Merge pull request #15 from LuisAPI/codex/find-and-fix-an-important-bug

Fix channel parsing in fetch_channels_needing_updates

Author: LuisAPI

src/utils.py

@@ -68,21 +68,37 @@ def generate_wiki_content(channel_info, username, join_date, statistics):
 *         Please leave the bottom part alone         *
 *        The channel box is updated regularly        *
 ■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■-->"""
-    infobox_content = f"""{{Infobox channel
+    infobox_content_top = f"""{{{{Infobox channel
 |channel_name    = {channel_info['title']}
 |subscribers     = {statistics.get('subscriberCount', '')}
 |views           = {statistics.get('viewCount', '')}
 |years_active    = {channel_info.get('years_active', '')}
-|as_of           = {{subst:#time:j F Y}}
+|as_of           = {{{{subst:#time:j F Y}}}}
 """
 
     if new_top_marker in boilerplate and new_bottom_marker in boilerplate:
         # Extract content between new markers
-        parts = boilerplate.split(new_bottom_marker)
-        top_part = parts[0].split(new_top_marker)[0] + new_top_marker
+        
+        # Split at the top marker
+        top_split = boilerplate.split(new_top_marker, 1)
+        top = top_split[0]
+        rest = top_split[1] if len(top_split) > 1 else ""
+
+        # Split the rest at the bottom marker
+        middle_split = rest.split(new_bottom_marker, 1)
+        middle = middle_split[0]
+        bottom = middle_split[1] if len(middle_split) > 1 else ""
+
+        # Now you have:
+        # top    -> before new_top_marker
+        # middle -> between new_top_marker and new_bottom_marker
+        # bottom -> after new_bottom_marker
+
+        top_part = infobox_content_top + new_top_marker + rest
+        bottom_part = new_bottom_marker + bottom
 
         # Combine the top part, autogenerated paragraph, and bottom part
-        wiki_content = f"{top_part}\n\n{auto_paragraph}\n\n{parts[1]}"
+        wiki_content = f"{top_part}\n\n{auto_paragraph}\n\n{bottom_part}"
         marker_comment = new_top_marker
 
     elif old_bottom_marker in boilerplate:
@@ -95,8 +111,23 @@ def generate_wiki_content(channel_info, username, join_date, statistics):
         marker_comment = old_bottom_marker
 
     else:
-        # Fallback if no markers are found
-        wiki_content = boilerplate
+        # Split at the top marker
+        top_split = boilerplate.split(new_top_marker, 1)
+        top = top_split[0]
+        rest = top_split[1] if len(top_split) > 1 else ""
+
+        # Split the rest at the bottom marker
+        middle_split = rest.split(new_bottom_marker, 1)
+        middle = middle_split[0]
+        bottom = middle_split[1] if len(middle_split) > 1 else ""
+
+        # Now you have:
+        # top    -> before new_top_marker
+        # middle -> between new_top_marker and new_bottom_marker
+        # bottom -> after new_bottom_marker
+
+        wiki_content = infobox_content_top + new_top_marker + "\n\n" + auto_paragraph + "\n\n" + middle + bottom
+        # No marker comment since this is a new page
         marker_comment = ""
 
     return marker_comment, wiki_content

src/wiki_bot.py

@@ -1,6 +1,7 @@
 # wiki_bot.py
 
 import mwclient
+import re
 
 def connect_to_wiki(wiki_url, wiki_path, username, password):
     """Connects to the MediaWiki site."""
@@ -21,20 +22,14 @@ def check_page_exists(site, page_title):
 
 def fetch_channels_needing_updates(site, page_title):
     """Fetches the list of channels needing updates from the specified wiki page."""
-    # Access the specified page
     page = site.pages[page_title]
     content = page.text()
-    
-    # Parse the content to find channels (assumes a simple list format)
+
     channels = []
-    lines = content.splitlines()
-    for line in lines:
-        if line.startswith('*'):
-            # Extract the channel handle from the line
-            channel_info = line.split('(')
-            if len(channel_info) > 1:
-                # Get the handle without "Handle: " label
-                channel_handle = channel_info[1].replace('Handle: ', '').strip(') ')
-                channels.append(channel_handle)
-    
+    for line in content.splitlines():
+        if line.lstrip().startswith('*'):
+            match = re.search(r'@[^\s)]+', line)
+            if match:
+                channels.append(match.group(0))
+
     return channels

tests/test_wiki_bot.py

@@ -17,45 +17,44 @@
 def mock_site():
     """Fixture to mock mwclient.Site."""
     with patch('mwclient.Site') as MockSite:
-        mock_site_instance = MockSite.return_value
-        yield mock_site_instance
+        yield MockSite
 
 def test_connect_to_wiki(mock_site):
     """Test connecting to the wiki."""
-    mock_site.login.return_value = True  # Mock successful login
+    mock_site.return_value.login.return_value = True  # Mock successful login
 
     # Call the function with test settings
     site = connect_to_wiki(TEST_WIKI_URL, TEST_WIKI_PATH, TEST_USERNAME, TEST_PASSWORD)
 
     # Assertions
     mock_site.assert_called_once_with(TEST_WIKI_URL, path=TEST_WIKI_PATH)
-    mock_site.login.assert_called_once_with(TEST_USERNAME, TEST_PASSWORD)
-    assert site == mock_site
+    mock_site.return_value.login.assert_called_once_with(TEST_USERNAME, TEST_PASSWORD)
+    assert site == mock_site.return_value
 
 def test_update_wiki_page(mock_site):
     """Test updating a wiki page."""
     mock_page = MagicMock()
-    mock_site.pages.__getitem__.return_value = mock_page  # Mock accessing the page
+    mock_site.return_value.pages.__getitem__.return_value = mock_page  # Mock accessing the page
 
     # Call the function
-    update_wiki_page(mock_site, 'Test Page', 'New content', 'Test summary')
+    update_wiki_page(mock_site.return_value, 'Test Page', 'New content', 'Test summary')
 
     # Assertions
-    mock_site.pages.__getitem__.assert_called_once_with('Test Page')
+    mock_site.return_value.pages.__getitem__.assert_called_once_with('Test Page')
     mock_page.save.assert_called_once_with('New content', summary='Test summary')
 
 def test_check_page_exists(mock_site):
     """Test checking if a wiki page exists."""
     mock_page = MagicMock()
     mock_page.exists = True
     mock_page.text.return_value = "Page content"
-    mock_site.pages.__getitem__.return_value = mock_page  # Mock accessing the page
+    mock_site.return_value.pages.__getitem__.return_value = mock_page  # Mock accessing the page
 
     # Call the function
-    exists, content = check_page_exists(mock_site, 'Test Page')
+    exists, content = check_page_exists(mock_site.return_value, 'Test Page')
 
     # Assertions
-    mock_site.pages.__getitem__.assert_called_once_with('Test Page')
+    mock_site.return_value.pages.__getitem__.assert_called_once_with('Test Page')
     assert exists is True
     assert content == "Page content"
 
@@ -66,13 +65,13 @@ def test_fetch_channels_needing_updates(mock_site):
     * Handle: @example_channel
     * Handle: @another_channel
     """
-    mock_site.pages.__getitem__.return_value = mock_page  # Mock accessing the page
+    mock_site.return_value.pages.__getitem__.return_value = mock_page  # Mock accessing the page
 
     # Call the function
-    channels = fetch_channels_needing_updates(TEST_WIKI_URL, 'Channel Update Requests', TEST_USERNAME, TEST_PASSWORD)
+    channels = fetch_channels_needing_updates(mock_site.return_value, 'Channel Update Requests')
 
     # Assertions
-    mock_site.pages.__getitem__.assert_called_once_with('Channel Update Requests')
+    mock_site.return_value.pages.__getitem__.assert_called_once_with('Channel Update Requests')
     assert channels == ['@example_channel', '@another_channel']
 
 @pytest.mark.parametrize("page_text, expected_channels", [
@@ -85,11 +84,11 @@ def test_fetch_channels_needing_updates_various(mock_site, page_text, expected_c
     """Test fetching channels under various content scenarios."""
     mock_page = MagicMock()
     mock_page.text.return_value = page_text
-    mock_site.pages.__getitem__.return_value = mock_page  # Mock accessing the page
+    mock_site.return_value.pages.__getitem__.return_value = mock_page  # Mock accessing the page
 
     # Call the function
-    channels = fetch_channels_needing_updates(mock_site, 'Channel Update Requests')
+    channels = fetch_channels_needing_updates(mock_site.return_value, 'Channel Update Requests')
 
     # Assertions
-    mock_site.pages.__getitem__.assert_called_once_with('Channel Update Requests')
+    mock_site.return_value.pages.__getitem__.assert_called_once_with('Channel Update Requests')
     assert channels == expected_channels