Merge branch 'develop'

lerna.json

@@ -1,7 +1,12 @@
 {
 	"$schema": "node_modules/lerna/schemas/lerna-schema.json",
-	"useWorkspaces": true,
 	"workspaces": ["packages/*"],
 	"version": "independent",
-	"npmClient": "pnpm"
+	"npmClient": "pnpm",
+	"includeMergedTags": true,
+	"command": {
+		"version": {
+			"allowBranch": "master"
+		}
+	}
 }

package.json

@@ -26,12 +26,12 @@
 	},
 	"homepage": "https://github.com/alexandercerutti/sub37#readme",
 	"devDependencies": {
-		"@jest/globals": "^29.4.3",
-		"@types/jest": "^29.4.0",
+		"@jest/globals": "^29.7.0",
+		"@types/jest": "^29.5.14",
 		"jest": "^29.4.3",
-		"jest-environment-jsdom": "^29.4.3",
-		"lerna": "^6.6.1",
-		"prettier": "^2.8.4",
+		"jest-environment-jsdom": "^29.7.0",
+		"lerna": "^8.1.9",
+		"prettier": "^3.5.0",
 		"typescript": "^4.9.5"
 	}
 }

packages/sample/package.json

@@ -1,7 +1,7 @@
 {
 	"name": "@sub37/sample-app",
 	"private": true,
-	"version": "1.0.2",
+	"version": "1.0.3",
 	"type": "module",
 	"scripts": {
 		"dev": "pnpm vite",

packages/webvtt-adapter/CHANGELOG.md

@@ -1,5 +1,13 @@
 # @sub37/webvtt-adapter
 
+## **1.1.0** (10 Feb 2025)
+
+**Changes**:
+
+- Added missing exclusion of cues with the same ids, when available, with error emission;
+
+---
+
 ## **1.0.4** (08 Feb 2024)
 
 **Changes**:

packages/webvtt-adapter/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@sub37/webvtt-adapter",
-	"version": "1.0.4",
+	"version": "1.1.0",
 	"description": "A subtitles adapter for WebVTT subtitles",
 	"main": "lib/index.js",
 	"type": "module",

packages/webvtt-adapter/specs/Adapter.spec.mjs

@@ -129,6 +129,52 @@ This cue should never appear, right?
 			expect(result.data[1].endTime).toEqual(10000);
 		});
 
+		it("should exclude cues with the same ids except timestamps", () => {
+			const SAME_IDS_CONTENT = `
+WEBVTT
+
+id1
+00:00:06.000 --> 00:00:07.000
+...
+
+id1
+00:00:06.050 --> 00:00:07.050
+...
+
+123
+00:00:08.000 --> 00:00:10.000
+...Right?
+
+123
+00:00:08.050 --> 00:00:10.050
+...Right?
+
+456
+00:00:08.000 --> 00:00:10.000
+<00:00:08.250> Test t1
+<00:00:08.500> Test t2
+`;
+
+			const result = adapter.parse(SAME_IDS_CONTENT);
+			expect(result.data.length).toEqual(4);
+
+			expect(result.data[0].id).toEqual("id1");
+			expect(result.data[0].startTime).toEqual(6000);
+			expect(result.data[0].endTime).toEqual(7000);
+
+			expect(result.data[1].id).toEqual("123");
+			expect(result.data[1].startTime).toEqual(8000);
+			expect(result.data[1].endTime).toEqual(10000);
+
+			expect(result.data[2].id).toEqual("456");
+			expect(result.data[2].startTime).toEqual(8250);
+			expect(result.data[2].endTime).toEqual(10000);
+
+			expect(result.data[3].id).toEqual("456");
+			expect(result.data[3].startTime).toEqual(8500);
+			expect(result.data[3].endTime).toEqual(10000);
+		});
+
 		it("should return an array containing two cues", () => {
 			const CLASSIC_CONTENT = `
 WEBVTT

packages/webvtt-adapter/src/Adapter.ts

@@ -38,6 +38,7 @@ export default class WebVTTAdapter extends BaseAdapter {
 			]);
 		}
 
+		const cueIdsList: Set<string> = new Set();
 		const cues: CueNode[] = [];
 		const content = String(rawContent).replace(/\r?\n/g, "\n");
 		const block = {
@@ -127,11 +128,44 @@ export default class WebVTTAdapter extends BaseAdapter {
 
 					for (const parsedCue of parsedContent) {
 						if (parsedCue.startTime >= parsedCue.endTime) {
+							failures.push({
+								error: new Error(`A cue cannot start (${parsedCue.startTime}) after its end time (${parsedCue.endTime})`),
+								failedChunk: content.substring(block.start, block.cursor),
+								isCritical: false,
+							});
+
 							continue;
 						}
 
+						if (parsedCue.id) {
+							/**
+							 * "A WebVTT cue identifier must be unique amongst
+							 * all the WebVTT cue identifiers of all WebVTT
+							 * cues of a WebVTT file."
+							 * 
+							 * @see https://www.w3.org/TR/webvtt1/#webvtt-cue-identifier
+							 */
+
+							if (!parsedCue.isTimestamp && cueIdsList.has(parsedCue.id)) {
+								failures.push({
+									error: new Error(`A WebVTT cue identifier must be unique amongst all the cue identifiers of a WebVTT file. Double id found: '${parsedCue.id}'`),
+									failedChunk: content.substring(block.start, block.cursor),
+									isCritical: false,
+								});
+
+								continue;
+							}
+
+							/**
+							 * ... however, when we generate a custom identifier
+							 * for the cue, we re-use the same for the timestamps
+							 * because they must appear on the same line.
+							 */
+							cueIdsList.add(parsedCue.id);
+						}
+
 						const cue = CueNode.from(latestRootCue, {
-							id: parsedCue.id,
+							id: parsedCue.id || `cue-${block.start}-${block.cursor}`,
 							startTime: parsedCue.startTime,
 							endTime: parsedCue.endTime,
 							content: parsedCue.text,
@@ -310,13 +344,15 @@ function evaluateBlock(
 		return new InvalidFormatError("UNKNOWN_BLOCK_ENTITY", contentSection);
 	}
 
-	const { attributes, cueid, endtime, starttime, text } = cueMatch.groups as {
-		[K in keyof Parser.CueRawData]: Parser.CueRawData[K];
+	type CueMatchGroups = {
+		[K in keyof Omit<Parser.CueRawData, "startCharPosition" | "endCharPosition">]: Parser.CueRawData[K];
 	};
 
+	const { attributes, cueid, endtime, starttime, text } = cueMatch.groups as CueMatchGroups;
+
 	const cueParsingResult = Parser.parseCue({
 		attributes,
-		cueid: cueid || `cue-${start}-${end}`,
+		cueid: cueid,
 		endtime,
 		starttime,
 		text: text.replace(TABS_REGEX, ""),

packages/webvtt-adapter/src/Parser/parseCue.ts

@@ -23,6 +23,7 @@ export interface CueParsedData {
 	tags: Entities.Tag[];
 	text: string;
 	renderingModifiers: RenderingModifiers;
+	isTimestamp?: boolean;
 }
 
 export function parseCue(data: CueRawData): CueParsedData[] {
@@ -96,6 +97,7 @@ export function parseCue(data: CueRawData): CueParsedData[] {
 					 */
 
 					addCueEntities(currentCue, Tags.createTagEntitiesFromUnpaired(openTagsQueue, currentCue));
+					currentCue.isTimestamp = true;
 					hsCues.push(currentCue);
 				}
 
@@ -104,6 +106,7 @@ export function parseCue(data: CueRawData): CueParsedData[] {
 					currentCue.endTime,
 					currentCue.id,
 					currentCue.renderingModifiers,
+					currentCue.isTimestamp,
 				);
 
 				break;
@@ -143,6 +146,7 @@ function createCue(
 	endTime: number,
 	id?: string,
 	renderingModifiers?: RenderingModifiers,
+	isTimestamp: boolean = false,
 ): CueParsedData {
 	return {
 		startTime,
@@ -151,5 +155,6 @@ function createCue(
 		tags: [],
 		id,
 		renderingModifiers,
+		isTimestamp,
 	};
 }