Android框架之OkHttp3源碼的示例分析

這篇文章將為大家詳細(xì)講解有關(guān)Android框架之OkHttp3源碼的示例分析，小編覺得挺實(shí)用的，因此分享給大家做個(gè)參考，希望大家閱讀完這篇文章后可以有所收獲。

OkHttp流程圖

OkHttp基本使用

gradle依賴

implementation 'com.squareup.okhttp3:okhttp:3.11.0'

implementation 'com.squareup.okio:okio:1.15.0'

/**
    *這里拿get請(qǐng)求來
     * 異步的get請(qǐng)求
     */
    public void okhttpAsyn() {
        //設(shè)置超時(shí)的時(shí)間
        OkHttpClient.Builder builder = new OkHttpClient.Builder()
                .connectTimeout(15, TimeUnit.SECONDS)
                .writeTimeout(20, TimeUnit.SECONDS)
                .readTimeout(20, TimeUnit.SECONDS)
                ;
        OkHttpClient okHttpClient = builder.build();
        Request request = new Request.Builder()
                .get() //設(shè)置請(qǐng)求模式
                .url("https://www.baidu.com/")
                .build();

        Call call = okHttpClient.newCall(request);
        call.enqueue(new Callback() {
            @Override
            public void onFailure(Call call, IOException e) {
                Log.d("MainActivity", "-----------onFailure-----------");
            }

            @Override
            public void onResponse(Call call, Response response) throws IOException {
                Log.d("MainActivity", "----onResponse----" + response.body().toString());
                runOnUiThread(new Runnable() {
                    @Override
                    public void run() {
                        Toast.makeText(MainActivity.this, "請(qǐng)求成功", Toast.LENGTH_LONG).show();
                    }
                });

            }
        });
    }

OkHttp源碼分析

從OkHttp的基本使用中，我們看到，通過okHttpClient.newCall()方法，拿到這個(gè)call對(duì)象，我們看看newCall是怎么走的

/**
   * Prepares the {@code request} to be executed at some point in the future.
   */
  @Override public Call newCall(Request request) {
    return RealCall.newRealCall(this, request, false /* for web socket */);
  }


  static RealCall newRealCall(OkHttpClient client, Request originalRequest, boolean forWebSocket) {
    // Safely publish the Call instance to the EventListener.
    RealCall call = new RealCall(client, originalRequest, forWebSocket);
    call.eventListener = client.eventListenerFactory().create(call);
    return call;
  }

從這里的源碼知道，okHttpClient.newCall()實(shí)際上返回的是RealCall對(duì)象，而call.enqueue()，實(shí)際上是調(diào)用的了RealCall中的enqueue()方法，我們看看enqueue()方法方法怎么走。

@Override public void enqueue(Callback responseCallback) {
    synchronized (this) {
      if (executed) throw new IllegalStateException("Already Executed");
      executed = true;
    }
    captureCallStackTrace();
    eventListener.callStart(this);
    client.dispatcher().enqueue(new AsyncCall(responseCallback));
  }

可以看到client.dispatcher().enqueue(new AsyncCall(responseCallback));這句代碼，也就是說，最終是有的請(qǐng)求是有dispatcher來完成，我們看看dispatcher。

/*
 * Copyright (C) 2013 Square, Inc.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
package okhttp3;

import java.util.ArrayDeque;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Deque;
import java.util.Iterator;
import java.util.List;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.SynchronousQueue;
import java.util.concurrent.ThreadPoolExecutor;
import java.util.concurrent.TimeUnit;
import javax.annotation.Nullable;
import okhttp3.RealCall.AsyncCall;
import okhttp3.internal.Util;

/**
 * Policy on when async requests are executed.
 *
 * <p>Each dispatcher uses an {@link ExecutorService} to run calls internally. If you supply your
 * own executor, it should be able to run {@linkplain #getMaxRequests the configured maximum} number
 * of calls concurrently.
 */
public final class Dispatcher {
  //最大請(qǐng)求的并發(fā)數(shù)
  private int maxRequests = 64;
  //每個(gè)主機(jī)最大請(qǐng)求數(shù)
  private int maxRequestsPerHost = 5;
  private @Nullable Runnable idleCallback;

  /** 消費(fèi)線程池 */
  private @Nullable ExecutorService executorService;

  /** 準(zhǔn)備運(yùn)行的異步請(qǐng)求隊(duì)列 */
  private final Deque<AsyncCall> readyAsyncCalls = new ArrayDeque<>();

  /** 正在運(yùn)行的異步請(qǐng)求隊(duì)列 */
  private final Deque<AsyncCall> runningAsyncCalls = new ArrayDeque<>();

  /** 正在運(yùn)行的同步請(qǐng)求隊(duì)列 */
  private final Deque<RealCall> runningSyncCalls = new ArrayDeque<>();

  /** 構(gòu)造方法 */
  public Dispatcher(ExecutorService executorService) {
    this.executorService = executorService;
  }

  public Dispatcher() {
  }

  public synchronized ExecutorService executorService() {
    if (executorService == null) {
      executorService = new ThreadPoolExecutor(0, Integer.MAX_VALUE, 60, TimeUnit.SECONDS,
          new SynchronousQueue<Runnable>(), Util.threadFactory("OkHttp Dispatcher", false));
    }
    return executorService;
  }




  /**
   *
   *設(shè)置并發(fā)執(zhí)行最大的請(qǐng)求數(shù)量
   * <p>If more than {@code maxRequests} requests are in flight when this is invoked, those requests
   * will remain in flight.
   */
  public synchronized void setMaxRequests(int maxRequests) {
    if (maxRequests < 1) {
      throw new IllegalArgumentException("max < 1: " + maxRequests);
    }
    this.maxRequests = maxRequests;
    promoteCalls();
  }

  //獲取到最大請(qǐng)求的數(shù)量
  public synchronized int getMaxRequests() {
    return maxRequests;
  }

  /**
   * 設(shè)置每個(gè)主機(jī)并發(fā)執(zhí)行的請(qǐng)求的最大數(shù)量
   * <p>If more than {@code maxRequestsPerHost} requests are in flight when this is invoked, those
   * requests will remain in flight.
   *
   * <p>WebSocket connections to hosts <b>do not</b> count against this limit.
   */
  public synchronized void setMaxRequestsPerHost(int maxRequestsPerHost) {
    if (maxRequestsPerHost < 1) {
      throw new IllegalArgumentException("max < 1: " + maxRequestsPerHost);
    }
    this.maxRequestsPerHost = maxRequestsPerHost;
    promoteCalls();
  }

  //獲取每個(gè)主機(jī)最大并發(fā)數(shù)量
  public synchronized int getMaxRequestsPerHost() {
    return maxRequestsPerHost;
  }

  /**
   * Set a callback to be invoked each time the dispatcher becomes idle (when the number of running
   * calls returns to zero).
   *
   * <p>Note: The time at which a {@linkplain Call call} is considered idle is different depending
   * on whether it was run {@linkplain Call#enqueue(Callback) asynchronously} or
   * {@linkplain Call#execute() synchronously}. Asynchronous calls become idle after the
   * {@link Callback#onResponse onResponse} or {@link Callback#onFailure onFailure} callback has
   * returned. Synchronous calls become idle once {@link Call#execute() execute()} returns. This
   * means that if you are doing synchronous calls the network layer will not truly be idle until
   * every returned {@link Response} has been closed.
   */
  public synchronized void setIdleCallback(@Nullable Runnable idleCallback) {
    this.idleCallback = idleCallback;
  }

  synchronized void enqueue(AsyncCall call) {
    if (runningAsyncCalls.size() < maxRequests && runningCallsForHost(call) < maxRequestsPerHost) {
      runningAsyncCalls.add(call);
      executorService().execute(call);
    } else {
      readyAsyncCalls.add(call);
    }
  }

  /**
   * Cancel all calls currently enqueued or executing. Includes calls executed both {@linkplain
   * Call#execute() synchronously} and {@linkplain Call#enqueue asynchronously}.
   */
  public synchronized void cancelAll() {
    for (AsyncCall call : readyAsyncCalls) {
      call.get().cancel();
    }

    for (AsyncCall call : runningAsyncCalls) {
      call.get().cancel();
    }

    for (RealCall call : runningSyncCalls) {
      call.cancel();
    }
  }

  private void promoteCalls() {
    if (runningAsyncCalls.size() >= maxRequests) return; // Already running max capacity.
    if (readyAsyncCalls.isEmpty()) return; // No ready calls to promote.

    for (Iterator<AsyncCall> i = readyAsyncCalls.iterator(); i.hasNext(); ) {
      AsyncCall call = i.next();

      if (runningCallsForHost(call) < maxRequestsPerHost) {
        i.remove();
        runningAsyncCalls.add(call);
        executorService().execute(call);
      }

      if (runningAsyncCalls.size() >= maxRequests) return; // Reached max capacity.
    }
  }

 //----------------省略若干代碼-----------------------

}

我們來找到這段代碼

synchronized void enqueue(AsyncCall call) {
    if (runningAsyncCalls.size() < maxRequests && runningCallsForHost(call) < maxRequestsPerHost) {
      runningAsyncCalls.add(call);
      executorService().execute(call);
    } else {
      readyAsyncCalls.add(call);
    }
  }

當(dāng)正在運(yùn)行的異步請(qǐng)求隊(duì)列中的數(shù)量小于64并且正在運(yùn)行的請(qǐng)求主機(jī)數(shù)小于5時(shí)則把請(qǐng)求加載到runningAsyncCalls中并在線程池中執(zhí)行，否則就再入到readyAsyncCalls中進(jìn)行緩存等待。而runningAsyncCalls這個(gè)請(qǐng)求隊(duì)列存放的就是AsyncCall對(duì)象，而這個(gè)AsyncCall就是RealCall的內(nèi)部類，也就是說executorService().execute(call);實(shí)際上走的是RealCall類中的execute()方法.

@Override protected void execute() {
      boolean signalledCallback = false;
      try {
        Response response = getResponseWithInterceptorChain();
        if (retryAndFollowUpInterceptor.isCanceled()) {
          signalledCallback = true;
          responseCallback.onFailure(RealCall.this, new IOException("Canceled"));
        } else {
          signalledCallback = true;
          responseCallback.onResponse(RealCall.this, response);
        }
      } catch (IOException e) {
        if (signalledCallback) {
          // Do not signal the callback twice!
          Platform.get().log(INFO, "Callback failure for " + toLoggableString(), e);
        } else {
          eventListener.callFailed(RealCall.this, e);
          responseCallback.onFailure(RealCall.this, e);
        }
      } finally {
        client.dispatcher().finished(this);
      }
    }

這部分的代碼，相信很多人都能夠看的明白，無非就是一些成功，失敗的回調(diào)，這段代碼，最重要的是esponse response = getResponseWithInterceptorChain();和client.dispatcher().finished(this);我們先來看看client.dispatcher().finished(this);這句代碼是怎么執(zhí)行的。

/** Used by {@code AsyncCall#run} to signal completion. */
  void finished(AsyncCall call) {
    finished(runningAsyncCalls, call, true);
  }

  /** Used by {@code Call#execute} to signal completion. */
  void finished(RealCall call) {
    finished(runningSyncCalls, call, false);
  }

  private <T> void finished(Deque<T> calls, T call, boolean promoteCalls) {
    int runningCallsCount;
    Runnable idleCallback;
    synchronized (this) {
      if (!calls.remove(call)) throw new AssertionError("Call wasn't in-flight!");
      if (promoteCalls) promoteCalls();
      runningCallsCount = runningCallsCount();
      idleCallback = this.idleCallback;
    }

    if (runningCallsCount == 0 && idleCallback != null) {
      idleCallback.run();
    }
  }

private void promoteCalls() {
    if (runningAsyncCalls.size() >= maxRequests) return; // Already running max capacity.
    if (readyAsyncCalls.isEmpty()) return; // No ready calls to promote.

    for (Iterator<AsyncCall> i = readyAsyncCalls.iterator(); i.hasNext(); ) {
      AsyncCall call = i.next();

      if (runningCallsForHost(call) < maxRequestsPerHost) {
        i.remove();
        runningAsyncCalls.add(call);
        executorService().execute(call);
      }

      if (runningAsyncCalls.size() >= maxRequests) return; // Reached max capacity.
    }
  }

由于client.dispatcher().finished(this);這句代碼是放到finally中執(zhí)行的，所以無論什么情況，都會(huì)執(zhí)行上面的promoteCalls()方法，而從promoteCalls()方法中可以看出通過遍歷來獲取到下一個(gè)請(qǐng)求從而執(zhí)行下一個(gè)網(wǎng)絡(luò)請(qǐng)求。

回過頭來，我們看看這一句代碼Response response = getResponseWithInterceptorChain(); 通過getResponseWithInterceptorChain();來獲取到response，然后回調(diào)返回。很明顯getResponseWithInterceptorChain()這句代碼里面進(jìn)行了網(wǎng)絡(luò)請(qǐng)求。我們看看是怎么執(zhí)行的。

Response getResponseWithInterceptorChain() throws IOException {
    // Build a full stack of interceptors.
    List<Interceptor> interceptors = new ArrayList<>();
    interceptors.addAll(client.interceptors());
    interceptors.add(retryAndFollowUpInterceptor);
    interceptors.add(new BridgeInterceptor(client.cookieJar()));
    interceptors.add(new CacheInterceptor(client.internalCache()));
    interceptors.add(new ConnectInterceptor(client));
    if (!forWebSocket) {
      interceptors.addAll(client.networkInterceptors());
    }
    interceptors.add(new CallServerInterceptor(forWebSocket));

    Interceptor.Chain chain = new RealInterceptorChain(interceptors, null, null, null, 0,
        originalRequest, this, eventListener, client.connectTimeoutMillis(),
        client.readTimeoutMillis(), client.writeTimeoutMillis());

    return chain.proceed(originalRequest);
  }
}

從上面代碼可以知道，緩存，網(wǎng)絡(luò)請(qǐng)求，都封裝成攔截器的形式。攔截器主要用來觀察，修改以及可能短路的請(qǐng)求輸出和響應(yīng)的回來。最后return chain.proceed，而chain是通過new RealInterceptorChain來獲取到的，我們來看看RealInterceptorChain對(duì)象，然后找到proceed()方法。

public Response proceed(Request request, StreamAllocation streamAllocation, HttpCodec httpCodec,
      RealConnection connection) throws IOException {
    if (index >= interceptors.size()) throw new AssertionError();

    calls++;

    // If we already have a stream, confirm that the incoming request will use it.
    if (this.httpCodec != null && !this.connection.supportsUrl(request.url())) {
      throw new IllegalStateException("network interceptor " + interceptors.get(index - 1)
          + " must retain the same host and port");
    }

    // If we already have a stream, confirm that this is the only call to chain.proceed().
    if (this.httpCodec != null && calls > 1) {
      throw new IllegalStateException("network interceptor " + interceptors.get(index - 1)
          + " must call proceed() exactly once");
    }

    // 調(diào)用下一個(gè)攔截器
    RealInterceptorChain next = new RealInterceptorChain(interceptors, streamAllocation, httpCodec,
        connection, index + 1, request, call, eventListener, connectTimeout, readTimeout,
        writeTimeout);
    Interceptor interceptor = interceptors.get(index);
    Response response = interceptor.intercept(next); //調(diào)用攔截器中的intercept()方法

    // Confirm that the next interceptor made its required call to chain.proceed().
    if (httpCodec != null && index + 1 < interceptors.size() && next.calls != 1) {
      throw new IllegalStateException("network interceptor " + interceptor
          + " must call proceed() exactly once");
    }

    // Confirm that the intercepted response isn't null.
    if (response == null) {
      throw new NullPointerException("interceptor " + interceptor + " returned null");
    }

    if (response.body() == null) {
      throw new IllegalStateException(
          "interceptor " + interceptor + " returned a response with no body");
    }

    return response;
  }

從上面的代碼可以看出來，chain.proceed主要是講集合中的攔截器遍歷出來，然后通過調(diào)用每一個(gè)攔截器中的intercept()方法，然后獲取到response結(jié)果，返回。

我們看看CacheInterceptor這個(gè)類，找到intercept()方法。

@Override public Response intercept(Chain chain) throws IOException {
    Response cacheCandidate = cache != null
        ? cache.get(chain.request())
        : null;

    long now = System.currentTimeMillis();

    //創(chuàng)建CacheStrategy.Factory對(duì)象，進(jìn)行緩存配置
    CacheStrategy strategy = new CacheStrategy.Factory(now, chain.request(), cacheCandidate).get();
    //網(wǎng)絡(luò)請(qǐng)求
    Request networkRequest = strategy.networkRequest;
    //緩存響應(yīng)
    Response cacheResponse = strategy.cacheResponse;

    if (cache != null) {
    //記錄當(dāng)前請(qǐng)求是網(wǎng)絡(luò)發(fā)起還是緩存發(fā)起
      cache.trackResponse(strategy);
    }

    if (cacheCandidate != null && cacheResponse == null) {
      closeQuietly(cacheCandidate.body()); // The cache candidate wasn't applicable. Close it.
    }

    // 不進(jìn)行網(wǎng)絡(luò)請(qǐng)求并且緩存不存在或者過期則返回504錯(cuò)誤
    if (networkRequest == null && cacheResponse == null) {
      return new Response.Builder()
          .request(chain.request())
          .protocol(Protocol.HTTP_1_1)
          .code(504)
          .message("Unsatisfiable Request (only-if-cached)")
          .body(Util.EMPTY_RESPONSE)
          .sentRequestAtMillis(-1L)
          .receivedResponseAtMillis(System.currentTimeMillis())
          .build();
    }

    // 不進(jìn)行網(wǎng)絡(luò)請(qǐng)求，而且緩存可以使用，直接返回緩存
    if (networkRequest == null) {
      return cacheResponse.newBuilder()
          .cacheResponse(stripBody(cacheResponse))
          .build();
    }

    //進(jìn)行網(wǎng)絡(luò)請(qǐng)求
    Response networkResponse = null;
    try {
      networkResponse = chain.proceed(networkRequest);
    } finally {
      // If we're crashing on I/O or otherwise, don't leak the cache body.
      if (networkResponse == null && cacheCandidate != null) {
        closeQuietly(cacheCandidate.body());
      }
    }

    //---------省略若干代碼-------------

    return response;
  }

上面我做了很多注釋，基本的流程是有緩存就取緩存里面的，沒有緩存就請(qǐng)求網(wǎng)絡(luò)。我們來看看網(wǎng)絡(luò)請(qǐng)求的類CallServerInterceptor

/*
 * Copyright (C) 2016 Square, Inc.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
package okhttp3.internal.http;

import java.io.IOException;
import java.net.ProtocolException;
import okhttp3.Interceptor;
import okhttp3.Request;
import okhttp3.Response;
import okhttp3.internal.Util;
import okhttp3.internal.connection.RealConnection;
import okhttp3.internal.connection.StreamAllocation;
import okio.Buffer;
import okio.BufferedSink;
import okio.ForwardingSink;
import okio.Okio;
import okio.Sink;

/** This is the last interceptor in the chain. It makes a network call to the server. */
public final class CallServerInterceptor implements Interceptor {
  private final boolean forWebSocket;

  public CallServerInterceptor(boolean forWebSocket) {
    this.forWebSocket = forWebSocket;
  }

  @Override public Response intercept(Chain chain) throws IOException {
    RealInterceptorChain realChain = (RealInterceptorChain) chain;
    HttpCodec httpCodec = realChain.httpStream();
    StreamAllocation streamAllocation = realChain.streamAllocation();
    RealConnection connection = (RealConnection) realChain.connection();
    Request request = realChain.request();

    long sentRequestMillis = System.currentTimeMillis();

    realChain.eventListener().requestHeadersStart(realChain.call());
    httpCodec.writeRequestHeaders(request);
    realChain.eventListener().requestHeadersEnd(realChain.call(), request);

    Response.Builder responseBuilder = null;
    if (HttpMethod.permitsRequestBody(request.method()) && request.body() != null) {
      // If there's a "Expect: 100-continue" header on the request, wait for a "HTTP/1.1 100
      // Continue" response before transmitting the request body. If we don't get that, return
      // what we did get (such as a 4xx response) without ever transmitting the request body.
      if ("100-continue".equalsIgnoreCase(request.header("Expect"))) {
        httpCodec.flushRequest();
        realChain.eventListener().responseHeadersStart(realChain.call());
        responseBuilder = httpCodec.readResponseHeaders(true);
      }

      if (responseBuilder == null) {
        // Write the request body if the "Expect: 100-continue" expectation was met.
        realChain.eventListener().requestBodyStart(realChain.call());
        long contentLength = request.body().contentLength();
        CountingSink requestBodyOut =
            new CountingSink(httpCodec.createRequestBody(request, contentLength));
        BufferedSink bufferedRequestBody = Okio.buffer(requestBodyOut);

        request.body().writeTo(bufferedRequestBody);
        bufferedRequestBody.close();
        realChain.eventListener()
            .requestBodyEnd(realChain.call(), requestBodyOut.successfulCount);
      } else if (!connection.isMultiplexed()) {
        // If the "Expect: 100-continue" expectation wasn't met, prevent the HTTP/1 connection
        // from being reused. Otherwise we're still obligated to transmit the request body to
        // leave the connection in a consistent state.
        streamAllocation.noNewStreams();
      }
    }

    httpCodec.finishRequest();

    if (responseBuilder == null) {
      realChain.eventListener().responseHeadersStart(realChain.call());
      responseBuilder = httpCodec.readResponseHeaders(false);
    }

    Response response = responseBuilder
        .request(request)
        .handshake(streamAllocation.connection().handshake())
        .sentRequestAtMillis(sentRequestMillis)
        .receivedResponseAtMillis(System.currentTimeMillis())
        .build();

    int code = response.code();
    if (code == 100) {
      // server sent a 100-continue even though we did not request one.
      // try again to read the actual response
      responseBuilder = httpCodec.readResponseHeaders(false);

      response = responseBuilder
              .request(request)
              .handshake(streamAllocation.connection().handshake())
              .sentRequestAtMillis(sentRequestMillis)
              .receivedResponseAtMillis(System.currentTimeMillis())
              .build();

      code = response.code();
    }

    realChain.eventListener()
            .responseHeadersEnd(realChain.call(), response);

    if (forWebSocket && code == 101) {
      // Connection is upgrading, but we need to ensure interceptors see a non-null response body.
      response = response.newBuilder()
          .body(Util.EMPTY_RESPONSE)
          .build();
    } else {
      response = response.newBuilder()
          .body(httpCodec.openResponseBody(response))
          .build();
    }

    if ("close".equalsIgnoreCase(response.request().header("Connection"))
        || "close".equalsIgnoreCase(response.header("Connection"))) {
      streamAllocation.noNewStreams();
    }

    if ((code == 204 || code == 205) && response.body().contentLength() > 0) {
      throw new ProtocolException(
          "HTTP " + code + " had non-zero Content-Length: " + response.body().contentLength());
    }

    return response;
  }
}

關(guān)于“Android框架之OkHttp3源碼的示例分析”這篇文章就分享到這里了，希望以上內(nèi)容可以對(duì)大家有一定的幫助，使各位可以學(xué)到更多知識(shí)，如果覺得文章不錯(cuò)，請(qǐng)把它分享出去讓更多的人看到。