whisper_web/tests/export.test.ts

import { describe, expect, test } from "bun:test";
import {
  exportFilename,
  formatClockTimestamp,
  formatSrtTimestamp,
  transcriptToSrt,
  transcriptToTimestampedTxt,
  transcriptToTxt,
} from "../src/lib/export";
import type { TranscriptChunk } from "../src/worker/protocol";

describe("transcriptToTxt", () => {
  test("trims and appends a single trailing newline", () => {
    expect(transcriptToTxt({ text: "  Hello world. " })).toBe("Hello world.\n");
  });

  test("empty transcript becomes a bare newline", () => {
    expect(transcriptToTxt({ text: "" })).toBe("\n");
  });
});

describe("formatSrtTimestamp", () => {
  test("zero", () => {
    expect(formatSrtTimestamp(0)).toBe("00:00:00,000");
  });

  test("pads milliseconds to three digits", () => {
    expect(formatSrtTimestamp(1.5)).toBe("00:00:01,500");
    expect(formatSrtTimestamp(1.05)).toBe("00:00:01,050");
    expect(formatSrtTimestamp(1.005)).toBe("00:00:01,005");
  });

  test("rolls over past one hour", () => {
    expect(formatSrtTimestamp(3661.25)).toBe("01:01:01,250");
  });

  test("rounds sub-millisecond values", () => {
    expect(formatSrtTimestamp(0.9996)).toBe("00:00:01,000");
  });

  test("clamps negatives to zero", () => {
    expect(formatSrtTimestamp(-1)).toBe("00:00:00,000");
  });
});

describe("transcriptToSrt", () => {
  const chunk = (
    text: string,
    start: number,
    end: number | null,
  ): TranscriptChunk => ({ text, timestamp: [start, end] });

  test("formats numbered cues separated by blank lines", () => {
    const srt = transcriptToSrt([
      chunk(" Hello.", 0, 2.5),
      chunk(" World.", 2.5, 4),
    ]);
    expect(srt).toBe(
      "1\n00:00:00,000 --> 00:00:02,500\nHello.\n" +
        "\n" +
        "2\n00:00:02,500 --> 00:00:04,000\nWorld.\n",
    );
  });

  test("null end borrows the next chunk's start", () => {
    const srt = transcriptToSrt([
      chunk("One", 0, null),
      chunk("Two", 3, 5),
    ]);
    expect(srt).toContain("1\n00:00:00,000 --> 00:00:03,000\nOne");
  });

  test("null end on the final chunk falls back to start + 2s", () => {
    const srt = transcriptToSrt([chunk("Tail", 10, null)]);
    expect(srt).toBe("1\n00:00:10,000 --> 00:00:12,000\nTail\n");
  });

  test("skips empty and whitespace-only chunks (numbering stays contiguous)", () => {
    const srt = transcriptToSrt([
      chunk("  ", 0, 1),
      chunk("Kept", 1, null),
      chunk("", 2, 3),
      chunk("Also kept", 4, 5),
    ]);
    // The null end skips over the empty chunk to the next *captioned* start.
    expect(srt).toBe(
      "1\n00:00:01,000 --> 00:00:04,000\nKept\n" +
        "\n" +
        "2\n00:00:04,000 --> 00:00:05,000\nAlso kept\n",
    );
  });

  test("cue never ends before it starts", () => {
    // Out-of-order borrow: next captioned chunk starts before this one.
    const srt = transcriptToSrt([chunk("A", 5, null), chunk("B", 4, 6)]);
    expect(srt).toContain("00:00:05,000 --> 00:00:05,000");
  });

  test("no captioned chunks yields an empty string", () => {
    expect(transcriptToSrt([])).toBe("");
    expect(transcriptToSrt([chunk("  ", 0, 1)])).toBe("");
  });
});

describe("formatClockTimestamp", () => {
  test("zero", () => {
    expect(formatClockTimestamp(0)).toBe("00:00.0");
  });

  test("pads minutes and seconds", () => {
    expect(formatClockTimestamp(3.21)).toBe("00:03.2");
    expect(formatClockTimestamp(65.4)).toBe("01:05.4");
  });

  test("rounding to tenths carries into the minutes", () => {
    expect(formatClockTimestamp(59.96)).toBe("01:00.0");
  });

  test("minutes keep growing past an hour", () => {
    expect(formatClockTimestamp(3661.5)).toBe("61:01.5");
  });

  test("clamps negatives to zero", () => {
    expect(formatClockTimestamp(-1)).toBe("00:00.0");
  });
});

describe("transcriptToTimestampedTxt", () => {
  const chunk = (
    text: string,
    start: number,
    end: number | null,
  ): TranscriptChunk => ({ text, timestamp: [start, end] });

  test("one bracketed range per line, trailing newline", () => {
    const txt = transcriptToTimestampedTxt([
      chunk(" Hello.", 3.21, 7.9),
      chunk(" World.", 7.9, 12),
    ]);
    expect(txt).toBe("[00:03.2 – 00:07.9] Hello.\n[00:07.9 – 00:12.0] World.\n");
  });

  test("null end omits the range end", () => {
    expect(transcriptToTimestampedTxt([chunk("Tail", 10, null)])).toBe(
      "[00:10.0] Tail\n",
    );
  });

  test("skips empty and whitespace-only chunks", () => {
    const txt = transcriptToTimestampedTxt([
      chunk("  ", 0, 1),
      chunk("Kept", 1, 2),
    ]);
    expect(txt).toBe("[00:01.0 – 00:02.0] Kept\n");
  });

  test("no captioned chunks yields an empty string", () => {
    expect(transcriptToTimestampedTxt([])).toBe("");
    expect(transcriptToTimestampedTxt([chunk(" ", 0, 1)])).toBe("");
  });
});

describe("exportFilename", () => {
  test("swaps the extension", () => {
    expect(exportFilename("talk.m4a", "srt")).toBe("talk.srt");
    expect(exportFilename("talk.m4a", "txt")).toBe("talk.txt");
  });

  test("only the last dot counts", () => {
    expect(exportFilename("2026.07.14 meeting.wav", "srt")).toBe(
      "2026.07.14 meeting.srt",
    );
  });

  test("no extension keeps the whole name", () => {
    expect(exportFilename("recording", "txt")).toBe("recording.txt");
  });

  test("leading dot is a hidden file, not an extension", () => {
    expect(exportFilename(".bashrc", "txt")).toBe(".bashrc.txt");
  });

  test("compound extension keeps timestamped exports distinct", () => {
    expect(exportFilename("talk.m4a", "timestamps.txt")).toBe(
      "talk.timestamps.txt",
    );
    expect(exportFilename(null, "timestamps.txt")).toBe(
      "transcript.timestamps.txt",
    );
  });

  test("null, undefined, and blank fall back to transcript.*", () => {
    expect(exportFilename(null, "srt")).toBe("transcript.srt");
    expect(exportFilename(undefined, "txt")).toBe("transcript.txt");
    expect(exportFilename("   ", "srt")).toBe("transcript.srt");
  });
});