From 78255e51a28409b939d8146a7d96bd137b21bba5 Mon Sep 17 00:00:00 2001
From: vvzvlad <git@vvzvlad.xyz>
Date: Fri, 11 Apr 2025 02:48:42 +0300
Subject: [PATCH] Enhance post parser to flag foreign channels for boost links

---
 post_parser.py                  | 16 ++++++++++++--
 tests/test_post_parser_flags.py | 39 +++++++++++++++++++++++++++++++++
 2 files changed, 53 insertions(+), 2 deletions(-)

diff --git a/post_parser.py b/post_parser.py
index 06603cd..d275e16 100644
--- a/post_parser.py
+++ b/post_parser.py
@@ -353,15 +353,27 @@ class PostParser:
             hidden_links = re.findall(r'https?://(?:www\.)?t\.me/\+([A-Za-z0-9]+)', message_text)
             # Find links without a '+' after t.me/ indicating an open (foreign) channel link.
             open_links = re.findall(r'https?://(?:www\.)?t\.me/(?!\+)([A-Za-z0-9_]+)', message_text)
+            
+            # Find links with pattern t.me/boost/channel_name
+            boost_links = re.findall(r'https?://(?:www\.)?t\.me/boost/([A-Za-z0-9_]+)', message_text)
+            
             if hidden_links:
                 flags.append("hid_channel")
 
             current_channel = self.get_channel_username(message)
-            # Add the flag "foreign_channel" if there's an open link that differs from the current channel.
+            
+            # Check regular open links
             for open_link in open_links:
-                if current_channel is None or open_link.lower() != current_channel.lower():
+                if current_channel is None or (open_link.lower() != current_channel.lower() and open_link.lower() != 'boost'):
                     flags.append("foreign_channel")
                     break
+                    
+            # Check boost links separately - only consider as foreign if the boosted channel is not current channel
+            if not "foreign_channel" in flags:
+                for boost_channel in boost_links:
+                    if current_channel is None or boost_channel.lower() != current_channel.lower():
+                        flags.append("foreign_channel")
+                        break
         except Exception as e:
             logger.error(f"tme_link_extraction_error: message_id {message.id}, error {str(e)}")
 
diff --git a/tests/test_post_parser_flags.py b/tests/test_post_parser_flags.py
index a4efd98..363b3c9 100644
--- a/tests/test_post_parser_flags.py
+++ b/tests/test_post_parser_flags.py
@@ -295,6 +295,45 @@ class TestPostParserExtractFlags(unittest.TestCase):
         self.parser.get_channel_username = MagicMock(return_value=None) # Simulate no current channel found
         self.assertIn("foreign_channel", self.parser._extract_flags(message)) # Should still flag as foreign if current unknown
 
+    def test_flag_foreign_channel_boost_other_channel(self):
+        """Test that boost link to foreign channel is flagged."""
+        message = self._create_mock_message(text="Boost another channel: https://t.me/boost/other_channel", chat_username="test_channel")
+        self.parser.get_channel_username = MagicMock(return_value="test_channel") # Ensure current channel is mocked
+        self.assertIn("foreign_channel", self.parser._extract_flags(message))
+    
+    def test_flag_foreign_channel_boost_same_channel(self):
+        """Test that boost link to own channel is not flagged."""
+        message = self._create_mock_message(text="Boost our channel: https://t.me/boost/test_channel", chat_username="test_channel")
+        self.parser.get_channel_username = MagicMock(return_value="test_channel") # Ensure current channel is mocked
+        self.assertNotIn("foreign_channel", self.parser._extract_flags(message))
+    
+    def test_flag_foreign_channel_boost_link_case_insensitive(self):
+        """Test that boost link to own channel is not flagged (case insensitive)."""
+        message = self._create_mock_message(text="Boost our channel: https://t.me/boost/TEST_CHANNEL", chat_username="test_channel")
+        self.parser.get_channel_username = MagicMock(return_value="test_channel") # Ensure current channel is mocked
+        self.assertNotIn("foreign_channel", self.parser._extract_flags(message))
+    
+    def test_flag_foreign_channel_multiple_links(self):
+        """Test with multiple links - should flag if any is foreign."""
+        message = self._create_mock_message(
+            text="Links: https://t.me/test_channel and https://t.me/other_channel and https://t.me/boost/test_channel", 
+            chat_username="test_channel"
+        )
+        self.parser.get_channel_username = MagicMock(return_value="test_channel") # Ensure current channel is mocked
+        self.assertIn("foreign_channel", self.parser._extract_flags(message))
+    
+    def test_flag_foreign_channel_only_boost_word(self):
+        """Test that the word 'boost' is not flagged."""
+        message = self._create_mock_message(text="Check out https://t.me/boost", chat_username="test_channel")
+        self.parser.get_channel_username = MagicMock(return_value="test_channel") # Ensure current channel is mocked
+        self.assertNotIn("foreign_channel", self.parser._extract_flags(message))
+        
+    def test_flag_foreign_channel_boost_no_channel(self):
+        """Test with boost/ but no channel after - should be flagged as foreign."""
+        message = self._create_mock_message(text="Strange link: https://t.me/boost/", chat_username="test_channel")
+        self.parser.get_channel_username = MagicMock(return_value="test_channel") # Ensure current channel is mocked
+        self.assertNotIn("foreign_channel", self.parser._extract_flags(message))
+
     def test_flag_multiple_flags(self):
         message = self._create_mock_message(
             media=MessageMediaType.VIDEO,