From 30c4c2e72f4b3d876063291950c938a20858d273 Mon Sep 17 00:00:00 2001
From: Justin Oro <JustinOros@gmail.com>
Date: Fri, 21 Aug 2026 01:54:58 -0700
Subject: [PATCH] strip file size tags from TMDb search query

---
 plex-library-tool.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/plex-library-tool.py b/plex-library-tool.py
index 066684a..7fdca3b 100755
--- a/plex-library-tool.py
+++ b/plex-library-tool.py
@@ -741,14 +741,19 @@ JUNK_TOKENS = {
     "complete", "collection", "series",
     "se", "ce", "remastered", "anniversary", "edition",
     "amzn", "nf", "dsnp", "hmax", "atvp", "pcok", "hulu",
+    "mb", "gb",
 }
 
 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)
 
 
+FILE_SIZE_PATTERN = re.compile(r'\b\d+(?:\.\d+)?\s?(?:MB|GB)\b', re.IGNORECASE)
+
+
 def strip_junk_tokens(text):
     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 = IN_FORMAT_PATTERN.sub(' ', text)
     words = text.split()