Refactor extractors and formatters for improved structure and functionality

- Converted static methods to instance methods in FileInfoExtractor and FilenameExtractor for better encapsulation.
- Enhanced MediaInfoExtractor to initialize with file path and extract media information upon instantiation.
- Updated MetadataExtractor to handle metadata extraction with improved error handling and added methods for meta type detection.
- Introduced ColorFormatter for consistent text formatting across the application.
- Refactored MediaFormatter to utilize the new extractor structure and improve output formatting.
- Removed redundant utility functions and replaced them with direct calls in extractors.
- Added ProposedNameFormatter for better handling of proposed filename formatting.
- Updated extension handling to use MEDIA_TYPES for descriptions instead of VIDEO_EXT_DESCRIPTIONS.
This commit is contained in:
sha
2025-12-25 23:35:59 +00:00
parent 37efdf60d3
commit d2ec235458
14 changed files with 664 additions and 500 deletions
+98 -61
View File
@@ -2,70 +2,107 @@ from pathlib import Path
from .extractors.filename_extractor import FilenameExtractor
from .extractors.metadata_extractor import MetadataExtractor
from .extractors.mediainfo_extractor import MediaInfoExtractor
from .extractors.fileinfo_extractor import FileInfoExtractor
class MediaExtractor:
"""Class to extract various metadata from media files using specialized extractors"""
def __init__(self):
self.mediainfo_extractor = MediaInfoExtractor()
def __init__(self, file_path: Path):
self.file_path = file_path
self.filename_extractor = FilenameExtractor(file_path)
self.metadata_extractor = MetadataExtractor(file_path)
self.mediainfo_extractor = MediaInfoExtractor(file_path)
self.fileinfo_extractor = FileInfoExtractor(file_path)
# Define sources for each data type
self._sources = {
'title': [
('metadata', lambda: self.metadata_extractor.extract_title()),
('filename', lambda: self.filename_extractor.extract_title())
],
'year': [
('filename', lambda: self.filename_extractor.extract_year())
],
'source': [
('filename', lambda: self.filename_extractor.extract_source())
],
'frame_class': [
('mediainfo', lambda: self.mediainfo_extractor.extract_frame_class()),
('filename', lambda: self.filename_extractor.extract_frame_class())
],
'resolution': [
('mediainfo', lambda: self.mediainfo_extractor.extract_resolution())
],
'aspect_ratio': [
('mediainfo', lambda: self.mediainfo_extractor.extract_aspect_ratio())
],
'hdr': [
('mediainfo', lambda: self.mediainfo_extractor.extract_hdr())
],
'audio_langs': [
('mediainfo', lambda: self.mediainfo_extractor.extract_audio_langs())
],
'metadata': [
('metadata', lambda: self.metadata_extractor.extract_all_metadata())
],
'meta_type': [
('metadata', lambda: self.metadata_extractor.extract_meta_type())
],
'meta_description': [
('metadata', lambda: self.metadata_extractor.extract_meta_description())
],
'file_size': [
('fileinfo', lambda: self.fileinfo_extractor.extract_size())
],
'modification_time': [
('fileinfo', lambda: self.fileinfo_extractor.extract_modification_time())
],
'file_name': [
('fileinfo', lambda: self.fileinfo_extractor.extract_file_name())
],
'file_path': [
('fileinfo', lambda: self.fileinfo_extractor.extract_file_path())
],
'extension': [
('fileinfo', lambda: self.fileinfo_extractor.extract_extension())
],
'tracks': [
('mediainfo', lambda: self.mediainfo_extractor.extract_tracks())
]
}
# Conditions for when a value is considered valid
self._conditions = {
'title': lambda x: x is not None,
'year': lambda x: x is not None,
'source': lambda x: x is not None,
'frame_class': lambda x: x and x != 'Unclassified',
'resolution': lambda x: x is not None,
'aspect_ratio': lambda x: x is not None,
'hdr': lambda x: x is not None,
'audio_langs': lambda x: x is not None,
'metadata': lambda x: x is not None,
'tracks': lambda x: x != ""
}
def extract_title(self, file_path: Path) -> str | None:
"""Extract movie title from metadata or filename"""
# Try metadata first
title = MetadataExtractor.extract_title(file_path)
if title:
return title
# Fallback to filename
return FilenameExtractor.extract_title(file_path)
def extract_year(self, file_path: Path) -> str | None:
"""Extract year from filename"""
return FilenameExtractor.extract_year(file_path)
def extract_source(self, file_path: Path) -> str | None:
"""Extract video source from filename"""
return FilenameExtractor.extract_source(file_path)
def extract_frame_class(self, file_path: Path) -> str | None:
"""Extract frame class from media info or filename"""
# Try media info first
frame_class = self.mediainfo_extractor.extract_frame_class(file_path)
if frame_class:
return frame_class
# Fallback to filename
return FilenameExtractor.extract_frame_class(file_path)
def extract_resolution(self, file_path: Path) -> str | None:
"""Extract actual video resolution (WIDTHxHEIGHT) from media info"""
return self.mediainfo_extractor.extract_resolution(file_path)
def extract_aspect_ratio(self, file_path: Path) -> str | None:
"""Extract video aspect ratio from media info"""
return self.mediainfo_extractor.extract_aspect_ratio(file_path)
def extract_hdr(self, file_path: Path) -> str | None:
"""Extract HDR info from media info"""
return self.mediainfo_extractor.extract_hdr(file_path)
def extract_audio_langs(self, file_path: Path) -> str:
"""Extract audio languages from media info"""
return self.mediainfo_extractor.extract_audio_langs(file_path)
def extract_metadata(self, file_path: Path) -> dict:
"""Extract general metadata"""
return MetadataExtractor.extract_all_metadata(file_path)
def extract_all(self, file_path: Path) -> dict:
"""Extract all rename-related data"""
return {
'title': self.extract_title(file_path),
'year': self.extract_year(file_path),
'source': self.extract_source(file_path),
'frame_class': self.extract_frame_class(file_path),
'resolution': self.extract_resolution(file_path),
'aspect_ratio': self.extract_aspect_ratio(file_path),
'hdr': self.extract_hdr(file_path),
'audio_langs': self.extract_audio_langs(file_path),
'metadata': self.extract_metadata(file_path)
}
def get(self, key: str, source: str | None = None):
"""Get extracted data by key, optionally from specific source"""
if key not in self._sources:
raise ValueError(f"Unknown key: {key}")
condition = self._conditions.get(key, lambda x: x is not None)
if source:
for src, func in self._sources[key]:
if src == source:
val = func()
return val if condition(val) else None
raise ValueError(f"No such source '{source}' for key '{key}'")
else:
# Use fallback: return first valid value
for src, func in self._sources[key]:
val = func()
if condition(val):
return val
return None