srs/trunk/src/app/srs_app_srt_source.cpp
OSSRS-AI 3f2539d8fb
AI: SRT: Stop TS parsing in SrsSrtFormat after codec detection. v7.0.125 (#4569) (#4571)
Fix log flooding issue when processing SRT streams containing SCTE-35
PIDs or other unrecognized stream types.

The `SrsSrtFormat::on_srt_packet()` method continuously parses TS
packets throughout the entire stream lifetime. The TS parser logs
warnings for every unrecognized stream type (like SCTE-35) in the PMT,
causing log flooding.

However, `SrsFormat` is only used to detect audio/video codec
information. Once both codecs are detected, there's no need to continue
parsing TS packets.

Note: This fix mitigates the problem - there will still be some warning
logs during the initial codec detection phase (typically 5-10 seconds),
but the continuous log flooding after codec detection is completely
eliminated.
2025-11-11 00:24:01 -05:00

1459 lines
40 KiB
C++

//
// Copyright (c) 2013-2025 The SRS Authors
//
// SPDX-License-Identifier: MIT
//
#include <srs_app_srt_source.hpp>
#include <algorithm>
using namespace std;
#include <srs_app_rtmp_source.hpp>
#include <srs_app_statistic.hpp>
#include <srs_core_autofree.hpp>
#include <srs_kernel_buffer.hpp>
#include <srs_kernel_flv.hpp>
#include <srs_kernel_pithy_print.hpp>
#include <srs_kernel_stream.hpp>
#include <srs_kernel_ts.hpp>
#include <srs_kernel_utility.hpp>
#include <srs_protocol_format.hpp>
#include <srs_protocol_raw_avc.hpp>
#include <srs_protocol_rtmp_stack.hpp>
// the time to cleanup source.
#define SRS_SRT_SOURCE_CLEANUP (3 * SRS_UTIME_SECONDS)
SrsSrtPacket::SrsSrtPacket()
{
shared_buffer_ = NULL;
actual_buffer_size_ = 0;
}
SrsSrtPacket::~SrsSrtPacket()
{
srs_freep(shared_buffer_);
}
char *SrsSrtPacket::wrap(int size)
{
// The buffer size is larger or equals to the size of packet.
actual_buffer_size_ = size;
// If the buffer is large enough, reuse it.
if (shared_buffer_ && shared_buffer_->size() >= size) {
return shared_buffer_->payload();
}
// Create a large enough message, with under-layer buffer.
srs_freep(shared_buffer_);
shared_buffer_ = new SrsMediaPacket();
char *buf = new char[size];
shared_buffer_->wrap(buf, size);
return shared_buffer_->payload();
}
char *SrsSrtPacket::wrap(char *data, int size)
{
char *buf = wrap(size);
memcpy(buf, data, size);
return buf;
}
char *SrsSrtPacket::wrap(SrsMediaPacket *msg)
{
// Generally, the wrap(msg) is used for RTMP to SRT, where the msg
// is not generated by SRT.
srs_freep(shared_buffer_);
// Copy from the new message.
shared_buffer_ = msg->copy();
// If we wrap a message, the size of packet equals to the message size.
actual_buffer_size_ = shared_buffer_->size();
return msg->payload();
}
SrsSrtPacket *SrsSrtPacket::copy()
{
SrsSrtPacket *cp = new SrsSrtPacket();
cp->shared_buffer_ = shared_buffer_ ? shared_buffer_->copy() : NULL;
cp->actual_buffer_size_ = actual_buffer_size_;
return cp;
}
char *SrsSrtPacket::data()
{
return shared_buffer_->payload();
}
int SrsSrtPacket::size()
{
return shared_buffer_->size();
}
ISrsSrtSourceManager::ISrsSrtSourceManager()
{
}
ISrsSrtSourceManager::~ISrsSrtSourceManager()
{
}
SrsSrtSourceManager::SrsSrtSourceManager()
{
lock_ = srs_mutex_new();
timer_ = new SrsHourGlass("sources", this, 1 * SRS_UTIME_SECONDS);
}
SrsSrtSourceManager::~SrsSrtSourceManager()
{
srs_mutex_destroy(lock_);
srs_freep(timer_);
}
srs_error_t SrsSrtSourceManager::initialize()
{
return setup_ticks();
}
srs_error_t SrsSrtSourceManager::setup_ticks()
{
srs_error_t err = srs_success;
if ((err = timer_->tick(1, 3 * SRS_UTIME_SECONDS)) != srs_success) {
return srs_error_wrap(err, "tick");
}
if ((err = timer_->start()) != srs_success) {
return srs_error_wrap(err, "timer");
}
return err;
}
srs_error_t SrsSrtSourceManager::notify(int event, srs_utime_t interval, srs_utime_t tick)
{
srs_error_t err = srs_success;
std::map<std::string, SrsSharedPtr<SrsSrtSource> >::iterator it;
for (it = pool_.begin(); it != pool_.end();) {
SrsSharedPtr<SrsSrtSource> &source = it->second;
// When source expired, remove it.
// @see https://github.com/ossrs/srs/issues/713
if (source->stream_is_dead()) {
SrsContextId cid = source->source_id();
if (cid.empty())
cid = source->pre_source_id();
srs_trace("SRT: cleanup die source, id=[%s], total=%d", cid.c_str(), (int)pool_.size());
pool_.erase(it++);
} else {
++it;
}
}
return err;
}
srs_error_t SrsSrtSourceManager::fetch_or_create(ISrsRequest *r, SrsSharedPtr<SrsSrtSource> &pps)
{
srs_error_t err = srs_success;
bool created = false;
// Should never invoke any function during the locking.
if (true) {
// Use lock to protect coroutine switch.
// @bug https://github.com/ossrs/srs/issues/1230
SrsLocker(&lock_);
string stream_url = r->get_stream_url();
std::map<std::string, SrsSharedPtr<SrsSrtSource> >::iterator it = pool_.find(stream_url);
if (it != pool_.end()) {
SrsSharedPtr<SrsSrtSource> source = it->second;
pps = source;
} else {
SrsSharedPtr<SrsSrtSource> source(new SrsSrtSource());
srs_trace("new srt source, stream_url=%s", stream_url.c_str());
pps = source;
pool_[stream_url] = source;
created = true;
}
}
// Initialize source.
if (created && (err = pps->initialize(r)) != srs_success) {
return srs_error_wrap(err, "init source %s", r->get_stream_url().c_str());
}
// we always update the request of resource,
// for origin auth is on, the token in request maybe invalid,
// and we only need to update the token of request, it's simple.
if (!created) {
pps->update_auth(r);
}
return err;
}
SrsSharedPtr<SrsSrtSource> SrsSrtSourceManager::fetch(ISrsRequest *r)
{
// Use lock to protect coroutine switch.
// @bug https://github.com/ossrs/srs/issues/1230
SrsLocker(&lock_);
string stream_url = r->get_stream_url();
std::map<std::string, SrsSharedPtr<SrsSrtSource> >::iterator it = pool_.find(stream_url);
SrsSharedPtr<SrsSrtSource> source;
if (it == pool_.end()) {
return source;
}
source = it->second;
return source;
}
SrsSrtSourceManager *_srs_srt_sources = NULL;
ISrsSrtConsumer::ISrsSrtConsumer()
{
}
ISrsSrtConsumer::~ISrsSrtConsumer()
{
}
SrsSrtConsumer::SrsSrtConsumer(ISrsSrtSource *s)
{
source_ = s;
should_update_source_id_ = false;
mw_wait_ = srs_cond_new();
mw_min_msgs_ = 0;
mw_waiting_ = false;
}
SrsSrtConsumer::~SrsSrtConsumer()
{
source_->on_consumer_destroy(this);
vector<SrsSrtPacket *>::iterator it;
for (it = queue_.begin(); it != queue_.end(); ++it) {
SrsSrtPacket *pkt = *it;
srs_freep(pkt);
}
srs_cond_destroy(mw_wait_);
}
void SrsSrtConsumer::update_source_id()
{
should_update_source_id_ = true;
}
srs_error_t SrsSrtConsumer::enqueue(SrsSrtPacket *packet)
{
srs_error_t err = srs_success;
queue_.push_back(packet);
if (mw_waiting_) {
if ((int)queue_.size() > mw_min_msgs_) {
srs_cond_signal(mw_wait_);
mw_waiting_ = false;
return err;
}
}
return err;
}
srs_error_t SrsSrtConsumer::dump_packet(SrsSrtPacket **ppkt)
{
srs_error_t err = srs_success;
if (should_update_source_id_) {
srs_trace("update source_id=%s/%s", source_->source_id().c_str(), source_->pre_source_id().c_str());
should_update_source_id_ = false;
}
// TODO: FIXME: Refine performance by ring buffer.
if (!queue_.empty()) {
*ppkt = queue_.front();
queue_.erase(queue_.begin());
}
return err;
}
void SrsSrtConsumer::wait(int nb_msgs, srs_utime_t timeout)
{
mw_min_msgs_ = nb_msgs;
// when duration ok, signal to flush.
if ((int)queue_.size() > mw_min_msgs_) {
return;
}
// the enqueue will notify this cond.
mw_waiting_ = true;
// use cond block wait for high performance mode.
srs_cond_timedwait(mw_wait_, timeout);
}
SrsSrtFrameBuilder::SrsSrtFrameBuilder(ISrsFrameTarget *target)
{
ts_ctx_ = new SrsTsContext();
sps_pps_change_ = false;
sps_ = "";
pps_ = "";
req_ = NULL;
frame_target_ = target;
video_streamid_ = 1;
audio_streamid_ = 2;
pp_audio_duration_ = new SrsAlonePithyPrint();
}
SrsSrtFrameBuilder::~SrsSrtFrameBuilder()
{
srs_freep(ts_ctx_);
srs_freep(req_);
srs_freep(pp_audio_duration_);
}
srs_error_t SrsSrtFrameBuilder::on_publish()
{
return srs_success;
}
srs_error_t SrsSrtFrameBuilder::on_srt_packet(SrsSrtPacket *pkt)
{
srs_error_t err = srs_success;
char *buf = pkt->data();
int nb_buf = pkt->size();
// use stream to parse ts packet.
int nb_packet = nb_buf / SRS_TS_PACKET_SIZE;
for (int i = 0; i < nb_packet; i++) {
char *p = buf + (i * SRS_TS_PACKET_SIZE);
SrsUniquePtr<SrsBuffer> stream(new SrsBuffer(p, SRS_TS_PACKET_SIZE));
// Process each ts packet. Note that the jitter of UDP may cause video glitch when packet loss or wrong seq. We
// don't handle it because SRT will, see tlpktdrop at https://ossrs.io/lts/en-us/docs/v7/doc/srt
if ((err = ts_ctx_->decode(stream.get(), this)) != srs_success) {
srs_warn("parse ts packet err=%s", srs_error_desc(err).c_str());
srs_freep(err);
continue;
}
}
return err;
}
void SrsSrtFrameBuilder::on_unpublish()
{
}
srs_error_t SrsSrtFrameBuilder::initialize(ISrsRequest *req)
{
srs_error_t err = srs_success;
// TODO: FIXME: check srt2rtmp enable in config.
req_ = req->copy();
return err;
}
srs_error_t SrsSrtFrameBuilder::on_ts_message(SrsTsMessage *msg)
{
srs_error_t err = srs_success;
// When the audio SID is private stream 1, we use common audio.
// @see https://github.com/ossrs/srs/issues/740
if (msg->channel_->apply_ == SrsTsPidApplyAudio && msg->sid_ == SrsTsPESStreamIdPrivateStream1) {
msg->sid_ = SrsTsPESStreamIdAudioCommon;
}
// when not audio/video, or not adts/annexb format, donot support.
if (msg->stream_number() != 0) {
return srs_error_new(ERROR_STREAM_CASTER_TS_ES, "ts: unsupported stream format, sid=%#x(%s-%d)",
msg->sid_, msg->is_audio() ? "A" : msg->is_video() ? "V"
: "N",
msg->stream_number());
}
// check supported codec
if (msg->channel_->stream_ != SrsTsStreamVideoH264 && msg->channel_->stream_ != SrsTsStreamVideoHEVC && msg->channel_->stream_ != SrsTsStreamAudioAAC) {
return srs_error_new(ERROR_STREAM_CASTER_TS_CODEC, "ts: unsupported stream codec=%d", msg->channel_->stream_);
}
// parse the stream.
SrsBuffer avs(msg->payload_->bytes(), msg->payload_->length());
// publish audio or video.
if (msg->channel_->stream_ == SrsTsStreamVideoH264) {
if ((err = on_ts_video_avc(msg, &avs)) != srs_success) {
return srs_error_wrap(err, "ts: consume video");
}
}
if (msg->channel_->stream_ == SrsTsStreamAudioAAC) {
if ((err = on_ts_audio(msg, &avs)) != srs_success) {
return srs_error_wrap(err, "ts: consume audio");
}
}
// TODO: FIXME: implements other codec?
if (msg->channel_->stream_ == SrsTsStreamVideoHEVC) {
if ((err = on_ts_video_hevc(msg, &avs)) != srs_success) {
return srs_error_wrap(err, "ts: consume hevc video");
}
}
return err;
}
srs_error_t SrsSrtFrameBuilder::on_ts_video_avc(SrsTsMessage *msg, SrsBuffer *avs)
{
srs_error_t err = srs_success;
vector<pair<char *, int> > ipb_frames;
SrsUniquePtr<SrsRawH264Stream> avc(new SrsRawH264Stream());
// send each frame.
while (!avs->empty()) {
char *frame = NULL;
int frame_size = 0;
if ((err = avc->annexb_demux(avs, &frame, &frame_size)) != srs_success) {
return srs_error_wrap(err, "demux annexb");
}
if (frame == NULL || frame_size == 0) {
continue;
}
// for sps
if (avc->is_sps(frame, frame_size)) {
std::string sps;
if ((err = avc->sps_demux(frame, frame_size, sps)) != srs_success) {
return srs_error_wrap(err, "demux sps");
}
if (!sps.empty() && sps_ != sps) {
sps_pps_change_ = true;
}
sps_ = sps;
continue;
}
// for pps
if (avc->is_pps(frame, frame_size)) {
std::string pps;
if ((err = avc->pps_demux(frame, frame_size, pps)) != srs_success) {
return srs_error_wrap(err, "demux pps");
}
if (!pps.empty() && pps_ != pps) {
sps_pps_change_ = true;
}
pps_ = pps;
continue;
}
ipb_frames.push_back(make_pair(frame, frame_size));
}
if ((err = check_sps_pps_change(msg)) != srs_success) {
return srs_error_wrap(err, "check sps pps");
}
return on_h264_frame(msg, ipb_frames);
}
srs_error_t SrsSrtFrameBuilder::check_sps_pps_change(SrsTsMessage *msg)
{
srs_error_t err = srs_success;
if (!sps_pps_change_) {
return err;
}
if (sps_.empty() || pps_.empty()) {
return srs_error_new(ERROR_SRT_TO_RTMP_EMPTY_SPS_PPS, "sps or pps empty");
}
// sps/pps changed, generate new video sh frame and dispatch it.
sps_pps_change_ = false;
// ts tbn to flv tbn.
uint32_t dts = (uint32_t)(msg->dts_ / 90);
std::string sh;
SrsUniquePtr<SrsRawH264Stream> avc(new SrsRawH264Stream());
if ((err = avc->mux_sequence_header(sps_, pps_, sh)) != srs_success) {
return srs_error_wrap(err, "mux sequence header");
}
// h264 packet to flv packet.
char *flv = NULL;
int nb_flv = 0;
if ((err = avc->mux_avc2flv(sh, SrsVideoAvcFrameTypeKeyFrame, SrsVideoAvcFrameTraitSequenceHeader, dts, dts, &flv, &nb_flv)) != srs_success) {
return srs_error_wrap(err, "avc to flv");
}
SrsMessageHeader header;
header.initialize_video(nb_flv, dts, video_streamid_);
SrsRtmpCommonMessage rtmp;
if ((err = rtmp.create(&header, flv, nb_flv)) != srs_success) {
return srs_error_wrap(err, "create rtmp");
}
SrsMediaPacket frame;
rtmp.to_msg(&frame);
if ((err = frame_target_->on_frame(&frame)) != srs_success) {
return srs_error_wrap(err, "srt to rtmp sps/pps");
}
return err;
}
srs_error_t SrsSrtFrameBuilder::on_h264_frame(SrsTsMessage *msg, vector<pair<char *, int> > &ipb_frames)
{
srs_error_t err = srs_success;
if (ipb_frames.empty()) {
return srs_error_new(ERROR_SRT_CONN, "empty frame");
}
bool is_keyframe = false;
// ts tbn to flv tbn.
uint32_t dts = (uint32_t)(msg->dts_ / 90);
uint32_t pts = (uint32_t)(msg->pts_ / 90);
int32_t cts = pts - dts;
int frame_size = 5; // 5bytes video tag header
for (size_t i = 0; i != ipb_frames.size(); ++i) {
// 4 bytes for nalu length.
frame_size += 4 + ipb_frames[i].second;
if (((SrsAvcNaluType)(ipb_frames[i].first[0] & 0x1f)) == SrsAvcNaluTypeIDR) {
is_keyframe = true;
}
}
SrsRtmpCommonMessage rtmp;
rtmp.header_.initialize_video(frame_size, dts, video_streamid_);
rtmp.create_payload(frame_size);
SrsBuffer payload(rtmp.payload(), rtmp.size());
// Write 5bytes video tag header.
if (is_keyframe) {
payload.write_1bytes(0x17); // type(4 bits): key frame; code(4bits): avc
} else {
payload.write_1bytes(0x27); // type(4 bits): inter frame; code(4bits): avc
}
payload.write_1bytes(0x01); // avc_type: nalu
payload.write_3bytes(cts); // composition time
// Write video nalus.
for (size_t i = 0; i != ipb_frames.size(); ++i) {
char *nal = ipb_frames[i].first;
int nal_size = ipb_frames[i].second;
// write 4 bytes of nalu length.
payload.write_4bytes(nal_size);
// write nalu
payload.write_bytes(nal, nal_size);
}
SrsMediaPacket frame;
rtmp.to_msg(&frame);
if ((err = frame_target_->on_frame(&frame)) != srs_success) {
return srs_error_wrap(err, "srt ts video to rtmp");
}
return err;
}
srs_error_t SrsSrtFrameBuilder::on_ts_video_hevc(SrsTsMessage *msg, SrsBuffer *avs)
{
srs_error_t err = srs_success;
vector<pair<char *, int> > ipb_frames;
SrsUniquePtr<SrsRawHEVCStream> hevc(new SrsRawHEVCStream());
std::vector<std::string> hevc_pps;
// send each frame.
while (!avs->empty()) {
char *frame = NULL;
int frame_size = 0;
if ((err = hevc->annexb_demux(avs, &frame, &frame_size)) != srs_success) {
return srs_error_wrap(err, "demux hevc annexb");
}
if (frame == NULL || frame_size == 0) {
continue;
}
// for vps
if (hevc->is_vps(frame, frame_size)) {
std::string vps;
if ((err = hevc->vps_demux(frame, frame_size, vps)) != srs_success) {
return srs_error_wrap(err, "demux vps");
}
if (!vps.empty() && hevc_vps_ != vps) {
vps_sps_pps_change_ = true;
}
hevc_vps_ = vps;
continue;
}
// for sps
if (hevc->is_sps(frame, frame_size)) {
std::string sps;
if ((err = hevc->sps_demux(frame, frame_size, sps)) != srs_success) {
return srs_error_wrap(err, "demux sps");
}
if (!sps.empty() && hevc_sps_ != sps) {
vps_sps_pps_change_ = true;
}
hevc_sps_ = sps;
continue;
}
// for pps
if (hevc->is_pps(frame, frame_size)) {
std::string pps;
if ((err = hevc->pps_demux(frame, frame_size, pps)) != srs_success) {
return srs_error_wrap(err, "demux pps");
}
if (!pps.empty()) {
vps_sps_pps_change_ = true;
}
hevc_pps.push_back(pps);
continue;
}
ipb_frames.push_back(make_pair(frame, frame_size));
}
if (!hevc_pps.empty()) {
hevc_pps_ = hevc_pps;
}
if ((err = check_vps_sps_pps_change(msg)) != srs_success) {
return srs_error_wrap(err, "check vps sps pps");
}
return on_hevc_frame(msg, ipb_frames);
}
srs_error_t SrsSrtFrameBuilder::check_vps_sps_pps_change(SrsTsMessage *msg)
{
srs_error_t err = srs_success;
if (!vps_sps_pps_change_) {
return err;
}
if (hevc_vps_.empty() || hevc_sps_.empty() || hevc_pps_.empty()) {
return err;
}
// vps/sps/pps changed, generate new video sh frame and dispatch it.
vps_sps_pps_change_ = false;
// ts tbn to flv tbn.
uint32_t dts = (uint32_t)(msg->dts_ / 90);
uint32_t pts = (uint32_t)(msg->pts_ / 90);
std::string sh;
SrsUniquePtr<SrsRawHEVCStream> hevc(new SrsRawHEVCStream());
if ((err = hevc->mux_sequence_header(hevc_vps_, hevc_sps_, hevc_pps_, sh)) != srs_success) {
return srs_error_wrap(err, "mux sequence header");
}
// h265 packet to flv packet.
char *flv = NULL;
int nb_flv = 0;
if ((err = hevc->mux_hevc2flv_enhanced(sh,
SrsVideoAvcFrameTypeKeyFrame,
SrsVideoHEVCFrameTraitPacketTypeSequenceStart,
dts,
pts,
&flv,
&nb_flv)) != srs_success) {
return srs_error_wrap(err, "hevc sh to flv");
}
SrsMessageHeader header;
header.initialize_video(nb_flv, dts, video_streamid_);
SrsRtmpCommonMessage rtmp;
if ((err = rtmp.create(&header, flv, nb_flv)) != srs_success) {
return srs_error_wrap(err, "create rtmp");
}
SrsMediaPacket frame;
rtmp.to_msg(&frame);
if ((err = frame_target_->on_frame(&frame)) != srs_success) {
return srs_error_wrap(err, "srt to rtmp vps/sps/pps");
}
return err;
}
srs_error_t SrsSrtFrameBuilder::on_hevc_frame(SrsTsMessage *msg, vector<pair<char *, int> > &ipb_frames)
{
srs_error_t err = srs_success;
if (ipb_frames.empty()) {
return err;
}
// ts tbn to flv tbn.
uint32_t dts = (uint32_t)(msg->dts_ / 90);
// for IDR frame, the frame is keyframe.
SrsVideoAvcFrameType frame_type = SrsVideoAvcFrameTypeInterFrame;
// 5bytes video tag header
int frame_size = 5;
for (size_t i = 0; i != ipb_frames.size(); ++i) {
// 4 bytes for nalu length.
frame_size += 4 + ipb_frames[i].second;
SrsHevcNaluType nalu_type = SrsHevcNaluTypeParse(ipb_frames[i].first[0]);
if (SrsIsIRAP(nalu_type)) {
frame_type = SrsVideoAvcFrameTypeKeyFrame;
}
}
SrsRtmpCommonMessage rtmp;
rtmp.header_.initialize_video(frame_size, dts, video_streamid_);
rtmp.create_payload(frame_size);
SrsBuffer payload(rtmp.payload(), rtmp.size());
// Write 5bytes video tag header.
// @see: https://veovera.org/docs/enhanced/enhanced-rtmp-v1.pdf, page 8
payload.write_1bytes(SRS_FLV_IS_EX_HEADER | (frame_type << 4) | SrsVideoHEVCFrameTraitPacketTypeCodedFramesX);
payload.write_4bytes(0x68766331); // 'h' 'v' 'c' '1'
// Write video nalus.
for (size_t i = 0; i != ipb_frames.size(); ++i) {
char *nal = ipb_frames[i].first;
int nal_size = ipb_frames[i].second;
// write 4 bytes of nalu length.
payload.write_4bytes(nal_size);
// write nalu
payload.write_bytes(nal, nal_size);
}
SrsMediaPacket frame;
rtmp.to_msg(&frame);
if ((err = frame_target_->on_frame(&frame)) != srs_success) {
return srs_error_wrap(err, "srt ts hevc video to rtmp");
}
return err;
}
srs_error_t SrsSrtFrameBuilder::on_ts_audio(SrsTsMessage *msg, SrsBuffer *avs)
{
srs_error_t err = srs_success;
SrsUniquePtr<SrsRawAacStream> aac(new SrsRawAacStream());
// ts tbn to flv tbn.
uint32_t pts = (uint32_t)(msg->pts_ / 90);
int frame_idx = 0;
int duration_ms = 0;
// send each frame.
while (!avs->empty()) {
char *frame = NULL;
int frame_size = 0;
SrsRawAacStreamCodec codec;
if ((err = aac->adts_demux(avs, &frame, &frame_size, codec)) != srs_success) {
return srs_error_wrap(err, "demux adts");
}
// ignore invalid frame,
// * atleast 1bytes for aac to decode the data.
if (frame_size <= 0) {
continue;
}
std::string sh;
if ((err = aac->mux_sequence_header(&codec, sh)) != srs_success) {
return srs_error_wrap(err, "mux sequence header");
}
if (!sh.empty() && sh != audio_sh_) {
audio_sh_ = sh;
audio_sh_change_ = true;
}
// May have more than one aac frame in PES packet, and shared same timestamp,
// so we must calculate each aac frame's timestamp.
int sample_rate = 44100;
switch (codec.sound_rate_) {
case SrsAudioSampleRate5512:
sample_rate = 5512;
break;
case SrsAudioSampleRate11025:
sample_rate = 11025;
break;
case SrsAudioSampleRate22050:
sample_rate = 22050;
break;
case SrsAudioSampleRate44100:
default:
sample_rate = 44100;
break;
}
uint32_t frame_pts = (double)pts + (frame_idx * (1024.0 * 1000.0 / sample_rate));
duration_ms += 1024.0 * 1000.0 / sample_rate;
++frame_idx;
if ((err = check_audio_sh_change(msg, frame_pts)) != srs_success) {
return srs_error_wrap(err, "audio sh");
}
if ((err = on_aac_frame(msg, frame_pts, frame, frame_size)) != srs_success) {
return srs_error_wrap(err, "audio frame");
}
}
pp_audio_duration_->elapse();
if ((duration_ms >= 200) && pp_audio_duration_->can_print()) {
// MPEG-TS always merge multi audio frame into one pes packet, may cause high latency and AV synchronization errors
// @see https://github.com/ossrs/srs/issues/3164
srs_warn("srt to rtmp, audio duration=%dms too large, audio frames=%d, may cause high latency and AV synchronization errors, "
"read https://ossrs.io/lts/en-us/docs/v7/doc/srt#ffmpeg-push-srt-stream",
duration_ms, frame_idx);
}
return err;
}
srs_error_t SrsSrtFrameBuilder::check_audio_sh_change(SrsTsMessage *msg, uint32_t pts)
{
srs_error_t err = srs_success;
if (!audio_sh_change_) {
return err;
}
// audio specific config changed, generate new audio sh and dispatch it.
audio_sh_change_ = false;
int rtmp_len = audio_sh_.size() + 2;
SrsRtmpCommonMessage rtmp;
rtmp.header_.initialize_audio(rtmp_len, pts, audio_streamid_);
rtmp.create_payload(rtmp_len);
SrsBuffer stream(rtmp.payload(), rtmp_len);
uint8_t aac_flag = (SrsAudioCodecIdAAC << 4) | (SrsAudioSampleRate44100 << 2) | (SrsAudioSampleBits16bit << 1) | SrsAudioChannelsStereo;
stream.write_1bytes(aac_flag);
stream.write_1bytes(0);
stream.write_bytes((char *)audio_sh_.data(), audio_sh_.size());
SrsMediaPacket frame;
rtmp.to_msg(&frame);
if ((err = frame_target_->on_frame(&frame)) != srs_success) {
return srs_error_wrap(err, "srt to rtmp audio sh");
}
return err;
}
srs_error_t SrsSrtFrameBuilder::on_aac_frame(SrsTsMessage *msg, uint32_t pts, char *data, int data_size)
{
srs_error_t err = srs_success;
int rtmp_len = data_size + 2 /* 2 bytes of flv audio tag header*/;
SrsRtmpCommonMessage rtmp;
rtmp.header_.initialize_audio(rtmp_len, pts, audio_streamid_);
rtmp.create_payload(rtmp_len);
SrsBuffer stream(rtmp.payload(), rtmp_len);
uint8_t aac_flag = (SrsAudioCodecIdAAC << 4) | (SrsAudioSampleRate44100 << 2) | (SrsAudioSampleBits16bit << 1) | SrsAudioChannelsStereo;
// Write 2bytes audio tag header.
stream.write_1bytes(aac_flag);
stream.write_1bytes(1);
// Write audio frame.
stream.write_bytes(data, data_size);
SrsMediaPacket frame;
rtmp.to_msg(&frame);
if ((err = frame_target_->on_frame(&frame)) != srs_success) {
return srs_error_wrap(err, "srt to rtmp audio sh");
}
return err;
}
ISrsSrtFormat::ISrsSrtFormat()
{
}
ISrsSrtFormat::~ISrsSrtFormat()
{
}
SrsSrtFormat::SrsSrtFormat()
{
req_ = NULL;
ts_ctx_ = new SrsTsContext();
format_ = new SrsRtmpFormat();
video_codec_reported_ = false;
audio_codec_reported_ = false;
nn_video_frames_ = 0;
nn_audio_frames_ = 0;
stat_ = _srs_stat;
}
SrsSrtFormat::~SrsSrtFormat()
{
req_ = NULL;
srs_freep(ts_ctx_);
srs_freep(format_);
stat_ = NULL;
}
srs_error_t SrsSrtFormat::initialize(ISrsRequest *req)
{
srs_error_t err = srs_success;
req_ = req;
return err;
}
srs_error_t SrsSrtFormat::on_srt_packet(SrsSrtPacket *pkt)
{
srs_error_t err = srs_success;
// Skip TS parsing if both video and audio codecs have been reported
// This avoids unnecessary TS decoding and log flooding from unrecognized stream types
if (video_codec_reported_ && audio_codec_reported_) {
return err;
}
char *buf = pkt->data();
int nb_buf = pkt->size();
// Parse TS packets to extract codec information
int nb_packet = nb_buf / SRS_TS_PACKET_SIZE;
for (int i = 0; i < nb_packet; i++) {
char *p = buf + (i * SRS_TS_PACKET_SIZE);
SrsUniquePtr<SrsBuffer> stream(new SrsBuffer(p, SRS_TS_PACKET_SIZE));
// Decode TS packet and call on_ts_message for each message
if ((err = ts_ctx_->decode(stream.get(), this)) != srs_success) {
// Ignore parse errors, just log and continue
srs_warn("srt format parse ts packet err=%s", srs_error_desc(err).c_str());
srs_freep(err);
continue;
}
}
return err;
}
void SrsSrtFormat::update_ts_message_stats(bool is_audio)
{
srs_error_t err = srs_success;
// Count TS messages for statistics.
if (is_audio) {
++nn_audio_frames_;
} else {
++nn_video_frames_;
}
// Update the stat for video frames, counting TS messages as frames.
if (nn_video_frames_ > 288) {
if ((err = stat_->on_video_frames(req_, nn_video_frames_)) != srs_success) {
srs_warn("SRT: stat video frames err %s", srs_error_desc(err).c_str());
srs_freep(err);
}
nn_video_frames_ = 0;
}
// Update the stat for audio frames periodically.
if (nn_audio_frames_ > 288) {
if ((err = stat_->on_audio_frames(req_, nn_audio_frames_)) != srs_success) {
srs_warn("SRT: stat audio frames err %s", srs_error_desc(err).c_str());
srs_freep(err);
}
nn_audio_frames_ = 0;
}
}
srs_error_t SrsSrtFormat::on_ts_message(SrsTsMessage *msg)
{
srs_error_t err = srs_success;
// Only parse video and audio messages
if (msg->channel_->stream_ == SrsTsStreamVideoH264 || msg->channel_->stream_ == SrsTsStreamVideoHEVC) {
// Update statistics for video frames
update_ts_message_stats(false);
if (video_codec_reported_) {
return err;
}
if ((err = parse_video_codec(msg)) != srs_success) {
return srs_error_wrap(err, "parse video codec");
}
if (format_->vcodec_) {
video_codec_reported_ = true;
SrsVideoCodecConfig *c = format_->vcodec_;
if ((err = stat_->on_video_info(req_, c->id_, c->avc_profile_, c->avc_level_, c->width_, c->height_)) != srs_success) {
return srs_error_wrap(err, "stat video info");
}
srs_trace("srt: parsed %s codec, profile=%s, level=%s, %dx%d",
srs_video_codec_id2str(c->id_).c_str(),
srs_avc_profile2str(c->avc_profile_).c_str(), srs_avc_level2str(c->avc_level_).c_str(),
c->width_, c->height_);
}
} else if (msg->channel_->stream_ == SrsTsStreamAudioAAC) {
// Update statistics for audio frames
update_ts_message_stats(true);
if (audio_codec_reported_) {
return err;
}
if ((err = parse_audio_codec(msg)) != srs_success) {
return srs_error_wrap(err, "parse audio codec");
}
if (format_->acodec_) {
audio_codec_reported_ = true;
SrsAudioCodecConfig *c = format_->acodec_;
SrsAudioChannels channels = c->sound_type_;
if (c->id_ == SrsAudioCodecIdAAC) {
channels = (c->aac_channels_ == 1) ? SrsAudioChannelsMono : SrsAudioChannelsStereo;
}
if ((err = stat_->on_audio_info(req_, c->id_, c->sound_rate_, channels, c->aac_object_)) != srs_success) {
return srs_error_wrap(err, "stat audio info");
}
srs_trace("srt: parsed %s codec, sample_rate=%dHZ, channels=%d, profile=%s",
srs_audio_codec_id2str(c->id_).c_str(),
srs_audio_sample_rate2number(c->sound_rate_), (int)channels + 1,
srs_aac_object2str(c->aac_object_).c_str());
}
}
return err;
}
srs_error_t SrsSrtFormat::parse_video_codec(SrsTsMessage *msg)
{
srs_error_t err = srs_success;
// Parse the video codec from TS message payload
SrsBuffer avs(msg->payload_->bytes(), msg->payload_->length());
// Parse H.265/HEVC to extract VPS/SPS/PPS
// TODO: FIXME: Implement HEVC codec parsing similar to H.264
if (msg->channel_->stream_ == SrsTsStreamVideoHEVC) {
video_codec_reported_ = true;
return err;
}
// Only parse H.264 video messages
if (msg->channel_->stream_ == SrsTsStreamVideoH264) {
// Parse H.264 to extract SPS/PPS
SrsUniquePtr<SrsRawH264Stream> avc(new SrsRawH264Stream());
std::string sps, pps;
while (!avs.empty()) {
char *data = NULL;
int data_size = 0;
if ((err = avc->annexb_demux(&avs, &data, &data_size)) != srs_success) {
return srs_error_wrap(err, "demux annexb");
}
if (data == NULL || data_size == 0) {
continue;
}
// Extract SPS
if (avc->is_sps(data, data_size)) {
if ((err = avc->sps_demux(data, data_size, sps)) != srs_success) {
return srs_error_wrap(err, "demux sps");
}
}
// Extract PPS
if (avc->is_pps(data, data_size)) {
if ((err = avc->pps_demux(data, data_size, pps)) != srs_success) {
return srs_error_wrap(err, "demux pps");
}
}
// Skip until we have both SPS and PPS
if (sps.empty() || pps.empty()) {
continue;
}
// If we have both SPS and PPS, parse codec details
std::string sh;
if ((err = avc->mux_sequence_header(sps, pps, sh)) != srs_success) {
return srs_error_wrap(err, "mux sequence header");
}
// Create a temporary media packet to parse codec info
char *flv = NULL;
int nb_flv = 0;
uint32_t dts = 0;
if ((err = avc->mux_avc2flv(sh, SrsVideoAvcFrameTypeKeyFrame, SrsVideoAvcFrameTraitSequenceHeader, dts, dts, &flv, &nb_flv)) != srs_success) {
return srs_error_wrap(err, "avc to flv");
}
SrsMediaPacket frame;
frame.wrap(flv, nb_flv);
// Parse the video format to extract codec details
if ((err = format_->on_video(&frame)) != srs_success) {
return srs_error_wrap(err, "format parse video");
}
return err;
}
}
return err;
}
srs_error_t SrsSrtFormat::parse_audio_codec(SrsTsMessage *msg)
{
srs_error_t err = srs_success;
// Parse the audio codec from TS message payload
SrsBuffer avs(msg->payload_->bytes(), msg->payload_->length());
if (msg->channel_->stream_ == SrsTsStreamAudioAAC) {
// Parse AAC to extract audio specific config
SrsUniquePtr<SrsRawAacStream> aac(new SrsRawAacStream());
char *frame = NULL;
int frame_size = 0;
SrsRawAacStreamCodec codec;
if ((err = aac->adts_demux(&avs, &frame, &frame_size, codec)) != srs_success) {
return srs_error_wrap(err, "demux adts");
}
if (frame_size <= 0) {
return err;
}
std::string sh;
if ((err = aac->mux_sequence_header(&codec, sh)) != srs_success) {
return srs_error_wrap(err, "mux sequence header");
}
// Create a temporary media packet to parse codec info
int rtmp_len = sh.size() + 2;
char *buf = new char[rtmp_len];
SrsBuffer stream(buf, rtmp_len);
uint8_t aac_flag = (SrsAudioCodecIdAAC << 4) | (codec.sound_rate_ << 2) | (codec.sound_size_ << 1) | codec.sound_type_;
stream.write_1bytes(aac_flag);
stream.write_1bytes(0);
stream.write_bytes((char *)sh.data(), sh.size());
SrsMediaPacket frame_pkt;
frame_pkt.wrap(buf, rtmp_len);
// Parse the audio format to extract codec details
if ((err = format_->on_audio(&frame_pkt)) != srs_success) {
return srs_error_wrap(err, "format parse audio");
}
}
return err;
}
ISrsSrtSource::ISrsSrtSource()
{
}
ISrsSrtSource::~ISrsSrtSource()
{
}
SrsSrtSource::SrsSrtSource()
{
req_ = NULL;
can_publish_ = true;
srt_bridge_ = NULL;
stream_die_at_ = 0;
stat_ = _srs_stat;
format_ = new SrsSrtFormat();
}
SrsSrtSource::~SrsSrtSource()
{
// never free the consumers,
// for all consumers are auto free.
consumers_.clear();
srs_freep(srt_bridge_);
srs_freep(req_);
srs_freep(format_);
SrsContextId cid = _source_id;
if (cid.empty())
cid = _pre_source_id;
srs_trace("free srt source id=[%s]", cid.c_str());
stat_ = NULL;
}
// CRITICAL: This method is called AFTER the source has been added to the source pool
// in the fetch_or_create pattern (see PR 4449).
//
// IMPORTANT: All field initialization in this method MUST NOT cause coroutine context switches
// before completing the basic field setup.
//
// If context switches occur before all fields are properly initialized, other coroutines
// accessing this source from the pool may encounter uninitialized state, leading to crashes
// or undefined behavior.
//
// This prevents the race condition where multiple coroutines could create duplicate sources
// for the same stream when context switches occurred during initialization.
srs_error_t SrsSrtSource::initialize(ISrsRequest *r)
{
srs_error_t err = srs_success;
req_ = r->copy();
// Initialize format parser for codec detection
if ((err = format_->initialize(req_)) != srs_success) {
return srs_error_wrap(err, "format initialize");
}
return err;
}
bool SrsSrtSource::stream_is_dead()
{
// still publishing?
if (!can_publish_) {
return false;
}
// has any consumers?
if (!consumers_.empty()) {
return false;
}
// Delay cleanup source.
srs_utime_t now = srs_time_now_cached();
if (now < stream_die_at_ + SRS_SRT_SOURCE_CLEANUP) {
return false;
}
return true;
}
srs_error_t SrsSrtSource::on_source_id_changed(SrsContextId id)
{
srs_error_t err = srs_success;
if (!_source_id.compare(id)) {
return err;
}
if (_pre_source_id.empty()) {
_pre_source_id = id;
}
_source_id = id;
// notice all consumer
std::vector<ISrsSrtConsumer *>::iterator it;
for (it = consumers_.begin(); it != consumers_.end(); ++it) {
ISrsSrtConsumer *consumer = *it;
SrsSrtConsumer *consumer_impl = dynamic_cast<SrsSrtConsumer *>(consumer);
if (consumer_impl) {
consumer_impl->update_source_id();
}
}
return err;
}
SrsContextId SrsSrtSource::source_id()
{
return _source_id;
}
SrsContextId SrsSrtSource::pre_source_id()
{
return _pre_source_id;
}
void SrsSrtSource::update_auth(ISrsRequest *r)
{
req_->update_auth(r);
}
void SrsSrtSource::set_bridge(ISrsSrtBridge *bridge)
{
srs_freep(srt_bridge_);
srt_bridge_ = bridge;
}
srs_error_t SrsSrtSource::create_consumer(ISrsSrtConsumer *&consumer)
{
srs_error_t err = srs_success;
consumer = new SrsSrtConsumer(this);
consumers_.push_back(consumer);
stream_die_at_ = 0;
return err;
}
srs_error_t SrsSrtSource::consumer_dumps(ISrsSrtConsumer *consumer)
{
srs_error_t err = srs_success;
// print status.
srs_trace("create ts consumer, no gop cache");
return err;
}
void SrsSrtSource::on_consumer_destroy(ISrsSrtConsumer *consumer)
{
std::vector<ISrsSrtConsumer *>::iterator it;
it = std::find(consumers_.begin(), consumers_.end(), consumer);
if (it != consumers_.end()) {
it = consumers_.erase(it);
}
// Destroy and cleanup source when no publishers and consumers.
if (can_publish_ && consumers_.empty()) {
stream_die_at_ = srs_time_now_cached();
}
}
bool SrsSrtSource::can_publish()
{
return can_publish_;
}
srs_error_t SrsSrtSource::on_publish()
{
srs_error_t err = srs_success;
can_publish_ = false;
if ((err = on_source_id_changed(_srs_context->get_id())) != srs_success) {
return srs_error_wrap(err, "source id change");
}
if (srt_bridge_ && (err = srt_bridge_->on_publish()) != srs_success) {
return srs_error_wrap(err, "bridge on publish");
}
stat_->on_stream_publish(req_, _source_id.c_str());
return err;
}
void SrsSrtSource::on_unpublish()
{
// ignore when already unpublished.
if (can_publish_) {
return;
}
stat_->on_stream_close(req_);
if (srt_bridge_) {
srt_bridge_->on_unpublish();
srs_freep(srt_bridge_);
}
// Destroy and cleanup source when no publishers and consumers.
if (consumers_.empty()) {
stream_die_at_ = srs_time_now_cached();
}
// Should never change the final state before all cleanup is done.
can_publish_ = true;
}
srs_error_t SrsSrtSource::on_srt_packet(SrsSrtPacket *packet)
{
srs_error_t err = srs_success;
// Parse packet to extract codec information for statistics
// This is lightweight and only parses until codec info is found
if ((err = format_->on_srt_packet(packet)) != srs_success) {
// Don't fail on parse errors, just log and continue
srs_warn("srt source parse packet err=%s", srs_error_desc(err).c_str());
srs_freep(err);
}
for (int i = 0; i < (int)consumers_.size(); i++) {
ISrsSrtConsumer *consumer = consumers_.at(i);
if ((err = consumer->enqueue(packet->copy())) != srs_success) {
return srs_error_wrap(err, "consume ts packet");
}
}
if (srt_bridge_ && (err = srt_bridge_->on_srt_packet(packet)) != srs_success) {
return srs_error_wrap(err, "bridge consume message");
}
return err;
}