From 1f61446dfb03c4104251213bc3cd0beca2db3cf1 Mon Sep 17 00:00:00 2001 From: vvzvlad Date: Thu, 27 Feb 2025 23:59:49 +0900 Subject: [PATCH] Enhance post parsing with webinar detection and expanded advertisement flags --- post_parser.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/post_parser.py b/post_parser.py index 52bf6e4..09b1ebf 100644 --- a/post_parser.py +++ b/post_parser.py @@ -214,6 +214,10 @@ class PostParser: # (e.g., "стрим", "стримы", etc.) or "livestream" in a case-insensitive manner. if re.search(r'(?i)\b(стрим\w*|livestream)\b', message_text): flags.append("stream") + + # Check if the message text contains variations of the word "вебинар" in a case-insensitive manner. + if re.search(r'(?i)\bвебинар\w*\b', message_text): + flags.append("vebinar") # Check if the message text contains the word "донат" in a case-insensitive manner. if re.search(r'(?i)\bдонат\w*\b', message_text): @@ -234,7 +238,7 @@ class PostParser: break # Check if the message text contains "#реклама", "Партнерский пост" or "по промокоду" in a case-insensitive manner. - if re.search(r'(?i)(#реклама|партнерский\s+пост|по\s+промокоду|erid)', message_text): + if re.search(r'(?i)(#реклама|#промо|О\s+рекламодателе|партнерский\s+пост|по\s+промокоду|erid)', message_text): flags.append("advert") try: