修改 IllegalWordsSearch

2026-10-17 04:40:38 +08:00 · 2020-05-24 11:55:03 +08:00
parent a9afd1eb8e
commit c1347e028f
8 changed files with 540 additions and 393 deletions
@@ -155,7 +155,7 @@ ToolGood.Words:                 本项目源代码
 ## 性能对比
 执行10万次性能对比，结果如下：

-![10W次性能对比](https://toolgood.github.io/ToolGood.Words/10W次性能对比_20200411.png)
+![10W次性能对比](https://toolgood.github.io/ToolGood.Words/10W次性能对比_20200524.png)

 注:C#自带正则很慢，`StringSearchEx2.ContainsAny`是`Regex.IsMatch`效率的8.8万倍多，跟关键字数量有关。

@@ -25,7 +25,7 @@ namespace ToolGood.Words.Contrast
        static WordsSearchEx3 wordsSearchEx3 = new WordsSearchEx3();


-        //static IllegalWordsSearch illegalWordsSearch = new IllegalWordsSearch();
+        static IllegalWordsSearch illegalWordsSearch = new IllegalWordsSearch();



@@ -45,7 +45,7 @@ namespace ToolGood.Words.Contrast
            Run("StringSearchEx（ContainsAny）--- WordsSearchEx（ContainsAny）代码相同", () => { stringSearchEx.ContainsAny(text); });
            Run("StringSearchEx2（ContainsAny）--- WordsSearchEx2（ContainsAny）代码相同", () => { stringSearchEx2.ContainsAny(text); });
            Run("StringSearchEx3（ContainsAny）--- WordsSearchEx3（ContainsAny）代码相同", () => { stringSearchEx3.ContainsAny(text); });
-            //Run("IllegalWordsSearch（ContainsAny）", () => { illegalWordsSearch.ContainsAny(text); });
+            Run("IllegalWordsSearch（ContainsAny）", () => { illegalWordsSearch.ContainsAny(text); });

            Run("StringSearch（FindFirst）", () => { stringSearch.FindFirst(text); });
            Run("StringSearchEx（FindFirst）", () => { stringSearchEx.FindFirst(text); });
@@ -55,7 +55,7 @@ namespace ToolGood.Words.Contrast
            Run("WordsSearchEx（FindFirst）", () => { wordsSearchEx.FindFirst(text); });
            Run("WordsSearchEx2（FindFirst）", () => { wordsSearchEx2.FindFirst(text); });
            Run("WordsSearchEx3（FindFirst）", () => { wordsSearchEx3.FindFirst(text); });
-            //Run("IllegalWordsSearch（FindFirst）", () => { illegalWordsSearch.FindFirst(text); });
+            Run("IllegalWordsSearch（FindFirst）", () => { illegalWordsSearch.FindFirst(text); });


            Console.Write("-------------------- Find All 100000次 --------------------\r\n");
@@ -70,7 +70,7 @@ namespace ToolGood.Words.Contrast
            Run("WordsSearchEx（FindAll）", () => { wordsSearchEx.FindAll(text); });
            Run("WordsSearchEx2（FindAll）", () => { wordsSearchEx2.FindAll(text); });
            Run("WordsSearchEx3（FindAll）", () => { wordsSearchEx3.FindAll(text); });
-            //Run("IllegalWordsSearch（FindAll）", () => { illegalWordsSearch.FindAll(text); });
+            Run("IllegalWordsSearch（FindAll）", () => { illegalWordsSearch.FindAll(text); });

            Console.Write("-------------------- Replace  100000次 --------------------\r\n");
            Run("TrieFilter（Replace）", () => { tf1.Replace(text); });
@@ -80,7 +80,7 @@ namespace ToolGood.Words.Contrast
            Run("StringSearchEx（Replace）--- WordsSearchEx（Replace）代码相同", () => { stringSearchEx.Replace(text); });
            Run("StringSearchEx2（Replace）--- WordsSearchEx2（Replace）代码相同", () => { stringSearchEx2.Replace(text); });
            Run("StringSearchEx3（Replace）--- WordsSearchEx3（Replace）代码相同", () => { stringSearchEx3.Replace(text); });
-            //Run("IllegalWordsSearch（Replace）", () => { illegalWordsSearch.Replace(text); });
+            Run("IllegalWordsSearch（Replace）", () => { illegalWordsSearch.Replace(text); });

            Console.Write("-------------------- Regex  100次 --------------------\r\n");
            Run(100, "Regex.IsMatch", () => { re.IsMatch(text); });
@@ -142,7 +142,7 @@ namespace ToolGood.Words.Contrast
            wordsSearchEx.SetKeywords(list);
            wordsSearchEx2.SetKeywords(list);
            wordsSearchEx3.SetKeywords(list);
-            //illegalWordsSearch.SetKeywords(list);
+            illegalWordsSearch.SetKeywords(list);

            list = list.OrderBy(q => q).ToList();
            var str = string.Join("|", list);
@@ -54,6 +54,7 @@ namespace ToolGood.Words.Test
            Assert.AreEqual("zg人", all[0].Keyword);

            test = "fuck al[]l"; //未启用跳词
+            iwords.UseSkipWordFilter = false; 
            all = iwords.FindAll(test);
            Assert.AreEqual("fuck", all[0].Keyword);
            Assert.AreEqual(1, all.Count);
@@ -73,6 +74,7 @@ namespace ToolGood.Words.Test
            Assert.AreEqual(1, all.Count);

            test = "asssert all";
+            iwords.UseDuplicateWordFilter = false; //启用重复词
            all = iwords.FindAll(test); //未启用重复词
            Assert.AreEqual("all", all[0].Keyword);
            Assert.AreEqual(1, all.Count);
@@ -113,12 +115,12 @@ namespace ToolGood.Words.Test
            var ss = iwords.Replace(test, '*');
            Assert.AreEqual("我是【****", ss);

-            test = "我是中国人"; //使用黑名单
-            iwords.SetBlacklist(bl);
-            iwords.UseBlacklistFilter = true;
-            all = iwords.FindAll(test, 1);
-            Assert.AreEqual("中国", all[0].Keyword);
-            Assert.AreEqual(1, all.Count);
+            //test = "我是中国人"; //使用黑名单
+            //iwords.SetBlacklist(bl);
+            //iwords.UseBlacklistFilter = true;
+            //all = iwords.FindAll(test, 1);
+            //Assert.AreEqual("中国", all[0].Keyword);
+            //Assert.AreEqual(1, all.Count);

        }

@@ -20,7 +20,9 @@ namespace ToolGood.Words.Test
            illegalWordsSearch.SetKeywords(s.Split('|'));
            var str = illegalWordsSearch.Replace("我是中美国人厉害中国完美ａｂｃddb好的", '*');

-            Assert.AreEqual("我是中美国人厉害**完美***ddb好的", str);
+            //Assert.AreEqual("我是中美国人厉害**完美***ddb好的", str);
+            //注，ａｂｃ先转abc,再判断abc左右是否为英文或数字,因为后面为d是英文，所以不能过滤
+            Assert.AreEqual("我是中美国人厉害**完美ａｂｃddb好的", str);
        }


@@ -40,16 +42,19 @@ namespace ToolGood.Words.Test
            iws.UseIgnoreCase = true;
            iws.UseDBCcaseConverter = true;
            var iwsFirst = iws.FindFirst(text);
+            Assert.AreEqual("吃", iwsFirst.Keyword);
            var iwsAll = iws.FindAll(text);
-            Assert.AreEqual(9, iwsAll.Count);
+            Assert.AreEqual(1, iwsAll.Count);// 因为1A20左右都是英文或数字，所以识别失败

            ss.SetKeywords(keywords);
            var ssFirst = ss.FindFirst(text);
+            Assert.AreEqual("吃", ssFirst);
            var ssAll = ss.FindAll(text);
            Assert.AreEqual(9, ssAll.Count);

            sse.SetKeywords(keywords);
            var sseFirst = sse.FindFirst(text);
+            Assert.AreEqual("吃", sseFirst);
            var sseAll = sse.FindAll(text);
            Assert.AreEqual(9, sseAll.Count);

@@ -2,20 +2,14 @@
 {
    public class IllegalWordsSearchResult
    {
-        internal IllegalWordsSearchResult(string keyword, int start, int end, string srcText)
+
+        internal IllegalWordsSearchResult(string keyword, int start, int end, int index, string matchKeyword, int type)
        {
-            MatchKeyword = keyword;
+            MatchKeyword = matchKeyword;
            End = end;
            Start = start;
-            Keyword = srcText.Substring(Start, end - Start + 1);
-            BlacklistType = BlacklistType.None;
-        }
-        internal IllegalWordsSearchResult(string keyword, int start, int end, string srcText, BlacklistType type)
-        {
-            MatchKeyword = keyword;
-            End = end;
-            Start = start;
-            Keyword = srcText.Substring(Start, end - Start + 1);
+            Index = index;
+            Keyword = keyword;
            BlacklistType = type;
        }

@@ -31,6 +25,12 @@
        /// 原始文本
        /// </summary>
        public string Keyword { get; private set; }
+
+        /// <summary>
+        /// 索引
+        /// </summary>
+        public int Index { get; private set; }
+
        /// <summary>
        /// 关键字
        /// </summary>
@@ -39,7 +39,7 @@
        /// <summary>
        /// 黑名单类型
        /// </summary>
-        public BlacklistType BlacklistType { get; private set; }
+        public int BlacklistType { get; private set; }


        public override string ToString()
@@ -19,7 +19,7 @@
    <RootNamespace>ToolGood.Words</RootNamespace>
    <GeneratePackageOnBuild>true</GeneratePackageOnBuild>
    <PackageRequireLicenseAcceptance>true</PackageRequireLicenseAcceptance>
-    <Version>3.0.1.3</Version>
+    <Version>3.0.2.0</Version>
    <PackageReleaseNotes></PackageReleaseNotes>
    <PackageLicenseFile>LICENSE</PackageLicenseFile>
    <AllowUnsafeBlocks>true</AllowUnsafeBlocks>