评:
Apache HttpClient ( http://jakarta.apache.org/commons/httpclient/ ) 是一个纯 Java 的HTTP 协议的客户端编程工具包, 对 HTTP 协议的支持相当全面, 更多细节也可以参考IBM 网站上的这篇文章 HttpClient入门 ( http://www-128.ibm.com/developerworks/cn/opensource/os-httpclient/ ).

问题分析
不过在实际使用中, 还是发现按照最基本的方式调用 HttpClient 时, 并不支持 UTF-8 编码, 在网络上找过一些文章, 也不得要领, 于是查看了 commons-httpclient-3.0.1 的一些代码, 首先在 PostMethod 中找到了 generateRequestEntity() 方法:

/** 

 * Generates a request entity from the post parameters, if present. Calls 

 * {@link EntityEnclosingMethod#generateRequestBody()} if parameters have not been set. 

 * 

 * @since 3.0 

 */ 

 protected RequestEntity generateRequestEntity() { 

 if (!this.params.isEmpty()) { 

 // Use a ByteArrayRequestEntity instead of a StringRequestEntity. 

 // This is to avoid potential encoding issues. Form url encoded strings 

 // are ASCII by definition but the content type may not be. Treating the content 

 // as bytes allows us to keep the current charset without worrying about how 

 // this charset will effect the encoding of the form url encoded string. 

 String content = EncodingUtil.formUrlEncode(getParameters(), getRequestCharSet()); 

 ByteArrayRequestEntity entity = new ByteArrayRequestEntity( 

 EncodingUtil.getAsciiBytes(content), 

 FORM_URL_ENCODED_CONTENT_TYPE 

 ); 

 return entity; 

 } else { 

 return super.generateRequestEntity(); 

 } 

 }


原来使用 NameValuePair 加入的 HTTP 请求的参数最终都会转化为 RequestEntity 提交到 HTTP 服务器, 接着在 PostMethod 的父类 EntityEnclosingMethod 中找到了如下的代码:

/** 

 * Returns the request's charset. The charset is parsed from the request entity's 

 * content type, unless the content type header has been set manually. 

 * 

 * @see RequestEntity#getContentType() 

 * 

 * @since 3.0 

 */ 

 public String getRequestCharSet() { 

 if (getRequestHeader("Content-Type") == null) { 

 // check the content type from request entity 

 // We can't call getRequestEntity() since it will probably call 

 // this method. 

 if (this.requestEntity != null) { 

 return getContentCharSet( 

 new Header("Content-Type", requestEntity.getContentType())); 

 } else { 

 return super.getRequestCharSet(); 

 } 

 } else { 

 return super.getRequestCharSet(); 

 } 

 }




解决方案
从上面两段代码可以看出是 HttpClient 是如何依据 "Content-Type" 获得请求的编码(字符集), 而这个编码又是如何应用到提交内容的编码过程中去的. 按照这个原来, 其实我们只需要重载 getRequestCharSet() 方法, 返回我们需要的编码(字符集)名称, 就可以解决 UTF-8 或者其它非默认编码提交 POST 请求时的乱码问题了.

测试
首先在 Tomcat 的 ROOT WebApp 下部署一个页面 test.jsp, 作为测试页面, 主要代码片段如下:

<%@ page contentType="text/html;charset=UTF-8"%> 

<%@ page session="false" %> 

<% 

request.setCharacterEncoding("UTF-8"); 

String val = request.getParameter("TEXT"); 

System.out.println(">>>> The result is " + val); 

%>




接着写一个测试类, 主要代码如下:

public static void main(String[] args) throws Exception, IOException { 

 String url = "http://localhost:8080/test.jsp"; 

 PostMethod postMethod = new UTF8PostMethod(url); 

 //填入各个表单域的值 

 NameValuePair[] data = { 

 new NameValuePair("TEXT", "中文"), 

 }; 

 //将表单的值放入postMethod中 

 postMethod.setRequestBody(data); 

 //执行postMethod 

 HttpClient httpClient = new HttpClient(); 

 httpClient.executeMethod(postMethod); 

 } 


 //Inner class for UTF-8 support 

 public static class UTF8PostMethod extends PostMethod{ 

 public UTF8PostMethod(String url){ 

 super(url); 

 } 

 @Override 

 public String getRequestCharSet() { 

 //return super.getRequestCharSet(); 

 return "UTF-8"; 

 } 

 }




运行这个测试程序, 在 Tomcat 的后台输出中可以正确打印出 ">>>> The result is 中文" .

代码下载
本文所提到的所有代码, 以及测试程序(可直接导入 eclipse)提供打包下载: att:HttpClient POST 的 UTF-8 编码问题.httpClientUTF8.tar.bz2

END