浅谈Android网络请求的前世今生-元基础HttpConnection

963 阅读8分钟
原文链接: www.jianshu.com

之前写过一篇关于网络请求相关的文章,主要关于一些网络基础.这篇则重点讲一讲Android下httpUrlConnect的内容。

相关链接

浅谈Android网络通信的前世今生--网络基础

Android下的Http client

Android提供了三种Http client:

  1. HttpURLConnection
  2. Apache HttpClient
  3. okHttp

HttpURLConnection

HttpUrlConnection是JDK里提供的联网API,我们知道Android SDK是基于Java的,所以当然优先考虑HttpUrlConnection这种最原始最基本的API,其实大多数开源的联网框架基本上也是基于JDK的HttpUrlConnection进行的封装罢了

HttpClient(不建议使用)

HttpClient是开源组织Apache提供的Java请求网络框架,其最早是为了方便Java服务器开发而诞生的,是对JDK中的HttpUrlConnection各API进行了封装和简化,提高了性能并且降低了调用API的繁琐,不过官方已经不建议使用。

okhttp

OKHttp是现在主流应用使用的网络请求方式, 用来交换数据和内容, 有效的使用OKHttp可以使你的APP变的更快和减少流量的使用。

  • 支持SPDY,可以合并多个到同一个主机的请求
  • 使用连接池技术减少请求的延迟(如果SPDY是可用的话)
  • 使用GZIP压缩减少传输的数据量
  • 缓存响应避免重复的网络请求

当你的网络出现拥挤的时候,就是OKHttp大显身手的时候,它可以避免常见的网络问题,如果你的服务是部署在不同的IP上面的,如果第一个连接失败,OkHTtp会尝试其他的连接。这对现在IPv4+IPv6中常见的把服务冗余部署在不同的数据中心上也是很有必要的。OkHttp将使用现在TLS特性(SNI ALPN)来初始化新的连接,如果握手失败,将切换到TLS 1.0

目前使用率最高的当属okHttp,不过我们这篇文章还是要先讲讲Android下网络请求的先辈 HttpUrlConnect,本文基于Android API26

简单使用

URL url=new URL("www.baidu.com");
HttpURLConnection conn= (HttpURLConnection) url.openConnection();
InputStream in = new BufferedInputStream(conn.getInputStream());

其中关键的两个过程
1.openConnection() 建立tcp连接
2.getInputSteam() 发送http请求,并获取返回流

openConnection()

URL.java

public URLConnection openConnection() throws java.io.IOException {
    return handler.openConnection(this);
}

其中 handler是URLStreamHandler的实例,handler的创建是在URL的构造函数中,其中调用了getURLStreamHandler()方法、

static URLStreamHandler getURLStreamHandler(String protocol) {

    URLStreamHandler handler = handlers.get(protocol);
    if (handler == null) {

        boolean checkedWithFactory = false;

        // Use the factory (if any)
        if (factory != null) {
            handler = factory.createURLStreamHandler(protocol);
            checkedWithFactory = true;
        }

        // Try java protocol handler
        if (handler == null) {
            final String packagePrefixList = System.getProperty(protocolPathProp,"");
            StringTokenizer packagePrefixIter = new StringTokenizer(packagePrefixList, "|");

            while (handler == null &&
                   packagePrefixIter.hasMoreTokens()) {

                String packagePrefix = packagePrefixIter.nextToken().trim();
                try {
                    String clsName = packagePrefix + "." + protocol +
                      ".Handler";
                    Class<?> cls = null;
                    try {
                        ClassLoader cl = ClassLoader.getSystemClassLoader();
                        cls = Class.forName(clsName, true, cl);
                    } catch (ClassNotFoundException e) {
                        ClassLoader contextLoader = Thread.currentThread().getContextClassLoader();
                        if (contextLoader != null) {
                            cls = Class.forName(clsName, true, contextLoader);
                        }
                    }
                    if (cls != null) {
                        handler  =
                          (URLStreamHandler)cls.newInstance();
                    }
                } catch (ReflectiveOperationException ignored) {
                }
            }
        }

        // Fallback to built-in stream handler.
        // Makes okhttp the default http/https handler
        if (handler == null) {
            try {
                // BEGIN Android-changed
                // Use of okhttp for http and https
                // Removed unnecessary use of reflection for sun classes
                if (protocol.equals("file")) {
                    handler = new sun.net.www.protocol.file.Handler();
                } else if (protocol.equals("ftp")) {
                    handler = new sun.net.www.protocol.ftp.Handler();
                } else if (protocol.equals("jar")) {
                    handler = new sun.net.www.protocol.jar.Handler();
                } else if (protocol.equals("http")) {
                    handler = (URLStreamHandler)Class.
                        forName("com.android.okhttp.HttpHandler").newInstance();
                } else if (protocol.equals("https")) {
                    handler = (URLStreamHandler)Class.
                        forName("com.android.okhttp.HttpsHandler").newInstance();
                }
                // END Android-changed
            } catch (Exception e) {
                throw new AssertionError(e);
            }
        }

        synchronized (streamHandlerLock) {

            URLStreamHandler handler2 = null;

            // Check again with hashtable just in case another
            // thread created a handler since we last checked
            handler2 = handlers.get(protocol);

            if (handler2 != null) {
                return handler2;
            }

            // Check with factory if another thread set a
            // factory since our last check
            if (!checkedWithFactory && factory != null) {
                handler2 = factory.createURLStreamHandler(protocol);
            }

            if (handler2 != null) {
                // The handler from the factory must be given more
                // importance. Discard the default handler that
                // this thread created.
                handler = handler2;
            }

            // Insert this handler into the hashtable
            if (handler != null) {
                handlers.put(protocol, handler);
            }

        }
    }

    return handler;

}
  1. 从handlers中取

  2. 如果URLStreamHandlerFactory不为空.让URLStreamHandlerFactory生成

  3. 根据协议 protocol生成,注意老版本中http和https用的是

    streamHandler = new HttpHandler(); streamHandler = new HttpsHandler();

后来的版本用的是,底层换成了okhttp的类

handler = (URLStreamHandler)Class.forName("com.android.okhttp.HttpHandler").newInstance();

handler = (URLStreamHandler)Class.forName("com.android.okhttp.HttpsHandler").newInstance();

最后还有个并发检查,避免因为多线程的原因导致生成多个handler实例

 synchronized (streamHandlerLock) {

    URLStreamHandler handler2 = null;

    // Check again with hashtable just in case another
    // thread created a handler since we last checked
    handler2 = handlers.get(protocol);

    if (handler2 != null) {
        return handler2;
    }

    // Check with factory if another thread set a
    // factory since our last check
    if (!checkedWithFactory && factory != null) {
        handler2 = factory.createURLStreamHandler(protocol);
    }

    if (handler2 != null) {
        // The handler from the factory must be given more
        // importance. Discard the default handler that
        // this thread created.
        handler = handler2;
    }

    // Insert this handler into the hashtable
    if (handler != null) {
        handlers.put(protocol, handler);
    }

}

URLStreamHandler是一个抽象类,子类包括http,https,ftp等实现类,这里我们只看http的实现类

protected URLConnection openConnection(URL var1) throws IOException {
    return this.openConnection(var1, (Proxy)null);
}

protected URLConnection openConnection(URL var1, Proxy var2) throws IOException {
    return new HttpURLConnection(var1, var2, this);
}

这里会生成一个HttpUrlConnection()对象,注意这里的HttpUrlConnection()对象和最开始的HttpUrlConnection()对象不是一个,包名不同,一个是sun公司的包,一个是谷歌官方的包。

sun/HttpURLConnection.java

protected HttpURLConnection(URL var1, Proxy var2, Handler var3) {
    super(var1);
    ....
    if(this.instProxy instanceof ApplicationProxy) {
        try {
            this.cookieHandler = CookieHandler.getDefault();
        } catch (SecurityException var5) {
            ;
        }
    } else {
        this.cookieHandler = (CookieHandler)AccessController.doPrivileged(new PrivilegedAction() {
            public CookieHandler run() {
                return CookieHandler.getDefault();
            }
        });
    }

    this.cacheHandler = (ResponseCache)AccessController.doPrivileged(new PrivilegedAction() {
        public ResponseCache run() {
            return ResponseCache.getDefault();
        }
    });
}

去掉部分不重要的代码之后,剩下的代码貌似都在获取缓存值。没有进行正常的网络连接。

此时猜测 API26的HttpUrlConnection的网络请求发生在 getInputStream()中,而 openConnection()是为获取上次请求的缓存状态

getInputStream()

URLConnection.java中

public InputStream getInputStream() throws IOException {
    throw new UnknownServiceException("protocol doesn't support input");
}

同样URLConnection是个抽象类,跳到对应的HttpUrlConnection.java类中

public synchronized InputStream getInputStream() throws IOException {
    this.connecting = true;
    SocketPermission var1 = this.URLtoSocketPermission(this.url);
    if(var1 != null) {
        try {
            return (InputStream)AccessController.doPrivilegedWithCombiner(new PrivilegedExceptionAction() {
                public InputStream run() throws IOException {
                    return HttpURLConnection.this.getInputStream0();
                }
            }, (AccessControlContext)null, new Permission[]{var1});
        } catch (PrivilegedActionException var3) {
            throw (IOException)var3.getException();
        }
    } else {
        return this.getInputStream0();
    }
}

一系列处理最后都会跳转到getInputStream0()函数内

getInputStream0()函数中的代码过多,就不贴出来了。来大概讲下流程

//判断是否有缓存,有的话直接返回
if(this.inputStream != null) {
   return this.inputStream;
}


//是否正在进行输入输出字符流的行为,如果有的判断输出流,因为这时网络连接还未建立,所以只可能是输出流到缓存
 if(this.streaming()) {
    if(this.strOutputStream == null) {
        this.getOutputStream();
    }

    this.strOutputStream.close();
    if(!this.strOutputStream.writtenOK()) {
        throw new IOException("Incomplete output stream");
    }
  }

同样 getOutputStream 会跳到getOutputStream0()中

getOutputStream()

//会对URL做一次校验,主要针对host,header,protocol和authority等
SocketPermission var1 = this.URLtoSocketPermission(this.url);

getOutputStream0()

//有写流的行为的话,就一定是POST请求,因为GET请求不需要请求体
//所以强制修改method为POST
if(this.method.equals("GET")) {
    this.method = "POST";
}

//检查是否有可用的链接,没有则进行重新链接
if(!this.checkReuseConnection()) {
    this.connect();
}

checkReuseConnection()

//先检查connected,再检查reuseClient,说明reuseClient是connect的基础连接
private boolean checkReuseConnection() {
    if(this.connected) {
        return true;
    } else if(this.reuseClient != null) {
        this.http = this.reuseClient;
        this.http.setReadTimeout(this.getReadTimeout());
        this.http.reuse = false;
        this.reuseClient = null;
        this.connected = true;
        return true;
    } else {
        return false;
    }
}

回到getOutputStream0()

//添加请求部分
if(this.streaming() && this.strOutputStream == null) {
    this.writeRequests();
}

接下来回到主流程 connect()部分会调用 plainConnect0()方法

有缓存拿缓存,缓存信息完整的话就不需要重新进行网络连接了

if(this.cacheHandler != null && this.getUseCaches()) {
    try {
        //对url做一次处理,兼容一些缺少/等情况的url
        URI var1 = ParseUtil.toURI(this.url);
        if(var1 != null) {
            this.cachedResponse = this.cacheHandler.get(var1, this.getRequestMethod(), this.getUserSetHeaders().getHeaders());
            if("https".equalsIgnoreCase(var1.getScheme()) && !(this.cachedResponse instanceof SecureCacheResponse)) {
                this.cachedResponse = null;
            }

            if(logger.isLoggable(Level.FINEST)) {
                logger.finest("Cache Request for " + var1 + " / " + this.getRequestMethod());
                logger.finest("From cache: " + (this.cachedResponse != null?this.cachedResponse.toString():"null"));
            }

            if(this.cachedResponse != null) {
                this.cachedHeaders = this.mapToMessageHeader(this.cachedResponse.getHeaders());
                this.cachedInputStream = this.cachedResponse.getBody();
            }
        }
    } catch (IOException var6) {
        ;
    }

    if(this.cachedHeaders != null && this.cachedInputStream != null) {
        this.connected = true;
        return;
    }

    this.cachedResponse = null;
}

没有连接缓存的话

//第一次连接还是再次连接对应的方法不一样
if(!this.failedOnce) {
    this.http = this.getNewHttpClient(this.url, (Proxy)null, this.connectTimeout);
    this.http.setReadTimeout(this.readTimeout);
} else {
    this.http = this.getNewHttpClient(this.url, (Proxy)null, this.connectTimeout, false);
    this.http.setReadTimeout(this.readTimeout);
}

不管是否是第一次连接都会生成一个HttpClient对象,而这个对象才是连接网络的主要成员。其中变量var3表示是否是第一次发起连接,如果是,并且httpClient对象var5是有缓存的,这时需要做一些缓存清理,变量重置的操作。因为需要重新开启一个新的连接,需要先把老的连接清理掉

public static HttpClient New(URL var0, Proxy var1, int var2, boolean var3, HttpURLConnection var4) throws IOException {
    if(var1 == null) {
        var1 = Proxy.NO_PROXY;
    }

    HttpClient var5 = null;
    if(var3) {
        var5 = kac.get(var0, (Object)null);
        if(var5 != null && var4 != null && var4.streaming() && var4.getRequestMethod() == "POST" && !var5.available()) {
            var5.inCache = false;
            var5.closeServer();
            var5 = null;
        }

        if(var5 != null) {
            if(var5.proxy != null && var5.proxy.equals(var1) || var5.proxy == null && var1 == null) {
                synchronized(var5) {
                    var5.cachedHttpClient = true;

                    assert var5.inCache;

                    var5.inCache = false;
                    if(var4 != null && var5.needsTunneling()) {
                        var4.setTunnelState(TunnelState.TUNNELING);
                    }

                    logFinest("KeepAlive stream retrieved from the cache, " + var5);
                }
            } else {
                synchronized(var5) {
                    var5.inCache = false;
                    var5.closeServer();
                }

                var5 = null;
            }
        }
    }
    
    //生成HttpClient对象,进行网络连接
    if(var5 == null) {
        var5 = new HttpClient(var0, var1, var2);
    } else {
        SecurityManager var6 = System.getSecurityManager();
        if(var6 != null) {
            if(var5.proxy != Proxy.NO_PROXY && var5.proxy != null) {
                var6.checkConnect(var0.getHost(), var0.getPort());
            } else {
                var6.checkConnect(InetAddress.getByName(var0.getHost()).getHostAddress(), var0.getPort());
            }
        }

        var5.url = var0;
    }

    return var5;
}

**HttpClient.java **

protected HttpClient(URL var1, Proxy var2, int var3) throws IOException {
    this.cachedHttpClient = false;
    this.poster = null;
    this.failedOnce = false;
    this.ignoreContinue = true;
    this.usingProxy = false;
    this.keepingAlive = false;
    this.keepAliveConnections = -1;
    this.keepAliveTimeout = 0;
    this.cacheRequest = null;
    this.reuse = false;
    this.capture = null;
    this.proxy = var2 == null?Proxy.NO_PROXY:var2;
    this.host = var1.getHost();
    this.url = var1;
    this.port = var1.getPort();
    if(this.port == -1) {
        this.port = this.getDefaultPort();
    }

    this.setConnectTimeout(var3);
    this.capture = HttpCapture.getCapture(var1);
    //开启连接服务
    this.openServer();
}

openServer()函数内调用 openServer(this.host, this.port); 传入host和port开启 serverSocket.setTcpNoDelay(true);服务

网络连接成功,http请求参数也已经设置成功,接下来只等待网络回调把数据流写会缓存即可。

补充说明

  • HttpURLConnection的connect()函数,实际上只是建立了一个与服务器的tcp连接,并没有实际发送http请求。 无论是post还是get,http请求实际上直到HttpURLConnection的getInputStream()这个函数里面才正式发送出去。
  • 在用POST方式发送URL请求时,URL请求参数的设定顺序是重中之重, 对connection对象的一切配置(那一堆set函数) 都必须要在connect()函数执行之前完成。而对outputStream的写操作,又必须要在inputStream的读操作之前。 这些顺序实际上是由http请求的格式决定的。
  • http请求实际上由两部分组成, 一个是http头,所有关于此次http请求的配置都在http头里面定义, 一个是正文content。 connect()函数会根据HttpURLConnection对象的配置值生成http头部信息,因此在调用connect函数之前, 就必须把所有的配置准备好。
  • 在http头后面紧跟着的是http请求的正文,正文的内容是通过outputStream流写入的,实际上outputStream不是一个网络流,充其量是个字符串流,往里面写入的东西不会立即发送到网络, 而是存在于内存缓冲区中,待outputStream流关闭时,根据输入的内容生成http正文。 至此,http请求的东西已经全部准备就绪。在getInputStream()函数调用的时候,就会把准备好的http请求 正式发送到服务器了,然后返回一个输入流,用于读取服务器对于此次http请求的返回信息。由于http 请求在getInputStream的时候已经发送出去了(包括http头和正文),因此在getInputStream()函数 之后对connection对象进行设置(对http头的信息进行修改)或者写入outputStream(对正文进行修改) 都是没有意义的了,执行这些操作会导致异常的发生。

参考文章

Android-浅析-HttpURLConnection
网络请求HttpURLConnection剖析
Android每周一轮子:HttpURLConnection

另外

个人的github
闲暇之余写的故事