fix line protocol

apps/iot/src/data_format/line_format.erl

@@ -16,67 +16,53 @@
 -spec parse(Metric :: binary()) -> map().
 parse(Metric) when is_binary(Metric) ->
     case lexer(Metric) of
-        [Measurement, Fields, Timestamp] ->
+        [[Measurement|Tags], Fields, [Timestamp|_]] ->
-            Map = case binary:split(Measurement, <<",">>) of
+            {ok, #{
-                      [Measurement] ->
+                <<"measurement">> => Measurement,
-                          #{<<"measurement">> => Measurement, <<"tags">> => #{}};
+                <<"tags">> => parser_keys(Tags),
-                      [Measurement0, Tags0] ->
+                <<"fields">> => parser_keys(Fields),
-                          Tags = maps:from_list(parser_keys(Tags0)),
+                <<"timestamp">> => Timestamp
-                          #{<<"measurement">> => Measurement0, <<"tags">> => Tags}
+            }};
-                  end,
-            FieldsMap = maps:from_list(parser_keys(Fields)),
-            {ok, Map#{<<"fields">> => FieldsMap, <<"timestamp">> => binary_to_integer(Timestamp)}};
         _ ->
             error
     end.
 
--spec lexer(Input :: binary()) -> [Token :: binary()].
+%% 词法分析：将输入文本转换为标记序列
-lexer(Input) when is_binary(Input) ->
+lexer(Input) ->
-    lexer(Input, false, [], []).
+    lexer(Input, false, [], [], []).
-lexer(<<>>, _Quotes, Current, Acc) ->
+lexer(<<>>, _Quotes, Current, Acc, Groups) ->
     Last = list_to_binary(lists:reverse(Current)),
-    lists:reverse([Last|Acc]);
+    NAcc = lists:reverse([Last|Acc]),
-lexer(<<$\s, Rest/binary>>, Quotes, Current, Acc) ->
+    lists:reverse([NAcc|Groups]);
+%% 遇到逗号
+lexer(<<$,, Rest/binary>>, Quotes, Current, Acc, Groups) ->
     case Quotes of
         true ->
-            lexer(Rest, Quotes, [" "|Current], Acc);
+            lexer(Rest, Quotes, [","|Current], Acc, Groups);
         false ->
             Part = list_to_binary(lists:reverse(Current)),
-            lexer(Rest, Quotes, [], [Part|Acc])
+            lexer(Rest, Quotes, [], [Part|Acc], Groups)
     end;
-lexer(<<$\", Rest/binary>>, Quotes, Current, Acc) ->
+lexer(<<$\s, Rest/binary>>, Quotes, Current, Acc, Groups) ->
-    lexer(Rest, not Quotes, [$\"|Current], Acc);
+    case Quotes of
-lexer(<<Char/utf8, Rest/binary>>, Quotes, Current, Acc) ->
+        true ->
-    lexer(Rest, Quotes, [Char|Current], Acc).
+            lexer(Rest, Quotes, [$\s|Current], Acc, Groups);
+        false ->
+            Part = list_to_binary(lists:reverse(Current)),
+            NAcc = lists:reverse([Part|Acc]),
+            lexer(Rest, Quotes, [], [], [NAcc|Groups])
+    end;
+lexer(<<$\", Rest/binary>>, Quotes, Current, Acc, Groups) ->
+    lexer(Rest, not Quotes, [$\"|Current], Acc, Groups);
+lexer(<<Char/utf8, Rest/binary>>, Quotes, Current, Acc, Groups) ->
+    lexer(Rest, Quotes, [Char|Current], Acc, Groups).
 
--spec parser_keys(Input :: binary()) -> [{Key :: binary(), Val :: binary()}].
+parser_keys(Tokens) ->
-parser_keys(Input) when is_binary(Input) ->
+    maps:from_list(lists:flatmap(fun(Item) ->
-    Parts = split_keys(Input),
-    lists:flatmap(fun(Item) ->
         case binary:split(Item, <<"=">>) of
             [Key, Val] ->
                 [{Key, Val}];
             _ ->
                 []
         end
-                  end, Parts).
+                                 end, Tokens)).
-
-%% 将 "key=val,key1=val1"的格式转换成 ["key=val", "key1=val1"]
--spec split_keys(Input :: binary()) -> [binary()].
-split_keys(Input) when is_binary(Input) ->
-    split_keys(Input, false, [], []).
-split_keys(<<>>, _Quotes, Current, Acc) ->
-    Last = list_to_binary(lists:reverse(Current)),
-    lists:reverse([Last|Acc]);
-split_keys(<<$,, Rest/binary>>, Quotes, Current, Acc) ->
-    case Quotes of
-        true ->
-            split_keys(Rest, Quotes, [$,|Current], Acc);
-        false ->
-            Part = list_to_binary(lists:reverse(Current)),
-            split_keys(Rest, Quotes, [], [Part|Acc])
-    end;
-split_keys(<<$\", Rest/binary>>, Quotes, Current, Acc) ->
-    split_keys(Rest, not Quotes, [$\"|Current], Acc);
-split_keys(<<Char/utf8, Rest/binary>>, Quotes, Current, Acc) ->
-    split_keys(Rest, Quotes, [Char|Current], Acc).