From 66025c6dd3867af218bdd38037fc3a3cde6b9551 Mon Sep 17 00:00:00 2001
From: Roger Gonzalez <roger@rogs.me>
Date: Sat, 9 Mar 2024 11:04:39 -0300
Subject: Precompiled ads list

---
 src/subscleaner/subscleaner.py | 80 ++++++++++++++++++++----------------------
 1 file changed, 39 insertions(+), 41 deletions(-)

(limited to 'src')

diff --git a/src/subscleaner/subscleaner.py b/src/subscleaner/subscleaner.py
index 0896d6c..504d43f 100755
--- a/src/subscleaner/subscleaner.py
+++ b/src/subscleaner/subscleaner.py
@@ -27,46 +27,44 @@ import chardet
 import pysrt
 
 ADS = [
-    r".*nordvpn.*",
-    r".*a Card Shark AMERICASCARDROOM.*",
-    r".*OpenSubtitles.*",
-    r".*Advertise your product or brand here.*",
-    r".*Apóyanos y conviértete en miembro VIP Para.*",
-    r".*Addic7ed.*",
-    r".*argenteam.*",
-    r".*AllSubs.*",
-    r"Created and Encoded by.*",
-    r".*corrected.*by.*",
-    r".*Entre a AmericasCardroom. com Hoy.*",
-    r".*Everyone is intimidated by a shark. Become.*",
-    r".*Juegue Poker en Línea por Dinero Real.*",
-    r".*OpenSubtitles.*",
-    r".*Open Subtitles.*",
-    r".*MKV Player.*",
-    r".*MKV player.*",
-    r".*Resync.*for.*",
-    r".*Resync.*improved.*",
-    r".*Ripped?By.*",
-    r'.*Sigue "Community" en.*',
-    r".*Subtitles.*by.*",
-    r".*Subt?tulos.*por.*",
-    r".*Support us and become VIP member.*",
-    r".*Subs.*Team.*",
-    r".*subscene.*",
-    r".*Subtitulado por.*",
-    r".*subtitulamos.*",
-    r".*Synchronized.*by.*",
-    r".*Sincronizado y corregido por.*",
-    r".*subdivx.*",
-    r".*Sync.*Corrected.*",
-    r".*Sync.*corrections.*by.*",
-    r".*sync and corrections by.*" r".*Sync.*by.*",
-    r".*Una.*traducci?n.*de.*",
-    r".*tvsubtitles.*",
-    r".*Una.*traducci?n.*de.*",
-    "Tacho8",
-    r".*www. com.*",
-    r".*www. es.*",
+    re.compile(r"\bnordvpn\b", re.IGNORECASE),
+    re.compile(r"\ba Card Shark AMERICASCARDROOM\b", re.IGNORECASE),
+    re.compile(r"\bOpenSubtitles\b", re.IGNORECASE),
+    re.compile(r"\bAdvertise your product or brand here\b", re.IGNORECASE),
+    re.compile(r"\bApóyanos y conviértete en miembro VIP Para\b", re.IGNORECASE),
+    re.compile(r"\bAddic7ed\b", re.IGNORECASE),
+    re.compile(r"\bargenteam\b", re.IGNORECASE),
+    re.compile(r"\bAllSubs\b", re.IGNORECASE),
+    re.compile(r"\bCreated and Encoded by\b", re.IGNORECASE),
+    re.compile(r"\bcorrected\s+by\b", re.IGNORECASE),
+    re.compile(r"\bEntre a AmericasCardroom\.com Hoy\b", re.IGNORECASE),
+    re.compile(r"\bEveryone is intimidated by a shark\. Become\b", re.IGNORECASE),
+    re.compile(r"\bJuegue Poker en Línea por Dinero Real\b", re.IGNORECASE),
+    re.compile(r"\bOpen Subtitles\b", re.IGNORECASE),
+    re.compile(r"\bMKV Player\b", re.IGNORECASE),
+    re.compile(r"\bResync\s+for\b", re.IGNORECASE),
+    re.compile(r"\bResync\s+improved\b", re.IGNORECASE),
+    re.compile(r"\bRipped\s+By\b", re.IGNORECASE),
+    re.compile(r'\bSigue "Community" en\b', re.IGNORECASE),
+    re.compile(r"\bSubtitles\s+by\b", re.IGNORECASE),
+    re.compile(r"\bSubt[íi]tulos\s+por\b", re.IGNORECASE),
+    re.compile(r"\bSupport us and become VIP member\b", re.IGNORECASE),
+    re.compile(r"\bSubs\s+Team\b", re.IGNORECASE),
+    re.compile(r"\bsubscene\b", re.IGNORECASE),
+    re.compile(r"\bSubtitulado por\b", re.IGNORECASE),
+    re.compile(r"\bsubtitulamos\b", re.IGNORECASE),
+    re.compile(r"\bSynchronized\s+by\b", re.IGNORECASE),
+    re.compile(r"\bSincronizado y corregido por\b", re.IGNORECASE),
+    re.compile(r"\bsubdivx\b", re.IGNORECASE),
+    re.compile(r"\bSync\s+Corrected\b", re.IGNORECASE),
+    re.compile(r"\bSync\s+corrections\s+by\b", re.IGNORECASE),
+    re.compile(r"\bsync and corrections by\b", re.IGNORECASE),
+    re.compile(r"\bSync\s+by\b", re.IGNORECASE),
+    re.compile(r"\bUna\s+traducci[óo]n\s+de\b", re.IGNORECASE),
+    re.compile(r"\btvsubtitles\b", re.IGNORECASE),
+    re.compile(r"\bTacho8\b", re.IGNORECASE),
+    re.compile(r"\bwww\.\S+\.com\b", re.IGNORECASE),
+    re.compile(r"\bwww\.\S+\.es\b", re.IGNORECASE),
 ]
 
 
@@ -80,7 +78,7 @@ def ads_in_line(line: str) -> bool:
     Returns:
         bool: True if the line contains an ad, False otherwise.
     """
-    return any(re.match(ad, line, re.DOTALL) for ad in ADS)
+    return any(ad.search(line) for ad in ADS)
 
 
 def is_already_processed(filename: str) -> bool:
-- 
cgit v1.2.3