/*
* Copyright 2022 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.
*/
package androidx.media3.extractor.avi;
import static java.lang.annotation.ElementType.TYPE_USE;
import androidx.annotation.IntDef;
import androidx.annotation.Nullable;
import androidx.media3.common.C;
import androidx.media3.common.Format;
import androidx.media3.common.MimeTypes;
import androidx.media3.common.ParserException;
import androidx.media3.common.util.Assertions;
import androidx.media3.common.util.Log;
import androidx.media3.common.util.ParsableByteArray;
import androidx.media3.common.util.UnstableApi;
import androidx.media3.extractor.DummyExtractorOutput;
import androidx.media3.extractor.Extractor;
import androidx.media3.extractor.ExtractorInput;
import androidx.media3.extractor.ExtractorOutput;
import androidx.media3.extractor.PositionHolder;
import androidx.media3.extractor.SeekMap;
import androidx.media3.extractor.TrackOutput;
import java.io.IOException;
import java.lang.annotation.Documented;
import java.lang.annotation.Retention;
import java.lang.annotation.RetentionPolicy;
import java.lang.annotation.Target;
import java.util.ArrayList;
import org.checkerframework.checker.nullness.qual.MonotonicNonNull;
/**
* Extracts data from the AVI container format.
*
* <p>Spec: https://docs.microsoft.com/en-us/windows/win32/directshow/avi-riff-file-reference.
*/
@UnstableApi
public final class AviExtractor implements Extractor {
private static final String TAG = "AviExtractor";
public static final int FOURCC_RIFF = 0x46464952;
public static final int FOURCC_AVI_ = 0x20495641; // AVI<space>
public static final int FOURCC_LIST = 0x5453494c;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_avih = 0x68697661;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_hdrl = 0x6c726468;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_strl = 0x6c727473;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_movi = 0x69766f6d;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_idx1 = 0x31786469;
public static final int FOURCC_JUNK = 0x4b4e554a;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_strf = 0x66727473;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_strn = 0x6e727473;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_strh = 0x68727473;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_auds = 0x73647561;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_txts = 0x73747874;
@SuppressWarnings("ConstantCaseForConstants")
public static final int FOURCC_vids = 0x73646976;
/** Parser states. */
@Documented
@Retention(RetentionPolicy.SOURCE)
@Target(TYPE_USE)
@IntDef({
STATE_SKIPPING_TO_HDRL,
STATE_READING_HDRL_HEADER,
STATE_READING_HDRL_BODY,
STATE_FINDING_MOVI_HEADER,
STATE_FINDING_IDX1_HEADER,
STATE_READING_IDX1_BODY,
STATE_READING_SAMPLES,
})
private @interface State {}
private static final int STATE_SKIPPING_TO_HDRL = 0;
private static final int STATE_READING_HDRL_HEADER = 1;
private static final int STATE_READING_HDRL_BODY = 2;
private static final int STATE_FINDING_MOVI_HEADER = 3;
private static final int STATE_FINDING_IDX1_HEADER = 4;
private static final int STATE_READING_IDX1_BODY = 5;
private static final int STATE_READING_SAMPLES = 6;
private static final int AVIIF_KEYFRAME = 16;
/**
* Maximum size to skip using {@link ExtractorInput#skip}. Boxes larger than this size are skipped
* using {@link #RESULT_SEEK}.
*/
private static final long RELOAD_MINIMUM_SEEK_DISTANCE = 256 * 1024;
private final ParsableByteArray scratch;
private final ChunkHeaderHolder chunkHeaderHolder;
private @State int state;
private ExtractorOutput extractorOutput;
private @MonotonicNonNull AviMainHeaderChunk aviHeader;
private long durationUs;
private ChunkReader[] chunkReaders;
private long pendingReposition;
@Nullable private ChunkReader currentChunkReader;
private int hdrlSize;
private long moviStart;
private long moviEnd;
private int idx1BodySize;
private boolean seekMapHasBeenOutput;
public AviExtractor() {
scratch = new ParsableByteArray(/* limit= */ 12);
chunkHeaderHolder = new ChunkHeaderHolder();
extractorOutput = new DummyExtractorOutput();
chunkReaders = new ChunkReader[0];
moviStart = C.POSITION_UNSET;
moviEnd = C.POSITION_UNSET;
hdrlSize = C.LENGTH_UNSET;
durationUs = C.TIME_UNSET;
}
// Extractor implementation.
@Override
public void init(ExtractorOutput output) {
this.state = STATE_SKIPPING_TO_HDRL;
this.extractorOutput = output;
pendingReposition = C.POSITION_UNSET;
}
@Override
public boolean sniff(ExtractorInput input) throws IOException {
input.peekFully(scratch.getData(), /* offset= */ 0, /* length= */ 12);
scratch.setPosition(0);
if (scratch.readLittleEndianInt() != FOURCC_RIFF) {
return false;
}
scratch.skipBytes(4); // Skip the RIFF chunk length.
return scratch.readLittleEndianInt() == FOURCC_AVI_;
}
@Override
public int read(ExtractorInput input, PositionHolder positionHolder) throws IOException {
if (resolvePendingReposition(input, positionHolder)) {
return RESULT_SEEK;
}
switch (state) {
case STATE_SKIPPING_TO_HDRL:
// Check for RIFF and AVI fourcc's just in case the caller did not sniff, in order to
// provide a meaningful error if the input is not an AVI file.
if (sniff(input)) {
input.skipFully(/* length= */ 12);
} else {
throw ParserException.createForMalformedContainer(
/* message= */ "AVI Header List not found", /* cause= */ null);
}
state = STATE_READING_HDRL_HEADER;
return RESULT_CONTINUE;
case STATE_READING_HDRL_HEADER:
input.readFully(scratch.getData(), /* offset= */ 0, /* length= */ 12);
scratch.setPosition(0);
chunkHeaderHolder.populateWithListHeaderFrom(scratch);
if (chunkHeaderHolder.listType != FOURCC_hdrl) {
throw ParserException.createForMalformedContainer(
/* message= */ "hdrl expected, found: " + chunkHeaderHolder.listType,
/* cause= */ null);
}
hdrlSize = chunkHeaderHolder.size;
state = STATE_READING_HDRL_BODY;
return RESULT_CONTINUE;
case STATE_READING_HDRL_BODY:
// hdrlSize includes the LIST type (hdrl), so we subtract 4 to the size.
int bytesToRead = hdrlSize - 4;
ParsableByteArray hdrlBody = new ParsableByteArray(bytesToRead);
input.readFully(hdrlBody.getData(), /* offset= */ 0, bytesToRead);
parseHdrlBody(hdrlBody);
state = STATE_FINDING_MOVI_HEADER;
return RESULT_CONTINUE;
case STATE_FINDING_MOVI_HEADER:
if (moviStart != C.POSITION_UNSET && input.getPosition() != moviStart) {
pendingReposition = moviStart;
return RESULT_CONTINUE;
}
input.peekFully(scratch.getData(), /* offset= */ 0, /* length= */ 12);
input.resetPeekPosition();
scratch.setPosition(0);
chunkHeaderHolder.populateFrom(scratch);
int listType = scratch.readLittleEndianInt();
if (chunkHeaderHolder.chunkType == FOURCC_RIFF) {
// We are at the start of the file. The movi chunk is in the RIFF chunk, so we skip the
// header, so as to read the RIFF chunk's body.
input.skipFully(12);
return RESULT_CONTINUE;
}
if (chunkHeaderHolder.chunkType != FOURCC_LIST || listType != FOURCC_movi) {
// The chunk header (8 bytes) plus the whole body.
pendingReposition = input.getPosition() + chunkHeaderHolder.size + 8;
return RESULT_CONTINUE;
}
moviStart = input.getPosition();
// Size includes the list type, but not the LIST or size fields, so we add 8.
moviEnd = moviStart + chunkHeaderHolder.size + 8;
if (!seekMapHasBeenOutput) {
if (Assertions.checkNotNull(aviHeader).hasIndex()) {
state = STATE_FINDING_IDX1_HEADER;
pendingReposition = moviEnd;
return RESULT_CONTINUE;
} else {
extractorOutput.seekMap(new SeekMap.Unseekable(durationUs));
seekMapHasBeenOutput = true;
}
}
// No need to parse the idx1, so we start reading the samples from the movi chunk straight
// away. We skip 12 bytes to move to the start of the movi's body.
pendingReposition = input.getPosition() + 12;
state = STATE_READING_SAMPLES;
return RESULT_CONTINUE;
case STATE_FINDING_IDX1_HEADER:
input.readFully(scratch.getData(), /* offset= */ 0, /* length= */ 8);
scratch.setPosition(0);
int idx1Fourcc = scratch.readLittleEndianInt();
int boxSize = scratch.readLittleEndianInt();
if (idx1Fourcc == FOURCC_idx1) {
state = STATE_READING_IDX1_BODY;
idx1BodySize = boxSize;
} else {
// This one is not idx1, skip to the next box.
pendingReposition = input.getPosition() + boxSize;
}
return RESULT_CONTINUE;
case STATE_READING_IDX1_BODY:
ParsableByteArray idx1Body = new ParsableByteArray(idx1BodySize);
input.readFully(idx1Body.getData(), /* offset= */ 0, /* length= */ idx1BodySize);
parseIdx1Body(idx1Body);
state = STATE_READING_SAMPLES;
pendingReposition = moviStart;
return RESULT_CONTINUE;
case STATE_READING_SAMPLES:
return readMoviChunks(input);
default:
throw new AssertionError(); // Should never happen.
}
}
@Override
public void seek(long position, long timeUs) {
pendingReposition = C.POSITION_UNSET;
currentChunkReader = null;
for (ChunkReader chunkReader : chunkReaders) {
chunkReader.seekToPosition(position);
}
if (position == 0) {
if (chunkReaders.length == 0) {
// Still unprepared.
state = STATE_SKIPPING_TO_HDRL;
} else {
state = STATE_FINDING_MOVI_HEADER;
}
return;
}
state = STATE_READING_SAMPLES;
}
@Override
public void release() {
// Nothing to release.
}
// Internal methods.
/**
* Returns whether a {@link #RESULT_SEEK} is required for the pending reposition. A seek may not
* be necessary when the desired position (as held by {@link #pendingReposition}) is after the
* {@link ExtractorInput#getPosition() current position}, but not further than {@link
* #RELOAD_MINIMUM_SEEK_DISTANCE}.
*/
private boolean resolvePendingReposition(ExtractorInput input, PositionHolder positionHolder)
throws IOException {
boolean needSeek = false;
if (pendingReposition != C.POSITION_UNSET) {
long currentPosition = input.getPosition();
if (pendingReposition < currentPosition
|| pendingReposition > currentPosition + RELOAD_MINIMUM_SEEK_DISTANCE) {
positionHolder.position = pendingReposition;
needSeek = true;
} else {
// The distance to the target position is short enough that it makes sense to just skip the
// bytes, instead of doing a seek which might re-create an HTTP connection.
input.skipFully((int) (pendingReposition - currentPosition));
}
}
pendingReposition = C.POSITION_UNSET;
return needSeek;
}
private void parseHdrlBody(ParsableByteArray hrdlBody) throws IOException {
ListChunk headerList = ListChunk.parseFrom(FOURCC_hdrl, hrdlBody);
if (headerList.getType() != FOURCC_hdrl) {
throw ParserException.createForMalformedContainer(
/* message= */ "Unexpected header list type " + headerList.getType(), /* cause= */ null);
}
@Nullable AviMainHeaderChunk aviHeader = headerList.getChild(AviMainHeaderChunk.class);
if (aviHeader == null) {
throw ParserException.createForMalformedContainer(
/* message= */ "AviHeader not found", /* cause= */ null);
}
this.aviHeader = aviHeader;
// This is usually wrong, so it will be overwritten by video if present
durationUs = aviHeader.totalFrames * (long) aviHeader.frameDurationUs;
ArrayList<ChunkReader> chunkReaderList = new ArrayList<>();
int streamId = 0;
for (AviChunk aviChunk : headerList.children) {
if (aviChunk.getType() == FOURCC_strl) {
ListChunk streamList = (ListChunk) aviChunk;
// Note the streamId needs to increment even if the corresponding `strl` is discarded.
// See
// https://docs.microsoft.com/en-us/windows/win32/directshow/avi-riff-file-reference#avi-stream-headers.
@Nullable ChunkReader chunkReader = processStreamList(streamList, streamId++);
if (chunkReader != null) {
chunkReaderList.add(chunkReader);
}
}
}
chunkReaders = chunkReaderList.toArray(new ChunkReader[0]);
extractorOutput.endTracks();
}
/** Builds and outputs the {@link SeekMap} from the idx1 chunk. */
private void parseIdx1Body(ParsableByteArray body) {
long seekOffset = peekSeekOffset(body);
while (body.bytesLeft() >= 16) {
int chunkId = body.readLittleEndianInt();
int flags = body.readLittleEndianInt();
long offset = body.readLittleEndianInt() + seekOffset;
body.readLittleEndianInt(); // We ignore the size.
ChunkReader chunkReader = getChunkReader(chunkId);
if (chunkReader == null) {
// We ignore unknown chunk IDs.
continue;
}
if ((flags & AVIIF_KEYFRAME) == AVIIF_KEYFRAME) {
chunkReader.appendKeyFrameToIndex(offset);
}
chunkReader.incrementIndexChunkCount();
}
for (ChunkReader chunkReader : chunkReaders) {
chunkReader.compactIndex();
}
seekMapHasBeenOutput = true;
extractorOutput.seekMap(new AviSeekMap(durationUs));
}
private long peekSeekOffset(ParsableByteArray idx1Body) {
// The spec states the offset is based on the start of the movi list type fourcc, but it also
// says some files base the offset on the start of the file. We use a best effort approach to
// figure out which is the case. See:
// https://docs.microsoft.com/en-us/previous-versions/windows/desktop/api/Aviriff/ns-aviriff-avioldindex#dwoffset.
if (idx1Body.bytesLeft() < 16) {
// There are no full entries in the index, meaning we don't need to apply an offset.
return 0;
}
int startingPosition = idx1Body.getPosition();
idx1Body.skipBytes(8); // Skip chunkId (4 bytes) and flags (4 bytes).
int offset = idx1Body.readLittleEndianInt();
// moviStart poitns at the start of the LIST, while the seek offset is based at the start of the
// movi fourCC, so we add 8 to reconcile the difference.
long seekOffset = offset > moviStart ? 0L : moviStart + 8;
idx1Body.setPosition(startingPosition);
return seekOffset;
}
@Nullable
private ChunkReader getChunkReader(int chunkId) {
for (ChunkReader chunkReader : chunkReaders) {
if (chunkReader.handlesChunkId(chunkId)) {
return chunkReader;
}
}
return null;
}
private int readMoviChunks(ExtractorInput input) throws IOException {
if (input.getPosition() >= moviEnd) {
return C.RESULT_END_OF_INPUT;
} else if (currentChunkReader != null) {
if (currentChunkReader.onChunkData(input)) {
currentChunkReader = null;
}
} else {
alignInputToEvenPosition(input);
input.peekFully(scratch.getData(), /* offset= */ 0, 12);
scratch.setPosition(0);
int chunkType = scratch.readLittleEndianInt();
if (chunkType == FOURCC_LIST) {
scratch.setPosition(8);
int listType = scratch.readLittleEndianInt();
input.skipFully(listType == FOURCC_movi ? 12 : 8);
input.resetPeekPosition();
return RESULT_CONTINUE;
}
int size = scratch.readLittleEndianInt();
if (chunkType == FOURCC_JUNK) {
pendingReposition = input.getPosition() + size + 8;
return RESULT_CONTINUE;
}
input.skipFully(8);
input.resetPeekPosition();
ChunkReader chunkReader = getChunkReader(chunkType);
if (chunkReader == null) {
// No handler for this chunk. We skip it.
pendingReposition = input.getPosition() + size;
return RESULT_CONTINUE;
} else {
chunkReader.onChunkStart(size);
this.currentChunkReader = chunkReader;
}
}
return RESULT_CONTINUE;
}
@Nullable
private ChunkReader processStreamList(ListChunk streamList, int streamId) {
AviStreamHeaderChunk aviStreamHeaderChunk = streamList.getChild(AviStreamHeaderChunk.class);
StreamFormatChunk streamFormatChunk = streamList.getChild(StreamFormatChunk.class);
if (aviStreamHeaderChunk == null) {
Log.w(TAG, "Missing Stream Header");
return null;
}
if (streamFormatChunk == null) {
Log.w(TAG, "Missing Stream Format");
return null;
}
long durationUs = aviStreamHeaderChunk.getDurationUs();
Format streamFormat = streamFormatChunk.format;
Format.Builder builder = streamFormat.buildUpon();
builder.setId(streamId);
int suggestedBufferSize = aviStreamHeaderChunk.suggestedBufferSize;
if (suggestedBufferSize != 0) {
builder.setMaxInputSize(suggestedBufferSize);
}
StreamNameChunk streamName = streamList.getChild(StreamNameChunk.class);
if (streamName != null) {
builder.setLabel(streamName.name);
}
int trackType = MimeTypes.getTrackType(streamFormat.sampleMimeType);
if (trackType == C.TRACK_TYPE_AUDIO || trackType == C.TRACK_TYPE_VIDEO) {
TrackOutput trackOutput = extractorOutput.track(streamId, trackType);
trackOutput.format(builder.build());
ChunkReader chunkReader =
new ChunkReader(
streamId, trackType, durationUs, aviStreamHeaderChunk.length, trackOutput);
this.durationUs = durationUs;
return chunkReader;
} else {
// We don't currently support tracks other than video and audio.
return null;
}
}
/**
* Skips one byte from the given {@code input} if the current position is odd.
*
* <p>This isn't documented anywhere, but AVI files are aligned to even bytes and fill gaps with
* zeros.
*/
private static void alignInputToEvenPosition(ExtractorInput input) throws IOException {
if ((input.getPosition() & 1) == 1) {
input.skipFully(1);
}
}
// Internal classes.
private class AviSeekMap implements SeekMap {
private final long durationUs;
public AviSeekMap(long durationUs) {
this.durationUs = durationUs;
}
@Override
public boolean isSeekable() {
return true;
}
@Override
public long getDurationUs() {
return durationUs;
}
@Override
public SeekPoints getSeekPoints(long timeUs) {
SeekPoints result = chunkReaders[0].getSeekPoints(timeUs);
for (int i = 1; i < chunkReaders.length; i++) {
SeekPoints seekPoints = chunkReaders[i].getSeekPoints(timeUs);
if (seekPoints.first.position < result.first.position) {
result = seekPoints;
}
}
return result;
}
}
private static class ChunkHeaderHolder {
public int chunkType;
public int size;
public int listType;
public void populateWithListHeaderFrom(ParsableByteArray headerBytes) throws ParserException {
populateFrom(headerBytes);
if (chunkType != AviExtractor.FOURCC_LIST) {
throw ParserException.createForMalformedContainer(
/* message= */ "LIST expected, found: " + chunkType, /* cause= */ null);
}
listType = headerBytes.readLittleEndianInt();
}
public void populateFrom(ParsableByteArray headerBytes) {
chunkType = headerBytes.readLittleEndianInt();
size = headerBytes.readLittleEndianInt();
listType = 0;
}
}
}