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

strip file size tags from TMDb search query

Browse Source
This commit is contained in:
Justin Oro
2026-08-21 01:54:58 -07:00
parent 83f3ab72ed
commit 30c4c2e72f
1 changed files with 5 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
plex-library-tool.py
+5
View File
@@ -741,14 +741,19 @@ JUNK_TOKENS = {
"complete", "collection", "series", "complete", "collection", "series",
"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",
} }
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)
IN_FORMAT_PATTERN = re.compile(r'\bin\s+(?:full\s+)?(?:hd|4k|uhd|sd)\b', re.IGNORECASE) IN_FORMAT_PATTERN = re.compile(r'\bin\s+(?:full\s+)?(?:hd|4k|uhd|sd)\b', re.IGNORECASE)
FILE_SIZE_PATTERN = re.compile(r'\b\d+(?:\.\d+)?\s?(?:MB|GB)\b', re.IGNORECASE)
def strip_junk_tokens(text): def strip_junk_tokens(text):
text = re.sub(r'\b\d{3,4}p\b', ' ', text, flags=re.IGNORECASE) text = re.sub(r'\b\d{3,4}p\b', ' ', text, flags=re.IGNORECASE)
text = FILE_SIZE_PATTERN.sub(' ', text)
text = SEASON_RANGE_PATTERN.sub(' ', text) text = SEASON_RANGE_PATTERN.sub(' ', text)
text = IN_FORMAT_PATTERN.sub(' ', text) text = IN_FORMAT_PATTERN.sub(' ', text)
words = text.split() words = text.split()