为什么BufferedInputStream一次读取最多2048个字节？

Question

我正在尝试使用我的Android应用程序HttpUrlConnection中的网络读取文件AsyncTask.

虽然我注意到文件下载工作的速度比它应该有的慢一点,因为我使用的网络速度更快.

所以我检查并发现该BufferedInputStream对象一次最多只能读取2048个字节.没有我设置的缓冲区大小.甚至内部默认缓冲区大小为BufferedInputStream8192字节.

我在这里添加我的代码以供参考.

private class DownloadFileTask extends AsyncTask<String, Integer, String> {

  @Override
  protected String doInBackground(String... params) {
    HttpURLConnection connection = null;
    BufferedInputStream input = null;
    OutputStream output = null;
    int lengthOfFile;
    int totalBytesDownloaded = 0;
    int count;
    final int bufferSize = 8 * 1024; // 8KB
    try {
      // Create the URL
      URL url = new URL(params[0]);
      // Open connection
      connection = (HttpURLConnection) url.openConnection();
      // Get the file length
      lengthOfFile = connection.getContentLength();
      // Input stream to read file - with bufferSize buffer
      input = new BufferedInputStream(connection.getInputStream(), bufferSize);

      if (isCancelled()) {
        return null;
      }
      // Output stream to write file
      File parentFile = TestUtil.getStorageDir(getApplicationContext(),Constants.EXTRA_DIRECTORY_NAME_TEST_MEDIA);
      File file = new File(parentFile.getAbsolutePath() + "/" + "zipFile");

      if (!file.exists()) {
        file.getParentFile().mkdirs();
      }

      // Create the file o/p stream
      output = new FileOutputStream(file.getAbsolutePath());

      // Create the buffer o/p stream for performance
      BufferedOutputStream bos = new BufferedOutputStream(output, bufferSize);

      // Buffer
      byte data[] = new byte[bufferSize];
      while ((count = input.read(data, 0, bufferSize)) != -1 && !isCancelled()) {
        // Increase the total bytes downloaded
        totalBytesDownloaded += count;
        Log.d("DEBUG_LOG","total bytes read : " + count + " buffer size : " + data.length);

        // Write the data to the o/p buffer
        bos.write(data, 0, count);
      }

      // Publish update again since the loop may have skipped the last publish update
      publishProgress(totalBytesDownloaded, lengthOfFile);

      // Flush the o/p stream
      output.flush();
      return file.getAbsolutePath();
    } catch (SocketException | SocketTimeoutException e) {
      handler.sendEmptyMessage(Constants.CASE_INTERNET_FAILURE);
      Log.e("DEBUG_LOG", e);
    } catch (IOException e) {
      Log.e("DEBUG_LOG","Error: " + e.getMessage());
    } finally {
      // closing streams
      if (output != null) {
        try {
          output.close();
        } catch (IOException e) {
          AMLog.e(e);
        }
      }
      if (input != null) {
        try {
          input.close();
        } catch (IOException e) {
          AMLog.e(e);
        }
      }
      if (connection != null) {
        connection.disconnect();
      }
    }
    return null;
  }

  @Override
  protected void onProgressUpdate(Integer... values) {
    int percentage = (values[0] * 100) / values[1] ;
    textDownloadSizeMb.setText(String.format(getString(R.string.label_download_mb), String.valueOf(values[0]), String.valueOf(values[1])));
    textDownloadPercent.setText(String.format(Locale.getDefault(), "%s%s", percentage, " %"));
    progressBar.setMax(values[1]);
    progressBar.setProgress(values[0]);
  }

  @Override
  protected void onPostExecute(String value) {
  }
}

这是日志的一部分,显示读取的数据从不大于2048字节,而缓冲区大小为8192字节.

total bytes read : 1748 buffer size : 8192
total bytes read : 2048 buffer size : 8192
total bytes read : 2048 buffer size : 8192
total bytes read : 2048 buffer size : 8192
total bytes read : 1988 buffer size : 8192

我尝试过多个缓冲区大小超过2048但似乎没有任何改变读取速率.

这背后的原因是什么？我可以更改它以获取指定缓冲区大小的数据吗？

Answer 1

由于它的发送方式,因此可以随时读取所有内容.在这方面你没有什么可以做的,除了可能不经常阅读;-)没有特别的理由一次想要8192个字节:在正确编写的读取循环中你不应该关心它是一个字节还是兆字节,或者无论你的缓冲区大小是多少 合同read()仅仅是它传输'至少一个字节',除非发生异常或流的结束.