UI - Preview problem fix for extract_text/ignore_text #4138 (#4169)

changedetectionio/processors/text_json_diff/__init__.py

@@ -35,6 +35,50 @@ def _task(watch, update_handler):
     return text_after_filter
 
 
+def _compute_ignore_line_numbers_for_preview(text_pre_extract, ignore_patterns, extract_patterns):
+    """1-indexed output line numbers in the post-extract display that correspond
+    to input lines matching ignore_text patterns.
+
+    Needed because extract_text (#4138) transforms line content — e.g. "0.54.10"
+    becomes ".54.10" — so a substring match for "0.54.10" against the post-extract
+    text fails and the preview UI can no longer mark the line as ignored. We find
+    the ignored line numbers in the pre-extract text and replay extract_by_regex
+    line-by-line to map them forward.
+    """
+    from changedetectionio import html_tools
+    from changedetectionio.processors.text_json_diff.processor import ContentTransformer
+
+    if not text_pre_extract or not ignore_patterns:
+        return []
+
+    ignored_input_lines = set(
+        html_tools.strip_ignore_text(
+            content=text_pre_extract,
+            wordlist=ignore_patterns,
+            mode='line numbers'
+        )
+    )
+    if not ignored_input_lines:
+        return []
+
+    if not extract_patterns:
+        return sorted(ignored_input_lines)
+
+    # Replay extract_by_regex per-line. Each emitted match ends with exactly one
+    # '\n', so counting newlines tells us how many output lines this input produced.
+    output_line_counter = 0
+    result = []
+    for input_idx, line in enumerate(text_pre_extract.splitlines()):
+        is_ignored = (input_idx + 1) in ignored_input_lines
+        matches_in_line = ContentTransformer.extract_by_regex(line, extract_patterns).count('\n')
+        for _ in range(matches_in_line):
+            output_line_counter += 1
+            if is_ignored:
+                result.append(output_line_counter)
+
+    return result
+
+
 def prepare_filter_prevew(datastore, watch_uuid, form_data):
     '''Used by @app.route("/edit/<uuid_str:uuid>/preview-rendered", methods=['POST'])'''
     from changedetectionio import forms, html_tools
@@ -50,6 +94,7 @@ def prepare_filter_prevew(datastore, watch_uuid, form_data):
 
     text_after_filter = ''
     text_before_filter = ''
+    text_pre_extract = ''
     trigger_line_numbers = []
     ignore_line_numbers = []
     blocked_line_numbers = []
@@ -89,15 +134,22 @@ def prepare_filter_prevew(datastore, watch_uuid, form_data):
             update_handler.fetcher.content = str(decompressed_data) # str() because playwright/puppeteer/requests return string
             update_handler.fetcher.headers['content-type'] = tmp_watch.get('content-type')
 
-            # Process our watch with filters and the HTML from disk, and also a blank watch with no filters but also with the same HTML from disk
+            # Process our watch with filters and the HTML from disk, and also a blank watch with no filters but also with the same HTML from disk.
+            # The third task runs with extract_text cleared so we can compute ignore_line_numbers
+            # against the pre-extract text (extract_text transforms lines so post-extract substring
+            # matching for ignore patterns would otherwise fail — see #4138 follow-up).
             # Do this as parallel threads (not processes) to avoid pickle issues with Lock objects
+            tmp_watch_no_extract = deepcopy(tmp_watch)
+            tmp_watch_no_extract['extract_text'] = []
             try:
-                with ThreadPoolExecutor(max_workers=2) as executor:
+                with ThreadPoolExecutor(max_workers=3) as executor:
                     future1 = executor.submit(_task, tmp_watch, update_handler)
                     future2 = executor.submit(_task, blank_watch_no_filters, update_handler)
+                    future3 = executor.submit(_task, tmp_watch_no_extract, update_handler)
 
                     text_after_filter = future1.result()
                     text_before_filter = future2.result()
+                    text_pre_extract = future3.result()
             except Exception as e:
                 x=1
 
@@ -111,10 +163,11 @@ def prepare_filter_prevew(datastore, watch_uuid, form_data):
 
     try:
         text_to_ignore = tmp_watch.get('ignore_text', []) + datastore.data['settings']['application'].get('global_ignore_text', [])
-        ignore_line_numbers = html_tools.strip_ignore_text(content=text_after_filter,
-                                                           wordlist=text_to_ignore,
-                                                           mode='line numbers'
-                                                           )
+        ignore_line_numbers = _compute_ignore_line_numbers_for_preview(
+            text_pre_extract=text_pre_extract,
+            ignore_patterns=text_to_ignore,
+            extract_patterns=tmp_watch.get('extract_text', [])
+        )
     except Exception as e:
         text_before_filter = f"Error: {str(e)}"
 

changedetectionio/static/js/watch-settings.js

@@ -9,6 +9,10 @@ function request_textpreview_update() {
     $('textarea:visible, input:visible').each(function () {
         const $element = $(this); // Cache the jQuery object for the current element
         const name = $element.attr('name'); // Get the name attribute of the element
+        // Radios share a name across multiple inputs; .val() returns the value
+        // attribute regardless of checked state, so iterating would let the last
+        // unchecked radio overwrite the user's actual selection. Skip unchecked.
+        if ($element.is(':radio') && !$element.is(':checked')) return;
         data[name] = $element.is(':checkbox') ? ($element.is(':checked') ? $element.val() : false) : $element.val();
     });
 

changedetectionio/tests/test_live_preview.py

@@ -77,3 +77,82 @@ def test_content_filter_live_preview(client, live_server, measure_memory_usage,
     assert reply.get('trigger_line_numbers') == [1]  # Triggers "Awesome" in line 1
 
     delete_all_watches(client)
+
+
+def _setup_version_list_preview(datastore_path, client):
+    """Shared HTML fixture for #4138 preview regressions (version tag list)."""
+    import time
+
+    data = """<html><body>
+0.55.5<br>
+0.55.4<br>
+0.55.3<br>
+0.54.10<br>
+0.54.9<br>
+</body></html>"""
+    with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:
+        f.write(data)
+
+    test_url = url_for('test_endpoint', _external=True)
+    uuid = client.application.config.get('DATASTORE').add_watch(url=test_url)
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    time.sleep(0.5)
+    wait_for_all_checks(client)
+    return test_url, uuid
+
+
+def test_preview_ignore_highlight_with_extract_text(client, live_server, measure_memory_usage, datastore_path):
+    """Regression for #4138 follow-up: when extract_text rewrites a line (e.g. "0.54.10" → ".54.10"),
+    the preview must still highlight that row as 'ignored' even though substring matching against the
+    post-extract text fails."""
+    import json
+
+    test_url, uuid = _setup_version_list_preview(datastore_path, client)
+
+    res = client.post(
+        url_for("ui.ui_edit.watch_get_preview_rendered", uuid=uuid),
+        data={
+            "include_filters": "",
+            "fetch_backend": 'html_requests',
+            "ignore_text": "0.54.10",
+            "extract_text": r"/(.\d+\.\d+)/",
+            "url": test_url,
+        },
+    )
+    reply = json.loads(res.data.decode('utf-8'))
+    # The regex strips the leading "0", so the post-extract line for the ignored input is ".54.10".
+    # The preview should still mark its position (line 4) as ignored.
+    assert reply.get('ignore_line_numbers') == [4], \
+        f"Expected line 4 to be highlighted as ignored, got {reply.get('ignore_line_numbers')!r}"
+
+    delete_all_watches(client)
+
+
+def test_preview_strip_ignored_lines_with_extract_text(client, live_server, measure_memory_usage, datastore_path):
+    """Regression for #4138 follow-up: with strip_ignored_lines enabled, an ignored line must be
+    removed from the preview output even when extract_text would otherwise rewrite it (0.54.10 → .54.10)."""
+    import json
+
+    test_url, uuid = _setup_version_list_preview(datastore_path, client)
+
+    res = client.post(
+        url_for("ui.ui_edit.watch_get_preview_rendered", uuid=uuid),
+        data={
+            "include_filters": "",
+            "fetch_backend": 'html_requests',
+            "ignore_text": "0.54.10",
+            "extract_text": r"/(.\d+\.\d+)/",
+            "strip_ignored_lines": "true",
+            "url": test_url,
+        },
+    )
+    reply = json.loads(res.data.decode('utf-8'))
+    after_filter = reply.get('after_filter', '')
+
+    assert '.54.10' not in after_filter, \
+        f"Stripped ignored line should not appear in preview output, got:\n{after_filter!r}"
+    assert '0.54.10' not in after_filter
+    assert reply.get('ignore_line_numbers') == [], \
+        f"Stripped lines need no highlight, got {reply.get('ignore_line_numbers')!r}"
+
+    delete_all_watches(client)