youtube_abschrift_korrigieren/tests/test_yt_untertitel.py

"""Tests for yt_untertitel.py; yt-dlp is always replaced by fakes (no network)."""
import subprocess
import sys
from pathlib import Path

import pytest

import yt_untertitel as yt

URL = "https://www.youtube.com/watch?v=abc123"

AUTO_ONLY = {
    "id": "abc123",
    "title": "Eine Rede",
    "channel": "Kanal",
    "upload_date": "20260328",
    "subtitles": {},
    "automatic_captions": {"de-orig": [], "de": [], "en": []},
}


pytestmark = pytest.mark.usefixtures("yt_dlp_installed")


def run_main(monkeypatch, *args):
    monkeypatch.setattr(sys, "argv", ["yt_untertitel.py", *map(str, args)])
    yt.main()


# === Track selection ===

@pytest.mark.parametrize("info, lang, expected", [
    ({"subtitles": {"de": []}, "automatic_captions": {"de-orig": []}}, "de", ("de", False)),
    (AUTO_ONLY, "de", ("de-orig", True)),
    # "en" exists only as machine translation of the German original
    (AUTO_ONLY, "en", None),
    ({"subtitles": None, "automatic_captions": None}, "de", None),
])
def test_choose_track(info, lang, expected):
    assert yt.choose_track(info, lang) == expected


@pytest.mark.parametrize("info, expected", [
    (AUTO_ONLY, "de"),
    ({"language": "fr", "automatic_captions": {}}, "fr"),
    ({"subtitles": {"en": []}}, None),
])
def test_original_language(info, expected):
    assert yt.original_language(info) == expected


# English original with AI-dubbed audio tracks: every dub has its own "-orig"
DUBBED = {
    "id": "dub",
    "language": "en-US",
    "subtitles": {"en-US": []},
    "automatic_captions": {"ar-orig": [], "en-orig": [], "de-DE-orig": [], "de": []},
}
DUBBED_NO_METADATA = {
    "formats": [
        {"format_note": "Arabic, low", "language": "ar"},
        {"format_note": "English (US) original (default), low", "language": "en-US"},
    ],
    "subtitles": {},
    "automatic_captions": {"ar-orig": [], "en-orig": [], "de-DE-orig": []},
}


@pytest.mark.parametrize("info, lang, expected", [
    (DUBBED, "en", ("en-US", False)),          # region variant of a manual track
    (DUBBED, "en-US", ("en-US", False)),
    (DUBBED, "de", None),                      # only the German AI dub
    (DUBBED, "ar", None),
    (DUBBED_NO_METADATA, "en-US", ("en-orig", True)),
    (DUBBED_NO_METADATA, "en", ("en-orig", True)),
    (DUBBED_NO_METADATA, "de", None),
])
def test_choose_track_with_dubbed_audio(info, lang, expected):
    assert yt.choose_track(info, lang) == expected


@pytest.mark.parametrize("info, expected", [
    (DUBBED, "en-US"),
    (DUBBED_NO_METADATA, "en-US"),             # from the audio track marked original
    ({"automatic_captions": {"ar-orig": [], "en-orig": []}}, None),
])
def test_original_language_with_dubbed_audio(info, expected):
    assert yt.original_language(info) == expected


@pytest.mark.parametrize("info, expected", [
    ({"channel": "John Mearsheimer - Offensive Realism",
      "title": "Episode 1: Origins of the War In Ukraine", "upload_date": "20260730"},
     "John_Mearsheimer_Offensive_Realism_Episode_1_Origins_of_the_War_In_Ukraine_2026-07-30"),
    # Title cut at a word boundary after 60 characters; hyphens and dots in words stay
    ({"channel": "GGBLahnstein", "upload_date": "20260730",
      "title": "Prof. Gabriele Krone-Schmalz – Respekt geht anders – "
               "GGB-Gesundheitstage, Lahnstein 28.03.2026"},
     "GGBLahnstein_Prof_Gabriele_Krone-Schmalz_Respekt_geht_anders_2026-07-30"),
    ({"uploader": "Ein Kanal", "title": "A/B: Krieg & Frieden 🎤?"}, "Ein_Kanal_A_B_Krieg_Frieden"),
    ({"title": "Nur Titel", "upload_date": "kaputt"}, "Nur_Titel"),
    ({"id": "abc123", "title": "🎤 – …"}, "abc123"),
    ({"title": "x" * 80}, "x" * 60),
])
def test_default_name(info, expected):
    assert yt.default_name(info) == expected


def test_available_tracks_hides_dubs():
    assert yt.available_tracks(DUBBED) == (["en-US"], ["en"])


def test_available_tracks():
    info = {"subtitles": {"fr": []}, "automatic_captions": {"de-orig": [], "de": []}}
    assert yt.available_tracks(info) == (["fr"], ["de"])


# === Command line ===

def test_main_downloads_auto_original(fake_yt_dlp, tmp_path, monkeypatch, capsys):
    run = fake_yt_dlp(AUTO_ONLY)
    monkeypatch.setattr(yt.subprocess, "run", run)
    run_main(monkeypatch, "-u", URL, "-n", "Rede", "-d", tmp_path / "neu")

    assert (tmp_path / "neu" / "Rede.de-orig.vtt").is_file()
    download = run.calls[1]
    assert "--write-auto-subs" in download and "--write-subs" not in download
    assert download[-1] == URL
    out = capsys.readouterr().out
    assert "automatische Untertitel „de-orig“" in out
    assert "vtt2txt.py -i" in out


def test_yt_dlp_runs_in_temporary_directory(fake_yt_dlp, tmp_path, monkeypatch):
    # ffmpeg -bsfs, called by yt-dlp, leaves a file "3" in its working directory
    run = fake_yt_dlp(AUTO_ONLY)
    monkeypatch.setattr(yt.subprocess, "run", run)
    monkeypatch.chdir(tmp_path)
    run_main(monkeypatch, "-u", URL, "-n", "Rede")

    assert all(cwd is not None and Path(cwd) != tmp_path for cwd in run.cwds)
    template = run.calls[1][run.calls[1].index("-o") + 1]
    # Without -d: working folder abschriften/<Name>/, named after -n
    work = tmp_path / "abschriften" / "Rede"
    assert template == f"{work / 'Rede'}.%(ext)s"
    assert (work / "Rede.de-orig.vtt").is_file()
    assert (work / "Rede.quelle.json").is_file()


def test_main_prefers_manual_and_default_name(fake_yt_dlp, tmp_path, monkeypatch):
    info = dict(AUTO_ONLY, subtitles={"de": []})
    run = fake_yt_dlp(info)
    monkeypatch.setattr(yt.subprocess, "run", run)
    run_main(monkeypatch, "-u", URL, "-d", tmp_path)

    assert (tmp_path / "Kanal_Eine_Rede_2026-03-28.de.vtt").is_file()
    assert "--write-subs" in run.calls[1]


def test_main_no_matching_track(fake_yt_dlp, tmp_path, monkeypatch):
    monkeypatch.setattr(yt.subprocess, "run", fake_yt_dlp(AUTO_ONLY))
    with pytest.raises(SystemExit, match="Automatisch \\(Originalsprache\\): de"):
        run_main(monkeypatch, "-u", URL, "-l", "en", "-d", tmp_path)


def test_main_info_error(tmp_path, monkeypatch):
    def run(command, capture_output, text, cwd=None):
        return subprocess.CompletedProcess(command, 1, "", "ERROR: Video unavailable")

    monkeypatch.setattr(yt.subprocess, "run", run)
    with pytest.raises(SystemExit, match="Video unavailable"):
        run_main(monkeypatch, "-u", URL, "-d", tmp_path)


@pytest.mark.parametrize("kwargs, message", [
    ({"fail_download": True}, "HTTP Error 429"),
    ({"write": False}, "keine Datei"),
])
def test_main_download_errors(fake_yt_dlp, tmp_path, monkeypatch, kwargs, message):
    monkeypatch.setattr(yt.subprocess, "run", fake_yt_dlp(AUTO_ONLY, **kwargs))
    with pytest.raises(SystemExit, match=message):
        run_main(monkeypatch, "-u", URL, "-d", tmp_path)


@pytest.mark.parametrize("args, message", [
    ([], "required"),
    (["-u", URL, "-n", "sub/Rede"], "ohne Verzeichnis"),
    (["-u", URL, "-n", ""], "ohne Verzeichnis"),
])
def test_cli_errors(monkeypatch, capsys, args, message):
    with pytest.raises(SystemExit) as exc:
        run_main(monkeypatch, *args)
    assert exc.value.code == 2
    assert message in capsys.readouterr().err


def test_cli_without_yt_dlp(monkeypatch, capsys):
    monkeypatch.setattr(yt.shutil, "which", lambda name: None)
    with pytest.raises(SystemExit):
        run_main(monkeypatch, "-u", URL)
    assert "pip install yt-dlp" in capsys.readouterr().err