From b59992cb22c32a1ca0851240618da85da39d106a Mon Sep 17 00:00:00 2001 From: vvzvlad Date: Fri, 11 Apr 2025 02:01:42 +0300 Subject: [PATCH] Add paywall detection in post parser --- post_parser.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/post_parser.py b/post_parser.py index c23f2b8..b49a59c 100644 --- a/post_parser.py +++ b/post_parser.py @@ -320,6 +320,10 @@ class PostParser: if re.search(r'(?i)(#реклама|#промо|О\s+рекламодателе|партнерский\s+пост|по\s+промокоду|erid|скидка\s+на\s+курс|регистрируйтесь\s+тут)', message_text): flags.append("advert") + # Check for paywall-related words and tags + if re.search(r'(?i)(Дзен\.Премиум|Sponsr|Бусти|Boosty)', message_text): + flags.append("paywall") + # Check if the message contains any http/https links in text or href attributes if re.search(r'https?://[^\s<>"\']+', message_text) or re.search(r'href=["\']https?://[^"\']+["\']', message_text): flags.append("link")