htmlsubstring.java

一套完整的工商12315的源程序jsp部分在12315里,后台JAVA部分在gs12315src里,没有打包数据库.

/**
 * <p>Title: </p>
 * <p>Description: This class is written to intercept named length of HTML String</p>
 * <p>Copyright: Copyright (c) 2002</p>
 * <p>Company: 中关村软件</p>
 * @author 唐旭成
 * @version 1.0
 */
package com.gs.util;
import java.lang.*;
import java.io.*;

public class htmlSubstring
{
   public htmlSubstring()
   {
   }

   public String htmlSubstr(String htmlStr,int len)
	{
		
		String bakStr   = new String(htmlStr+" ");
		String finalStr = new String("");
		String tagStr   = new String("");
		String tmpStr   = new String("");
		char chBegin    = '<';
		char chEnd      = '>';
		char chSpace    = ' ';
		int  pos1       = -1;
		int  pos2       = -1;
		int  pos3       = -1;
		int  i          = 0;
		int  count      = 0;
		int remainder   = 0;
		String[] tagArray = new String[10];

		int num = bakStr.length();
		if(len >= num)
		{
			return bakStr; // 返回全部字符
		}

		pos1 = bakStr.indexOf(chBegin);
		if(pos1 >= len)
		{
			finalStr += bakStr.substring(0,len);
			return finalStr;
		}

		if(pos1 > -1)
		{
			finalStr += bakStr.substring(0,pos1);
			i += pos1;
		}

		while(true)
		{
			if(pos1 == -1 || pos1 == num -1)
			{
				remainder = len - i;
				if(pos2 > 0 && pos3 > 0)
				{
					finalStr += bakStr.substring(pos2,pos3);
				}
				tmpStr = bakStr.substring(pos2+1,pos2+1+remainder);
				finalStr += tmpStr;
				break;
			}

			remainder = len - i;
			if(remainder <= 0)
			{
				break;
			}

			pos2 = bakStr.indexOf(chEnd,pos1+1);
			if(pos2 == -1)
			{
				tmpStr = bakStr.substring(pos2+1,pos3);
				finalStr += tmpStr.substring(0,remainder);
				break;
			}

			pos3 = bakStr.indexOf(chSpace,pos1+1);
			if(pos3 > pos2 || pos3 == -1) /* "<"和">"之间没有空格，如：<TD> */
			{
				tagStr = bakStr.substring(pos1+1,pos2);
 			   tagStr = "<" + tagStr + ">";
			}
			else /* "<"和">"之间有空格，如：<TD ALIGN=LEFT> */
			{
				tagStr = bakStr.substring(pos1+1,pos3);
 			   tagStr = "<" + tagStr;
			}

			tagArray[count] = tagStr.toUpperCase();
			// System.out.println("The Tag is <" + tagStr + ">");

			pos3 = bakStr.indexOf(chBegin,pos2+1);
			if(pos3 == -1)
			{
			   tmpStr = bakStr.substring(pos2+1);
			   if(remainder > num - pos2 - 1)
			   {
			   	remainder = num - pos2 - 1;
			   }
			   finalStr += bakStr.substring(pos1,pos2+1) + tmpStr.substring(0,remainder);
			   break;
			}

			tmpStr = bakStr.substring(pos2+1,pos3);
			if(tmpStr.length() >= remainder)
			{
				finalStr += bakStr.substring(pos1,pos2+1) + tmpStr.substring(0,remainder);
				break;
			}
			else
			{
				i += tmpStr.length();
				finalStr += bakStr.substring(pos1,pos2+1) + tmpStr;
			}

			// System.out.println("finalStr is " + finalStr);
			pos1 = pos3;
			count++;

		}

		// System.out.println("finalStr is " + finalStr);

		// 以下程序自动追加结束Tag

		int num1 = 0;
		int num2 = 0;
		int j    = 0;
		String tagStr2 = new String();

		num = tagArray.length -1;

		for(i=num;i>=0;i--)
		{
			num1 = 0;
			num2 = 0;
			pos1 = 0;
			tagStr = tagArray[i];
			if(tagStr == null)
			{
			   continue;
			}
			if(tagStr.equalsIgnoreCase("<IMG") || tagStr.equalsIgnoreCase("<HR>") || tagStr.equalsIgnoreCase("<BR>")  )
			{
			   continue;  // 跳过本来就没有结束Tag的Tag
			}
			if(tagStr.charAt(1) == '/')
			{
			   continue; // 跳过结束Tag
			}

			tagStr2 = "</" + tagStr.substring(1,tagStr.length()) ; // 结束Tag
			if(tagStr.charAt(tagStr.length() -1) != '>')
			   tagStr2 += ">";

		   pos1 = finalStr.indexOf(tagStr,0);
			while(pos1 > -1)
			{
			   pos1 = finalStr.indexOf(tagStr,pos1);
				if(pos1 == -1)
				   break;
				pos1++;
				num1++;
			}

		   pos1 = finalStr.indexOf(tagStr2,0);
			while(pos1 >= 0)
			{
			   pos1 = finalStr.indexOf(tagStr2,pos1);
				if(pos1 == -1)
				   break;
				pos1++;
				num2++;
			}

			for(j= 1;j<=num1 - num2;j++)
			{
				finalStr += tagStr2; // 追加结束Tag
			}
		}

		return finalStr;

   }

   public static void main(String[] args)
   {
		int li_len = 0;
		String htmlStr = new String("");

		if ( args.length == 2 )
		{
			htmlStr = args[0].trim();
			li_len  = Integer.parseInt(args[1]);
		}
		else
		{
			System.out.println("Syntax is: java htmlSubstring [htmlString] [length]");
			System.out.println("                String htmlString --- HTML String to intercept");
			System.out.println("                int    length     --- The length of interception");
			System.out.println("Remark:    A chinese character's length is 1, not 2!!!\n");
		}

		if(htmlStr.equals(null) || htmlStr.equalsIgnoreCase(""))
		{
		   htmlStr = "编辑会议<a href=““><B>ss<font color=red size=+9t>辑会<font color=blue>议比较</font>ss</a>";
			li_len = 20;
		}

		htmlSubstring mySubstr = new htmlSubstring();
      String resultStr = mySubstr.htmlSubstr(htmlStr,li_len);
		System.out.println("Original String is: " + htmlStr);
		System.out.println("First " + li_len + " characters are: " + resultStr);
   }
}