A programmer's site

jetty处理 http response header中文字符，调试笔记

这段时间，我和几个同事在做另外一个美味书签，一个没有选辑，专注收藏的服务，和有选辑的美味书签并存。

大家讨论，觉得url应该是尽量清晰简单。比如张三用户登陆，它的url是/u/张三，之类的。能让高级用户直接输入url。url是可读的。

我们的程序用clojure开发。用ring，compojure, jetty。

我遇到了一个问题，在jetty返回给浏览器的http header中， 中文成了乱码。

场景如下：

1. 张三到达我们网站，他在美味书签的用户名也是张三，输入的网址是landing page的 url： http://meiweisq.com/
2. 程序发现，此用户已经登陆过，redirect到他的首页 /u/张三
3. 浏览器看见的却是 /u/??

在第2和第3步之间出问题了。jetty在encode

{
     :status 302
     :headers {"Location" "/u/张三"}
     :body nil
     }

时出现乱码。尝试各种方法无果。

无奈之下换上我手工写的一个ring adapter， 发现也出现乱码，赶快修改了encode header相关代码， 从ASCII改为UTF8，问题解决。

但这只是一个临时的解决方法。

下班回家后，还在琢磨这件事， download了jetty的代码：

git clone git://github.com/eclipse/jetty.project.git

checkout 我们用的相应版本7.6.1.v20120215。

经过一番查找，调试，发现了问题所在， jetty在encode http response header时，用了hardcode的 ISO-8859-1

这段代码是用来encode http header的:

// ./jetty-http/src/main/java/org/eclipse/jetty/http/HttpFields.java
// 322行到347行
  private Buffer convertValue(String value)
    {
        Buffer buffer = __cache.get(value);
        if (buffer!=null)
            return buffer;

        try
        {
            buffer = new ByteArrayBuffer(value,StringUtil.__ISO_8859_1);

            if (__cacheSize>0)
            {
                if (__cache.size()>__cacheSize)
                    __cache.clear();
                Buffer b=__cache.putIfAbsent(value,buffer);
                if (b!=null)
                    buffer=b;
            }

            return buffer;
        }
        catch (UnsupportedEncodingException e)
        {
            throw new RuntimeException(e);
        }
    }

StringUtil.__ISO_8859_1的定义

// ./jetty-util/src/main/java/org/eclipse/jetty/util/StringUtil.java

    public static final String __ISO_8859_1="ISO-8859-1";
    public final static String __UTF8="UTF-8";
    public final static String __UTF8Alt="UTF8";
    public final static String __UTF16="UTF-16";

解决办法

{
:status 302
:headers {"Location" (URLEncoder/encode "/u/张三")}
:body nil
}

后面想到可以用另外的方式绕过jetty编码http header的问题：

• 返回一个网页，网页里面用js

location.href="/u/张三"

• Meta refresh

 http-equiv="refresh" content="0;URL='/u/张三'">

Rssminer public beta

Rssminer is a simple online RSS reader written in Clojure, Java, Javascript.

Feature

• Not Feature rich, just a central place to reading RSS news.
• Concise UI.
• Read the orignal article, which may include valuable comments.
• Learing from reading history, rank all articles accordingly.
• Fast, Instant full text search.

A brief history

1. Initially, I randly pick a project to experiment one page javascript APP, client side unit testing. A RSS reader is choosen by accident.
2. Backbone is tried, I even read several times. , got read several times.
3. Function and Object literal are the best parts of Javascript. They complement each other. Backbone fail to utilize them efficiently.
4. I did many expriments and thinking. By dropping Backone.js, take my brain as the backone, structure the code the way I like it, I found a better world.
5. Writing javascript is fun. Writing Clojure is fun. Clojure mixed with JAVA is really great. They make me continue doing this project.
6. Rssminer is born.

Performance

I like fast application. In order to be fast,

• The application is carefully designed .
• A WEB server, written from scratch to power Rssminer, to achieve low lantency.
• A HTTP Client, written from scratch to power the fetcher module.
• A fast Template. With it, I can easily pack the minified CSS with the HTML to save a network round trip. It bootstrap the application.
• Browser talks to server in JSON. Render HTML in browser side achieves even better performance.
• Powered by Redis and MySQL

Demo

A demo account is created: http://rssminer.net/demo. Feel free to try it. Feedbacks are welcome.

Open Source

https://github.com/shenfeng/rssminer

About Javascript dispatch event and delegate event

(function () {
  var all_events = {};          // events data, private to this file
  function bind_event (ele, event, handler) {
    if(ele.addEventListener) {
      ele.addEventListener(event, handler, false);
    } else {                    // ie
      ele.attachEvent('on' + event, handler);
    }
  }
  // a Simple impl of dispatch event.
  function dispatch_event (ele, selector, eventName, handler) {
    var handlers = all_events[ele] || {};
    if(handlers[eventName]) {
      handlers[eventName].push(handler); // add more
    } else {
      handlers[eventName] = [handler]; // register
      var one_handler = function (e) {
        // find registerd handlers
        var handlers = (all_events[ele] || {})[eventName];
        if(!handlers) { return; }
        var target = e.target;
        // IE6, IE7 does not support querySelectorAll
        var nodes = ele.querySelectorAll(selector);
        for(var i = 0; i < nodes.length; i++) {
          // interested event hanpend on interested DOM node
          if(nodes[i] === target) {
            for(var j = 0; j < handlers.length; j++) { // call one by one
              // properly bind `this`
              var ret = handlers[j].apply(target, [e]);
              if(ret === false) {
                e.preventDefault(); e.stopPropagation();
              }
            }
          }
        }
      };
      bind_event(ele, eventName, one_handler);
    }
    all_events[ele] = handlers;
  }
  var EVENTSPLITTER = /^(\S+)\s*(.*)$/;
  // extracted from Backbone.js, brilliant
  function delegate_events (ele, events) {
    for (var key in events) {
      var method = events[key],
          match = key.match(EVENTSPLITTER),
          eventName = match[1],
          selector = match[2];
      if (selector === '') {
        bind_event(ele, eventName, method);
      } else {
        dispatch_event(ele, selector, eventName, method);
      }
    }
  }
  function click_handler1 () { console.log('click', this, e); }
  function click_handler2 () { console.log('click2', this, e); }
  // I like the syntax.
  // And it's very flexable, modify the DOM as you want,
  // event handler will work just the way you want it
  delegate_events(document, {
    'click li': click_handler1,
    'click ul li': click_handler2
  });
})();

jvm options

java -XX:+PrintFlagsFinal

2012要看的书， 要写的代码

管2012会不会有不有世界末日， 他并不妨碍我读几本书。

作为一个程序员，行业要求要无止境的学习，并且我还很喜欢这个行业

• 《深入理解Java虚拟机:JVM高级特性与最佳实践》 不管怎样， 我很喜欢java和JVM上的Clojure， 有必要深入了解一下。
• 《Computer Systems, A programmer’s perspective》 去年就开始翻了， 希望今年能看完。

• 《The c programming language》 大一就读了这本书， 现在有必要重读一下

• 《a global history from prehistory to the 21st century》 袁某人推荐的。去年就买了，可是没有看。 很是喜欢袁老师

想要写的代码

• 用C写一个Web Server, 用epoll，自己写内存分配，妥善处理超时，和各种意 外情况。 看看能到多快。

• 用纯JAVA， 从头写一个web server 和http client，给Rssminer用。 能给 production用。 妥善处理超时， 各种意外，控制latency， 控制内存使用。
• 写完Rssminer。 这个已经写了快一年了。 今年6月份以前应该publicly available