如何在超时之前中断Jsoup get线程？

Question

I have this problem: I am getting some data in an android asynctask from multiple urls using Jsoup.connect(url).get() and then showing it in an Activity. The problem is that sometimes i need to interrupt the threads and free the processor before getting the jsoup result or catching a timeout exception.

The important part of the code code looks like this:

    @Override
    protected Void doInBackground(String... urls) {
            int cpuCount = Runtime.getRuntime().availableProcessors();

            BlockingQueue<Runnable> queue = new LinkedBlockingQueue<>();
            ThreadPoolExecutor threadPoolExecutor = new ThreadPoolExecutor(
                    cpuCount,
                    cpuCount,
                    30,
                    TimeUnit.SECONDS,
                    queue);

            final CountDownLatch countDownLatch = new CountDownLatch(urls.length);

            for (final String url : urls) {
                threadPoolExecutor.execute(new Runnable() {
                    @Override
                    public void run() {
                        String html = null;
                        try {
                            html = Jsoup.connect(url).get().outerHtml();
                        } catch (Exception ignored) {
                        }

                        if (html != null) {
                            publishProgress(videoStream);
                        }
                        countDownLatch.countDown();
                    }
                });
            }
        countDownLatch.await();
        return null;
    }

    @Override
    protected void onProgressUpdate(String... result) {
        updateUiStuff(result[0]);
    }

The reason i am using a Thread pool executor is that it does the connections in multiple threads in parallel, instead of waiting for one result and the opening a second connection and so on.

I tried doing cancel(true); OR threadPoolExecutor.shutDown(); and threadPoolExecutor.shutDownNow(); on a button, which was set in onPreExecute but it did not stop the Runnable s. Even when i close the activity, the Runnable s continue to run until they timeout and until then i can not do anything on the network because the jsoup connections are still running. Please point me in the right direction. Many thanks and i apologize if this is a really stupid question but i could not find a similar one here.

In short: I need to stop the threads when the AsyncTask cancels.

Answer 1

ThreadPoolExecutor is a general purpose executor that needs some experience to configure. It's a gateway for those experienced programmers whose needs are not satisfied (after enough testing) by the executors returned by other factory methods in the Executors class. The Javadoc for one of the constructors of this class even documents it:

It may be more convenient to use one of the {@link Executors} factory methods instead of this general purpose constructor.

Are you convinced that you still want to use ThreadPoolExector ? Note that you can always provide to the methods of the Executors the ThreadFactory that creates the actual threads that run your async tasks. You do not have to go to ThreadPoolExecutor .

In your call above, you make every thread wait 30 seconds ( keepAliveTime ) which makes it wait idly before termination. Are you sure you want that?

You also use just Runnable instances as your tasks. Had you considered using Callable instances instead, you could have used submit method which returns a Future which can be canceled. This also facilitates the use of invokeAll() method that lets you pass in a collection of Callable s and query/cancel the returned Future s. You said above that you did cancel(true) , but note that that method is only available on a Future .

A reliable way to shut down an executor or executor service is:

• Shut it down to attempt orderly shutdown using the shutdown() method.
• Await its termination.
• Only if needed, shut it down now using the shutdownNow() method and if you did, handle the returned list of Runnable s properly.
• Handle the exceptions well (this needs to be emphasized). I am sure that Jsoup.connect throws various different exceptions and you should handle them correctly. In you case, sadly, you even name the exception variable ignored ! Please don't believe everything you read on the Internet ;).

After you have gotten some experience with ExecutorService this way, you can move on to CompletionService which provides Future s as the async tasks complete . This exploits parallelism rather better especially if the tasks consist of variable IO-bound activity.