diff options
Diffstat (limited to 'packages/meshbay-node/tests')
| -rw-r--r-- | packages/meshbay-node/tests/test_enrich.py | 41 | ||||
| -rw-r--r-- | packages/meshbay-node/tests/test_media_meta_request.py | 61 | ||||
| -rw-r--r-- | packages/meshbay-node/tests/test_title_parse.py | 22 |
3 files changed, 123 insertions, 1 deletions
diff --git a/packages/meshbay-node/tests/test_enrich.py b/packages/meshbay-node/tests/test_enrich.py index 4b957cd..a35c713 100644 --- a/packages/meshbay-node/tests/test_enrich.py +++ b/packages/meshbay-node/tests/test_enrich.py @@ -373,3 +373,44 @@ async def test_enricher_reads_a_three_digit_episode_number_correctly(tmp_path, m assert fields["season"] == 6 assert fields["episode"] == 100 + + +@pytestmark_ffmpeg +@pytest.mark.asyncio +async def test_a_movie_with_a_mangled_quality_tag_is_not_shelved_as_a_series( + tmp_path, media_cache): + """ + Found live 2026-08-29: a standalone film whose "1080p" tag was + truncated to "108" in the filename makes guessit invent S01E08, so + enrich (flat library, no season ancestor) filed it as a nonexistent + series. A real flat-dumped episode carries an explicit SxxExx / 1x08 / + "Episode N" marker; a movie has a "(2017)"-style year and none. + """ + clip = tmp_path / "Some.Film.2017.MULTI.108.grp.mkv" + _make_clip(clip) + entry = IndexEntry(id="fid-trunc", name=clip.name, path=clip.name, + size=clip.stat().st_size, type="video", added_at=0) + + enricher = Enricher(media_cache) + _, fields = await _run(enricher, entry, clip) + + assert fields.get("season") is None and fields.get("episode") is None, ( + "a movie with a mangled quality tag must not become a series") + assert fields["display_title"] + + +@pytestmark_ffmpeg +@pytest.mark.asyncio +async def test_a_flat_episode_with_a_real_marker_stays_a_show_even_with_a_year( + tmp_path, media_cache): + """The guard must not misfire: a genuine flat-dumped episode that also + carries a year has an explicit SxxExx marker and stays a show.""" + clip = tmp_path / "Some.Show.2022.S01E02.1080p.WEB.mkv" + _make_clip(clip) + entry = IndexEntry(id="fid-marker", name=clip.name, path=clip.name, + size=clip.stat().st_size, type="video", added_at=0) + + enricher = Enricher(media_cache) + _, fields = await _run(enricher, entry, clip) + + assert fields["season"] == 1 and fields["episode"] == 2 diff --git a/packages/meshbay-node/tests/test_media_meta_request.py b/packages/meshbay-node/tests/test_media_meta_request.py index a752825..0a3df12 100644 --- a/packages/meshbay-node/tests/test_media_meta_request.py +++ b/packages/meshbay-node/tests/test_media_meta_request.py @@ -9,7 +9,7 @@ real risk here too, since a season folder routinely holds many episodes. import pytest from cryptography.hazmat.primitives.asymmetric.ed25519 import Ed25519PrivateKey -from meshbay_common.protocol import IndexEntry +from meshbay_common.protocol import MNP, IndexEntry from meshbay_node.indexer.group_index import GroupIndex from meshbay_node.media_cache import MediaCache from meshbay_node.transport.webrtc_server import WebRTCPeerSession @@ -165,3 +165,62 @@ async def test_unknown_file_id_is_refused(media_cache): await session._do_media_meta_request({"file_id": "nope"}) assert session.sent == [{"type": "error", "detail": "File not found"}] + + +async def test_a_not_yet_enriched_entry_with_no_cache_is_answered_without_a_search(media_cache): + """ + A video the indexer has seen but not enriched yet has no display_title + (enrich.py always sets one) and season/episode still None — which the + movie/show split reads as "movie" and hands its raw filename to TMDB's + movie search. During a slow initial scan with a browser on the Videos + tab that is a storm of `search/movie?query=<raw filename>` and bogus + cached matches (found live 2026-08-29). With nothing cached it must + answer confidence 0 and let the client refetch once the enriched fields + arrive — never guess a match from the raw filename. + """ + index = GroupIndex(group_id="g" * 32, sk_node=Ed25519PrivateKey.generate()) + entry = IndexEntry(id="id-raw", name="Show.S01E01.1080p.WEB.mkv", + path="shows/Show", size=1, type="video", added_at=0) + index.add_entry(entry) + client = FakeTmdbClient() + session = _session(index, media_cache, client) + + await session._do_media_meta_request({"file_id": "id-raw"}) + + assert len(session.sent) == 1 + resp = session.sent[0] + assert resp["type"] == MNP.MEDIA_META_RESP + assert resp["file_id"] == "id-raw" + assert resp["confidence"] == 0 + assert "tmdb_id" not in resp + assert client.searched == [], "no TMDB search for a not-yet-enriched video" + + +async def test_a_not_yet_enriched_entry_is_served_from_cache_without_a_search(media_cache): + """ + The point the operator raised: a restart must not re-query TMDB for a + file already resolved. An un-enriched entry (season/episode not yet + populated) whose content hash already has a cached match is served + straight from that cache, honouring the cached *kind* rather than the + provisional "movie" the split would pick — so a show episode keeps its + real "tv" match instead of triggering a fresh movie search on its raw + filename. + """ + index = GroupIndex(group_id="g" * 32, sk_node=Ed25519PrivateKey.generate()) + entry = IndexEntry(id="id-known", name="Show.S02E05.1080p.WEB.mkv", + path="shows/Show", size=1, type="video", added_at=0) + index.add_entry(entry) + await media_cache.set_file_tmdb("id-known", "1396", "tv") + await media_cache.set_tmdb_meta("1396", "tv", { + "title": "The Cached Show", "original_title": "The Cached Show", + "first_air_date": "2008-01-20", "confidence": 1.0, + }) + client = FakeTmdbClient() + session = _session(index, media_cache, client) + + await session._do_media_meta_request({"file_id": "id-known"}) + + resp = session.sent[0] + assert resp["tmdb_id"] == "1396" + assert resp["title"] == "The Cached Show" + assert client.searched == [], "a cached match must not be re-searched on restart" diff --git a/packages/meshbay-node/tests/test_title_parse.py b/packages/meshbay-node/tests/test_title_parse.py index ac4d434..045635f 100644 --- a/packages/meshbay-node/tests/test_title_parse.py +++ b/packages/meshbay-node/tests/test_title_parse.py @@ -7,6 +7,7 @@ is a manual acceptance step (§11), not something this repo's corpus holds. from meshbay_node.indexer.title_parse import ( ParsedName, clean_query, + has_episode_marker, leading_episode_number, naive_title, parse_episode_filename, @@ -113,6 +114,27 @@ def test_clean_query_despaces_a_folder_name_without_eating_the_last_word(): assert naive_title("Some.Show.Name") != "Some Show Name" # the trap it avoids +# ── §10.1/V14: telling a real episode marker from a mangled number ────────── + +def test_has_episode_marker_accepts_real_markers(): + for name in ["Some.Show.S01E08.mkv", "some.show.s1.e8.mkv", + "Some Show 1x08.mkv", "Some Show 01x08.mkv", + "Some Show Episode 8.mkv", "Some Show ep08.mkv", + "Some Show ep.8.mkv", "Some Show Season 1.mkv", + "Une Serie Saison 3.mkv"]: + assert has_episode_marker(name), name + + +def test_has_episode_marker_rejects_bare_numbers_and_ordinary_words(): + # "1080p" truncated to "108", "1280" left in, a year, plain words that + # merely contain "ep" — none of these are episode markers. + for name in ["Some.Film.2017.MULTI.108.grp.mkv", + "Some.Flick.2013.1280.x264-grp.mkv", + "Some Movie 2017.mkv", "The Dark Knight.mkv", + "Sleep 8.mkv", "Deep 8 mm.mkv", "Ocean's 11.mkv"]: + assert not has_episode_marker(name), name + + # ── bug 2026-08-29: guessit peels "Volume N" off the title ───────────────── def test_movie_volume_number_is_folded_back_into_the_title(): |