经过前三篇文章的学习,Volley的用法我们已经掌握的差不多了,但是对于Volley的工作原理,恐怕有很多朋友还不是很清楚。因此,本篇文章中我们就来一起阅读一下Volley的源码,将它的工作流程整体地梳理一遍。同时,这也是Volley系列的最后一篇文章了。
其实,Volley的官方文档中本身就附有了一张Volley的工作流程图,如下图所示。
多数朋友突然看到一张这样的图,应该会和我一样,感觉一头雾水吧?没错,目前我们对Volley背后的工作原理还没有一个概念性的理解,直接就来看这张图自然会有些吃力。不过没关系,下面我们就去分析一下Volley的源码,之后再重新来看这张图就会好理解多了。
Volley.newRequestQueue(context)方法来获取一个RequestQueue对象,那么我们自然要从这个方法开始看起了,代码如下所示:
1. public static RequestQueue newRequestQueue(Context context) {
2. return newRequestQueue(context, null);
3. }
这个方法仅仅只有一行代码,只是调用了
newRequestQueue()的方法重载,并给第二个参数传入null。那我们看下带有两个参数的newRequestQueue()方法中的代码,如下所示:
1. public static RequestQueue newRequestQueue(Context context, HttpStack stack) {
2. new File(context.getCacheDir(), DEFAULT_CACHE_DIR);
3. "volley/0";
4. try {
5. String packageName = context.getPackageName();
6. 0);
7. "/" + info.versionCode;
8. catch (NameNotFoundException e) {
9. }
10. if (stack == null) {
11. if (Build.VERSION.SDK_INT >= 9) {
12. new HurlStack();
13. else {
14. new HttpClientStack(AndroidHttpClient.newInstance(userAgent));
15. }
16. }
17. new BasicNetwork(stack);
18. new RequestQueue(new DiskBasedCache(cacheDir), network);
19. queue.start();
20. return queue;
21. }
可以看到,这里在第10行判断如果stack是等于null的,则去创建一个HttpStack对象,这里会判断如果手机系统版本号是大于9的,则创建一个HurlStack的实例,否则就创建一个HttpClientStack的实例。实际上 HurlStack的内部就是使用HttpURLConnection进行网络通讯的,而HttpClientStack的内部则是使用HttpClient进行网络通讯的,这里为什么这样选择呢?可以参考我之前翻译的一篇文章Android访问网络,使用HttpURLConnection还是HttpClient?
创建好了HttpStack之后,接下来又创建了一个Network对象,它是用于根据传入的HttpStack对象来处理网络请求的,紧接着new出一个RequestQueue对象,并调用它的start()方法进行启动,然后将RequestQueue返回,这样newRequestQueue()的方法就执行结束了。
那么RequestQueue的start()方法内部到底执行了什么东西呢?我们跟进去瞧一瞧:
1. public void start() {
2. // Make sure any currently running dispatchers are stopped.
3. // Create the cache dispatcher and start it.
4. new CacheDispatcher(mCacheQueue, mNetworkQueue, mCache, mDelivery);
5. mCacheDispatcher.start();
6. // Create network dispatchers (and corresponding threads) up to the pool size.
7. for (int i = 0; i < mDispatchers.length; i++) {
8. new NetworkDispatcher(mNetworkQueue, mNetwork,
9. mCache, mDelivery);
10. mDispatchers[i] = networkDispatcher;
11. networkDispatcher.start();
12. }
13. }
这里先是创建了一个CacheDispatcher的实例,然后调用了它的start()方法,接着在一个for循环里去创建NetworkDispatcher的实例,并分别调用它们的start()方法。这里的CacheDispatcher和NetworkDispatcher都是继承自Thread的,而默认情况下for循环会执行四次,也就是说当调用了Volley.newRequestQueue(context)之后,就会有五个线程一直在后台运行,不断等待网络请求的到来,
其中
CacheDispatcher是缓存线程,NetworkDispatcher是网络请求线程。
得到了RequestQueue之后,我们只需要构建出相应的Request,然后调用RequestQueue的add()方法将Request传入就可以完成网络请求操作了,那么不用说,add()方法的内部肯定有着非常复杂的逻辑,我们来一起看一下:
1. public <T> Request<T> add(Request<T> request) {
2. // Tag the request as belonging to this queue and add it to the set of current requests.
3. this);
4. synchronized (mCurrentRequests) {
5. mCurrentRequests.add(request);
6. }
7. // Process requests in the order they are added.
8. request.setSequence(getSequenceNumber());
9. "add-to-queue");
10. // If the request is uncacheable, skip the cache queue and go straight to the network.
11. if (!request.shouldCache()) {
12. mNetworkQueue.add(request);
13. return request;
14. }
15. // Insert request into stage if there's already a request with the same cache key in flight.
16. synchronized (mWaitingRequests) {
17. String cacheKey = request.getCacheKey();
18. if (mWaitingRequests.containsKey(cacheKey)) {
19. // There is already a request in flight. Queue up.
20. Queue<Request<?>> stagedRequests = mWaitingRequests.get(cacheKey);
21. if (stagedRequests == null) {
22. new LinkedList<Request<?>>();
23. }
24. stagedRequests.add(request);
25. mWaitingRequests.put(cacheKey, stagedRequests);
26. if (VolleyLog.DEBUG) {
27. "Request for cacheKey=%s is in flight, putting on hold.", cacheKey);
28. }
29. else {
30. // Insert 'null' queue for this cacheKey, indicating there is now a request in
31. // flight.
32. null);
33. mCacheQueue.add(request);
34. }
35. return request;
36. }
37. }
可以看到,在第11行的时候会判断当前的请求是否可以缓存,如果不能缓存则在第12行直接将这条请求加入网络请求队列,可以缓存的话则在第33行将这条请求加入缓存队列。在默认情况下,每条请求都是可以缓存的,当然我们也可以调用Request的setShouldCache(false)方法来改变这一默认行为。
OK,那么既然默认每条请求都是可以缓存的,自然就被添加到了缓存队列中,于是一直在后台等待的缓存线程就要开始运行起来了,我们看下CacheDispatcher中的run()方法,代码如下所示:
1. public class CacheDispatcher extends Thread {
2.
3. ……
4.
5. @Override
6. public void run() {
7. if (DEBUG) VolleyLog.v("start new dispatcher");
8. Process.setThreadPriority(Process.THREAD_PRIORITY_BACKGROUND);
9. // Make a blocking call to initialize the cache.
10. mCache.initialize();
11. while (true) {
12. try {
13. // Get a request from the cache triage queue, blocking until
14. // at least one is available.
15. final Request<?> request = mCacheQueue.take();
16. "cache-queue-take");
17. // If the request has been canceled, don't bother dispatching it.
18. if (request.isCanceled()) {
19. "cache-discard-canceled");
20. continue;
21. }
22. // Attempt to retrieve this item from cache.
23. Cache.Entry entry = mCache.get(request.getCacheKey());
24. if (entry == null) {
25. "cache-miss");
26. // Cache miss; send off to the network dispatcher.
27. mNetworkQueue.put(request);
28. continue;
29. }
30. // If it is completely expired, just send it to the network.
31. if (entry.isExpired()) {
32. "cache-hit-expired");
33. request.setCacheEntry(entry);
34. mNetworkQueue.put(request);
35. continue;
36. }
37. // We have a cache hit; parse its data for delivery back to the request.
38. "cache-hit");
39. Response<?> response = request.parseNetworkResponse(
40. new NetworkResponse(entry.data, entry.responseHeaders));
41. "cache-hit-parsed");
42. if (!entry.refreshNeeded()) {
43. // Completely unexpired cache hit. Just deliver the response.
44. mDelivery.postResponse(request, response);
45. else {
46. // Soft-expired cache hit. We can deliver the cached response,
47. // but we need to also send the request to the network for
48. // refreshing.
49. "cache-hit-refresh-needed");
50. request.setCacheEntry(entry);
51. // Mark the response as intermediate.
52. true;
53. // Post the intermediate response back to the user and have
54. // the delivery then forward the request along to the network.
55. new Runnable() {
56. @Override
57. public void run() {
58. try {
59. mNetworkQueue.put(request);
60. catch (InterruptedException e) {
61. // Not much we can do about this.
62. }
63. }
64. });
65. }
66. catch (InterruptedException e) {
67. // We may have been interrupted because it was time to quit.
68. if (mQuit) {
69. return;
70. }
71. continue;
72. }
73. }
74. }
75. }
代码有点长,我们只挑重点看。首先在11行可以看到一个while(true)循环,说明缓存线程始终是在运行的,接着在第23行会尝试从缓存当中取出响应结果,如何为空的话则把这条请求加入到网络请求队列中,如果不为空的话再判断该缓存是否已过期,如果已经过期了则同样把这条请求加入到网络请求队列中,否则就认为不需要重发网络请求,直接使用缓存中的数据即可。之后会在第39行调用Request的
parseNetworkResponse()方法来对数据进行解析,再往后就是将解析出来的数据进行回调了,这部分代码我们先跳过,因为它的逻辑和NetworkDispatcher后半部分的逻辑是基本相同的,那么我们等下合并在一起看就好了,先来看一下NetworkDispatcher中是怎么处理网络请求队列的,代码如下所示:
1. public class NetworkDispatcher extends Thread {
2. ……
3. @Override
4. public void run() {
5. Process.setThreadPriority(Process.THREAD_PRIORITY_BACKGROUND);
6. Request<?> request;
7. while (true) {
8. try {
9. // Take a request from the queue.
10. request = mQueue.take();
11. catch (InterruptedException e) {
12. // We may have been interrupted because it was time to quit.
13. if (mQuit) {
14. return;
15. }
16. continue;
17. }
18. try {
19. "network-queue-take");
20. // If the request was cancelled already, do not perform the
21. // network request.
22. if (request.isCanceled()) {
23. "network-discard-cancelled");
24. continue;
25. }
26. addTrafficStatsTag(request);
27. // Perform the network request.
28. NetworkResponse networkResponse = mNetwork.performRequest(request);
29. "network-http-complete");
30. // If the server returned 304 AND we delivered a response already,
31. // we're done -- don't deliver a second identical response.
32. if (networkResponse.notModified && request.hasHadResponseDelivered()) {
33. "not-modified");
34. continue;
35. }
36. // Parse the response here on the worker thread.
37. Response<?> response = request.parseNetworkResponse(networkResponse);
38. "network-parse-complete");
39. // Write to cache if applicable.
40. // TODO: Only update cache metadata instead of entire record for 304s.
41. if (request.shouldCache() && response.cacheEntry != null) {
42. mCache.put(request.getCacheKey(), response.cacheEntry);
43. "network-cache-written");
44. }
45. // Post the response back.
46. request.markDelivered();
47. mDelivery.postResponse(request, response);
48. catch (VolleyError volleyError) {
49. parseAndDeliverNetworkError(request, volleyError);
50. catch (Exception e) {
51. "Unhandled exception %s", e.toString());
52. new VolleyError(e));
53. }
54. }
55. }
56. }
同样地,在第7行我们看到了类似的while(true)循环,说明网络请求线程也是在不断运行的。在第28行的时候会调用Network的performRequest()方法来去发送网络请求,而Network是一个接口,这里具体的实现是BasicNetwork,我们来看下它的
performRequest()方法,如下所示:
1. public class BasicNetwork implements Network {
2. ……
3. @Override
4. public NetworkResponse performRequest(Request<?> request) throws VolleyError {
5. long requestStart = SystemClock.elapsedRealtime();
6. while (true) {
7. null;
8. byte[] responseContents = null;
9. new HashMap<String, String>();
10. try {
11. // Gather headers.
12. new HashMap<String, String>();
13. addCacheHeaders(headers, request.getCacheEntry());
14. httpResponse = mHttpStack.performRequest(request, headers);
15. StatusLine statusLine = httpResponse.getStatusLine();
16. int statusCode = statusLine.getStatusCode();
17. responseHeaders = convertHeaders(httpResponse.getAllHeaders());
18. // Handle cache validation.
19. if (statusCode == HttpStatus.SC_NOT_MODIFIED) {
20. return new NetworkResponse(HttpStatus.SC_NOT_MODIFIED,
21. null ? null : request.getCacheEntry().data,
22. true);
23. }
24. // Some responses such as 204s do not have content. We must check.
25. if (httpResponse.getEntity() != null) {
26. responseContents = entityToBytes(httpResponse.getEntity());
27. else {
28. // Add 0 byte response as a way of honestly representing a
29. // no-content request.
30. new byte[0];
31. }
32. // if the request is slow, log it.
33. long requestLifetime = SystemClock.elapsedRealtime() - requestStart;
34. logSlowRequests(requestLifetime, request, responseContents, statusLine);
35. if (statusCode < 200 || statusCode > 299) {
36. throw new IOException();
37. }
38. return new NetworkResponse(statusCode, responseContents, responseHeaders, false);
39. catch (Exception e) {
40. ……
41. }
42. }
43. }
44. }
这段方法中大多都是一些网络请求细节方面的东西,我们并不需要太多关心,需要注意的是在第14行调用了HttpStack的performRequest()方法,这里的HttpStack就是在一开始调用newRequestQueue()方法是创建的实例,默认情况下如果系统版本号大于9就创建的HurlStack对象,否则创建HttpClientStack对象。前面已经说过,这两个对象的内部实际就是分别使用HttpURLConnection和HttpClient来发送网络请求的,我们就不再跟进去阅读了,之后会将服务器返回的数据组装成一个NetworkResponse对象进行返回。
在NetworkDispatcher中收到了NetworkResponse这个返回值后又会调用Request的parseNetworkResponse()方法来解析NetworkResponse中的数据,以及将数据写入到缓存,这个方法的实现是交给Request的子类来完成的,因为不同种类的Request解析的方式也肯定不同。还记得我们在上一篇文章中学习的自定义Request的方式吗?其中parseNetworkResponse()这个方法就是必须要重写的。
在解析完了NetworkResponse中的数据之后,又会调用ExecutorDelivery的postResponse()方法来回调解析出的数据,代码如下所示:
1. public void postResponse(Request<?> request, Response<?> response, Runnable runnable) {
2. request.markDelivered();
3. "post-response");
4. new ResponseDeliveryRunnable(request, response, runnable));
5. }
其中,在mResponsePoster的execute()方法中传入了一个ResponseDeliveryRunnable对象,就可以保证该对象中的run()方法就是在主线程当中运行的了,我们看下run()方法中的代码是什么样的:
1. private class ResponseDeliveryRunnable implements Runnable {
2. private final Request mRequest;
3. private final Response mResponse;
4. private final Runnable mRunnable;
5.
6. public ResponseDeliveryRunnable(Request request, Response response, Runnable runnable) {
7. mRequest = request;
8. mResponse = response;
9. mRunnable = runnable;
10. }
11.
12. @SuppressWarnings("unchecked")
13. @Override
14. public void run() {
15. // If this request has canceled, finish it and don't deliver.
16. if (mRequest.isCanceled()) {
17. "canceled-at-delivery");
18. return;
19. }
20. // Deliver a normal response or error, depending.
21. if (mResponse.isSuccess()) {
22. mRequest.deliverResponse(mResponse.result);
23. else {
24. mRequest.deliverError(mResponse.error);
25. }
26. // If this is an intermediate response, add a marker, otherwise we're done
27. // and the request can be finished.
28. if (mResponse.intermediate) {
29. "intermediate-response");
30. else {
31. "done");
32. }
33. // If we have been provided a post-delivery runnable, run it.
34. if (mRunnable != null) {
35. mRunnable.run();
36. }
37. }
38. }
代码虽然不多,但我们并不需要行行阅读,抓住重点看即可。其中在第22行调用了Request的deliverResponse()方法,有没有感觉很熟悉?没错,这个就是我们在自定义Request时需要重写的另外一个方法,每一条网络请求的响应都是回调到这个方法中,最后我们再在这个方法中将响应的数据回调到Response.Listener的onResponse()方法中就可以了。
好了,到这里我们就把Volley的完整执行流程全部梳理了一遍,你是不是已经感觉已经很清晰了呢?对了,还记得在文章一开始的那张流程图吗,刚才还不能理解,现在我们再来重新看下这张图:
其中蓝色部分代表主线程,绿色部分代表缓存线程,橙色部分代表网络线程。我们在主线程中调用RequestQueue的add()方法来添加一条网络请求,这条请求会先被加入到缓存队列当中,如果发现可以找到相应的缓存结果就直接读取缓存并解析,然后回调给主线程。如果在缓存中没有找到结果,则将这条请求加入到网络请求队列中,然后处理发送HTTP请求,解析响应结果,写入缓存,并回调主线程。
怎么样,是不是感觉现在理解这张图已经变得轻松简单了?好了,到此为止我们就把Volley的用法和源码全部学习完了,相信你已经对Volley非常熟悉并可以将它应用到实际项目当中了,那么Volley完全解析系列的文章到此结束,感谢大家有耐心看到最后。