apache的StringUtils方法与原生的string方法性能的巨大差别

apache的replace,trim方法 StringUtils.replace(),StringUtils.trimWhitespace()
java原生的replace,trim方法之间性能的巨大差别。

public class App 
{
	final static String teString = "    555555599999999999999999aaa sss ssaaaaaaaaaaaaaatasdtatastasrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrraaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa      "; 
    public static void main( String[] args )
    {
    	int t = 100000;
    	 long start1 = System.nanoTime();
    	 long stringUtilsReplaceTime = 0;
    	 long stringUtilsTrimTime = 0;
    	 long stringReplaceTime = 0;
    	 long stringTrimTime = 0;
         for(int i= 0;i<t;i++){
      	   StringUtils.replace(teString,"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa","b");
         }
         stringUtilsReplaceTime= System.nanoTime()-start1;
         
         
          start1 = System.nanoTime();
         for(int i= 0;i<t;i++){
      	   teString.replace("aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", "b");
         }
         
         stringReplaceTime = System.nanoTime()-start1;

        start1 = System.nanoTime();
       for(int i= 0;i<t;i++){
    	   teString.trim();
       }
       stringTrimTime = System.nanoTime()-start1;
       
       
        start1 = System.nanoTime();
       for(int i= 0;i<t;i++){
    	  
    	   StringUtils.trimWhitespace(teString);
       }
       
       stringUtilsTrimTime = System.nanoTime()-start1;
       
       System.out.println("stringUtilsReplaceTime:"+stringUtilsReplaceTime);
       System.out.println("stringReplaceTime     :"+stringReplaceTime);
       System.out.println("stringUtilsTrimTime   :"+stringUtilsTrimTime);
       System.out.println("stringTrimTime        :"+stringTrimTime);
    }
 
}

输出结果为：
stringUtilsReplaceTime:293389595
stringReplaceTime     :1019421816
stringUtilsTrimTime   :176730378
stringTrimTime        :3888271

public class App 
{
	final static String teString = "    555555599999999999999999aaa sss ssaaaaaaaaaaaaaatasdtatastasrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrraaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa      "; 
    public static void main( String[] args )
    {
    	int t = 100;
    	 long start1 = System.nanoTime();
    	 long stringUtilsReplaceTime = 0;
    	 long stringUtilsTrimTime = 0;
    	 long stringReplaceTime = 0;
    	 long stringTrimTime = 0;
         for(int i= 0;i<t;i++){
      	   StringUtils.replace(teString,"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa","b");
         }
         stringUtilsReplaceTime= System.nanoTime()-start1;
         
         
          start1 = System.nanoTime();
         for(int i= 0;i<t;i++){
      	   teString.replace("aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", "b");
         }
         
         stringReplaceTime = System.nanoTime()-start1;

        start1 = System.nanoTime();
       for(int i= 0;i<t;i++){
    	   teString.trim();
       }
       stringTrimTime = System.nanoTime()-start1;
       
       
        start1 = System.nanoTime();
       for(int i= 0;i<t;i++){
    	  
    	   StringUtils.trimWhitespace(teString);
       }
       
       stringUtilsTrimTime = System.nanoTime()-start1;
       
       System.out.println("stringUtilsReplaceTime:"+stringUtilsReplaceTime);
       System.out.println("stringReplaceTime     :"+stringReplaceTime);
       System.out.println("stringUtilsTrimTime   :"+stringUtilsTrimTime);
       System.out.println("stringTrimTime        :"+stringTrimTime);
    }
 
}

结果为：
stringUtilsReplaceTime:94864540
stringReplaceTime     :8341259
stringUtilsTrimTime   :953821
stringTrimTime        :57856

の，兄弟，这个测试有点不准确啊
1、有没有试过测试方法中将StringUtils.replace()和replace()的测试顺序对调下，输出下结果？
2、有没有考虑jvm刚启动时需要预热的情况？

4-5倍的性能差异应该是在预料之中的

靠，一个10000次，一个1000次，能一样吗？

+1

已经修改，悲剧。

这说明在数据执行次数比较多的超过10000次，java原生的replace性能会急剧下降，而apache的replace才能显现出优势来，呵呵。

对同一个字符串执行10000次相同的替换。。。

一个支持正则，一个不支持，性能上当然有差异。

我自己试了一下，如果数据执行次数比较少，少于200次，java原生的replace那个执行时间反而少，而apache的replace的执行时间却比较多一些，呵呵。

到底实际应用中，哪一种比较常见呢？俺想不到什么时候需要对一个字符串操作很多次。是不是改成对多个不同的字符串操作来测试才会比较有意义呢？