Add comprehensive tests for formatter classes, services, and utilities

- Introduced tests for various formatter classes including TextFormatter, DurationFormatter, SizeFormatter, DateFormatter, and more to ensure correct formatting behavior. - Added tests for service classes such as FileTreeService, MetadataService, and RenameService, covering directory validation, metadata extraction, and file renaming functionalities. - Implemented utility tests for LanguageCodeExtractor, PatternExtractor, and FrameClassMatcher to validate their extraction and matching capabilities. - Updated test cases to use datasets for better maintainability and clarity. - Enhanced error handling tests to ensure robustness against missing or invalid data.
2026-08-28 11:33:25 +00:00 · 2025-12-31 14:04:33 +00:00
parent c5fbd367fc
commit 262c0a7b7d
45 changed files with 3346 additions and 1057 deletions
@@ -0,0 +1,45 @@
+"""Constants package for Renamer.
+
+This package contains constants split into logical modules:
+- media_constants.py: Media type definitions (MEDIA_TYPES)
+- source_constants.py: Video source types (SOURCE_DICT)
+- frame_constants.py: Resolution/frame classes (FRAME_CLASSES)
+- moviedb_constants.py: Movie database identifiers (MOVIE_DB_DICT)
+- edition_constants.py: Special edition types (SPECIAL_EDITIONS)
+- lang_constants.py: Language-related constants (SKIP_WORDS)
+- year_constants.py: Year validation (CURRENT_YEAR, MIN_VALID_YEAR, etc.)
+- cyrillic_constants.py: Cyrillic character normalization (CYRILLIC_TO_ENGLISH)
+"""
+
+# Import from all constant modules
+from .media_constants import MEDIA_TYPES
+from .source_constants import SOURCE_DICT
+from .frame_constants import FRAME_CLASSES, NON_STANDARD_QUALITY_INDICATORS
+from .moviedb_constants import MOVIE_DB_DICT
+from .edition_constants import SPECIAL_EDITIONS
+from .lang_constants import SKIP_WORDS
+from .year_constants import CURRENT_YEAR, MIN_VALID_YEAR, YEAR_FUTURE_BUFFER, is_valid_year
+from .cyrillic_constants import CYRILLIC_TO_ENGLISH
+
+__all__ = [
+    # Media types
+    'MEDIA_TYPES',
+    # Source types
+    'SOURCE_DICT',
+    # Frame classes
+    'FRAME_CLASSES',
+    'NON_STANDARD_QUALITY_INDICATORS',
+    # Movie databases
+    'MOVIE_DB_DICT',
+    # Special editions
+    'SPECIAL_EDITIONS',
+    # Language constants
+    'SKIP_WORDS',
+    # Year validation
+    'CURRENT_YEAR',
+    'MIN_VALID_YEAR',
+    'YEAR_FUTURE_BUFFER',
+    'is_valid_year',
+    # Cyrillic normalization
+    'CYRILLIC_TO_ENGLISH',
+]
@@ -0,0 +1,21 @@
+"""Cyrillic character normalization constants.
+
+This module contains mappings for normalizing Cyrillic characters to their
+English equivalents for parsing filenames.
+"""
+
+# Cyrillic to English character mappings
+# Used for normalizing Cyrillic characters that look like English letters
+CYRILLIC_TO_ENGLISH = {
+    'р': 'p',  # Cyrillic 'er' looks like Latin 'p'
+    'і': 'i',  # Cyrillic 'i' looks like Latin 'i'
+    'о': 'o',  # Cyrillic 'o' looks like Latin 'o'
+    'с': 'c',  # Cyrillic 'es' looks like Latin 'c'
+    'е': 'e',  # Cyrillic 'ie' looks like Latin 'e'
+    'а': 'a',  # Cyrillic 'a' looks like Latin 'a'
+    'т': 't',  # Cyrillic 'te' looks like Latin 't'
+    'у': 'y',  # Cyrillic 'u' looks like Latin 'y'
+    'к': 'k',  # Cyrillic 'ka' looks like Latin 'k'
+    'х': 'x',  # Cyrillic 'ha' looks like Latin 'x
+    # Add more mappings as needed
+}
@@ -0,0 +1,59 @@
+"""Special edition constants.
+
+This module defines special edition types (Director's Cut, Extended Edition, etc.)
+and their aliases for detection in filenames.
+"""
+
+SPECIAL_EDITIONS = {
+    "Theatrical Cut": ["Theatrical Cut"],
+    "Director's Cut": ["Director's Cut", "Director Cut"],
+    "Extended Edition": ["Extended Edition", "Ultimate Extended Edition"],
+    "Special Edition": ["Special Edition"],
+    "Collector's Edition": ["Collector's Edition"],
+    "Criterion Collection": ["Criterion Collection"],
+    "Anniversary Edition": ["Anniversary Edition"],
+    "Redux": ["Redux"],
+    "Final Cut": ["Final Cut"],
+    "Alternate Cut": ["Alternate Cut"],
+    "International Cut": ["International Cut"],
+    "Restored Edition": [
+        "Restored Edition",
+        "Restored Version",
+        "4K Restoration",
+        "Restoration",
+    ],
+    "Remastered": ["Remastered", "Remaster", "HD Remaster"],
+    "Unrated": ["Unrated"],
+    "Uncensored": ["Uncensored"],
+    "Definitive Edition": ["Definitive Edition"],
+    "Platinum Edition": ["Platinum Edition"],
+    "Gold Edition": ["Gold Edition"],
+    "Diamond Edition": ["Diamond Edition"],
+    "Steelbook Edition": ["Steelbook Edition"],
+    "Limited Edition": ["Limited Edition"],
+    "Deluxe Edition": ["Deluxe Edition"],
+    "Premium Edition": ["Premium Edition"],
+    "Complete Edition": ["Complete Edition"],
+    "AI Remaster": ["AI Remaster", "AI Remastered"],
+    "Upscaled": [
+        "AI Upscaled",
+        "AI Enhanced",
+        "AI Upscale",
+        "Upscaled",
+        "Upscale",
+        "Upscaling",
+    ],
+    "Director's Definitive Cut": ["Director's Definitive Cut"],
+    "Extended Director's Cut": ["Extended Director's Cut", "Ultimate Director's Cut"],
+    "Original Cut": ["Original Cut"],
+    "Cinematic Cut": ["Cinematic Cut"],
+    "Roadshow Cut": ["Roadshow Cut"],
+    "Premiere Cut": ["Premiere Cut"],
+    "Festival Cut": ["Festival Cut"],
+    "Workprint": ["Workprint"],
+    "Rough Cut": ["Rough Cut"],
+    "Special Assembly Cut": ["Special Assembly Cut"],
+    "Amazon Edition": ["Amazon Edition", "Amazon", "AMZN"],
+    "Netflix Edition": ["Netflix Edition"],
+    "HBO Edition": ["HBO Edition"],
+}
@@ -0,0 +1,74 @@
+"""Frame class and resolution constants.
+
+This module defines video resolution frame classes (480p, 720p, 1080p, 4K, 8K, etc.)
+and their nominal heights and typical widths.
+
+Also includes non-standard quality indicators that appear in filenames but don't
+represent specific resolutions.
+"""
+
+# Non-standard quality indicators that don't have specific resolution values
+# These are used in filenames to indicate quality but aren't proper frame classes
+# When found, we return None instead of trying to classify them
+# Note: We have specific frame classes like "2160p" (4K) and "4320p" (8K),
+# but when files use just "4K" or "8K" without the "p" suffix, we can't determine
+# the exact resolution, so we treat them as non-standard indicators
+NON_STANDARD_QUALITY_INDICATORS = ['SD', 'LQ', 'HD', 'QHD', 'FHD', 'FullHD', '4K', '8K']
+
+FRAME_CLASSES = {
+    "480p": {
+        "nominal_height": 480,
+        "typical_widths": [640, 704, 720],
+        "description": "Standard Definition (SD) - DVD quality",
+    },
+    "480i": {
+        "nominal_height": 480,
+        "typical_widths": [640, 704, 720],
+        "description": "Standard Definition (SD) interlaced - NTSC quality",
+    },
+    "360p": {
+        "nominal_height": 360,
+        "typical_widths": [480, 640],
+        "description": "Low Definition (LD) - 360p",
+    },
+    "576p": {
+        "nominal_height": 576,
+        "typical_widths": [720, 768],
+        "description": "PAL Standard Definition (SD) - European DVD quality",
+    },
+    "576i": {
+        "nominal_height": 576,
+        "typical_widths": [720, 768],
+        "description": "PAL Standard Definition (SD) interlaced - European quality",
+    },
+    "720p": {
+        "nominal_height": 720,
+        "typical_widths": [1280],
+        "description": "High Definition (HD) - 720p HD",
+    },
+    "1080p": {
+        "nominal_height": 1080,
+        "typical_widths": [1920],
+        "description": "Full High Definition (FHD) - 1080p HD",
+    },
+    "1080i": {
+        "nominal_height": 1080,
+        "typical_widths": [1920],
+        "description": "Full High Definition (FHD) interlaced - 1080i HD",
+    },
+    "1440p": {
+        "nominal_height": 1440,
+        "typical_widths": [2560],
+        "description": "Quad High Definition (QHD) - 1440p 2K",
+    },
+    "2160p": {
+        "nominal_height": 2160,
+        "typical_widths": [3840],
+        "description": "Ultra High Definition (UHD) - 2160p 4K",
+    },
+    "4320p": {
+        "nominal_height": 4320,
+        "typical_widths": [7680],
+        "description": "Ultra High Definition (UHD) - 4320p 8K",
+    },
+}
@@ -0,0 +1,31 @@
+"""Language-related constants for filename parsing.
+
+This module contains sets of words and patterns used to identify and skip
+non-language codes when extracting language information from filenames.
+"""
+
+# Words to skip when looking for language codes in filenames
+# These are common words, file extensions, or technical terms that might
+# look like language codes but aren't
+SKIP_WORDS = {
+    # Common English words that might look like language codes (2-3 letters)
+    'the', 'and', 'for', 'are', 'but', 'not', 'you', 'all', 'can', 'had',
+    'her', 'was', 'one', 'our', 'out', 'day', 'get', 'has', 'him', 'his',
+    'how', 'its', 'may', 'new', 'now', 'old', 'see', 'two', 'way', 'who',
+    'boy', 'did', 'let', 'put', 'say', 'she', 'too', 'use',
+
+    # File extensions (video)
+    'avi', 'mkv', 'mp4', 'mpg', 'mov', 'wmv', 'flv', 'webm', 'm4v', 'm2ts',
+    'ts', 'vob', 'iso', 'img',
+
+    # Quality/resolution indicators
+    'sd', 'hd', 'lq', 'qhd', 'uhd', 'p', 'i', 'hdr', 'sdr', '4k', '8k',
+    '2160p', '1080p', '720p', '480p', '360p', '240p', '144p',
+
+    # Source/codec indicators
+    'web', 'dl', 'rip', 'bluray', 'dvd', 'hdtv', 'bdrip', 'dvdrip', 'xvid',
+    'divx', 'h264', 'h265', 'x264', 'x265', 'hevc', 'avc',
+
+    # Audio codecs
+    'ma', 'atmos', 'dts', 'aac', 'ac3', 'mp3', 'flac', 'wav', 'wma', 'ogg', 'opus'
+}
@@ -0,0 +1,46 @@
+"""Media type constants for supported video formats.
+
+This module defines all supported video container formats and their metadata.
+"""
+
+MEDIA_TYPES = {
+    "mkv": {
+        "description": "Matroska multimedia container",
+        "meta_type": "Matroska",
+        "mime": "video/x-matroska",
+    },
+    "mk3d": {
+        "description": "Matroska 3D multimedia container",
+        "meta_type": "Matroska",
+        "mime": "video/x-matroska",
+    },
+    "avi": {
+        "description": "Audio Video Interleave",
+        "meta_type": "AVI",
+        "mime": "video/x-msvideo",
+    },
+    "mov": {
+        "description": "QuickTime movie",
+        "meta_type": "QuickTime",
+        "mime": "video/quicktime",
+    },
+    "mp4": {
+        "description": "MPEG-4 video container",
+        "meta_type": "MP4",
+        "mime": "video/mp4",
+    },
+    "wmv": {
+        "description": "Windows Media Video",
+        "meta_type": "ASF",
+        "mime": "video/x-ms-wmv",
+    },
+    "flv": {"description": "Flash Video", "meta_type": "FLV", "mime": "video/x-flv"},
+    "webm": {
+        "description": "WebM multimedia",
+        "meta_type": "WebM",
+        "mime": "video/webm",
+    },
+    "m4v": {"description": "MPEG-4 video", "meta_type": "MP4", "mime": "video/mp4"},
+    "3gp": {"description": "3GPP multimedia", "meta_type": "MP4", "mime": "video/3gpp"},
+    "ogv": {"description": "Ogg Video", "meta_type": "Ogg", "mime": "video/ogg"},
+}
@@ -0,0 +1,32 @@
+"""Movie database identifier constants.
+
+This module defines movie and TV database services (TMDB, IMDB, Trakt, TVDB)
+and their identifier patterns.
+"""
+
+MOVIE_DB_DICT = {
+    "tmdb": {
+        "name": "The Movie Database (TMDb)",
+        "description": "Community built movie and TV database",
+        "url": "https://www.themoviedb.org/",
+        "patterns": ["tmdbid", "tmdb", "tmdbid-", "tmdb-"],
+    },
+    "imdb": {
+        "name": "Internet Movie Database (IMDb)",
+        "description": "Comprehensive movie, TV, and celebrity database",
+        "url": "https://www.imdb.com/",
+        "patterns": ["imdbid", "imdb", "imdbid-", "imdb-"],
+    },
+    "trakt": {
+        "name": "Trakt.tv",
+        "description": "Service that integrates with media centers for scrobbling",
+        "url": "https://trakt.tv/",
+        "patterns": ["traktid", "trakt", "traktid-", "trakt-"],
+    },
+    "tvdb": {
+        "name": "The TV Database (TVDB)",
+        "description": "Community driven TV database",
+        "url": "https://thetvdb.com/",
+        "patterns": ["tvdbid", "tvdb", "tvdbid-", "tvdb-"],
+    },
+}
@@ -0,0 +1,23 @@
+"""Video source type constants.
+
+This module defines video source types (WEB-DL, BDRip, etc.) and their aliases.
+"""
+
+SOURCE_DICT = {
+    "WEB-DL": ["WEB-DL", "WEBRip", "WEB-Rip", "WEB", "WEB-DLRip"],
+    "BDRip": ["BDRip", "BD-Rip", "BDRIP"],
+    "BDRemux": ["BDRemux", "BD-Remux", "BDREMUX"],
+    "DVDRip": ["DVDRip", "DVD-Rip", "DVDRIP"],
+    "HDTVRip": ["HDTVRip", "HDTV"],
+    "BluRay": ["BluRay", "BLURAY", "Blu-ray"],
+    "SATRip": ["SATRip", "SAT-Rip", "SATRIP"],
+    "VHSRecord": [
+        "VHSRecord",
+        "VHS Record",
+        "VHS-Rip",
+        "VHSRip",
+        "VHS",
+        "VHS Tape",
+        "VHS-Tape",
+    ],
+}
@@ -0,0 +1,20 @@
+"""Year validation constants for filename parsing.
+
+This module contains constants used for validating years extracted from filenames.
+"""
+
+import datetime
+
+# Current year for validation
+CURRENT_YEAR = datetime.datetime.now().year
+
+# Minimum valid year for movies/media (start of cinema era)
+MIN_VALID_YEAR = 1900
+
+# Allow years slightly into the future (for upcoming releases)
+YEAR_FUTURE_BUFFER = 10
+
+# Valid year range: MIN_VALID_YEAR to (CURRENT_YEAR + YEAR_FUTURE_BUFFER)
+def is_valid_year(year: int) -> bool:
+    """Check if a year is within the valid range for media files."""
+    return MIN_VALID_YEAR <= year <= CURRENT_YEAR + YEAR_FUTURE_BUFFER