| /* |
| * Copyright (C) 2009 The Android Open Source Project |
| * |
| * Licensed under the Apache License, Version 2.0 (the "License"); |
| * you may not use this file except in compliance with the License. |
| * You may obtain a copy of the License at |
| * |
| * http://www.apache.org/licenses/LICENSE-2.0 |
| * |
| * Unless required by applicable law or agreed to in writing, software |
| * distributed under the License is distributed on an "AS IS" BASIS, |
| * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| * See the License for the specific language governing permissions and |
| * limitations under the License. |
| */ |
| |
| //#define LOG_NDEBUG 0 |
| #define LOG_TAG "MPEG4Extractor" |
| |
| #include <ctype.h> |
| #include <inttypes.h> |
| #include <memory> |
| #include <stdint.h> |
| #include <stdlib.h> |
| #include <string.h> |
| |
| #include <utils/Log.h> |
| |
| #include "include/MPEG4Extractor.h" |
| #include "include/SampleTable.h" |
| #include "include/ItemTable.h" |
| #include "include/ESDS.h" |
| |
| #include <media/stagefright/foundation/ABitReader.h> |
| #include <media/stagefright/foundation/ABuffer.h> |
| #include <media/stagefright/foundation/ADebug.h> |
| #include <media/stagefright/foundation/AMessage.h> |
| #include <media/stagefright/foundation/AUtils.h> |
| #include <media/stagefright/foundation/ColorUtils.h> |
| #include <media/stagefright/foundation/hexdump.h> |
| #include <media/stagefright/MediaBuffer.h> |
| #include <media/stagefright/MediaBufferGroup.h> |
| #include <media/stagefright/MediaDefs.h> |
| #include <media/stagefright/MediaSource.h> |
| #include <media/stagefright/MetaData.h> |
| #include <utils/String8.h> |
| |
| #include <byteswap.h> |
| #include "include/ID3.h" |
| #include "include/avc_utils.h" |
| |
| #ifndef UINT32_MAX |
| #define UINT32_MAX (4294967295U) |
| #endif |
| |
| namespace android { |
| |
| enum { |
| // max track header chunk to return |
| kMaxTrackHeaderSize = 32, |
| |
| // maximum size of an atom. Some atoms can be bigger according to the spec, |
| // but we only allow up to this size. |
| kMaxAtomSize = 64 * 1024 * 1024, |
| }; |
| |
| class MPEG4Source : public MediaSource { |
| public: |
| // Caller retains ownership of both "dataSource" and "sampleTable". |
| MPEG4Source(const sp<MPEG4Extractor> &owner, |
| const sp<MetaData> &format, |
| const sp<DataSource> &dataSource, |
| int32_t timeScale, |
| const sp<SampleTable> &sampleTable, |
| Vector<SidxEntry> &sidx, |
| const Trex *trex, |
| off64_t firstMoofOffset, |
| const sp<ItemTable> &itemTable); |
| virtual status_t init(); |
| |
| virtual status_t start(MetaData *params = NULL); |
| virtual status_t stop(); |
| |
| virtual sp<MetaData> getFormat(); |
| |
| virtual status_t read(MediaBuffer **buffer, const ReadOptions *options = NULL); |
| virtual bool supportNonblockingRead() { return true; } |
| virtual status_t fragmentedRead(MediaBuffer **buffer, const ReadOptions *options = NULL); |
| |
| protected: |
| virtual ~MPEG4Source(); |
| |
| private: |
| Mutex mLock; |
| |
| // keep the MPEG4Extractor around, since we're referencing its data |
| sp<MPEG4Extractor> mOwner; |
| sp<MetaData> mFormat; |
| sp<DataSource> mDataSource; |
| int32_t mTimescale; |
| sp<SampleTable> mSampleTable; |
| uint32_t mCurrentSampleIndex; |
| uint32_t mCurrentFragmentIndex; |
| Vector<SidxEntry> &mSegments; |
| const Trex *mTrex; |
| off64_t mFirstMoofOffset; |
| off64_t mCurrentMoofOffset; |
| off64_t mNextMoofOffset; |
| uint32_t mCurrentTime; |
| int32_t mLastParsedTrackId; |
| int32_t mTrackId; |
| |
| int32_t mCryptoMode; // passed in from extractor |
| int32_t mDefaultIVSize; // passed in from extractor |
| uint8_t mCryptoKey[16]; // passed in from extractor |
| uint32_t mCurrentAuxInfoType; |
| uint32_t mCurrentAuxInfoTypeParameter; |
| int32_t mCurrentDefaultSampleInfoSize; |
| uint32_t mCurrentSampleInfoCount; |
| uint32_t mCurrentSampleInfoAllocSize; |
| uint8_t* mCurrentSampleInfoSizes; |
| uint32_t mCurrentSampleInfoOffsetCount; |
| uint32_t mCurrentSampleInfoOffsetsAllocSize; |
| uint64_t* mCurrentSampleInfoOffsets; |
| |
| bool mIsAVC; |
| bool mIsHEVC; |
| size_t mNALLengthSize; |
| |
| bool mStarted; |
| |
| MediaBufferGroup *mGroup; |
| |
| MediaBuffer *mBuffer; |
| |
| bool mWantsNALFragments; |
| |
| uint8_t *mSrcBuffer; |
| |
| bool mIsHEIF; |
| sp<ItemTable> mItemTable; |
| |
| size_t parseNALSize(const uint8_t *data) const; |
| status_t parseChunk(off64_t *offset); |
| status_t parseTrackFragmentHeader(off64_t offset, off64_t size); |
| status_t parseTrackFragmentRun(off64_t offset, off64_t size); |
| status_t parseSampleAuxiliaryInformationSizes(off64_t offset, off64_t size); |
| status_t parseSampleAuxiliaryInformationOffsets(off64_t offset, off64_t size); |
| |
| struct TrackFragmentHeaderInfo { |
| enum Flags { |
| kBaseDataOffsetPresent = 0x01, |
| kSampleDescriptionIndexPresent = 0x02, |
| kDefaultSampleDurationPresent = 0x08, |
| kDefaultSampleSizePresent = 0x10, |
| kDefaultSampleFlagsPresent = 0x20, |
| kDurationIsEmpty = 0x10000, |
| }; |
| |
| uint32_t mTrackID; |
| uint32_t mFlags; |
| uint64_t mBaseDataOffset; |
| uint32_t mSampleDescriptionIndex; |
| uint32_t mDefaultSampleDuration; |
| uint32_t mDefaultSampleSize; |
| uint32_t mDefaultSampleFlags; |
| |
| uint64_t mDataOffset; |
| }; |
| TrackFragmentHeaderInfo mTrackFragmentHeaderInfo; |
| |
| struct Sample { |
| off64_t offset; |
| size_t size; |
| uint32_t duration; |
| int32_t compositionOffset; |
| uint8_t iv[16]; |
| Vector<size_t> clearsizes; |
| Vector<size_t> encryptedsizes; |
| }; |
| Vector<Sample> mCurrentSamples; |
| |
| MPEG4Source(const MPEG4Source &); |
| MPEG4Source &operator=(const MPEG4Source &); |
| }; |
| |
| // This custom data source wraps an existing one and satisfies requests |
| // falling entirely within a cached range from the cache while forwarding |
| // all remaining requests to the wrapped datasource. |
| // This is used to cache the full sampletable metadata for a single track, |
| // possibly wrapping multiple times to cover all tracks, i.e. |
| // Each MPEG4DataSource caches the sampletable metadata for a single track. |
| |
| struct MPEG4DataSource : public DataSource { |
| explicit MPEG4DataSource(const sp<DataSource> &source); |
| |
| virtual status_t initCheck() const; |
| virtual ssize_t readAt(off64_t offset, void *data, size_t size); |
| virtual status_t getSize(off64_t *size); |
| virtual uint32_t flags(); |
| |
| status_t setCachedRange(off64_t offset, size_t size); |
| |
| protected: |
| virtual ~MPEG4DataSource(); |
| |
| private: |
| Mutex mLock; |
| |
| sp<DataSource> mSource; |
| off64_t mCachedOffset; |
| size_t mCachedSize; |
| uint8_t *mCache; |
| |
| void clearCache(); |
| |
| MPEG4DataSource(const MPEG4DataSource &); |
| MPEG4DataSource &operator=(const MPEG4DataSource &); |
| }; |
| |
| MPEG4DataSource::MPEG4DataSource(const sp<DataSource> &source) |
| : mSource(source), |
| mCachedOffset(0), |
| mCachedSize(0), |
| mCache(NULL) { |
| } |
| |
| MPEG4DataSource::~MPEG4DataSource() { |
| clearCache(); |
| } |
| |
| void MPEG4DataSource::clearCache() { |
| if (mCache) { |
| free(mCache); |
| mCache = NULL; |
| } |
| |
| mCachedOffset = 0; |
| mCachedSize = 0; |
| } |
| |
| status_t MPEG4DataSource::initCheck() const { |
| return mSource->initCheck(); |
| } |
| |
| ssize_t MPEG4DataSource::readAt(off64_t offset, void *data, size_t size) { |
| Mutex::Autolock autoLock(mLock); |
| |
| if (isInRange(mCachedOffset, mCachedSize, offset, size)) { |
| memcpy(data, &mCache[offset - mCachedOffset], size); |
| return size; |
| } |
| |
| return mSource->readAt(offset, data, size); |
| } |
| |
| status_t MPEG4DataSource::getSize(off64_t *size) { |
| return mSource->getSize(size); |
| } |
| |
| uint32_t MPEG4DataSource::flags() { |
| return mSource->flags(); |
| } |
| |
| status_t MPEG4DataSource::setCachedRange(off64_t offset, size_t size) { |
| Mutex::Autolock autoLock(mLock); |
| |
| clearCache(); |
| |
| mCache = (uint8_t *)malloc(size); |
| |
| if (mCache == NULL) { |
| return -ENOMEM; |
| } |
| |
| mCachedOffset = offset; |
| mCachedSize = size; |
| |
| ssize_t err = mSource->readAt(mCachedOffset, mCache, mCachedSize); |
| |
| if (err < (ssize_t)size) { |
| clearCache(); |
| |
| return ERROR_IO; |
| } |
| |
| return OK; |
| } |
| |
| //////////////////////////////////////////////////////////////////////////////// |
| |
| static const bool kUseHexDump = false; |
| |
| static const char *FourCC2MIME(uint32_t fourcc) { |
| switch (fourcc) { |
| case FOURCC('m', 'p', '4', 'a'): |
| return MEDIA_MIMETYPE_AUDIO_AAC; |
| |
| case FOURCC('s', 'a', 'm', 'r'): |
| return MEDIA_MIMETYPE_AUDIO_AMR_NB; |
| |
| case FOURCC('s', 'a', 'w', 'b'): |
| return MEDIA_MIMETYPE_AUDIO_AMR_WB; |
| |
| case FOURCC('m', 'p', '4', 'v'): |
| return MEDIA_MIMETYPE_VIDEO_MPEG4; |
| |
| case FOURCC('s', '2', '6', '3'): |
| case FOURCC('h', '2', '6', '3'): |
| case FOURCC('H', '2', '6', '3'): |
| return MEDIA_MIMETYPE_VIDEO_H263; |
| |
| case FOURCC('a', 'v', 'c', '1'): |
| return MEDIA_MIMETYPE_VIDEO_AVC; |
| |
| case FOURCC('h', 'v', 'c', '1'): |
| case FOURCC('h', 'e', 'v', '1'): |
| return MEDIA_MIMETYPE_VIDEO_HEVC; |
| default: |
| CHECK(!"should not be here."); |
| return NULL; |
| } |
| } |
| |
| static bool AdjustChannelsAndRate(uint32_t fourcc, uint32_t *channels, uint32_t *rate) { |
| if (!strcasecmp(MEDIA_MIMETYPE_AUDIO_AMR_NB, FourCC2MIME(fourcc))) { |
| // AMR NB audio is always mono, 8kHz |
| *channels = 1; |
| *rate = 8000; |
| return true; |
| } else if (!strcasecmp(MEDIA_MIMETYPE_AUDIO_AMR_WB, FourCC2MIME(fourcc))) { |
| // AMR WB audio is always mono, 16kHz |
| *channels = 1; |
| *rate = 16000; |
| return true; |
| } |
| return false; |
| } |
| |
| MPEG4Extractor::MPEG4Extractor(const sp<DataSource> &source) |
| : mMoofOffset(0), |
| mMoofFound(false), |
| mMdatFound(false), |
| mDataSource(source), |
| mInitCheck(NO_INIT), |
| mHeaderTimescale(0), |
| mIsQT(false), |
| mIsHEIF(false), |
| mFirstTrack(NULL), |
| mLastTrack(NULL), |
| mFileMetaData(new MetaData), |
| mFirstSINF(NULL), |
| mIsDrm(false) { |
| } |
| |
| MPEG4Extractor::~MPEG4Extractor() { |
| release(); |
| } |
| |
| void MPEG4Extractor::release() { |
| Track *track = mFirstTrack; |
| while (track) { |
| Track *next = track->next; |
| |
| delete track; |
| track = next; |
| } |
| mFirstTrack = mLastTrack = NULL; |
| |
| SINF *sinf = mFirstSINF; |
| while (sinf) { |
| SINF *next = sinf->next; |
| delete[] sinf->IPMPData; |
| delete sinf; |
| sinf = next; |
| } |
| mFirstSINF = NULL; |
| |
| for (size_t i = 0; i < mPssh.size(); i++) { |
| delete [] mPssh[i].data; |
| } |
| mPssh.clear(); |
| |
| if (mDataSource != NULL) { |
| mDataSource->close(); |
| mDataSource.clear(); |
| } |
| } |
| |
| uint32_t MPEG4Extractor::flags() const { |
| return CAN_PAUSE | |
| ((mMoofOffset == 0 || mSidxEntries.size() != 0) ? |
| (CAN_SEEK_BACKWARD | CAN_SEEK_FORWARD | CAN_SEEK) : 0); |
| } |
| |
| sp<MetaData> MPEG4Extractor::getMetaData() { |
| status_t err; |
| if ((err = readMetaData()) != OK) { |
| return new MetaData; |
| } |
| |
| return mFileMetaData; |
| } |
| |
| size_t MPEG4Extractor::countTracks() { |
| status_t err; |
| if ((err = readMetaData()) != OK) { |
| ALOGV("MPEG4Extractor::countTracks: no tracks"); |
| return 0; |
| } |
| |
| size_t n = 0; |
| Track *track = mFirstTrack; |
| while (track) { |
| ++n; |
| track = track->next; |
| } |
| |
| ALOGV("MPEG4Extractor::countTracks: %zu tracks", n); |
| return n; |
| } |
| |
| sp<MetaData> MPEG4Extractor::getTrackMetaData( |
| size_t index, uint32_t flags) { |
| status_t err; |
| if ((err = readMetaData()) != OK) { |
| return NULL; |
| } |
| |
| Track *track = mFirstTrack; |
| while (index > 0) { |
| if (track == NULL) { |
| return NULL; |
| } |
| |
| track = track->next; |
| --index; |
| } |
| |
| if (track == NULL) { |
| return NULL; |
| } |
| |
| if ((flags & kIncludeExtensiveMetaData) |
| && !track->includes_expensive_metadata) { |
| track->includes_expensive_metadata = true; |
| |
| const char *mime; |
| CHECK(track->meta->findCString(kKeyMIMEType, &mime)); |
| if (!strncasecmp("video/", mime, 6)) { |
| // MPEG2 tracks do not provide CSD, so read the stream header |
| if (!strcmp(mime, MEDIA_MIMETYPE_VIDEO_MPEG2)) { |
| off64_t offset; |
| size_t size; |
| if (track->sampleTable->getMetaDataForSample( |
| 0 /* sampleIndex */, &offset, &size, NULL /* sampleTime */) == OK) { |
| if (size > kMaxTrackHeaderSize) { |
| size = kMaxTrackHeaderSize; |
| } |
| uint8_t header[kMaxTrackHeaderSize]; |
| if (mDataSource->readAt(offset, &header, size) == (ssize_t)size) { |
| track->meta->setData(kKeyStreamHeader, 'mdat', header, size); |
| } |
| } |
| } |
| |
| if (mMoofOffset > 0) { |
| int64_t duration; |
| if (track->meta->findInt64(kKeyDuration, &duration)) { |
| // nothing fancy, just pick a frame near 1/4th of the duration |
| track->meta->setInt64( |
| kKeyThumbnailTime, duration / 4); |
| } |
| } else { |
| uint32_t sampleIndex; |
| uint32_t sampleTime; |
| if (track->timescale != 0 && |
| track->sampleTable->findThumbnailSample(&sampleIndex) == OK |
| && track->sampleTable->getMetaDataForSample( |
| sampleIndex, NULL /* offset */, NULL /* size */, |
| &sampleTime) == OK) { |
| track->meta->setInt64( |
| kKeyThumbnailTime, |
| ((int64_t)sampleTime * 1000000) / track->timescale); |
| } |
| } |
| } |
| } |
| |
| return track->meta; |
| } |
| |
| status_t MPEG4Extractor::readMetaData() { |
| if (mInitCheck != NO_INIT) { |
| return mInitCheck; |
| } |
| |
| off64_t offset = 0; |
| status_t err; |
| bool sawMoovOrSidx = false; |
| |
| while (!((sawMoovOrSidx && (mMdatFound || mMoofFound)) || |
| (mIsHEIF && (mItemTable != NULL) && mItemTable->isValid()))) { |
| off64_t orig_offset = offset; |
| err = parseChunk(&offset, 0); |
| |
| if (err != OK && err != UNKNOWN_ERROR) { |
| break; |
| } else if (offset <= orig_offset) { |
| // only continue parsing if the offset was advanced, |
| // otherwise we might end up in an infinite loop |
| ALOGE("did not advance: %lld->%lld", (long long)orig_offset, (long long)offset); |
| err = ERROR_MALFORMED; |
| break; |
| } else if (err == UNKNOWN_ERROR) { |
| sawMoovOrSidx = true; |
| } |
| } |
| |
| if (mInitCheck == OK) { |
| if (findTrackByMimePrefix("video/") != NULL) { |
| mFileMetaData->setCString( |
| kKeyMIMEType, MEDIA_MIMETYPE_CONTAINER_MPEG4); |
| } else if (findTrackByMimePrefix("audio/") != NULL) { |
| mFileMetaData->setCString(kKeyMIMEType, "audio/mp4"); |
| } else { |
| mFileMetaData->setCString(kKeyMIMEType, "application/octet-stream"); |
| } |
| } else { |
| mInitCheck = err; |
| } |
| |
| CHECK_NE(err, (status_t)NO_INIT); |
| |
| // copy pssh data into file metadata |
| uint64_t psshsize = 0; |
| for (size_t i = 0; i < mPssh.size(); i++) { |
| psshsize += 20 + mPssh[i].datalen; |
| } |
| if (psshsize > 0 && psshsize <= UINT32_MAX) { |
| char *buf = (char*)malloc(psshsize); |
| if (!buf) { |
| ALOGE("b/28471206"); |
| return NO_MEMORY; |
| } |
| char *ptr = buf; |
| for (size_t i = 0; i < mPssh.size(); i++) { |
| memcpy(ptr, mPssh[i].uuid, 20); // uuid + length |
| memcpy(ptr + 20, mPssh[i].data, mPssh[i].datalen); |
| ptr += (20 + mPssh[i].datalen); |
| } |
| mFileMetaData->setData(kKeyPssh, 'pssh', buf, psshsize); |
| free(buf); |
| } |
| |
| if (mIsHEIF) { |
| sp<MetaData> meta = mItemTable->getImageMeta(); |
| if (meta == NULL) { |
| return ERROR_MALFORMED; |
| } |
| |
| Track *track = mLastTrack; |
| if (track != NULL) { |
| ALOGW("track is set before metadata is fully processed"); |
| } else { |
| track = new Track; |
| track->next = NULL; |
| mFirstTrack = mLastTrack = track; |
| } |
| |
| track->meta = meta; |
| track->meta->setInt32(kKeyTrackID, 0); |
| track->includes_expensive_metadata = false; |
| track->skipTrack = false; |
| track->timescale = 0; |
| } |
| |
| return mInitCheck; |
| } |
| |
| char* MPEG4Extractor::getDrmTrackInfo(size_t trackID, int *len) { |
| if (mFirstSINF == NULL) { |
| return NULL; |
| } |
| |
| SINF *sinf = mFirstSINF; |
| while (sinf && (trackID != sinf->trackID)) { |
| sinf = sinf->next; |
| } |
| |
| if (sinf == NULL) { |
| return NULL; |
| } |
| |
| *len = sinf->len; |
| return sinf->IPMPData; |
| } |
| |
| // Reads an encoded integer 7 bits at a time until it encounters the high bit clear. |
| static int32_t readSize(off64_t offset, |
| const sp<DataSource> &DataSource, uint8_t *numOfBytes) { |
| uint32_t size = 0; |
| uint8_t data; |
| bool moreData = true; |
| *numOfBytes = 0; |
| |
| while (moreData) { |
| if (DataSource->readAt(offset, &data, 1) < 1) { |
| return -1; |
| } |
| offset ++; |
| moreData = (data >= 128) ? true : false; |
| size = (size << 7) | (data & 0x7f); // Take last 7 bits |
| (*numOfBytes) ++; |
| } |
| |
| return size; |
| } |
| |
| status_t MPEG4Extractor::parseDrmSINF( |
| off64_t * /* offset */, off64_t data_offset) { |
| uint8_t updateIdTag; |
| if (mDataSource->readAt(data_offset, &updateIdTag, 1) < 1) { |
| return ERROR_IO; |
| } |
| data_offset ++; |
| |
| if (0x01/*OBJECT_DESCRIPTOR_UPDATE_ID_TAG*/ != updateIdTag) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t numOfBytes; |
| int32_t size = readSize(data_offset, mDataSource, &numOfBytes); |
| if (size < 0) { |
| return ERROR_IO; |
| } |
| data_offset += numOfBytes; |
| |
| while(size >= 11 ) { |
| uint8_t descriptorTag; |
| if (mDataSource->readAt(data_offset, &descriptorTag, 1) < 1) { |
| return ERROR_IO; |
| } |
| data_offset ++; |
| |
| if (0x11/*OBJECT_DESCRIPTOR_ID_TAG*/ != descriptorTag) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t buffer[8]; |
| //ObjectDescriptorID and ObjectDescriptor url flag |
| if (mDataSource->readAt(data_offset, buffer, 2) < 2) { |
| return ERROR_IO; |
| } |
| data_offset += 2; |
| |
| if ((buffer[1] >> 5) & 0x0001) { //url flag is set |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt(data_offset, buffer, 8) < 8) { |
| return ERROR_IO; |
| } |
| data_offset += 8; |
| |
| if ((0x0F/*ES_ID_REF_TAG*/ != buffer[1]) |
| || ( 0x0A/*IPMP_DESCRIPTOR_POINTER_ID_TAG*/ != buffer[5])) { |
| return ERROR_MALFORMED; |
| } |
| |
| SINF *sinf = new SINF; |
| sinf->trackID = U16_AT(&buffer[3]); |
| sinf->IPMPDescriptorID = buffer[7]; |
| sinf->next = mFirstSINF; |
| mFirstSINF = sinf; |
| |
| size -= (8 + 2 + 1); |
| } |
| |
| if (size != 0) { |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt(data_offset, &updateIdTag, 1) < 1) { |
| return ERROR_IO; |
| } |
| data_offset ++; |
| |
| if(0x05/*IPMP_DESCRIPTOR_UPDATE_ID_TAG*/ != updateIdTag) { |
| return ERROR_MALFORMED; |
| } |
| |
| size = readSize(data_offset, mDataSource, &numOfBytes); |
| if (size < 0) { |
| return ERROR_IO; |
| } |
| data_offset += numOfBytes; |
| |
| while (size > 0) { |
| uint8_t tag; |
| int32_t dataLen; |
| if (mDataSource->readAt(data_offset, &tag, 1) < 1) { |
| return ERROR_IO; |
| } |
| data_offset ++; |
| |
| if (0x0B/*IPMP_DESCRIPTOR_ID_TAG*/ == tag) { |
| uint8_t id; |
| dataLen = readSize(data_offset, mDataSource, &numOfBytes); |
| if (dataLen < 0) { |
| return ERROR_IO; |
| } else if (dataLen < 4) { |
| return ERROR_MALFORMED; |
| } |
| data_offset += numOfBytes; |
| |
| if (mDataSource->readAt(data_offset, &id, 1) < 1) { |
| return ERROR_IO; |
| } |
| data_offset ++; |
| |
| SINF *sinf = mFirstSINF; |
| while (sinf && (sinf->IPMPDescriptorID != id)) { |
| sinf = sinf->next; |
| } |
| if (sinf == NULL) { |
| return ERROR_MALFORMED; |
| } |
| sinf->len = dataLen - 3; |
| sinf->IPMPData = new (std::nothrow) char[sinf->len]; |
| if (sinf->IPMPData == NULL) { |
| return ERROR_MALFORMED; |
| } |
| data_offset += 2; |
| |
| if (mDataSource->readAt(data_offset, sinf->IPMPData, sinf->len) < sinf->len) { |
| return ERROR_IO; |
| } |
| data_offset += sinf->len; |
| |
| size -= (dataLen + numOfBytes + 1); |
| } |
| } |
| |
| if (size != 0) { |
| return ERROR_MALFORMED; |
| } |
| |
| return UNKNOWN_ERROR; // Return a dummy error. |
| } |
| |
| struct PathAdder { |
| PathAdder(Vector<uint32_t> *path, uint32_t chunkType) |
| : mPath(path) { |
| mPath->push(chunkType); |
| } |
| |
| ~PathAdder() { |
| mPath->pop(); |
| } |
| |
| private: |
| Vector<uint32_t> *mPath; |
| |
| PathAdder(const PathAdder &); |
| PathAdder &operator=(const PathAdder &); |
| }; |
| |
| static bool underMetaDataPath(const Vector<uint32_t> &path) { |
| return path.size() >= 5 |
| && path[0] == FOURCC('m', 'o', 'o', 'v') |
| && path[1] == FOURCC('u', 'd', 't', 'a') |
| && path[2] == FOURCC('m', 'e', 't', 'a') |
| && path[3] == FOURCC('i', 'l', 's', 't'); |
| } |
| |
| static bool underQTMetaPath(const Vector<uint32_t> &path, int32_t depth) { |
| return path.size() >= 2 |
| && path[0] == FOURCC('m', 'o', 'o', 'v') |
| && path[1] == FOURCC('m', 'e', 't', 'a') |
| && (depth == 2 |
| || (depth == 3 |
| && (path[2] == FOURCC('h', 'd', 'l', 'r') |
| || path[2] == FOURCC('i', 'l', 's', 't') |
| || path[2] == FOURCC('k', 'e', 'y', 's')))); |
| } |
| |
| // Given a time in seconds since Jan 1 1904, produce a human-readable string. |
| static bool convertTimeToDate(int64_t time_1904, String8 *s) { |
| // delta between mpeg4 time and unix epoch time |
| static const int64_t delta = (((66 * 365 + 17) * 24) * 3600); |
| if (time_1904 < INT64_MIN + delta) { |
| return false; |
| } |
| time_t time_1970 = time_1904 - delta; |
| |
| char tmp[32]; |
| struct tm* tm = gmtime(&time_1970); |
| if (tm != NULL && |
| strftime(tmp, sizeof(tmp), "%Y%m%dT%H%M%S.000Z", tm) > 0) { |
| s->setTo(tmp); |
| return true; |
| } |
| return false; |
| } |
| |
| status_t MPEG4Extractor::parseChunk(off64_t *offset, int depth) { |
| ALOGV("entering parseChunk %lld/%d", (long long)*offset, depth); |
| |
| if (*offset < 0) { |
| ALOGE("b/23540914"); |
| return ERROR_MALFORMED; |
| } |
| if (depth > 100) { |
| ALOGE("b/27456299"); |
| return ERROR_MALFORMED; |
| } |
| uint32_t hdr[2]; |
| if (mDataSource->readAt(*offset, hdr, 8) < 8) { |
| return ERROR_IO; |
| } |
| uint64_t chunk_size = ntohl(hdr[0]); |
| int32_t chunk_type = ntohl(hdr[1]); |
| off64_t data_offset = *offset + 8; |
| |
| if (chunk_size == 1) { |
| if (mDataSource->readAt(*offset + 8, &chunk_size, 8) < 8) { |
| return ERROR_IO; |
| } |
| chunk_size = ntoh64(chunk_size); |
| data_offset += 8; |
| |
| if (chunk_size < 16) { |
| // The smallest valid chunk is 16 bytes long in this case. |
| return ERROR_MALFORMED; |
| } |
| } else if (chunk_size == 0) { |
| if (depth == 0) { |
| // atom extends to end of file |
| off64_t sourceSize; |
| if (mDataSource->getSize(&sourceSize) == OK) { |
| chunk_size = (sourceSize - *offset); |
| } else { |
| // XXX could we just pick a "sufficiently large" value here? |
| ALOGE("atom size is 0, and data source has no size"); |
| return ERROR_MALFORMED; |
| } |
| } else { |
| // not allowed for non-toplevel atoms, skip it |
| *offset += 4; |
| return OK; |
| } |
| } else if (chunk_size < 8) { |
| // The smallest valid chunk is 8 bytes long. |
| ALOGE("invalid chunk size: %" PRIu64, chunk_size); |
| return ERROR_MALFORMED; |
| } |
| |
| char chunk[5]; |
| MakeFourCCString(chunk_type, chunk); |
| ALOGV("chunk: %s @ %lld, %d", chunk, (long long)*offset, depth); |
| |
| if (kUseHexDump) { |
| static const char kWhitespace[] = " "; |
| const char *indent = &kWhitespace[sizeof(kWhitespace) - 1 - 2 * depth]; |
| printf("%sfound chunk '%s' of size %" PRIu64 "\n", indent, chunk, chunk_size); |
| |
| char buffer[256]; |
| size_t n = chunk_size; |
| if (n > sizeof(buffer)) { |
| n = sizeof(buffer); |
| } |
| if (mDataSource->readAt(*offset, buffer, n) |
| < (ssize_t)n) { |
| return ERROR_IO; |
| } |
| |
| hexdump(buffer, n); |
| } |
| |
| PathAdder autoAdder(&mPath, chunk_type); |
| |
| // (data_offset - *offset) is either 8 or 16 |
| off64_t chunk_data_size = chunk_size - (data_offset - *offset); |
| if (chunk_data_size < 0) { |
| ALOGE("b/23540914"); |
| return ERROR_MALFORMED; |
| } |
| if (chunk_type != FOURCC('m', 'd', 'a', 't') && chunk_data_size > kMaxAtomSize) { |
| char errMsg[100]; |
| sprintf(errMsg, "%s atom has size %" PRId64, chunk, chunk_data_size); |
| ALOGE("%s (b/28615448)", errMsg); |
| android_errorWriteWithInfoLog(0x534e4554, "28615448", -1, errMsg, strlen(errMsg)); |
| return ERROR_MALFORMED; |
| } |
| |
| if (chunk_type != FOURCC('c', 'p', 'r', 't') |
| && chunk_type != FOURCC('c', 'o', 'v', 'r') |
| && mPath.size() == 5 && underMetaDataPath(mPath)) { |
| off64_t stop_offset = *offset + chunk_size; |
| *offset = data_offset; |
| while (*offset < stop_offset) { |
| status_t err = parseChunk(offset, depth + 1); |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| if (*offset != stop_offset) { |
| return ERROR_MALFORMED; |
| } |
| |
| return OK; |
| } |
| |
| switch(chunk_type) { |
| case FOURCC('m', 'o', 'o', 'v'): |
| case FOURCC('t', 'r', 'a', 'k'): |
| case FOURCC('m', 'd', 'i', 'a'): |
| case FOURCC('m', 'i', 'n', 'f'): |
| case FOURCC('d', 'i', 'n', 'f'): |
| case FOURCC('s', 't', 'b', 'l'): |
| case FOURCC('m', 'v', 'e', 'x'): |
| case FOURCC('m', 'o', 'o', 'f'): |
| case FOURCC('t', 'r', 'a', 'f'): |
| case FOURCC('m', 'f', 'r', 'a'): |
| case FOURCC('u', 'd', 't', 'a'): |
| case FOURCC('i', 'l', 's', 't'): |
| case FOURCC('s', 'i', 'n', 'f'): |
| case FOURCC('s', 'c', 'h', 'i'): |
| case FOURCC('e', 'd', 't', 's'): |
| case FOURCC('w', 'a', 'v', 'e'): |
| { |
| if (chunk_type == FOURCC('m', 'o', 'o', 'v') && depth != 0) { |
| ALOGE("moov: depth %d", depth); |
| return ERROR_MALFORMED; |
| } |
| |
| if (chunk_type == FOURCC('m', 'o', 'o', 'v') && mInitCheck == OK) { |
| ALOGE("duplicate moov"); |
| return ERROR_MALFORMED; |
| } |
| |
| if (chunk_type == FOURCC('m', 'o', 'o', 'f') && !mMoofFound) { |
| // store the offset of the first segment |
| mMoofFound = true; |
| mMoofOffset = *offset; |
| } |
| |
| if (chunk_type == FOURCC('s', 't', 'b', 'l')) { |
| ALOGV("sampleTable chunk is %" PRIu64 " bytes long.", chunk_size); |
| |
| if (mDataSource->flags() |
| & (DataSource::kWantsPrefetching |
| | DataSource::kIsCachingDataSource)) { |
| sp<MPEG4DataSource> cachedSource = |
| new MPEG4DataSource(mDataSource); |
| |
| if (cachedSource->setCachedRange(*offset, chunk_size) == OK) { |
| mDataSource = cachedSource; |
| } |
| } |
| |
| if (mLastTrack == NULL) { |
| return ERROR_MALFORMED; |
| } |
| |
| mLastTrack->sampleTable = new SampleTable(mDataSource); |
| } |
| |
| bool isTrack = false; |
| if (chunk_type == FOURCC('t', 'r', 'a', 'k')) { |
| if (depth != 1) { |
| ALOGE("trak: depth %d", depth); |
| return ERROR_MALFORMED; |
| } |
| isTrack = true; |
| |
| Track *track = new Track; |
| track->next = NULL; |
| if (mLastTrack) { |
| mLastTrack->next = track; |
| } else { |
| mFirstTrack = track; |
| } |
| mLastTrack = track; |
| |
| track->meta = new MetaData; |
| track->includes_expensive_metadata = false; |
| track->skipTrack = false; |
| track->timescale = 0; |
| track->meta->setCString(kKeyMIMEType, "application/octet-stream"); |
| } |
| |
| off64_t stop_offset = *offset + chunk_size; |
| *offset = data_offset; |
| while (*offset < stop_offset) { |
| status_t err = parseChunk(offset, depth + 1); |
| if (err != OK) { |
| if (isTrack) { |
| mLastTrack->skipTrack = true; |
| break; |
| } |
| return err; |
| } |
| } |
| |
| if (*offset != stop_offset) { |
| return ERROR_MALFORMED; |
| } |
| |
| if (isTrack) { |
| int32_t trackId; |
| // There must be exact one track header per track. |
| if (!mLastTrack->meta->findInt32(kKeyTrackID, &trackId)) { |
| mLastTrack->skipTrack = true; |
| } |
| |
| status_t err = verifyTrack(mLastTrack); |
| if (err != OK) { |
| mLastTrack->skipTrack = true; |
| } |
| |
| if (mLastTrack->skipTrack) { |
| Track *cur = mFirstTrack; |
| |
| if (cur == mLastTrack) { |
| delete cur; |
| mFirstTrack = mLastTrack = NULL; |
| } else { |
| while (cur && cur->next != mLastTrack) { |
| cur = cur->next; |
| } |
| if (cur) { |
| cur->next = NULL; |
| } |
| delete mLastTrack; |
| mLastTrack = cur; |
| } |
| |
| return OK; |
| } |
| } else if (chunk_type == FOURCC('m', 'o', 'o', 'v')) { |
| mInitCheck = OK; |
| |
| if (!mIsDrm) { |
| return UNKNOWN_ERROR; // Return a dummy error. |
| } else { |
| return OK; |
| } |
| } |
| break; |
| } |
| |
| case FOURCC('e', 'l', 's', 't'): |
| { |
| *offset += chunk_size; |
| |
| // See 14496-12 8.6.6 |
| uint8_t version; |
| if (mDataSource->readAt(data_offset, &version, 1) < 1) { |
| return ERROR_IO; |
| } |
| |
| uint32_t entry_count; |
| if (!mDataSource->getUInt32(data_offset + 4, &entry_count)) { |
| return ERROR_IO; |
| } |
| |
| if (entry_count != 1) { |
| // we only support a single entry at the moment, for gapless playback |
| ALOGW("ignoring edit list with %d entries", entry_count); |
| } else if (mHeaderTimescale == 0) { |
| ALOGW("ignoring edit list because timescale is 0"); |
| } else { |
| off64_t entriesoffset = data_offset + 8; |
| uint64_t segment_duration; |
| int64_t media_time; |
| |
| if (version == 1) { |
| if (!mDataSource->getUInt64(entriesoffset, &segment_duration) || |
| !mDataSource->getUInt64(entriesoffset + 8, (uint64_t*)&media_time)) { |
| return ERROR_IO; |
| } |
| } else if (version == 0) { |
| uint32_t sd; |
| int32_t mt; |
| if (!mDataSource->getUInt32(entriesoffset, &sd) || |
| !mDataSource->getUInt32(entriesoffset + 4, (uint32_t*)&mt)) { |
| return ERROR_IO; |
| } |
| segment_duration = sd; |
| media_time = mt; |
| } else { |
| return ERROR_IO; |
| } |
| |
| uint64_t halfscale = mHeaderTimescale / 2; |
| segment_duration = (segment_duration * 1000000 + halfscale)/ mHeaderTimescale; |
| media_time = (media_time * 1000000 + halfscale) / mHeaderTimescale; |
| |
| int64_t duration; |
| int32_t samplerate; |
| if (!mLastTrack) { |
| return ERROR_MALFORMED; |
| } |
| if (mLastTrack->meta->findInt64(kKeyDuration, &duration) && |
| mLastTrack->meta->findInt32(kKeySampleRate, &samplerate)) { |
| |
| int64_t delay = (media_time * samplerate + 500000) / 1000000; |
| mLastTrack->meta->setInt32(kKeyEncoderDelay, delay); |
| |
| int64_t paddingus = duration - (int64_t)(segment_duration + media_time); |
| if (paddingus < 0) { |
| // track duration from media header (which is what kKeyDuration is) might |
| // be slightly shorter than the segment duration, which would make the |
| // padding negative. Clamp to zero. |
| paddingus = 0; |
| } |
| int64_t paddingsamples = (paddingus * samplerate + 500000) / 1000000; |
| mLastTrack->meta->setInt32(kKeyEncoderPadding, paddingsamples); |
| } |
| } |
| break; |
| } |
| |
| case FOURCC('f', 'r', 'm', 'a'): |
| { |
| *offset += chunk_size; |
| |
| uint32_t original_fourcc; |
| if (mDataSource->readAt(data_offset, &original_fourcc, 4) < 4) { |
| return ERROR_IO; |
| } |
| original_fourcc = ntohl(original_fourcc); |
| ALOGV("read original format: %d", original_fourcc); |
| |
| if (mLastTrack == NULL) { |
| return ERROR_MALFORMED; |
| } |
| |
| mLastTrack->meta->setCString(kKeyMIMEType, FourCC2MIME(original_fourcc)); |
| uint32_t num_channels = 0; |
| uint32_t sample_rate = 0; |
| if (AdjustChannelsAndRate(original_fourcc, &num_channels, &sample_rate)) { |
| mLastTrack->meta->setInt32(kKeyChannelCount, num_channels); |
| mLastTrack->meta->setInt32(kKeySampleRate, sample_rate); |
| } |
| break; |
| } |
| |
| case FOURCC('t', 'e', 'n', 'c'): |
| { |
| *offset += chunk_size; |
| |
| if (chunk_size < 32) { |
| return ERROR_MALFORMED; |
| } |
| |
| // tenc box contains 1 byte version, 3 byte flags, 3 byte default algorithm id, one byte |
| // default IV size, 16 bytes default KeyID |
| // (ISO 23001-7) |
| char buf[4]; |
| memset(buf, 0, 4); |
| if (mDataSource->readAt(data_offset + 4, buf + 1, 3) < 3) { |
| return ERROR_IO; |
| } |
| uint32_t defaultAlgorithmId = ntohl(*((int32_t*)buf)); |
| if (defaultAlgorithmId > 1) { |
| // only 0 (clear) and 1 (AES-128) are valid |
| return ERROR_MALFORMED; |
| } |
| |
| memset(buf, 0, 4); |
| if (mDataSource->readAt(data_offset + 7, buf + 3, 1) < 1) { |
| return ERROR_IO; |
| } |
| uint32_t defaultIVSize = ntohl(*((int32_t*)buf)); |
| |
| if ((defaultAlgorithmId == 0 && defaultIVSize != 0) || |
| (defaultAlgorithmId != 0 && defaultIVSize == 0)) { |
| // only unencrypted data must have 0 IV size |
| return ERROR_MALFORMED; |
| } else if (defaultIVSize != 0 && |
| defaultIVSize != 8 && |
| defaultIVSize != 16) { |
| // only supported sizes are 0, 8 and 16 |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t defaultKeyId[16]; |
| |
| if (mDataSource->readAt(data_offset + 8, &defaultKeyId, 16) < 16) { |
| return ERROR_IO; |
| } |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| mLastTrack->meta->setInt32(kKeyCryptoMode, defaultAlgorithmId); |
| mLastTrack->meta->setInt32(kKeyCryptoDefaultIVSize, defaultIVSize); |
| mLastTrack->meta->setData(kKeyCryptoKey, 'tenc', defaultKeyId, 16); |
| break; |
| } |
| |
| case FOURCC('t', 'k', 'h', 'd'): |
| { |
| *offset += chunk_size; |
| |
| status_t err; |
| if ((err = parseTrackHeader(data_offset, chunk_data_size)) != OK) { |
| return err; |
| } |
| |
| break; |
| } |
| |
| case FOURCC('p', 's', 's', 'h'): |
| { |
| *offset += chunk_size; |
| |
| PsshInfo pssh; |
| |
| if (mDataSource->readAt(data_offset + 4, &pssh.uuid, 16) < 16) { |
| return ERROR_IO; |
| } |
| |
| uint32_t psshdatalen = 0; |
| if (mDataSource->readAt(data_offset + 20, &psshdatalen, 4) < 4) { |
| return ERROR_IO; |
| } |
| pssh.datalen = ntohl(psshdatalen); |
| ALOGV("pssh data size: %d", pssh.datalen); |
| if (chunk_size < 20 || pssh.datalen > chunk_size - 20) { |
| // pssh data length exceeds size of containing box |
| return ERROR_MALFORMED; |
| } |
| |
| pssh.data = new (std::nothrow) uint8_t[pssh.datalen]; |
| if (pssh.data == NULL) { |
| return ERROR_MALFORMED; |
| } |
| ALOGV("allocated pssh @ %p", pssh.data); |
| ssize_t requested = (ssize_t) pssh.datalen; |
| if (mDataSource->readAt(data_offset + 24, pssh.data, requested) < requested) { |
| delete[] pssh.data; |
| return ERROR_IO; |
| } |
| mPssh.push_back(pssh); |
| |
| break; |
| } |
| |
| case FOURCC('m', 'd', 'h', 'd'): |
| { |
| *offset += chunk_size; |
| |
| if (chunk_data_size < 4 || mLastTrack == NULL) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t version; |
| if (mDataSource->readAt( |
| data_offset, &version, sizeof(version)) |
| < (ssize_t)sizeof(version)) { |
| return ERROR_IO; |
| } |
| |
| off64_t timescale_offset; |
| |
| if (version == 1) { |
| timescale_offset = data_offset + 4 + 16; |
| } else if (version == 0) { |
| timescale_offset = data_offset + 4 + 8; |
| } else { |
| return ERROR_IO; |
| } |
| |
| uint32_t timescale; |
| if (mDataSource->readAt( |
| timescale_offset, ×cale, sizeof(timescale)) |
| < (ssize_t)sizeof(timescale)) { |
| return ERROR_IO; |
| } |
| |
| if (!timescale) { |
| ALOGE("timescale should not be ZERO."); |
| return ERROR_MALFORMED; |
| } |
| |
| mLastTrack->timescale = ntohl(timescale); |
| |
| // 14496-12 says all ones means indeterminate, but some files seem to use |
| // 0 instead. We treat both the same. |
| int64_t duration = 0; |
| if (version == 1) { |
| if (mDataSource->readAt( |
| timescale_offset + 4, &duration, sizeof(duration)) |
| < (ssize_t)sizeof(duration)) { |
| return ERROR_IO; |
| } |
| if (duration != -1) { |
| duration = ntoh64(duration); |
| } |
| } else { |
| uint32_t duration32; |
| if (mDataSource->readAt( |
| timescale_offset + 4, &duration32, sizeof(duration32)) |
| < (ssize_t)sizeof(duration32)) { |
| return ERROR_IO; |
| } |
| if (duration32 != 0xffffffff) { |
| duration = ntohl(duration32); |
| } |
| } |
| if (duration != 0 && mLastTrack->timescale != 0) { |
| mLastTrack->meta->setInt64( |
| kKeyDuration, (duration * 1000000) / mLastTrack->timescale); |
| } |
| |
| uint8_t lang[2]; |
| off64_t lang_offset; |
| if (version == 1) { |
| lang_offset = timescale_offset + 4 + 8; |
| } else if (version == 0) { |
| lang_offset = timescale_offset + 4 + 4; |
| } else { |
| return ERROR_IO; |
| } |
| |
| if (mDataSource->readAt(lang_offset, &lang, sizeof(lang)) |
| < (ssize_t)sizeof(lang)) { |
| return ERROR_IO; |
| } |
| |
| // To get the ISO-639-2/T three character language code |
| // 1 bit pad followed by 3 5-bits characters. Each character |
| // is packed as the difference between its ASCII value and 0x60. |
| char lang_code[4]; |
| lang_code[0] = ((lang[0] >> 2) & 0x1f) + 0x60; |
| lang_code[1] = ((lang[0] & 0x3) << 3 | (lang[1] >> 5)) + 0x60; |
| lang_code[2] = (lang[1] & 0x1f) + 0x60; |
| lang_code[3] = '\0'; |
| |
| mLastTrack->meta->setCString( |
| kKeyMediaLanguage, lang_code); |
| |
| break; |
| } |
| |
| case FOURCC('s', 't', 's', 'd'): |
| { |
| uint8_t buffer[8]; |
| if (chunk_data_size < (off64_t)sizeof(buffer)) { |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, 8) < 8) { |
| return ERROR_IO; |
| } |
| |
| if (U32_AT(buffer) != 0) { |
| // Should be version 0, flags 0. |
| return ERROR_MALFORMED; |
| } |
| |
| uint32_t entry_count = U32_AT(&buffer[4]); |
| |
| if (entry_count > 1) { |
| // For 3GPP timed text, there could be multiple tx3g boxes contain |
| // multiple text display formats. These formats will be used to |
| // display the timed text. |
| // For encrypted files, there may also be more than one entry. |
| const char *mime; |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| CHECK(mLastTrack->meta->findCString(kKeyMIMEType, &mime)); |
| if (strcasecmp(mime, MEDIA_MIMETYPE_TEXT_3GPP) && |
| strcasecmp(mime, "application/octet-stream")) { |
| // For now we only support a single type of media per track. |
| mLastTrack->skipTrack = true; |
| *offset += chunk_size; |
| break; |
| } |
| } |
| off64_t stop_offset = *offset + chunk_size; |
| *offset = data_offset + 8; |
| for (uint32_t i = 0; i < entry_count; ++i) { |
| status_t err = parseChunk(offset, depth + 1); |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| if (*offset != stop_offset) { |
| return ERROR_MALFORMED; |
| } |
| break; |
| } |
| case FOURCC('m', 'e', 't', 't'): |
| { |
| *offset += chunk_size; |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| sp<ABuffer> buffer = new ABuffer(chunk_data_size); |
| if (buffer->data() == NULL) { |
| return NO_MEMORY; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer->data(), chunk_data_size) < chunk_data_size) { |
| return ERROR_IO; |
| } |
| |
| String8 mimeFormat((const char *)(buffer->data()), chunk_data_size); |
| mLastTrack->meta->setCString(kKeyMIMEType, mimeFormat.string()); |
| |
| break; |
| } |
| |
| case FOURCC('m', 'p', '4', 'a'): |
| case FOURCC('e', 'n', 'c', 'a'): |
| case FOURCC('s', 'a', 'm', 'r'): |
| case FOURCC('s', 'a', 'w', 'b'): |
| { |
| if (mIsQT && chunk_type == FOURCC('m', 'p', '4', 'a') |
| && depth >= 1 && mPath[depth - 1] == FOURCC('w', 'a', 'v', 'e')) { |
| // Ignore mp4a embedded in QT wave atom |
| *offset += chunk_size; |
| break; |
| } |
| |
| uint8_t buffer[8 + 20]; |
| if (chunk_data_size < (ssize_t)sizeof(buffer)) { |
| // Basic AudioSampleEntry size. |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, sizeof(buffer)) < (ssize_t)sizeof(buffer)) { |
| return ERROR_IO; |
| } |
| |
| uint16_t data_ref_index __unused = U16_AT(&buffer[6]); |
| uint16_t version = U16_AT(&buffer[8]); |
| uint32_t num_channels = U16_AT(&buffer[16]); |
| |
| uint16_t sample_size = U16_AT(&buffer[18]); |
| uint32_t sample_rate = U32_AT(&buffer[24]) >> 16; |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| off64_t stop_offset = *offset + chunk_size; |
| *offset = data_offset + sizeof(buffer); |
| |
| if (mIsQT && chunk_type == FOURCC('m', 'p', '4', 'a')) { |
| if (version == 1) { |
| if (mDataSource->readAt(*offset, buffer, 16) < 16) { |
| return ERROR_IO; |
| } |
| |
| #if 0 |
| U32_AT(buffer); // samples per packet |
| U32_AT(&buffer[4]); // bytes per packet |
| U32_AT(&buffer[8]); // bytes per frame |
| U32_AT(&buffer[12]); // bytes per sample |
| #endif |
| *offset += 16; |
| } else if (version == 2) { |
| uint8_t v2buffer[36]; |
| if (mDataSource->readAt(*offset, v2buffer, 36) < 36) { |
| return ERROR_IO; |
| } |
| |
| #if 0 |
| U32_AT(v2buffer); // size of struct only |
| sample_rate = (uint32_t)U64_AT(&v2buffer[4]); // audio sample rate |
| num_channels = U32_AT(&v2buffer[12]); // num audio channels |
| U32_AT(&v2buffer[16]); // always 0x7f000000 |
| sample_size = (uint16_t)U32_AT(&v2buffer[20]); // const bits per channel |
| U32_AT(&v2buffer[24]); // format specifc flags |
| U32_AT(&v2buffer[28]); // const bytes per audio packet |
| U32_AT(&v2buffer[32]); // const LPCM frames per audio packet |
| #endif |
| *offset += 36; |
| } |
| } |
| |
| if (chunk_type != FOURCC('e', 'n', 'c', 'a')) { |
| // if the chunk type is enca, we'll get the type from the sinf/frma box later |
| mLastTrack->meta->setCString(kKeyMIMEType, FourCC2MIME(chunk_type)); |
| AdjustChannelsAndRate(chunk_type, &num_channels, &sample_rate); |
| } |
| ALOGV("*** coding='%s' %d channels, size %d, rate %d\n", |
| chunk, num_channels, sample_size, sample_rate); |
| mLastTrack->meta->setInt32(kKeyChannelCount, num_channels); |
| mLastTrack->meta->setInt32(kKeySampleRate, sample_rate); |
| |
| while (*offset < stop_offset) { |
| status_t err = parseChunk(offset, depth + 1); |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| if (*offset != stop_offset) { |
| return ERROR_MALFORMED; |
| } |
| break; |
| } |
| |
| case FOURCC('m', 'p', '4', 'v'): |
| case FOURCC('e', 'n', 'c', 'v'): |
| case FOURCC('s', '2', '6', '3'): |
| case FOURCC('H', '2', '6', '3'): |
| case FOURCC('h', '2', '6', '3'): |
| case FOURCC('a', 'v', 'c', '1'): |
| case FOURCC('h', 'v', 'c', '1'): |
| case FOURCC('h', 'e', 'v', '1'): |
| { |
| uint8_t buffer[78]; |
| if (chunk_data_size < (ssize_t)sizeof(buffer)) { |
| // Basic VideoSampleEntry size. |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, sizeof(buffer)) < (ssize_t)sizeof(buffer)) { |
| return ERROR_IO; |
| } |
| |
| uint16_t data_ref_index __unused = U16_AT(&buffer[6]); |
| uint16_t width = U16_AT(&buffer[6 + 18]); |
| uint16_t height = U16_AT(&buffer[6 + 20]); |
| |
| // The video sample is not standard-compliant if it has invalid dimension. |
| // Use some default width and height value, and |
| // let the decoder figure out the actual width and height (and thus |
| // be prepared for INFO_FOMRAT_CHANGED event). |
| if (width == 0) width = 352; |
| if (height == 0) height = 288; |
| |
| // printf("*** coding='%s' width=%d height=%d\n", |
| // chunk, width, height); |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| if (chunk_type != FOURCC('e', 'n', 'c', 'v')) { |
| // if the chunk type is encv, we'll get the type from the sinf/frma box later |
| mLastTrack->meta->setCString(kKeyMIMEType, FourCC2MIME(chunk_type)); |
| } |
| mLastTrack->meta->setInt32(kKeyWidth, width); |
| mLastTrack->meta->setInt32(kKeyHeight, height); |
| |
| off64_t stop_offset = *offset + chunk_size; |
| *offset = data_offset + sizeof(buffer); |
| while (*offset < stop_offset) { |
| status_t err = parseChunk(offset, depth + 1); |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| if (*offset != stop_offset) { |
| return ERROR_MALFORMED; |
| } |
| break; |
| } |
| |
| case FOURCC('s', 't', 'c', 'o'): |
| case FOURCC('c', 'o', '6', '4'): |
| { |
| if ((mLastTrack == NULL) || (mLastTrack->sampleTable == NULL)) { |
| return ERROR_MALFORMED; |
| } |
| |
| status_t err = |
| mLastTrack->sampleTable->setChunkOffsetParams( |
| chunk_type, data_offset, chunk_data_size); |
| |
| *offset += chunk_size; |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| break; |
| } |
| |
| case FOURCC('s', 't', 's', 'c'): |
| { |
| if ((mLastTrack == NULL) || (mLastTrack->sampleTable == NULL)) |
| return ERROR_MALFORMED; |
| |
| status_t err = |
| mLastTrack->sampleTable->setSampleToChunkParams( |
| data_offset, chunk_data_size); |
| |
| *offset += chunk_size; |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| break; |
| } |
| |
| case FOURCC('s', 't', 's', 'z'): |
| case FOURCC('s', 't', 'z', '2'): |
| { |
| if ((mLastTrack == NULL) || (mLastTrack->sampleTable == NULL)) { |
| return ERROR_MALFORMED; |
| } |
| |
| status_t err = |
| mLastTrack->sampleTable->setSampleSizeParams( |
| chunk_type, data_offset, chunk_data_size); |
| |
| *offset += chunk_size; |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| size_t max_size; |
| err = mLastTrack->sampleTable->getMaxSampleSize(&max_size); |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| if (max_size != 0) { |
| // Assume that a given buffer only contains at most 10 chunks, |
| // each chunk originally prefixed with a 2 byte length will |
| // have a 4 byte header (0x00 0x00 0x00 0x01) after conversion, |
| // and thus will grow by 2 bytes per chunk. |
| if (max_size > SIZE_MAX - 10 * 2) { |
| ALOGE("max sample size too big: %zu", max_size); |
| return ERROR_MALFORMED; |
| } |
| mLastTrack->meta->setInt32(kKeyMaxInputSize, max_size + 10 * 2); |
| } else { |
| // No size was specified. Pick a conservatively large size. |
| uint32_t width, height; |
| if (!mLastTrack->meta->findInt32(kKeyWidth, (int32_t*)&width) || |
| !mLastTrack->meta->findInt32(kKeyHeight,(int32_t*) &height)) { |
| ALOGE("No width or height, assuming worst case 1080p"); |
| width = 1920; |
| height = 1080; |
| } else { |
| // A resolution was specified, check that it's not too big. The values below |
| // were chosen so that the calculations below don't cause overflows, they're |
| // not indicating that resolutions up to 32kx32k are actually supported. |
| if (width > 32768 || height > 32768) { |
| ALOGE("can't support %u x %u video", width, height); |
| return ERROR_MALFORMED; |
| } |
| } |
| |
| const char *mime; |
| CHECK(mLastTrack->meta->findCString(kKeyMIMEType, &mime)); |
| if (!strcmp(mime, MEDIA_MIMETYPE_VIDEO_AVC) |
| || !strcmp(mime, MEDIA_MIMETYPE_VIDEO_HEVC)) { |
| // AVC & HEVC requires compression ratio of at least 2, and uses |
| // macroblocks |
| max_size = ((width + 15) / 16) * ((height + 15) / 16) * 192; |
| } else { |
| // For all other formats there is no minimum compression |
| // ratio. Use compression ratio of 1. |
| max_size = width * height * 3 / 2; |
| } |
| // HACK: allow 10% overhead |
| // TODO: read sample size from traf atom for fragmented MPEG4. |
| max_size += max_size / 10; |
| mLastTrack->meta->setInt32(kKeyMaxInputSize, max_size); |
| } |
| |
| // NOTE: setting another piece of metadata invalidates any pointers (such as the |
| // mimetype) previously obtained, so don't cache them. |
| const char *mime; |
| CHECK(mLastTrack->meta->findCString(kKeyMIMEType, &mime)); |
| // Calculate average frame rate. |
| if (!strncasecmp("video/", mime, 6)) { |
| size_t nSamples = mLastTrack->sampleTable->countSamples(); |
| if (nSamples == 0) { |
| int32_t trackId; |
| if (mLastTrack->meta->findInt32(kKeyTrackID, &trackId)) { |
| for (size_t i = 0; i < mTrex.size(); i++) { |
| Trex *t = &mTrex.editItemAt(i); |
| if (t->track_ID == (uint32_t) trackId) { |
| if (t->default_sample_duration > 0) { |
| int32_t frameRate = |
| mLastTrack->timescale / t->default_sample_duration; |
| mLastTrack->meta->setInt32(kKeyFrameRate, frameRate); |
| } |
| break; |
| } |
| } |
| } |
| } else { |
| int64_t durationUs; |
| if (mLastTrack->meta->findInt64(kKeyDuration, &durationUs)) { |
| if (durationUs > 0) { |
| int32_t frameRate = (nSamples * 1000000LL + |
| (durationUs >> 1)) / durationUs; |
| mLastTrack->meta->setInt32(kKeyFrameRate, frameRate); |
| } |
| } |
| } |
| } |
| |
| break; |
| } |
| |
| case FOURCC('s', 't', 't', 's'): |
| { |
| if ((mLastTrack == NULL) || (mLastTrack->sampleTable == NULL)) |
| return ERROR_MALFORMED; |
| |
| *offset += chunk_size; |
| |
| status_t err = |
| mLastTrack->sampleTable->setTimeToSampleParams( |
| data_offset, chunk_data_size); |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| break; |
| } |
| |
| case FOURCC('c', 't', 't', 's'): |
| { |
| if ((mLastTrack == NULL) || (mLastTrack->sampleTable == NULL)) |
| return ERROR_MALFORMED; |
| |
| *offset += chunk_size; |
| |
| status_t err = |
| mLastTrack->sampleTable->setCompositionTimeToSampleParams( |
| data_offset, chunk_data_size); |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| break; |
| } |
| |
| case FOURCC('s', 't', 's', 's'): |
| { |
| if ((mLastTrack == NULL) || (mLastTrack->sampleTable == NULL)) |
| return ERROR_MALFORMED; |
| |
| *offset += chunk_size; |
| |
| status_t err = |
| mLastTrack->sampleTable->setSyncSampleParams( |
| data_offset, chunk_data_size); |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| break; |
| } |
| |
| // \xA9xyz |
| case FOURCC(0xA9, 'x', 'y', 'z'): |
| { |
| *offset += chunk_size; |
| |
| // Best case the total data length inside "\xA9xyz" box would |
| // be 9, for instance "\xA9xyz" + "\x00\x05\x15\xc7" + "+0+0/", |
| // where "\x00\x05" is the text string length with value = 5, |
| // "\0x15\xc7" is the language code = en, and "+0+0/" is a |
| // location (string) value with longitude = 0 and latitude = 0. |
| // Since some devices encountered in the wild omit the trailing |
| // slash, we'll allow that. |
| if (chunk_data_size < 8) { // 8 instead of 9 to allow for missing / |
| return ERROR_MALFORMED; |
| } |
| |
| uint16_t len; |
| if (!mDataSource->getUInt16(data_offset, &len)) { |
| return ERROR_IO; |
| } |
| |
| // allow "+0+0" without trailing slash |
| if (len < 4 || len > chunk_data_size - 4) { |
| return ERROR_MALFORMED; |
| } |
| // The location string following the language code is formatted |
| // according to ISO 6709:2008 (https://en.wikipedia.org/wiki/ISO_6709). |
| // Allocate 2 extra bytes, in case we need to add a trailing slash, |
| // and to add a terminating 0. |
| std::unique_ptr<char[]> buffer(new (std::nothrow) char[len+2]()); |
| if (!buffer) { |
| return NO_MEMORY; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset + 4, &buffer[0], len) < len) { |
| return ERROR_IO; |
| } |
| |
| len = strlen(&buffer[0]); |
| if (len < 4) { |
| return ERROR_MALFORMED; |
| } |
| // Add a trailing slash if there wasn't one. |
| if (buffer[len - 1] != '/') { |
| buffer[len] = '/'; |
| } |
| mFileMetaData->setCString(kKeyLocation, &buffer[0]); |
| break; |
| } |
| |
| case FOURCC('e', 's', 'd', 's'): |
| { |
| *offset += chunk_size; |
| |
| if (chunk_data_size < 4) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t buffer[256]; |
| if (chunk_data_size > (off64_t)sizeof(buffer)) { |
| return ERROR_BUFFER_TOO_SMALL; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, chunk_data_size) < chunk_data_size) { |
| return ERROR_IO; |
| } |
| |
| if (U32_AT(buffer) != 0) { |
| // Should be version 0, flags 0. |
| return ERROR_MALFORMED; |
| } |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| mLastTrack->meta->setData( |
| kKeyESDS, kTypeESDS, &buffer[4], chunk_data_size - 4); |
| |
| if (mPath.size() >= 2 |
| && mPath[mPath.size() - 2] == FOURCC('m', 'p', '4', 'a')) { |
| // Information from the ESDS must be relied on for proper |
| // setup of sample rate and channel count for MPEG4 Audio. |
| // The generic header appears to only contain generic |
| // information... |
| |
| status_t err = updateAudioTrackInfoFromESDS_MPEG4Audio( |
| &buffer[4], chunk_data_size - 4); |
| |
| if (err != OK) { |
| return err; |
| } |
| } |
| if (mPath.size() >= 2 |
| && mPath[mPath.size() - 2] == FOURCC('m', 'p', '4', 'v')) { |
| // Check if the video is MPEG2 |
| ESDS esds(&buffer[4], chunk_data_size - 4); |
| |
| uint8_t objectTypeIndication; |
| if (esds.getObjectTypeIndication(&objectTypeIndication) == OK) { |
| if (objectTypeIndication >= 0x60 && objectTypeIndication <= 0x65) { |
| mLastTrack->meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_VIDEO_MPEG2); |
| } |
| } |
| } |
| break; |
| } |
| |
| case FOURCC('b', 't', 'r', 't'): |
| { |
| *offset += chunk_size; |
| if (mLastTrack == NULL) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t buffer[12]; |
| if (chunk_data_size != sizeof(buffer)) { |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, chunk_data_size) < chunk_data_size) { |
| return ERROR_IO; |
| } |
| |
| uint32_t maxBitrate = U32_AT(&buffer[4]); |
| uint32_t avgBitrate = U32_AT(&buffer[8]); |
| if (maxBitrate > 0 && maxBitrate < INT32_MAX) { |
| mLastTrack->meta->setInt32(kKeyMaxBitRate, (int32_t)maxBitrate); |
| } |
| if (avgBitrate > 0 && avgBitrate < INT32_MAX) { |
| mLastTrack->meta->setInt32(kKeyBitRate, (int32_t)avgBitrate); |
| } |
| break; |
| } |
| |
| case FOURCC('a', 'v', 'c', 'C'): |
| { |
| *offset += chunk_size; |
| |
| sp<ABuffer> buffer = new ABuffer(chunk_data_size); |
| |
| if (buffer->data() == NULL) { |
| ALOGE("b/28471206"); |
| return NO_MEMORY; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer->data(), chunk_data_size) < chunk_data_size) { |
| return ERROR_IO; |
| } |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| mLastTrack->meta->setData( |
| kKeyAVCC, kTypeAVCC, buffer->data(), chunk_data_size); |
| |
| break; |
| } |
| case FOURCC('h', 'v', 'c', 'C'): |
| { |
| sp<ABuffer> buffer = new ABuffer(chunk_data_size); |
| |
| if (buffer->data() == NULL) { |
| ALOGE("b/28471206"); |
| return NO_MEMORY; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer->data(), chunk_data_size) < chunk_data_size) { |
| return ERROR_IO; |
| } |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| mLastTrack->meta->setData( |
| kKeyHVCC, kTypeHVCC, buffer->data(), chunk_data_size); |
| |
| *offset += chunk_size; |
| break; |
| } |
| |
| case FOURCC('d', '2', '6', '3'): |
| { |
| *offset += chunk_size; |
| /* |
| * d263 contains a fixed 7 bytes part: |
| * vendor - 4 bytes |
| * version - 1 byte |
| * level - 1 byte |
| * profile - 1 byte |
| * optionally, "d263" box itself may contain a 16-byte |
| * bit rate box (bitr) |
| * average bit rate - 4 bytes |
| * max bit rate - 4 bytes |
| */ |
| char buffer[23]; |
| if (chunk_data_size != 7 && |
| chunk_data_size != 23) { |
| ALOGE("Incorrect D263 box size %lld", (long long)chunk_data_size); |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, chunk_data_size) < chunk_data_size) { |
| return ERROR_IO; |
| } |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| mLastTrack->meta->setData(kKeyD263, kTypeD263, buffer, chunk_data_size); |
| |
| break; |
| } |
| |
| case FOURCC('m', 'e', 't', 'a'): |
| { |
| off64_t stop_offset = *offset + chunk_size; |
| *offset = data_offset; |
| bool isParsingMetaKeys = underQTMetaPath(mPath, 2); |
| if (!isParsingMetaKeys) { |
| uint8_t buffer[4]; |
| if (chunk_data_size < (off64_t)sizeof(buffer)) { |
| *offset = stop_offset; |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, 4) < 4) { |
| *offset = stop_offset; |
| return ERROR_IO; |
| } |
| |
| if (U32_AT(buffer) != 0) { |
| // Should be version 0, flags 0. |
| |
| // If it's not, let's assume this is one of those |
| // apparently malformed chunks that don't have flags |
| // and completely different semantics than what's |
| // in the MPEG4 specs and skip it. |
| *offset = stop_offset; |
| return OK; |
| } |
| *offset += sizeof(buffer); |
| } |
| |
| while (*offset < stop_offset) { |
| status_t err = parseChunk(offset, depth + 1); |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| if (*offset != stop_offset) { |
| return ERROR_MALFORMED; |
| } |
| break; |
| } |
| |
| case FOURCC('i', 'l', 'o', 'c'): |
| case FOURCC('i', 'i', 'n', 'f'): |
| case FOURCC('i', 'p', 'r', 'p'): |
| case FOURCC('p', 'i', 't', 'm'): |
| case FOURCC('i', 'd', 'a', 't'): |
| case FOURCC('i', 'r', 'e', 'f'): |
| case FOURCC('i', 'p', 'r', 'o'): |
| { |
| if (mIsHEIF) { |
| if (mItemTable == NULL) { |
| mItemTable = new ItemTable(mDataSource); |
| } |
| status_t err = mItemTable->parse( |
| chunk_type, data_offset, chunk_data_size); |
| if (err != OK) { |
| return err; |
| } |
| } |
| *offset += chunk_size; |
| break; |
| } |
| |
| case FOURCC('m', 'e', 'a', 'n'): |
| case FOURCC('n', 'a', 'm', 'e'): |
| case FOURCC('d', 'a', 't', 'a'): |
| { |
| *offset += chunk_size; |
| |
| if (mPath.size() == 6 && underMetaDataPath(mPath)) { |
| status_t err = parseITunesMetaData(data_offset, chunk_data_size); |
| |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| break; |
| } |
| |
| case FOURCC('m', 'v', 'h', 'd'): |
| { |
| *offset += chunk_size; |
| |
| if (depth != 1) { |
| ALOGE("mvhd: depth %d", depth); |
| return ERROR_MALFORMED; |
| } |
| if (chunk_data_size < 32) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t header[32]; |
| if (mDataSource->readAt( |
| data_offset, header, sizeof(header)) |
| < (ssize_t)sizeof(header)) { |
| return ERROR_IO; |
| } |
| |
| uint64_t creationTime; |
| uint64_t duration = 0; |
| if (header[0] == 1) { |
| creationTime = U64_AT(&header[4]); |
| mHeaderTimescale = U32_AT(&header[20]); |
| duration = U64_AT(&header[24]); |
| if (duration == 0xffffffffffffffff) { |
| duration = 0; |
| } |
| } else if (header[0] != 0) { |
| return ERROR_MALFORMED; |
| } else { |
| creationTime = U32_AT(&header[4]); |
| mHeaderTimescale = U32_AT(&header[12]); |
| uint32_t d32 = U32_AT(&header[16]); |
| if (d32 == 0xffffffff) { |
| d32 = 0; |
| } |
| duration = d32; |
| } |
| if (duration != 0 && mHeaderTimescale != 0 && duration < UINT64_MAX / 1000000) { |
| mFileMetaData->setInt64(kKeyDuration, duration * 1000000 / mHeaderTimescale); |
| } |
| |
| String8 s; |
| if (convertTimeToDate(creationTime, &s)) { |
| mFileMetaData->setCString(kKeyDate, s.string()); |
| } |
| |
| |
| break; |
| } |
| |
| case FOURCC('m', 'e', 'h', 'd'): |
| { |
| *offset += chunk_size; |
| |
| if (chunk_data_size < 8) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t flags[4]; |
| if (mDataSource->readAt( |
| data_offset, flags, sizeof(flags)) |
| < (ssize_t)sizeof(flags)) { |
| return ERROR_IO; |
| } |
| |
| uint64_t duration = 0; |
| if (flags[0] == 1) { |
| // 64 bit |
| if (chunk_data_size < 12) { |
| return ERROR_MALFORMED; |
| } |
| mDataSource->getUInt64(data_offset + 4, &duration); |
| if (duration == 0xffffffffffffffff) { |
| duration = 0; |
| } |
| } else if (flags[0] == 0) { |
| // 32 bit |
| uint32_t d32; |
| mDataSource->getUInt32(data_offset + 4, &d32); |
| if (d32 == 0xffffffff) { |
| d32 = 0; |
| } |
| duration = d32; |
| } else { |
| return ERROR_MALFORMED; |
| } |
| |
| if (duration != 0 && mHeaderTimescale != 0) { |
| mFileMetaData->setInt64(kKeyDuration, duration * 1000000 / mHeaderTimescale); |
| } |
| |
| break; |
| } |
| |
| case FOURCC('m', 'd', 'a', 't'): |
| { |
| ALOGV("mdat chunk, drm: %d", mIsDrm); |
| |
| mMdatFound = true; |
| |
| if (!mIsDrm) { |
| *offset += chunk_size; |
| break; |
| } |
| |
| if (chunk_size < 8) { |
| return ERROR_MALFORMED; |
| } |
| |
| return parseDrmSINF(offset, data_offset); |
| } |
| |
| case FOURCC('h', 'd', 'l', 'r'): |
| { |
| *offset += chunk_size; |
| |
| if (underQTMetaPath(mPath, 3)) { |
| break; |
| } |
| |
| uint32_t buffer; |
| if (mDataSource->readAt( |
| data_offset + 8, &buffer, 4) < 4) { |
| return ERROR_IO; |
| } |
| |
| uint32_t type = ntohl(buffer); |
| // For the 3GPP file format, the handler-type within the 'hdlr' box |
| // shall be 'text'. We also want to support 'sbtl' handler type |
| // for a practical reason as various MPEG4 containers use it. |
| if (type == FOURCC('t', 'e', 'x', 't') || type == FOURCC('s', 'b', 't', 'l')) { |
| if (mLastTrack != NULL) { |
| mLastTrack->meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_TEXT_3GPP); |
| } |
| } |
| |
| break; |
| } |
| |
| case FOURCC('k', 'e', 'y', 's'): |
| { |
| *offset += chunk_size; |
| |
| if (underQTMetaPath(mPath, 3)) { |
| status_t err = parseQTMetaKey(data_offset, chunk_data_size); |
| if (err != OK) { |
| return err; |
| } |
| } |
| break; |
| } |
| |
| case FOURCC('t', 'r', 'e', 'x'): |
| { |
| *offset += chunk_size; |
| |
| if (chunk_data_size < 24) { |
| return ERROR_IO; |
| } |
| Trex trex; |
| if (!mDataSource->getUInt32(data_offset + 4, &trex.track_ID) || |
| !mDataSource->getUInt32(data_offset + 8, &trex.default_sample_description_index) || |
| !mDataSource->getUInt32(data_offset + 12, &trex.default_sample_duration) || |
| !mDataSource->getUInt32(data_offset + 16, &trex.default_sample_size) || |
| !mDataSource->getUInt32(data_offset + 20, &trex.default_sample_flags)) { |
| return ERROR_IO; |
| } |
| mTrex.add(trex); |
| break; |
| } |
| |
| case FOURCC('t', 'x', '3', 'g'): |
| { |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| uint32_t type; |
| const void *data; |
| size_t size = 0; |
| if (!mLastTrack->meta->findData( |
| kKeyTextFormatData, &type, &data, &size)) { |
| size = 0; |
| } |
| |
| if ((chunk_size > SIZE_MAX) || (SIZE_MAX - chunk_size <= size)) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t *buffer = new (std::nothrow) uint8_t[size + chunk_size]; |
| if (buffer == NULL) { |
| return ERROR_MALFORMED; |
| } |
| |
| if (size > 0) { |
| memcpy(buffer, data, size); |
| } |
| |
| if ((size_t)(mDataSource->readAt(*offset, buffer + size, chunk_size)) |
| < chunk_size) { |
| delete[] buffer; |
| buffer = NULL; |
| |
| // advance read pointer so we don't end up reading this again |
| *offset += chunk_size; |
| return ERROR_IO; |
| } |
| |
| mLastTrack->meta->setData( |
| kKeyTextFormatData, 0, buffer, size + chunk_size); |
| |
| delete[] buffer; |
| |
| *offset += chunk_size; |
| break; |
| } |
| |
| case FOURCC('c', 'o', 'v', 'r'): |
| { |
| *offset += chunk_size; |
| |
| if (mFileMetaData != NULL) { |
| ALOGV("chunk_data_size = %" PRId64 " and data_offset = %" PRId64, |
| chunk_data_size, data_offset); |
| |
| if (chunk_data_size < 0 || static_cast<uint64_t>(chunk_data_size) >= SIZE_MAX - 1) { |
| return ERROR_MALFORMED; |
| } |
| sp<ABuffer> buffer = new ABuffer(chunk_data_size + 1); |
| if (buffer->data() == NULL) { |
| ALOGE("b/28471206"); |
| return NO_MEMORY; |
| } |
| if (mDataSource->readAt( |
| data_offset, buffer->data(), chunk_data_size) != (ssize_t)chunk_data_size) { |
| return ERROR_IO; |
| } |
| const int kSkipBytesOfDataBox = 16; |
| if (chunk_data_size <= kSkipBytesOfDataBox) { |
| return ERROR_MALFORMED; |
| } |
| |
| mFileMetaData->setData( |
| kKeyAlbumArt, MetaData::TYPE_NONE, |
| buffer->data() + kSkipBytesOfDataBox, chunk_data_size - kSkipBytesOfDataBox); |
| } |
| |
| break; |
| } |
| |
| case FOURCC('c', 'o', 'l', 'r'): |
| { |
| *offset += chunk_size; |
| // this must be in a VisualSampleEntry box under the Sample Description Box ('stsd') |
| // ignore otherwise |
| if (depth >= 2 && mPath[depth - 2] == FOURCC('s', 't', 's', 'd')) { |
| status_t err = parseColorInfo(data_offset, chunk_data_size); |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| break; |
| } |
| |
| case FOURCC('t', 'i', 't', 'l'): |
| case FOURCC('p', 'e', 'r', 'f'): |
| case FOURCC('a', 'u', 't', 'h'): |
| case FOURCC('g', 'n', 'r', 'e'): |
| case FOURCC('a', 'l', 'b', 'm'): |
| case FOURCC('y', 'r', 'r', 'c'): |
| { |
| *offset += chunk_size; |
| |
| status_t err = parse3GPPMetaData(data_offset, chunk_data_size, depth); |
| |
| if (err != OK) { |
| return err; |
| } |
| |
| break; |
| } |
| |
| case FOURCC('I', 'D', '3', '2'): |
| { |
| *offset += chunk_size; |
| |
| if (chunk_data_size < 6) { |
| return ERROR_MALFORMED; |
| } |
| |
| parseID3v2MetaData(data_offset + 6); |
| |
| break; |
| } |
| |
| case FOURCC('-', '-', '-', '-'): |
| { |
| mLastCommentMean.clear(); |
| mLastCommentName.clear(); |
| mLastCommentData.clear(); |
| *offset += chunk_size; |
| break; |
| } |
| |
| case FOURCC('s', 'i', 'd', 'x'): |
| { |
| status_t err = parseSegmentIndex(data_offset, chunk_data_size); |
| if (err != OK) { |
| return err; |
| } |
| *offset += chunk_size; |
| return UNKNOWN_ERROR; // stop parsing after sidx |
| } |
| |
| case FOURCC('a', 'c', '-', '3'): |
| { |
| *offset += chunk_size; |
| return parseAC3SampleEntry(data_offset); |
| } |
| |
| case FOURCC('f', 't', 'y', 'p'): |
| { |
| if (chunk_data_size < 8 || depth != 0) { |
| return ERROR_MALFORMED; |
| } |
| |
| off64_t stop_offset = *offset + chunk_size; |
| uint32_t numCompatibleBrands = (chunk_data_size - 8) / 4; |
| std::set<uint32_t> brandSet; |
| for (size_t i = 0; i < numCompatibleBrands + 2; ++i) { |
| if (i == 1) { |
| // Skip this index, it refers to the minorVersion, |
| // not a brand. |
| continue; |
| } |
| |
| uint32_t brand; |
| if (mDataSource->readAt(data_offset + 4 * i, &brand, 4) < 4) { |
| return ERROR_MALFORMED; |
| } |
| |
| brand = ntohl(brand); |
| brandSet.insert(brand); |
| } |
| |
| if (brandSet.count(FOURCC('q', 't', ' ', ' ')) > 0) { |
| mIsQT = true; |
| } else if (brandSet.count(FOURCC('m', 'i', 'f', '1')) > 0 |
| && brandSet.count(FOURCC('h', 'e', 'i', 'c')) > 0) { |
| mIsHEIF = true; |
| ALOGV("identified HEIF image"); |
| } |
| |
| *offset = stop_offset; |
| |
| break; |
| } |
| |
| default: |
| { |
| // check if we're parsing 'ilst' for meta keys |
| // if so, treat type as a number (key-id). |
| if (underQTMetaPath(mPath, 3)) { |
| status_t err = parseQTMetaVal(chunk_type, data_offset, chunk_data_size); |
| if (err != OK) { |
| return err; |
| } |
| } |
| |
| *offset += chunk_size; |
| break; |
| } |
| } |
| |
| return OK; |
| } |
| |
| status_t MPEG4Extractor::parseAC3SampleEntry(off64_t offset) { |
| // skip 16 bytes: |
| // + 6-byte reserved, |
| // + 2-byte data reference index, |
| // + 8-byte reserved |
| offset += 16; |
| uint16_t channelCount; |
| if (!mDataSource->getUInt16(offset, &channelCount)) { |
| return ERROR_MALFORMED; |
| } |
| // skip 8 bytes: |
| // + 2-byte channelCount, |
| // + 2-byte sample size, |
| // + 4-byte reserved |
| offset += 8; |
| uint16_t sampleRate; |
| if (!mDataSource->getUInt16(offset, &sampleRate)) { |
| ALOGE("MPEG4Extractor: error while reading ac-3 block: cannot read sample rate"); |
| return ERROR_MALFORMED; |
| } |
| |
| // skip 4 bytes: |
| // + 2-byte sampleRate, |
| // + 2-byte reserved |
| offset += 4; |
| return parseAC3SpecificBox(offset, sampleRate); |
| } |
| |
| status_t MPEG4Extractor::parseAC3SpecificBox( |
| off64_t offset, uint16_t sampleRate) { |
| uint32_t size; |
| // + 4-byte size |
| // + 4-byte type |
| // + 3-byte payload |
| const uint32_t kAC3SpecificBoxSize = 11; |
| if (!mDataSource->getUInt32(offset, &size) || size < kAC3SpecificBoxSize) { |
| ALOGE("MPEG4Extractor: error while reading ac-3 block: cannot read specific box size"); |
| return ERROR_MALFORMED; |
| } |
| |
| offset += 4; |
| uint32_t type; |
| if (!mDataSource->getUInt32(offset, &type) || type != FOURCC('d', 'a', 'c', '3')) { |
| ALOGE("MPEG4Extractor: error while reading ac-3 specific block: header not dac3"); |
| return ERROR_MALFORMED; |
| } |
| |
| offset += 4; |
| const uint32_t kAC3SpecificBoxPayloadSize = 3; |
| uint8_t chunk[kAC3SpecificBoxPayloadSize]; |
| if (mDataSource->readAt(offset, chunk, sizeof(chunk)) != sizeof(chunk)) { |
| ALOGE("MPEG4Extractor: error while reading ac-3 specific block: bitstream fields"); |
| return ERROR_MALFORMED; |
| } |
| |
| ABitReader br(chunk, sizeof(chunk)); |
| static const unsigned channelCountTable[] = {2, 1, 2, 3, 3, 4, 4, 5}; |
| static const unsigned sampleRateTable[] = {48000, 44100, 32000}; |
| |
| unsigned fscod = br.getBits(2); |
| if (fscod == 3) { |
| ALOGE("Incorrect fscod (3) in AC3 header"); |
| return ERROR_MALFORMED; |
| } |
| unsigned boxSampleRate = sampleRateTable[fscod]; |
| if (boxSampleRate != sampleRate) { |
| ALOGE("sample rate mismatch: boxSampleRate = %d, sampleRate = %d", |
| boxSampleRate, sampleRate); |
| return ERROR_MALFORMED; |
| } |
| |
| unsigned bsid = br.getBits(5); |
| if (bsid > 8) { |
| ALOGW("Incorrect bsid in AC3 header. Possibly E-AC-3?"); |
| return ERROR_MALFORMED; |
| } |
| |
| // skip |
| unsigned bsmod __unused = br.getBits(3); |
| |
| unsigned acmod = br.getBits(3); |
| unsigned lfeon = br.getBits(1); |
| unsigned channelCount = channelCountTable[acmod] + lfeon; |
| |
| if (mLastTrack == NULL) { |
| return ERROR_MALFORMED; |
| } |
| mLastTrack->meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_AUDIO_AC3); |
| mLastTrack->meta->setInt32(kKeyChannelCount, channelCount); |
| mLastTrack->meta->setInt32(kKeySampleRate, sampleRate); |
| return OK; |
| } |
| |
| status_t MPEG4Extractor::parseSegmentIndex(off64_t offset, size_t size) { |
| ALOGV("MPEG4Extractor::parseSegmentIndex"); |
| |
| if (size < 12) { |
| return -EINVAL; |
| } |
| |
| uint32_t flags; |
| if (!mDataSource->getUInt32(offset, &flags)) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint32_t version = flags >> 24; |
| flags &= 0xffffff; |
| |
| ALOGV("sidx version %d", version); |
| |
| uint32_t referenceId; |
| if (!mDataSource->getUInt32(offset + 4, &referenceId)) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint32_t timeScale; |
| if (!mDataSource->getUInt32(offset + 8, &timeScale)) { |
| return ERROR_MALFORMED; |
| } |
| ALOGV("sidx refid/timescale: %d/%d", referenceId, timeScale); |
| if (timeScale == 0) |
| return ERROR_MALFORMED; |
| |
| uint64_t earliestPresentationTime; |
| uint64_t firstOffset; |
| |
| offset += 12; |
| size -= 12; |
| |
| if (version == 0) { |
| if (size < 8) { |
| return -EINVAL; |
| } |
| uint32_t tmp; |
| if (!mDataSource->getUInt32(offset, &tmp)) { |
| return ERROR_MALFORMED; |
| } |
| earliestPresentationTime = tmp; |
| if (!mDataSource->getUInt32(offset + 4, &tmp)) { |
| return ERROR_MALFORMED; |
| } |
| firstOffset = tmp; |
| offset += 8; |
| size -= 8; |
| } else { |
| if (size < 16) { |
| return -EINVAL; |
| } |
| if (!mDataSource->getUInt64(offset, &earliestPresentationTime)) { |
| return ERROR_MALFORMED; |
| } |
| if (!mDataSource->getUInt64(offset + 8, &firstOffset)) { |
| return ERROR_MALFORMED; |
| } |
| offset += 16; |
| size -= 16; |
| } |
| ALOGV("sidx pres/off: %" PRIu64 "/%" PRIu64, earliestPresentationTime, firstOffset); |
| |
| if (size < 4) { |
| return -EINVAL; |
| } |
| |
| uint16_t referenceCount; |
| if (!mDataSource->getUInt16(offset + 2, &referenceCount)) { |
| return ERROR_MALFORMED; |
| } |
| offset += 4; |
| size -= 4; |
| ALOGV("refcount: %d", referenceCount); |
| |
| if (size < referenceCount * 12) { |
| return -EINVAL; |
| } |
| |
| uint64_t total_duration = 0; |
| for (unsigned int i = 0; i < referenceCount; i++) { |
| uint32_t d1, d2, d3; |
| |
| if (!mDataSource->getUInt32(offset, &d1) || // size |
| !mDataSource->getUInt32(offset + 4, &d2) || // duration |
| !mDataSource->getUInt32(offset + 8, &d3)) { // flags |
| return ERROR_MALFORMED; |
| } |
| |
| if (d1 & 0x80000000) { |
| ALOGW("sub-sidx boxes not supported yet"); |
| } |
| bool sap = d3 & 0x80000000; |
| uint32_t saptype = (d3 >> 28) & 7; |
| if (!sap || (saptype != 1 && saptype != 2)) { |
| // type 1 and 2 are sync samples |
| ALOGW("not a stream access point, or unsupported type: %08x", d3); |
| } |
| total_duration += d2; |
| offset += 12; |
| ALOGV(" item %d, %08x %08x %08x", i, d1, d2, d3); |
| SidxEntry se; |
| se.mSize = d1 & 0x7fffffff; |
| se.mDurationUs = 1000000LL * d2 / timeScale; |
| mSidxEntries.add(se); |
| } |
| |
| uint64_t sidxDuration = total_duration * 1000000 / timeScale; |
| |
| if (mLastTrack == NULL) |
| return ERROR_MALFORMED; |
| |
| int64_t metaDuration; |
| if (!mLastTrack->meta->findInt64(kKeyDuration, &metaDuration) || metaDuration == 0) { |
| mLastTrack->meta->setInt64(kKeyDuration, sidxDuration); |
| } |
| return OK; |
| } |
| |
| status_t MPEG4Extractor::parseQTMetaKey(off64_t offset, size_t size) { |
| if (size < 8) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint32_t count; |
| if (!mDataSource->getUInt32(offset + 4, &count)) { |
| return ERROR_MALFORMED; |
| } |
| |
| if (mMetaKeyMap.size() > 0) { |
| ALOGW("'keys' atom seen again, discarding existing entries"); |
| mMetaKeyMap.clear(); |
| } |
| |
| off64_t keyOffset = offset + 8; |
| off64_t stopOffset = offset + size; |
| for (size_t i = 1; i <= count; i++) { |
| if (keyOffset + 8 > stopOffset) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint32_t keySize; |
| if (!mDataSource->getUInt32(keyOffset, &keySize) |
| || keySize < 8 |
| || keyOffset + keySize > stopOffset) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint32_t type; |
| if (!mDataSource->getUInt32(keyOffset + 4, &type) |
| || type != FOURCC('m', 'd', 't', 'a')) { |
| return ERROR_MALFORMED; |
| } |
| |
| keySize -= 8; |
| keyOffset += 8; |
| |
| sp<ABuffer> keyData = new ABuffer(keySize); |
| if (keyData->data() == NULL) { |
| return ERROR_MALFORMED; |
| } |
| if (mDataSource->readAt( |
| keyOffset, keyData->data(), keySize) < (ssize_t) keySize) { |
| return ERROR_MALFORMED; |
| } |
| |
| AString key((const char *)keyData->data(), keySize); |
| mMetaKeyMap.add(i, key); |
| |
| keyOffset += keySize; |
| } |
| return OK; |
| } |
| |
| status_t MPEG4Extractor::parseQTMetaVal( |
| int32_t keyId, off64_t offset, size_t size) { |
| ssize_t index = mMetaKeyMap.indexOfKey(keyId); |
| if (index < 0) { |
| // corresponding key is not present, ignore |
| return ERROR_MALFORMED; |
| } |
| |
| if (size <= 16) { |
| return ERROR_MALFORMED; |
| } |
| uint32_t dataSize; |
| if (!mDataSource->getUInt32(offset, &dataSize) |
| || dataSize > size || dataSize <= 16) { |
| return ERROR_MALFORMED; |
| } |
| uint32_t atomFourCC; |
| if (!mDataSource->getUInt32(offset + 4, &atomFourCC) |
| || atomFourCC != FOURCC('d', 'a', 't', 'a')) { |
| return ERROR_MALFORMED; |
| } |
| uint32_t dataType; |
| if (!mDataSource->getUInt32(offset + 8, &dataType) |
| || ((dataType & 0xff000000) != 0)) { |
| // not well-known type |
| return ERROR_MALFORMED; |
| } |
| |
| dataSize -= 16; |
| offset += 16; |
| |
| if (dataType == 23 && dataSize >= 4) { |
| // BE Float32 |
| uint32_t val; |
| if (!mDataSource->getUInt32(offset, &val)) { |
| return ERROR_MALFORMED; |
| } |
| if (!strcasecmp(mMetaKeyMap[index].c_str(), "com.android.capture.fps")) { |
| mFileMetaData->setFloat(kKeyCaptureFramerate, *(float *)&val); |
| } |
| } else if (dataType == 67 && dataSize >= 4) { |
| // BE signed int32 |
| uint32_t val; |
| if (!mDataSource->getUInt32(offset, &val)) { |
| return ERROR_MALFORMED; |
| } |
| if (!strcasecmp(mMetaKeyMap[index].c_str(), "com.android.video.temporal_layers_count")) { |
| mFileMetaData->setInt32(kKeyTemporalLayerCount, val); |
| } |
| } else { |
| // add more keys if needed |
| ALOGV("ignoring key: type %d, size %d", dataType, dataSize); |
| } |
| |
| return OK; |
| } |
| |
| status_t MPEG4Extractor::parseTrackHeader( |
| off64_t data_offset, off64_t data_size) { |
| if (data_size < 4) { |
| return ERROR_MALFORMED; |
| } |
| |
| uint8_t version; |
| if (mDataSource->readAt(data_offset, &version, 1) < 1) { |
| return ERROR_IO; |
| } |
| |
| size_t dynSize = (version == 1) ? 36 : 24; |
| |
| uint8_t buffer[36 + 60]; |
| |
| if (data_size != (off64_t)dynSize + 60) { |
| return ERROR_MALFORMED; |
| } |
| |
| if (mDataSource->readAt( |
| data_offset, buffer, data_size) < (ssize_t)data_size) { |
| return ERROR_IO; |
| } |
| |
| uint64_t ctime __unused, mtime __unused, duration __unused; |
| int32_t id; |
| |
| if (version == 1) { |
| ctime = U64_AT(&buffer[4]); |
| mtime = U64_AT(&buffer[12]); |
| id = U32_AT(&buffer[20]); |
| duration |