dasemu/Transcriptarr
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

renamed function and cleaned up readability

Browse Source
This commit is contained in:
McCloudS
2025-02-03 10:49:34 -07:00
committed by GitHub
parent 4401c72732
commit 574347f5ea
1 changed files with 64 additions and 54 deletions
Download Patch File Download Diff File Expand all files Collapse all files

98
subgen.py
View File

@@ -1056,7 +1056,7 @@ def gen_subtitles_queue(file_path: str, transcription_type: str, force_language:
return return
if have_to_skip(file_path, force_language): if should_skip_file(file_path, force_language):
return return
task = { task = {
@@ -1068,66 +1068,76 @@ def gen_subtitles_queue(file_path: str, transcription_type: str, force_language:
task_queue.put(task) task_queue.put(task)
logging.info(f"task_queue.put(task)({task['path']}, {task['transcribe_or_translate']}, {task['force_language']})") logging.info(f"task_queue.put(task)({task['path']}, {task['transcribe_or_translate']}, {task['force_language']})")
def have_to_skip(file_path: str, transcribe_language: LanguageCode) -> bool: def should_skip_file(file_path: str, transcribe_language: LanguageCode) -> bool:
""" """Determine if subtitle generation should be skipped for a file."""
Determines whether subtitle generation should be skipped for a given file. base_name = os.path.basename(file_path)
file_name, file_ext = os.path.splitext(base_name)
Args: # 1. Check for existing LRC files for audio files
file_path: The path to the file to check for existing subtitles. if isAudioFileExtension(file_ext) and lrc_for_audio_files:
transcribe_language: The language intended for transcription. lrc_path = f"{file_name}.lrc"
if os.path.exists(lrc_path):
Returns: logging.info(f"Skipping {base_name}: LRC file already exists at {lrc_path}")
True if subtitle generation should be skipped; otherwise, False.
"""
file_name, file_extension = os.path.splitext(file_path)
if isAudioFileExtension(file_extension) and lrc_for_audio_files:
lrc_file_path = file_name + '.lrc'
# Check if the file already exists
if os.path.exists(lrc_file_path):
logging.debug(f"The file {lrc_file_path} already exists. Skipping LRC creation.")
return True return True
# 2. Skip if unknown language and configured to skip unknowns
if skip_unknown_language and transcribe_language == LanguageCode.NONE: if skip_unknown_language and transcribe_language == LanguageCode.NONE:
logging.debug(f"{file_path} has unknown language, skipping.") logging.info(f"Skipping {base_name}: Language detection failed (unknown language)")
return True return True
# Check if subtitles in the desired transcription language already exist # 3. Check subtitle existence in target language
if skip_if_to_transcribe_sub_already_exist and has_subtitle_language(file_path, transcribe_language): if skip_if_to_transcribe_sub_already_exist:
logging.debug(f"{file_path} already has subtitles in {transcribe_language}, skipping.") if has_subtitle_language(file_path, transcribe_language):
return True lang_name = transcribe_language.to_name()
lang_code = transcribe_language.to_iso_639_1()
logging.info(f"Skipping {base_name}: Existing {lang_name} ({lang_code}) subtitles found")
return True
# Check if subtitles in the specified internal language(s) should skip processing # 4. Check against internal subtitle skip list
if skipifinternalsublang and has_subtitle_language(file_path, skipifinternalsublang): if skipifinternalsublang and has_subtitle_language(file_path, skipifinternalsublang):
logging.debug(f"{file_path} has internal subtitles matching skip condition, skipping.") lang_name = skipifinternalsublang.to_name()
lang_code = skipifinternalsublang.to_iso_639_1()
logging.info(f"Skipping {base_name}: Internal {lang_name} ({lang_code}) subtitles present")
return True return True
# Check if external subtitles exist for the specified language # 5. Check for external subtitles in configured language
# Probably not use LanguageCode for this, but just check with strings, to be able to skip with custom named languages. if skipifexternalsub and LanguageCode.is_valid_language(namesublang):
if LanguageCode.is_valid_language(namesublang): target_lang = LanguageCode.from_string(namesublang)
if skipifexternalsub and has_subtitle_language(file_path, LanguageCode.from_string(namesublang)): if has_subtitle_language(file_path, target_lang):
logging.debug(f"{file_path} has external subtitles in {namesublang}, skipping.") lang_name = target_lang.to_name()
lang_code = target_lang.to_iso_639_1()
logging.info(f"Skipping {base_name}: External {lang_name} ({lang_code}) subtitles exist")
return True return True
# Skip if any language in the skip list is detected in existing subtitles # 6. Check against global subtitle language skip list
existing_sub_langs = get_subtitle_languages(file_path) existing_sub_langs = get_subtitle_languages(file_path)
if any(lang in skip_lang_codes_list for lang in existing_sub_langs): for lang in existing_sub_langs:
logging.debug(f"Languages in skip list {skip_lang_codes_list} detected in {file_path}, skipping.") if lang in skip_lang_codes_list:
return True lang_name = lang.to_name()
lang_code = lang.to_iso_639_1()
logging.info(f"Skipping {base_name}: Contains skipped subtitle language {lang_name} ({lang_code})")
return True
# 7. Audio language checks
audio_langs = get_audio_languages(file_path) audio_langs = get_audio_languages(file_path)
if preferred_audio_languages in audio_langs:
logging.debug(f"Preferred audio language {preferred_audio_languages} detected in {file_path}.") # 7a. Limit to preferred audio languages
# maybe not skip if subtitle exist in preferred audio language, but not in another preferred audio language if the file has multiple audio tracks matching the preferred audio languages if limit_to_preferred_audio_languages:
else: preferred_names = [lang.to_name() for lang in preferred_audio_languages]
if limit_to_preferred_audio_languages: found_audio = any(lang in preferred_audio_languages for lang in audio_langs)
logging.debug(f"Only non-preferred audio language detected in {file_path}, skipping.") if not found_audio:
return True logging.info(f"Skipping {base_name}: No preferred audio tracks found (looking for {', '.join(preferred_names)})")
if any(lang in skip_if_audio_track_is_in_list for lang in audio_langs):
logging.debug(f"Audio language in skip list {skip_if_audio_track_is_in_list} detected in {file_path}, skipping.")
return True return True
# If none of the conditions matched, do not skip # 7b. Check for audio languages in skip list
for lang in audio_langs:
if lang in skip_if_audio_track_is_in_list:
lang_name = lang.to_name()
lang_code = lang.to_iso_639_1()
logging.info(f"Skipping {base_name}: Contains skipped audio language {lang_name} ({lang_code})")
return True
logging.debug(f"Proceeding with {base_name}: No skip conditions met (Language: {transcribe_language.to_name() if transcribe_language else 'auto-detect'})")
return False return False
def get_subtitle_languages(video_path): def get_subtitle_languages(video_path):