ft: 代码调整

README.md

@@ -1,17 +1,17 @@
 eAcs
 =====
-    基于ac算法实现的快速高效的erl版本敏感词匹配,检查,过滤代码
+    基于ac算法实现的快速高效的敏感词匹配,检查,过滤功能
 
 Build
 -----
-    $ rebar3 escriptize   -> genAcs
+    $ rebar3 escriptize   ->   genAcs
     $ rebar3 compile
 
 Uses
 -----
     创建 acsTree.erl
-        脚本生成：./genAcs SWord.txtFile OutputDir  或者函数嗲用
-        函数调用： genAcs:main([SWord.txtFile, OutputDir])
+        脚本生成：./genAcs SWordFile OutputDir
+        函数调用： genAcs:main([SWordFile, OutputDir])
     匹配 检查 过滤 敏感词
         eAcs:matchSw/1               %% 返回匹配的敏感词列表
         eAcs:isHasSw/1               %% 检查是否包含敏感词

src/eAcs.erl

@@ -4,6 +4,7 @@
    matchSw/1                 %% 返回匹配的敏感词列表
    , isHasSw/1               %% 检查是否包含敏感词
    , replaceSw/1             %% 替换敏感词
+   , strSize/2               %% 获取utf8字符串的长度
 ]).
 
 %% state 0 is the root node
@@ -17,10 +18,10 @@ matchSw(BinStr) ->
 doMatch(<<>>, _, _Index, MatchList) ->
    MatchList;
 doMatch(<<Word/utf8, Tail/binary>>, State, Index, MatchList) ->
-   {NewState, NewMatchList} = matchInner(Word, State, Index, MatchList),
+   {NewState, NewMatchList} = matchWord(Word, State, Index, MatchList),
    doMatch(Tail, NewState, Index + 1, NewMatchList).
 
-matchInner(Word, State, Index, MatchList) ->
+matchWord(Word, State, Index, MatchList) ->
    Node = acsTree:goto(State),
    case Node of
       undefined ->
@@ -29,7 +30,7 @@ matchInner(Word, State, Index, MatchList) ->
                {State, MatchList};
             _ ->
                {NextState, _} = acsTree:failOut(State),
-               matchInner(Word, NextState, Index, MatchList)
+               matchWord(Word, NextState, Index, MatchList)
          end;
       _ ->
          case Node of
@@ -42,7 +43,7 @@ matchInner(Word, State, Index, MatchList) ->
                      {State, MatchList};
                   _ ->
                      {NextState, _} = acsTree:failOut(State),
-                     matchInner(Word, NextState, Index, MatchList)
+                     matchWord(Word, NextState, Index, MatchList)
                end
          end
    end.
@@ -55,7 +56,7 @@ getOutput(State, Index, MatchList) ->
       undefined ->
          getOutput(FailState, Index, MatchList);
       _ ->
-         NewMatchList = [{Index - patternSize(Pattern, 0) + 1, Index, Pattern} | MatchList],
+         NewMatchList = [{Index - Pattern + 1, Pattern} | MatchList],
          getOutput(FailState, Index, NewMatchList)
    end.
 
@@ -68,14 +69,14 @@ isHasSw(BinStr) ->
 doMatch(<<>>, _) ->
    false;
 doMatch(<<Word/utf8, Tail/binary>>, State) ->
-   case matchInner(Word, State) of
+   case matchWord(Word, State) of
       true ->
          true;
       NewState ->
          doMatch(Tail, NewState)
    end.
 
-matchInner(Word, State) ->
+matchWord(Word, State) ->
    Node = acsTree:goto(State),
    case Node of
       undefined ->
@@ -84,7 +85,7 @@ matchInner(Word, State) ->
                State;
             _ ->
                {NextState, _} = acsTree:failOut(State),
-               matchInner(Word, NextState)
+               matchWord(Word, NextState)
          end;
       _ ->
          case Node of
@@ -101,7 +102,7 @@ matchInner(Word, State) ->
                      State;
                   _ ->
                      {NextState, _} = acsTree:failOut(State),
-                     matchInner(Word, NextState)
+                     matchWord(Word, NextState)
                end
          end
    end.
@@ -110,11 +111,11 @@ getOutput(0) ->
    false;
 getOutput(State) ->
    {FailState, Pattern} = acsTree:failOut(State),
-   case Pattern == undefined orelse FailState == 0 of
-      true ->
-         false;
+   case Pattern of
+      undefined ->
+         getOutput(FailState);
       _ ->
-         getOutput(FailState)
+         true
    end.
 %% *************************************** matchSw end   ***************************************************************
 %% *************************************** replaceSw start *************************************************************
@@ -122,7 +123,7 @@ replaceSw(_BinStr) ->
    ok.
 %% *************************************** replaceSw end   *************************************************************
 
-patternSize(<<>>, Cnt) ->
+strSize(<<>>, Cnt) ->
    Cnt;
-patternSize(<<_Word/utf8, Left/binary>>, Cnt) ->
-   patternSize(Left, Cnt + 1).
+strSize(<<_Word/utf8, Left/binary>>, Cnt) ->
+   strSize(Left, Cnt + 1).

src/genAcs.erl

@@ -23,7 +23,7 @@ dealEveryLine(IoDevice, Goto, Output, MaxState) ->
          case BinStr =/= <<>> of
             true ->
                {NewGoto, NewState, NewMaxState} = addGoto(BinStr, Goto, 0, MaxState),
-               NewOutput = Output#{NewState => BinStr},
+               NewOutput = Output#{NewState => eAcs:strSize(BinStr, 0)},
                dealEveryLine(IoDevice, NewGoto, NewOutput, NewMaxState);
             _ ->
                dealEveryLine(IoDevice, Goto, Output, MaxState)
@@ -146,15 +146,15 @@ doGenGoto([{K, V} | SortKvs], StrAcc) ->
    end.
 
 genFailOut([], _Fail, _Output, StrAcc) ->
-   <<StrAcc/binary, "\nfailOut(_) -> {0, undefined}.\n\n">>;
+   <<StrAcc/binary, "\nfailOut(_) -> {0, undefined}.">>;
 genFailOut([State], Fail, Output, StrAcc) ->
    FailState = maps:get(State, Fail, 0),
    Pattern = maps:get(State, Output, undefined),
    case FailState /= 0 orelse Pattern /= undefined of
       true ->
-         <<StrAcc/binary, "failOut(", (integer_to_binary(State))/binary, ") -> ", (iolist_to_binary(io_lib:format(<<"~w">>, [{FailState, Pattern}])))/binary, ";\nfailOut(_) -> {0, undefined}.\n\n">>;
+         <<StrAcc/binary, "failOut(", (integer_to_binary(State))/binary, ") -> ", (iolist_to_binary(io_lib:format(<<"~w">>, [{FailState, Pattern}])))/binary, ";\nfailOut(_) -> {0, undefined}.">>;
       _ ->
-         <<StrAcc/binary, ";\nfailOut(_) -> {0, undefined}.\n\n">>
+         <<StrAcc/binary, ";\nfailOut(_) -> {0, undefined}.">>
    end;
 genFailOut([State | SortStates], Fail, Output, StrAcc) ->
    FailState = maps:get(State, Fail, 0),

src/test/test.txt

@@ -1,7 +1,6 @@
-去你妈的
-
-
-你妈
+abcd
+cd
+c