UI - CSS - Ensure 'difference' 'preview' both wraps by word and by very long strings

Fix: Some SPAs with long content - Stripping tags must also find matching close tag (#3895 )
Fix: Some SPA's also set body content to display: none which breaks text output
2026-02-17 19:56:03 +00:00 · 2026-02-17 17:08:44 +01:00 · 2026-02-17 16:57:29 +01:00 · 2026-02-17 15:38:54 +01:00 · 2026-02-17 14:44:03 +01:00 · 2026-02-17 02:55:18 +01:00
12 changed files with 826 additions and 27 deletions
--- a/changedetectionio/blueprint/ui/init.py
+++ b/changedetectionio/blueprint/ui/init.py
@@ -194,9 +194,9 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, worker_pool,
        tag_limit = request.args.get('tag')
        now = int(time.time())

-        # Mark watches as viewed in background thread to avoid blocking
-        def mark_viewed_background():
-            """Background thread to mark watches as viewed - discarded after completion."""
+        # Mark watches as viewed - use background thread only for large watch counts
+        def mark_viewed_impl():
+            """Mark watches as viewed - can run synchronously or in background thread."""
            marked_count = 0
            try:
                for watch_uuid, watch in datastore.data['watching'].items():
@@ -209,15 +209,21 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, worker_pool,
                    datastore.set_last_viewed(watch_uuid, now)
                    marked_count += 1

-                logger.info(f"Background marking complete: {marked_count} watches marked as viewed")
+                logger.info(f"Marking complete: {marked_count} watches marked as viewed")
            except Exception as e:
-                logger.error(f"Error in background mark as viewed: {e}")
+                logger.error(f"Error marking as viewed: {e}")

-        # Start background thread and return immediately
-        thread = threading.Thread(target=mark_viewed_background, daemon=True)
-        thread.start()
+        # For small watch counts (< 10), run synchronously to avoid race conditions in tests
+        # For larger counts, use background thread to avoid blocking the UI
+        watch_count = len(datastore.data['watching'])
+        if watch_count < 10:
+            # Run synchronously for small watch counts
+            mark_viewed_impl()
+        else:
+            # Start background thread for large watch counts
+            thread = threading.Thread(target=mark_viewed_impl, daemon=True)
+            thread.start()

-        flash(gettext("Marking watches as viewed in background..."))
        return redirect(url_for('watchlist.index', tag=tag_limit))

    @ui_blueprint.route("/delete", methods=['GET'])
--- a/changedetectionio/html_tools.py
+++ b/changedetectionio/html_tools.py
@@ -565,6 +565,27 @@ def html_to_text(html_content: str, render_anchor_tag_content=False, is_rss=Fals
    if is_rss:
        html_content = re.sub(r'<title([\s>])', r'<h1\1', html_content)
        html_content = re.sub(r'</title>', r'</h1>', html_content)
+    else:
+        # Strip bloat in one pass, SPA's often dump 10Mb+ into the <head> for styles, which is not needed
+        # Causing inscriptis to silently exit when more than ~10MB is found.
+        # All we are doing here is converting the HTML to text, no CSS layout etc
+        # Use backreference (\1) to ensure opening/closing tags match (prevents <style> matching </svg> in CSS data URIs)
+        html_content = re.sub(r'<(style|script|svg|noscript)[^>]*>.*?</\1>|<(?:link|meta)[^>]*/?>|<!--.*?-->',
+                              '', html_content, flags=re.DOTALL | re.IGNORECASE)
+
+        # SPAs often use <body style="display:none"> to hide content until JS loads
+        # inscriptis respects CSS display rules, so we need to remove these hiding styles
+        # to extract the actual page content
+        body_style_pattern = r'(<body[^>]*)\s+style\s*=\s*["\']([^"\']*\b(?:display\s*:\s*none|visibility\s*:\s*hidden)\b[^"\']*)["\']'
+
+        # Check if body has hiding styles that need to be fixed
+        body_match = re.search(body_style_pattern, html_content, flags=re.IGNORECASE)
+        if body_match:
+            from loguru import logger
+            logger.debug(f"html_to_text: Removing hiding styles from body tag (found: '{body_match.group(2)}')")
+
+        html_content = re.sub(body_style_pattern, r'\1', html_content, flags=re.IGNORECASE)
+

    text_content = get_text(html_content, config=parser_config)
    return text_content
--- a/changedetectionio/pluggy_interface.py
+++ b/changedetectionio/pluggy_interface.py
@@ -129,6 +129,51 @@ class ChangeDetectionSpec:
        """
        pass

+    @hookspec
+    def update_handler_alter(update_handler, watch, datastore):
+        """Modify or wrap the update_handler before it processes a watch.
+
+        This hook is called after the update_handler (perform_site_check instance) is created
+        but before it calls call_browser() and run_changedetection(). Plugins can use this to:
+        - Wrap the handler to add logging/metrics
+        - Modify handler configuration
+        - Add custom preprocessing logic
+
+        Args:
+            update_handler: The perform_site_check instance that will process the watch
+            watch: The watch dict being processed
+            datastore: The application datastore
+
+        Returns:
+            object or None: Return a modified/wrapped handler, or None to keep the original.
+                           If multiple plugins return handlers, they are chained in registration order.
+        """
+        pass
+
+    @hookspec
+    def update_finalize(update_handler, watch, datastore, processing_exception):
+        """Called after watch processing completes (success or failure).
+
+        This hook is called in the finally block after all processing is complete,
+        allowing plugins to perform cleanup, update metrics, or log final status.
+
+        The plugin can access update_handler.last_logging_insert_id if it was stored
+        during update_handler_alter, and use processing_exception to determine if
+        the processing succeeded or failed.
+
+        Args:
+            update_handler: The perform_site_check instance (may be None if creation failed)
+            watch: The watch dict that was processed (may be None if not loaded)
+            datastore: The application datastore
+            processing_exception: The exception from the main processing block, or None if successful.
+                                 This does NOT include cleanup exceptions - only exceptions from
+                                 the actual watch processing (fetch, diff, etc).
+
+        Returns:
+            None: This hook doesn't return a value
+        """
+        pass
+

 # Set up Plugin Manager
 plugin_manager = pluggy.PluginManager(PLUGIN_NAMESPACE)
@@ -499,4 +544,66 @@ def get_plugin_template_paths():
                template_paths.append(templates_dir)
                logger.debug(f"Added plugin template path: {templates_dir}")

-    return template_paths
+    return template_paths
+
+
+def apply_update_handler_alter(update_handler, watch, datastore):
+    """Apply update_handler_alter hooks from all plugins.
+
+    Allows plugins to wrap or modify the update_handler before it processes a watch.
+    Multiple plugins can chain modifications - each plugin receives the result from
+    the previous plugin.
+
+    Args:
+        update_handler: The perform_site_check instance to potentially modify
+        watch: The watch dict being processed
+        datastore: The application datastore
+
+    Returns:
+        object: The (potentially modified/wrapped) update_handler
+    """
+    # Get all plugins that implement the update_handler_alter hook
+    results = plugin_manager.hook.update_handler_alter(
+        update_handler=update_handler,
+        watch=watch,
+        datastore=datastore
+    )
+
+    # Chain results - each plugin gets the result from the previous one
+    current_handler = update_handler
+    if results:
+        for result in results:
+            if result is not None:
+                logger.debug(f"Plugin modified update_handler for watch {watch.get('uuid')}")
+                current_handler = result
+
+    return current_handler
+
+
+def apply_update_finalize(update_handler, watch, datastore, processing_exception):
+    """Apply update_finalize hooks from all plugins.
+
+    Called in the finally block after watch processing completes, allowing plugins
+    to perform cleanup, update metrics, or log final status.
+
+    Args:
+        update_handler: The perform_site_check instance (may be None)
+        watch: The watch dict that was processed (may be None)
+        datastore: The application datastore
+        processing_exception: The exception from processing, or None if successful
+
+    Returns:
+        None
+    """
+    try:
+        # Call all plugins that implement the update_finalize hook
+        plugin_manager.hook.update_finalize(
+            update_handler=update_handler,
+            watch=watch,
+            datastore=datastore,
+            processing_exception=processing_exception
+        )
+    except Exception as e:
+        # Don't let plugin errors crash the worker
+        logger.error(f"Error in update_finalize hook: {e}")
+        logger.exception(f"update_finalize hook exception details:")
--- a/changedetectionio/processors/text_json_diff/processor.py
+++ b/changedetectionio/processors/text_json_diff/processor.py
@@ -347,6 +347,7 @@ class ContentProcessor:
    def extract_text_from_html(self, html_content, stream_content_type):
        """Convert HTML to plain text."""
        do_anchor = self.datastore.data["settings"]["application"].get("render_anchor_tag_content", False)
+
        return html_tools.html_to_text(
            html_content=html_content,
            render_anchor_tag_content=do_anchor,
--- a/changedetectionio/static/styles/diff.css
+++ b/changedetectionio/static/styles/diff.css
@@ -1 +1 @@
-#diff-form{background:rgba(0,0,0,.05);padding:1em;border-radius:10px;margin-bottom:1em;color:#fff;font-size:.9rem;text-align:center}#diff-form label.from-to-label{width:4rem;text-decoration:none;padding:.5rem}#diff-form label.from-to-label#change-from{color:#b30000;background:#fadad7}#diff-form label.from-to-label#change-to{background:#eaf2c2;color:#406619}#diff-form #diff-style>span{display:inline-block;padding:.3em}#diff-form #diff-style>span label{font-weight:normal}#diff-form *{vertical-align:middle}body.difference-page section.content{padding-top:40px}#diff-ui{background:var(--color-background);padding:1rem;border-radius:5px}@media(min-width: 767px){#diff-ui{min-width:50%}}#diff-ui #text{font-size:11px}#diff-ui pre{white-space:break-spaces}#diff-ui h1{display:inline;font-size:100%}#diff-ui #result{white-space:pre-wrap;word-break:break-word;overflow-wrap:break-word}#diff-ui .source{position:absolute;right:1%;top:.2em}@-moz-document url-prefix(){#diff-ui body{height:99%}}#diff-ui td#diff-col div{text-align:justify;white-space:pre-wrap}#diff-ui .ignored{background-color:#ccc;opacity:.7}#diff-ui .triggered{background-color:#1b98f8}#diff-ui .ignored.triggered{background-color:red}#diff-ui .tab-pane-inner#screenshot{text-align:center}#diff-ui .tab-pane-inner#screenshot img{max-width:99%}#diff-ui .pure-form button.reset-margin{margin:0px}#diff-ui .diff-fieldset{display:flex;align-items:center;gap:4px;flex-wrap:wrap}#diff-ui ul#highlightSnippetActions{list-style-type:none;display:flex;align-items:center;justify-content:center;gap:1.5rem;flex-wrap:wrap;padding:0;margin:0}#diff-ui ul#highlightSnippetActions li{display:flex;flex-direction:column;align-items:center;text-align:center;padding:.5rem;gap:.3rem}#diff-ui ul#highlightSnippetActions li button,#diff-ui ul#highlightSnippetActions li a{white-space:nowrap}#diff-ui ul#highlightSnippetActions span{font-size:.8rem;color:var(--color-text-input-description)}#diff-ui #cell-diff-jump-visualiser{display:flex;flex-direction:row;gap:1px;background:var(--color-background);border-radius:3px;overflow-x:hidden;position:sticky;top:0;z-index:10;padding-top:1rem;padding-bottom:1rem;justify-content:center}#diff-ui #cell-diff-jump-visualiser>div{flex:1;min-width:1px;max-width:10px;height:10px;background:var(--color-background-button-cancel);opacity:.3;border-radius:1px;transition:opacity .2s;position:relative}#diff-ui #cell-diff-jump-visualiser>div.deletion{background:#b30000;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.insertion{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.note{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.mixed{background:linear-gradient(to right, #b30000 50%, #406619 50%);opacity:1}#diff-ui #cell-diff-jump-visualiser>div.current-position::after{content:"";position:absolute;bottom:-6px;left:50%;transform:translateX(-50%);width:0;height:0;border-left:4px solid rgba(0,0,0,0);border-right:4px solid rgba(0,0,0,0);border-bottom:4px solid var(--color-text)}#diff-ui #cell-diff-jump-visualiser>div:hover{opacity:.8;cursor:pointer}#text-diff-heading-area .snapshot-age{padding:4px;margin:.5rem 0;background-color:var(--color-background-snapshot-age);border-radius:3px;font-weight:bold;margin-bottom:4px}#text-diff-heading-area .snapshot-age.error{background-color:var(--color-error-background-snapshot-age);color:var(--color-error-text-snapshot-age)}#text-diff-heading-area .snapshot-age>*{padding-right:1rem}
+#diff-form{background:rgba(0,0,0,.05);padding:1em;border-radius:10px;margin-bottom:1em;color:#fff;font-size:.9rem;text-align:center}#diff-form label.from-to-label{width:4rem;text-decoration:none;padding:.5rem}#diff-form label.from-to-label#change-from{color:#b30000;background:#fadad7}#diff-form label.from-to-label#change-to{background:#eaf2c2;color:#406619}#diff-form #diff-style>span{display:inline-block;padding:.3em}#diff-form #diff-style>span label{font-weight:normal}#diff-form *{vertical-align:middle}body.difference-page section.content{padding-top:40px}#diff-ui{background:var(--color-background);padding:1rem;border-radius:5px}@media(min-width: 767px){#diff-ui{min-width:50%}}#diff-ui #text{font-size:11px}#diff-ui pre{white-space:break-spaces;overflow-wrap:anywhere}#diff-ui h1{display:inline;font-size:100%}#diff-ui #result{white-space:pre-wrap;word-break:break-word;overflow-wrap:break-word}#diff-ui .source{position:absolute;right:1%;top:.2em}@-moz-document url-prefix(){#diff-ui body{height:99%}}#diff-ui td#diff-col div{text-align:justify;white-space:pre-wrap}#diff-ui .ignored{background-color:#ccc;opacity:.7}#diff-ui .triggered{background-color:#1b98f8}#diff-ui .ignored.triggered{background-color:red}#diff-ui .tab-pane-inner#screenshot{text-align:center}#diff-ui .tab-pane-inner#screenshot img{max-width:99%}#diff-ui .pure-form button.reset-margin{margin:0px}#diff-ui .diff-fieldset{display:flex;align-items:center;gap:4px;flex-wrap:wrap}#diff-ui ul#highlightSnippetActions{list-style-type:none;display:flex;align-items:center;justify-content:center;gap:1.5rem;flex-wrap:wrap;padding:0;margin:0}#diff-ui ul#highlightSnippetActions li{display:flex;flex-direction:column;align-items:center;text-align:center;padding:.5rem;gap:.3rem}#diff-ui ul#highlightSnippetActions li button,#diff-ui ul#highlightSnippetActions li a{white-space:nowrap}#diff-ui ul#highlightSnippetActions span{font-size:.8rem;color:var(--color-text-input-description)}#diff-ui #cell-diff-jump-visualiser{display:flex;flex-direction:row;gap:1px;background:var(--color-background);border-radius:3px;overflow-x:hidden;position:sticky;top:0;z-index:10;padding-top:1rem;padding-bottom:1rem;justify-content:center}#diff-ui #cell-diff-jump-visualiser>div{flex:1;min-width:1px;max-width:10px;height:10px;background:var(--color-background-button-cancel);opacity:.3;border-radius:1px;transition:opacity .2s;position:relative}#diff-ui #cell-diff-jump-visualiser>div.deletion{background:#b30000;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.insertion{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.note{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.mixed{background:linear-gradient(to right, #b30000 50%, #406619 50%);opacity:1}#diff-ui #cell-diff-jump-visualiser>div.current-position::after{content:"";position:absolute;bottom:-6px;left:50%;transform:translateX(-50%);width:0;height:0;border-left:4px solid rgba(0,0,0,0);border-right:4px solid rgba(0,0,0,0);border-bottom:4px solid var(--color-text)}#diff-ui #cell-diff-jump-visualiser>div:hover{opacity:.8;cursor:pointer}#text-diff-heading-area .snapshot-age{padding:4px;margin:.5rem 0;background-color:var(--color-background-snapshot-age);border-radius:3px;font-weight:bold;margin-bottom:4px}#text-diff-heading-area .snapshot-age.error{background-color:var(--color-error-background-snapshot-age);color:var(--color-error-text-snapshot-age)}#text-diff-heading-area .snapshot-age>*{padding-right:1rem}
--- a/changedetectionio/static/styles/scss/diff.scss
+++ b/changedetectionio/static/styles/scss/diff.scss
@@ -62,6 +62,7 @@ body.difference-page {

  pre {
    white-space: break-spaces;
+    overflow-wrap: anywhere;
  }


--- a/changedetectionio/static/styles/styles.css
+++ b/changedetectionio/static/styles/styles.css
--- a/changedetectionio/templates/edit/text-options.html
+++ b/changedetectionio/templates/edit/text-options.html
@@ -10,6 +10,7 @@
                        <li>{{ _('Trigger text is processed from the result-text that comes out of any CSS/JSON Filters for this monitor') }}</li>
                        <li>{{ _('Each line is processed separately (think of each line as "OR")') }}</li>
                        <li>{{ _('Note: Wrap in forward slash / to use regex example:') }} <code>/foo\d/</code></li>
+                        <li>{{ _('You can also use')}} <a href="#conditions">{{ _('conditions')}}</a> - {{ _('"Page text" - with Contains, Starts With, Not Contains and many more' ) }} <code>/foo\d/</code></li>
                    </ul>
                        </span>
                    </div>
--- a/changedetectionio/tests/unit/test_html_to_text.py
+++ b/changedetectionio/tests/unit/test_html_to_text.py
@@ -199,6 +199,259 @@ class TestHtmlToText(unittest.TestCase):

        print(f"✓ Basic thread-safety test passed: {len(results)} threads, no errors")

+    def test_large_html_with_bloated_head(self):
+        """
+        Test that html_to_text can handle large HTML documents with massive <head> bloat.
+
+        SPAs often dump 10MB+ of styles, scripts, and other bloat into the <head> section.
+        This can cause inscriptis to silently exit when processing very large documents.
+        The fix strips <style>, <script>, <svg>, <noscript>, <link>, <meta>, and HTML comments
+        before processing, allowing extraction of actual body content.
+        """
+        # Generate massive style block (~5MB)
+        large_style = '<style>' + '.class{color:red;}\n' * 200000 + '</style>\n'
+
+        # Generate massive script block (~5MB)
+        large_script = '<script>' + 'console.log("bloat");\n' * 200000 + '</script>\n'
+
+        # Generate lots of SVG bloat (~3MB)
+        svg_bloat = '<svg><path d="M0,0 L100,100"/></svg>\n' * 50000
+
+        # Generate meta/link tags (~2MB)
+        meta_bloat = '<meta name="description" content="bloat"/>\n' * 50000
+        link_bloat = '<link rel="stylesheet" href="bloat.css"/>\n' * 50000
+
+        # Generate HTML comments (~1MB)
+        comment_bloat = '<!-- This is bloat -->\n' * 50000
+
+        # Generate noscript bloat
+        noscript_bloat = '<noscript>Enable JavaScript</noscript>\n' * 10000
+
+        # Build the large HTML document
+        html = f'''<!DOCTYPE html>
+<html>
+<head>
+    <title>Test Page</title>
+    {large_style}
+    {large_script}
+    {svg_bloat}
+    {meta_bloat}
+    {link_bloat}
+    {comment_bloat}
+    {noscript_bloat}
+</head>
+<body>
+    <h1>Important Heading</h1>
+    <p>This is the actual content that should be extracted.</p>
+    <div>
+        <p>First paragraph with meaningful text.</p>
+        <p>Second paragraph with more content.</p>
+    </div>
+    <footer>Footer text</footer>
+</body>
+</html>
+'''
+
+        # Verify the HTML is actually large (should be ~20MB+)
+        html_size_mb = len(html) / (1024 * 1024)
+        assert html_size_mb > 15, f"HTML should be >15MB, got {html_size_mb:.2f}MB"
+
+        print(f"  Testing {html_size_mb:.2f}MB HTML document with bloated head...")
+
+        # This should not crash or silently exit
+        text = html_to_text(html)
+
+        # Verify we got actual text output (not empty/None)
+        assert text is not None, "html_to_text returned None"
+        assert len(text) > 0, "html_to_text returned empty string"
+
+        # Verify the actual body content was extracted
+        assert 'Important Heading' in text, "Failed to extract heading"
+        assert 'actual content that should be extracted' in text, "Failed to extract paragraph"
+        assert 'First paragraph with meaningful text' in text, "Failed to extract first paragraph"
+        assert 'Second paragraph with more content' in text, "Failed to extract second paragraph"
+        assert 'Footer text' in text, "Failed to extract footer"
+
+        # Verify bloat was stripped (output should be tiny compared to input)
+        text_size_kb = len(text) / 1024
+        assert text_size_kb < 1, f"Output too large ({text_size_kb:.2f}KB), bloat not stripped"
+
+        # Verify no CSS, script content, or SVG leaked through
+        assert 'color:red' not in text, "Style content leaked into text output"
+        assert 'console.log' not in text, "Script content leaked into text output"
+        assert '<path' not in text, "SVG content leaked into text output"
+        assert 'bloat.css' not in text, "Link href leaked into text output"
+
+        print(f"  ✓ Successfully processed {html_size_mb:.2f}MB HTML -> {text_size_kb:.2f}KB text")
+
+    def test_body_display_none_spa_pattern(self):
+        """
+        Test that html_to_text can extract content from pages with display:none body.
+
+        SPAs (Single Page Applications) often use <body style="display:none"> to hide content
+        until JavaScript loads and renders the page. inscriptis respects CSS display rules,
+        so without preprocessing, it would skip all content and return only newlines.
+
+        The fix strips display:none and visibility:hidden styles from the body tag before
+        processing, allowing text extraction from client-side rendered applications.
+        """
+        # Test case 1: Basic display:none
+        html1 = '''<!DOCTYPE html>
+<html lang="en">
+<head><title>What's New – Fluxguard</title></head>
+<body style="display:none">
+    <h1>Important Heading</h1>
+    <p>This is actual content that should be extracted.</p>
+    <div>
+        <p>First paragraph with meaningful text.</p>
+        <p>Second paragraph with more content.</p>
+    </div>
+</body>
+</html>'''
+
+        text1 = html_to_text(html1)
+
+        # Before fix: would return ~33 newlines, len(text) ~= 33
+        # After fix: should extract actual content, len(text) > 100
+        assert len(text1) > 100, f"Expected substantial text output, got {len(text1)} chars"
+        assert 'Important Heading' in text1, "Failed to extract heading from display:none body"
+        assert 'actual content' in text1, "Failed to extract paragraph from display:none body"
+        assert 'First paragraph' in text1, "Failed to extract nested content"
+
+        # Should not be mostly newlines
+        newline_ratio = text1.count('\n') / len(text1)
+        assert newline_ratio < 0.5, f"Output is mostly newlines ({newline_ratio:.2%}), content not extracted"
+
+        # Test case 2: visibility:hidden (another hiding pattern)
+        html2 = '<html><body style="visibility:hidden"><h1>Hidden Content</h1><p>Test paragraph.</p></body></html>'
+        text2 = html_to_text(html2)
+
+        assert 'Hidden Content' in text2, "Failed to extract content from visibility:hidden body"
+        assert 'Test paragraph' in text2, "Failed to extract paragraph from visibility:hidden body"
+
+        # Test case 3: Mixed styles (display:none with other CSS)
+        html3 = '<html><body style="color: red; display:none; font-size: 12px"><p>Mixed style content</p></body></html>'
+        text3 = html_to_text(html3)
+
+        assert 'Mixed style content' in text3, "Failed to extract content from body with mixed styles"
+
+        # Test case 4: Case insensitivity (DISPLAY:NONE uppercase)
+        html4 = '<html><body style="DISPLAY:NONE"><p>Uppercase style</p></body></html>'
+        text4 = html_to_text(html4)
+
+        assert 'Uppercase style' in text4, "Failed to handle uppercase DISPLAY:NONE"
+
+        # Test case 5: Space variations (display: none vs display:none)
+        html5 = '<html><body style="display: none"><p>With spaces</p></body></html>'
+        text5 = html_to_text(html5)
+
+        assert 'With spaces' in text5, "Failed to handle 'display: none' with space"
+
+        # Test case 6: Body with other attributes (class, id)
+        html6 = '<html><body class="foo" style="display:none" id="bar"><p>With attributes</p></body></html>'
+        text6 = html_to_text(html6)
+
+        assert 'With attributes' in text6, "Failed to extract from body with multiple attributes"
+
+        # Test case 7: Should NOT affect opacity:0 (which doesn't hide from inscriptis)
+        html7 = '<html><body style="opacity:0"><p>Transparent content</p></body></html>'
+        text7 = html_to_text(html7)
+
+        # Opacity doesn't affect inscriptis text extraction, content should be there
+        assert 'Transparent content' in text7, "Incorrectly stripped opacity:0 style"
+
+        print("  ✓ All display:none body tag tests passed")
+
+    def test_style_tag_with_svg_data_uri(self):
+        """
+        Test that style tags containing SVG data URIs are properly stripped.
+
+        Some WordPress and modern sites embed SVG as data URIs in CSS, which contains
+        <svg> and </svg> tags within the style content. The regex must use backreferences
+        to ensure <style> matches </style> (not </svg> inside the CSS).
+
+        This was causing errors where the regex would match <style> and stop at the first
+        </svg> it encountered inside a CSS data URI, breaking the HTML structure.
+        """
+        # Real-world example from WordPress wp-block-image styles
+        html = '''<!DOCTYPE html>
+<html>
+<head>
+    <style id='wp-block-image-inline-css'>
+.wp-block-image>a,.wp-block-image>figure>a{display:inline-block}.wp-block-image img{box-sizing:border-box;height:auto;max-width:100%;vertical-align:bottom}@supports ((-webkit-mask-image:none) or (mask-image:none)) or (-webkit-mask-image:none){.wp-block-image.is-style-circle-mask img{border-radius:0;-webkit-mask-image:url('data:image/svg+xml;utf8,<svg viewBox="0 0 100 100" xmlns="http://www.w3.org/2000/svg"><circle cx="50" cy="50" r="50"/></svg>');mask-image:url('data:image/svg+xml;utf8,<svg viewBox="0 0 100 100" xmlns="http://www.w3.org/2000/svg"><circle cx="50" cy="50" r="50"/></svg>');mask-mode:alpha}}
+    </style>
+</head>
+<body>
+    <h1>Test Heading</h1>
+    <p>This is the actual content that should be extracted.</p>
+    <div class="wp-block-image">
+        <img src="test.jpg" alt="Test image">
+    </div>
+</body>
+</html>'''
+
+        # This should not crash and should extract the body content
+        text = html_to_text(html)
+
+        # Verify the actual body content was extracted
+        assert text is not None, "html_to_text returned None"
+        assert len(text) > 0, "html_to_text returned empty string"
+        assert 'Test Heading' in text, "Failed to extract heading"
+        assert 'actual content that should be extracted' in text, "Failed to extract paragraph"
+
+        # Verify CSS content was stripped (including the SVG data URI)
+        assert '.wp-block-image' not in text, "CSS class selector leaked into text"
+        assert 'mask-image' not in text, "CSS property leaked into text"
+        assert 'data:image/svg+xml' not in text, "SVG data URI leaked into text"
+        assert 'viewBox' not in text, "SVG attributes leaked into text"
+
+        # Verify no broken HTML structure
+        assert '<style' not in text, "Unclosed style tag in output"
+        assert '</svg>' not in text, "SVG closing tag leaked into text"
+
+        print("  ✓ Style tag with SVG data URI test passed")
+
+    def test_style_tag_closes_correctly(self):
+        """
+        Test that each tag type (style, script, svg) closes with the correct closing tag.
+
+        Before the fix, the regex used (?:style|script|svg|noscript) for both opening and
+        closing tags, which meant <style> could incorrectly match </svg> as its closing tag.
+        With backreferences, <style> must close with </style>, <svg> with </svg>, etc.
+        """
+        # Test nested tags where incorrect matching would break
+        html = '''<!DOCTYPE html>
+<html>
+<head>
+    <style>
+        body { background: url('data:image/svg+xml,<svg><rect/></svg>'); }
+    </style>
+    <script>
+        const svg = '<svg><path d="M0,0"/></svg>';
+    </script>
+</head>
+<body>
+    <h1>Content</h1>
+    <svg><circle cx="50" cy="50" r="40"/></svg>
+    <p>After SVG</p>
+</body>
+</html>'''
+
+        text = html_to_text(html)
+
+        # Should extract body content
+        assert 'Content' in text, "Failed to extract heading"
+        assert 'After SVG' in text, "Failed to extract content after SVG"
+
+        # Should strip all style/script/svg content
+        assert 'background:' not in text, "Style content leaked"
+        assert 'const svg' not in text, "Script content leaked"
+        assert '<circle' not in text, "SVG element leaked"
+        assert 'data:image/svg+xml' not in text, "Data URI leaked"
+
+        print("  ✓ Tag closing validation test passed")
+
+

 if __name__ == '__main__':
    # Can run this file directly for quick testing
--- a/changedetectionio/tests/unit/test_time_handler.py
+++ b/changedetectionio/tests/unit/test_time_handler.py
@@ -8,6 +8,7 @@ python3 -m pytest changedetectionio/tests/unit/test_time_handler.py -v
 """

 import unittest
+import unittest.mock
 import arrow
 from changedetectionio import time_handler

@@ -240,6 +241,211 @@ class TestAmIInsideTime(unittest.TestCase):
        # Result depends on current time
        self.assertIsInstance(result, bool)

+    def test_24_hour_schedule_from_midnight(self):
+        """Test 24-hour schedule starting at midnight covers entire day."""
+        timezone_str = 'UTC'
+        # Test at a specific time: Monday 00:00
+        test_time = arrow.get('2024-01-01 00:00:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')  # Monday
+
+        # Mock current time for testing
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="00:00",
+                timezone_str=timezone_str,
+                duration=1440  # 24 hours
+            )
+            self.assertTrue(result, "Should be active at start of 24-hour schedule")
+
+    def test_24_hour_schedule_at_end_of_day(self):
+        """Test 24-hour schedule is active at 23:59:59."""
+        timezone_str = 'UTC'
+        # Test at Monday 23:59:59
+        test_time = arrow.get('2024-01-01 23:59:59', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')  # Monday
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="00:00",
+                timezone_str=timezone_str,
+                duration=1440  # 24 hours
+            )
+            self.assertTrue(result, "Should be active at end of 24-hour schedule")
+
+    def test_24_hour_schedule_at_midnight_transition(self):
+        """Test 24-hour schedule at exactly midnight transition."""
+        timezone_str = 'UTC'
+        # Test at Tuesday 00:00:00 (end of Monday's 24-hour schedule)
+        test_time = arrow.get('2024-01-02 00:00:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        monday = test_time.shift(days=-1).format('dddd')  # Monday
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=monday,
+                time_str="00:00",
+                timezone_str=timezone_str,
+                duration=1440  # 24 hours
+            )
+            self.assertTrue(result, "Should include exactly midnight at end of 24-hour schedule")
+
+    def test_schedule_crosses_midnight_before_midnight(self):
+        """Test schedule crossing midnight - before midnight."""
+        timezone_str = 'UTC'
+        # Monday 23:30
+        test_time = arrow.get('2024-01-01 23:30:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')  # Monday
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="23:00",
+                timezone_str=timezone_str,
+                duration=120  # 2 hours (until 01:00 next day)
+            )
+            self.assertTrue(result, "Should be active before midnight in cross-midnight schedule")
+
+    def test_schedule_crosses_midnight_after_midnight(self):
+        """Test schedule crossing midnight - after midnight."""
+        timezone_str = 'UTC'
+        # Tuesday 00:30
+        test_time = arrow.get('2024-01-02 00:30:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        monday = test_time.shift(days=-1).format('dddd')  # Monday
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=monday,
+                time_str="23:00",
+                timezone_str=timezone_str,
+                duration=120  # 2 hours (until 01:00 Tuesday)
+            )
+            self.assertTrue(result, "Should be active after midnight in cross-midnight schedule")
+
+    def test_schedule_crosses_midnight_at_exact_end(self):
+        """Test schedule crossing midnight at exact end time."""
+        timezone_str = 'UTC'
+        # Tuesday 01:00 (exact end of Monday 23:00 + 120 minutes)
+        test_time = arrow.get('2024-01-02 01:00:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        monday = test_time.shift(days=-1).format('dddd')  # Monday
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=monday,
+                time_str="23:00",
+                timezone_str=timezone_str,
+                duration=120  # 2 hours
+            )
+            self.assertTrue(result, "Should include exact end time of schedule")
+
+    def test_duration_60_minutes(self):
+        """Test that duration of 60 minutes works correctly."""
+        timezone_str = 'UTC'
+        test_time = arrow.get('2024-01-01 12:30:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="12:00",
+                timezone_str=timezone_str,
+                duration=60  # Exactly 60 minutes
+            )
+            self.assertTrue(result, "60-minute duration should work")
+
+    def test_duration_at_exact_end_minute(self):
+        """Test at exact end of 60-minute window."""
+        timezone_str = 'UTC'
+        # Exactly 13:00 (end of 12:00 + 60 minutes)
+        test_time = arrow.get('2024-01-01 13:00:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="12:00",
+                timezone_str=timezone_str,
+                duration=60
+            )
+            self.assertTrue(result, "Should include exact end minute")
+
+    def test_one_second_after_schedule_ends(self):
+        """Test one second after schedule should end."""
+        timezone_str = 'UTC'
+        # 13:00:01 (one second after 12:00 + 60 minutes)
+        test_time = arrow.get('2024-01-01 13:00:01', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="12:00",
+                timezone_str=timezone_str,
+                duration=60
+            )
+            self.assertFalse(result, "Should be False one second after schedule ends")
+
+    def test_multi_day_schedule(self):
+        """Test schedule longer than 24 hours (48 hours)."""
+        timezone_str = 'UTC'
+        # Tuesday 12:00 (36 hours after Monday 00:00)
+        test_time = arrow.get('2024-01-02 12:00:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        monday = test_time.shift(days=-1).format('dddd')
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=monday,
+                time_str="00:00",
+                timezone_str=timezone_str,
+                duration=2880  # 48 hours
+            )
+            self.assertTrue(result, "Should support multi-day schedules")
+
+    def test_schedule_one_minute_duration(self):
+        """Test very short 1-minute schedule."""
+        timezone_str = 'UTC'
+        test_time = arrow.get('2024-01-01 12:00:30', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="12:00",
+                timezone_str=timezone_str,
+                duration=1  # Just 1 minute
+            )
+            self.assertTrue(result, "1-minute schedule should work")
+
+    def test_schedule_at_exact_start_time(self):
+        """Test at exact start time (00:00:00.000000)."""
+        timezone_str = 'UTC'
+        test_time = arrow.get('2024-01-01 12:00:00.000000', 'YYYY-MM-DD HH:mm:ss.SSSSSS').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="12:00",
+                timezone_str=timezone_str,
+                duration=30
+            )
+            self.assertTrue(result, "Should include exact start time")
+
+    def test_schedule_one_microsecond_before_start(self):
+        """Test one microsecond before schedule starts."""
+        timezone_str = 'UTC'
+        test_time = arrow.get('2024-01-01 11:59:59.999999', 'YYYY-MM-DD HH:mm:ss.SSSSSS').replace(tzinfo=timezone_str)
+        day_of_week = test_time.format('dddd')
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.am_i_inside_time(
+                day_of_week=day_of_week,
+                time_str="12:00",
+                timezone_str=timezone_str,
+                duration=30
+            )
+            self.assertFalse(result, "Should not include time before start")
+

 class TestIsWithinSchedule(unittest.TestCase):
    """Tests for the is_within_schedule function."""
@@ -405,6 +611,175 @@ class TestIsWithinSchedule(unittest.TestCase):
        result = time_handler.is_within_schedule(time_schedule_limit)
        self.assertTrue(result, "Should handle timezone with whitespace")

+    def test_schedule_with_60_minutes(self):
+        """Test schedule with duration of 0 hours and 60 minutes."""
+        timezone_str = 'UTC'
+        now = arrow.now(timezone_str)
+        current_day = now.format('dddd').lower()
+        current_hour = now.format('HH:00')
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            current_day: {
+                'enabled': True,
+                'start_time': current_hour,
+                'duration': {'hours': 0, 'minutes': 60}  # 60 minutes
+            }
+        }
+
+        result = time_handler.is_within_schedule(time_schedule_limit)
+        self.assertTrue(result, "Should accept 60 minutes as valid duration")
+
+    def test_schedule_with_24_hours(self):
+        """Test schedule with duration of 24 hours and 0 minutes."""
+        timezone_str = 'UTC'
+        now = arrow.now(timezone_str)
+        current_day = now.format('dddd').lower()
+        start_hour = now.format('HH:00')
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            current_day: {
+                'enabled': True,
+                'start_time': start_hour,
+                'duration': {'hours': 24, 'minutes': 0}  # Full 24 hours
+            }
+        }
+
+        result = time_handler.is_within_schedule(time_schedule_limit)
+        self.assertTrue(result, "Should accept 24 hours as valid duration")
+
+    def test_schedule_with_90_minutes(self):
+        """Test schedule with duration of 0 hours and 90 minutes."""
+        timezone_str = 'UTC'
+        now = arrow.now(timezone_str)
+        current_day = now.format('dddd').lower()
+        current_hour = now.format('HH:00')
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            current_day: {
+                'enabled': True,
+                'start_time': current_hour,
+                'duration': {'hours': 0, 'minutes': 90}  # 90 minutes = 1.5 hours
+            }
+        }
+
+        result = time_handler.is_within_schedule(time_schedule_limit)
+        self.assertTrue(result, "Should accept 90 minutes as valid duration")
+
+    def test_schedule_24_hours_from_midnight(self):
+        """Test 24-hour schedule from midnight using is_within_schedule."""
+        timezone_str = 'UTC'
+        test_time = arrow.get('2024-01-01 12:00:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        current_day = test_time.format('dddd').lower()  # monday
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            current_day: {
+                'enabled': True,
+                'start_time': '00:00',
+                'duration': {'hours': 24, 'minutes': 0}
+            }
+        }
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.is_within_schedule(time_schedule_limit)
+            self.assertTrue(result, "24-hour schedule from midnight should cover entire day")
+
+    def test_schedule_24_hours_at_end_of_day(self):
+        """Test 24-hour schedule at 23:59 using is_within_schedule."""
+        timezone_str = 'UTC'
+        test_time = arrow.get('2024-01-01 23:59:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        current_day = test_time.format('dddd').lower()
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            current_day: {
+                'enabled': True,
+                'start_time': '00:00',
+                'duration': {'hours': 24, 'minutes': 0}
+            }
+        }
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.is_within_schedule(time_schedule_limit)
+            self.assertTrue(result, "Should be active at 23:59 in 24-hour schedule")
+
+    def test_schedule_crosses_midnight_with_is_within_schedule(self):
+        """Test schedule crossing midnight using is_within_schedule."""
+        timezone_str = 'UTC'
+        # Tuesday 00:30
+        test_time = arrow.get('2024-01-02 00:30:00', 'YYYY-MM-DD HH:mm:ss').replace(tzinfo=timezone_str)
+        # Get Monday as that's when the schedule started
+        monday = test_time.shift(days=-1).format('dddd').lower()
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            'monday': {
+                'enabled': True,
+                'start_time': '23:00',
+                'duration': {'hours': 2, 'minutes': 0}  # Until 01:00 Tuesday
+            },
+            'tuesday': {
+                'enabled': False,
+                'start_time': '09:00',
+                'duration': {'hours': 8, 'minutes': 0}
+            }
+        }
+
+        with unittest.mock.patch('arrow.now', return_value=test_time):
+            result = time_handler.is_within_schedule(time_schedule_limit)
+            # Note: This checks Tuesday's schedule, not Monday's overlap
+            # So it should be False because Tuesday is disabled
+            self.assertFalse(result, "Should check current day (Tuesday), which is disabled")
+
+    def test_schedule_with_mixed_hours_minutes(self):
+        """Test schedule with both hours and minutes (23 hours 60 minutes = 24 hours)."""
+        timezone_str = 'UTC'
+        now = arrow.now(timezone_str)
+        current_day = now.format('dddd').lower()
+        current_hour = now.format('HH:00')
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            current_day: {
+                'enabled': True,
+                'start_time': current_hour,
+                'duration': {'hours': 23, 'minutes': 60}  # = 1440 minutes = 24 hours
+            }
+        }
+
+        result = time_handler.is_within_schedule(time_schedule_limit)
+        self.assertTrue(result, "Should handle 23 hours + 60 minutes = 24 hours")
+
+    def test_schedule_48_hours(self):
+        """Test schedule with 48-hour duration."""
+        timezone_str = 'UTC'
+        now = arrow.now(timezone_str)
+        current_day = now.format('dddd').lower()
+        start_hour = now.format('HH:00')
+
+        time_schedule_limit = {
+            'enabled': True,
+            'timezone': timezone_str,
+            current_day: {
+                'enabled': True,
+                'start_time': start_hour,
+                'duration': {'hours': 48, 'minutes': 0}  # 2 full days
+            }
+        }
+
+        result = time_handler.is_within_schedule(time_schedule_limit)
+        self.assertTrue(result, "Should support 48-hour (multi-day) schedules")
+

 class TestWeekdayEnum(unittest.TestCase):
    """Tests for the Weekday enum."""
--- a/changedetectionio/time_handler.py
+++ b/changedetectionio/time_handler.py
@@ -62,19 +62,19 @@ def am_i_inside_time(
        # Calculate start and end times for the overlap from the previous day
        start_datetime_tz = start_datetime_tz.shift(days=-1)
        end_datetime_tz = start_datetime_tz.shift(minutes=duration)
-        if start_datetime_tz <= now_tz < end_datetime_tz:
+        if start_datetime_tz <= now_tz <= end_datetime_tz:
            return True

    # Handle current day's range
    if target_weekday == current_weekday:
        end_datetime_tz = start_datetime_tz.shift(minutes=duration)
-        if start_datetime_tz <= now_tz < end_datetime_tz:
+        if start_datetime_tz <= now_tz <= end_datetime_tz:
            return True

    # Handle next day's overlap
    if target_weekday == (current_weekday + 1) % 7:
        end_datetime_tz = start_datetime_tz.shift(minutes=duration)
-        if now_tz < start_datetime_tz and now_tz.shift(days=1) < end_datetime_tz:
+        if now_tz < start_datetime_tz and now_tz.shift(days=1) <= end_datetime_tz:
            return True

    return False
--- a/changedetectionio/worker.py
+++ b/changedetectionio/worker.py
@@ -4,11 +4,10 @@ import changedetectionio.content_fetchers.exceptions as content_fetchers_excepti
 from changedetectionio.processors.text_json_diff.processor import FilterNotFoundInResponse
 from changedetectionio import html_tools
 from changedetectionio import worker_pool
-from changedetectionio.flask_app import watch_check_update
 from changedetectionio.queuedWatchMetaData import PrioritizedItem
+from changedetectionio.pluggy_interface import apply_update_handler_alter, apply_update_finalize

 import asyncio
-import importlib
 import os
 import sys
 import time
@@ -56,6 +55,7 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
    while not app.config.exit.is_set():
        update_handler = None
        watch = None
+        processing_exception = None  # Reset at start of each iteration to prevent state bleeding

        try:
            # Efficient blocking via run_in_executor (no polling overhead!)
@@ -119,7 +119,7 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
        # to prevent race condition with wait_for_all_checks()

        fetch_start_time = round(time.time())
-        
+
        try:
            if uuid in list(datastore.data['watching'].keys()) and datastore.data['watching'][uuid].get('url'):
                changed_detected = False
@@ -136,6 +136,8 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
                logger.info(f"Worker {worker_id} processing watch UUID {uuid} Priority {queued_item_data.priority} URL {watch['url']}")

                try:
+                    # Retrieve signal by name to ensure thread-safe access across worker threads
+                    watch_check_update = signal('watch_check_update')
                    watch_check_update.send(watch_uuid=uuid)

                    # Processor is what we are using for detecting the "Change"
@@ -154,6 +156,9 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
                    update_handler = processor_module.perform_site_check(datastore=datastore,
                                                                         watch_uuid=uuid)

+                    # Allow plugins to modify/wrap the update_handler
+                    update_handler = apply_update_handler_alter(update_handler, watch, datastore)
+
                    update_signal = signal('watch_small_status_comment')
                    update_signal.send(watch_uuid=uuid, status="Fetching page..")

@@ -473,8 +478,6 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
                            logger.exception(f"Worker {worker_id} full exception details:")
                            logger.warning(f"UUID: {uuid} Exception when extracting <title> - {str(e)}")

-
-
                    # Store favicon if necessary
                    if update_handler.fetcher.favicon_blob and update_handler.fetcher.favicon_blob.get('base64'):
                        watch.bump_favicon(url=update_handler.fetcher.favicon_blob.get('url'),
@@ -498,6 +501,8 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
                gc.collect()

        except Exception as e:
+            # Store the processing exception for plugin finalization hook
+            processing_exception = e

            logger.error(f"Worker {worker_id} unexpected error processing {uuid}: {e}")
            logger.exception(f"Worker {worker_id} full exception details:")
@@ -509,6 +514,11 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
        finally:
            # Always cleanup - this runs whether there was an exception or not
            if uuid:
+                # Capture references for plugin finalize hook BEFORE cleanup
+                # (cleanup may delete these variables, but plugins need the original references)
+                finalize_handler = update_handler  # Capture now, before cleanup deletes it
+                finalize_watch = watch              # Capture now, before any modifications
+
                # Call quit() as backup (Puppeteer/Playwright have internal cleanup, but this acts as safety net)
                try:
                    if update_handler and hasattr(update_handler, 'fetcher') and update_handler.fetcher:
@@ -518,12 +528,6 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
                    logger.exception(f"Worker {worker_id} full exception details:")

                try:
-                    # Release UUID from processing (thread-safe)
-                    worker_pool.release_uuid_from_processing(uuid, worker_id=worker_id)
-
-                    # Send completion signal
-                    if watch:
-                        watch_check_update.send(watch_uuid=watch['uuid'])

                    # Clean up all memory references BEFORE garbage collection
                    if update_handler:
@@ -547,7 +551,37 @@ async def async_update_worker(worker_id, q, notification_q, app, datastore, exec
                    logger.error(f"Worker {worker_id} error during cleanup: {cleanup_error}")
                    logger.exception(f"Worker {worker_id} full exception details:")

-            del(uuid)
+                # Call plugin finalization hook after all cleanup is done
+                # Use captured references from before cleanup
+                try:
+                    apply_update_finalize(
+                        update_handler=finalize_handler,
+                        watch=finalize_watch,
+                        datastore=datastore,
+                        processing_exception=processing_exception
+                    )
+                except Exception as finalize_error:
+                    logger.error(f"Worker {worker_id} error in finalize hook: {finalize_error}")
+                    logger.exception(f"Worker {worker_id} full exception details:")
+                finally:
+                    # Clean up captured references to allow immediate garbage collection
+                    del finalize_handler
+                    del finalize_watch
+
+                # Release UUID from processing AFTER all cleanup and hooks complete (thread-safe)
+                # This ensures wait_for_all_checks() waits for finalize hooks to complete
+                try:
+                    worker_pool.release_uuid_from_processing(uuid, worker_id=worker_id)
+                except Exception as release_error:
+                    logger.error(f"Worker {worker_id} error releasing UUID: {release_error}")
+                    logger.exception(f"Worker {worker_id} full exception details:")
+                finally:
+                    # Send completion signal - retrieve by name to ensure thread-safe access
+                    if watch:
+                        watch_check_update = signal('watch_check_update')
+                        watch_check_update.send(watch_uuid=watch['uuid'])
+
+            del (uuid)

            # Brief pause before continuing to avoid tight error loops (only on error)
            if 'e' in locals():
Author	SHA1	Message	Date
dgtlmoon	a6e55aaba9	UI - CSS - Ensure 'difference' 'preview' both wraps by word and by very long strings Some checks are pending Build and push containers / metadata (push) Waiting to run Details Build and push containers / build-push-containers (push) Waiting to run Details Publish Python 🐍distribution 📦 to PyPI and TestPyPI / Build distribution 📦 (push) Waiting to run Details Publish Python 🐍distribution 📦 to PyPI and TestPyPI / Test the built package works basically. (push) Blocked by required conditions Details Publish Python 🐍distribution 📦 to PyPI and TestPyPI / Publish Python 🐍 distribution 📦 to PyPI (push) Blocked by required conditions Details ChangeDetection.io App Test / lint-code (push) Waiting to run Details ChangeDetection.io App Test / test-application-3-10 (push) Blocked by required conditions Details ChangeDetection.io App Test / test-application-3-11 (push) Blocked by required conditions Details ChangeDetection.io App Test / test-application-3-12 (push) Blocked by required conditions Details ChangeDetection.io App Test / test-application-3-13 (push) Blocked by required conditions Details	2026-02-17 17:08:44 +01:00
dgtlmoon	25a17bd49d	Fix: Some SPAs with long content - Stripping tags must also find matching close tag (#3895 )	2026-02-17 16:57:29 +01:00
dgtlmoon	954582a581	Fix: Some SPA's also set body content to display: none which breaks text output	2026-02-17 15:38:54 +01:00
dgtlmoon	d8ef86a8b5	"Error 200 no content" - Some very large SPA pages make HTML to Text fail by dumping 10Mb+ into page header, strip extras. (#3892 )	2026-02-17 14:44:03 +01:00
dgtlmoon	8711d29861	UI - Filters & Triggers - Adding reminder that you can also use 'Conditions' for trigger rules Some checks failed ChangeDetection.io Container Build Test / Build linux/amd64 (alpine) (push) Waiting to run Details ChangeDetection.io Container Build Test / Build linux/arm64 (alpine) (push) Waiting to run Details ChangeDetection.io Container Build Test / Build linux/amd64 (main) (push) Waiting to run Details ChangeDetection.io Container Build Test / Build linux/arm/v7 (main) (push) Waiting to run Details ChangeDetection.io Container Build Test / Build linux/arm/v8 (main) (push) Waiting to run Details ChangeDetection.io Container Build Test / Build linux/arm64 (main) (push) Waiting to run Details Build and push containers / metadata (push) Has been cancelled Details Build and push containers / build-push-containers (push) Has been cancelled Details Publish Python 🐍distribution 📦 to PyPI and TestPyPI / Build distribution 📦 (push) Has been cancelled Details Publish Python 🐍distribution 📦 to PyPI and TestPyPI / Test the built package works basically. (push) Has been cancelled Details Publish Python 🐍distribution 📦 to PyPI and TestPyPI / Publish Python 🐍 distribution 📦 to PyPI (push) Has been cancelled Details ChangeDetection.io App Test / lint-code (push) Has been cancelled Details ChangeDetection.io App Test / test-application-3-10 (push) Has been cancelled Details ChangeDetection.io App Test / test-application-3-11 (push) Has been cancelled Details ChangeDetection.io App Test / test-application-3-12 (push) Has been cancelled Details ChangeDetection.io App Test / test-application-3-13 (push) Has been cancelled Details	2026-02-17 02:55:18 +01:00
dgtlmoon	2343ddd88a	Minor code tidy	2026-02-17 02:46:22 +01:00
dgtlmoon	c6d6ef0e0c	Fix time schedule off-by-one bug at exact end times for all durations and add comprehensive edge case tests Re #846 (#3890 )	2026-02-17 02:38:16 +01:00
dgtlmoon	23063ad8a1	UI - More fixes for realtime updates	2026-02-17 02:37:03 +01:00
dgtlmoon	27b8a2d178	UI - Fixing realtime updates for status updates when checking (#3889 )	2026-02-17 02:26:38 +01:00
dgtlmoon	a53f2a784d	Pluggy plugin hook for before and after a watch is processed (#3888 )	2026-02-17 01:58:41 +01:00