- package taotao.montao.demo.util;
- /**
- * 獲取文章的字數或則字符數
- * @author montao
- */
- public class StatWordCount {
- private final char[] CHS = {',',';','!','.','!','?',';','+',',','?','!','/'}; //符號數組
- private final char[] CHN = {'\r','\n'}; //轉義符數組
- private final char[] SPACE = {' ',' '}; //空格的數組(前半角,後全角)
- /**
- * 根據指定條件來篩選文章的字數
- * @param wordContent 文章內容
- * @param compriseInterpunction 是否包含指定字符
- * @param compriseSpace 是否包含空格
- * @return 返回文章經過指定篩選後的長度
- */
- public int getWordCount(String wordContent,boolean compriseInterpunction,boolean compriseSpace)
- {
- if(wordContent==null){
- return 0;
- }else if(wordContent.length()==0){
- return 0;
- }else{
- //既要包含符號又要包含空格
- if(compriseInterpunction && compriseSpace){
- //清除轉義符
- String regex = "["+new String(CHN)+"]";
- wordContent = wordContent.replaceAll(regex," ");
- return this.getWordCount(wordContent);
- }
- //不包含符號包含空格
- else if(!compriseInterpunction && compriseSpace){
- //使用正則表達式去掉指定的符號和轉義符
- String regex1 = "["+new String(CHN)+"]";
- String regex2 = "["+new String(CHS)+"]";
- wordContent = wordContent.replaceAll(regex1," ");
- wordContent = wordContent.replaceAll(regex2," ");
- return this.getWordCount(wordContent);
- }
- //包含指定符號不包含空格
- else if(compriseInterpunction && !compriseSpace){
- //使用正則表達式去掉空格和轉義符
- String regex1 = "["+new String(CHN)+"]";
- String regex2 = "["+new String(SPACE)+"]";
- wordContent = wordContent.replaceAll(regex1," ");
- wordContent = wordContent.replaceAll(regex2," ");
- return this.getWordCount(wordContent);
- }
- //空格和指定符號都不包含
- else{
- //使用正則表達式去掉空格,指定符號和轉義符
- 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);
- }
- }
- }
- /**
- * 返回文章中的字數
- * @param wordCount 文章內容
- * @return
- */
- @SuppressWarnings("unused")
- private int getWordCount(String wordContent){
- int count = 0;
- if(wordContent==null){ //判斷是否爲null,如果爲null直接返回0
- count = 0;
- }else if(wordContent.length()==0){ //判斷是否爲空,如果爲空直接返回0
- count = 0;
- }else{ //判斷獲取字數
- wordContent = wordContent.trim(); //清空空格
- //臨時變量
- String s4 = "";
- String s3 = "";
- String s2 = "";
- 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()){
- s1 = String.valueOf(wordContent.charAt(i+1));
- if((s1.hashCode()==32 && (s3.hashCode()!=32)) || ((s1.getBytes().length==2)&& (s3.hashCode()!=32))){
- count++;
- }
- }
- }
- }
- if(!bb){
- count++;
- }else{
- if(s4.getBytes().length==1){
- count++;
- }
- }
- }
- return count;
- }
- /**
- * 根據條件來獲取文章的字符數
- * @param wordContent 文章內容
- * @param compriseInterpunction 是否包含指定符號
- * @param compriseSpace 是否包含空格
- * @return 返回字符長度
- */
- public int getWordCharacter(String wordContent,boolean compriseInterpunction,boolean compriseSpace)
- {
- //既要包含符號又要包含空格
- if(compriseInterpunction && compriseSpace){
- //清除轉義符
- String regex = "["+new String(CHN)+"]";
- wordContent = wordContent.replaceAll(regex,"");
- //首部的空格不算
- wordContent = wordContent.replaceAll("^\\s+","");
- return wordContent.length();
- }//不包含符號包含空格
- else if(!compriseInterpunction && compriseSpace){
- //首部的空格不算
- wordContent = wordContent.replaceAll("^\\s+","");
- //使用正則表達式去掉指定的符號和轉義符
- String regex1 = "["+new String(CHN)+"]";
- String regex2 = "["+new String(CHS)+"]";
- wordContent = wordContent.replaceAll(regex1,"");
- wordContent = wordContent.replaceAll(regex2,"");
- return wordContent.length();
- }//包含指定符號不包含空格
- else if(compriseInterpunction && !compriseSpace){
- //使用正則表達式去掉空格和轉義符
- String regex = "["+new String(SPACE)+"]";
- wordContent = wordContent.replaceAll(regex," ");
- return this.getNoSpaceCount(wordContent);
- }//空格和指定符號都不包含
- else{
- //使用正則表達式去掉指定符號
- String regex1 = "["+new String(CHS)+"]";
- String regex2 = "["+new String(SPACE)+"]";
- wordContent = wordContent.replaceAll(regex1," ");
- wordContent = wordContent.replaceAll(regex2," ");
- return this.getNoSpaceCount(wordContent);
- }
- }
- /**
- * 獲取文章中非空格的字符總數
- * @param wordContent 文章內容
- * @return
- */
- private int getNoSpaceCount(String wordContent) {
- int spaceCount = 0;
- if(wordContent==null)
- {
- spaceCount = 0;
- }else if(wordContent.length()==0)
- {
- spaceCount = 0;
- }else
- {
- //替換首部的
- wordContent = wordContent.replaceAll("^\\s+","");
- wordContent = wordContent.replaceAll(" ","");
- //使用正則替換轉義符
- String regex = "["+new String(CHN)+"]";
- wordContent = wordContent.replaceAll(regex,"");
- spaceCount = wordContent.length();
- }
- return spaceCount;
- }
- }