Open source Loom alternative. Beautiful, shareable screen recordings.
Rust appcapcossloommacnextjsnextjs15open-sourceossreactrecordrustscreen-capturescreen-recorderscreenshotsolidjstauritauri-apptypescriptvite
import { describe, expect, it } from "vitest";
import { formatTimestamp, formatToWebVTT } from "@/lib/transcribe-utils";
import {
longVideoAssemblyAIResponse,
realAssemblyAIResponse,
silentVideoAssemblyAIResponse,
} from "../fixtures/assemblyai-responses";
describe("formatTimestamp", () => {
it("formats zero correctly", () => {
expect(formatTimestamp(0)).toBe("00:00:00.000");
});
it("formats sub-second timestamps", () => {
expect(formatTimestamp(80)).toBe("00:00:00.080");
expect(formatTimestamp(500)).toBe("00:00:00.500");
expect(formatTimestamp(999)).toBe("00:00:00.999");
});
it("formats seconds correctly", () => {
expect(formatTimestamp(1_000)).toBe("00:00:01.000");
expect(formatTimestamp(30_500)).toBe("00:00:30.500");
expect(formatTimestamp(59_999)).toBe("00:00:59.999");
});
it("formats minutes correctly", () => {
expect(formatTimestamp(60_000)).toBe("00:01:00.000");
expect(formatTimestamp(90_250)).toBe("00:01:30.250");
expect(formatTimestamp(3_599_500)).toBe("00:59:59.500");
});
it("formats hours correctly", () => {
expect(formatTimestamp(3_600_000)).toBe("01:00:00.000");
expect(formatTimestamp(7_261_123)).toBe("02:01:01.123");
});
it("handles AssemblyAI word timestamps", () => {
const firstWord = realAssemblyAIResponse.words[0];
expect(formatTimestamp(firstWord?.start ?? 0)).toBe("00:00:00.080");
expect(formatTimestamp(firstWord?.end ?? 0)).toBe("00:00:00.320");
});
});
describe("formatToWebVTT", () => {
it("matches the completed AssemblyAI response shape", () => {
expect(realAssemblyAIResponse).toMatchObject({
status: "completed",
speech_models: ["universal-3-5-pro", "universal-2"],
speech_model_used: "universal-3-5-pro",
language_detection: true,
utterances: null,
});
expect(realAssemblyAIResponse.words[0]).toEqual({
text: "Hello",
start: 80,
end: 320,
confidence: 0.99,
speaker: null,
});
});
it("generates valid WebVTT header", () => {
expect(formatToWebVTT({ words: [] })).toBe("WEBVTT\n\n");
});
it("handles null words", () => {
expect(formatToWebVTT({ words: null })).toBe("WEBVTT\n\n");
});
it("generates correct VTT from an AssemblyAI response", () => {
const vtt = formatToWebVTT(realAssemblyAIResponse);
expect(vtt).toMatch(/^WEBVTT\n\n/);
expect(vtt).toContain("-->");
expect(vtt).toContain("Hello everyone.");
expect(vtt).toContain("Welcome to this demo video.");
});
it("creates separate captions at punctuation", () => {
const vtt = formatToWebVTT(realAssemblyAIResponse);
const captionBlocks = vtt
.split("\n\n")
.filter((block) => block.includes("-->"));
expect(captionBlocks.length).toBeGreaterThan(1);
});
it("formats timestamps correctly in VTT format", () => {
const vtt = formatToWebVTT(realAssemblyAIResponse);
const timestampPattern =
/\d{2}:\d{2}:\d{2}\.\d{3} --> \d{2}:\d{2}:\d{2}\.\d{3}/;
expect(vtt).toMatch(timestampPattern);
});
it("handles silent video", () => {
expect(formatToWebVTT(silentVideoAssemblyAIResponse)).toBe("WEBVTT\n\n");
});
it("breaks long captions at eight words", () => {
const vtt = formatToWebVTT(longVideoAssemblyAIResponse);
const captionTextLines = vtt
.split("\n")
.filter(
(line) =>
line.trim() &&
!line.startsWith("WEBVTT") &&
!line.includes("-->") &&
!/^\d+$/.test(line.trim()),
);
for (const line of captionTextLines) {
expect(line.split(" ").length).toBeLessThanOrEqual(8);
}
});
it("breaks captions at commas", () => {
const vtt = formatToWebVTT({
words: [
{ text: "First,", start: 0, end: 300 },
{ text: "then", start: 400, end: 600 },
{ text: "second.", start: 700, end: 1_000 },
],
});
expect(vtt).toContain("First,");
expect(vtt).toContain("then second.");
});
it("breaks captions at long pauses", () => {
const vtt = formatToWebVTT({
words: [
{ text: "Before", start: 0, end: 300 },
{ text: "after.", start: 1_000, end: 1_300 },
],
});
const captionBlocks = vtt
.split("\n\n")
.filter((block) => block.includes("-->"));
expect(captionBlocks).toHaveLength(2);
});
it("preserves caption ordering with sequential numbers", () => {
const lines = formatToWebVTT(realAssemblyAIResponse).split("\n");
const captionNumbers = lines
.filter((line) => /^\d+$/.test(line.trim()))
.map(Number);
for (let i = 0; i < captionNumbers.length; i++) {
expect(captionNumbers[i]).toBe(i + 1);
}
});
it("includes remaining words after the last punctuation", () => {
const vtt = formatToWebVTT({
words: [
{ text: "Hello.", start: 0, end: 300 },
{ text: "Some", start: 500, end: 700 },
{ text: "trailing", start: 800, end: 1_100 },
{ text: "words", start: 1_200, end: 1_500 },
],
});
expect(vtt).toContain("Hello.");
expect(vtt).toContain("Some trailing words");
});
});