Question

我想从网址播放一个只有一段时间才有效的音频流。由于缓冲机制（在第二次填充缓冲区时url将会死亡），因此使用内置的stagefright流功能不能很好地发挥作用，所以我实现了一个代理流，类似于在npr app中做了什么

http://code.google.com/p/npr-android-app/source/browse/trunk/Npr/src/org/npr/android/news/StreamProxy.java

这实际上非常好用，但有一个例外，任何搜索调用都会有效地破坏代理流。我很难确定搜寻在stagefright中的确切运作方式。每当我寻求时，我都会收到

的消息

01-12 13:35:57.201: ERROR/(4870): Connection reset by peer
01-12 13:35:57.201: ERROR/(4870): java.net.SocketException: Connection reset by peer
01-12 13:35:57.201: ERROR/(4870): at org.apache.harmony.luni.platform.OSNetworkSystem.writeSocketImpl(Native Method)
01-12 13:35:57.201: ERROR/(4870):     at org.apache.harmony.luni.platform.OSNetworkSystem.write(OSNetworkSystem.java:723)
01-12 13:35:57.201: ERROR/(4870):     at org.apache.harmony.luni.net.PlainSocketImpl.write(PlainSocketImpl.java:578)
01-12 13:35:57.201: ERROR/(4870):     at org.apache.harmony.luni.net.SocketOutputStream.write(SocketOutputStream.java:59)
01-12 13:35:57.201: ERROR/(4870):     at com.soundcloud.utils.StreamProxy.processRequest(StreamProxy.java:209)

然后暂停几秒钟，之后哪个stagefright重试连接到同一个url，并且通常会抛出一个错误（我会想象因为代理流没有被重置）。另一个潜在的问题是，似乎代理流将始终线性地读取数据源：

while (isRunning && (readBytes = data.read(buff, 0, buff.length)) != -1)

我只是猜测，但我认为为了支持搜索，代理必须能够在从缓冲区读取时提供偏移量。有没有办法找出套接字客户端请求的偏移量（预期的搜索位置）？

我对套接字的体验有限。有没有人对此有实施建议？

Answer 1

当您寻找或跳过或连接丢失且MediaPlayer不断重新连接到代理服务器时，您必须在从客户端获取请求和范围（int）后将此响应发送到状态206。

String headers += "HTTP/1.1 206 OK\r\n";
headers += "Content-Type: audio/mpeg\r\n";
headers += "Accept-Ranges: bytes\r\n";
headers += "Content-Length: " + (fileSize-range) + "\r\n";
headers += "Content-Range: bytes "+range + "-" + fileSize + "/*\r\n";
headers += "\r\n";

当您收到来自MediaPlayer的请求时，HTTP标头中不包含Range，那么它正在请求新的流文件，在这种情况下，您的响应标头应如下所示：

String headers = "HTTP/1.1 200 OK\r\n";
headers += "Content-Type: audio/mpeg\r\n";
headers += "Accept-Ranges: bytes\r\n";
headers += "Content-Length: " + fileSize + "\r\n";
headers += "\r\n";

享受！

Answer 2

MediaPlayer正在发出HTTP范围请求，指定播放器期望的字节偏移量。您可以从提交给代理服务器的请求的标头中读取这些范围值。然后，您可以打开将这些范围发送回MediaPlayer。

Answer 3

不确定你是否想过这个，但我需要做的就是将原始请求中的标题传递给代理请求：

  private HttpResponse download(String url, Header[] headers) {
DefaultHttpClient seed = new DefaultHttpClient();
SchemeRegistry registry = new SchemeRegistry();
registry.register(
        new Scheme("http", PlainSocketFactory.getSocketFactory(), 80));
SingleClientConnManager mgr = new MyClientConnManager(seed.getParams(),
    registry);
DefaultHttpClient http = new DefaultHttpClient(mgr, seed.getParams());
HttpGet method = new HttpGet(url);
for (Header header : headers) {
    method.addHeader(header);
}
HttpResponse response = null;
try {
  Log.d(getClass().getName(), "starting download");
  response = http.execute(method);
  Log.d(getClass().getName(), "downloaded");

}catch(java.net.UnknownHostException e)
{
    Intent i=new Intent("org.prx.errorInStream");
    mContext.sendBroadcast(i);
}
catch (ClientProtocolException e) {
  Log.e(getClass().getName(), "Error downloading", e);
} catch (IOException e) {
  Log.e(getClass().getName(), "Error downloading", e);
}
return response;

}

private void processRequest(HttpRequest request, Socket client)
  throws IllegalStateException, IOException {
if (request == null) {
  return;
}
Log.d(getClass().getName(), "processing");
String url = request.getRequestLine().getUri();

HttpResponse realResponse = download(url, request.getAllHeaders());

if (realResponse == null) {
  return;
}

...

}

Answer 4

如@Roberto所述，您必须修改方法download和processRequest才能绕过HTTP请求标头。但实际上要完成它，或者至少在我的情况下，我必须进一步修改方法private HttpRequest readRequest(Socket client)来解析socket中的标题，如下所示。

private HttpRequest readRequest(Socket client) {
        HttpRequest request = null;
        InputStream is;
        String firstLine;
        List<Header> headers = new ArrayList<Header>();
        try {
            is = client.getInputStream();
            BufferedReader reader = new BufferedReader(
                    new InputStreamReader(is), 8192);
            firstLine = reader.readLine();
            /**
             * read HTTP request headers from socket
             * */
            String line;
            while (((line = reader.readLine()) != null) && !(line.equals(""))) { // HTTP header ends at ""
                try {
                    StringTokenizer st = new StringTokenizer(line, ":");
                    String h = st.nextToken();
                    String v = st.nextToken();
                    Header header = new BasicHeader(h, v);
                    headers.add(header);
                } catch (NoSuchElementException e) {
                }
            }
        } catch (IOException e) {
            Log.e(LOG_TAG, "Error parsing request", e);
            return request;
        }

        if (firstLine == null) {
            Log.i(LOG_TAG, "Proxy client closed connection without a request.");
            return request;
        }

        /**
         * retrieve REAL url
         * */
        StringTokenizer st = new StringTokenizer(firstLine);
        String method = st.nextToken();
        String uri = st.nextToken();
        Log.d(LOG_TAG, uri);
        String realUri = uri.substring(1);
        Log.d(LOG_TAG, realUri);
        request = new BasicHttpRequest(method, realUri);
        /**
         * add headers back to HTTP request
         * */
        for (Header header : headers) {
            request.addHeader(header);
        }
        return request;
    }

android寻求代理流和stagefright播放器的问题

4 个答案: