From 7a95211fe73df167432f5059f63c0ae931842dac Mon Sep 17 00:00:00 2001
From: vvzvlad <git@vvzvlad.xyz>
Date: Fri, 11 Apr 2025 02:50:53 +0300
Subject: [PATCH] Update post parser to exclude t.me links from flag detection

---
 post_parser.py                  |  5 +++--
 tests/test_post_parser_flags.py | 18 +++++++++++++++---
 2 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/post_parser.py b/post_parser.py
index d275e16..9d3f2b0 100644
--- a/post_parser.py
+++ b/post_parser.py
@@ -340,8 +340,9 @@ class PostParser:
         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):
+        # Check if the message contains any http/https links in text or href attributes (excluding t.me links)
+        if (re.search(r'https?://(?!(?:www\.)?t\.me)[^\s<>"\']+', message_text) or 
+            re.search(r'href=["\']https?://(?!(?:www\.)?t\.me)[^"\']+["\']', message_text)):
             flags.append("link")
 
         # Check if the message contains channel mentions in the format @name
diff --git a/tests/test_post_parser_flags.py b/tests/test_post_parser_flags.py
index 363b3c9..ff3eebf 100644
--- a/tests/test_post_parser_flags.py
+++ b/tests/test_post_parser_flags.py
@@ -380,7 +380,7 @@ class TestPostParserExtractFlags(unittest.TestCase):
             text="Join our livestream! Full access on Boosty. Check out https://t.me/other_channel",
         )
         flags = self.parser._extract_flags(message)
-        expected_flags = ["no_image", "stream", "paywall", "link", "foreign_channel"]
+        expected_flags = ["no_image", "stream", "paywall", "foreign_channel"]
         self.assertCountEqual(flags, expected_flags)
 
     def test_flag_multiple_sticker_donat_poo_hid(self):
@@ -391,7 +391,7 @@ class TestPostParserExtractFlags(unittest.TestCase):
         )
         flags = self.parser._extract_flags(message)
         # Sticker implies no 'no_image' flag
-        expected_flags = ["sticker", "donat", "poo", "link", "hid_channel"]
+        expected_flags = ["sticker", "donat", "poo", "hid_channel"]
         self.assertCountEqual(flags, expected_flags)
 
     def test_flag_multiple_text_only_advert_stream_clown_foreign(self):
@@ -401,7 +401,7 @@ class TestPostParserExtractFlags(unittest.TestCase):
             reactions_data=[("🤡", 31)]
         )
         flags = self.parser._extract_flags(message)
-        expected_flags = ["no_image", "advert", "stream", "clown", "link", "foreign_channel"]
+        expected_flags = ["no_image", "advert", "stream", "clown", "foreign_channel"]
         self.assertCountEqual(flags, expected_flags)
 
     def test_flag_multiple_video_note_fwd_paywall_mention(self):
@@ -414,6 +414,18 @@ class TestPostParserExtractFlags(unittest.TestCase):
         expected_flags = ["video", "fwd", "paywall", "mention"]
         self.assertCountEqual(flags, expected_flags)
 
+    def test_flag_link_with_tme_excluded(self):
+        message = self._create_mock_message(text="Check out this Telegram link https://t.me/channel_name")
+        # Use _generate_html_body to ensure link detection logic runs
+        _ = self.parser._generate_html_body(message)
+        self.assertNotIn("link", self.parser._extract_flags(message))
+    
+    def test_flag_link_with_mixed_links(self):
+        message = self._create_mock_message(text="Check both links: https://example.com and https://t.me/channel_name")
+        # Use _generate_html_body to ensure link detection logic runs
+        _ = self.parser._generate_html_body(message)
+        self.assertIn("link", self.parser._extract_flags(message))
+
 
 if __name__ == '__main__':
     unittest.main() 
\ No newline at end of file