Implement MP3 demuxer

src/index.ts

@@ -74,12 +74,14 @@ export {
 	QuickTimeInputFormat,
 	MatroskaInputFormat,
 	WebMInputFormat,
+	Mp3InputFormat,
 	WaveInputFormat,
 	ALL_FORMATS,
 	MP4,
 	QTFF,
 	MATROSKA,
 	WEBM,
+	MP3,
 	WAVE,
 } from './input-format';
 export { Input, InputOptions } from './input';

src/input-format.ts

@@ -4,6 +4,8 @@ import { IsobmffDemuxer } from './isobmff/isobmff-demuxer';
 import { IsobmffReader } from './isobmff/isobmff-reader';
 import { EBMLId, EBMLReader } from './matroska/ebml';
 import { MatroskaDemuxer } from './matroska/matroska-demuxer';
+import { Mp3Demuxer } from './mp3/mp3-demuxer';
+import { Mp3Reader } from './mp3/mp3-reader';
 import { RiffReader } from './wave/riff-reader';
 import { WaveDemuxer } from './wave/wave-demuxer';
 
@@ -210,6 +212,42 @@ export class WaveInputFormat extends InputFormat {
 	}
 }
 
+/** @public */
+export class Mp3InputFormat extends InputFormat {
+	async _canReadInput(input: Input) {
+		const sourceSize = await input._mainReader.source._getSize();
+		if (sourceSize < 4) {
+			return false;
+		}
+
+		const mp3Reader = new Mp3Reader(input._mainReader);
+		mp3Reader.fileSize = sourceSize;
+
+		const id3Tag = mp3Reader.readId3();
+
+		if (id3Tag) {
+			mp3Reader.pos += id3Tag.size;
+		}
+
+		await mp3Reader.reader.loadRange(mp3Reader.pos, mp3Reader.pos + 4096);
+
+		return mp3Reader.readNextFrameHeader(mp3Reader.pos + 4096) !== null;
+	}
+
+	/** @internal */
+	_createDemuxer(input: Input) {
+		return new Mp3Demuxer(input);
+	}
+
+	getName() {
+		return 'MP3';
+	}
+
+	getMimeType() {
+		return 'audio/mpeg';
+	}
+}
+
 /** @public */
 export const MP4 = new Mp4InputFormat();
 /** @public */
@@ -219,7 +257,9 @@ export const MATROSKA = new MatroskaInputFormat();
 /** @public */
 export const WEBM = new WebMInputFormat();
 /** @public */
+export const MP3 = new Mp3InputFormat();
+/** @public */
 export const WAVE = new WaveInputFormat();
 
 /** @public */
-export const ALL_FORMATS: InputFormat[] = [MP4, QTFF, MATROSKA, WEBM, WAVE];
+export const ALL_FORMATS: InputFormat[] = [MP4, QTFF, MATROSKA, WEBM, MP3, WAVE];

src/mp3/mp3-demuxer.ts

@@ -0,0 +1,217 @@
+import { AudioCodec } from '../codec';
+import { Demuxer } from '../demuxer';
+import { Input } from '../input';
+import { InputAudioTrack, InputAudioTrackBacking } from '../input-track';
+import { SampleRetrievalOptions } from '../media-sink';
+import { assert, binarySearchExact, binarySearchLessOrEqual, last, UNDETERMINED_LANGUAGE } from '../misc';
+import { EncodedAudioSample, PLACEHOLDER_DATA } from '../sample';
+import { FrameHeader, Mp3Reader } from './mp3-reader';
+
+const AUDIO_SAMPLES_PER_FRAME = 1152;
+
+type Sample = {
+	timestamp: number;
+	duration: number;
+	dataStart: number;
+	dataSize: number;
+};
+
+export class Mp3Demuxer extends Demuxer {
+	reader: Mp3Reader;
+
+	metadataPromise: Promise<void> | null = null;
+	firstFrameHeader: FrameHeader | null = null;
+	allSamples: Sample[] = [];
+
+	tracks: InputAudioTrack[] = [];
+
+	constructor(input: Input) {
+		super(input);
+
+		this.reader = new Mp3Reader(input._mainReader);
+	}
+
+	async readMetadata() {
+		return this.metadataPromise ??= (async () => {
+			const fileSize = await this.input._source._getSize();
+			this.reader.fileSize = fileSize;
+
+			// Just load the entire file. Primitive, but the only way to actually ensure 100% correct timestamps.
+			// Random access in MP3 can be flaky and unreliable.
+			await this.reader.reader.loadRange(0, fileSize);
+
+			const id3Tag = this.reader.readId3();
+			if (id3Tag) {
+				this.reader.pos += id3Tag.size;
+			}
+
+			let nextTimestamp = 0;
+
+			// Let's read all samples
+			while (true) {
+				const header = this.reader.readNextFrameHeader();
+				if (!header) {
+					break;
+				}
+
+				const xingOffset = header.mpegVersionId === 3
+					? (header.channelCount === 1 ? 21 : 36)
+					: (header.channelCount === 1 ? 13 : 21);
+				this.reader.pos = header.startPos + xingOffset;
+				const word = this.reader.readU32();
+				const isXing = word === 0x58696e67 // 'Xing'
+					|| word === 0x496e666f; // 'Info'
+
+				this.reader.pos = header.startPos + header.totalSize - 1; // -1 in case the frame is 1 byte too short
+
+				if (isXing) {
+					// There's no actual audio data in this frame, so let's skip it
+					continue;
+				}
+
+				if (!this.firstFrameHeader) {
+					this.firstFrameHeader = header;
+				}
+
+				const sampleDuration = AUDIO_SAMPLES_PER_FRAME / header.sampleRate;
+				const sample: Sample = {
+					timestamp: nextTimestamp,
+					duration: sampleDuration,
+					dataStart: header.startPos,
+					dataSize: header.totalSize,
+				};
+
+				this.allSamples.push(sample);
+				nextTimestamp += sampleDuration;
+			}
+
+			if (!this.firstFrameHeader) {
+				throw new Error('No MP3 frames found.');
+			}
+
+			this.tracks = [new InputAudioTrack(new Mp3AudioTrackBacking(this))];
+		})();
+	}
+
+	async getMimeType() {
+		return 'audio/mpeg';
+	}
+
+	async getTracks() {
+		await this.readMetadata();
+		return this.tracks;
+	}
+
+	async computeDuration() {
+		await this.readMetadata();
+
+		const lastSample = last(this.allSamples);
+		assert(lastSample);
+
+		return lastSample.timestamp + lastSample.duration;
+	}
+}
+
+class Mp3AudioTrackBacking implements InputAudioTrackBacking {
+	constructor(public demuxer: Mp3Demuxer) {}
+
+	getId() {
+		return 1;
+	}
+
+	async getFirstTimestamp() {
+		return 0;
+	}
+
+	computeDuration() {
+		return this.demuxer.computeDuration();
+	}
+
+	async getLanguageCode() {
+		return UNDETERMINED_LANGUAGE;
+	}
+
+	async getCodec(): Promise<AudioCodec> {
+		return 'mp3';
+	}
+
+	async getNumberOfChannels() {
+		assert(this.demuxer.firstFrameHeader);
+		return this.demuxer.firstFrameHeader.channelCount;
+	}
+
+	async getSampleRate() {
+		assert(this.demuxer.firstFrameHeader);
+		return this.demuxer.firstFrameHeader.sampleRate;
+	}
+
+	async getDecoderConfig(): Promise<AudioDecoderConfig> {
+		assert(this.demuxer.firstFrameHeader);
+
+		return {
+			codec: 'mp3',
+			numberOfChannels: this.demuxer.firstFrameHeader.channelCount,
+			sampleRate: this.demuxer.firstFrameHeader.sampleRate,
+		};
+	}
+
+	getSampleAtIndex(sampleIndex: number, options: SampleRetrievalOptions) {
+		if (sampleIndex === -1) {
+			return null;
+		}
+
+		const rawSample = this.demuxer.allSamples[sampleIndex];
+		if (!rawSample) {
+			return null;
+		}
+
+		let data: Uint8Array;
+		if (options.metadataOnly) {
+			data = PLACEHOLDER_DATA;
+		} else {
+			this.demuxer.reader.pos = rawSample.dataStart;
+			data = this.demuxer.reader.readBytes(rawSample.dataSize);
+		}
+
+		return new EncodedAudioSample(
+			data,
+			'key',
+			rawSample.timestamp,
+			rawSample.duration,
+		);
+	}
+
+	async getFirstSample(options: SampleRetrievalOptions) {
+		return this.getSampleAtIndex(0, options);
+	}
+
+	async getNextSample(sample: EncodedAudioSample, options: SampleRetrievalOptions) {
+		const sampleIndex = binarySearchExact(
+			this.demuxer.allSamples,
+			sample.timestamp,
+			x => x.timestamp,
+		);
+		if (sampleIndex === -1) {
+			throw new Error('Sample was not created from this track.');
+		}
+
+		return this.getSampleAtIndex(sampleIndex + 1, options);
+	}
+
+	async getSample(timestamp: number, options: SampleRetrievalOptions) {
+		const index = binarySearchLessOrEqual(
+			this.demuxer.allSamples,
+			timestamp,
+			x => x.timestamp,
+		);
+		return this.getSampleAtIndex(index, options);
+	}
+
+	getKeySample(timestamp: number, options: SampleRetrievalOptions) {
+		return this.getSample(timestamp, options);
+	}
+
+	getNextKeySample(sample: EncodedAudioSample, options: SampleRetrievalOptions) {
+		return this.getNextSample(sample, options);
+	}
+}

src/mp3/mp3-reader.ts

@@ -0,0 +1,176 @@
+import { assert } from '../misc';
+import { Reader } from '../reader';
+
+const FRAME_HEADER_SIZE = 4;
+
+// These are in kbps:
+const MPEG_V1_BITRATES: Record<number, number[]> = {
+	// Layer 3
+	1: [-1, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, -1],
+	// Layer 2
+	2: [-1, 32, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384, -1],
+	// Layer 1
+	3: [-1, 32, 64, 96, 128, 160, 192, 224, 256, 288, 320, 352, 384, 416, 448, -1],
+};
+const MPEG_V2_BITRATES: Record<number, number[]> = {
+	// Layer 3
+	1: [-1, 32, 48, 56, 64, 80, 96, 112, 128, 144, 160, 176, 192, 224, 256, -1],
+	// Layer 2
+	2: [-1, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160, -1],
+	// Layer 1
+	3: [-1, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160, -1],
+};
+const SAMPLING_RATES: Record<number, number[]> = {
+	// MPEG Version 2.5
+	0: [11025, 12000, 8000, -1],
+	// MPEG Version 2 (ISO/IEC 13818-3)
+	2: [22050, 24000, 16000, -1],
+	// MPEG Version 1 (ISO/IEC 11172-3)
+	3: [44100, 48000, 32000, -1],
+};
+
+export type FrameHeader = {
+	startPos: number;
+	mpegVersionId: number;
+	bitrate: number;
+	sampleRate: number;
+	channelCount: number;
+	totalSize: number;
+	dataStart: number;
+	dataSize: number;
+};
+
+export class Mp3Reader {
+	pos = 0;
+	fileSize: number | null = null;
+
+	constructor(public reader: Reader) {}
+
+	readBytes(length: number) {
+		const { view, offset } = this.reader.getViewAndOffset(this.pos, this.pos + length);
+		this.pos += length;
+
+		return new Uint8Array(view.buffer, offset, length);
+	}
+
+	readU16() {
+		const { view, offset } = this.reader.getViewAndOffset(this.pos, this.pos + 2);
+		this.pos += 2;
+
+		return view.getUint16(offset, false);
+	}
+
+	readU32() {
+		const { view, offset } = this.reader.getViewAndOffset(this.pos, this.pos + 4);
+		this.pos += 4;
+
+		return view.getUint32(offset, false);
+	}
+
+	readAscii(length: number) {
+		const { view, offset } = this.reader.getViewAndOffset(this.pos, this.pos + length);
+		this.pos += length;
+
+		let str = '';
+		for (let i = 0; i < length; i++) {
+			str += String.fromCharCode(view.getUint8(offset + i));
+		}
+		return str;
+	}
+
+	readId3() {
+		const tag = this.readAscii(3);
+		if (tag !== 'ID3') {
+			this.pos -= 3;
+			return null;
+		}
+
+		this.pos += 3;
+
+		const size = decodeSynchsafe(this.readU32());
+		return { size };
+	}
+
+	readNextFrameHeader(until?: number): FrameHeader | null {
+		assert(this.fileSize);
+		until ??= this.fileSize;
+
+		while (this.pos < until - FRAME_HEADER_SIZE) {
+			const startPos = this.pos;
+			const word = this.readU32();
+
+			const firstByte = word >>> 24;
+			const secondByte = (word >>> 16) & 0xff;
+			const thirdByte = (word >>> 8) & 0xff;
+			const fourthByte = word & 0xff;
+
+			if (firstByte !== 0xff && secondByte !== 0xff && thirdByte !== 0xff && fourthByte !== 0xff) {
+				continue;
+			}
+
+			this.pos -= 3; // For when we continue
+
+			if ((secondByte & 0xe0) !== 0xe0) {
+				continue;
+			}
+
+			const mpegVersionId = (secondByte >> 3) & 0x3;
+			const layer = (secondByte >> 1) & 0x3;
+
+			const bitrateIndex = (thirdByte >> 4) & 0xf;
+			const frequencyIndex = (thirdByte >> 2) & 0x3;
+			const padding = (thirdByte >> 1) & 0x1;
+
+			const channel = (fourthByte >> 6) & 0x3;
+
+			const kilobitRate = mpegVersionId === 3
+				? MPEG_V1_BITRATES[layer]?.[bitrateIndex]
+				: MPEG_V2_BITRATES[layer]?.[bitrateIndex];
+			if (!kilobitRate || kilobitRate === -1) {
+				continue;
+			}
+
+			const bitrate = kilobitRate * 1000;
+
+			const sampleRate = SAMPLING_RATES[mpegVersionId]?.[frequencyIndex];
+			if (!sampleRate || sampleRate === -1) {
+				continue;
+			}
+
+			const channelCount = channel === 3 ? 1 : 2;
+
+			const frameLength = Math.floor((144 * bitrate / sampleRate) + padding);
+
+			if (this.fileSize - startPos < frameLength) {
+				// The frame doesn't fit into the rest of the file
+				return null;
+			}
+
+			return {
+				startPos,
+				mpegVersionId,
+				bitrate,
+				sampleRate,
+				channelCount,
+				totalSize: frameLength,
+				dataStart: startPos + 4,
+				dataSize: frameLength - 4,
+			};
+		}
+
+		return null;
+	}
+}
+
+export const decodeSynchsafe = (synchsafed: number) => {
+	let mask = 0x7f000000;
+	let unsynchsafed = 0;
+
+	while (mask !== 0) {
+		unsynchsafed >>= 1;
+		unsynchsafed |= synchsafed & mask;
+		mask >>= 8;
+	}
+
+	return unsynchsafed;
+};

src/reader.ts

@@ -27,6 +27,10 @@ export class Reader {
 	async loadRange(start: number, end: number) {
 		end = Math.min(end, await this.source._getSize());
 
+		if (start >= end) {
+			return;
+		}
+
 		const matchingLoadingSegment = this.loadingSegments.find(x => x.start <= start && x.end >= end);
 		if (matchingLoadingSegment) {
 			// Simply wait for the existing promise to finish to avoid loading the same range twice

src/wave/wave-demuxer.ts

@@ -17,7 +17,7 @@ export enum WaveFormat {
 }
 
 export class WaveDemuxer extends Demuxer {
-	riffReader: RiffReader;
+	metadataReader: RiffReader;
 	chunkReader: RiffReader;
 
 	metadataPromise: Promise<void> | null = null;
@@ -36,38 +36,38 @@ export class WaveDemuxer extends Demuxer {
 	constructor(input: Input) {
 		super(input);
 
-		this.riffReader = new RiffReader(input._mainReader);
+		this.metadataReader = new RiffReader(input._mainReader);
 		this.chunkReader = new RiffReader(new Reader(input._source, 64 * 2 ** 20));
 	}
 
 	async readMetadata() {
 		return this.metadataPromise ??= (async () => {
-			const riffType = this.riffReader.readAscii(4);
-			this.riffReader.littleEndian = riffType === 'RIFF';
+			const riffType = this.metadataReader.readAscii(4);
+			this.metadataReader.littleEndian = riffType === 'RIFF';
 
-			const totalFileSize = this.riffReader.readU32() + 8;
-			const format = this.riffReader.readAscii(4);
+			const totalFileSize = this.metadataReader.readU32() + 8;
+			const format = this.metadataReader.readAscii(4);
 
 			if (format !== 'WAVE') {
 				throw new Error('Invalid WAVE file - wrong format');
 			}
 
-			this.riffReader.pos = 12;
-			while (this.riffReader.pos < totalFileSize) {
-				await this.riffReader.reader.loadRange(this.riffReader.pos, this.riffReader.pos + 8);
+			this.metadataReader.pos = 12;
+			while (this.metadataReader.pos < totalFileSize) {
+				await this.metadataReader.reader.loadRange(this.metadataReader.pos, this.metadataReader.pos + 8);
 
-				const chunkId = this.riffReader.readAscii(4);
-				const chunkSize = this.riffReader.readU32();
-				const startPos = this.riffReader.pos;
+				const chunkId = this.metadataReader.readAscii(4);
+				const chunkSize = this.metadataReader.readU32();
+				const startPos = this.metadataReader.pos;
 
 				if (chunkId === 'fmt ') {
 					await this.parseFmtChunk(chunkSize);
 				} else if (chunkId === 'data') {
-					this.dataStart = this.riffReader.pos;
+					this.dataStart = this.metadataReader.pos;
 					this.dataSize = chunkSize;
 				}
 
-				this.riffReader.pos = startPos + chunkSize + (chunkSize & 1); // Handle padding
+				this.metadataReader.pos = startPos + chunkSize + (chunkSize & 1); // Handle padding
 			}
 
 			if (!this.audioInfo) {
@@ -85,32 +85,32 @@ export class WaveDemuxer extends Demuxer {
 	}
 
 	private async parseFmtChunk(size: number) {
-		await this.riffReader.reader.loadRange(this.riffReader.pos, this.riffReader.pos + size);
+		await this.metadataReader.reader.loadRange(this.metadataReader.pos, this.metadataReader.pos + size);
 
-		let formatTag = this.riffReader.readU16();
-		const numChannels = this.riffReader.readU16();
-		const sampleRate = this.riffReader.readU32();
-		this.riffReader.pos += 4;
-		const blockAlign = this.riffReader.readU16();
+		let formatTag = this.metadataReader.readU16();
+		const numChannels = this.metadataReader.readU16();
+		const sampleRate = this.metadataReader.readU32();
+		this.metadataReader.pos += 4;
+		const blockAlign = this.metadataReader.readU16();
 
 		let bitsPerSample: number;
 
 		if (size === 14) { // Plain WAVEFORMAT
 			bitsPerSample = 8;
 		} else {
-			bitsPerSample = this.riffReader.readU16();
+			bitsPerSample = this.metadataReader.readU16();
 		}
 
 		// Handle WAVEFORMATEXTENSIBLE
 		if (size >= 18 && formatTag !== 0x0165) {
-			const cbSize = this.riffReader.readU16();
+			const cbSize = this.metadataReader.readU16();
 			const remainingSize = size - 18;
 			const extensionSize = Math.min(remainingSize, cbSize);
 
 			if (extensionSize >= 22 && formatTag === WaveFormat.EXTENSIBLE) {
 				// Parse WAVEFORMATEXTENSIBLE
-				this.riffReader.pos += 2 + 4;
-				const subFormat = this.riffReader.readBytes(16);
+				this.metadataReader.pos += 2 + 4;
+				const subFormat = this.metadataReader.readBytes(16);
 
 				// Get actual format from subFormat GUID
 				formatTag = subFormat[0]! | (subFormat[1]! << 8);