Allow all Matroska AttachedFiles to be read & written via metadata tags

src/matroska/ebml.ts

@@ -16,6 +16,7 @@ export interface EBMLElement {
 	size?: number;
 	data:
 		| number
+		| bigint
 		| string
 		| Uint8Array
 		| EBMLFloat32
@@ -198,6 +199,26 @@ export const measureUnsignedInt = (value: number) => {
 	}
 };
 
+export const measureUnsignedBigInt = (value: bigint) => {
+	if (value < (1n << 8n)) {
+		return 1;
+	} else if (value < (1n << 16n)) {
+		return 2;
+	} else if (value < (1n << 24n)) {
+		return 3;
+	} else if (value < (1n << 32n)) {
+		return 4;
+	} else if (value < (1n << 40n)) {
+		return 5;
+	} else if (value < (1n << 48n)) {
+		return 6;
+	} else if (value < (1n << 56n)) {
+		return 7;
+	} else {
+		return 8;
+	}
+};
+
 export const measureSignedInt = (value: number) => {
 	if (value >= -(1 << 6) && value < (1 << 6)) {
 		return 1;
@@ -276,16 +297,16 @@ export class EBMLWriter {
 			// eslint-disable-next-line no-fallthrough
 			case 5:
 				this.helperView.setUint8(pos++, (value / 2 ** 32) | 0);
-				// eslint-disable-next-line no-fallthrough
+			// eslint-disable-next-line no-fallthrough
 			case 4:
 				this.helperView.setUint8(pos++, value >> 24);
-				// eslint-disable-next-line no-fallthrough
+			// eslint-disable-next-line no-fallthrough
 			case 3:
 				this.helperView.setUint8(pos++, value >> 16);
-				// eslint-disable-next-line no-fallthrough
+			// eslint-disable-next-line no-fallthrough
 			case 2:
 				this.helperView.setUint8(pos++, value >> 8);
-				// eslint-disable-next-line no-fallthrough
+			// eslint-disable-next-line no-fallthrough
 			case 1:
 				this.helperView.setUint8(pos++, value);
 				break;
@@ -296,6 +317,16 @@ export class EBMLWriter {
 		this.writer.write(this.helper.subarray(0, pos));
 	}
 
+	writeUnsignedBigInt(value: bigint, width = measureUnsignedBigInt(value)) {
+		let pos = 0;
+
+		for (let i = width - 1; i >= 0; i--) {
+			this.helperView.setUint8(pos++, Number((value >> BigInt(i * 8)) & 0xffn));
+		}
+
+		this.writer.write(this.helper.subarray(0, pos));
+	}
+
 	writeSignedInt(value: number, width = measureSignedInt(value)) {
 		if (value < 0) {
 			// Two's complement stuff
@@ -398,6 +429,10 @@ export class EBMLWriter {
 				const size = data.size ?? measureUnsignedInt(data.data);
 				this.writeVarInt(size);
 				this.writeUnsignedInt(data.data, size);
+			} else if (typeof data.data === 'bigint') {
+				const size = data.size ?? measureUnsignedBigInt(data.data);
+				this.writeVarInt(size);
+				this.writeUnsignedBigInt(data.data, size);
 			} else if (typeof data.data === 'string') {
 				this.writeVarInt(data.data.length);
 				this.writeAsciiString(data.data);
@@ -491,6 +526,21 @@ export const readUnsignedInt = (slice: FileSlice, width: number) => {
 	return value;
 };
 
+export const readUnsignedBigInt = (slice: FileSlice, width: number) => {
+	if (width < 1) {
+		throw new Error('Bad unsigned int size ' + width);
+	}
+
+	let value = 0n;
+
+	for (let i = 0; i < width; i++) {
+		value <<= 8n;
+		value += BigInt(readU8(slice));
+	}
+
+	return value;
+};
+
 export const readSignedInt = (slice: FileSlice, width: number) => {
 	let value = readUnsignedInt(slice, width);
 

src/matroska/matroska-demuxer.ts

@@ -31,7 +31,7 @@ import {
 	InputVideoTrack,
 	InputVideoTrackBacking,
 } from '../input-track';
-import { MetadataTags } from '../tags';
+import { AttachedFile, MetadataTags } from '../tags';
 import { PacketRetrievalOptions } from '../media-sink';
 import {
 	assert,
@@ -69,6 +69,7 @@ import {
 	readVarInt,
 	resync,
 	searchForNextElementId,
+	readUnsignedBigInt,
 } from './ebml';
 import { buildMatroskaMimeType } from './matroska-misc';
 import { FileSlice, readBytes, Reader, readI16Be, readU8 } from '../reader';
@@ -240,6 +241,7 @@ export class MatroskaDemuxer extends Demuxer {
 	currentTagTargetIsMovie: boolean = true;
 	currentSimpleTagName: string | null = null;
 	currentAttachedFile: {
+		fileUid: bigint | null;
 		fileName: string | null;
 		fileMediaType: string | null;
 		fileData: Uint8Array | null;
@@ -1478,6 +1480,7 @@ export class MatroskaDemuxer extends Demuxer {
 				if (!this.currentSegment) break;
 
 				this.currentAttachedFile = {
+					fileUid: null,
 					fileName: null,
 					fileMediaType: null,
 					fileData: null,
@@ -1486,6 +1489,19 @@ export class MatroskaDemuxer extends Demuxer {
 
 				this.readContiguousElements(slice.slice(dataStartPos, size));
 
+				const tags = this.currentSegment.metadataTags;
+
+				if (this.currentAttachedFile.fileUid && this.currentAttachedFile.fileData) {
+					// All attached files get surfaced in the `raw` metadata tags
+					tags.raw ??= {};
+					tags.raw[this.currentAttachedFile.fileUid.toString()] = new AttachedFile(
+						this.currentAttachedFile.fileData,
+						this.currentAttachedFile.fileMediaType ?? undefined,
+						this.currentAttachedFile.fileName ?? undefined,
+						this.currentAttachedFile.fileDescription ?? undefined,
+					);
+				}
+
 				// Only process image attachments
 				if (this.currentAttachedFile.fileMediaType?.startsWith('image/') && this.currentAttachedFile.fileData) {
 					const fileName = this.currentAttachedFile.fileName;
@@ -1500,8 +1516,8 @@ export class MatroskaDemuxer extends Demuxer {
 						}
 					}
 
-					this.currentSegment.metadataTags.images ??= [];
-					this.currentSegment.metadataTags.images.push({
+					tags.images ??= [];
+					tags.images.push({
 						data: this.currentAttachedFile.fileData,
 						mimeType: this.currentAttachedFile.fileMediaType,
 						kind,
@@ -1513,6 +1529,12 @@ export class MatroskaDemuxer extends Demuxer {
 				this.currentAttachedFile = null;
 			}; break;
 
+			case EBMLId.FileUID: {
+				if (!this.currentAttachedFile) break;
+
+				this.currentAttachedFile.fileUid = readUnsignedBigInt(slice, size);
+			}; break;
+
 			case EBMLId.FileName: {
 				if (!this.currentAttachedFile) break;
 

src/matroska/matroska-muxer.ts

@@ -22,6 +22,7 @@ import {
 	roundToMultiple,
 	textEncoder,
 	toUint8Array,
+	uint8ArraysAreEqual,
 	writeBits,
 } from '../misc';
 import {
@@ -62,6 +63,7 @@ import { Muxer } from '../muxer';
 import { Writer } from '../writer';
 import { EncodedPacket } from '../packet';
 import { parseOpusIdentificationHeader } from '../codec-data';
+import { AttachedFile } from '../tags';
 
 const MIN_CLUSTER_TIMESTAMP_MS = -(2 ** 15);
 const MAX_CLUSTER_TIMESTAMP_MS = 2 ** 15 - 1;
@@ -537,13 +539,12 @@ export class MatroskaMuxer extends Muxer {
 
 	private maybeCreateAttachments() {
 		const metadataTags = this.output._metadataTags;
-		if (!metadataTags.images || metadataTags.images.length === 0) {
-			return;
-		}
+		const elements: EBMLElement[] = [];
 
 		const existingFileUids = new Set<number>();
+		const images = metadataTags.images ?? [];
 
-		this.attachmentsElement = { id: EBMLId.Attachments, data: metadataTags.images.map((image): EBMLElement => {
+		for (const image of images) {
 			let imageName = image.name;
 			if (imageName === undefined) {
 				const baseName = image.kind === 'coverFront' ? 'cover' : image.kind === 'coverBack' ? 'back' : 'image';
@@ -561,7 +562,7 @@ export class MatroskaMuxer extends Muxer {
 
 			existingFileUids.add(fileUid);
 
-			return {
+			elements.push({
 				id: EBMLId.AttachedFile,
 				data: [
 					image.description !== undefined
@@ -572,8 +573,45 @@ export class MatroskaMuxer extends Muxer {
 					{ id: EBMLId.FileData, data: image.data },
 					{ id: EBMLId.FileUID, data: fileUid },
 				],
-			};
-		}) };
+			});
+		}
+
+		// Add all AttachedFiles from the raw metadata
+		for (const [key, value] of Object.entries(metadataTags.raw ?? {})) {
+			if (!(value instanceof AttachedFile)) {
+				continue;
+			}
+
+			const keyIsNumeric = /^\d+$/.test(key);
+			if (!keyIsNumeric) {
+				continue;
+			}
+
+			if (images.find(x => x.mimeType === value.mimeType && uint8ArraysAreEqual(x.data, value.data))) {
+				// This attached file has very likely already been added as an image above
+				// (happens when remuxing Matroska)
+				continue;
+			}
+
+			elements.push({
+				id: EBMLId.AttachedFile,
+				data: [
+					value.description !== undefined
+						? { id: EBMLId.FileDescription, data: new EBMLUnicodeString(value.description) }
+						: null,
+					{ id: EBMLId.FileName, data: new EBMLUnicodeString(value.name ?? '') },
+					{ id: EBMLId.FileMediaType, data: value.mimeType ?? '' },
+					{ id: EBMLId.FileData, data: value.data },
+					{ id: EBMLId.FileUID, data: BigInt(key) },
+				],
+			});
+		}
+
+		if (elements.length === 0) {
+			return;
+		}
+
+		this.attachmentsElement = { id: EBMLId.Attachments, data: elements };
 	}
 
 	private createSegment() {