rogs/subscleaner
rogs/subscleaner
1
0
Fork 0
Code Issues 1 Pull Requests Actions Packages Projects Releases 11 Wiki Activity

Precompiled ads list

Browse Source
This commit is contained in:
Roger Gonzalez 2024-03-09 11:04:39 -03:00
parent 9e8394f9ad
commit 66025c6dd3
Signed by: rogs
GPG Key ID: C7ECE9C6C36EC2E6
1 changed files with 39 additions and 41 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

80
src/subscleaner/subscleaner.py
View File

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