From c0090f7678704d67f4f022fb4b009ac0e2d8c460 Mon Sep 17 00:00:00 2001 From: vsxd Date: Mon, 13 Jan 2025 15:11:07 +0800 Subject: [PATCH 1/3] rename local var --- hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java b/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java index b52133be7..fa193ddad 100755 --- a/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java +++ b/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java @@ -98,7 +98,7 @@ public class WordTree extends HashMap { * @return this */ public WordTree addWord(String word) { - final Filter charFilter = this.charFilter; + final Filter localCharFilter = this.charFilter; WordTree parent = null; WordTree current = this; WordTree child; @@ -106,7 +106,7 @@ public class WordTree extends HashMap { final int length = word.length(); for (int i = 0; i < length; i++) { currentChar = word.charAt(i); - if (charFilter.accept(currentChar)) {//只处理合法字符 + if (localCharFilter.accept(currentChar)) {//只处理合法字符 child = current.get(currentChar); if (child == null) { //无子类，新建一个子节点后存放下一个字符 @@ -245,7 +245,7 @@ public class WordTree extends HashMap { List foundWords = new ArrayList<>(); WordTree current = this; final int length = text.length(); - final Filter charFilter = this.charFilter; + final Filter localCharFilter = this.charFilter; //存放查找到的字符缓存。完整出现一个词时加到findedWords中，否则清空 final StringBuilder wordBuffer = StrUtil.builder(); final StringBuilder keyBuffer = StrUtil.builder(); @@ -256,7 +256,7 @@ public class WordTree extends HashMap { for (int j = i; j < length; j++) { currentChar = text.charAt(j); // Console.log("i: {}, j: {}, currentChar: {}", i, j, currentChar); - if (false == charFilter.accept(currentChar)) { + if (false == localCharFilter.accept(currentChar)) { if (wordBuffer.length() > 0) { //做为关键词中间的停顿词被当作关键词的一部分被返回 wordBuffer.append(currentChar); From 86ebc3bc2d69813da775e6f400a3eee07e454e9b Mon Sep 17 00:00:00 2001 From: vsxd Date: Mon, 13 Jan 2025 15:11:36 +0800 Subject: [PATCH 2/3] rm repeated stop chars --- .../src/main/java/cn/hutool/dfa/StopChar.java | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/hutool-dfa/src/main/java/cn/hutool/dfa/StopChar.java b/hutool-dfa/src/main/java/cn/hutool/dfa/StopChar.java index 9aaa7b6a9..583380653 100755 --- a/hutool-dfa/src/main/java/cn/hutool/dfa/StopChar.java +++ b/hutool-dfa/src/main/java/cn/hutool/dfa/StopChar.java @@ -20,14 +20,14 @@ public class StopChar { '※', '→', '←', '↑', '↓', '↖', '↗', '↘', '↙', '〓', 'ⅰ', 'ⅱ', 'ⅲ', 'ⅳ', 'ⅴ', 'ⅵ', 'ⅶ', 'ⅷ', 'ⅸ', 'ⅹ', '①', // '②', '③', '④', '⑤', '⑥', '⑦', '⑧', '⑨', '⑩', '⒈', '⒉', '⒊', '⒋', '⒌', '⒍', '⒎', '⒏', '⒐', '⒑', '⒒', '⒓', // '⒔', '⒕', '⒖', '⒗', '⒘', '⒙', '⒚', '⒛', '⑴', '⑵', '⑶', '⑷', '⑸', '⑹', '⑺', '⑻', '⑼', '⑽', '⑾', '⑿', '⒀', // - '⒁', '⒂', '⒃', '⒄', '⒅', '⒆', '⒇', 'Ⅰ', 'Ⅱ', 'Ⅲ', 'Ⅳ', 'Ⅴ', 'Ⅵ', 'Ⅶ', 'Ⅷ', 'Ⅸ', 'Ⅹ', 'Ⅺ', 'Ⅻ', '！', '”', // - '＃', '￥', '％', '＆', '’', '（', '）', '＊', '＋', '，', '－', '．', '／', '０', '１', '２', '３', '４', '５', '６', '７', // - '８', '９', '：', '；', '＜', '＝', '＞', '？', '＠', '〔', '＼', '〕', '＾', '＿', '‘', '｛', '｜', '｝', '∏', 'Ρ', '∑', // + '⒁', '⒂', '⒃', '⒄', '⒅', '⒆', '⒇', 'Ⅰ', 'Ⅱ', 'Ⅲ', 'Ⅳ', 'Ⅴ', 'Ⅵ', 'Ⅶ', 'Ⅷ', 'Ⅸ', 'Ⅹ', 'Ⅺ', 'Ⅻ', '！', // + '＃', '￥', '％', '＆', '（', '）', '＊', '，', '．', '／', '０', '１', '２', '３', '４', '５', '６', '７', // + '８', '９', '：', '；', '＜', '＝', '＞', '？', '＠', '＼', '＾', '＿', '｛', '｜', '｝', 'Ρ', // 'Υ', 'Φ', 'Χ', 'Ψ', 'Ω', 'α', 'β', 'γ', 'δ', 'ε', 'ζ', 'η', 'θ', 'ι', 'κ', 'λ', 'μ', 'ν', 'ξ', 'ο', 'π', // - 'ρ', 'σ', 'τ', 'υ', 'φ', 'χ', 'ψ', 'ω', '（', '）', '〔', '〕', '＾', '﹊', '﹍', '╭', '╮', '╰', '╯', '', '_', // - '', '^', '（', '^', '：', '！', '/', '\\', '\"', '<', '>', '`', '·', '。', '{', '}', '~', '～', '(', ')', '-', // - '√', '$', '@', '*', '&', '#', '卐', '㎎', '㎏', '㎜', '㎝', '㎞', '㎡', '㏄', '㏎', '㏑', '㏒', '㏕', '+', '=', '?', - ':', '.', '!', ';', ']','|','%'); + 'ρ', 'σ', 'τ', 'υ', 'φ', 'χ', 'ψ', 'ω', '﹊', '﹍', '╭', '╮', '╰', '╯', '', '_', // + '/', '\\', '\"', '<', '>', '`', '{', '}', '~', '(', ')', '-', // + '$', '@', '*', '&', '#', '卐', '㎎', '㎏', '㎜', '㎝', '㎞', '㎡', '㏄', '㏎', '㏑', '㏒', '㏕', '+', '=', '?', + ':', '.', '!', ';', ']','|','%'); /** * 判断指定的词是否是不处理的词。 如果参数为空，则返回true，因为空也属于不处理的字符。 From 7467a2cb5baf2df2bd8f80b0bf952b18cfcf986a Mon Sep 17 00:00:00 2001 From: vsxd Date: Tue, 14 Jan 2025 10:02:27 +0800 Subject: [PATCH 3/3] Revert "rename local var" This reverts commit c0090f7678704d67f4f022fb4b009ac0e2d8c460. --- hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java b/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java index fa193ddad..b52133be7 100755 --- a/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java +++ b/hutool-dfa/src/main/java/cn/hutool/dfa/WordTree.java @@ -98,7 +98,7 @@ public class WordTree extends HashMap { * @return this */ public WordTree addWord(String word) { - final Filter localCharFilter = this.charFilter; + final Filter charFilter = this.charFilter; WordTree parent = null; WordTree current = this; WordTree child; @@ -106,7 +106,7 @@ public class WordTree extends HashMap { final int length = word.length(); for (int i = 0; i < length; i++) { currentChar = word.charAt(i); - if (localCharFilter.accept(currentChar)) {//只处理合法字符 + if (charFilter.accept(currentChar)) {//只处理合法字符 child = current.get(currentChar); if (child == null) { //无子类，新建一个子节点后存放下一个字符 @@ -245,7 +245,7 @@ public class WordTree extends HashMap { List foundWords = new ArrayList<>(); WordTree current = this; final int length = text.length(); - final Filter localCharFilter = this.charFilter; + final Filter charFilter = this.charFilter; //存放查找到的字符缓存。完整出现一个词时加到findedWords中，否则清空 final StringBuilder wordBuffer = StrUtil.builder(); final StringBuilder keyBuffer = StrUtil.builder(); @@ -256,7 +256,7 @@ public class WordTree extends HashMap { for (int j = i; j < length; j++) { currentChar = text.charAt(j); // Console.log("i: {}, j: {}, currentChar: {}", i, j, currentChar); - if (false == localCharFilter.accept(currentChar)) { + if (false == charFilter.accept(currentChar)) { if (wordBuffer.length() > 0) { //做为关键词中间的停顿词被当作关键词的一部分被返回 wordBuffer.append(currentChar);