Files
dl-organizer/tests/test_analysis.py
T

589 lines
22 KiB
Python
Raw Normal View History

2026-02-09 17:43:35 +08:00
"""Unit tests for the analysis engine.
Tests series completeness analysis, duplicate detection, and quality comparison.
"""
from datetime import datetime, timezone
from pathlib import Path
from vlm.analysis import analyze_series_completeness, compare_quality, detect_duplicates
from vlm.models import MovieIdentity, SeriesIdentity, VideoFile
2026-02-09 17:43:35 +08:00
def _movie(title="Movie", year=2020, **kw):
return MovieIdentity(
title=title, year=year, confidence=kw.pop("confidence", 0.9),
needs_review=kw.pop("needs_review", False),
original_filename=kw.pop("original_filename", f"{title}.{year}.mkv"),
**kw,
)
def _series(title="Show", season=1, episodes=None, **kw):
if episodes is None:
episodes = [1]
return SeriesIdentity(
title=title, season=season, episodes=episodes,
confidence=kw.pop("confidence", 0.9),
needs_review=kw.pop("needs_review", False),
original_filename=kw.pop(
"original_filename",
f"{title.replace(' ', '.')}.S{season:02d}E01.mkv"
if season is not None
else f"{title.replace(' ', '.')}.E01.mkv",
),
**kw,
)
def _video(filename="file.mkv", size=1000, category="movie", **kw):
return VideoFile(
path=kw.pop("path", Path(f"/tmp/{filename}")),
filename=filename, size_bytes=size,
modified_timestamp=kw.pop("modified_timestamp", datetime(2023, 1, 1)),
category=category, **kw,
)
2026-02-09 17:43:35 +08:00
class TestSeriesCompletenessAnalysis:
"""Test series completeness analysis functionality."""
2026-02-09 17:43:35 +08:00
def test_detect_single_gap(self):
"""Test detection of a single missing episode."""
episodes = [
_series("Show Name", episodes=[1],
original_filename="Show.Name.S01E01.mkv"),
_series("Show Name", episodes=[2],
original_filename="Show.Name.S01E02.mkv"),
_series("Show Name", episodes=[4],
original_filename="Show.Name.S01E04.mkv"),
_series("Show Name", episodes=[5],
original_filename="Show.Name.S01E05.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
assert len(result) == 1
assert result[0].series_title == "Show Name"
assert result[0].season == 1
assert result[0].episodes_found == [1, 2, 4, 5]
assert result[0].episodes_missing == [3]
2026-02-09 17:43:35 +08:00
def test_detect_multiple_gaps(self):
"""Test detection of multiple missing episodes."""
episodes = [
_series("Show Name", episodes=[1],
original_filename="Show.Name.S01E01.mkv"),
_series("Show Name", episodes=[3],
original_filename="Show.Name.S01E03.mkv"),
_series("Show Name", episodes=[5],
original_filename="Show.Name.S01E05.mkv"),
_series("Show Name", episodes=[7],
original_filename="Show.Name.S01E07.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
assert len(result) == 1
assert result[0].episodes_found == [1, 3, 5, 7]
assert result[0].episodes_missing == [2, 4, 6]
2026-02-09 17:43:35 +08:00
def test_no_gaps_returns_empty(self):
"""Test that complete seasons are not included in results."""
episodes = [
_series("Show Name", episodes=[1],
original_filename="Show.Name.S01E01.mkv"),
_series("Show Name", episodes=[2],
original_filename="Show.Name.S01E02.mkv"),
_series("Show Name", episodes=[3],
original_filename="Show.Name.S01E03.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_multi_season_independence(self):
"""Test that gap detection for one season doesn't affect others."""
episodes = [
# Season 1 - has gap at episode 2
_series("Show Name", season=1, episodes=[1],
original_filename="Show.Name.S01E01.mkv"),
_series("Show Name", season=1, episodes=[3],
original_filename="Show.Name.S01E03.mkv"),
2026-02-09 17:43:35 +08:00
# Season 2 - complete
_series("Show Name", season=2, episodes=[1],
original_filename="Show.Name.S02E01.mkv"),
_series("Show Name", season=2, episodes=[2],
original_filename="Show.Name.S02E02.mkv"),
2026-02-09 17:43:35 +08:00
# Season 3 - has gap at episode 5
_series("Show Name", season=3, episodes=[4],
original_filename="Show.Name.S03E04.mkv"),
_series("Show Name", season=3, episodes=[6],
original_filename="Show.Name.S03E06.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
# Should have 2 results (seasons 1 and 3 with gaps)
assert len(result) == 2
2026-02-09 17:43:35 +08:00
# Find season 1 result
season1 = next(r for r in result if r.season == 1)
assert season1.episodes_found == [1, 3]
assert season1.episodes_missing == [2]
2026-02-09 17:43:35 +08:00
# Find season 3 result
season3 = next(r for r in result if r.season == 3)
assert season3.episodes_found == [4, 6]
assert season3.episodes_missing == [5]
2026-02-09 17:43:35 +08:00
def test_multi_episode_files(self):
"""Test handling of multi-episode files."""
episodes = [
_series("Show Name", episodes=[1, 2],
original_filename="Show.Name.S01E01-E02.mkv"),
_series("Show Name", episodes=[4],
original_filename="Show.Name.S01E04.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
assert len(result) == 1
assert result[0].episodes_found == [1, 2, 4]
assert result[0].episodes_missing == [3]
2026-02-09 17:43:35 +08:00
def test_different_series_separate_analysis(self):
"""Test that different series are analyzed separately."""
episodes = [
# Series A - has gap
_series("Series A", episodes=[1],
original_filename="Series.A.S01E01.mkv"),
_series("Series A", episodes=[3],
original_filename="Series.A.S01E03.mkv"),
2026-02-09 17:43:35 +08:00
# Series B - complete
_series("Series B", episodes=[1],
original_filename="Series.B.S01E01.mkv"),
_series("Series B", episodes=[2],
original_filename="Series.B.S01E02.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
# Only Series A should be in results
assert len(result) == 1
assert result[0].series_title == "Series A"
assert result[0].episodes_missing == [2]
2026-02-09 17:43:35 +08:00
def test_skip_episodes_without_season(self):
"""Test that episodes with season=None are excluded from analysis."""
episodes = [
_series("Show Name", episodes=[1],
original_filename="Show.Name.S01E01.mkv"),
_series("Show Name", episodes=[2],
original_filename="Show.Name.S01E02.mkv"),
_series("Show Name", season=None, confidence=0.3,
needs_review=True,
original_filename="Show.Name.Episode.1.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
# Should only analyze season 1, which is complete
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_skip_episodes_with_empty_episode_list(self):
"""Test that episodes with empty episode list are excluded from analysis."""
episodes = [
_series("Show Name", episodes=[1],
original_filename="Show.Name.S01E01.mkv"),
_series("Show Name", episodes=[3],
original_filename="Show.Name.S01E03.mkv"),
_series("Show Name", episodes=[], confidence=0.3,
needs_review=True,
original_filename="Show.Name.S01.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
# Should detect gap at episode 2
assert len(result) == 1
assert result[0].episodes_missing == [2]
2026-02-09 17:43:35 +08:00
def test_non_sequential_start(self):
"""Test gap detection when episodes don't start at 1."""
episodes = [
_series("Show Name", episodes=[5],
original_filename="Show.Name.S01E05.mkv"),
_series("Show Name", episodes=[6],
original_filename="Show.Name.S01E06.mkv"),
_series("Show Name", episodes=[8],
original_filename="Show.Name.S01E08.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
# Should detect gap at episode 7 in range [5, 8]
assert len(result) == 1
assert result[0].episodes_found == [5, 6, 8]
assert result[0].episodes_missing == [7]
2026-02-09 17:43:35 +08:00
def test_empty_input(self):
"""Test handling of empty episode list."""
result = analyze_series_completeness([])
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_single_episode_no_gap(self):
"""Test that a single episode has no gaps."""
episodes = [
_series("Show Name", episodes=[1],
original_filename="Show.Name.S01E01.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = analyze_series_completeness(episodes)
2026-02-09 17:43:35 +08:00
# Single episode has no gaps
assert len(result) == 0
class TestDuplicateDetection:
"""Test duplicate detection functionality."""
2026-02-09 17:43:35 +08:00
def test_detect_movie_duplicates(self):
"""Test detection of duplicate movies with identical title and year."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_movie("The Matrix", 1999,
original_filename="The.Matrix.1999.1080p.mkv"),
_movie("The Matrix", 1999,
original_filename="The.Matrix.1999.720p.mkv"),
_movie("Inception", 2010),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("The.Matrix.1999.1080p.mkv", 2_000_000_000,
modified_timestamp=now, resolution="1920x1080",
codec="h264"),
_video("The.Matrix.1999.720p.mkv", 1_000_000_000,
modified_timestamp=now, resolution="1280x720",
codec="h264"),
_video("Inception.2010.mkv", 1_500_000_000,
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
# Should find one duplicate group (The Matrix)
assert len(result) == 1
assert isinstance(result[0].identity, MovieIdentity)
assert result[0].identity.title == "The Matrix"
assert result[0].identity.year == 1999
assert len(result[0].files) == 2
assert len(result[0].quality_comparison) == 2
2026-02-09 17:43:35 +08:00
def test_detect_series_duplicates(self):
"""Test detection of duplicate series episodes."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_series("Breaking Bad", episodes=[1],
original_filename="Breaking.Bad.S01E01.1080p.mkv"),
_series("Breaking Bad", episodes=[1],
original_filename="Breaking.Bad.S01E01.720p.mkv"),
_series("Breaking Bad", episodes=[2],
original_filename="Breaking.Bad.S01E02.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Breaking.Bad.S01E01.1080p.mkv", 1_500_000_000,
"series", modified_timestamp=now,
resolution="1920x1080"),
_video("Breaking.Bad.S01E01.720p.mkv", 800_000_000,
"series", modified_timestamp=now,
resolution="1280x720"),
_video("Breaking.Bad.S01E02.mkv", 1_200_000_000,
"series", modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
# Should find one duplicate group (S01E01)
assert len(result) == 1
assert isinstance(result[0].identity, SeriesIdentity)
assert result[0].identity.title == "Breaking Bad"
assert result[0].identity.season == 1
assert 1 in result[0].identity.episodes
assert len(result[0].files) == 2
2026-02-09 17:43:35 +08:00
def test_no_duplicates(self):
"""Test that unique files are not flagged as duplicates."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_movie("Movie A", 2020),
_movie("Movie B", 2021),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Movie.A.2020.mkv", 1_000_000_000,
modified_timestamp=now),
_video("Movie.B.2021.mkv", 1_000_000_000,
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_skip_movies_without_year(self):
"""Test that movies without year are excluded from duplicate detection."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_movie("Unknown Movie", year=None, confidence=0.3,
needs_review=True,
original_filename="Unknown.Movie.mkv"),
_movie("Unknown Movie", year=None, confidence=0.3,
needs_review=True,
original_filename="Unknown.Movie.2.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Unknown.Movie.mkv", 1_000_000_000,
modified_timestamp=now),
_video("Unknown.Movie.2.mkv", 1_000_000_000,
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
# Should not detect duplicates for files needing review
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_skip_series_without_season(self):
"""Test that series without season are excluded from duplicate detection."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_series("Unknown Show", season=None, confidence=0.3,
needs_review=True,
original_filename="Unknown.Show.E01.mkv"),
_series("Unknown Show", season=None, confidence=0.3,
needs_review=True,
original_filename="Unknown.Show.Episode.1.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Unknown.Show.E01.mkv", 1_000_000_000, "series",
modified_timestamp=now),
_video("Unknown.Show.Episode.1.mkv", 1_000_000_000, "series",
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_skip_series_with_empty_episodes(self):
"""Test that series with empty episode list are excluded."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_series("Show Name", episodes=[], confidence=0.3,
needs_review=True,
original_filename="Show.Name.S01.mkv"),
_series("Show Name", episodes=[], confidence=0.3,
needs_review=True,
original_filename="Show.Name.Season.1.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Show.Name.S01.mkv", 1_000_000_000, "series",
modified_timestamp=now),
_video("Show.Name.Season.1.mkv", 1_000_000_000, "series",
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_quality_comparison_includes_all_metadata(self):
"""Test that quality comparison includes all available metadata."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_movie("Test Movie", 2020,
original_filename="Test.Movie.2020.1080p.mkv"),
_movie("Test Movie", 2020,
original_filename="Test.Movie.2020.720p.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Test.Movie.2020.1080p.mkv", 2_000_000_000,
modified_timestamp=now, resolution="1920x1080",
codec="h264", duration_seconds=7200.0,
bitrate_kbps=5000),
_video("Test.Movie.2020.720p.mkv", 1_000_000_000,
modified_timestamp=now, resolution="1280x720",
codec="h264", duration_seconds=7200.0,
bitrate_kbps=2500),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
assert len(result) == 1
comparison = result[0].quality_comparison
assert len(comparison) == 2
2026-02-09 17:43:35 +08:00
# Check first file comparison data
assert comparison[0]['filename'] == "Test.Movie.2020.1080p.mkv"
assert comparison[0]['size_bytes'] == 2000000000
assert comparison[0]['resolution'] == "1920x1080"
assert comparison[0]['codec'] == "h264"
assert comparison[0]['duration_seconds'] == 7200.0
assert comparison[0]['bitrate_kbps'] == 5000
2026-02-09 17:43:35 +08:00
# Check second file comparison data
assert comparison[1]['filename'] == "Test.Movie.2020.720p.mkv"
assert comparison[1]['size_bytes'] == 1000000000
assert comparison[1]['resolution'] == "1280x720"
2026-02-09 17:43:35 +08:00
def test_multi_episode_file_duplicates(self):
"""Test duplicate detection for multi-episode files."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_series("Show", episodes=[1, 2],
original_filename="Show.S01E01-E02.mkv"),
_series("Show", episodes=[1],
original_filename="Show.S01E01.mkv"),
_series("Show", episodes=[2],
original_filename="Show.S01E02.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Show.S01E01-E02.mkv", 2_000_000_000, "series",
modified_timestamp=now),
_video("Show.S01E01.mkv", 1_000_000_000, "series",
modified_timestamp=now),
_video("Show.S01E02.mkv", 1_000_000_000, "series",
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
# Should find duplicates for both E01 and E02
assert len(result) == 2
2026-02-09 17:43:35 +08:00
def test_different_years_not_duplicates(self):
"""Test that same title with different years are not duplicates."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_movie("The Thing", 1982),
_movie("The Thing", 2011),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("The.Thing.1982.mkv", 1_000_000_000,
modified_timestamp=now),
_video("The.Thing.2011.mkv", 1_000_000_000,
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_different_seasons_not_duplicates(self):
"""Test that same series/episode in different seasons are not duplicates."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
identities = [
_series("Show", season=1,
original_filename="Show.S01E01.mkv"),
_series("Show", season=2,
original_filename="Show.S02E01.mkv"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
files = [
_video("Show.S01E01.mkv", 1_000_000_000, "series",
modified_timestamp=now),
_video("Show.S02E01.mkv", 1_000_000_000, "series",
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
result = detect_duplicates(list(zip(identities, files)))
2026-02-09 17:43:35 +08:00
assert len(result) == 0
class TestQualityComparison:
"""Test quality comparison functionality."""
2026-02-09 17:43:35 +08:00
def test_compare_quality_with_all_metadata(self):
"""Test quality comparison with all metadata available."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
files = [
_video("file1.mkv", 2_000_000_000,
modified_timestamp=now, resolution="1920x1080",
codec="h264", duration_seconds=7200.0,
bitrate_kbps=5000),
_video("file2.mkv", 1_000_000_000,
modified_timestamp=now, resolution="1280x720",
codec="h265", duration_seconds=7200.0,
bitrate_kbps=2500),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = compare_quality(files)
2026-02-09 17:43:35 +08:00
assert len(result) == 2
assert result[0]['filename'] == "file1.mkv"
assert result[0]['size_bytes'] == 2000000000
assert result[0]['resolution'] == "1920x1080"
assert result[0]['codec'] == "h264"
assert result[0]['duration_seconds'] == 7200.0
assert result[0]['bitrate_kbps'] == 5000
2026-02-09 17:43:35 +08:00
assert result[1]['filename'] == "file2.mkv"
assert result[1]['size_bytes'] == 1000000000
assert result[1]['resolution'] == "1280x720"
assert result[1]['codec'] == "h265"
2026-02-09 17:43:35 +08:00
def test_compare_quality_with_partial_metadata(self):
"""Test quality comparison when some metadata is missing."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
files = [
_video("file1.mkv", 2_000_000_000,
modified_timestamp=now, resolution="1920x1080"),
_video("file2.mkv", 1_000_000_000,
modified_timestamp=now),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = compare_quality(files)
2026-02-09 17:43:35 +08:00
assert len(result) == 2
assert result[0]['filename'] == "file1.mkv"
assert result[0]['size_bytes'] == 2000000000
assert result[0]['resolution'] == "1920x1080"
assert 'codec' not in result[0]
assert 'duration_seconds' not in result[0]
assert 'bitrate_kbps' not in result[0]
2026-02-09 17:43:35 +08:00
assert result[1]['filename'] == "file2.mkv"
assert result[1]['size_bytes'] == 1000000000
assert 'resolution' not in result[1]
assert 'codec' not in result[1]
2026-02-09 17:43:35 +08:00
def test_compare_quality_empty_list(self):
"""Test quality comparison with empty file list."""
result = compare_quality([])
assert len(result) == 0
2026-02-09 17:43:35 +08:00
def test_compare_quality_single_file(self):
"""Test quality comparison with single file."""
now = datetime.now(timezone.utc)
2026-02-09 17:43:35 +08:00
files = [
_video("file.mkv", 1_500_000_000,
modified_timestamp=now, resolution="1920x1080",
codec="h264"),
2026-02-09 17:43:35 +08:00
]
2026-02-09 17:43:35 +08:00
result = compare_quality(files)
2026-02-09 17:43:35 +08:00
assert len(result) == 1
assert result[0]['filename'] == "file.mkv"
assert result[0]['size_bytes'] == 1500000000