justin/plex-library-tool
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

strip domains/URLs and add cam-release junk tokens to TMDb query

Browse Source
This commit is contained in:
Justin Oro
2026-08-22 16:52:01 -07:00
parent ce3516c76e
commit 11de696807
1 changed files with 22 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
plex-library-tool.py
+22
View File
@@ -742,6 +742,9 @@ JUNK_TOKENS = {
"se", "ce", "remastered", "anniversary", "edition", "se", "ce", "remastered", "anniversary", "edition",
"amzn", "nf", "dsnp", "hmax", "atvp", "pcok", "hulu", "amzn", "nf", "dsnp", "hmax", "atvp", "pcok", "hulu",
"mb", "gb", "mb", "gb",
"subbed", "dubbed", "subs", "multisub",
"telesync", "telecine", "camrip", "hdcam", "hdts", "hdtc",
"r5", "screener", "dvdscr", "workprint",
} }
SEASON_RANGE_PATTERN = re.compile(r'\bSeasons?\b(?:\s+\d{1,2})+', re.IGNORECASE) SEASON_RANGE_PATTERN = re.compile(r'\bSeasons?\b(?:\s+\d{1,2})+', re.IGNORECASE)
@@ -772,11 +775,24 @@ RELEASE_GROUP_SUFFIX_PATTERN = re.compile(
) )
AUDIO_CHANNELS_PATTERN = re.compile(r'(?<![0-9])\d\.\d(?![0-9])') AUDIO_CHANNELS_PATTERN = re.compile(r'(?<![0-9])\d\.\d(?![0-9])')
AUDIO_CODEC_CHANNELS_PATTERN = re.compile(
r'\b(DDP?|EAC3|AC3|TrueHD|Atmos|DTS(?:-HD)?|FLAC|AAC)[.\s]?\d(?:[.\s]\d)?\b',
re.IGNORECASE,
)
VIDEO_CODEC_SPACED_PATTERN = re.compile(r'\b[Hh][.\s]?(26[45])\b')
LEADING_TAG_DASH_CHARS = '-' + chr(8211) + chr(8212)
LEADING_TAG_PATTERN = re.compile(r'^\s*\[[^\[\]]*\](?:\([^()]*\))?\s*[' + re.escape(LEADING_TAG_DASH_CHARS) + r']+\s*')
URL_PATTERN = re.compile(r'https?://\S+', re.IGNORECASE)
WWW_DOMAIN_PATTERN = re.compile(r'\bwww\.[A-Za-z0-9-]+(?:\.[A-Za-z0-9-]+)+\b', re.IGNORECASE)
TRAILING_PAREN_BLOCK_PATTERN = re.compile(r'\([^()]*\)\s*$') TRAILING_PAREN_BLOCK_PATTERN = re.compile(r'\([^()]*\)\s*$')
TRAILING_BRACKET_TAG_PATTERN = re.compile(r'\[[^\[\]]*\]\s*$') TRAILING_BRACKET_TAG_PATTERN = re.compile(r'\[[^\[\]]*\]\s*$')
BARE_YEAR_PATTERN = re.compile(r'(19|20)\d{2}$') BARE_YEAR_PATTERN = re.compile(r'(19|20)\d{2}$')
def strip_leading_watermark(raw_name):
return LEADING_TAG_PATTERN.sub('', raw_name, count=1)
def strip_trailing_bracket_tags(name): def strip_trailing_bracket_tags(name):
stripped = name stripped = name
while True: while True:
@@ -817,6 +833,10 @@ def strip_junk_trailing_paren(raw_name):
def build_query(raw_name, year): def build_query(raw_name, year):
raw_name = strip_leading_watermark(raw_name)
raw_name = URL_PATTERN.sub(' ', raw_name)
raw_name = WWW_DOMAIN_PATTERN.sub(' ', raw_name)
cut_points = [] cut_points = []
match = SEASON_TRUNCATE_PATTERN.search(raw_name) match = SEASON_TRUNCATE_PATTERN.search(raw_name)
if match: if match:
@@ -829,6 +849,8 @@ def build_query(raw_name, year):
if cut > 0: if cut > 0:
raw_name = raw_name[:cut] raw_name = raw_name[:cut]
raw_name = VIDEO_CODEC_SPACED_PATTERN.sub(r'H\1', raw_name)
raw_name = AUDIO_CODEC_CHANNELS_PATTERN.sub(' ', raw_name)
raw_name = AUDIO_CHANNELS_PATTERN.sub(' ', raw_name) raw_name = AUDIO_CHANNELS_PATTERN.sub(' ', raw_name)
raw_name = strip_junk_trailing_paren(raw_name) raw_name = strip_junk_trailing_paren(raw_name)