代码语言
.
CSharp
.
JS
Java
Asp.Net
C
MSSQL
PHP
Css
PLSQL
Python
Shell
EBS
ASP
Perl
ObjC
VB.Net
VBS
MYSQL
GO
Delphi
AS
DB2
Domino
Rails
ActionScript
Scala
代码分类
文件
系统
字符串
数据库
网络相关
图形/GUI
多媒体
算法
游戏
Jquery
Extjs
Android
HTML5
菜单
网页交互
WinForm
控件
企业应用
安全与加密
脚本/批处理
开放平台
其它
【
Java
】
取汉字拼音首字母含多音字及不常见的字
作者:
善培
/ 发布于
2016/4/26
/
683
package com.anxin.ssk.common; import java.io.UnsupportedEncodingException; import java.util.HashMap; import java.util.Map; import java.util.Map.Entry; import java.util.regex.Matcher; import java.util.regex.Pattern; /** * 取得给定汉字串的首字母串,即声母串 Title: ChineseCharToEn(含常用汉字,不常见汉字及多音字) * * @author: (le.qiao) * @e-mail: qiaolevip@gmail.com * @myblog: <a href="http://qiaolevip.iteye.com">http://qiaolevip.iteye.com</a> * @date: 2014-1-15 注:只支持GB2312字符集中的汉字 * */ public class LetterUtil { private final static int[] li_SecPosValue = { 1601, 1637, 1833, 2078, 2274, 2302, 2433, 2594, 2787, 3106, 3212, 3472, 3635, 3722, 3730, 3858, 4027, 4086, 4390, 4558, 4684, 4925, 5249, 5590 }; private final static String[] lc_FirstLetter = { "a", "b", "c", "d", "e", "f", "g", "h", "j", "k", "l", "m", "n", "o", "p", "q", "r", "s", "t", "w", "x", "y", "z" }; private static Map<String, String> exceptWords = new HashMap<String, String>(); static { exceptWords.put("a", "庵鳌"); exceptWords.put("b", "璧亳並侼別匂"); exceptWords.put("c", "茌丞丒丳刅"); exceptWords.put("d", "渎砀棣儋丟"); exceptWords.put("e", ""); exceptWords.put("f", "邡冹兝"); exceptWords.put("g", "崮藁莞丐丱乢亁仠冮匃匄"); exceptWords.put("h", "骅珲潢湟丆冴匢"); exceptWords.put("j", "泾蛟暨缙旌莒鄄丌丩丮丯丼亅伋冏匊匛匞"); exceptWords.put("k", "丂匟"); exceptWords.put("l", "崂涞栾溧漯浏耒醴泸阆崃両刢劽啰"); exceptWords.put("m", "渑汨丏冐冺兞冇"); exceptWords.put("n", ""); exceptWords.put("o", "瓯"); exceptWords.put("p", "邳濮郫丕伂冸"); exceptWords.put("q", "喬綦衢岐朐邛丠丬亝冾兛匤"); exceptWords.put("r", "榕刄"); exceptWords.put("s", "泗睢沭嵊歙莘嵩鄯丄丗侺兙"); exceptWords.put("t", "潼滕郯亣侹侻"); exceptWords.put("w", "婺涠汶亾仼卍卐"); exceptWords.put("x", "鑫盱浔荥淅浠亵丅伈兇"); exceptWords.put("y", "懿眙黟颍兖郓偃鄢晏丣亜伇偐円匜"); exceptWords.put("z", "梓涿诏柘秭圳伀冑刣"); } private final static String polyphoneTxt = "重庆|cq,音乐|yy"; /** * 取得给定汉字串的首字母串,即声母串 * * @param str 给定汉字串 * @return 声母串 */ public static String getAllFirstLetter(String str) { if (str == null || str.trim().length() == 0) { return ""; } // 多音字判定 for (String polyphone : polyphoneTxt.split(",")) { String[] chinese = polyphone.split("[|]"); if (str.indexOf(chinese[0]) != -1) { str = str.replace(chinese[0], chinese[1]); } } String _str = ""; for (int i = 0; i < str.length(); i++) { _str = _str + getFirstLetter(str.substring(i, i + 1)); } return _str; } /** * 取得给定汉字的首字母,即声母 * * @param chinese 给定的汉字 * @return 给定汉字的声母 */ public static String getFirstLetter(String chinese) { if (chinese == null || chinese.trim().length() == 0) { return ""; } String chineseTemp = chinese; chinese = conversionStr(chinese, "GB2312", "ISO8859-1"); if (chinese.length() > 1) { // 判断是不是汉字 int li_SectorCode = (int) chinese.charAt(0); // 汉字区码 int li_PositionCode = (int) chinese.charAt(1); // 汉字位码 li_SectorCode = li_SectorCode - 160; li_PositionCode = li_PositionCode - 160; int li_SecPosCode = li_SectorCode * 100 + li_PositionCode; // 汉字区位码 if (li_SecPosCode > 1600 && li_SecPosCode < 5590) { for (int i = 0; i < 23; i++) { if (li_SecPosCode >= li_SecPosValue[i] && li_SecPosCode < li_SecPosValue[i + 1]) { chinese = lc_FirstLetter[i]; break; } } } else { // 非汉字字符,如图形符号或ASCII码 chinese = matchPinYin(chinese); } } // 如还是无法匹配,再次进行拼音匹配 if (chinese.equals("?")) { chinese = matchPinYin(chineseTemp, false); } return chinese; } /** * 汉字匹配拼音对照 * * @param chinese * @return */ private static String matchPinYin(String chinese, boolean needConvert) { String chineseTemp = chinese; if (needConvert) { chinese = conversionStr(chinese, "ISO8859-1", "GB2312"); } chinese = chinese.substring(0, 1); // findRepeatWord(exceptWords); for (Entry<String, String> letterSet : exceptWords.entrySet()) { if (letterSet.getValue().indexOf(chinese) != -1) { chinese = letterSet.getKey(); break; } } chinese = chineseTemp.equals(chinese) ? "?" : chinese; return chinese; } private static String matchPinYin(String chinese) { return matchPinYin(chinese, true); } /** * 字符串编码转换 * * @param str 要转换编码的字符串 * @param charsetName 原来的编码 * @param toCharsetName 转换后的编码 * @return 经过编码转换后的字符串 */ private static String conversionStr(String str, String charsetName, String toCharsetName) { try { str = new String(str.getBytes(charsetName), toCharsetName); } catch (UnsupportedEncodingException ex) { System.out.println("字符串编码转换异常:" + ex.getMessage()); } return str; } public static void main(String[] args) { String content = "丬"; String pinyin = ""; String contentAll = ""; String pinyinAll = ""; for (int i = 21000; i <= 22000; i++) { content = ((char) i) + ""; pinyin = LetterUtil.getFirstLetter(content); if (pinyin.equals("?")) { contentAll += content; pinyinAll += pinyin; } } System.out.println("获取拼音首字母:" + contentAll + ":" + pinyinAll); // String address = "(金浜小区)栖山路1689弄"; // address = address.substring(address.indexOf(")") + 1); // System.out.println("获取拼音首字母:" + LetterUtil.getFirstLetter(address)); } @SuppressWarnings("unused") private static void findRepeatWord(Map<String, String> wordsMap) { String words = wordsMap.values().toString().replaceAll("[, ]", ""); words = words.substring(1, words.length() - 1); for (char word : words.toCharArray()) { int count = findStr2(words, String.valueOf(word)); if (count > 1) { System.out.println("汉字:【" + word + "】出现了" + count + "次!"); } } } private static int findStr2(String srcText, String keyword) { int count = 0; Pattern p = Pattern.compile(keyword); Matcher m = p.matcher(srcText); while (m.find()) { count++; } return count; } }
试试其它关键字
汉字拼音
同语言下
.
List 切割成几份 工具类
.
一行一行读取txt的内容
.
Java PDF转换成图片并输出给前台展示
.
java 多线程框架
.
double类型如果小数点后为零则显示整数否则保留两位小
.
将图片转换为Base64字符串公共类抽取
.
sqlParser 处理SQL(增删改查) 替换schema 用于多租户
.
JAVA 月份中的第几周处理 1-7属于第一周 依次类推 29-
.
java计算两个经纬度之间的距离
.
输入时间参数计算年龄
可能有用的
.
C#实现的html内容截取
.
List 切割成几份 工具类
.
SQL查询 多列合并成一行用逗号隔开
.
一行一行读取txt的内容
.
C#动态修改文件夹名称(FSO实现,不移动文件)
.
c# 移动文件或文件夹
.
c#图片添加水印
.
Java PDF转换成图片并输出给前台展示
.
网站后台修改图片尺寸代码
.
处理大图片在缩略图时的展示
善培
贡献的其它代码
(
14
)
.
删除文件夹,指定文件夹,删除文件夹和所有文件,删除
.
查询字符串内有多少个指定字符
.
Node.js 连接池
.
如何检查cookie是否启用
.
取汉字拼音首字母含多音字及不常见的字
.
二分图最大匹配(hungary邻接表形式,邻接阵接口)
.
重命名文件
.
计算权重
.
配置文件读取
.
微信信息servlet端的处理
Copyright © 2004 - 2024 dezai.cn. All Rights Reserved
站长博客
粤ICP备13059550号-3