import { test } from "node:test";
import assert from "node:assert/strict";
import { mkdtemp, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import path from "node:path";
import {
  provenanceFromMetadata,
  readVttProvenance,
  resolveVttProvenance,
  sniffVttProvenance,
  vttTrackName,
} from "./subtitleProvenance";

// Run with: pnpm --filter yt-dlp-transcript-common exec tsx --test lib/subtitleProvenance.test.ts

// Shaped after real yt-dlp --write-auto-subs output: every cue carries
// `align:start position:0%` and inline word timings.
const ASR_HEAD = `WEBVTT
Kind: captions
Language: en

00:00:00.030 --> 00:00:03.919 align:start position:0%

so<00:00:00.719> today<00:00:01.199> we're<00:00:01.439> going<00:00:01.680> to

00:00:03.919 --> 00:00:03.929 align:start position:0%
so today we're going to

00:00:03.929 --> 00:00:07.070 align:start position:0%
so today we're going to
talk<00:00:04.320> about<00:00:04.639> the<00:00:04.879> whole<00:00:05.199> thing
`;

// Shaped after a human-uploaded/manual track: plain cues, no cue settings, no
// inline word timings.
const MANUAL_HEAD = `WEBVTT
Kind: captions
Language: en

00:00:01.000 --> 00:00:04.000
So today we're going to talk about the whole thing.

00:00:04.000 --> 00:00:08.500
It's a long story, but bear with me.

00:00:08.500 --> 00:00:12.000
Here we go.
`;

test("sniffVttProvenance classifies YouTube ASR output as asr", () => {
  assert.equal(sniffVttProvenance(ASR_HEAD), "asr");
});

test("sniffVttProvenance classifies a manual caption track as manual", () => {
  assert.equal(sniffVttProvenance(MANUAL_HEAD), "manual");
});

test("sniffVttProvenance detects asr from word timings alone", () => {
  // Some ASR tracks are served without the align/position cue settings but keep
  // the inline word timings — still machine-generated.
  const head = ASR_HEAD.replace(/ align:start position:\d+%/g, "");
  assert.equal(sniffVttProvenance(head), "asr");
});

test("sniffVttProvenance returns unknown for too little evidence", () => {
  assert.equal(sniffVttProvenance(""), "unknown");
  assert.equal(sniffVttProvenance("not a vtt at all"), "unknown");
  assert.equal(
    sniffVttProvenance("WEBVTT\n\n00:00:00.000 --> 00:00:02.000\nOnly one cue.\n"),
    "unknown",
  );
});

test("sniffVttProvenance returns unknown when markers are present but sparse", () => {
  const head = `WEBVTT

00:00:00.000 --> 00:00:02.000 align:start position:0%
one

00:00:02.000 --> 00:00:04.000
two

00:00:04.000 --> 00:00:06.000
three

00:00:06.000 --> 00:00:08.000
four
`;
  assert.equal(sniffVttProvenance(head), "unknown");
});

test("provenanceFromMetadata prefers subtitles over automatic_captions", () => {
  const meta = {
    subtitles: { en: [{ ext: "vtt" }] },
    automatic_captions: { en: [{ ext: "vtt" }], "en-orig": [{ ext: "vtt" }] },
  };
  assert.equal(provenanceFromMetadata(meta, "en"), "manual");
  assert.equal(provenanceFromMetadata(meta, "en-orig"), "asr");
  assert.equal(provenanceFromMetadata(meta, "fr"), "unknown");
  assert.equal(provenanceFromMetadata(null, "en"), "unknown");
  assert.equal(provenanceFromMetadata({}, "en"), "unknown");
});

test("vttTrackName extracts the language segment", () => {
  assert.equal(vttTrackName("transcript.en.vtt"), "en");
  assert.equal(vttTrackName("transcript.en-US.vtt"), "en-US");
  assert.equal(vttTrackName("transcript.json"), null);
  assert.equal(vttTrackName("transcript.cues.json"), null);
});

test("readVttProvenance sniffs only the head of a file on disk", async () => {
  const dir = await mkdtemp(path.join(tmpdir(), "subprov-"));
  await writeFile(path.join(dir, "transcript.en.vtt"), ASR_HEAD);
  await writeFile(path.join(dir, "transcript.en-US.vtt"), MANUAL_HEAD);
  assert.equal(await readVttProvenance(dir, "transcript.en.vtt"), "asr");
  assert.equal(await readVttProvenance(dir, "transcript.en-US.vtt"), "manual");
  assert.equal(await readVttProvenance(dir, "missing.vtt"), "unknown");
});

test("resolveVttProvenance falls back to metadata for an undecidable file", async () => {
  const dir = await mkdtemp(path.join(tmpdir(), "subprov-"));
  // One cue only: the sniff can't judge it.
  await writeFile(
    path.join(dir, "transcript.en.vtt"),
    "WEBVTT\n\n00:00:00.000 --> 00:00:02.000\nhi\n",
  );
  assert.equal(await resolveVttProvenance(dir, "transcript.en.vtt"), "unknown");
  await writeFile(
    path.join(dir, "metadata.info.json"),
    JSON.stringify({ automatic_captions: { en: [{ ext: "vtt" }] } }),
  );
  assert.equal(await resolveVttProvenance(dir, "transcript.en.vtt"), "asr");
});