mirror
/
bazarr
mirror of https://github.com/morpheus65535/bazarr
1
0
Fork 0
Code Issues Releases Wiki Activity

Fix false positive zh detection

This commit is contained in:
zx900930 2021-02-19 17:14:31 +01:00
parent 1aee4770ca
commit cee1ce6214
No known key found for this signature in database
GPG Key ID: 662326EBB1A60A73
1 changed files with 15 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
bazarr/list_subtitles.py
View File

@ -58,11 +58,11 @@ def store_subtitles(original_path, reversed_path):
brazilian_portuguese = [".pt-br", ".pob", "pb"] brazilian_portuguese = [".pt-br", ".pob", "pb"]
brazilian_portuguese_forced = [".pt-br.forced", ".pob.forced", "pb.forced"] brazilian_portuguese_forced = [".pt-br.forced", ".pob.forced", "pb.forced"]
simplified_chinese_long = [".simplified"] simplified_chinese_fuzzy = ["hans", u"简", u"双语"]
simplified_chinese_short = [".chs", ".sc", ".zhs", "hans", ".gb", u"简", u"双语"] simplified_chinese = [".chs", ".sc", ".zhs", ".gb",".simplified"]
simplified_chinese_forced = [".chs.forced", ".sc.forced", ".zhs.forced", "hans.forced", ".gb.forced", u"简体中文.forced", u"双语.forced"] simplified_chinese_forced = [".chs.forced", ".sc.forced", ".zhs.forced", "hans.forced", ".gb.forced", u"简体中文.forced", u"双语.forced"]
traditional_chinese_long = [".traditional"] traditional_chinese_fuzzy = ["hant", u"繁", u"雙語"]
traditional_chinese_short = [".cht", ".tc", ".zh-tw", ".zht", "hant", "big5", u"繁", u"雙語"] traditional_chinese = [".cht", ".tc", ".zh-tw", ".zht", ".big5", ".traditional"]
traditional_chinese_forced = [".cht.forced", ".tc.forced", ".zht.forced", "hant.forced", ".big5.forced", u"繁體中文.forced", u"雙語.forced", "zh-tw.forced"] traditional_chinese_forced = [".cht.forced", ".tc.forced", ".zht.forced", "hant.forced", ".big5.forced", u"繁體中文.forced", u"雙語.forced", "zh-tw.forced"]
try: try:
@ -91,7 +91,7 @@ def store_subtitles(original_path, reversed_path):
logging.debug("BAZARR external subtitles detected: " + "pb:forced") logging.debug("BAZARR external subtitles detected: " + "pb:forced")
actual_subtitles.append( actual_subtitles.append(
[str("pb:forced"), path_mappings.path_replace_reverse(subtitle_path)]) [str("pb:forced"), path_mappings.path_replace_reverse(subtitle_path)])
elif any(ext in (str(subtitle_path).lower())[:-7] for ext in simplified_chinese_short) or (str(subtitle_path).lower())[:-11] in simplified_chinese_long: elif str(os.path.splitext(subtitle)[0]).lower().endswith(tuple(simplified_chinese)) or (str(subtitle_path).lower())[:-7] in simplified_chinese_fuzzy:
logging.debug("BAZARR external subtitles detected: " + "zh") logging.debug("BAZARR external subtitles detected: " + "zh")
actual_subtitles.append( actual_subtitles.append(
[str("zh"), path_mappings.path_replace_reverse(subtitle_path)]) [str("zh"), path_mappings.path_replace_reverse(subtitle_path)])
@ -99,7 +99,7 @@ def store_subtitles(original_path, reversed_path):
logging.debug("BAZARR external subtitles detected: " + "zh:forced") logging.debug("BAZARR external subtitles detected: " + "zh:forced")
actual_subtitles.append( actual_subtitles.append(
[str("zh:forced"), path_mappings.path_replace_reverse(subtitle_path)]) [str("zh:forced"), path_mappings.path_replace_reverse(subtitle_path)])
elif any(ext in (str(subtitle_path).lower())[:-7] for ext in traditional_chinese_short) or (str(subtitle_path).lower())[:-12] in traditional_chinese_long: elif str(os.path.splitext(subtitle)[0]).lower().endswith(tuple(traditional_chinese)) or (str(subtitle_path).lower())[:-7] in traditional_chinese_fuzzy:
logging.debug("BAZARR external subtitles detected: " + "zt") logging.debug("BAZARR external subtitles detected: " + "zt")
actual_subtitles.append( actual_subtitles.append(
[str("zt"), path_mappings.path_replace_reverse(subtitle_path)]) [str("zt"), path_mappings.path_replace_reverse(subtitle_path)])
@ -172,11 +172,11 @@ def store_subtitles_movie(original_path, reversed_path):
brazilian_portuguese = [".pt-br", ".pob", "pb"] brazilian_portuguese = [".pt-br", ".pob", "pb"]
brazilian_portuguese_forced = [".pt-br.forced", ".pob.forced", "pb.forced"] brazilian_portuguese_forced = [".pt-br.forced", ".pob.forced", "pb.forced"]
simplified_chinese_long = [".simplified"] simplified_chinese_fuzzy = ["hans", u"简", u"双语"]
simplified_chinese_short = [".chs", ".sc", ".zhs", "hans", ".gb", u"简", u"双语"] simplified_chinese = [".chs", ".sc", ".zhs", ".gb",".simplified"]
simplified_chinese_forced = [".chs.forced", ".sc.forced", ".zhs.forced", "hans.forced", ".gb.forced", u"简体中文.forced", u"双语.forced"] simplified_chinese_forced = [".chs.forced", ".sc.forced", ".zhs.forced", "hans.forced", ".gb.forced", u"简体中文.forced", u"双语.forced"]
traditional_chinese_long = [".traditional"] traditional_chinese_fuzzy = ["hant", u"繁", u"雙語"]
traditional_chinese_short = [".cht", ".tc", ".zh-tw", ".zht", "hant", "big5", u"繁", u"雙語"] traditional_chinese = [".cht", ".tc", ".zh-tw", ".zht", ".big5", ".traditional"]
traditional_chinese_forced = [".cht.forced", ".tc.forced", ".zht.forced", "hant.forced", ".big5.forced", u"繁體中文.forced", u"雙語.forced", "zh-tw.forced"] traditional_chinese_forced = [".cht.forced", ".tc.forced", ".zht.forced", "hant.forced", ".big5.forced", u"繁體中文.forced", u"雙語.forced", "zh-tw.forced"]
try: try:
dest_folder = get_subtitle_destination_folder() or '' dest_folder = get_subtitle_destination_folder() or ''
@ -201,13 +201,13 @@ def store_subtitles_movie(original_path, reversed_path):
elif str(os.path.splitext(subtitle)[0]).lower().endswith(tuple(brazilian_portuguese_forced)): elif str(os.path.splitext(subtitle)[0]).lower().endswith(tuple(brazilian_portuguese_forced)):
logging.debug("BAZARR external subtitles detected: " + "pb:forced") logging.debug("BAZARR external subtitles detected: " + "pb:forced")
actual_subtitles.append([str("pb:forced"), path_mappings.path_replace_reverse_movie(subtitle_path)]) actual_subtitles.append([str("pb:forced"), path_mappings.path_replace_reverse_movie(subtitle_path)])
elif any(ext in (str(subtitle_path).lower())[:-7] for ext in simplified_chinese_short) or (str(subtitle_path).lower())[:-11] in simplified_chinese_long: elif str(os.path.splitext(subtitle)[0]).lower().endswith(tuple(simplified_chinese)) or (str(subtitle_path).lower())[:-7] in simplified_chinese_fuzzy:
logging.debug("BAZARR external subtitles detected: " + "zh") logging.debug("BAZARR external subtitles detected: " + "zh")
actual_subtitles.append([str("zh"), path_mappings.path_replace_reverse_movie(subtitle_path)]) actual_subtitles.append([str("zh"), path_mappings.path_replace_reverse_movie(subtitle_path)])
elif any(ext in (str(os.path.splitext(subtitle)[0]).lower())[-12:] for ext in simplified_chinese_forced): elif any(ext in (str(os.path.splitext(subtitle)[0]).lower())[-12:] for ext in simplified_chinese_forced):
logging.debug("BAZARR external subtitles detected: " + "zh:forced") logging.debug("BAZARR external subtitles detected: " + "zh:forced")
actual_subtitles.append([str("zh:forced"), path_mappings.path_replace_reverse_movie(subtitle_path)]) actual_subtitles.append([str("zh:forced"), path_mappings.path_replace_reverse_movie(subtitle_path)])
elif any(ext in (str(subtitle_path).lower())[:-7] for ext in traditional_chinese_short) or (str(subtitle_path).lower())[:-12] in traditional_chinese_long: elif str(os.path.splitext(subtitle)[0]).lower().endswith(tuple(traditional_chinese)) or (str(subtitle_path).lower())[:-7] in traditional_chinese_fuzzy:
logging.debug("BAZARR external subtitles detected: " + "zt") logging.debug("BAZARR external subtitles detected: " + "zt")
actual_subtitles.append([str("zt"), path_mappings.path_replace_reverse_movie(subtitle_path)]) actual_subtitles.append([str("zt"), path_mappings.path_replace_reverse_movie(subtitle_path)])
elif any(ext in (str(os.path.splitext(subtitle)[0]).lower())[-12:] for ext in traditional_chinese_forced): elif any(ext in (str(os.path.splitext(subtitle)[0]).lower())[-12:] for ext in traditional_chinese_forced):
@ -549,9 +549,9 @@ def guess_external_subtitles(dest_folder, subtitles):
detected_language = guess_language(text) detected_language = guess_language(text)
#add simplified and traditional chinese detection #add simplified and traditional chinese detection
if detected_language == 'zh': if detected_language == 'zh':
traditional_chinese_long = [".traditional"] traditional_chinese_fuzzy = ["hant", u"繁", u"雙語"]
traditional_chinese_short = [".cht", ".tc", ".zh-tw", ".zht", "hant", "big5", u"繁", u"雙語"] traditional_chinese = [".cht", ".tc", ".zh-tw", ".zht", ".big5", ".traditional"]
if any(ext in (str(subtitle_path).lower())[:-7] for ext in traditional_chinese_short) or (str(subtitle_path).lower())[:-12] in traditional_chinese_long: if str(os.path.splitext(subtitle)[0]).lower().endswith(tuple(traditional_chinese)) or (str(subtitle_path).lower())[:-7] in traditional_chinese_fuzzy:
detected_language == 'zt' detected_language == 'zt'
except UnicodeDecodeError: except UnicodeDecodeError:
detector = Detector() detector = Detector()