gecko-dev/dom/media/mediasource/ContainerParser.cpp
Chris Pearce 42fd2adbaa Bug 1558364 - Add MediaSpan and use it for TrackBuffersManager::mInputBuffer. r=jya
As seen in this profile of a Twitch replay: https://perfht.ml/2K9Ydb3 we can
often end up spending time in TrackBuffersManager::CodedFrameProcessing()
shaving off bytes from the front off TrackBuffersManager::mInputBuffer. This
requires all the remaining bytes to be memmove'd down to the start of this
array. Sometimes we have close to 1MB in that buffer, and when we're just
trying to consume a few hundred bytes, that becomes high overhead.

So intead of using this "slice off, shuffle down" approach change
TrackBuffersManager::mInputBuffer to be a new type MediaSpan, which maintains a
RefPtr to a MediaByteBuffer and a span defining the subregion of the buffer we
care about. This means the RemoveElementsAt(0,N) operation becomes basically
free, and we can eliminate a few other copies we were doing as well.

Differential Revision: https://phabricator.services.mozilla.com/D34661

--HG--
extra : moz-landing-system : lando
2019-06-14 02:10:09 +00:00

767 lines
27 KiB
C++

/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#include "ContainerParser.h"
#include "WebMBufferedParser.h"
#include "mozilla/EndianUtils.h"
#include "mozilla/IntegerPrintfMacros.h"
#include "mozilla/ErrorResult.h"
#include "MoofParser.h"
#include "mozilla/Logging.h"
#include "mozilla/Maybe.h"
#include "mozilla/Result.h"
#include "MediaData.h"
#include "nsMimeTypes.h"
#ifdef MOZ_FMP4
# include "AtomType.h"
# include "BufferReader.h"
# include "Index.h"
# include "MP4Interval.h"
# include "ByteStream.h"
#endif
#include "nsAutoPtr.h"
#include "SourceBufferResource.h"
#include <algorithm>
extern mozilla::LogModule* GetMediaSourceSamplesLog();
#define STRINGIFY(x) #x
#define TOSTRING(x) STRINGIFY(x)
#define MSE_DEBUG(arg, ...) \
DDMOZ_LOG(GetMediaSourceSamplesLog(), mozilla::LogLevel::Debug, \
"(%s)::%s: " arg, mType.OriginalString().Data(), __func__, \
##__VA_ARGS__)
#define MSE_DEBUGV(arg, ...) \
DDMOZ_LOG(GetMediaSourceSamplesLog(), mozilla::LogLevel::Verbose, \
"(%s)::%s: " arg, mType.OriginalString().Data(), __func__, \
##__VA_ARGS__)
#define MSE_DEBUGVEX(_this, arg, ...) \
DDMOZ_LOGEX(_this, GetMediaSourceSamplesLog(), mozilla::LogLevel::Verbose, \
"(%s)::%s: " arg, mType.OriginalString().Data(), __func__, \
##__VA_ARGS__)
namespace mozilla {
ContainerParser::ContainerParser(const MediaContainerType& aType)
: mHasInitData(false), mTotalParsed(0), mGlobalOffset(0), mType(aType) {}
ContainerParser::~ContainerParser() = default;
MediaResult ContainerParser::IsInitSegmentPresent(const MediaSpan& aData) {
MSE_DEBUG(
"aLength=%zu [%x%x%x%x]", aData.Length(),
aData.Length() > 0 ? aData[0] : 0, aData.Length() > 1 ? aData[1] : 0,
aData.Length() > 2 ? aData[2] : 0, aData.Length() > 3 ? aData[3] : 0);
return NS_ERROR_NOT_AVAILABLE;
}
MediaResult ContainerParser::IsMediaSegmentPresent(const MediaSpan& aData) {
MSE_DEBUG(
"aLength=%zu [%x%x%x%x]", aData.Length(),
aData.Length() > 0 ? aData[0] : 0, aData.Length() > 1 ? aData[1] : 0,
aData.Length() > 2 ? aData[2] : 0, aData.Length() > 3 ? aData[3] : 0);
return NS_ERROR_NOT_AVAILABLE;
}
MediaResult ContainerParser::ParseStartAndEndTimestamps(const MediaSpan& aData,
int64_t& aStart,
int64_t& aEnd) {
return NS_ERROR_NOT_AVAILABLE;
}
bool ContainerParser::TimestampsFuzzyEqual(int64_t aLhs, int64_t aRhs) {
return llabs(aLhs - aRhs) <= GetRoundingError();
}
int64_t ContainerParser::GetRoundingError() {
NS_WARNING("Using default ContainerParser::GetRoundingError implementation");
return 0;
}
bool ContainerParser::HasCompleteInitData() {
return mHasInitData && !!mInitData->Length();
}
MediaByteBuffer* ContainerParser::InitData() { return mInitData; }
MediaByteRange ContainerParser::InitSegmentRange() {
return mCompleteInitSegmentRange;
}
MediaByteRange ContainerParser::MediaHeaderRange() {
return mCompleteMediaHeaderRange;
}
MediaByteRange ContainerParser::MediaSegmentRange() {
return mCompleteMediaSegmentRange;
}
DDLoggedTypeDeclNameAndBase(WebMContainerParser, ContainerParser);
class WebMContainerParser
: public ContainerParser,
public DecoderDoctorLifeLogger<WebMContainerParser> {
public:
explicit WebMContainerParser(const MediaContainerType& aType)
: ContainerParser(aType), mParser(0), mOffset(0) {}
static const unsigned NS_PER_USEC = 1000;
MediaResult IsInitSegmentPresent(const MediaSpan& aData) override {
ContainerParser::IsInitSegmentPresent(aData);
if (aData.Length() < 4) {
return NS_ERROR_NOT_AVAILABLE;
}
WebMBufferedParser parser(0);
nsTArray<WebMTimeDataOffset> mapping;
ReentrantMonitor dummy("dummy");
bool result =
parser.Append(aData.Elements(), aData.Length(), mapping, dummy);
if (!result) {
return MediaResult(NS_ERROR_FAILURE,
RESULT_DETAIL("Invalid webm content"));
}
return parser.mInitEndOffset > 0 ? NS_OK : NS_ERROR_NOT_AVAILABLE;
}
MediaResult IsMediaSegmentPresent(const MediaSpan& aData) override {
ContainerParser::IsMediaSegmentPresent(aData);
if (aData.Length() < 4) {
return NS_ERROR_NOT_AVAILABLE;
}
WebMBufferedParser parser(0);
nsTArray<WebMTimeDataOffset> mapping;
ReentrantMonitor dummy("dummy");
parser.AppendMediaSegmentOnly();
bool result =
parser.Append(aData.Elements(), aData.Length(), mapping, dummy);
if (!result) {
return MediaResult(NS_ERROR_FAILURE,
RESULT_DETAIL("Invalid webm content"));
}
return parser.GetClusterOffset() >= 0 ? NS_OK : NS_ERROR_NOT_AVAILABLE;
}
MediaResult ParseStartAndEndTimestamps(const MediaSpan& aData,
int64_t& aStart,
int64_t& aEnd) override {
bool initSegment = NS_SUCCEEDED(IsInitSegmentPresent(aData));
if (mLastMapping &&
(initSegment || NS_SUCCEEDED(IsMediaSegmentPresent(aData)))) {
// The last data contained a complete cluster but we can only detect it
// now that a new one is starting.
// We use mOffset as end position to ensure that any blocks not reported
// by WebMBufferParser are properly skipped.
mCompleteMediaSegmentRange =
MediaByteRange(mLastMapping.ref().mSyncOffset, mOffset) +
mGlobalOffset;
mLastMapping.reset();
MSE_DEBUG("New cluster found at start, ending previous one");
return NS_ERROR_NOT_AVAILABLE;
}
if (initSegment) {
mOffset = 0;
mParser = WebMBufferedParser(0);
mOverlappedMapping.Clear();
mInitData = new MediaByteBuffer();
mResource = new SourceBufferResource();
DDLINKCHILD("resource", mResource.get());
mCompleteInitSegmentRange = MediaByteRange();
mCompleteMediaHeaderRange = MediaByteRange();
mCompleteMediaSegmentRange = MediaByteRange();
mGlobalOffset = mTotalParsed;
}
// XXX if it only adds new mappings, overlapped but not available
// (e.g. overlap < 0) frames are "lost" from the reported mappings here.
nsTArray<WebMTimeDataOffset> mapping;
mapping.AppendElements(mOverlappedMapping);
mOverlappedMapping.Clear();
ReentrantMonitor dummy("dummy");
mParser.Append(aData.Elements(), aData.Length(), mapping, dummy);
if (mResource) {
mResource->AppendData(aData);
}
// XXX This is a bit of a hack. Assume if there are no timecodes
// present and it's an init segment that it's _just_ an init segment.
// We should be more precise.
if (initSegment || !HasCompleteInitData()) {
if (mParser.mInitEndOffset > 0) {
MOZ_ASSERT(mParser.mInitEndOffset <= mResource->GetLength());
if (!mInitData->SetLength(mParser.mInitEndOffset, fallible)) {
// Super unlikely OOM
return NS_ERROR_OUT_OF_MEMORY;
}
mCompleteInitSegmentRange =
MediaByteRange(0, mParser.mInitEndOffset) + mGlobalOffset;
char* buffer = reinterpret_cast<char*>(mInitData->Elements());
mResource->ReadFromCache(buffer, 0, mParser.mInitEndOffset);
MSE_DEBUG("Stashed init of %" PRId64 " bytes.", mParser.mInitEndOffset);
mResource = nullptr;
} else {
MSE_DEBUG("Incomplete init found.");
}
mHasInitData = true;
}
mOffset += aData.Length();
mTotalParsed += aData.Length();
if (mapping.IsEmpty()) {
return NS_ERROR_NOT_AVAILABLE;
}
// Calculate media range for first media segment.
// Check if we have a cluster finishing in the current data.
uint32_t endIdx = mapping.Length() - 1;
bool foundNewCluster = false;
while (mapping[0].mSyncOffset != mapping[endIdx].mSyncOffset) {
endIdx -= 1;
foundNewCluster = true;
}
int32_t completeIdx = endIdx;
while (completeIdx >= 0 && mOffset < mapping[completeIdx].mEndOffset) {
MSE_DEBUG("block is incomplete, missing: %" PRId64,
mapping[completeIdx].mEndOffset - mOffset);
completeIdx -= 1;
}
// Save parsed blocks for which we do not have all data yet.
mOverlappedMapping.AppendElements(mapping.Elements() + completeIdx + 1,
mapping.Length() - completeIdx - 1);
if (completeIdx < 0) {
mLastMapping.reset();
return NS_ERROR_NOT_AVAILABLE;
}
if (mCompleteMediaHeaderRange.IsEmpty()) {
mCompleteMediaHeaderRange =
MediaByteRange(mapping[0].mSyncOffset, mapping[0].mEndOffset) +
mGlobalOffset;
}
if (foundNewCluster && mOffset >= mapping[endIdx].mEndOffset) {
// We now have all information required to delimit a complete cluster.
int64_t endOffset = mapping[endIdx + 1].mSyncOffset;
if (mapping[endIdx + 1].mInitOffset > mapping[endIdx].mInitOffset) {
// We have a new init segment before this cluster.
endOffset = mapping[endIdx + 1].mInitOffset;
}
mCompleteMediaSegmentRange =
MediaByteRange(mapping[endIdx].mSyncOffset, endOffset) +
mGlobalOffset;
} else if (mapping[endIdx].mClusterEndOffset >= 0 &&
mOffset >= mapping[endIdx].mClusterEndOffset) {
mCompleteMediaSegmentRange =
MediaByteRange(
mapping[endIdx].mSyncOffset,
mParser.EndSegmentOffset(mapping[endIdx].mClusterEndOffset)) +
mGlobalOffset;
}
Maybe<WebMTimeDataOffset> previousMapping;
if (completeIdx) {
previousMapping = Some(mapping[completeIdx - 1]);
} else {
previousMapping = mLastMapping;
}
mLastMapping = Some(mapping[completeIdx]);
if (!previousMapping && completeIdx + 1u >= mapping.Length()) {
// We have no previous nor next block available,
// so we can't estimate this block's duration.
return NS_ERROR_NOT_AVAILABLE;
}
uint64_t frameDuration =
(completeIdx + 1u < mapping.Length())
? mapping[completeIdx + 1].mTimecode -
mapping[completeIdx].mTimecode
: mapping[completeIdx].mTimecode - previousMapping.ref().mTimecode;
aStart = mapping[0].mTimecode / NS_PER_USEC;
aEnd = (mapping[completeIdx].mTimecode + frameDuration) / NS_PER_USEC;
MSE_DEBUG("[%" PRId64 ", %" PRId64 "] [fso=%" PRId64 ", leo=%" PRId64
", l=%zu processedIdx=%u fs=%" PRId64 "]",
aStart, aEnd, mapping[0].mSyncOffset,
mapping[completeIdx].mEndOffset, mapping.Length(), completeIdx,
mCompleteMediaSegmentRange.mEnd);
return NS_OK;
}
int64_t GetRoundingError() override {
int64_t error = mParser.GetTimecodeScale() / NS_PER_USEC;
return error * 2;
}
private:
WebMBufferedParser mParser;
nsTArray<WebMTimeDataOffset> mOverlappedMapping;
int64_t mOffset;
Maybe<WebMTimeDataOffset> mLastMapping;
};
#ifdef MOZ_FMP4
DDLoggedTypeDeclNameAndBase(MP4Stream, ByteStream);
class MP4Stream : public ByteStream, public DecoderDoctorLifeLogger<MP4Stream> {
public:
explicit MP4Stream(SourceBufferResource* aResource);
virtual ~MP4Stream();
bool ReadAt(int64_t aOffset, void* aBuffer, size_t aCount,
size_t* aBytesRead) override;
bool CachedReadAt(int64_t aOffset, void* aBuffer, size_t aCount,
size_t* aBytesRead) override;
bool Length(int64_t* aSize) override;
const uint8_t* GetContiguousAccess(int64_t aOffset, size_t aSize) override;
private:
RefPtr<SourceBufferResource> mResource;
};
MP4Stream::MP4Stream(SourceBufferResource* aResource) : mResource(aResource) {
MOZ_COUNT_CTOR(MP4Stream);
MOZ_ASSERT(aResource);
DDLINKCHILD("resource", aResource);
}
MP4Stream::~MP4Stream() { MOZ_COUNT_DTOR(MP4Stream); }
bool MP4Stream::ReadAt(int64_t aOffset, void* aBuffer, size_t aCount,
size_t* aBytesRead) {
return CachedReadAt(aOffset, aBuffer, aCount, aBytesRead);
}
bool MP4Stream::CachedReadAt(int64_t aOffset, void* aBuffer, size_t aCount,
size_t* aBytesRead) {
nsresult rv = mResource->ReadFromCache(reinterpret_cast<char*>(aBuffer),
aOffset, aCount);
if (NS_FAILED(rv)) {
*aBytesRead = 0;
return false;
}
*aBytesRead = aCount;
return true;
}
const uint8_t* MP4Stream::GetContiguousAccess(int64_t aOffset, size_t aSize) {
return mResource->GetContiguousAccess(aOffset, aSize);
}
bool MP4Stream::Length(int64_t* aSize) {
if (mResource->GetLength() < 0) return false;
*aSize = mResource->GetLength();
return true;
}
DDLoggedTypeDeclNameAndBase(MP4ContainerParser, ContainerParser);
class MP4ContainerParser : public ContainerParser,
public DecoderDoctorLifeLogger<MP4ContainerParser> {
public:
explicit MP4ContainerParser(const MediaContainerType& aType)
: ContainerParser(aType) {}
MediaResult IsInitSegmentPresent(const MediaSpan& aData) override {
ContainerParser::IsInitSegmentPresent(aData);
// Each MP4 atom has a chunk size and chunk type. The root chunk in an MP4
// file is the 'ftyp' atom followed by a file type. We just check for a
// vaguely valid 'ftyp' atom.
if (aData.Length() < 8) {
return NS_ERROR_NOT_AVAILABLE;
}
AtomParser parser(*this, aData, AtomParser::StopAt::eInitSegment);
if (!parser.IsValid()) {
return MediaResult(
NS_ERROR_FAILURE,
RESULT_DETAIL("Invalid Top-Level Box:%s", parser.LastInvalidBox()));
}
return parser.StartWithInitSegment() ? NS_OK : NS_ERROR_NOT_AVAILABLE;
}
MediaResult IsMediaSegmentPresent(const MediaSpan& aData) override {
if (aData.Length() < 8) {
return NS_ERROR_NOT_AVAILABLE;
}
AtomParser parser(*this, aData, AtomParser::StopAt::eMediaSegment);
if (!parser.IsValid()) {
return MediaResult(
NS_ERROR_FAILURE,
RESULT_DETAIL("Invalid Box:%s", parser.LastInvalidBox()));
}
return parser.StartWithMediaSegment() ? NS_OK : NS_ERROR_NOT_AVAILABLE;
}
private:
class AtomParser {
public:
enum class StopAt { eInitSegment, eMediaSegment, eEnd };
AtomParser(const MP4ContainerParser& aParser, const MediaSpan& aData,
StopAt aStop = StopAt::eEnd) {
mValid = Init(aParser, aData, aStop).isOk();
}
Result<Ok, nsresult> Init(const MP4ContainerParser& aParser,
const MediaSpan& aData, StopAt aStop) {
const MediaContainerType mType(
aParser.ContainerType()); // for logging macro.
BufferReader reader(aData);
AtomType initAtom("moov");
AtomType mediaAtom("moof");
AtomType dataAtom("mdat");
// Valid top-level boxes defined in ISO/IEC 14496-12 (Table 1)
static const AtomType validBoxes[] = {
"ftyp", "moov", // init segment
"pdin", "free", "sidx", // optional prior moov box
"styp", "moof", "mdat", // media segment
"mfra", "skip", "meta", "meco", "ssix", "prft", // others.
"pssh", // optional with encrypted EME, though ignored.
"emsg", // ISO23009-1:2014 Section 5.10.3.3
"bloc", "uuid" // boxes accepted by chrome.
};
while (reader.Remaining() >= 8) {
uint32_t tmp;
MOZ_TRY_VAR(tmp, reader.ReadU32());
uint64_t size = tmp;
const uint8_t* typec = reader.Peek(4);
MOZ_TRY_VAR(tmp, reader.ReadU32());
AtomType type(tmp);
MSE_DEBUGVEX(&aParser, "Checking atom:'%c%c%c%c' @ %u", typec[0],
typec[1], typec[2], typec[3],
(uint32_t)reader.Offset() - 8);
if (std::find(std::begin(validBoxes), std::end(validBoxes), type) ==
std::end(validBoxes)) {
// No valid box found, no point continuing.
mLastInvalidBox[0] = typec[0];
mLastInvalidBox[1] = typec[1];
mLastInvalidBox[2] = typec[2];
mLastInvalidBox[3] = typec[3];
mLastInvalidBox[4] = '\0';
return Err(NS_ERROR_FAILURE);
}
if (mInitOffset.isNothing() && AtomType(type) == initAtom) {
mInitOffset = Some(reader.Offset());
}
if (mMediaOffset.isNothing() && AtomType(type) == mediaAtom) {
mMediaOffset = Some(reader.Offset());
}
if (mDataOffset.isNothing() && AtomType(type) == dataAtom) {
mDataOffset = Some(reader.Offset());
}
if (size == 1) {
// 64 bits size.
MOZ_TRY_VAR(size, reader.ReadU64());
} else if (size == 0) {
// Atom extends to the end of the buffer, it can't have what we're
// looking for.
break;
}
if (reader.Remaining() < size - 8) {
// Incomplete atom.
break;
}
reader.Read(size - 8);
if (aStop == StopAt::eInitSegment && (mInitOffset || mMediaOffset)) {
// When we're looking for an init segment, if we encountered a media
// segment, it we will need to be processed first. So we can stop
// right away if we have found a media segment.
break;
}
if (aStop == StopAt::eMediaSegment &&
(mInitOffset || (mMediaOffset && mDataOffset))) {
// When we're looking for a media segment, if we encountered an init
// segment, it we will need to be processed first. So we can stop
// right away if we have found an init segment.
break;
}
}
return Ok();
}
bool StartWithInitSegment() const {
return mInitOffset.isSome() && (mMediaOffset.isNothing() ||
mInitOffset.ref() < mMediaOffset.ref());
}
bool StartWithMediaSegment() const {
return mMediaOffset.isSome() && (mInitOffset.isNothing() ||
mMediaOffset.ref() < mInitOffset.ref());
}
bool IsValid() const { return mValid; }
const char* LastInvalidBox() const { return mLastInvalidBox; }
private:
Maybe<size_t> mInitOffset;
Maybe<size_t> mMediaOffset;
Maybe<size_t> mDataOffset;
bool mValid;
char mLastInvalidBox[5];
};
public:
MediaResult ParseStartAndEndTimestamps(const MediaSpan& aData,
int64_t& aStart,
int64_t& aEnd) override {
bool initSegment = NS_SUCCEEDED(IsInitSegmentPresent(aData));
if (initSegment) {
mResource = new SourceBufferResource();
DDLINKCHILD("resource", mResource.get());
mStream = new MP4Stream(mResource);
// We use a timestampOffset of 0 for ContainerParser, and require
// consumers of ParseStartAndEndTimestamps to add their timestamp offset
// manually. This allows the ContainerParser to be shared across different
// timestampOffsets.
mParser = new MoofParser(mStream, AsVariant(ParseAllTracks{}),
/* aIsAudio = */ false);
DDLINKCHILD("parser", mParser.get());
mInitData = new MediaByteBuffer();
mCompleteInitSegmentRange = MediaByteRange();
mCompleteMediaHeaderRange = MediaByteRange();
mCompleteMediaSegmentRange = MediaByteRange();
mGlobalOffset = mTotalParsed;
} else if (!mStream || !mParser) {
mTotalParsed += aData.Length();
return NS_ERROR_NOT_AVAILABLE;
}
mResource->AppendData(aData);
MediaByteRangeSet byteRanges;
byteRanges +=
MediaByteRange(int64_t(mParser->mOffset), mResource->GetLength());
mParser->RebuildFragmentedIndex(byteRanges);
if (initSegment || !HasCompleteInitData()) {
MediaByteRange& range = mParser->mInitRange;
if (range.Length()) {
mCompleteInitSegmentRange = range + mGlobalOffset;
if (!mInitData->SetLength(range.Length(), fallible)) {
// Super unlikely OOM
return NS_ERROR_OUT_OF_MEMORY;
}
char* buffer = reinterpret_cast<char*>(mInitData->Elements());
mResource->ReadFromCache(buffer, range.mStart, range.Length());
MSE_DEBUG("Stashed init of %" PRIu64 " bytes.", range.Length());
} else {
MSE_DEBUG("Incomplete init found.");
}
mHasInitData = true;
}
mTotalParsed += aData.Length();
MP4Interval<Microseconds> compositionRange =
mParser->GetCompositionRange(byteRanges);
mCompleteMediaHeaderRange =
mParser->FirstCompleteMediaHeader() + mGlobalOffset;
mCompleteMediaSegmentRange =
mParser->FirstCompleteMediaSegment() + mGlobalOffset;
if (HasCompleteInitData()) {
mResource->EvictData(mParser->mOffset, mParser->mOffset);
}
if (compositionRange.IsNull()) {
return NS_ERROR_NOT_AVAILABLE;
}
aStart = compositionRange.start;
aEnd = compositionRange.end;
MSE_DEBUG("[%" PRId64 ", %" PRId64 "]", aStart, aEnd);
return NS_OK;
}
// Gaps of up to 35ms (marginally longer than a single frame at 30fps) are
// considered to be sequential frames.
int64_t GetRoundingError() override { return 35000; }
private:
RefPtr<MP4Stream> mStream;
nsAutoPtr<MoofParser> mParser;
};
#endif // MOZ_FMP4
#ifdef MOZ_FMP4
DDLoggedTypeDeclNameAndBase(ADTSContainerParser, ContainerParser);
class ADTSContainerParser
: public ContainerParser,
public DecoderDoctorLifeLogger<ADTSContainerParser> {
public:
explicit ADTSContainerParser(const MediaContainerType& aType)
: ContainerParser(aType) {}
typedef struct {
size_t header_length; // Length of just the initialization data.
size_t frame_length; // Includes header_length;
uint8_t aac_frames; // Number of AAC frames in the ADTS frame.
bool have_crc;
} Header;
/// Helper to parse the ADTS header, returning data we care about.
/// Returns true if the header is parsed successfully.
/// Returns false if the header is invalid or incomplete,
/// without modifying the passed-in Header object.
bool Parse(const MediaSpan& aData, Header& header) {
// ADTS initialization segments are just the packet header.
if (aData.Length() < 7) {
MSE_DEBUG("buffer too short for header.");
return false;
}
// Check 0xfffx sync word plus layer 0.
if ((aData[0] != 0xff) || ((aData[1] & 0xf6) != 0xf0)) {
MSE_DEBUG("no syncword.");
return false;
}
bool have_crc = !(aData[1] & 0x01);
if (have_crc && aData.Length() < 9) {
MSE_DEBUG("buffer too short for header with crc.");
return false;
}
uint8_t frequency_index = (aData[2] & 0x3c) >> 2;
MOZ_ASSERT(frequency_index < 16);
if (frequency_index == 15) {
MSE_DEBUG("explicit frequency disallowed.");
return false;
}
size_t header_length = have_crc ? 9 : 7;
size_t data_length = ((aData[3] & 0x03) << 11) | ((aData[4] & 0xff) << 3) |
((aData[5] & 0xe0) >> 5);
uint8_t frames = (aData[6] & 0x03) + 1;
MOZ_ASSERT(frames > 0);
MOZ_ASSERT(frames < 4);
// Return successfully parsed data.
header.header_length = header_length;
header.frame_length = header_length + data_length;
header.aac_frames = frames;
header.have_crc = have_crc;
return true;
}
MediaResult IsInitSegmentPresent(const MediaSpan& aData) override {
// Call superclass for logging.
ContainerParser::IsInitSegmentPresent(aData);
Header header;
if (!Parse(aData, header)) {
return NS_ERROR_NOT_AVAILABLE;
}
MSE_DEBUGV("%llu byte frame %d aac frames%s",
(unsigned long long)header.frame_length, (int)header.aac_frames,
header.have_crc ? " crc" : "");
return NS_OK;
}
MediaResult IsMediaSegmentPresent(const MediaSpan& aData) override {
// Call superclass for logging.
ContainerParser::IsMediaSegmentPresent(aData);
// Make sure we have a header so we know how long the frame is.
// NB this assumes the media segment buffer starts with an
// initialization segment. Since every frame has an ADTS header
// this is a normal place to divide packets, but we can re-parse
// mInitData if we need to handle separate media segments.
Header header;
if (!Parse(aData, header)) {
return NS_ERROR_NOT_AVAILABLE;
}
// We're supposed to return true as long as aData contains the
// start of a media segment, whether or not it's complete. So
// return true if we have any data beyond the header.
if (aData.Length() <= header.header_length) {
return NS_ERROR_NOT_AVAILABLE;
}
// We should have at least a partial frame.
return NS_OK;
}
MediaResult ParseStartAndEndTimestamps(const MediaSpan& aData,
int64_t& aStart,
int64_t& aEnd) override {
// ADTS header.
Header header;
if (!Parse(aData, header)) {
return NS_ERROR_NOT_AVAILABLE;
}
mHasInitData = true;
mCompleteInitSegmentRange =
MediaByteRange(0, int64_t(header.header_length));
// Cache raw header in case the caller wants a copy.
mInitData = new MediaByteBuffer(header.header_length);
mInitData->AppendElements(aData.Elements(), header.header_length);
// Check that we have enough data for the frame body.
if (aData.Length() < header.frame_length) {
MSE_DEBUGV(
"Not enough data for %llu byte frame"
" in %llu byte buffer.",
(unsigned long long)header.frame_length,
(unsigned long long)(aData.Length()));
return NS_ERROR_NOT_AVAILABLE;
}
mCompleteMediaSegmentRange =
MediaByteRange(header.header_length, header.frame_length);
// The ADTS MediaSource Byte Stream Format document doesn't
// define media header. Just treat it the same as the whole
// media segment.
mCompleteMediaHeaderRange = mCompleteMediaSegmentRange;
MSE_DEBUG("[%" PRId64 ", %" PRId64 "]", aStart, aEnd);
// We don't update timestamps, regardless.
return NS_ERROR_NOT_AVAILABLE;
}
// Audio shouldn't have gaps.
// Especially when we generate the timestamps ourselves.
int64_t GetRoundingError() override { return 0; }
};
#endif // MOZ_FMP4
/*static*/
ContainerParser* ContainerParser::CreateForMIMEType(
const MediaContainerType& aType) {
if (aType.Type() == MEDIAMIMETYPE(VIDEO_WEBM) ||
aType.Type() == MEDIAMIMETYPE(AUDIO_WEBM)) {
return new WebMContainerParser(aType);
}
#ifdef MOZ_FMP4
if (aType.Type() == MEDIAMIMETYPE(VIDEO_MP4) ||
aType.Type() == MEDIAMIMETYPE(AUDIO_MP4)) {
return new MP4ContainerParser(aType);
}
if (aType.Type() == MEDIAMIMETYPE("audio/aac")) {
return new ADTSContainerParser(aType);
}
#endif
return new ContainerParser(aType);
}
#undef MSE_DEBUG
#undef MSE_DEBUGV
#undef MSE_DEBUGVEX
} // namespace mozilla