Formerly KrillinAI, VideoTranslation and Dubbing Tool for 120+ Languages. Open-source AI workspace for creators, powered by Codex. Video downloader, video translation, create videos, images, audios, and edits with Agents in one place. For your Social Media like YouTube, TikTok, Instagram, etc.
TypeScript agent-skillscodex-clideskop-appdubbingimage-generationlocalizationskillsttsvideo-editingvideo-generationvideo-transcriptionvideo-translation
import { describe, expect, it } from 'vitest';
import type { SubtitleCue } from './VideoTranslationResultWorkspace.js';
import { mergeSubtitle, retimeSubtitle, sameSubtitleCues, splitSubtitle, subtitleSeconds, subtitleTimestamp, waveformPeaks } from './subtitle-editing.js';
const cues: SubtitleCue[] = [
{ id: 4, start: '00:00:01,250', end: '00:00:05,750', text: 'Hello world, welcome home.', sourceText: '你好世界,欢迎回家。' },
{ id: 9, start: '00:00:06,000', end: '00:00:08,000', text: 'Next sentence.', sourceText: '下一句。' }
];
describe('subtitle editing', () => {
it('round-trips millisecond boundaries and long recordings without rolling seconds past 59', () => {
for (const value of [0, .001, 59.999, 60, 3599.999, 3600, 90061.123]) expect(subtitleSeconds(subtitleTimestamp(value))).toBeCloseTo(value, 3);
expect(subtitleTimestamp(59.9996)).toBe('00:01:00,000');
expect(subtitleSeconds('00:61:00,000')).toBeNaN();
expect(subtitleSeconds('00:00:01.25')).toBe(1.25);
});
it('rejects invalid intervals and preserves text and identity during retiming', () => {
for (const [start, end] of [[-1, 5], [4, 4], [5, 4], [Number.NaN, 5], [0, .01]]) expect(retimeSubtitle(cues, 4, start!, end!)).toBe(cues);
const retimed = retimeSubtitle(cues, 4, 2.125, 5.625);
expect(retimed[0]).toEqual({ ...cues[0], start: '00:00:02,125', end: '00:00:05,625' });
expect(cues[0]!.start).toBe('00:00:01,250');
expect(sameSubtitleCues(cues, retimed)).toBe(false);
});
it('splits both languages at one boundary with a fresh ID, retaining every word', () => {
const split = splitSubtitle(cues, 4, 3.5);
expect(split).toHaveLength(3);
expect(split[0]!.end).toBe(split[1]!.start);
expect(split[0]!.end).toBe('00:00:03,500');
expect(split[1]!.id).toBe(10);
expect(split[0]!.text + ' ' + split[1]!.text).toBe(cues[0]!.text);
expect(split[0]!.sourceText! + split[1]!.sourceText!).toBe(cues[0]!.sourceText);
expect(split[2]).toBe(cues[1]);
expect(splitSubtitle(cues, 4, 1.25)).toBe(cues);
expect(splitSubtitle(cues, 4, 7)).toBe(cues);
});
it('merges nested or overlapping intervals without losing their duration or bilingual text', () => {
const nested = [cues[0]!, { ...cues[1]!, start: '00:00:02,000', end: '00:00:03,000' }];
const merged = mergeSubtitle(nested, 4);
expect(merged).toHaveLength(1);
expect(merged[0]!.end).toBe('00:00:05,750');
expect(merged[0]!.text).toBe(cues.map(cue => cue.text).join('\n'));
expect(merged[0]!.sourceText).toBe(cues.map(cue => cue.sourceText).join('\n'));
expect(mergeSubtitle(cues, 9)).toBe(cues);
});
it('detects structural and source edits, including reordering and changed IDs', () => {
expect(sameSubtitleCues(cues, cues.map(cue => ({ ...cue })))).toBe(true);
expect(sameSubtitleCues(cues, [...cues].reverse())).toBe(false);
expect(sameSubtitleCues(cues, [{ ...cues[0]!, sourceText: 'Changed' }, cues[1]!])).toBe(false);
expect(sameSubtitleCues(cues, splitSubtitle(cues, 4, 3))).toBe(false);
});
it('uses real amplitude from every channel and keeps silent audio flat', () => {
expect(waveformPeaks([new Float32Array([0, -.8, 0, 0]), new Float32Array([0, 0, 0, .5])], 2)).toEqual([expect.closeTo(.8), .5]);
expect(waveformPeaks([new Float32Array(10)], 4)).toEqual([0, 0, 0, 0]);
expect(waveformPeaks([])).toEqual([]);
});
});