Android流媒体的实现

来源：互联网发布：网络机柜为什么用u 编辑：程序博客网时间：2024/06/03 17:41

在mps的setDataSource中，根据url判断应该生成那种dataSource，然后获取默认的player，即nuplayer。进入NuPlayerDriver的setDataSource，

再进入void NuPlayer::setDataSourceAsync(const sp<IStreamSource> &source) {

#ifdef MTK_AOSP_ENHANCEMENT

mIsStreamSource = true;

#endif

sp<AMessage> msg = new AMessage(kWhatSetDataSource, id());

sp<AMessage> notify = new AMessage(kWhatSourceNotify, id());

msg->setObject("source", new StreamingSource(notify, source));

msg->post();

}

到了NuPlayer异步消息的世界，进入setDataSourceAsync

首先判断是什么流媒体

if (!strncasecmp(url, "http://", 7)

|| !strncasecmp(url, "https://", 8)) {

mDataSourceType = SOURCE_Http;

ALOGI("Is http Streaming");

生成一个sp<GenericSource> genericSource =

new GenericSource(notify, mUIDValid, mUID);

然后触发了kWhatSetDataSource

进入NuPlayerDriver的

void NuPlayerDriver::notifySetDataSourceCompleted(status_t err) {

Mutex::Autolock autoLock(mLock);

CHECK_EQ(mState, STATE_SET_DATASOURCE_PENDING);

mAsyncResult = err;

mState = (err == OK) ? STATE_UNPREPARED : STATE_IDLE;

#ifdef MTK_AOSP_ENHANCEMENT

ALOGD("after notifySetDataSourceCompleted mState=%d", mState);

#endif

mCondition.broadcast();

}

注意，mCondition是在MPS调用NuPlayerDriver的API的时候就已经wait了的

status_t NuPlayerDriver::setDataSource(int fd, int64_t offset, int64_t length) {

ALOGV("setDataSource(%p) file(%d)", this, fd);

Mutex::Autolock autoLock(mLock);

if (mState != STATE_IDLE) {

return INVALID_OPERATION;

}

mState = STATE_SET_DATASOURCE_PENDING;

mPlayer->setDataSourceAsync(fd, offset, length);

while (mState == STATE_SET_DATASOURCE_PENDING) {

mCondition.wait(mLock);

}

return mAsyncResult;

}

好，下面该prepare了

status_t NuPlayerDriver::prepare_l() {

switch (mState) {

case STATE_UNPREPARED:

mState = STATE_PREPARING;

// Make sure we're not posting any notifications, success or

// failure information is only communicated through our result

// code.

mIsAsyncPrepare = false;

mPlayer->prepareAsync();

while (mState == STATE_PREPARING) {

mCondition.wait(mLock);

}

return (mState == STATE_PREPARED) ? OK : UNKNOWN_ERROR;

case STATE_STOPPED:

// this is really just paused. handle as seek to start

mAtEOS = false;

mState = STATE_STOPPED_AND_PREPARING;

mIsAsyncPrepare = false;

mPlayer->seekToAsync(0, true /* needNotify */);

while (mState == STATE_STOPPED_AND_PREPARING) {

mCondition.wait(mLock);

}

return (mState == STATE_STOPPED_AND_PREPARED) ? OK : UNKNOWN_ERROR;

default:

return INVALID_OPERATION;

};

}

进到NuPlayer的消息处理机制

case kWhatPrepare:

{

#ifdef MTK_AOSP_ENHANCEMENT

ATRACE_ASYNC_BEGIN("Prepare",mPlayerCnt);

ALOGD("kWhatPrepare, source type = %d", (int)mDataSourceType);

if (mPrepare == PREPARING)

break;

mPrepare = PREPARING;

if (mSource == NULL) {

ALOGW("prepare error: source is not ready");

finishPrepare(UNKNOWN_ERROR);

break;

}

#endif

mSource->prepareAsync();

break;

}

这个mSource是GenericSource

进入了onPrepareAsync

此时去create 一个dataSource

mDataSource = DataSource::CreateFromURI(

mHTTPService, uri, &mUriHeaders, &mContentType,

static_cast<HTTPBase *>(mHttpSource.get()));

此时就涉及到两个Source，一个是从网络获取数据的Http，一个是cache在本地的流媒体数据

Http的Source被这样创建

mHttpSource = DataSource::CreateMediaHTTP(mHTTPService);

mHttpSource继承于DataSource

第一个Source是聚合在NuCachedSource2中的一个Source

cache的Source被这样创建，本质上也是一个DataSource，因为NuCachedSource2是dataSource的子类

source = new NuCachedSource2(

httpSource,

cacheConfig.isEmpty() ? NULL : cacheConfig.string(),

disconnectAtHighwatermark);

然后进入initFromDataSource，也就是去选择extractor的类型，这个需要一些数据来判断是哪种流媒体格式

extractor = MediaExtractor::Create(mDataSource,

mSniffedMIME.empty() ? NULL: mSniffedMIME.c_str());

那些初始下载的数据是什么时候下的呢？在创建NuCachedSource2的时候就开始了

这是另外一个线程在做的事情

ssize_t n = mSource->readAt(

mCacheOffset + mCache->totalSize(), page->mData, kPageSize);

一次性获取kPageSize大小的数据

此时是靠IMediaHTTPConnection来通过匿名共享内存和binder获取的

Bn的部分在MediaHTTPConnection.java (base\media\java\android\media)

而它使用的又是 HttpURLConnection，是现在libcore下，属于基础架构

回到Extractor，它会调用ssize_t NuCachedSource2::readAt(off64_t offset, void *data, size_t size) {

发kWhatRead调用onRead，然后调用readInternal，从mCache里面copy数据

mCache->copy(delta, data, avail);

有多少拷多少，data是输出参数，给Extractor

注意，判断是什么类型是需要一点数据，这点数据往往就是tmd的tts数据了

bool DataSource::sniff(

String8 *mimeType, float *confidence, sp<AMessage> *meta) {

策略是所有Sniff，挨个闻一遍，哪个confidence高用谁的，Sniff是Extractor的一个副产品

0 0