Revision: 7711
http://languagetool.svn.sourceforge.net/languagetool/?rev=7711&view=rev
Author: taolin2011
Date: 2012-07-17 04:17:59 +0000 (Tue, 17 Jul 2012)
Log Message:
-----------
add some Chinese rules
Modified Paths:
--------------
trunk/JLanguageTool/src/rules/zh/grammar.xml
Added Paths:
-----------
trunk/JLanguageTool/src/java/org/languagetool/rules/zh/
trunk/JLanguageTool/src/java/org/languagetool/rules/zh/ChineseWordRepeatBeginningRule.java
Added:
trunk/JLanguageTool/src/java/org/languagetool/rules/zh/ChineseWordRepeatBeginningRule.java
===================================================================
---
trunk/JLanguageTool/src/java/org/languagetool/rules/zh/ChineseWordRepeatBeginningRule.java
(rev 0)
+++
trunk/JLanguageTool/src/java/org/languagetool/rules/zh/ChineseWordRepeatBeginningRule.java
2012-07-17 04:17:59 UTC (rev 7711)
@@ -0,0 +1,110 @@
+package org.languagetool.rules.zh;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.ResourceBundle;
+
+import org.languagetool.AnalyzedSentence;
+import org.languagetool.AnalyzedTokenReadings;
+import org.languagetool.Language;
+import org.languagetool.rules.Category;
+import org.languagetool.rules.Rule;
+import org.languagetool.rules.RuleMatch;
+
+public class ChineseWordRepeatBeginningRule extends Rule{
+
+ private String lastEndToken ="" ;
+ private String lastToken = "";
+ private String beforeLastToken = "";
+
+ public ChineseWordRepeatBeginningRule(final ResourceBundle messages,
final Language language) {
+ super(messages);
+ super.setCategory(new
Category(messages.getString("category_misc")));
+ }
+
+ @Override
+ public String getId() {
+ return "WORD_REPEAT_BEGINNING_RULE";
+ }
+
+ @Override
+ public String getDescription() {
+ return messages.getString("desc_repetition_beginning");
+ }
+
+ protected boolean isAdverb(AnalyzedTokenReadings token) {
+ return false;
+ }
+
+ public boolean isException(String token) {
+ // avoid warning when having lists like "2007: ..." or the like
+ if (token.equals(":") || token.equals("–") || token.equals("-")) {
+ return true;
+ }
+ return false;
+ }
+
+ @Override
+ public RuleMatch[] match(final AnalyzedSentence text) {
+
+ final List<RuleMatch> ruleMatches = new ArrayList<RuleMatch>();
+ final AnalyzedTokenReadings[] tokens =
text.getTokensWithoutWhitespace();
+
+ if (lastEndToken.equals("，")){
+ lastEndToken = tokens[tokens.length -1].getToken();
+ return toRuleMatchArray(ruleMatches);
+ }
+
+
+
+ if (tokens.length>3) {
+ final AnalyzedTokenReadings analyzedToken = tokens[1];
+ final String token = analyzedToken.getToken();
+ // avoid "..." etc. to be matched:
+ boolean isWord = true;
+ if (token.length() == 1) {
+ final char c = token.charAt(0);
+ if (!Character.isLetter(c)) {
+ isWord = false;
+ }
+ }
+
+ if (isWord && lastToken.equals(token)
+ && !isException(token) && !isException(tokens[2].getToken())
&& !isException(tokens[3].getToken())) {
+ final String shortMsg;
+ if (isAdverb(analyzedToken)) {
+ shortMsg =
messages.getString("desc_repetition_beginning_adv");
+ } else if (beforeLastToken.equals(token)) {
+ shortMsg =
messages.getString("desc_repetition_beginning_word");
+ } else {
+ shortMsg = "";
+ }
+
+ if (!shortMsg.equals("")) {
+ final String msg = shortMsg + " " +
messages.getString("desc_repetition_beginning_thesaurus");
+ final int startPos = analyzedToken.getStartPos();
+ final int endPos = startPos + token.length();
+ final RuleMatch ruleMatch = new RuleMatch(this, startPos,
endPos, msg, shortMsg);
+ ruleMatches.add(ruleMatch);
+ }
+ }
+ beforeLastToken = lastToken;
+ lastToken = token;
+
+ lastEndToken = tokens[tokens.length -1].getToken();
+ }
+
+ //TODO should we ignore repetitions involving multiple paragraphs?
+ //if (tokens[tokens.length - 1].isParaEnd()) beforeLastToken = "";
+
+ return toRuleMatchArray(ruleMatches);
+ }
+
+ @Override
+ public void reset() {
+ lastToken = "";
+ beforeLastToken = "";
+ }
+
+
+}
Modified: trunk/JLanguageTool/src/rules/zh/grammar.xml
===================================================================
--- trunk/JLanguageTool/src/rules/zh/grammar.xml 2012-07-17 03:22:15 UTC
(rev 7710)
+++ trunk/JLanguageTool/src/rules/zh/grammar.xml 2012-07-17 04:17:59 UTC
(rev 7711)
@@ -20,6 +20,671 @@
<!-- 拼写： 词语错误 -->
<!--
====================================================================== -->
<category name="词语错误">
+
+ <rulegroup id="ZHEN1_ZHEN2" name="震/振">
+ <rule>
+ <pattern>
+ <marker>
+ <token>振</token>
+ </marker>
+ <token regexp="yes" postag="v|vg"
postag_regexp="yes">惊|吓|怒|慑|撼
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>震</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="震">
+ 他大为
+ <marker>振</marker>
+ 怒。
+ </example>
+ <example type="correct">他大为震怒。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>震</token>
+ </marker>
+ <token regexp="yes" postag="v|vg"
postag_regexp="yes">动|奋</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>振</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="振">
+ 这首歌
+ <marker>震</marker>
+ 奋人心。
+ </example>
+ <example type="correct">这首歌振奋人心。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="ZUO1_ZUO2" name="坐/座">
+ <rule>
+ <pattern>
+ <token>坐位</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>座位</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="座位">
+ 这就是您的
+ <marker>坐位</marker>
+ 。
+ </example>
+ <example type="correct">这就是您的座位。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token>座标</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>坐标</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="坐标">
+ 他的
+ <marker>座标</marker>
+ 是多少？
+ </example>
+ <example type="correct">他的坐标是多少？</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="ZUN1_ZUN2" name="尊/遵">
+ <rule>
+ <pattern>
+ <marker>
+ <token>遵</token>
+ </marker>
+ <token regexp="yes" postag="v|a"
postag_regexp="yes">敬|重|称|奉
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>尊</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="尊">
+ 你是我最
+ <marker>遵</marker>
+ 重的人.
+ </example>
+ <example type="correct">你是我最尊重的人.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>尊</token>
+ </marker>
+ <token postag="v">守</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>遵</suggestion>
+ 守”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="遵">
+ 你怎么能不
+ <marker>尊</marker>
+ 守交通规则？
+ </example>
+ <example type="correct">你怎么能不遵守交通规则？</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="ZUO11_ZUO2" name="做/作">
+ <rule>
+ <pattern>
+ <marker>
+ <token>作</token>
+ </marker>
+ <token postag="n|v" regexp="yes"
postag_regexp="yes">官|戏|生意|功课|文章|买卖|鬼脸
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>做</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="做">
+ 我来此地是
+ <marker>作</marker>
+ 生意的。
+ </example>
+ <example type="correct">我来此地是做生意的。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>做</token>
+ </marker>
+ <token postag="n|v|k" regexp="yes"
postag_regexp="yes">画|文|业|风|品|者
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>作</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="作">
+ 他的
+ <marker>做</marker>
+ 品很受欢迎。
+ </example>
+ <example type="correct">他的作品很受欢迎。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="BAN1_BAN2" name="绊/拌">
+ <rule>
+ <pattern>
+ <marker>
+ <token>拌</token>
+ </marker>
+ <token postag="n|v" regexp="yes"
postag_regexp="yes">脚|倒|住</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>绊</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="绊">
+ 你愿意做一个
+ <marker>拌</marker>
+ 脚石吗？
+ </example>
+ <example type="correct">你愿意做一个绊脚石？</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token>羁</token>
+ <marker>
+ <token>拌</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“\1
+ <suggestion>绊</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="绊">
+ 他们被家俗的习惯所羁
+ <marker>拌</marker>
+ 。
+ </example>
+ <example type="correct">他们被家俗的习惯所羁绊。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token postag="v">搅</token>
+ <marker>
+ <token>绊</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“\1
+ <suggestion>拌</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="拌">
+ 快把布丁搅
+ <marker>绊</marker>
+ 一下。
+ </example>
+ <example type="correct">快把布丁搅拌一下。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="BAO11_BAO2" name="暴/曝">
+ <rule>
+ <pattern>
+ <marker>
+ <token>曝</token>
+ </marker>
+ <token postag="v|n" regexp="yes"
postag_regexp="yes">君|力|风|雨|雪|露|漏|掠|跌
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>暴</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="暴">
+ 夏桀是中国历史上第一位
+ <marker>曝</marker>
+ 君。
+ </example>
+ <example type="correct">夏桀是中国历史上第一位暴君。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>曝</token>
+ </marker>
+ <token postag="a|n" regexp="yes"
postag_regexp="yes">乱|沙|风</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>暴</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="暴">
+ 警方把
+ <marker>曝</marker>
+ 乱镇压下去了。
+ </example>
+ <example type="correct">警方把暴乱镇压下去了。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token>暴光</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>曝光</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="曝光">
+ 胶卷不小心
+ <marker>暴光</marker>
+ 了。
+ </example>
+ <example type="correct">胶卷不小心曝光了。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="CAI1_CAI2" name="采/彩">
+ <rule>
+ <pattern>
+ <marker>
+ <token>彩</token>
+ </marker>
+ <token postag="v|n" regexp="yes"
postag_regexp="yes">访|伐|集|收|购|取|选|矿|药|木
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>采</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="采">
+ 校长上午接受了记者的
+ <marker>彩</marker>
+ 访。
+ </example>
+ <example type="correct">校长上午接受了记者的采访。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>采</token>
+ </marker>
+ <token postag="n|v|ng" regexp="yes"
postag_regexp="yes">虹|笔|绸|色|灯|画|票|电|排|带
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>彩</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="彩">
+ 雨后的天边出现一道
+ <marker>采</marker>
+ 虹。
+ </example>
+ <example type="correct">雨后的天边出现一道彩虹。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token>色</token>
+ <marker>
+ <token postag="v">采</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“\1
+ <suggestion>彩</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="彩">
+ 这幅画色
+ <marker>采</marker>
+ 鲜明。
+ </example>
+ <example type="correct">这幅画色彩鲜明。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>精采</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>精彩</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="精彩">
+ 这场球赛踢得很
+ <marker>精采</marker>
+ 。
+ </example>
+ <example type="correct">这场球赛踢得很精彩。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="CHI1_CHI2" name="驰/弛">
+ <rule>
+ <pattern>
+ <token postag="vg|v|dg" regexp="yes"
postag_regexp="yes">疾|奔
+ </token>
+ <marker>
+ <token>弛</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“\1
+ <suggestion>驰</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="驰">
+ 汽车在公路上疾
+ <marker>弛</marker>
+ 。
+ </example>
+ <example type="correct">汽车在公路上疾驰。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>松驰</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>松弛</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="松弛">
+ 微笑能化紧张为
+ <marker>松驰</marker>
+ 。
+ </example>
+ <example type="correct">微笑能化紧张为松弛。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="BAO111_BAO2" name="薄/簙">
+ <rule>
+ <pattern>
+ <marker>
+ <token>簙</token>
+ </marker>
+ <token postag="n"
regexp="yes">饼|板</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>薄</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="薄">
+ 这家店以做
+ <marker>簙</marker>
+ 饼好吃而闻名。
+ </example>
+ <example type="correct">这家店以做薄饼好吃而闻名。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token postag="n|v" regexp="yes"
postag_regexp="yes">刻|轻</token>
+ <marker>
+ <token>簙</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“\1
+ <suggestion>薄</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect">
+ 你为何对我这样刻
+ <marker>簙</marker>
+ ？
+ </example>
+ <example type="correct">你为何对我这样刻薄？</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="BAN1_BAN2_BAN3" name="板/扳/版">
+ <rule>
+ <pattern>
+ <token postag="n|mg|v|j" regexp="yes"
postag_regexp="yes">甲|画|篮|黑
+ </token>
+ <marker>
+ <token regexp="yes">版|扳</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“\1
+ <suggestion>板</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="板">
+ 海浪拍在甲
+ <marker>版</marker>
+ 上。
+ </example>
+ <example type="correct">海浪拍在甲板上。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token regexp="yes">扳|板</token>
+ </marker>
+ <token postag="n"
regexp="yes">权|署</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>版</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect">
+ 他拥有这本书的
+ <marker>板</marker>
+ 权。
+ </example>
+ <example type="correct">他拥有这本书的版权。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token regexp="yes">版|板</token>
+ </marker>
+ <token postag="n|v" regexp="yes"
postag_regexp="yes">手|倒</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>扳</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect">
+ 请把
+ <marker>板</marker>
+ 手递给我。
+ </example>
+ <example type="correct">请把扳手递给我。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="BIAN1_BIAN2_BIAN3_BAN4" name="辩/辨/辫/瓣">
+ <rule>
+ <pattern>
+ <marker>
+ <token
regexp="yes">辨|辫|瓣</token>
+ </marker>
+ <token postag="v">护</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>辩</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect">
+ 被告拒绝律师为自己
+ <marker>辨</marker>
+ 护。
+ </example>
+ <example type="correct">被告拒绝律师为自己辩护。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token
regexp="yes">辩|辫|瓣</token>
+ </marker>
+ <token>别</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>辨</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect">
+ 作为成年人，我们要有
+ <marker>辩</marker>
+ 别是非的能力。
+ </example>
+ <example
type="correct">作为成年人，我们要有辨别是非的能力。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token
regexp="yes">辩|辨|瓣</token>
+ </marker>
+ <token>子</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>辫</suggestion>
+ \”2吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect">
+ 清朝男人都留着
+ <marker>辨</marker>
+ 子。
+ </example>
+ <example type="correct">清朝男人都留着辫子。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token postag="n"
regexp="yes">花|豆|蒜</token>
+ <marker>
+ <token
regexp="yes">辩|辨|辫</token>
+ </marker>
+ </pattern>
+ <message>
+ 您的意思是“\1
+ <suggestion>瓣</suggestion>
+ ”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect">
+ 粉色的花蕊被包裹在花
+ <marker>辫</marker>
+ 中。
+ </example>
+ <example type="correct">粉色的花蕊被包裹在花瓣中。</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="DI1_DI2_DI3" name="诋/抵/砥">
+ <rule>
+ <pattern>
+ <marker>
+ <token>抵</token>
+ </marker>
+ <token postag="v">毁</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>诋</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="诋">
+ 他的个人声望受到
+ <marker>抵</marker>
+ 毁。
+ </example>
+ <example type="correct">他的个人声望受到诋毁。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>诋</token>
+ </marker>
+ <token postag="v">
+ 达
+ <exception>达洞湾村</exception>
+ </token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>抵</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="抵">
+ 顺着这条小溪便可
+ <marker>诋</marker>
+ 达小河村。
+ </example>
+ <example type="correct">顺着这条小溪便可抵达小河村。</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>抵</token>
+ </marker>
+ <token postag="vg">砺</token>
+ </pattern>
+ <message>
+ 您的意思是“
+ <suggestion>砥</suggestion>
+ \2”吗？
+ </message>
+ <short>词语错误</short>
+ <example type="incorrect" correction="砥">
+ <marker>抵</marker>
+ 砺革命意志。
+ </example>
+ <example type="correct">砥砺革命意志。</example>
+ </rule>
+ </rulegroup>
<rulegroup id="SHA1_SHA2" name="霎/刹">
<rule>
<pattern>
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Languagetool-cvs mailing list
Languagetool-cvs@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-cvs
