthcheck 重试次数。 |
+
+时间单位：
+
+| 单位 | 含义 |
+| --- | --- |
+| `ns` | 纳秒 |
+| `us` | 微秒 |
+| `ms` | 毫秒 |
+| `s` | 秒 |
+| `m` | 分钟 |
+| `h` | 小时 |
+| 无单位 | 秒 |
+
+示例：
+
+- `"30s"` -> `30000000000`
+- `"10ms"` -> `10000000`
+- `"2m"` -> `120000000000`
+- `"5"` -> `5000000000`
+
+### mem_limit 和 mem_reservation
+
+输入：
+
+```json
+"mem_limit": "512m",
+"mem_reservation": "1g"
+```
+
+转换：
+
+```erlang
+<<"memory">> => 536870912,
+<<"memory_reservation">> => 1073741824
+```
+
+支持单位：
+
+| 单位 | 倍数 |
+| --- | --- |
+| `b` 或无单位 | 1 |
+| `k`, `kb`, `ki`, `kib` | 1024 |
+| `m`, `mb`, `mi`, `mib` | 1048576 |
+| `g`, `gb`, `gi`, `gib` | 1073741824 |
+| `t`, `tb`, `ti`, `tib` | 1099511627776 |
+
+数值支持整数或小数，例如 `"1.5g"`。
+
+### cpus
+
+输入：
+
+```json
+"cpus": 1.5
+```
+
+转换：
+
+```erlang
+<<"nano_cpus">> => 1500000000
+```
+
+规则：
+
+- integer 或 float 都可以。
+- 必须大于等于 0。
+- 转换公式：`trunc(Cpus * 1000000000)`。
+
+## 11. 最小请求示例
+
+```json
+{
+  "uuid": "qbxmjyzrkpntfgswaevodhluicqzxplkm",
+  "task_id": 1001,
+  "config": {
+    "image": "docker.io/library/nginx:latest",
+    "container_name": "my_nginx",
+    "command": ["nginx", "-g", "daemon off;"],
+    "restart": "always"
+  }
+}
+```
+
+对应下发给 `efka` 的协议命令示意：
+
+```erlang
+{<<"command">>, Ref, {<<"container">>, #{
+    <<"action">> => <<"deploy">>,
+    <<"task_id">> => 1001,
+    <<"params">> => #{
+        <<"container_name">> => <<"my_nginx">>,
+        <<"create">> => #{
+            <<"config">> => #{
+                <<"image">> => <<"docker.io/library/nginx:latest">>,
+                <<"cmd">> => [<<"nginx">>, <<"-g">>, <<"daemon off;">>],
+                <<"entrypoint">> => [],
+                <<"env">> => [],
+                <<"labels">> => #{},
+                <<"volumes">> => [],
+                <<"user">> => <<>>,
+                <<"working_dir">> => <<>>,
+                <<"hostname">> => <<>>,
+                <<"exposed_ports">> => [],
+                <<"healthcheck">> => undefined
+            },
+            <<"host_config">> => #{
+                <<"binds">> => [],
+                <<"network_mode">> => <<>>,
+                <<"restart_policy">> => #{<<"name">> => <<"always">>, <<"maximum_retry_count">> => 0},
+                <<"privileged">> => false,
+                <<"cap_add">> => [],
+                <<"cap_drop">> => [],
+                <<"devices">> => [],
+                <<"memory">> => 0,
+                <<"memory_reservation">> => 0,
+                <<"nano_cpus">> => 0,
+                <<"cpu_shares">> => 0,
+                <<"port_bindings">> => [],
+                <<"ulimits">> => [],
+                <<"tmpfs">> => #{},
+                <<"sysctls">> => #{},
+                <<"extra_hosts">> => []
+            },
+            <<"networking_config">> => #{<<"endpoints">> => []}
+        }
+    }
+}}}
+```
+
+## 12. 响应
+
+成功时 HTTP body 由 `iot_util:json_data/1` 包装：
+
+```json
+{
+  "result": "ok"
+}
+```
+
+如果 efka 返回的是 JSON binary，handler 会尝试解码后再放入 `result`。
+
+错误示例：
+
+```json
+{
+  "error": {
+    "code": 400,
+    "message": "invalid port binding"
+  }
+}
+```
+
+常见状态：
+
+| HTTP 状态 | 场景 |
+| --- | --- |
+| `200` | host not found 或构造阶段返回的业务错误也可能通过 200 包装业务错误。 |
+| `400` | 参数校验失败、efka 返回普通错误。 |
+| `504` | 等待 efka command response 超时。 |

docs/efka_iot_protocol.md

@@ -0,0 +1,268 @@
+# EFKA 与 IOT 交互协议
+
+本文档描述 `efka` 与 `iot` 之间的 TLS 长连接协议。当前协议由 Erlang term 直接序列化，发送端使用 `term_to_binary/1`，接收端使用 `binary_to_term(PacketBin, [safe])`。
+
+协议帧只使用 safe external term：顶层 label、业务 label、map key 使用 binary；`Ref` 使用 `crypto:strong_rand_bytes(16)` 生成，是 16 字节 binary。网络协议里不发送 Erlang `reference()`，也不依赖动态创建 atom。
+
+## 传输层
+
+- `efka` 作为 TLS client 连接 `iot`。
+- `iot` 作为 TLS server 接收多个 `efka` 连接，一个连接对应一个 `ssl_channel` 进程。
+- socket 使用 `{packet, 4}`，每个 Erlang term binary 作为一个完整包发送。
+- `Ref` 使用 `crypto:strong_rand_bytes(16)` 生成，只在当前连接的 inflight 表内匹配。
+
+## 顶层帧
+
+协议顶层 tuple 用来表达交互语义：
+
+```erlang
+{<<"request">>, Ref, Body}
+{<<"response">>, Ref, Reply}
+{<<"command">>, Ref, {Domain, Payload}}
+{<<"command_response">>, Ref, {Domain, Reply}}
+{<<"message">>, Body}
+```
+
+语义说明：
+
+| 帧 | 方向 | 语义 |
+| --- | --- | --- |
+| `{<<"request">>, Ref, Body}` | efka -> iot | efka 发起请求，需要 iot 回复 |
+| `{<<"response">>, Ref, Reply}` | iot -> efka | iot 对 efka request 的回复 |
+| `{<<"command">>, Ref, {Domain, Payload}}` | iot -> efka | iot 下发命令，需要 efka 回复 |
+| `{<<"command_response">>, Ref, {Domain, Reply}}` | efka -> iot | efka 对 iot command 的回复 |
+| `{<<"message">>, Body}` | 双向 | 异步消息，不要求回复 |
+
+`command` 和 `command_response` 的 `Domain` 表示业务域，目前支持：
+
+- `<<"container">>`
+
+## 鉴权请求
+
+初始连接由 `efka` 发起鉴权 request。每条 TLS 连接只允许一次鉴权；`iot` 侧鉴权成功后会在 `ssl_channel` 标记该连接已鉴权，如果同一连接再次发送 `auth_request`，`iot` 会直接关闭连接。
+
+```erlang
+{<<"request">>, Ref, {<<"auth_request">>, #{
+    <<"uuid">> => UUID,
+    <<"token">> => Token,
+    <<"timestamp">> => Timestamp
+}}}
+```
+
+`iot` 回复：
+
+```erlang
+{<<"response">>, Ref, {<<"auth_response">>, <<"ok">>}}
+{<<"response">>, Ref, {<<"auth_response">>, {<<"error">>, {<<"failed">>, Reason}}}}
+```
+
+处理语义：
+
+- `<<"ok">>`：`efka` 进入 `activated` 状态。
+- `{<<"error">>, {<<"failed">>, Reason}}`：鉴权失败，`iot` 返回失败响应后关闭连接；`efka` 进入重连流程。
+
+## 授权控制
+
+`/host/activate` 只修改 `iot` 本地和持久化的 host 授权状态，不再向 `efka` 下发 auth command。`efka` 可以继续保持连接并发送数据，是否处理这些数据由 `iot_host` 当前状态决定。
+
+因此当前协议没有 `{<<"command">>, Ref, {<<"auth">>, ...}}` 和 `{<<"command_response">>, Ref, {<<"auth">>, ...}}`。授权关闭时，`iot_host` 保持 channel 在线，但不处理上报数据；授权重新打开后，已在线的 channel 可以继续使用。
+
+## 容器管理命令
+
+`iot` 对 `efka` 的容器管理使用 command 语义：
+
+```erlang
+{<<"command">>, Ref, {<<"container">>, CommandMap}}
+```
+
+`efka` 回复：
+
+```erlang
+{<<"command_response">>, Ref, {<<"container">>, Reply}}
+```
+
+`Reply` 取值：
+
+```erlang
+<<"ok">>
+{<<"ok">>, Result}
+{<<"error">>, Reason}
+```
+
+`CommandMap` 使用 binary key 和 binary action；`efka` 接收后直接按 binary key/action 匹配，Docker 参数链路继续使用 binary-key map，不再转换成 atom-key map。
+
+### list
+
+```erlang
+#{<<"action">> => <<"list">>}
+```
+
+返回当前 `efka` 主机上的容器列表。
+
+### deploy
+
+```erlang
+#{
+    <<"action">> => <<"deploy">>,
+    <<"task_id">> => TaskId,
+    <<"params">> => Params
+}
+```
+
+触发容器部署。部署过程中的流式日志不通过该 command response 返回，而是通过 `message` 的 `task_event` 上报。
+
+### start
+
+```erlang
+#{
+    <<"action">> => <<"start">>,
+    <<"target">> => Target
+}
+```
+
+### stop
+
+```erlang
+#{
+    <<"action">> => <<"stop">>,
+    <<"target">> => Target,
+    <<"timeout_seconds">> => TimeoutSeconds
+}
+```
+
+### kill
+
+```erlang
+#{
+    <<"action">> => <<"kill">>,
+    <<"target">> => Target,
+    <<"signal">> => Signal
+}
+```
+
+### remove
+
+```erlang
+#{
+    <<"action">> => <<"remove">>,
+    <<"target">> => Target,
+    <<"force">> => Force,
+    <<"remove_volumes">> => RemoveVolumes
+}
+```
+
+### config
+
+```erlang
+#{
+    <<"action">> => <<"config">>,
+    <<"target">> => Target,
+    <<"config">> => Config
+}
+```
+
+更新容器配置文件。
+
+### Target
+
+容器目标使用 map 表示：
+
+```erlang
+#{
+    <<"name">> => ContainerName,
+    <<"id">> => ContainerId
+}
+```
+
+`name` 和 `id` 至少一个非空；优先使用 `name`，`name` 为空时使用 `id`。
+
+## 异步消息
+
+`message` 不带 `Ref`，不要求对端回复。
+
+### efka -> iot: data
+
+```erlang
+{<<"message">>, {<<"data">>, #{
+    <<"route_key">> => RouteKey,
+    <<"metric">> => Metric
+}}}
+```
+
+用于 `efka` 上报业务指标数据。
+
+### efka -> iot: task_event
+
+```erlang
+{<<"message">>, {<<"task_event">>, #{
+    <<"task_id">> => TaskId,
+    <<"type">> => Type,
+    <<"stream">> => Stream
+}}}
+```
+
+任务事件流关闭时：
+
+```erlang
+{<<"message">>, {<<"task_event">>, #{
+    <<"task_id">> => TaskId,
+    <<"type">> => <<"close">>,
+    <<"stream">> => Reason
+}}}
+```
+
+`task_event` 本身不携带 `uuid`。`iot` 在接收该消息时使用当前已鉴权 `ssl_channel` 绑定的 host UUID，把事件路由到内部任务进程 `{UUID, TaskId}`。HTTP 页面通过 SSE 订阅时也必须使用同一组参数：
+
+```http
+GET /event_stream?uuid=<host_uuid>&task_id=<task_id>
+```
+
+`iot` 会为每个 `{UUID, TaskId}` 维护一个独立的任务进程，用于缓存最近的部署日志、支持多个 SSE listener，并在收到 close 事件后结束事件流。
+
+### efka -> iot: ping
+
+```erlang
+{<<"message">>, <<"ping">>}
+```
+
+用于 TLS 长连接的应用层保活。`efka` 在鉴权成功后周期发送，当前发送间隔为 30 秒。
+
+`iot` 收到后回复：
+
+```erlang
+{<<"message">>, <<"pong">>}
+```
+
+`ping/pong` 只表示 TLS 连接仍可读写，不参与 host online/offline 判定。host 上下线仍由 UDP 心跳和 `iot_host` 本地连接状态共同维护。
+
+### iot -> efka: pub
+
+```erlang
+{<<"message">>, {<<"pub">>, #{
+    <<"topic">> => Topic,
+    <<"qos">> => Qos,
+    <<"content">> => Content
+}}}
+```
+
+用于 `iot` 向 `efka` 本地订阅系统发布 topic 消息。
+
+## 状态与超时
+
+- `efka` 鉴权超时时间：5 秒。
+- `iot` command inflight 超时时间：60 秒。
+- `iot` SSL channel 空闲超时时间：120 秒。120 秒内没有收到任何 TLS 包，包括 `ping`、业务 `message`、`request`、`command_response`，`iot` 会主动关闭该连接。
+- `iot` 管理多个 `efka` 时，每个连接有独立 `ssl_channel` 和独立 inflight 表。
+- command 超时后，`iot` 删除 inflight 记录；之后如果迟到的 `command_response` 到达，会被视为未预期响应。
+
+## 兼容性
+
+当前协议不兼容旧 tuple：
+
+- 旧容器管理：`{request, Ref, {container_request, ...}}`
+- 旧容器回复：`{response, Ref, {container_response, ...}}`
+- 旧授权控制：`{message, {auth_control, Command}}`
+- 已移除的 auth command：`{command, Ref, {auth, activate | deactivate}}`
+- 旧 Ref：Erlang `reference()`，例如 `make_ref()` 生成的值。
+
+如果需要滚动升级，应先增加临时兼容分支或引入协议版本协商。

docs/heartbeat.md

@@ -1,11 +1,44 @@
 ## 心跳机制
 * 边缘主机通过心跳机制来判断主机是否存活(解决弱网环境下websocket链接会经常断开的问题)
-* 边缘主机每隔5秒发送一次心跳包，服务端每隔2分钟检测一下，判断是否有收到心跳包；如果没有收到则认为主机离线
+* 边缘主机定期发送 UDP 心跳包，服务端按照 `iot_host` 的心跳检测周期判断是否有收到心跳包。
+* 如果 UDP 心跳丢失但 SSL channel 仍然存在，服务端不会把 host 标记为离线。
+* 如果 UDP 心跳丢失且 SSL channel 也不存在，服务端会把 host 标记为离线。
 
 ### udp服务器
 * 端口: 18080
 
 ### 心跳包格式
-*  <<Len:2, HostUUID/binary>>
+旧格式 `<<Len:16, HostUUID/binary>>` 已不再接受。
-* Len表示HostUUID对应的字节数，Len本身占用2字节长度(HostUUID不一定是固定长度，因此需要标注)
+
-* 注解：采用这种格式是为了方便后续扩展别的心跳信息字段
+当前心跳包格式：
+
+```erlang
+<<
+    Version:8,
+    UuidLen:16,
+    UUID:UuidLen/binary,
+    Timestamp:64/unsigned-big,
+    Nonce:16/binary,
+    Mac:32/binary
+>>
+```
+
+字段说明：
+
+| 字段 | 说明 |
+| --- | --- |
+| `Version` | 当前固定为 `1`。 |
+| `UuidLen` | `UUID` 的字节数。 |
+| `UUID` | efka 对应的 host UUID。 |
+| `Timestamp` | 秒级 Unix 时间戳。服务端只接受 120 秒时间窗内的心跳。 |
+| `Nonce` | 16 字节随机数，必须加入 HMAC payload，使同一秒内多次心跳的 Mac 不同。当前服务端不保存 nonce。 |
+| `Mac` | HMAC-SHA256 结果，32 字节。 |
+
+`Mac` 的计算内容为前面所有字段：
+
+```erlang
+Payload = <<Version:8, UuidLen:16, UUID:UuidLen/binary, Timestamp:64/unsigned-big, Nonce:16/binary>>,
+Mac = crypto:mac(hmac, sha256, HeartbeatSecret, Payload)
+```
+
+服务端在 `efka_client_store:verify_heartbeat/4` 中校验时间窗和 HMAC。`HeartbeatSecret` 使用 `SHA256(Token)`，其中 `Token` 和 TLS 鉴权使用同一个 auth token。`iot` 侧只保存派生后的 `heartbeat_secret`，不保存明文 token。

docs/iot_api.md

@@ -3,7 +3,7 @@
 ```markdown
 # 📘 IoT API 接口文档
 
-> 模块：`iot_api`  
+> 模块：`iot_api_client`  
 > 作者：**anlicheng**  
 > 创建时间：2023-12-24  
 > 数据格式：`application/json`  
@@ -133,7 +133,36 @@ POST /change_host_status
 
 ---
 
-### 5. 获取主机下的设备列表
+### 5. 修改主机授权状态
+
+**接口：**
+
+```
+POST /change_host_authorize_status
+```
+
+**请求体：**
+
+```json
+{
+  "uuid": "uuid-1",
+  "new_authorize_status": 1
+}
+```
+
+`new_authorize_status` 为 `1` 表示允许 `iot_host` 处理该 host 的上报数据，为 `0` 表示保持连接但不处理上报数据。
+
+**返回示例：**
+
+```json
+{
+  "result": "ok"
+}
+```
+
+---
+
+### 6. 获取主机下的设备列表
 
 **接口：**
 
@@ -162,7 +191,7 @@ GET /get_host_devices?host_id=<id>
 
 ## 🔧 设备（Device）相关接口
 
-### 6. 获取设备详情
+### 7. 获取设备详情
 
 **接口：**
 
@@ -190,7 +219,7 @@ GET /get_device_by_uuid?device_uuid=<uuid>
 
 ---
 
-### 7. 修改设备状态
+### 8. 修改设备状态
 
 **接口：**
 
@@ -219,7 +248,7 @@ POST /change_device_status
 
 ## 🌐 Endpoint（数据终端）相关接口
 
-### 8. 获取所有 Endpoint
+### 9. 获取所有 Endpoint
 
 **接口：**
 
@@ -312,7 +341,7 @@ GET /get_all_endpoints
 
 ---
 
-### 9. 获取指定 Endpoint 信息
+### 10. 获取指定 Endpoint 信息
 
 **接口：**
 

rebar.config

@@ -1,4 +1,7 @@
 {erl_opts, [debug_info]}.
+
+{project_app_dirs, ["apps/*"]}.
+
 {deps, [
     {poolboy, ".*", {git, "https://github.com/devinus/poolboy.git", {tag, "1.5.1"}}},
     {hackney, ".*", {git, "https://github.com/benoitc/hackney.git", {tag, "1.25.0"}}},
@@ -6,27 +9,25 @@
     {cowboy, ".*", {git, "https://github.com/ninenines/cowboy.git", {tag, "2.14.0"}}},
     {ranch, ".*", {git, "https://github.com/ninenines/ranch.git", {tag, "2.2.0"}}},
     {brod, ".*", {git, "https://github.com/kafka4beam/brod.git", {tag, "4.4.5"}}},
-    {jiffy, ".*", {git, "https://github.com/davisp/jiffy.git", {tag, "1.1.1"}}},
-    {mysql, ".*", {git, "https://github.com/mysql-otp/mysql-otp", {tag, "1.8.0"}}},
     {eredis, ".*", {git, "https://github.com/wooga/eredis.git", {tag, "v1.2.0"}}},
-    {emqtt, ".*", {git, "https://gitea.s5s8.com/anlicheng/emqtt.git", {branch, "main"}}},
+    {emqtt, ".*", {git, "https://gitea.s5s8.com/anlicheng/emqtt.git", {tag, "v1.2"}}},
-    {gproc, ".*", {git, "https://github.com/uwiger/gproc.git", {tag, "0.9.1"}}},
+    {gproc, ".*", {git, "https://github.com/uwiger/gproc.git", {tag, "0.9.1"}}}
-    {parse_trans, ".*", {git, "https://github.com/uwiger/parse_trans", {tag, "3.0.0"}}},
-    {lager, ".*", {git,"https://github.com/erlang-lager/lager.git", {tag, "3.9.2"}}}
 ]}.
 
 {relx, [{release, {iot, "0.1.0"},
          [iot,
+          endpoint,
           sasl]},
 
-        {mode, dev},
+        {include_erts, true},
+        {mode, prod},
         %{mode, prod},
 
     %% automatically picked up if the files
         %% exist but can be set manually, which
         %% is required if the names aren't exactly
         %% sys.config and vm.args
-        {sys_config, "./config/sys.config"},
+        {sys_config, "./config/sys-dev.config"},
         {vm_args, "./config/vm.args"}
 
         %% the .src form of the configuration files do
@@ -39,14 +40,12 @@
                      [%% prod is the default mode when prod
                       %% profile is used, so does not have
                       %% to be explicitly included like this
-                      {mode, prod}
+                      {mode, prod},
+                      {sys_config, "./config/sys-prod.config"}
 
                       %% use minimal mode to exclude ERTS
                       %% {mode, minimal}
                      ]
             }]}]}.
 
-{erl_opts, [{parse_transform,lager_transform}]}.
-
 {rebar_packages_cdn, "https://hexpm.upyun.com"}.
-

rebar.lock

@@ -15,14 +15,13 @@
  {<<"crc32cer">>,{pkg,<<"crc32cer">>,<<"1.0.3">>},2},
  {<<"emqtt">>,
   {git,"https://gitea.s5s8.com/anlicheng/emqtt.git",
-       {ref,"5111914a9b1b92b0b497f825c77bdd365e3989b0"}},
+       {ref,"c5a52dcb57cd23a4318e342b705f9f726e8d67a1"}},
   0},
  {<<"eredis">>,
   {git,"https://github.com/wooga/eredis.git",
        {ref,"9ad91f149310a7d002cb966f62b7e2c3330abb04"}},
   0},
  {<<"fs">>,{pkg,<<"fs">>,<<"6.1.1">>},1},
- {<<"goldrush">>,{pkg,<<"goldrush">>,<<"0.1.9">>},1},
  {<<"gproc">>,
   {git,"https://github.com/uwiger/gproc.git",
        {ref,"4ca45e0a97722a418a31eb1753f4e3b953f7fb1d"}},
@@ -32,25 +31,10 @@
        {ref,"8c00789e411d7c09a9808d720232098da1f19d69"}},
   0},
  {<<"idna">>,{pkg,<<"idna">>,<<"6.1.1">>},1},
- {<<"jiffy">>,
-  {git,"https://github.com/davisp/jiffy.git",
-       {ref,"9ea1b35b6e60ba21dfd4adbd18e7916a831fd7d4"}},
-  0},
  {<<"kafka_protocol">>,{pkg,<<"kafka_protocol">>,<<"4.2.7">>},1},
- {<<"lager">>,
-  {git,"https://github.com/erlang-lager/lager.git",
-       {ref,"459a3b2cdd9eadd29e5a7ce5c43932f5ccd6eb88"}},
-  0},
  {<<"metrics">>,{pkg,<<"metrics">>,<<"1.0.1">>},1},
  {<<"mimerl">>,{pkg,<<"mimerl">>,<<"1.4.0">>},1},
- {<<"mysql">>,
+ {<<"parse_trans">>,{pkg,<<"parse_trans">>,<<"3.4.1">>},1},
-  {git,"https://github.com/mysql-otp/mysql-otp",
-       {ref,"caf5ff96c677a8fe0ce6f4082bc036c8fd27dd62"}},
-  0},
- {<<"parse_trans">>,
-  {git,"https://github.com/uwiger/parse_trans",
-       {ref,"6f3645afb43c7c57d61b54ef59aecab288ce1013"}},
-  0},
  {<<"poolboy">>,
   {git,"https://github.com/devinus/poolboy.git",
        {ref,"3bb48a893ff5598f7c73731ac17545206d259fac"}},
@@ -70,22 +54,22 @@
  {<<"certifi">>, <<"0E6E882FCDAAA0A5A9F2B3DB55B1394DBA07E8D6D9BCAD08318FB604C6839712">>},
  {<<"crc32cer">>, <<"AD0E42BED8603F2C72DE2A00F1B5063FFE12D5988615CAD984096900431D1C1A">>},
  {<<"fs">>, <<"9D147B944D60CFA48A349F12D06C8EE71128F610C90870BDF9A6773206452ED0">>},
- {<<"goldrush">>, <<"F06E5D5F1277DA5C413E84D5A2924174182FB108DABB39D5EC548B27424CD106">>},
  {<<"idna">>, <<"8A63070E9F7D0C62EB9D9FCB360A7DE382448200FBBD1B106CC96D3D8099DF8D">>},
  {<<"kafka_protocol">>, <<"6F53B15CD6F6A12C1D0010DB074B4A15985C71BC7F594BC2D67D9837B3B378A1">>},
  {<<"metrics">>, <<"25F094DEA2CDA98213CECC3AEFF09E940299D950904393B2A29D191C346A8486">>},
  {<<"mimerl">>, <<"3882A5CA67FBBE7117BA8947F27643557ADEC38FA2307490C4C4207624CB213B">>},
+ {<<"parse_trans">>, <<"6E6AA8167CB44CC8F39441D05193BE6E6F4E7C2946CB2759F015F8C56B76E5FF">>},
  {<<"ssl_verify_fun">>, <<"354C321CF377240C7B8716899E182CE4890C5938111A1296ADD3EC74CF1715DF">>},
  {<<"unicode_util_compat">>, <<"BC84380C9AB48177092F43AC89E4DFA2C6D62B40B8BD132B1059ECC7232F9A78">>}]},
 {pkg_hash_ext,[
  {<<"certifi">>, <<"B147ED22CE71D72EAFDAD94F055165C1C182F61A2FF49DF28BCC71D1D5B94A60">>},
  {<<"crc32cer">>, <<"08FDCD5CE51ACD839A12E98742F0F0EDA19A2A679FC9FBFAF6AAB958310FB70E">>},
  {<<"fs">>, <<"EF94E95FFE79916860649FED80AC62B04C322B0BB70F5128144C026B4D171F8B">>},
- {<<"goldrush">>, <<"99CB4128CFFCB3227581E5D4D803D5413FA643F4EB96523F77D9E6937D994CEB">>},
  {<<"idna">>, <<"92376EB7894412ED19AC475E4A86F7B413C1B9FBB5BD16DCCD57934157944CEA">>},
  {<<"kafka_protocol">>, <<"1D5E9597AD3C0776C86DC5E08D3BAAEA7DB805A52E5FD35E3F071AAD7789FC4C">>},
  {<<"metrics">>, <<"69B09ADDDC4F74A40716AE54D140F93BEB0FB8978D8636EADED0C31B6F099F16">>},
  {<<"mimerl">>, <<"13AF15F9F68C65884ECCA3A3891D50A7B57D82152792F3E19D88650AA126B144">>},
+ {<<"parse_trans">>, <<"620A406CE75DADA827B82E453C19CF06776BE266F5A67CFF34E1EF2CBB60E49A">>},
  {<<"ssl_verify_fun">>, <<"FE4C190E8F37401D30167C8C405EDA19469F34577987C76DDE613E838BBC67F8">>},
  {<<"unicode_util_compat">>, <<"25EEE6D67DF61960CF6A794239566599B09E17E668D3700247BC498638152521">>}]}
 ].

todo_docs/mysql.md

@@ -1,37 +0,0 @@
-# 待完成
-
-## 1. endpoint需要存储在数据库
-```mysql
-
-CREATE TABLE `endpoint` (
-`id` bigint unsigned NOT NULL AUTO_INCREMENT,
-`name` varchar(191) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '' COMMENT '名称,路由时基于名称',
-`title` varchar(191) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '' COMMENT '序列号',
-`type` varchar(60) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci DEFAULT '' COMMENT '类型',
-`config_json` text CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci COMMENT '配置信息，基于json格式存储',
-`status` smallint NOT NULL DEFAULT '-1',
-`creator` smallint NOT NULL DEFAULT '0' COMMENT '创建人',
-`created_at` timestamp NULL DEFAULT NULL,
-`updated_at` timestamp NULL DEFAULT NULL,
-PRIMARY KEY (`id`),
-KEY `idx_name` (`name`)
-) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci
-```
-
-## 2. service_config微服务配置也要存在于数据库
-```mysql
-CREATE TABLE `service_config` (
-`id` bigint unsigned NOT NULL AUTO_INCREMENT,
-`service_id` bigint unsigned NOT NULL COMMENT '服务的id',
-`host_uuid` varchar(191) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '' COMMENT '名称,路由时基于名称',
-`config_json` text CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci COMMENT '配置信息，基于json格式存储',
-`last_config_json` text CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci COMMENT '配置信息，基于json格式存储',
-`creator` smallint NOT NULL DEFAULT '0' COMMENT '创建人',
-`created_at` timestamp NULL DEFAULT NULL,
-`updated_at` timestamp NULL DEFAULT NULL,
-PRIMARY KEY (`id`),
-KEY `idx_service_id` (`service_id`)
-) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci
-
-
-```