Make MP3 next frame location more strict for better recovery (fixes #382), fix Date timezone for ID3 tags

packages/mp3-encoder/src/index.ts

@@ -7,7 +7,7 @@
  */
 
 import { CustomAudioEncoder, AudioCodec, AudioSample, EncodedPacket, registerEncoder } from 'mediabunny';
-import { FRAME_HEADER_SIZE, readMp3FrameHeader, SAMPLING_RATES } from '../../../shared/mp3-misc';
+import { MP3_FRAME_HEADER_SIZE, readMp3FrameHeader, SAMPLING_RATES } from '../../../shared/mp3-misc';
 import type { WorkerCommand, WorkerResponse, WorkerResponseData } from './shared';
 // @ts-expect-error An esbuild plugin handles this, TypeScript doesn't need to understand
 import createWorker from './encode.worker';
@@ -160,7 +160,7 @@ class Mp3Encoder extends CustomAudioEncoder {
 		this.currentBufferOffset = requiredBufferSize;
 
 		let pos = 0;
-		while (pos <= this.currentBufferOffset - FRAME_HEADER_SIZE) {
+		while (pos <= this.currentBufferOffset - MP3_FRAME_HEADER_SIZE) {
 			const word = new DataView(this.buffer.buffer).getUint32(pos, false);
 			const header = readMp3FrameHeader(word, null).header;
 			if (!header) {

shared/mp3-misc.ts

@@ -6,7 +6,7 @@
  * file, You can obtain one at https://mozilla.org/MPL/2.0/.
  */
 
-export const FRAME_HEADER_SIZE = 4;
+export const MP3_FRAME_HEADER_SIZE = 4;
 export const SAMPLING_RATES = [44100, 48000, 32000];
 export const KILOBIT_RATES = [
 	// lowSamplingFrequency === 0
@@ -122,7 +122,8 @@ export const readMp3FrameHeader = (word: number, remainingBytes: number | null):
 	const layer = (secondByte >> 1) & 0x3;
 
 	const bitrateIndex = (thirdByte >> 4) & 0xf;
-	const frequencyIndex = ((thirdByte >> 2) & 0x3) % 3;
+	const frequencyIndex = ((thirdByte >> 2) & 0x3) % 3; // FFmpeg effectively does % 3 (but in a roundabout way)
+
 	const padding = (thirdByte >> 1) & 0x1;
 
 	const channel = (fourthByte >> 6) & 0x3;
@@ -212,3 +213,7 @@ export enum XingFlags {
 	FileSize = 1 << 1,
 	Toc = 1 << 2,
 }
+
+export const getMp3ChannelCount = (channel: number) => {
+	return channel === 3 ? 1 : 2;
+};

src/id3.ts

@@ -93,7 +93,7 @@ export const parseId3V1Tag = (slice: FileSlice, tags: MetadataTags) => {
 	const yearText = readId3V1String(slice, 4);
 	const year = Number.parseInt(yearText, 10);
 	if (Number.isInteger(year) && year > 0) {
-		tags.date ??= new Date(year, 0, 1);
+		tags.date ??= new Date(String(year)); // String so that it parses as UTC
 	}
 
 	const commentBytes = readBytes(slice, 30);
@@ -339,7 +339,7 @@ export const parseId3V2Tag = (slice: FileSlice, header: Id3V2Header, tags: Metad
 				const year = Number.parseInt(yearText, 10);
 
 				if (Number.isInteger(year)) {
-					tags.date ??= new Date(year, 0, 1);
+					tags.date ??= new Date(String(year)); // String so that it parses as UTC
 				}
 			}; break;
 

src/input-format.ts

@@ -22,7 +22,7 @@ import {
 } from './matroska/ebml';
 import { MatroskaDemuxer } from './matroska/matroska-demuxer';
 import { Mp3Demuxer } from './mp3/mp3-demuxer';
-import { FRAME_HEADER_SIZE, getXingOffset, INFO, XING } from '../shared/mp3-misc';
+import { MP3_FRAME_HEADER_SIZE, getXingOffset, INFO, XING } from '../shared/mp3-misc';
 import { ID3_V2_HEADER_SIZE, readId3V2Header } from './id3';
 import { readNextMp3FrameHeader } from './mp3/mp3-reader';
 import { OggDemuxer } from './ogg/ogg-demuxer';
@@ -330,7 +330,7 @@ export class Mp3InputFormat extends InputFormat {
 
 		// Fine, we found one frame header, but we're still not entirely sure this is MP3. Let's check if we can find
 		// another header right after it:
-		const secondResult = await readNextMp3FrameHeader(input._reader, currentPos, currentPos + FRAME_HEADER_SIZE);
+		const secondResult = await readNextMp3FrameHeader(input._reader, currentPos, currentPos + MP3_FRAME_HEADER_SIZE);
 		if (!secondResult) {
 			return false;
 		}

src/mp3/mp3-demuxer.ts

@@ -28,6 +28,7 @@ import {
 	XING,
 	XingFlags,
 	computeAverageMp3FrameSize,
+	getMp3ChannelCount,
 } from '../../shared/mp3-misc';
 import {
 	ID3_V1_TAG_SIZE,
@@ -108,7 +109,12 @@ export class Mp3Demuxer extends Demuxer {
 			}
 		}
 
-		const result = await readNextMp3FrameHeader(this.reader, this.lastLoadedPos, this.reader.fileSize);
+		const result = await readNextMp3FrameHeader(
+			this.reader,
+			this.lastLoadedPos,
+			this.reader.fileSize,
+			this.firstFrameHeader,
+		);
 		if (!result) {
 			this.lastSampleLoaded = true;
 			return;
@@ -157,13 +163,6 @@ export class Mp3Demuxer extends Demuxer {
 			this.firstFrameHeaderPos = result.startPos;
 		}
 
-		if (header.sampleRate !== this.firstFrameHeader.sampleRate) {
-			console.warn(
-				`MP3 changed sample rate mid-file: ${this.firstFrameHeader.sampleRate} Hz to ${header.sampleRate} Hz.`
-				+ ` Might be a bug, so please report this file.`,
-			);
-		}
-
 		const sampleDuration = header.audioSamplesInFrame / this.firstFrameHeader.sampleRate;
 		const sample: Sample = {
 			timestamp: this.nextTimestampInSamples / this.firstFrameHeader.sampleRate,
@@ -332,7 +331,7 @@ class Mp3AudioTrackBacking implements InputAudioTrackBacking {
 
 	getNumberOfChannels() {
 		assert(this.demuxer.firstFrameHeader);
-		return this.demuxer.firstFrameHeader.channel === 3 ? 1 : 2;
+		return getMp3ChannelCount(this.demuxer.firstFrameHeader.channel);
 	}
 
 	getSampleRate() {
@@ -351,7 +350,7 @@ class Mp3AudioTrackBacking implements InputAudioTrackBacking {
 
 		return {
 			codec: 'mp3',
-			numberOfChannels: this.demuxer.firstFrameHeader.channel === 3 ? 1 : 2,
+			numberOfChannels: getMp3ChannelCount(this.demuxer.firstFrameHeader.channel),
 			sampleRate: this.demuxer.firstFrameHeader.sampleRate,
 		};
 	}

src/mp3/mp3-reader.ts

@@ -6,10 +6,15 @@
  * file, You can obtain one at https://mozilla.org/MPL/2.0/.
  */
 
-import { FRAME_HEADER_SIZE, Mp3FrameHeader, readMp3FrameHeader } from '../../shared/mp3-misc';
+import { MP3_FRAME_HEADER_SIZE, getMp3ChannelCount, Mp3FrameHeader, readMp3FrameHeader } from '../../shared/mp3-misc';
 import { Reader, readU32Be } from '../reader';
 
-export const readNextMp3FrameHeader = async (reader: Reader, startPos: number, until: number | null): Promise<{
+export const readNextMp3FrameHeader = async (
+	reader: Reader,
+	startPos: number,
+	until: number | null,
+	ref: Mp3FrameHeader | null = null,
+): Promise<{
 	header: Mp3FrameHeader;
 	startPos: number;
 } | null> => {
@@ -21,11 +26,11 @@ export const readNextMp3FrameHeader = async (reader: Reader, startPos: number, u
 			? Math.min(CHUNK_SIZE, until - currentPos)
 			: CHUNK_SIZE;
 
-		let slice = reader.requestSliceRange(currentPos, FRAME_HEADER_SIZE, maxLength);
+		let slice = reader.requestSliceRange(currentPos, MP3_FRAME_HEADER_SIZE, maxLength);
 		if (slice instanceof Promise) slice = await slice;
-		if (!slice || slice.length < FRAME_HEADER_SIZE) break;
+		if (!slice || slice.length < MP3_FRAME_HEADER_SIZE) break;
 
-		while (slice.remainingLength >= FRAME_HEADER_SIZE) {
+		while (slice.remainingLength >= MP3_FRAME_HEADER_SIZE) {
 			const posBeforeRead = slice.filePos;
 			const word = readU32Be(slice);
 			const remainingBytes = reader.fileSize !== null
@@ -33,7 +38,17 @@ export const readNextMp3FrameHeader = async (reader: Reader, startPos: number, u
 				: null;
 
 			const result = readMp3FrameHeader(word, remainingBytes);
-			if (result.header) {
+			if (
+				result.header
+				&& (!ref || (
+					// This condition helps us recover malformed streams
+					// https://stackoverflow.com/a/20884944
+					result.header.sampleRate === ref.sampleRate
+					&& result.header.mpegVersionId === ref.mpegVersionId
+					&& result.header.layer === ref.layer
+					&& getMp3ChannelCount(result.header.channel) === getMp3ChannelCount(ref.channel)
+				))
+			) {
 				return { header: result.header, startPos: currentPos };
 			}
 

src/mpeg-ts/mpeg-ts-demuxer.ts

@@ -63,7 +63,11 @@ import {
 	TRANSFER_CHARACTERISTICS_MAP_INVERSE,
 	UNDETERMINED_LANGUAGE,
 } from '../misc';
-import { FRAME_HEADER_SIZE as MP3_FRAME_HEADER_SIZE, readMp3FrameHeader } from '../../shared/mp3-misc';
+import {
+	MP3_FRAME_HEADER_SIZE,
+	getMp3ChannelCount,
+	readMp3FrameHeader,
+} from '../../shared/mp3-misc';
 import { EncodedPacket, PacketType, PLACEHOLDER_DATA } from '../packet';
 import { FileSlice, readBytes, Reader, readU16Be, readU32Be, readU8 } from '../reader';
 import { buildMpegTsMimeType, MpegTsStreamType, TIMESCALE, TS_PACKET_SIZE } from './mpeg-ts-misc';
@@ -629,7 +633,7 @@ export class MpegTsDemuxer extends Demuxer {
 									);
 								}
 
-								elementaryStream.info.numberOfChannels = result.header.channel === 3 ? 1 : 2;
+								elementaryStream.info.numberOfChannels = getMp3ChannelCount(result.header.channel);
 								elementaryStream.info.sampleRate = result.header.sampleRate;
 							} else if (elementaryStream.info.codec === 'ac3') {
 								const frameInfo = parseAc3SyncFrame(context.suppliedPacket.data);

test/browser/mp3.test.ts

@@ -0,0 +1,21 @@
+import { test } from 'vitest';
+import { Input } from '../../src/input.js';
+import { UrlSource } from '../../src/source.js';
+import { ALL_FORMATS } from '../../src/input-format.js';
+import { assert } from '../../src/misc.js';
+import { AudioSampleSink } from '../../src/media-sink.js';
+
+// "joined" in the sense that it was two separate MP3s that were spliced together (I think)
+test('Can decode malformed joined MP3', async () => {
+	using input = new Input({
+		source: new UrlSource('/malformed-join.mp3'),
+		formats: ALL_FORMATS,
+	});
+
+	const audioTrack = await input.getPrimaryAudioTrack();
+	assert(audioTrack);
+
+	const sink = new AudioSampleSink(audioTrack);
+	// eslint-disable-next-line @typescript-eslint/no-unused-vars
+	for await (using sample of sink.samples());
+});