Android xUtils3源码解析之网络模块
Posted 一口仨馍
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了Android xUtils3源码解析之网络模块相关的知识,希望对你有一定的参考价值。
本文已授权微信公众号《非著名程序员》原创首发,转载请务必注明出处。
xUtils3源码解析系列
一. Android xUtils3源码解析之网络模块
二. Android xUtils3源码解析之图片模块
三. Android xUtils3源码解析之注解模块
四. Android xUtils3源码解析之数据库模块
关于xUtils3
- xUtils 包含了很多实用的android工具.
- xUtils 支持超大文件(超过2G)上传,更全面的http请求协议支持(11种谓词),拥有更加灵活的ORM,更多的事件注解支持且不受混淆影响…
- xUtils 最低兼容Android 4.0 (api level 14). (Android 2.3?)
- xUtils3变化较多所以建立了新的项目不在旧版(github.com/wyouflf/xUtils)上继续维护, 相对于旧版本:
- HTTP实现替换HttpClient为UrlConnection, 自动解析回调泛型, 更安全的断点续传策略.
- 支持标准的Cookie策略, 区分domain, path…
- 事件注解去除不常用的功能, 提高性能.
- 数据库api简化提高性能, 达到和greenDao一致的性能.
- 图片绑定支持gif(受系统兼容性影响, 部分gif文件只能静态显示), webp; 支持圆角, 圆形, 方形等裁剪, 支持自动旋转…
xUtils3四大模块:网络请求、图片加载、ORM框架和事件注解。本文阅读分析网络请求相关代码。
使用版本:compile 'org.xutils:xutils:3.3.36'
xUtils3项目地址 : https://github.com/wyouflf/xUtils3
初始化
Ext.init(this);
public static class Ext
private static Application app;
public static void init(Application app)
if (Ext.app == null)
Ext.app = app;
获取ApplicationCotext,方便以后调用。在Ext中有个静态代码块。详情如下:
public static class Ext
private static TaskController taskController;
static
TaskControllerImpl.registerInstance();
// 默认信任所有https域名
HttpsURLConnection.setDefaultHostnameVerifier(new HostnameVerifier()
@Override
public boolean verify(String hostname, SSLSession session)
return true;
);
public static void setTaskController(TaskController taskController)
if (Ext.taskController == null)
Ext.taskController = taskController;
public static void registerInstance()
if (instance == null)
synchronized (TaskController.class)
if (instance == null)
instance = new TaskControllerImpl();
x.Ext.setTaskController(instance);
这段静态代码块的作用是注册TaskController对象为TaskControllerImpl实例。并设置信任所有https域名。ps:还是中文注释看着爽。
添加参数
RequestParams param = new RequestParams(url);
param.addParameter("name","一口仨馍");
public RequestParams(String uri)
this(uri, null, null, null);
public RequestParams(String uri, ParamsBuilder builder, String[] signs, String[] cacheKeys)
if (uri != null && builder == null)
builder = new DefaultParamsBuilder();
this.uri = uri;
this.signs = signs;
this.cacheKeys = cacheKeys;
this.builder = builder;
public void addParameter(String name, Object value)
if (value == null) return;
if (method == null || HttpMethod.permitsRequestBody(method))
if (!TextUtils.isEmpty(name))
if (value instanceof File
|| value instanceof InputStream
|| value instanceof byte[])
this.fileParams.add(new KeyValue(name, value));
else
if (value instanceof List)
...
else if (value instanceof JSONArray)
...
else if (value.getClass().isArray())
...
else
this.bodyParams.add(new KeyValue(name, value));
else
this.bodyContent = value.toString();
else
...
这个时候还没有设置请求的方式,例如GET、POST等,所以mothod属性为null。而value为String类型的参数,所以name和value被以KeyValue结构的形式保存在bodyParams中。
网络请求
下文以x.http().post(requestParams, new Callback.CommonCallback<String>()
过程为例,逐步查看xUtils3调用流程。
x.http()
public static HttpManager http()
if (Ext.httpManager == null)
HttpManagerImpl.registerInstance();
return Ext.httpManager;
public static void registerInstance()
if (instance == null)
synchronized (lock)
if (instance == null)
instance = new HttpManagerImpl();
x.Ext.setHttpManager(instance);
public static void setHttpManager(HttpManager httpManager)
Ext.httpManager = httpManager;
第一次调用的时候会初始化Ext#imageManager对象为HttpManagerImpl,以后所有HTTP/HTTPS相关调用都通过HttpManagerImpl管理。
HttpManagerImpl.post()
public final class HttpManagerImpl implements HttpManager
@Override
public <T> Callback.Cancelable post(RequestParams entity, Callback.CommonCallback<T> callback)
return request(HttpMethod.POST, entity, callback);
@Override
public <T> Callback.Cancelable request(HttpMethod method, RequestParams entity, Callback.CommonCallback<T> callback)
entity.setMethod(method);
Callback.Cancelable cancelable = null;
if (callback instanceof Callback.Cancelable)
cancelable = (Callback.Cancelable) callback;
HttpTask<T> task = new HttpTask<T>(entity, cancelable, callback);
return x.task().start(task);
这里以HttpMethod == HttpMethod.POST,entity为构造的请求参数,Callback.CommonCallback泛型为String查看流程。
首先设置RequestParams中请求方式为HttpMethod.POST,Callback.CommonCallback没有实现Callback.Cancelable接口,所以这里的if语句不成立,即cancelable为null。接下来构建HttpTask对象。跟进。
public class HttpTask<ResultType> extends AbsTask<ResultType> implements ProgressHandler
private static final PriorityExecutor HTTP_EXECUTOR = new PriorityExecutor(5, true);
public HttpTask(RequestParams params, Callback.Cancelable cancelHandler,
Callback.CommonCallback<ResultType> callback)
super(cancelHandler);
this.params = params;
this.callback = callback;
if (callback instanceof Callback.CacheCallback)
this.cacheCallback = (Callback.CacheCallback<ResultType>) callback;
...
if (params.getExecutor() != null)
this.executor = params.getExecutor();
else
if (cacheCallback != null)
this.executor = CACHE_EXECUTOR;
else
this.executor = HTTP_EXECUTOR;
我们没有使用自定义的Executor(可以通过RequestParams.setExecutor()设置),所以params.getExecutor()返回是null,前文也提过CommonCallback没有实现CacheCallback,所以第一个if语句不成立,即cacheCallback为null。即,在HttpTask的构造函数中除了赋值params、callback之外,最主要的是指定了执行请求的线程池为HTTP_EXECUTOR。下面跟进看下这个HTTP_EXECUTOR。
FIFO线程池
public class PriorityExecutor implements Executor
private static final int CORE_POOL_SIZE = 5;
private static final int MAXIMUM_POOL_SIZE = 256;
private static final int KEEP_ALIVE = 1;
private static final AtomicLong SEQ_SEED = new AtomicLong(0);
private static final ThreadFactory sThreadFactory = new ThreadFactory()
private final AtomicInteger mCount = new AtomicInteger(1);
@Override
public Thread newThread(Runnable runnable)
return new Thread(runnable, "xTID#" + mCount.getAndIncrement());
;
/**
* @param poolSize 工作线程数
* @param fifo 优先级相同时, 等待队列的是否优先执行先加入的任务.
*/
public PriorityExecutor(int poolSize, boolean fifo)
BlockingQueue<Runnable> mPoolWorkQueue =
new PriorityBlockingQueue<Runnable>(MAXIMUM_POOL_SIZE, fifo ? FIFO_CMP : FILO_CMP);
mThreadPoolExecutor = new ThreadPoolExecutor(
poolSize,
MAXIMUM_POOL_SIZE,
KEEP_ALIVE,
TimeUnit.SECONDS,
mPoolWorkQueue,
sThreadFactory);
private static final Comparator<Runnable> FIFO_CMP = new Comparator<Runnable>()
@Override
public int compare(Runnable lhs, Runnable rhs)
if (lhs instanceof PriorityRunnable && rhs instanceof PriorityRunnable)
PriorityRunnable lpr = ((PriorityRunnable) lhs);
PriorityRunnable rpr = ((PriorityRunnable) rhs);
int result = lpr.priority.ordinal() - rpr.priority.ordinal();
return result == 0 ? (int) (lpr.SEQ - rpr.SEQ) : result;
else
return 0;
;
@Override
public void execute(Runnable runnable)
if (runnable instanceof PriorityRunnable)
((PriorityRunnable) runnable).SEQ = SEQ_SEED.getAndIncrement();
mThreadPoolExecutor.execute(runnable);
自定义了一个线程池,核心线程数是5,最大256,线程存活时间为1s,fifo(first in first out)类型。在执行Runnable之前,给PriorityRunnable的SEQ属性赋值(每次+1),并对比SEQ的值实现优先级。优先级相同时,SEQ值小的先执行。
初始化HttpTask之后,调用了x.task().start(task)
,x.task()返回Ext.taskController,实际返回是TaskControllerImpl对象,详见x$Ext中static代码块。所以实际上调用的是TaskControllerImpl.start()。
管理任务
public final class TaskControllerImpl implements TaskController
@Override
public <T> AbsTask<T> start(AbsTask<T> task)
TaskProxy<T> proxy = null;
if (task instanceof TaskProxy)
proxy = (TaskProxy<T>) task;
else
proxy = new TaskProxy<T>(task);
try
proxy.doBackground();
catch (Throwable ex)
LogUtil.e(ex.getMessage(), ex);
return proxy;
/*package*/ TaskProxy(AbsTask<ResultType> task)
super(task);
this.task = task;
this.task.setTaskProxy(this);
this.setTaskProxy(null);
Executor taskExecutor = task.getExecutor();
if (taskExecutor == null)
taskExecutor = sDefaultExecutor;
this.executor = taskExecutor;
任务代理
首先,将HttpTask包装成TaskProxy,然后执行TaskProxy.doBackground()。包装成TaskProxy对象的过程无非就是设置代理任务。ps:目前没看出这个TaskProxy存在的意义,只有一个HttpTask,难道是为了可拓展?重点看TaskProxy.doBackground()。
/*package*/ class TaskProxy<ResultType> extends AbsTask<ResultType>
@Override
protected final ResultType doBackground() throws Throwable
this.onWaiting();
PriorityRunnable runnable = new PriorityRunnable(
task.getPriority(),
new Runnable()
@Override
public void run()
try
// 等待过程中取消
if (callOnCanceled || TaskProxy.this.isCancelled())
throw new Callback.CancelledException("");
// start running
TaskProxy.this.onStarted();
if (TaskProxy.this.isCancelled()) // 开始时取消
throw new Callback.CancelledException("");
// 执行task, 得到结果.
task.setResult(task.doBackground());
TaskProxy.this.setResult(task.getResult());
// 未在doBackground过程中取消成功
if (TaskProxy.this.isCancelled())
throw new Callback.CancelledException("");
// 执行成功
TaskProxy.this.onSuccess(task.getResult());
catch (Callback.CancelledException cex)
TaskProxy.this.onCancelled(cex);
catch (Throwable ex)
TaskProxy.this.onError(ex, false);
finally
TaskProxy.this.onFinished();
);
this.executor.execute(runnable);
return null;
this.onWaiting()的作用是将任务置为等待状态,对阅读代码无影响,继续。PriorityRunnable实现了Runnable接口,为传递进来的Runnable对象添加了priority属性,priority默认为3(优先级为0、1、2、3、4、5、6,数字越小,优先级越高)。之后会将PriorityRunnable添加进HTTP_EXECUTOR并依据优先级执行。callOnCanceled默认为false,之后设置任务状态为started,回调onStarted()方法。这些都不是重点,重点在下面两行:
// 执行task, 得到结果.
task.setResult(task.doBackground());
TaskProxy.this.setResult(task.getResult());
正室不死,代理终究还是代理。在这里调用了HttpTask.doBackground()。看样子真正的执行请求都在这里,跟进。
protected ResultType doBackground() throws Throwable
...
ResultType result = null;
// 获取LoadType
resolveLoadType();
// 创建真正的网络请求
request = createNewRequest();
...
// 是否重试,默认2次
boolean retry = true;
// 已经重试的次数
int retryCount = 0;
Throwable exception = null;
HttpRetryHandler retryHandler = this.params.getHttpRetryHandler();
if (retryHandler == null)
retryHandler = new HttpRetryHandler();
// 设置最大重试次数
retryHandler.setMaxRetryCount(this.params.getMaxRetryCount());
...
Object cacheResult = null;
...
if (trustCache == null)
trustCache = false;
...
// 发起请求
retry = true;
while (retry)
retry = false;
try
if (this.isCancelled())
throw new Callback.CancelledException("cancelled before request");
// 由loader发起请求, 拿到结果.
this.request.close(); // retry 前关闭上次请求
try
clearRawResult();
// 开始请求工作
LogUtil.d("load: " + this.request.getRequestUri());
requestWorker = new RequestWorker();
// 真正开始请求
requestWorker.request();
if (requestWorker.ex != null)
throw requestWorker.ex;
rawResult = requestWorker.result;
catch (Throwable ex)
clearRawResult();
i...
if (prepareCallback != null)
...
else
result = (ResultType) rawResult;
...
catch (HttpRedirectException redirectEx)
retry = true;
LogUtil.w("Http Redirect:" + params.getUri());
catch (Throwable ex)
switch (this.request.getResponseCode())
case 204: // empty content
case 205: // empty content
case 304: // disk cache is valid.
return null;
default:
exception = ex;
if (this.isCancelled() && !(exception instanceof Callback.CancelledException))
exception = new Callback.CancelledException("canceled by user");
retry = retryHandler.canRetry(this.request, exception, ++retryCount);
if (exception != null && result == null && !trustCache)
hasException = true;
throw exception;
return result;
有些长,我们一点点的看。首先,ResultType肯定是我们传递进来的泛型String。resolveLoadType()为loadType赋值,港真,这里的loadType和ResultType是一样的。没搞明白为什么ResultType能解决的事情,又定义了一个loadType属性。难道是为了好区分ResultType是要返回的类型,loadType是要解析的类型?实际上两者是一样的,在这里都是String。非要说区别的话,ResultType是String,loadType为String.class。
请求参数的处理
// 初始化请求参数
private UriRequest createNewRequest() throws Throwable
// init request
params.init();
UriRequest result = UriRequestFactory.getUriRequest(params, loadType);
result.setCallingClassLoader(callback.getClass().getClassLoader());
result.setProgressHandler(this);
this.loadingUpdateMaxTimeSpan = params.getLoadingUpdateMaxTimeSpan();
this.update(FLAG_REQUEST_CREATED, result);
return result;
在params.init()中,主要是设置信任所有证书。主要关注点在下面的创建UriRequest对象。
public static UriRequest getUriRequest(RequestParams params, Type loadType) throws Throwable
// get scheme
String scheme = null;
String uri = params.getUri();
int index = uri.indexOf(":");
if (index > 0)
scheme = uri.substring(0, index);
else if (uri.startsWith("/"))
scheme = "file";
// get UriRequest
if (!TextUtils.isEmpty(scheme))
Class<? extends UriRequest> cls = SCHEME_CLS_MAP.get(scheme);
if (cls != null)
Constructor<? extends UriRequest> constructor
= cls.getConstructor(RequestParams.class, Class.class);
return constructor.newInstance(params, loadType);
else
if (scheme.startsWith("http"))
return new HttpRequest(params, loadType);
else if (scheme.equals("assets"))
return new AssetsRequest(params, loadType);
else if (scheme.equals("file"))
return new LocalFileRequest(params, loadType);
else
throw new IllegalArgumentException("The url not be support: " + uri);
else
throw new IllegalArgumentException("The url not be support: " + uri);
获取scheme,这里以https分析。这里好像还整个缓存Map,不管有没有缓存,返回的都是HttpRequest对象,只是来源不一样,这里就不具体分析这个存取的过程。实例化HttpRequest的时候,还有些文章。下面跟进。
public class HttpRequest extends UriRequest
/*package*/ HttpRequest(RequestParams params, Type loadType) throws Throwable
super(params, loadType);
public abstract class UriRequest implements Closeable
protected final Loader<?> loader;
/*package*/ UriRequest(RequestParams params, Type loadType) throws Throwable
this.params = params;
this.queryUrl = buildQueryUrl(params);
this.loader = LoaderFactory.getLoader(loadType, params);
运用工厂模式,通过LoaderFactory获取了一个Loader对象,这个对象在后面有很大的作用。先跟进看下。
public final class LoaderFactory
private LoaderFactory()
private static final HashMap<Type, Loader> converterHashMap = new HashMap<Type, Loader>();
static
converterHashMap.put(String.class, new StringLoader());
...
@SuppressWarnings("unchecked")
public static Loader<?> getLoader(Type type, RequestParams params)
Loader<?> result = converterHashMap.get(type);
if (result == null)
result = new ObjectLoader(type);
else
result = result.newInstance();
result.setParams(params);
return result;
传递进来的type为String.class,所以调用StringLoader.newInstance()。这里并不是反射,newInstance()只是个普通方法,返回StringLoader对象。拓展的有些远了,回到HttpTask.doBackground()。大概屡一下思路:现在已经将解析出loadType为String.class,HttpRequest对象赋值给request,并在在实例化HttpRequest对象的过程中设置Loader
retry机制
在HttpTask.doBackground()中定义了一个局部变量retry,默认为true。并通过retryHandler.setMaxRetryCount(this.params.getMaxRetryCount())
设置retry数量为2(默认)。white(retry)
中首先把retry设置为false,即“期望”一次请求成功。如果中途出现HttpRedirectException异常或者抛出Throwable并且responseCode不等于204、205、304,那么会再执行一遍while循环。其中HttpRedirectException异常是无限次retry(这点感觉还是计数要好一些),抛出Throwable才会对retry次数进行处理。下面是整个流程的分析。
// 发起请求
retry = true;
while (retry)
retry = false;
...
try
requestWorker = new RequestWorker();
requestWorker.request();
if (requestWorker.ex != null)
throw requestWorker.ex;
catch (Throwable ex)
...
throw ex;
...
catch (HttpRedirectException redirectEx)
retry = true;
LogUtil.w("Http Redirect:" + params.getUri());
catch (Throwable ex)
switch (this.request.getResponseCode())
case 204: // empty content
case 205: // empty content
case 304: // disk cache is valid.
return null;
default:
...
retry = retryHandler.canRetry(this.request, exception, ++retryCount);
private final class RequestWorker
/*private*/ Object result;
/*private*/ Throwable ex;
private RequestWorker()
public void request()
try
...
try
this.result = request.loadResult();
catch (Throwable ex)
this.ex = ex;
if (this.ex != null)
throw this.ex;
catch (Throwable ex)
this.ex = ex;
...
if (errorCode == 301 || errorCode == 302) //重定向
HttpTask.this.params = redirectParams;
HttpTask.this.request = createNewRequest();
this.ex = new HttpRedirectException(errorCode, httpEx.getMessage(), httpEx.getResult());
finally
...
正常的请求失败,会通过retryHandler.canRetry(),将retry置为true,最多执行两次retry。让我疑惑的地方在于重定向的处理。在requestWorker.request()中,如果有重定向,会抛出HttpRedirectException。但是在HttpTask#doBackground()中
try
if (requestWorker.ex != null)
throw requestWorker.ex;
catch (Throwable ex)
clearRawResult();
if (this.isCancelled())
throw new Callback.CancelledException("cancelled during request");
else
throw ex;
这里向上转型为Throwable并抛出,从而进入外层的
catch (HttpRedirectException redirectEx)
retry = true;
LogUtil.w("Http Redirect:" + params.getUri());
catch (Throwable ex)
switch (this.request.getResponseCode())
case 204: // empty content
case 205: // empty content
case 304: // disk cache is valid.
return null;
default:
exception = ex;
if (this.isCancelled() && !(exception instanceof Callback.CancelledException))
exception = new Callback.CancelledException("canceled by user");
retry = retryHandler.canRetry(this.request, exception, ++retryCount);
正常应该进入HttpRedirectException的,但是向上转型成Throwable之后就只能进入下面的catch代码块中。那么HttpRedirectException存在的意义在哪里?希望明白的给指点下。
真正的请求
requestWorker = new RequestWorker();
requestWorker.request();
HttpTask.doBackground()中请求其实只有这两行代码。RequestWorker是HttpTask的一个final内部类,requestWorker.request()方法内部会设置拦截器,处理重定向等,这些暂时不是关注的重点。总之先把流程跑通再说。
private final class RequestWorker
/*private*/ Object result;
public void request()
...
this.result = request.loadResult();
...
request实际上是HttpRequest对象。
public class HttpRequest extends UriRequest
@Override
public Object loadResult() throws Throwable
isLoading = true;
return super.loadResult();
public abstract class UriRequest implements Closeable
public Object loadResult() throws Throwable
return this.loader.load(this);
/*package*/ class StringLoader extends Loader<String>
@Override
public String load(final UriRequest request) throws Throwable
request.sendRequest();
return this.load(request.getInputStream());
@Override
public String load(final InputStream in) throws Throwable
resultStr = IOUtil.readStr(in, charset);
return resultStr;
经过层层调用最终调用request.sendRequest(),即HttpRequest.sendRequest()。看方法名字也知道干撒的了,代码太长这里就不贴了,概述下主要作用:依据之前的各种参数设置请求,发起请求,获取返回状态码,如果有自定义拦截器的话,还会调用beforeRequest()、afterRequest()。接下来调用request.getInputStream()获取输入流,并使用IOUtil.readStr()转换之后返回String,最后在RequestWorker.request()中将返回的String赋值给result。一切顺利的话,接着回到Task.doBackground()。执行task.setResult(task.doBackground())
。之后翻来覆去的调用get/setResult好像暂时没有用,可能是其他情况下的处理吧。不过也没有什么影响,接着会调用TaskProxy.this.onSuccess(task.getResult())
。在这个方法中通过Handler(获取了MainLooper)在主线程中调用callback.onSuccess(result)
。如果出现异常会则调用TaskProxy.this.onError(ex, false)
,不管成功还是失败都会调用finally中的TaskProxy.this.onFinish(ex, false)
过程都是类似的,这里就不再赘述。至此,整个网络请求包括回调结束。
取消网络请求
取消网络请求会调用cancelable.cancel()
,这里的cancelable是HttpManagerImpl.request()返回的TaskProxy对象。即等价于执行TaskProxy.cancle()。这个方法的具体实现在父类AbsTask中。
@Override
public final synchronized void cancel()
if (!this.isCancelled)
this.isCancelled = true;
cancelWorks();
if (cancelHandler != null && !cancelHandler.isCancelled())
cancelHandler.cancel();
if (this.state == State.WAITING || (this.state == State.STARTED && isCancelFast()))
if (taskProxy != null)
taskProxy.onCancelled(new Callback.CancelledException("cancelled by user"));
taskProxy.onFinished();
else if (this instanceof TaskProxy)
this.onCancelled(new Callback.CancelledException("cancelled by user"));
this.onFinished();
AbsTask实现了Cancelable接口,所以这里的cancelHandler不为null,但是cancelHandler.isCancelled()返回true,所以不会循环调用cancelHandler.cancel()。任务被创建后就进入了WAITING状态,所以会调用TaskProxy.onCancelled()和TaskProxy.onFinished()。在这两个回调方法中分别通过Hanlder在主线程中调用HttpTask的onCanclled()和onFinished()方法,之后再调用接口回调中的onCanclled()和onFinished()方法。需要注意的是HttpTask.onFinished()方法。
protected void onFinished()
if (tracker != null)
tracker.onFinished(request);
x.task().run(new Runnable()
@Override
public void run()
closeRequestSync();
);
callback.onFinished();
private void closeRequestSync()
clearRawResult();
IOUtil.closeQuietly(request);
IoUtil#
public static void closeQuietly(Closeable closeable)
if (closeable != null)
try
closeable.close();
catch (Throwable ignored)
LogUtil.d(ignored.getMessage(), ignored);
HttpRequest#
public void close() throws IOException
if (inputStream != null)
IOUtil.closeQuietly(inputStream);
inputStream = null;
if (connection != null)
connection.disconnect();
//connection = null;
在调用callback.onFinished()的同时异步执行线程,清除请求结果和关闭链接。这个过程对调用者来说是异步的无感知的。
总结
- xUtils3使用PriorityExecutor(FIFO线程池)+PriorityRunnable(带优先级的Runnable)实现网络请求。
- 通过传入MainLooper的Handler实现由子线程到主线程的切换,并调用相应的回调方法。
- 支持retry,默认2次。
- 通过给Task设置不同状态,实现不同状态下的处理,主要是为了实现cancle()。
- 自带拓展了不同类型的Callback和Loader,满足日常开发(特殊需求可仿照实现)。
以上是关于Android xUtils3源码解析之网络模块的主要内容,如果未能解决你的问题,请参考以下文章