日期:2014-05-20  浏览次数:20824 次

求字数统计代码
RT:类似下面这个网站上的工具代码,用Java该怎么写啊,求高手。
http://www.mb5u.com/tool/zishutongji/

------解决方案--------------------
JScript code

function CountChineseCharacters(Words) {
    var W = new Object();
    var Result = new Array();
    var iNumwords = 0;
    var sNumwords = 0;
    var sTotal = 0;//??????????
    var iTotal = 0;//???????訒?
    var eTotal = 0;//????????
    var otherTotal = 0;
    var bTotal = 0;

    for (i=0; i<Words.length; i++) {
        var c = Words.charAt(i);
        if (c.match(/[\u4e00-\u9fa5]/)) {
            if (isNaN(W[c])) {
                iNumwords++;
                W[c] = 1;
            }
            iTotal++;
        }

    }
    for (i=0; i<Words.length; i++) {
        var c = Words.charAt(i);
        if (c.match(/[^\x00-\xff]/)) {
            if (isNaN(W[c])) {
                sNumwords++;

            }
            sTotal++;
        }
        else
        {
        eTotal++;
        }
//        else if(c.match(/[\w\d]/)) {
//        eTotal++;}
//        else{
//        bTotal++;
//        }
    }
//    Result["numwords"] = iNumwords;
//    Result["total"] = iTotal;
//    avg = parseInt(iTotal/iNumwords*1000)/1000;
//    return Result;
document.getElementById("iTotal").innerHTML=iTotal;
document.getElementById("eTotal").innerHTML=eTotal;
document.getElementById("sTotal").innerHTML=sTotal-iTotal;
document.getElementById("numwords").innerHTML=iTotal*2+(sTotal-iTotal)*2+eTotal;
//document.getElementById("avg").innerHTML=parseInt(iTotal/iNumwords*1000)/1000;
}
//var res = CountChineseCharacters(document.getElementById("id_entrytext").innerHTML);
//document.getElementById("numwords").innerHTML=res["numwords"];
//document.getElementById("avg").innerHTML=res["avg"];

------解决方案--------------------
给你转个

Java code

package taotao.montao.demo;
/**
* 获取文章的字数或则字符数
* @author montao
*/
public class StatWordCount {

   private final char[] CHS = {',',';','.','!','?',';','+','。','?','!'};  //符号数组

   private final char[] CHN = {'\n','\r'}; //转义符数组

   private final char[] SPACE = {' ',' '}; //空格的数组(前半角,后全角)

   /**
    * 根据指定条件来筛选文章的字数
    * @param wordContent  文章内容
    * @param compriseInterpunction  是否包含指定字符
    * @param compriseSpace  是否包含空格
    * @return 返回文章经过指定筛选后的长度
    */
   public int getWordCount(String wordContent,boolean compriseInterpunction,boolean compriseSpace)
   {
       if(wordContent==null){
           return 0;
       }else if(wordContent.length()==0){
           return 0;
       }else{
           //既要包含符号又要包含空格
           if(compriseInterpunction && compriseSpace){
               //清除转义符
               String regex = "["+new String(CHN)+"]";
               wordContent = wordContent.replaceAll(regex," ");
               return this.getWordCount(wordContent);
           }
           //不包含符号包含空格
           else if(!compriseInterpunction && compriseSpace){
               //使用正则表达式去掉指定的符号和转义符
               String regex1 = "["+new String(CHN)+"]";
               String regex2 = "["+new String(CHS)+"]";
               wordContent = wordContent.replaceAll(regex1," ");
               wordContent = wordContent.replaceAll(regex2," ");
               return this.getWordCount(wordContent);
           }
           //包含指定符号不包含空格
           else if(compriseInterpunction && !compriseSpace){
               //使用正则表达式去掉空格和转义符
               String regex1 = "["+new String(CHN)+"]";
               String regex2 = "["+new String(SPACE)+"]";
               wordContent = wordContent.replaceAll(regex1," ");
               wordContent = wordContent.replaceAll(regex2," ");
               return this.getWordCount(wordContent);
           }
           //空格和指定符号都不包含
           else{
               //使用正则表达式去掉空格,指定符号和转义符
               String regex1 = "["+new String(CHN)+"]";
               String regex3 = "["+new String(CHS)+"]";
               String regex2 = "["+new String(SPACE)+"]";
               wordContent = wordContent.replaceAll(regex1," ");
               wordContent = wordContent.replaceAll(regex2," ");
               wordContent = wordContent.replaceAll(regex3," ");
               return this.getWordCount(wordContent);
           }
       }
   }

   /**
    * 返回文章中的字数
    * @param wordCount 文章内容
    * @return
    */
   @SuppressWarnings("unused")
   private int getWordCount(String wordContent){
       int count = 0;
       if(wordContent==null){ //判断是否为null,如果为null直接返回0
           count = 0;
       }else if(wordContent.length()==0){ //判断是否为空,如果为空直接返回0
           count = 0;
       }else{ //判断获取字数
           wordContent = wordContent.trim(); //清空空格
           //临时变量
           String s4 = "";
           String s3 = "";
           String s1 = "";
           boolean bb = false;
           if(wordContent.length()>0){
               s4 = String.valueOf(wordContent.charAt(wordContent.length()-1));
           }
           for(int i=0;i<wordContent.length();i++){
               s3 = String.valueOf(wordContent.charAt(i));
               int num = s3.getBytes().length;
               if(s3.hashCode()==32||s3.getBytes().length==2){
                   bb=true;
               }if(num==2){
                   count++;
               }else{
                       if(i+1<wordContent.length()&&(i>1)){
                           s1 = String.valueOf(wordContent.charAt(i+1));
                           if((s1.hashCode()==32||s1.getBytes().length==2)&&(s3.hashCode()!=32)){
                               count++;
                           }
                       }
               }
           }
           if(!bb){
               count++;
           }else{
               if(s4.getBytes().length==1){
                   count++;
               }
           }
       }
       return count;
   }

   /**
    * 根据条件来获取文章的字符数
    * @param wordContent 文章内容
    * @param compriseInterpunction 是否包含指定符号
    * @param compriseSpace 是否包含空格
    * @return 返回字符长度
    */
   public int getWordCharacter(String wordContent,boolean compriseInterpunction,boolean compriseSpace)
   {
       //既要包含符号又要包含空格
       if(compriseInterpunction && compriseSpace){
           //清除转义符
           String regex = "["+new String(CHN)+"]";
           wordContent = wordContent.replaceAll(regex," ");
           //首部的空格不算
           wordContent = wordContent.replaceAll("^\\s+","");
           return wordContent.length();
       }//不包含符号包含空格
       else if(!compriseInterpunction && compriseSpace){
           //首部的空格不算
           wordContent = wordContent.replaceAll("^\\s+","");
           //使用正则表达式去掉指定的符号和转义符
           String regex1 = "["+new String(CHN)+"]";
           String regex2 = "["+new String(CHS)+"]";
           wordContent = wordContent.replaceAll(regex1," ");
           wordContent = wordContent.replaceAll(regex2," ");
           return wordContent.length();
       }//包含指定符号不包含空格
       else if(compriseInterpunction && !compriseSpace){
           //使用正则表达式去掉空格和转义符
           return this.getNoSpaceCount(wordContent);
       }//空格和指定符号都不包含
       else{
           //使用正则表达式去掉指定符号
           String regex1 = "["+new String(CHS)+"]";
           wordContent = wordContent.replaceAll(regex1," ");
           return this.getNoSpaceCount(wordContent);
       }
   }
   /**
    * 获取文章中非空格的字符总数
    * @param wordContent 文章内容
    * @return
    */
   private int getNoSpaceCount(String wordContent) {
       int spaceCount = 0;
       if(wordContent==null)
       {
           spaceCount = 0;
       }else if(wordContent.length()==0)
       {
           spaceCount = 0;
       }else
       {
           //替换首部的
           wordContent = wordContent.replaceAll("^\\s+","");
           wordContent = wordContent.replaceAll(" ","");
           //使用正则替换转义符
           String regex = "["+new String(CHN)+"]";
           wordContent = wordContent.replaceAll(regex,"");
           spaceCount = wordContent.length();
       }
       return spaceCount;
   }
}