hostname connect fix for playwright etc

.github/workflows/containers.yml

@@ -66,27 +66,27 @@ jobs:
           echo ${{ github.ref }} > changedetectionio/tag.txt
 
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@v3
+        uses: docker/setup-qemu-action@v4
         with:
           image: tonistiigi/binfmt:latest
           platforms: all
 
       - name: Login to GitHub Container Registry
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           registry: ghcr.io
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Login to Docker Hub Container Registry
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           username: ${{ secrets.DOCKER_HUB_USERNAME }}
           password: ${{ secrets.DOCKER_HUB_ACCESS_TOKEN }}
 
       - name: Set up Docker Buildx
         id: buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@v4
         with:
           install: true
           version: latest
@@ -95,7 +95,7 @@ jobs:
       # master branch -> :dev container tag
       - name: Docker meta :dev
         if: ${{ github.ref == 'refs/heads/master' && github.event_name != 'release' }}
-        uses: docker/metadata-action@v5
+        uses: docker/metadata-action@v6
         id: meta_dev
         with:
           images: |
@@ -103,11 +103,19 @@ jobs:
             ghcr.io/${{ github.repository }}
           tags: |
             type=raw,value=dev
+          labels: |
+            org.opencontainers.image.created=${{ github.event.release.published_at }}
+            org.opencontainers.image.description=Website, webpage change detection, monitoring and notifications.
+            org.opencontainers.image.documentation=https://changedetection.io
+            org.opencontainers.image.revision=${{ github.sha }}
+            org.opencontainers.image.source=https://github.com/dgtlmoon/changedetection.io
+            org.opencontainers.image.title=changedetection.io
+            org.opencontainers.image.url=https://changedetection.io
 
       - name: Build and push :dev
         id: docker_build
         if: ${{ github.ref == 'refs/heads/master' && github.event_name != 'release' }}
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@v7
         with:
           context: ./
           file: ./Dockerfile
@@ -128,10 +136,10 @@ jobs:
           echo "Release tag: ${{ github.event.release.tag_name }}"
           echo "Github ref: ${{ github.ref }}"
           echo "Github ref name: ${{ github.ref_name }}"
-          
+
       - name: Docker meta :tag
         if: github.event_name == 'release' && startsWith(github.event.release.tag_name, '0.')
-        uses: docker/metadata-action@v5
+        uses: docker/metadata-action@v6
         id: meta
         with:
             images: |
@@ -142,11 +150,20 @@ jobs:
                 type=semver,pattern={{major}}.{{minor}},value=${{ github.event.release.tag_name }}
                 type=semver,pattern={{major}},value=${{ github.event.release.tag_name }}
                 type=raw,value=latest
+            labels: |
+              org.opencontainers.image.created=${{ github.event.release.published_at }}
+              org.opencontainers.image.description=Website, webpage change detection, monitoring and notifications.
+              org.opencontainers.image.documentation=https://changedetection.io
+              org.opencontainers.image.revision=${{ github.sha }}
+              org.opencontainers.image.source=https://github.com/dgtlmoon/changedetection.io
+              org.opencontainers.image.title=changedetection.io
+              org.opencontainers.image.url=https://changedetection.io
+              org.opencontainers.image.version=${{ github.event.release.tag_name }}
 
       - name: Build and push :tag
         id: docker_build_tag_release
         if: github.event_name == 'release' && startsWith(github.event.release.tag_name, '0.')
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@v7
         with:
           context: ./
           file: ./Dockerfile

.github/workflows/test-container-build.yml

@@ -60,14 +60,14 @@ jobs:
 
         # Just test that the build works, some libraries won't compile on ARM/rPi etc
         - name: Set up QEMU
-          uses: docker/setup-qemu-action@v3
+          uses: docker/setup-qemu-action@v4
           with:
             image: tonistiigi/binfmt:latest
             platforms: all
 
         - name: Set up Docker Buildx
           id: buildx
-          uses: docker/setup-buildx-action@v3
+          uses: docker/setup-buildx-action@v4
           with:
             install: true
             version: latest
@@ -75,7 +75,7 @@ jobs:
 
         - name: Test that the docker containers can build (${{ matrix.platform }} - ${{ matrix.dockerfile }})
           id: docker_build
-          uses: docker/build-push-action@v6
+          uses: docker/build-push-action@v7
           # https://github.com/docker/build-push-action#customizing
           with:
             context: ./

.github/workflows/test-only.yml

@@ -52,4 +52,13 @@ jobs:
     uses: ./.github/workflows/test-stack-reusable-workflow.yml
     with:
       python-version: '3.13'
-      skip-pypuppeteer: true
+      skip-pypuppeteer: true
+
+
+  test-application-3-14:
+    #if: github.event_name == 'push' && github.ref == 'refs/heads/master'
+    needs: lint-code
+    uses: ./.github/workflows/test-stack-reusable-workflow.yml
+    with:
+      python-version: '3.14'
+      skip-pypuppeteer: false

.github/workflows/test-stack-reusable-workflow.yml

@@ -42,10 +42,10 @@ jobs:
         run: echo "date=$(date +'%Y-%m-%d')" >> $GITHUB_OUTPUT
 
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@v4
 
       - name: Build changedetection.io container for testing under Python ${{ env.PYTHON_VERSION }}
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@v7
         with:
           context: ./
           file: ./Dockerfile
@@ -292,8 +292,8 @@ jobs:
 
       - name: Specific tests in built container for Selenium
         run: |
-          docker run --rm -e "WEBDRIVER_URL=http://selenium:4444/wd/hub" --network changedet-network test-changedetectionio bash -c 'cd changedetectionio;pytest tests/fetchers/test_content.py && pytest tests/test_errorhandling.py'
-
+          docker run --name "changedet" --hostname changedet --rm -e "FLASK_SERVER_NAME=changedet" -e "WEBDRIVER_URL=http://selenium:4444/wd/hub" --network changedet-network test-changedetectionio bash -c 'cd changedetectionio; pytest  --live-server-host=0.0.0.0 --live-server-port=5004 tests/fetchers/test_content.py'
+          docker run --name "changedet" --hostname changedet --rm -e "FLASK_SERVER_NAME=changedet" -e "WEBDRIVER_URL=http://selenium:4444/wd/hub" --network changedet-network test-changedetectionio bash -c 'cd changedetectionio; pytest  --live-server-host=0.0.0.0 --live-server-port=5004 tests/test_errorhandling.py'
 
   # SMTP tests
   smtp-tests:
@@ -706,7 +706,19 @@ jobs:
       - name: Check upgrade works without error
         run: |
           echo "=== Testing upgrade path from 0.49.1 to ${{ github.ref_name }} (${{ github.sha }}) ==="
-
+          sudo apt-get update && sudo apt-get install -y --no-install-recommends \
+              g++ \
+              gcc \
+              libc-dev \
+              libffi-dev \
+              libjpeg-dev \
+              libssl-dev \
+              libxslt-dev \
+              make \
+              patch \
+              pkg-config \
+              zlib1g-dev
+          
           # Checkout old version and create datastore
           git checkout 0.49.1
           python3 -m venv .venv

changedetectionio/__init__.py

@@ -2,7 +2,7 @@
 
 # Read more https://github.com/dgtlmoon/changedetection.io/wiki
 # Semver means never use .01, or 00. Should be .1.
-__version__ = '0.54.3'
+__version__ = '0.54.7'
 
 from changedetectionio.strtobool import strtobool
 from json.decoder import JSONDecodeError
@@ -10,7 +10,6 @@ from json.decoder import JSONDecodeError
 from loguru import logger
 import getopt
 import logging
-import os
 import platform
 import signal
 import threading
@@ -61,8 +60,22 @@ import time
 # ==============================================================================
 
 import multiprocessing
+import os
 import sys
 
+# Limit glibc malloc arena count to prevent RSS growth from concurrent requests.
+# Default: glibc creates up to 8×CPU_cores arenas. Each concurrent thread/connection
+# can trigger a new arena, and freed memory stays mapped in those arenas as RSS forever.
+# With MALLOC_ARENA_MAX=2, at most 2 arenas are used; freed pages return to the OS faster.
+# Must be set before worker threads start; env var is read lazily by glibc on first arena creation.
+if 'MALLOC_ARENA_MAX' not in os.environ:
+    os.environ['MALLOC_ARENA_MAX'] = '2'
+    try:
+        import ctypes as _ctypes
+        _ctypes.CDLL('libc.so.6').mallopt(-8, 2)  # M_ARENA_MAX = -8
+    except Exception:
+        pass
+
 # Set spawn as global default (safety net - all our code uses explicit contexts anyway)
 # Skip in tests to avoid breaking pytest-flask's LiveServer fixture (uses unpicklable local functions)
 if 'pytest' not in sys.modules:

changedetectionio/api/Import.py

@@ -154,11 +154,10 @@ class Import(Resource):
             if extras['processor'] not in available:
                 return f"Invalid processor '{extras['processor']}'. Available processors: {', '.join(available)}", 400
 
-        # Validate fetch_backend if provided
+        # Validate fetch_backend if provided (legacy API compat — still accepted, stored as-is)
         if 'fetch_backend' in extras:
             from changedetectionio.content_fetchers import available_fetchers
             available = [f[0] for f in available_fetchers()]
-            # Also allow 'system' and extra_browser_* patterns
             is_valid = (
                 extras['fetch_backend'] == 'system' or
                 extras['fetch_backend'] in available or
@@ -167,6 +166,14 @@ class Import(Resource):
             if not is_valid:
                 return f"Invalid fetch_backend '{extras['fetch_backend']}'. Available: system, {', '.join(available)}", 400
 
+        # Validate browser_profile if provided
+        if 'browser_profile' in extras:
+            from changedetectionio.model.browser_profile import get_builtin_profiles, RESERVED_MACHINE_NAMES
+            store_profiles = self.datastore.data['settings']['application'].get('browser_profiles', {})
+            known = set(get_builtin_profiles().keys()) | set(store_profiles.keys()) | {'system', None}
+            if extras['browser_profile'] not in known:
+                return f"Invalid browser_profile '{extras['browser_profile']}'. Available: {', '.join(str(k) for k in known)}", 400
+
         # Validate notification_urls if provided
         if 'notification_urls' in extras:
             from wtforms import ValidationError

changedetectionio/api/Tags.py

@@ -85,6 +85,9 @@ class Tag(Resource):
         # Create clean tag dict without Watch-specific fields
         clean_tag = {k: v for k, v in tag.items() if k not in watch_only_fields}
 
+        # fetch_backend is a legacy field superseded by browser_profile — omit from API response
+        clean_tag.pop('fetch_backend', None)
+
         return clean_tag
 
     @auth.check_token
@@ -177,6 +180,13 @@ class Tag(Resource):
 
         new_uuid = self.datastore.add_tag(title=title)
         if new_uuid:
+            # Apply any extra fields (e.g. processor_config_restock_diff) beyond just title
+            extra = {k: v for k, v in json_data.items() if k != 'title'}
+            if extra:
+                tag = self.datastore.data['settings']['application']['tags'].get(new_uuid)
+                if tag:
+                    tag.update(extra)
+                    tag.commit()
             return {'uuid': new_uuid}, 201
         else:
             return "Invalid or unsupported tag", 400

changedetectionio/api/Watch.py

@@ -105,6 +105,9 @@ class Watch(Resource):
         watch['viewed'] = watch_obj.viewed
         watch['link'] = watch_obj.link,
 
+        # fetch_backend is a legacy field superseded by browser_profile — omit from API response
+        watch.pop('fetch_backend', None)
+
         return watch
 
     @auth.check_token
@@ -338,7 +341,7 @@ class WatchHistoryDiff(Resource):
             word_diff = True
 
         # Get boolean diff preferences with defaults from DIFF_PREFERENCES_CONFIG
-        changes_only = strtobool(request.args.get('changesOnly', 'true'))
+        changes_only = strtobool(request.args.get('changesOnly', 'false'))
         ignore_whitespace = strtobool(request.args.get('ignoreWhitespace', 'false'))
         include_removed = strtobool(request.args.get('removed', 'true'))
         include_added = strtobool(request.args.get('added', 'true'))
@@ -349,7 +352,7 @@ class WatchHistoryDiff(Resource):
             previous_version_file_contents=from_version_file_contents,
             newest_version_file_contents=to_version_file_contents,
             ignore_junk=ignore_whitespace,
-            include_equal=changes_only,
+            include_equal=not changes_only,
             include_removed=include_removed,
             include_added=include_added,
             include_replaced=include_replaced,
@@ -567,4 +570,4 @@ class CreateWatch(Resource):
 
                 return {'status': f'OK, queueing {len(watches_to_queue)} watches in background'}, 202
 
-        return list, 200
+        return list, 200

changedetectionio/blueprint/backups/__init__.py

@@ -40,11 +40,6 @@ def create_backup(datastore_path, watches: dict, tags: dict = None):
             zipObj.write(url_watches_json, arcname="url-watches.json")
             logger.debug("Added url-watches.json to backup")
 
-        # Add the flask app secret (if it exists)
-        secret_file = os.path.join(datastore_path, "secret.txt")
-        if os.path.isfile(secret_file):
-            zipObj.write(secret_file, arcname="secret.txt")
-
         # Add tag data directories (each tag has its own {uuid}/tag.json)
         for uuid, tag in (tags or {}).items():
             for f in Path(tag.data_dir).glob('*'):
@@ -151,19 +146,22 @@ def construct_blueprint(datastore: ChangeDetectionStore):
     def download_backup(filename):
         import re
         filename = filename.strip()
-        backup_filename_regex = BACKUP_FILENAME_FORMAT.format("\d+")
-
-        full_path = os.path.join(os.path.abspath(datastore.datastore_path), filename)
-        if not full_path.startswith(os.path.abspath(datastore.datastore_path)):
-            abort(404)
+        backup_filename_regex = BACKUP_FILENAME_FORMAT.format(r"\d+")
 
+        # Resolve 'latest' before any validation so checks run against the real filename.
         if filename == 'latest':
             backups = find_backups()
+            if not backups:
+                abort(404)
             filename = backups[0]['filename']
 
         if not re.match(r"^" + backup_filename_regex + "$", filename):
             abort(400)  # Bad Request if the filename doesn't match the pattern
 
+        full_path = os.path.join(os.path.abspath(datastore.datastore_path), filename)
+        if not full_path.startswith(os.path.abspath(datastore.datastore_path) + os.sep):
+            abort(404)
+
         logger.debug(f"Backup download request for '{full_path}'")
         return send_from_directory(os.path.abspath(datastore.datastore_path), filename, as_attachment=True)
 

changedetectionio/blueprint/backups/restore.py

@@ -1,6 +1,7 @@
 import io
 import json
 import os
+import re
 import shutil
 import tempfile
 import threading
@@ -14,6 +15,16 @@ from loguru import logger
 
 from changedetectionio.flask_app import login_optionally_required
 
+# Maximum size of the uploaded zip file. Override via env var MAX_RESTORE_UPLOAD_MB.
+_MAX_UPLOAD_BYTES = int(os.getenv("MAX_RESTORE_UPLOAD_MB", 256)) * 1024 * 1024
+# Maximum total uncompressed size of all entries (zip-bomb guard). Override via MAX_RESTORE_DECOMPRESSED_MB.
+_MAX_DECOMPRESSED_BYTES = int(os.getenv("MAX_RESTORE_DECOMPRESSED_MB", 1024)) * 1024 * 1024
+# Only top-level directories whose name is a valid UUID are treated as watch/tag entries.
+_UUID_RE = re.compile(
+    r'^[0-9a-f]{8}-[0-9a-f]{4}-4[0-9a-f]{3}-[89ab][0-9a-f]{3}-[0-9a-f]{12}$',
+    re.IGNORECASE,
+)
+
 
 class RestoreForm(Form):
     zip_file = FileField(_l('Backup zip file'), validators=[
@@ -50,7 +61,18 @@ def import_from_zip(zip_stream, datastore, include_groups, include_groups_replac
     with tempfile.TemporaryDirectory() as tmpdir:
         logger.debug(f"Restore: extracting zip to {tmpdir}")
         with zipfile.ZipFile(zip_stream, 'r') as zf:
-            zf.extractall(tmpdir)
+            total_uncompressed = sum(m.file_size for m in zf.infolist())
+            if total_uncompressed > _MAX_DECOMPRESSED_BYTES:
+                raise ValueError(
+                    f"Backup archive decompressed size ({total_uncompressed // (1024 * 1024)} MB) "
+                    f"exceeds the {_MAX_DECOMPRESSED_BYTES // (1024 * 1024)} MB limit"
+                )
+            resolved_dest = os.path.realpath(tmpdir)
+            for member in zf.infolist():
+                member_dest = os.path.realpath(os.path.join(resolved_dest, member.filename))
+                if not member_dest.startswith(resolved_dest + os.sep) and member_dest != resolved_dest:
+                    raise ValueError(f"Zip Slip path traversal detected in backup archive: {member.filename!r}")
+                zf.extract(member, tmpdir)
         logger.debug("Restore: zip extracted, scanning UUID directories")
 
         for entry in os.scandir(tmpdir):
@@ -58,6 +80,9 @@ def import_from_zip(zip_stream, datastore, include_groups, include_groups_replac
                 continue
 
             uuid = entry.name
+            if not _UUID_RE.match(uuid):
+                logger.warning(f"Restore: skipping non-UUID directory {uuid!r}")
+                continue
             tag_json_path = os.path.join(entry.path, 'tag.json')
             watch_json_path = os.path.join(entry.path, 'watch.json')
 
@@ -155,7 +180,9 @@ def construct_restore_blueprint(datastore):
         form = RestoreForm()
         return render_template("backup_restore.html",
                                form=form,
-                               restore_running=any(t.is_alive() for t in restore_threads))
+                               restore_running=any(t.is_alive() for t in restore_threads),
+                               max_upload_mb=_MAX_UPLOAD_BYTES // (1024 * 1024),
+                               max_decompressed_mb=_MAX_DECOMPRESSED_BYTES // (1024 * 1024))
 
     @login_optionally_required
     @restore_blueprint.route("/restore/start", methods=['POST'])
@@ -173,10 +200,22 @@ def construct_restore_blueprint(datastore):
             flash(gettext("File must be a .zip backup file"), "error")
             return redirect(url_for('backups.restore.restore'))
 
-        # Read into memory now — the request stream is gone once we return
+        # Reject oversized uploads before reading the stream into memory.
+        content_length = request.content_length
+        if content_length and content_length > _MAX_UPLOAD_BYTES:
+            flash(gettext("Backup file is too large (max %(mb)s MB)", mb=_MAX_UPLOAD_BYTES // (1024 * 1024)), "error")
+            return redirect(url_for('backups.restore.restore'))
+
+        # Read into memory now — the request stream is gone once we return.
+        # Read one byte beyond the limit so we can detect truncated-but-still-oversized streams.
         try:
-            zip_bytes = io.BytesIO(zip_file.read())
-            zipfile.ZipFile(zip_bytes)  # quick validity check before spawning
+            raw = zip_file.read(_MAX_UPLOAD_BYTES + 1)
+            if len(raw) > _MAX_UPLOAD_BYTES:
+                flash(gettext("Backup file is too large (max %(mb)s MB)", mb=_MAX_UPLOAD_BYTES // (1024 * 1024)), "error")
+                return redirect(url_for('backups.restore.restore'))
+            zip_bytes = io.BytesIO(raw)
+            with zipfile.ZipFile(zip_bytes):  # quick validity check before spawning
+                pass
             zip_bytes.seek(0)
         except zipfile.BadZipFile:
             flash(gettext("Invalid or corrupted zip file"), "error")
@@ -201,6 +240,7 @@ def construct_restore_blueprint(datastore):
             name="BackupRestore"
         )
         restore_thread.start()
+        restore_threads[:] = [t for t in restore_threads if t.is_alive()]
         restore_threads.append(restore_thread)
         flash(gettext("Restore started in background, check back in a few minutes."))
         return redirect(url_for('backups.restore.restore'))

changedetectionio/blueprint/backups/templates/backup_restore.html

@@ -19,6 +19,10 @@
 
                 <p>{{ _('Restore a backup. Must be a .zip backup file created on/after v0.53.1 (new database layout).') }}</p>
                 <p>{{ _('Note: This does not override the main application settings, only watches and groups.') }}</p>
+                <p class="pure-form-message">
+                    {{ _('Max upload size: %(upload)s MB &nbsp;·&nbsp; Max decompressed size: %(decomp)s MB',
+                         upload=max_upload_mb, decomp=max_decompressed_mb) }}
+                </p>
 
                 <form class="pure-form pure-form-stacked settings"
                       action="{{ url_for('backups.restore.backups_restore_start') }}"

changedetectionio/blueprint/browser_steps/__init__.py

@@ -102,6 +102,35 @@ def run_async_in_browser_loop(coro):
     else:
         raise RuntimeError("Browser steps event loop is not available")
 
+async def _close_session_resources(session_data, label=''):
+    """Close all browser resources for a session in the correct order.
+
+    browserstepper.cleanup() closes page+context but not the browser itself.
+    For CloakBrowser, browser.close() is what stops the local Chromium process via pw.stop().
+    For the default CDP path, playwright_context.stop() shuts down the playwright instance.
+    """
+    browserstepper = session_data.get('browserstepper')
+    if browserstepper:
+        try:
+            await browserstepper.cleanup()
+        except Exception as e:
+            logger.error(f"Error cleaning up browserstepper{label}: {e}")
+
+    browser = session_data.get('browser')
+    if browser:
+        try:
+            await asyncio.wait_for(browser.close(), timeout=5.0)
+        except Exception as e:
+            logger.warning(f"Error closing browser{label}: {e}")
+
+    playwright_context = session_data.get('playwright_context')
+    if playwright_context:
+        try:
+            await playwright_context.stop()
+        except Exception as e:
+            logger.warning(f"Error stopping playwright context{label}: {e}")
+
+
 def cleanup_expired_sessions():
     """Remove expired browsersteps sessions and cleanup their resources"""
     global browsersteps_sessions, browsersteps_watch_to_session
@@ -119,13 +148,10 @@ def cleanup_expired_sessions():
         logger.debug(f"Cleaning up expired browsersteps session {session_id}")
         session_data = browsersteps_sessions[session_id]
 
-        # Cleanup playwright resources asynchronously
-        browserstepper = session_data.get('browserstepper')
-        if browserstepper:
-            try:
-                run_async_in_browser_loop(browserstepper.cleanup())
-            except Exception as e:
-                logger.error(f"Error cleaning up session {session_id}: {e}")
+        try:
+            run_async_in_browser_loop(_close_session_resources(session_data, label=f" for session {session_id}"))
+        except Exception as e:
+            logger.error(f"Error cleaning up session {session_id}: {e}")
 
         # Remove from sessions dict
         del browsersteps_sessions[session_id]
@@ -152,12 +178,10 @@ def cleanup_session_for_watch(watch_uuid):
 
     session_data = browsersteps_sessions.get(session_id)
     if session_data:
-        browserstepper = session_data.get('browserstepper')
-        if browserstepper:
-            try:
-                run_async_in_browser_loop(browserstepper.cleanup())
-            except Exception as e:
-                logger.error(f"Error cleaning up session {session_id} for watch {watch_uuid}: {e}")
+        try:
+            run_async_in_browser_loop(_close_session_resources(session_data, label=f" for watch {watch_uuid}"))
+        except Exception as e:
+            logger.error(f"Error cleaning up session {session_id} for watch {watch_uuid}: {e}")
 
         # Remove from sessions dict
         del browsersteps_sessions[session_id]
@@ -178,59 +202,64 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         import time
         from playwright.async_api import async_playwright
 
-        # We keep the playwright session open for many minutes
         keepalive_seconds = int(os.getenv('BROWSERSTEPS_MINUTES_KEEPALIVE', 10)) * 60
+        keepalive_ms = ((keepalive_seconds + 3) * 1000)
 
         browsersteps_start_session = {'start_time': time.time()}
 
-        # Create a new async playwright instance for browser steps
-        playwright_instance = async_playwright()
-        playwright_context = await playwright_instance.start()
+        # Build proxy dict first — needed by both the CDP path and fetcher-specific launchers
+        proxy_url = datastore.get_proxy_url_for_watch(uuid=watch_uuid)
+        proxy = None
+        if proxy_url:
+            from urllib.parse import urlparse
+            parsed = urlparse(proxy_url)
+            proxy = {'server': proxy_url}
+            if parsed.username:
+                proxy['username'] = parsed.username
+            if parsed.password:
+                proxy['password'] = parsed.password
+            logger.debug(f"Browser Steps: UUID {watch_uuid} selected proxy {proxy_url}")
 
-        keepalive_ms = ((keepalive_seconds + 3) * 1000)
-        base_url = os.getenv('PLAYWRIGHT_DRIVER_URL', '').strip('"')
-        a = "?" if not '?' in base_url else '&'
-        base_url += a + f"timeout={keepalive_ms}"
+        # Resolve the fetcher class for this watch so we can ask it to launch its own browser
+        # if it supports that (e.g. CloakBrowser, which runs locally rather than via CDP)
+        watch = datastore.data['watching'][watch_uuid]
+        from changedetectionio import content_fetchers
+        fetcher_class = content_fetchers.get_fetcher(watch.effective_browser_profile.fetch_backend)
+
+        browser = None
+        playwright_context = None
+
+        # If the fetcher has its own browser launch for the live steps UI, use it.
+        # get_browsersteps_browser(proxy, keepalive_ms) returns (browser, playwright_context_or_None)
+        # or None to fall back to the default CDP path.
+        if fetcher_class and hasattr(fetcher_class, 'get_browsersteps_browser'):
+            result = await fetcher_class.get_browsersteps_browser(proxy=proxy, keepalive_ms=keepalive_ms)
+            if result is not None:
+                browser, playwright_context = result
+                logger.debug(f"Browser Steps: using fetcher-specific browser for '{fetcher_class.__name__}'")
+
+        # Default: connect to the remote Playwright/sockpuppetbrowser via CDP
+        if browser is None:
+            playwright_instance = async_playwright()
+            playwright_context = await playwright_instance.start()
+            base_url = os.getenv('PLAYWRIGHT_DRIVER_URL', '').strip('"')
+            a = "?" if '?' not in base_url else '&'
+            base_url += a + f"timeout={keepalive_ms}"
+            browser = await playwright_context.chromium.connect_over_cdp(base_url, timeout=keepalive_ms)
+            logger.debug(f"Browser Steps: using CDP connection to {base_url}")
 
-        browser = await playwright_context.chromium.connect_over_cdp(base_url, timeout=keepalive_ms)
         browsersteps_start_session['browser'] = browser
         browsersteps_start_session['playwright_context'] = playwright_context
 
-        proxy_id = datastore.get_preferred_proxy_for_watch(uuid=watch_uuid)
-        proxy = None
-        if proxy_id:
-            proxy_url = datastore.proxy_list.get(proxy_id).get('url')
-            if proxy_url:
-
-                # Playwright needs separate username and password values
-                from urllib.parse import urlparse
-                parsed = urlparse(proxy_url)
-                proxy = {'server': proxy_url}
-
-                if parsed.username:
-                    proxy['username'] = parsed.username
-
-                if parsed.password:
-                    proxy['password'] = parsed.password
-
-                logger.debug(f"Browser Steps: UUID {watch_uuid} selected proxy {proxy_url}")
-
-        # Tell Playwright to connect to Chrome and setup a new session via our stepper interface
         browserstepper = browser_steps.browsersteps_live_ui(
             playwright_browser=browser,
             proxy=proxy,
-            start_url=datastore.data['watching'][watch_uuid].link,
-            headers=datastore.data['watching'][watch_uuid].get('headers')
+            start_url=watch.link,
+            headers=watch.get('headers')
         )
-        
-        # Initialize the async connection
         await browserstepper.connect(proxy=proxy)
-        
         browsersteps_start_session['browserstepper'] = browserstepper
 
-        # For test
-        #await browsersteps_start_session['browserstepper'].action_goto_url(value="http://example.com?time="+str(time.time()))
-
         return browsersteps_start_session
 
 

changedetectionio/blueprint/check_proxies/__init__.py

@@ -40,12 +40,14 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         contents = ''
         now = time.time()
         try:
+            import asyncio
             processor_module = importlib.import_module("changedetectionio.processors.text_json_diff.processor")
             update_handler = processor_module.perform_site_check(datastore=datastore,
                                                                  watch_uuid=uuid
                                                                  )
 
-            update_handler.call_browser(preferred_proxy_id=preferred_proxy)
+            update_handler.preferred_proxy_override = preferred_proxy
+            asyncio.run(update_handler.call_browser())
         # title, size is len contents not len xfer
         except content_fetcher_exceptions.Non200ErrorCodeReceived as e:
             if e.status_code == 404:

changedetectionio/blueprint/imports/importer.py

@@ -175,9 +175,9 @@ class import_xlsx_wachete(Importer):
                 dynamic_wachet = str(data.get('dynamic wachet', '')).strip().lower()  # Convert bool to str to cover all cases
                 # libreoffice and others can have it as =FALSE() =TRUE(), or bool(true)
                 if 'true' in dynamic_wachet or dynamic_wachet == '1':
-                    extras['fetch_backend'] = 'html_webdriver'
+                    extras['browser_profile'] = 'browser_chromeplaywright'
                 elif 'false' in dynamic_wachet or dynamic_wachet == '0':
-                    extras['fetch_backend'] = 'html_requests'
+                    extras['browser_profile'] = 'direct_http_requests'
 
                 if data.get('xpath'):
                     # @todo split by || ?

changedetectionio/blueprint/rss/tag.py

@@ -7,7 +7,7 @@ def construct_tag_routes(rss_blueprint, datastore):
         datastore: The ChangeDetectionStore instance
     """
 
-    @rss_blueprint.route("/tag/<string:tag_uuid>", methods=['GET'])
+    @rss_blueprint.route("/tag/<uuid_str:tag_uuid>", methods=['GET'])
     def rss_tag_feed(tag_uuid):
 
         from flask import make_response, request, url_for

changedetectionio/blueprint/settings/__init__.py

@@ -15,6 +15,9 @@ from changedetectionio.auth_decorator import login_optionally_required
 def construct_blueprint(datastore: ChangeDetectionStore):
     settings_blueprint = Blueprint('settings', __name__, template_folder="templates")
 
+    from changedetectionio.blueprint.settings.browser_profile import construct_blueprint as construct_browser_profile_blueprint
+    settings_blueprint.register_blueprint(construct_browser_profile_blueprint(datastore), url_prefix='/browsers')
+
     @settings_blueprint.route("", methods=['GET', "POST"])
     @login_optionally_required
     def settings_page():

changedetectionio/blueprint/settings/browser_profile/__init__.py

@@ -0,0 +1,200 @@
+import flask_login
+from flask import Blueprint, render_template, request, redirect, url_for, flash
+from flask_babel import gettext
+
+from changedetectionio.store import ChangeDetectionStore
+from changedetectionio.auth_decorator import login_optionally_required
+
+
+def construct_blueprint(datastore: ChangeDetectionStore):
+    settings_browser_profile_blueprint = Blueprint(
+        'settings_browsers',
+        __name__,
+        template_folder="templates"
+    )
+
+    def _render_index(browser_profile_form=None, editing_machine_name=None):
+        from changedetectionio import forms
+        from changedetectionio import content_fetchers as cf
+        from changedetectionio.model.browser_profile import BrowserProfile, RESERVED_MACHINE_NAMES
+
+        # Only browser-capable fetchers are valid profile types
+        fetcher_choices = cf.available_browser_fetchers()
+        if browser_profile_form is None:
+            browser_profile_form = forms.BrowserProfileForm()
+        browser_profile_form.fetch_backend.choices = fetcher_choices
+
+        fetcher_supports_screenshots = {name: True for name, _ in fetcher_choices}
+        fetcher_requires_connection_url = {name: True for name, cls in cf.FETCHERS.items()
+                                           if getattr(cls, 'requires_connection_url', False)}
+
+        # Table shows default built-in profiles first, then user-created profiles
+        store_profiles = datastore.data['settings']['application'].get('browser_profiles', {})
+        user_profiles = dict(cf.DEFAULT_BROWSER_PROFILES)
+        for machine_name, raw in store_profiles.items():
+            try:
+                user_profiles[machine_name] = BrowserProfile(**raw) if isinstance(raw, dict) else raw
+            except Exception:
+                pass
+
+        current_default = datastore.data['settings']['application'].get('browser_profile') or 'direct_http_requests'
+
+        return render_template(
+            "browser_profiles.html",
+            browser_profiles=user_profiles,
+            browser_profile_form=browser_profile_form,
+            reserved_browser_profile_names=RESERVED_MACHINE_NAMES,
+            fetcher_choices=fetcher_choices,
+            fetcher_supports_screenshots=fetcher_supports_screenshots,
+            fetcher_requires_connection_url=fetcher_requires_connection_url,
+            current_default_profile=current_default,
+            editing_machine_name=editing_machine_name,
+        )
+
+    @settings_browser_profile_blueprint.route("", methods=['GET'])
+    @login_optionally_required
+    def index():
+        return _render_index()
+
+    @settings_browser_profile_blueprint.route("/<string:machine_name>/edit", methods=['GET'])
+    @login_optionally_required
+    def edit(machine_name):
+        from changedetectionio import forms
+        from changedetectionio.model.browser_profile import BrowserProfile, RESERVED_MACHINE_NAMES
+
+        if machine_name in RESERVED_MACHINE_NAMES:
+            flash(gettext("Built-in browser profiles cannot be edited."), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        store_profiles = datastore.data['settings']['application'].get('browser_profiles', {})
+        raw = store_profiles.get(machine_name)
+        if raw is None:
+            flash(gettext("Browser profile not found."), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        profile = BrowserProfile(**raw) if isinstance(raw, dict) else raw
+        form = forms.BrowserProfileForm(data=profile.model_dump())
+        return _render_index(browser_profile_form=form, editing_machine_name=machine_name)
+
+    @settings_browser_profile_blueprint.route("/save", methods=['POST'])
+    @login_optionally_required
+    def save():
+        from changedetectionio import forms
+        from changedetectionio import content_fetchers as cf
+        from changedetectionio.model.browser_profile import BrowserProfile, RESERVED_MACHINE_NAMES
+
+        fetcher_choices = [(name, desc) for name, desc in cf.available_fetchers()]
+        browser_profile_form = forms.BrowserProfileForm(formdata=request.form)
+        browser_profile_form.fetch_backend.choices = fetcher_choices
+
+        if not browser_profile_form.validate():
+            flash(gettext("Browser profile error: {}").format(
+                '; '.join(str(e) for errs in browser_profile_form.errors.values() for e in errs)
+            ), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        name = browser_profile_form.name.data.strip()
+        machine_name = BrowserProfile.machine_name_from_str(name)
+
+        if machine_name in RESERVED_MACHINE_NAMES:
+            flash(gettext("Cannot use reserved profile name '{}'. Please choose a different name.").format(name), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        original_machine_name = request.form.get('original_machine_name', '').strip()
+        store_profiles = datastore.data['settings']['application'].setdefault('browser_profiles', {})
+
+        if machine_name != original_machine_name and machine_name in store_profiles:
+            flash(gettext("A browser profile named '{}' already exists.").format(name), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        profile_data = {
+            'name': name,
+            'fetch_backend': browser_profile_form.fetch_backend.data,
+            'browser_connection_url': browser_profile_form.browser_connection_url.data or None,
+            'viewport_width': browser_profile_form.viewport_width.data or 1280,
+            'viewport_height': browser_profile_form.viewport_height.data or 1000,
+            'block_images': bool(browser_profile_form.block_images.data),
+            'block_fonts': bool(browser_profile_form.block_fonts.data),
+            'ignore_https_errors': bool(browser_profile_form.ignore_https_errors.data),
+            'user_agent': browser_profile_form.user_agent.data or None,
+            'locale': browser_profile_form.locale.data or None,
+            'custom_headers': browser_profile_form.custom_headers.data or '',
+            'is_builtin': False,
+        }
+
+        try:
+            BrowserProfile(**profile_data)
+        except Exception as e:
+            flash(gettext("Browser profile validation error: {}").format(str(e)), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        # Handle rename: remove old key, cascade-update watches and tags
+        if original_machine_name and original_machine_name != machine_name and original_machine_name in store_profiles:
+            del store_profiles[original_machine_name]
+            for watch in datastore.data['watching'].values():
+                if watch.get('browser_profile') == original_machine_name:
+                    watch['browser_profile'] = machine_name
+            for tag in datastore.data.get('settings', {}).get('application', {}).get('tags', {}).values():
+                if tag.get('browser_profile') == original_machine_name:
+                    tag['browser_profile'] = machine_name
+
+        store_profiles[machine_name] = profile_data
+        datastore.commit()
+        flash(gettext("Browser profile '{}' saved.").format(name), 'notice')
+        return redirect(url_for('settings.settings_browsers.index'))
+
+    @settings_browser_profile_blueprint.route("/<string:machine_name>/delete", methods=['GET'])
+    @login_optionally_required
+    def delete(machine_name):
+        from changedetectionio.model.browser_profile import RESERVED_MACHINE_NAMES
+
+        if machine_name in RESERVED_MACHINE_NAMES:
+            flash(gettext("Built-in browser profiles cannot be deleted."), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        store_profiles = datastore.data['settings']['application'].get('browser_profiles', {})
+        if machine_name not in store_profiles:
+            flash(gettext("Browser profile not found."), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        raw = store_profiles[machine_name]
+        profile_name = raw.get('name', machine_name) if isinstance(raw, dict) else machine_name
+
+        for watch in datastore.data['watching'].values():
+            if watch.get('browser_profile') == machine_name:
+                watch['browser_profile'] = None
+
+        for tag in datastore.data.get('settings', {}).get('application', {}).get('tags', {}).values():
+            if tag.get('browser_profile') == machine_name:
+                tag['browser_profile'] = None
+
+        if datastore.data['settings']['application'].get('browser_profile') == machine_name:
+            datastore.data['settings']['application']['browser_profile'] = None
+
+        del store_profiles[machine_name]
+        datastore.commit()
+        flash(gettext("Browser profile '{}' deleted.").format(profile_name), 'notice')
+        return redirect(url_for('settings.settings_browsers.index'))
+
+    @settings_browser_profile_blueprint.route("/set-default", methods=['POST'])
+    @login_optionally_required
+    def set_default():
+        from changedetectionio import content_fetchers as cf
+
+        machine_name = request.form.get('machine_name', '').strip()
+        if not machine_name:
+            flash(gettext("No profile specified."), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        from changedetectionio.model.browser_profile import get_profile
+        store_profiles = datastore.data['settings']['application'].get('browser_profiles', {})
+        if get_profile(machine_name, store_profiles) is None:
+            flash(gettext("Unknown browser profile '{}'.").format(machine_name), 'error')
+            return redirect(url_for('settings.settings_browsers.index'))
+
+        datastore.data['settings']['application']['browser_profile'] = machine_name
+        datastore.commit()
+        flash(gettext("Default browser profile set to '{}'.").format(machine_name), 'notice')
+        return redirect(url_for('settings.settings_browsers.index'))
+
+    return settings_browser_profile_blueprint

changedetectionio/blueprint/settings/browser_profile/templates/browser_profiles.html

@@ -0,0 +1,163 @@
+{% extends 'base.html' %}
+{% block content %}
+{% from '_helpers.html' import render_field, render_checkbox_field, render_button %}
+
+<div class="edit-form">
+    <div class="box-wrap inner">
+        <h2>{{ _('Browser Profiles') }}</h2>
+        <p>{{ _('Create named profiles to configure browser settings — viewport size, connection URL, image/font blocking, and more. Each profile is based on an available browser type.') }}</p>
+
+        <form id="set-default-form" action="{{ url_for('settings.settings_browsers.set_default') }}" method="POST">
+            <input type="hidden" name="csrf_token" value="{{ csrf_token() }}">
+            <input type="hidden" name="machine_name" id="default-machine-name" value="">
+        </form>
+        {% if browser_profiles %}
+        <table class="pure-table pure-table-striped" style="width:100%; margin-bottom:1.5em;">
+            <thead>
+                <tr>
+                    <th style="width:2.5em; text-align:center;" title="{{ _('System default') }}">{{ _('Default') }}</th>
+                    <th>{{ _('Name') }}</th>
+                    <th>{{ _('Type') }}</th>
+                    <th style="width:3em; text-align:center;"></th>
+                    <th>{{ _('Viewport') }}</th>
+                    <th>{{ _('Options') }}</th>
+                    <th></th>
+                </tr>
+            </thead>
+            <tbody>
+            {% for machine_name, profile in browser_profiles.items() %}
+                <tr>
+                    <td style="text-align:center;">
+                        <input type="radio"
+                               name="default_profile"
+                               value="{{ machine_name }}"
+                               title="{{ _('Set as system default') }}"
+                               {% if machine_name == current_default_profile %}checked{% endif %}
+                               onchange="setDefaultProfile('{{ machine_name }}')">
+                    </td>
+                    <td>{{ profile.name }}</td>
+                    <td><code>{{ profile.fetch_backend }}</code></td>
+                    <td style="text-align:center;">{{ profile.get_fetcher_class_name()|fetcher_status_icons }}</td>
+                    <td>{{ profile.viewport_width }}×{{ profile.viewport_height }}</td>
+                    <td style="font-size:0.8em; line-height:1.6;">
+                        {% if profile.block_images %}{{ _('No images') }}<br>{% endif %}
+                        {% if profile.block_fonts %}{{ _('No fonts') }}<br>{% endif %}
+                        {% if profile.ignore_https_errors %}{{ _('Ignore TLS') }}<br>{% endif %}
+                        {% if profile.browser_connection_url %}<span title="{{ profile.browser_connection_url }}">{{ _('Custom URL') }}</span>{% endif %}
+                    </td>
+                    <td style="white-space:nowrap;">
+                        {% if not profile.is_builtin %}
+                        <a href="{{ url_for('settings.settings_browsers.edit', machine_name=machine_name) }}"
+                           class="pure-button button-small">{{ _('Edit') }}</a>
+                        <a href="{{ url_for('settings.settings_browsers.delete', machine_name=machine_name) }}"
+                           class="pure-button button-small button-error"
+                           onclick="return confirm('{{ _('Delete this browser profile?') }}')">{{ _('Delete') }}</a>
+                        {% endif %}
+                    </td>
+                </tr>
+            {% endfor %}
+            </tbody>
+        </table>
+        {% else %}
+        <p style="color:#888; font-style:italic;">{{ _('No browser profiles configured yet. Add one below.') }}</p>
+        {% endif %}
+
+        <div class="border-fieldset">
+            <h3 id="profile-form-heading">{{ _('Edit browser profile') if editing_machine_name else _('Add new browser profile') }}</h3>
+            {% if not editing_machine_name %}
+            <p style="font-size:0.9em; color:#666;">{{ _('Choose a browser type, give it a name, and configure its settings. You can create multiple profiles from the same type with different connection URLs or options.') }}</p>
+            {% endif %}
+            <form class="pure-form pure-form-stacked"
+                  id="browser-profile-form"
+                  action="{{ url_for('settings.settings_browsers.save') }}"
+                  method="POST">
+                <input type="hidden" name="csrf_token" value="{{ csrf_token() }}">
+                <input type="hidden" name="original_machine_name" id="original_machine_name" value="{{ editing_machine_name or '' }}">
+                <fieldset>
+                    <div class="pure-control-group">
+                        {{ render_field(browser_profile_form.name) }}
+                    </div>
+                    <div class="pure-control-group inline-radio">
+                        {{ render_field(browser_profile_form.fetch_backend, id="profile-fetch-backend") }}
+                    </div>
+                    <div class="pure-control-group browser-only-field cdp-only-field">
+                        {{ render_field(browser_profile_form.browser_connection_url) }}
+                        <span class="pure-form-message-inline">{{ _('Optional — override the system CDP/WebSocket URL for this profile only (e.g.') }} <code>ws://my-chrome:3000</code>).</span>
+                    </div>
+                    <div class="pure-control-group browser-only-field" style="display:flex; gap:1em; flex-wrap:wrap;">
+                        <div>{{ render_field(browser_profile_form.viewport_width) }}</div>
+                        <div>{{ render_field(browser_profile_form.viewport_height) }}</div>
+                    </div>
+                    <div class="pure-control-group browser-only-field">
+                        {{ render_checkbox_field(browser_profile_form.block_images) }}
+                        <span class="pure-form-message-inline">{{ _('Block image downloads — speeds up loads on image-heavy pages.') }}</span>
+                    </div>
+                    <div class="pure-control-group browser-only-field">
+                        {{ render_checkbox_field(browser_profile_form.block_fonts) }}
+                        <span class="pure-form-message-inline">{{ _('Block web font downloads.') }}</span>
+                    </div>
+                    <div class="pure-control-group browser-only-field">
+                        {{ render_checkbox_field(browser_profile_form.ignore_https_errors) }}
+                        <span class="pure-form-message-inline">{{ _('Ignore TLS/HTTPS certificate errors (useful for self-signed certs on staging sites).') }}</span>
+                    </div>
+                    <div class="pure-control-group browser-only-field">
+                        {{ render_field(browser_profile_form.user_agent) }}
+                        <span class="pure-form-message-inline">{{ _("Leave blank to use the fetcher's default User-Agent.") }}</span>
+                    </div>
+                    <div class="pure-control-group browser-only-field">
+                        {{ render_field(browser_profile_form.locale) }}
+                        <span class="pure-form-message-inline">{{ _('Sets Accept-Language and navigator.language (e.g. en-US, de-DE).') }}</span>
+                    </div>
+                    <div class="pure-control-group">
+                        {{ render_field(browser_profile_form.custom_headers) }}
+                        <span class="pure-form-message-inline">{{ _('Extra HTTP headers for all requests using this profile (one per line, Key: Value). Applied before per-watch headers.') }}</span>
+                    </div>
+                    <div class="pure-control-group">
+                        <button type="submit" class="pure-button pure-button-primary" id="profile-submit-btn">{{ _('Save profile') }}</button>
+                        {% if editing_machine_name %}
+                        <a href="{{ url_for('settings.settings_browsers.index') }}" class="pure-button button-cancel">{{ _('Cancel') }}</a>
+                        {% endif %}
+                        <a href="{{ url_for('settings.settings_page') }}" class="pure-button button-cancel">{{ _('Back to Settings') }}</a>
+                    </div>
+                </fieldset>
+            </form>
+        </div>
+    </div>
+</div>
+
+<script>
+function setDefaultProfile(machineName) {
+    document.getElementById('default-machine-name').value = machineName;
+    document.getElementById('set-default-form').submit();
+}
+
+const fetcherSupportsBrowser = {{ fetcher_supports_screenshots | tojson }};
+const fetcherRequiresConnectionUrl = {{ fetcher_requires_connection_url | tojson }};
+
+function updateBrowserFieldVisibility() {
+    const fetchBackend = document.getElementById('profile-fetch-backend').value;
+    const isBrowser = !!fetcherSupportsBrowser[fetchBackend];
+    const isCdp = !!fetcherRequiresConnectionUrl[fetchBackend];
+    document.querySelectorAll('.browser-only-field').forEach(function(el) {
+        el.style.display = isBrowser ? '' : 'none';
+    });
+    document.querySelectorAll('.cdp-only-field').forEach(function(el) {
+        el.style.display = isCdp ? '' : 'none';
+    });
+}
+
+document.addEventListener('DOMContentLoaded', function() {
+    const sel = document.getElementById('profile-fetch-backend');
+    if (sel) {
+        sel.addEventListener('change', updateBrowserFieldVisibility);
+        updateBrowserFieldVisibility();
+    }
+});
+
+{% if editing_machine_name %}
+document.addEventListener('DOMContentLoaded', function() {
+    document.getElementById('browser-profile-form').scrollIntoView({behavior: 'smooth'});
+});
+{% endif %}
+</script>
+{% endblock %}

changedetectionio/blueprint/settings/templates/settings.html

@@ -28,6 +28,7 @@
             <li class="tab"><a href="{{ url_for('backups.create') }}">{{ _('Backups') }}</a></li>
             <li class="tab"><a href="#timedate">{{ _('Time & Date') }}</a></li>
             <li class="tab"><a href="#proxies">{{ _('CAPTCHA & Proxies') }}</a></li>
+            <li class="tab"><a href="{{ url_for('settings.settings_browsers.index') }}">{{ _('Browsers') }}</a></li>
             {% if plugin_tabs %}
                 {% for tab in plugin_tabs %}
             <li class="tab"><a href="#plugin-{{ tab.plugin_id }}">{{ tab.tab_label }}</a></li>
@@ -115,14 +116,7 @@
             </div>
 
             <div class="tab-pane-inner" id="fetching">
-                <div class="pure-control-group inline-radio">
-                    {{ render_field(form.application.form.fetch_backend, class="fetch-backend") }}
-                    <span class="pure-form-message-inline">
-                        <p>{{ _('Use the') }} <strong>{{ _('Basic') }}</strong> {{ _('method (default) where your watched sites don\'t need Javascript to render.') }}</p>
-                        <p>{{ _('The') }} <strong>{{ _('Chrome/Javascript') }}</strong> {{ _('method requires a network connection to a running WebDriver+Chrome server, set by the ENV var') }} 'WEBDRIVER_URL'. </p>
-                    </span>
-                </div>
-                <fieldset class="pure-group" id="webdriver-override-options" data-visible-for="application-fetch_backend=html_webdriver">
+                <fieldset class="pure-group" id="webdriver-override-options">
                     <div class="pure-form-message-inline">
                         <strong>{{ _('If you\'re having trouble waiting for the page to be fully rendered (text missing etc), try increasing the \'wait\' time here.') }}</strong>
                         <br>
@@ -146,17 +140,9 @@
                     {{ render_field(form.requests.form.timeout) }}
                     <span class="pure-form-message-inline">{{ _('For regular plain requests (not chrome based), maximum number of seconds until timeout, 1-999.') }}</span><br>
                 </div>
-                <div class="pure-control-group inline-radio">
-                    {{ render_field(form.requests.form.default_ua) }}
-                    <span class="pure-form-message-inline">
-                        {{ _('Applied to all requests.') }}<br><br>
-                        {{ _('Note: Simply changing the User-Agent often does not defeat anti-robot technologies, it\'s important to consider') }} <a href="https://changedetection.io/tutorial/what-are-main-types-anti-robot-mechanisms">{{ _('all of the ways that the browser is detected') }}</a>.
-                    </span>
-                </div>
                 <div class="pure-control-group">
-                                        <br>
-                    {{ _('Tip:') }} <a href="https://github.com/dgtlmoon/changedetection.io/wiki/Proxy-configuration#brightdata-proxy-support">{{ _('Connect using Bright Data and Oxylabs Proxies, find out more here.') }}</a>
-
+                <br>
+                    {{ _('Tip:') }} <a href="{{ url_for('settings.settings_page')}}#proxies">{{ _('Connect using Bright Data proxies, find out more here.') }}</a>
                 </div>
             </div>
 
@@ -352,7 +338,7 @@ nav
                     </div>
                 </div>
 
-                <p><strong>{{ _('Tip') }}</strong>: {{ _('"Residential" and "Mobile" proxy type can be more successfull than "Data Center" for blocked websites.') }}</p>
+                <p><strong>{{ _('Tip') }}</strong>: {{ _('"Residential" and "Mobile" proxy type can be more successful than "Data Center" for blocked websites.') }}</p>
 
                 <div class="pure-control-group" id="extra-proxies-setting">
                 {{ render_fieldlist_with_inline_errors(form.requests.form.extra_proxies) }}

changedetectionio/blueprint/ui/__init__.py

@@ -156,9 +156,9 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, worker_pool,
     @login_optionally_required
     def clear_all_history():
         if request.method == 'POST':
-            confirmtext = request.form.get('confirmtext')
+            confirmtext = request.form.get('confirmtext', '')
 
-            if confirmtext == 'clear':
+            if confirmtext.strip().lower() == gettext('clear').strip().lower():
                 # Run in background thread to avoid blocking
                 def clear_history_background():
                     # Capture UUIDs first to avoid race conditions

changedetectionio/blueprint/ui/edit.py

@@ -67,6 +67,10 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
                 default['proxy'] = ''
         # proxy_override set to the json/text list of the items
 
+        # browser_profile: None means "use system default" — map to 'system' so the radio pre-selects correctly
+        if not default.get('browser_profile'):
+            default['browser_profile'] = 'system'
+
         # Does it use some custom form? does one exist?
         processor_name = datastore.data['watching'][uuid].get('processor', '')
         processor_classes = next((tpl for tpl in processors.find_processors() if tpl[1] == processor_name), None)
@@ -139,10 +143,37 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
             except Exception as e:
                 logger.warning(f"Failed to load processor config: {e}")
 
-        for p in datastore.extra_browsers:
-            form.fetch_backend.choices.append(p)
+        from changedetectionio.model.browser_profile import BrowserProfile
+        from changedetectionio import content_fetchers as cf
+        store_profiles = datastore.data['settings']['application'].get('browser_profiles', {})
 
-        form.fetch_backend.choices.append(("system", 'System settings default'))
+        # Resolve the name of the system-level default profile for the label
+        from changedetectionio.model.browser_profile import get_profile
+        _system_default_machine_name = datastore.data['settings']['application'].get('browser_profile') or 'direct_http_requests'
+        _all_store_profiles = datastore.data['settings']['application'].get('browser_profiles', {})
+        _default_profile = get_profile(_system_default_machine_name, _all_store_profiles)
+        if _default_profile:
+            _system_label = gettext('System settings default') + ' \u2013 ' + _default_profile.name
+        else:
+            _system_label = gettext('System settings default')
+
+        # Choices: system default + always-present defaults (requests) + user-created profiles
+        form.browser_profile.choices = [('system', _system_label)] + [
+            (p.get_machine_name(), p.name)
+            for p in cf.DEFAULT_BROWSER_PROFILES.values()
+        ] + [
+            (machine_name, raw.get('name', machine_name) if isinstance(raw, dict) else getattr(raw, 'name', machine_name))
+            for machine_name, raw in store_profiles.items()
+        ]
+
+        # Build a map of machine_name → fetcher class name for the JS visibility system
+        all_profiles = dict(cf.DEFAULT_BROWSER_PROFILES)
+        for machine_name, raw in store_profiles.items():
+            try:
+                all_profiles[machine_name] = BrowserProfile(**raw) if isinstance(raw, dict) else raw
+            except Exception:
+                pass
+        browser_profile_fetchers = {mn: p.get_fetcher_class_name() for mn, p in all_profiles.items()}
 
         # form.browser_steps[0] can be assumed that we 'goto url' first
 
@@ -210,7 +241,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
 
             # Recast it if need be to right data Watch handler
             watch_class = processors.get_custom_watch_obj_for_processor(form.data.get('processor'))
-            datastore.data['watching'][uuid] = watch_class(datastore_path=datastore.datastore_path, __datastore=datastore.data, default=datastore.data['watching'][uuid])
+            datastore.data['watching'][uuid] = watch_class(datastore_path=datastore.datastore_path, __datastore=datastore, default=datastore.data['watching'][uuid])
 
             # Save the watch immediately
             datastore.data['watching'][uuid].commit()
@@ -296,6 +327,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
             template_args = {
                 'available_processors': processors.available_processors(),
                 'available_timezones': sorted(available_timezones()),
+                'browser_profile_fetchers': browser_profile_fetchers,
                 'browser_steps_config': browser_step_ui_config,
                 'emailprefix': os.getenv('NOTIFICATION_MAIL_BUTTON_PREFIX', False),
                 'extra_classes': ' '.join(c),

changedetectionio/blueprint/ui/preview.py

@@ -10,7 +10,8 @@ from changedetectionio import html_tools
 def construct_blueprint(datastore: ChangeDetectionStore):
     preview_blueprint = Blueprint('ui_preview', __name__, template_folder="../ui/templates")
 
-    @preview_blueprint.route("/preview/<uuid_str:uuid>", methods=['GET'])
+
+    @preview_blueprint.route("/preview/<uuid_str:uuid>", methods=['GET', 'POST'])
     @login_optionally_required
     def preview_page(uuid):
         """
@@ -59,12 +60,8 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         versions = []
         timestamp = None
 
-        system_uses_webdriver = datastore.data['settings']['application']['fetch_backend'] == 'html_webdriver'
         extra_stylesheets = [url_for('static_content', group='styles', filename='diff.css')]
-
-        is_html_webdriver = False
-        if (watch.get('fetch_backend') == 'system' and system_uses_webdriver) or watch.get('fetch_backend') == 'html_webdriver' or watch.get('fetch_backend', '').startswith('extra_browser_'):
-            is_html_webdriver = True
+        fetcher_supports_screenshots = watch.fetcher_supports_screenshots
 
         triggered_line_numbers = []
         ignored_line_numbers = []
@@ -74,7 +71,9 @@ def construct_blueprint(datastore: ChangeDetectionStore):
             flash(gettext("Preview unavailable - No fetch/check completed or triggers not reached"), "error")
         else:
             # So prepare the latest preview or not
-            preferred_version = request.args.get('version')
+            preferred_version = request.values.get('version') if request.method == 'POST' else request.args.get('version')
+
+
             versions = list(watch.history.keys())
             timestamp = versions[-1]
             if preferred_version and preferred_version in versions:
@@ -113,7 +112,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
                                  highlight_triggered_line_numbers=triggered_line_numbers,
                                  highlight_blocked_line_numbers=blocked_line_numbers,
                                  history_n=watch.history_n,
-                                 is_html_webdriver=is_html_webdriver,
+                                 fetcher_supports_screenshots=fetcher_supports_screenshots,
                                  last_error=watch['last_error'],
                                  last_error_screenshot=watch.get_error_snapshot(),
                                  last_error_text=watch.get_error_text(),

changedetectionio/blueprint/ui/templates/diff.html

@@ -143,7 +143,7 @@
          <div class="tip">
              {{ _('For now, Differences are performed on text, not graphically, only the latest screenshot is available.') }}
          </div>
-         {% if is_html_webdriver %}
+         {% if fetcher_supports_screenshots %}
            {% if screenshot %}
             <div class="snapshot-age">{{watch_a.snapshot_screenshot_ctime|format_timestamp_timeago}}</div>
             <img style="max-width: 80%" id="screenshot-img" alt="{{ _('Current screenshot from most recent request') }}" >

changedetectionio/blueprint/ui/templates/edit.html

@@ -27,7 +27,8 @@
     const proxy_recheck_status_url="{{url_for('check_proxies.get_recheck_status', uuid=uuid)}}";
     const screenshot_url="{{url_for('static_content', group='screenshot', filename=uuid)}}";
     const watch_visual_selector_data_url="{{url_for('static_content', group='visual_selector_data', filename=uuid)}}";
-    const default_system_fetch_backend="{{ settings_application['fetch_backend'] }}";
+    const default_system_fetch_backend = {{ (browser_profile_fetchers.get(settings_application.get('browser_profile') or 'direct_http_requests', 'requests')) | tojson }};
+    const browserProfileFetcherMap = {{ browser_profile_fetchers | tojson }};
 </script>
 <script src="{{url_for('static_content', group='js', filename='plugins.js')}}" defer></script>
 <script src="{{url_for('static_content', group='js', filename='watch-settings.js')}}" defer></script>
@@ -131,11 +132,19 @@
             {% if capabilities.supports_request_type %}
             <div class="tab-pane-inner" id="request">
                     <div class="pure-control-group inline-radio">
-                        {{ render_field(form.fetch_backend, class="fetch-backend") }}
+                        <div><label for="browser_profile">{{ form.browser_profile.label.text }}</label></div>
+                        <div><ul class="fetch-backend" id="browser_profile">
+                        {%- for subfield in form.browser_profile %}
+                            <li>
+                                {{ subfield() }}
+                                {{ browser_profile_fetchers.get(subfield.data, '')|fetcher_status_icons }}
+                                <label for="{{ subfield.id }}">{{ subfield.label.text }}</label>
+                            </li>
+                        {%- endfor %}
+                        </ul></div>
                         <span class="pure-form-message-inline">
-                            <p>{{ _('Use the') }} <strong>{{ _('Basic') }}</strong> {{ _('method (default) where your watched site doesn\'t need Javascript to render.') }}</p>
-                            <p>{{ _('The') }} <strong>{{ _('Chrome/Javascript') }}</strong> {{ _('method requires a network connection to a running WebDriver+Chrome server, set by the ENV var \'WEBDRIVER_URL\'.') }} </p>
-                            {{ _('Tip:') }} <a href="https://github.com/dgtlmoon/changedetection.io/wiki/Proxy-configuration#brightdata-proxy-support">{{ _('Connect using Bright Data and Oxylabs Proxies, find out more here.') }}</a>
+                            <p>{{ _('Choose how this watch fetches its target URL. \'System settings default\' inherits the global setting.') }}</p>
+                            <p>{{ _('Manage browser profiles in') }} <a href="{{ url_for('settings.settings_browsers.index') }}">{{ _('Settings → Browsers') }}</a>.</p>
                         </span>
                     </div>
                 {% if form.proxy %}
@@ -149,7 +158,7 @@
                 {% endif %}
 
                 <!-- webdriver always -->
-                <fieldset data-visible-for="fetch_backend=html_webdriver"  style="display: none;">
+                <fieldset data-visible-for="fetch_backend=playwright fetch_backend=selenium fetch_backend=puppeteer fetch_backend=cloakbrowser"  style="display: none;">
                     <div class="pure-control-group">
                         {{ render_field(form.webdriver_delay) }}
                         <div class="pure-form-message-inline">
@@ -172,8 +181,8 @@
                         </div>
                     </div>
                 </fieldset>
-                <!-- html requests always -->
-                <fieldset data-visible-for="fetch_backend=html_requests">
+                <!-- requests always -->
+                <fieldset data-visible-for="fetch_backend=requests">
                     <div class="pure-control-group">
                         <a class="pure-button button-secondary button-xsmall show-advanced">{{ _('Show advanced options') }}</a>
                     </div>
@@ -210,7 +219,7 @@ Math: {{ 1 + 1 }}") }}
                             ({{ _('Not supported by Selenium browser') }})
                         </div>
                     </div>
-            <fieldset data-visible-for="fetch_backend=html_requests fetch_backend=html_webdriver" >
+            <fieldset data-visible-for="fetch_backend=requests fetch_backend=playwright fetch_backend=selenium fetch_backend=puppeteer fetch_backend=cloakbrowser" >
                     <div class="pure-control-group inline-radio advanced-options"  style="display: none;">
                     {{ render_checkbox_field(form.ignore_status_codes) }}
                     </div>

changedetectionio/blueprint/ui/templates/preview.html

@@ -17,7 +17,7 @@
     <script src="{{ url_for('static_content', group='js', filename='tabs.js') }}" defer></script>
     {% if versions|length >= 2 %}
         <div id="diff-form" style="text-align: center;">
-            <form class="pure-form " action="" method="POST">
+            <form class="pure-form "  action="{{url_for('ui.ui_preview.preview_page', uuid=uuid)}}" method="POST">
                 <fieldset>
                     <label for="preview-version">{{ _('Select timestamp') }}</label> <select id="preview-version"
                                                                                  name="from_version"
@@ -28,6 +28,7 @@
                         </option>
                     {% endfor %}
                 </select>
+                    <input type="hidden" name="csrf_token" value="{{ csrf_token() }}">
                     <button type="submit" class="pure-button pure-button-primary">{{ _('Go') }}</button>
 
                 </fieldset>

changedetectionio/blueprint/watchlist/__init__.py

@@ -81,6 +81,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
 
         sorted_tags = sorted(datastore.data['settings']['application'].get('tags').items(), key=lambda x: x[1]['title'])
 
+        proxy_list = datastore.proxy_list
         output = render_template(
             "watch-overview.html",
             active_tag=active_tag,
@@ -92,7 +93,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
             form=form,
             generate_tag_colors=processors.generate_processor_badge_colors,
             guid=datastore.data['app_guid'],
-            has_proxies=datastore.proxy_list,
+            has_proxies=proxy_list,
             hosted_sticky=os.getenv("SALTED_PASS", False) == False,
             now_time_server=round(time.time()),
             pagination=pagination,
@@ -104,12 +105,22 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
             search_q=request.args.get('q', '').strip(),
             sort_attribute=request.args.get('sort') if request.args.get('sort') else request.cookies.get('sort'),
             sort_order=request.args.get('order') if request.args.get('order') else request.cookies.get('order'),
-            system_default_fetcher=datastore.data['settings']['application'].get('fetch_backend'),
+            system_default_fetcher=datastore.data['settings']['application'].get('browser_profile'),
             tags=sorted_tags,
             unread_changes_count=datastore.unread_changes_count,
             watches=sorted_watches
         )
 
+        # Return freed template-building memory to the OS immediately.
+        # render_template allocates ~20MB of intermediate strings that are freed on return,
+        # but glibc keeps those pages mapped in its arenas as RSS. malloc_trim() forces
+        # glibc to release them, preventing RSS growth from concurrent Chrome connections.
+        try:
+            import ctypes
+            ctypes.CDLL('libc.so.6').malloc_trim(0)
+        except Exception:
+            pass
+
         if session.get('share-link'):
             del (session['share-link'])
 

changedetectionio/blueprint/watchlist/templates/watch-overview.html

@@ -213,12 +213,13 @@ html[data-darkmode="true"] .watch-tag-list.tag-{{ class_name }} {
                 {%- set checking_now = is_checking_now(watch) -%}
                 {%- set history_n = watch.history_n -%}
                 {%- set favicon = watch.get_favicon_filename() -%}
+                {%- set error_texts = watch.compile_error_texts(has_proxies=has_proxies) -%}
                 {%- set system_use_url_watchlist = datastore.data['settings']['application']['ui'].get('use_page_title_in_list')  -%}
                 {#  Class settings mirrored in changedetectionio/static/js/realtime.js for the frontend #}
                 {%- set row_classes = [
                     loop.cycle('pure-table-odd', 'pure-table-even'),
                     'processor-' ~ watch['processor'],
-                    'has-error' if watch.compile_error_texts()|length > 2 else '',
+                    'has-error' if error_texts|length > 2 else '',
                     'paused' if watch.paused is defined and watch.paused != False else '',
                     'unviewed' if watch.has_unviewed else '',
                     'has-restock-info' if watch.has_restock_info else 'no-restock-info',
@@ -271,7 +272,7 @@ html[data-darkmode="true"] .watch-tag-list.tag-{{ class_name }} {
                                 {% endif %}
                                <a class="external" target="_blank" rel="noopener" href="{{ watch.link.replace('source:','') }}">&nbsp;</a>
                             </span>
-                            <div class="error-text" style="display:none;">{{ watch.compile_error_texts(has_proxies=datastore.proxy_list)|safe }}</div>
+                            <div class="error-text" style="display:none;">{{ error_texts|safe }}</div>
                             {%- if watch['processor'] == 'text_json_diff'  -%}
                                 {%- if watch['has_ldjson_price_data'] and not watch['track_ldjson_price_data']  -%}
                                 <div class="ldjson-price-track-offer">Switch to Restock & Price watch mode? <a href="{{url_for('price_data_follower.accept', uuid=watch.uuid)}}" class="pure-button button-xsmall">Yes</a> <a href="{{url_for('price_data_follower.reject', uuid=watch.uuid)}}" class="">No</a></div>
@@ -284,10 +285,7 @@ html[data-darkmode="true"] .watch-tag-list.tag-{{ class_name }} {
                         </div>
                     <div class="status-icons">
                             <a class="link-spread" href="{{url_for('ui.form_share_put_watch', uuid=watch.uuid)}}"><img src="{{url_for('static_content', group='images', filename='spread.svg')}}" class="status-icon icon icon-spread" title="Create a link to share watch config with others" ></a>
-                            {%- set effective_fetcher = watch.get_fetch_backend if watch.get_fetch_backend != "system" else system_default_fetcher -%}
-                            {%- if effective_fetcher and ("html_webdriver" in effective_fetcher or "html_" in effective_fetcher or "extra_browser_" in effective_fetcher) -%}
-                                {{ effective_fetcher|fetcher_status_icons }}
-                            {%- endif -%}
+                            {{ watch.effective_browser_profile.get_fetcher_class_name()|fetcher_status_icons }}
                             {%- if watch.is_pdf  -%}<img class="status-icon" src="{{url_for('static_content', group='images', filename='pdf-icon.svg')}}" alt="Converting PDF to text" >{%- endif -%}
                             {%- if watch.has_browser_steps -%}<img class="status-icon status-browsersteps" src="{{url_for('static_content', group='images', filename='steps.svg')}}" alt="Browser Steps is enabled" >{%- endif -%}
 
@@ -305,12 +303,20 @@ html[data-darkmode="true"] .watch-tag-list.tag-{{ class_name }} {
                         {%- endif -%}
 
                         {%- if watch.get('restock') and watch['restock'].get('price') -%}
-                                {%- if watch['restock']['price'] is number -%}
-                                <span class="restock-label price" title="{{ _('Price') }}">
-                                {{ watch['restock']['price']|format_number_locale if watch['restock'].get('price') else '' }} {{ watch['restock'].get('currency','') }}
-                                </span>
-                                {%- else -%} <!-- watch['restock']['price']' is not a number, cant output it -->
+                            {%- set restock = watch['restock'] -%}
+                            {%- set price = restock.get('price') -%}
+                            {%- set cur = restock.get('currency','') -%}
+
+                            {%- if price is not none and (price|string)|regex_search('\d') -%}
+                              <span class="restock-label price" title="{{ _('Price') }}">
+                              {# @todo: make parse_currency/parse_decimal aware of the locale of the actual web page and use that instead changedetectionio/processors/restock_diff/__init__.py #}
+                                {%- if price is number -%}{# It's a number so we can convert it to their locale' #}
+                                  {{ price|format_number_locale }} {{ cur }}<!-- as number -->
+                                {%- else -%}{# It's totally fine if it arrives as something else, the website might be something weird in this field #}
+                                  {{ price }} {{ cur }}<!-- as string -->
                                 {%- endif -%}
+                              </span>
+                            {%- endif -%}
                         {%- elif not watch.has_restock_info -%}
                             <span class="restock-label error">{{ _('No information') }}</span>
                         {%- endif -%}

changedetectionio/content_fetchers/__init__.py

@@ -1,5 +1,4 @@
 import sys
-from changedetectionio.strtobool import strtobool
 from loguru import logger
 from changedetectionio.content_fetchers.exceptions import BrowserStepsStepException
 import os
@@ -25,87 +24,71 @@ SCREENSHOT_MAX_TOTAL_HEIGHT = int(os.getenv("SCREENSHOT_MAX_HEIGHT", SCREENSHOT_
 # Most modern GPUs support 16384x16384 textures, so 1280x10000 is safe
 SCREENSHOT_SIZE_STITCH_THRESHOLD = int(os.getenv("SCREENSHOT_CHUNK_HEIGHT", 10000))
 
-# available_fetchers() will scan this implementation looking for anything starting with html_
-# this information is used in the form selections
-from changedetectionio.content_fetchers.requests import fetcher as html_requests
-
-
 import importlib.resources
 XPATH_ELEMENT_JS = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('xpath_element_scraper.js').read_text(encoding='utf-8')
 INSTOCK_DATA_JS = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('stock-not-in-stock.js').read_text(encoding='utf-8')
 FAVICON_FETCHER_JS = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('favicon-fetcher.js').read_text(encoding='utf-8')
 
 
+# Registry: clean fetcher name → fetcher class (e.g. 'requests', 'playwright', 'cloakbrowser')
+FETCHERS: dict = {}
+
+
+def register_fetcher(name: str, cls) -> None:
+    """Register a fetcher class under its clean name (no html_ prefix)."""
+    FETCHERS[name] = cls
+
+
+def get_fetcher(name: str):
+    """Return the fetcher class for a clean name, or None."""
+    return FETCHERS.get(name)
+
+
 def available_fetchers():
-    # See the if statement at the bottom of this file for how we switch between playwright and webdriver
-    import inspect
-    p = []
-
-    # Get built-in fetchers (but skip plugin fetchers that were added via setattr)
-    for name, obj in inspect.getmembers(sys.modules[__name__], inspect.isclass):
-        if inspect.isclass(obj):
-            # @todo html_ is maybe better as fetcher_ or something
-            # In this case, make sure to edit the default one in store.py and fetch_site_status.py
-            if name.startswith('html_'):
-                # Skip plugin fetchers that were already registered
-                if name not in _plugin_fetchers:
-                    t = tuple([name, obj.fetcher_description])
-                    p.append(t)
-
-    # Get plugin fetchers from cache (already loaded at module init)
-    for name, fetcher_class in _plugin_fetchers.items():
-        if hasattr(fetcher_class, 'fetcher_description'):
-            t = tuple([name, fetcher_class.fetcher_description])
-            p.append(t)
-        else:
-            logger.warning(f"Plugin fetcher '{name}' does not have fetcher_description attribute")
-
-    return p
+    """Return list of (name, description) for all registered fetchers."""
+    return [(name, cls.fetcher_description) for name, cls in FETCHERS.items()
+            if hasattr(cls, 'fetcher_description')]
 
 
-def get_plugin_fetchers():
-    """Load and return all plugin fetchers from the centralized plugin manager."""
-    from changedetectionio.pluggy_interface import plugin_manager
+def available_browser_fetchers():
+    """Return list of (name, description) for fetchers that support screenshots (browser-type fetchers)."""
+    return [(name, cls.fetcher_description) for name, cls in FETCHERS.items()
+            if cls.supports_screenshots]
 
-    fetchers = {}
+
+def _load_fetchers():
+    """Load all fetchers (built-ins + plugins) into the FETCHERS registry."""
+    from changedetectionio.pluggy_interface import plugin_manager, register_builtin_fetchers
+
+    # Built-ins must be registered first
+    register_builtin_fetchers()
+
+    # Then external plugins
     try:
-        # Call the register_content_fetcher hook from all registered plugins
         results = plugin_manager.hook.register_content_fetcher()
         for result in results:
             if result:
                 name, fetcher_class = result
-                fetchers[name] = fetcher_class
-                # Register in current module so hasattr() checks work
-                setattr(sys.modules[__name__], name, fetcher_class)
-                logger.info(f"Registered plugin fetcher: {name} - {getattr(fetcher_class, 'fetcher_description', 'No description')}")
+                register_fetcher(name, fetcher_class)
+                logger.info(f"Registered fetcher: {name} - {getattr(fetcher_class, 'fetcher_description', '?')}")
     except Exception as e:
         logger.error(f"Error loading plugin fetchers: {e}")
 
-    return fetchers
 
 
-# Initialize plugins at module load time
-_plugin_fetchers = get_plugin_fetchers()
+# Default browser profiles always shown in the browser profiles table (keyed by machine name)
+DEFAULT_BROWSER_PROFILES: dict = {}
 
 
-# Decide which is the 'real' HTML webdriver, this is more a system wide config
-# rather than site-specific.
-use_playwright_as_chrome_fetcher = os.getenv('PLAYWRIGHT_DRIVER_URL', False)
-if use_playwright_as_chrome_fetcher:
-    # @note - For now, browser steps always uses playwright
-    if not strtobool(os.getenv('FAST_PUPPETEER_CHROME_FETCHER', 'False')):
-        logger.debug('Using Playwright library as fetcher')
-        from .playwright import fetcher as html_webdriver
-    else:
-        logger.debug('Using direct Python Puppeteer library as fetcher')
-        from .puppeteer import fetcher as html_webdriver
-
-else:
-    logger.debug("Falling back to selenium as fetcher")
-    from .webdriver_selenium import fetcher as html_webdriver
+def _register_default_browser_profiles():
+    """Register browser profiles that are always present in the profiles table."""
+    from changedetectionio.model.browser_profile import BUILTIN_REQUESTS
+    DEFAULT_BROWSER_PROFILES[BUILTIN_REQUESTS.get_machine_name()] = BUILTIN_REQUESTS
 
 
-# Register built-in fetchers as plugins after all imports are complete
-from changedetectionio.pluggy_interface import register_builtin_fetchers
-register_builtin_fetchers()
+# Populate the registry at module load time
+_load_fetchers()
+
+
+_register_default_browser_profiles()
 

changedetectionio/content_fetchers/base.py

@@ -70,37 +70,41 @@ class Fetcher():
     supports_screenshots = False        # Can capture page screenshots
     supports_xpath_element_data = False # Can extract xpath element positions/data for visual selector
 
+    # Icon shown in the watch list when this fetcher is the effective fetcher.
+    # Set to a dict with 'filename', 'alt', 'title' keys (image served from static/images/).
+    # None means no icon is shown (e.g. plain HTTP requests fetcher).
+    status_icon = None
+
     # Screenshot element locking - prevents layout shifts during screenshot capture
     # Only needed for visual comparison (image_ssim_diff processor)
     # Locks element dimensions in the first viewport to prevent headers/ads from resizing
     lock_viewport_elements = False      # Default: disabled for performance
 
+    # BrowserProfile-derived settings — applied by browser fetchers, ignored by html_requests
+    viewport_width: int = 1280
+    viewport_height: int = 1000
+    block_images: bool = False
+    block_fonts: bool = False
+    profile_user_agent: str = None   # Profile-level UA; lower priority than request_headers User-Agent
+    ignore_https_errors: bool = False
+    locale: str = None
+    service_workers: str = 'allow'
+    extra_delay: int = 0
+
     def __init__(self, **kwargs):
         if kwargs and 'screenshot_format' in kwargs:
             self.screenshot_format = kwargs.get('screenshot_format')
 
-        # Allow lock_viewport_elements to be set via kwargs
         if kwargs and 'lock_viewport_elements' in kwargs:
             self.lock_viewport_elements = kwargs.get('lock_viewport_elements')
 
+        # BrowserProfile fields — store whatever was passed, subclasses use them
+        for field in ('viewport_width', 'viewport_height', 'block_images', 'block_fonts',
+                      'profile_user_agent', 'ignore_https_errors', 'locale',
+                      'service_workers', 'extra_delay'):
+            if field in kwargs:
+                setattr(self, field, kwargs[field])
 
-    @classmethod
-    def get_status_icon_data(cls):
-        """Return data for status icon to display in the watch overview.
-
-        This method can be overridden by subclasses to provide custom status icons.
-
-        Returns:
-            dict or None: Dictionary with icon data:
-                {
-                    'filename': 'icon-name.svg',  # Icon filename
-                    'alt': 'Alt text',            # Alt attribute
-                    'title': 'Tooltip text',      # Title attribute
-                    'style': 'height: 1em;'       # Optional inline CSS
-                }
-                Or None if no icon
-        """
-        return None
 
     def clear_content(self):
         """
@@ -198,6 +202,16 @@ class Fetcher():
                     # Stop processing here
                     raise BrowserStepsStepException(step_n=step_n, original_e=e)
 
+    def disk_cleanup_after_fetch(self):
+        """Remove any temporary files written to disk during a fetch.
+
+        The default implementation is a no-op.  Browser-based fetchers
+        override this to delete browser-step screenshots and any other
+        ephemeral files they create.  Called by the processor after
+        ``quit()`` regardless of whether the fetch succeeded or failed.
+        """
+        pass
+
     # It's always good to reset these
     def delete_browser_steps_screenshots(self):
         import glob

changedetectionio/content_fetchers/playwright.py

@@ -1,471 +0,0 @@
-import asyncio
-import gc
-import json
-import os
-from urllib.parse import urlparse
-
-from loguru import logger
-
-from changedetectionio.content_fetchers import SCREENSHOT_MAX_HEIGHT_DEFAULT, visualselector_xpath_selectors, \
-    SCREENSHOT_SIZE_STITCH_THRESHOLD, SCREENSHOT_MAX_TOTAL_HEIGHT, XPATH_ELEMENT_JS, INSTOCK_DATA_JS, FAVICON_FETCHER_JS
-from changedetectionio.content_fetchers.base import Fetcher, manage_user_agent
-from changedetectionio.content_fetchers.exceptions import PageUnloadable, Non200ErrorCodeReceived, EmptyReply, ScreenshotUnavailable, \
-    BrowserStepsStepException
-
-
-async def capture_full_page_async(page, screenshot_format='JPEG', watch_uuid=None, lock_viewport_elements=False):
-    import os
-    import time
-
-    start = time.time()
-    watch_info = f"[{watch_uuid}] " if watch_uuid else ""
-
-    setup_start = time.time()
-    page_height = await page.evaluate("document.documentElement.scrollHeight")
-    page_width = await page.evaluate("document.documentElement.scrollWidth")
-    original_viewport = page.viewport_size
-    dimensions_time = time.time() - setup_start
-
-    logger.debug(f"{watch_info}Playwright viewport size {page.viewport_size} page height {page_height} page width {page_width} (got dimensions in {dimensions_time:.2f}s)")
-
-    # Use an approach similar to puppeteer: set a larger viewport and take screenshots in chunks
-    step_size = SCREENSHOT_SIZE_STITCH_THRESHOLD # Size that won't cause GPU to overflow
-    screenshot_chunks = []
-    y = 0
-    elements_locked = False
-
-    # Only lock viewport elements if explicitly enabled (for image_ssim_diff processor)
-    # This prevents headers/ads from resizing when viewport changes
-    if lock_viewport_elements and page_height > page.viewport_size['height']:
-        lock_start = time.time()
-        lock_elements_js_path = os.path.join(os.path.dirname(__file__), 'res', 'lock-elements-sizing.js')
-        with open(lock_elements_js_path, 'r') as f:
-            lock_elements_js = f.read()
-        await page.evaluate(lock_elements_js)
-        elements_locked = True
-        lock_time = time.time() - lock_start
-        logger.debug(f"{watch_info}Viewport element locking enabled (took {lock_time:.2f}s)")
-
-    if page_height > page.viewport_size['height']:
-        if page_height < step_size:
-            step_size = page_height # Incase page is bigger than default viewport but smaller than proposed step size
-        viewport_start = time.time()
-        logger.debug(f"{watch_info}Setting bigger viewport to step through large page width W{page.viewport_size['width']}xH{step_size} because page_height > viewport_size")
-        # Set viewport to a larger size to capture more content at once
-        await page.set_viewport_size({'width': page.viewport_size['width'], 'height': step_size})
-        viewport_time = time.time() - viewport_start
-        logger.debug(f"{watch_info}Viewport changed to {page.viewport_size['width']}x{step_size} (took {viewport_time:.2f}s)")
-
-    # Capture screenshots in chunks up to the max total height
-    capture_start = time.time()
-    chunk_times = []
-    # Use PNG for better quality (no compression artifacts), JPEG for smaller size
-    screenshot_type = screenshot_format.lower() if screenshot_format else 'jpeg'
-    # PNG should use quality 100, JPEG uses configurable quality
-    screenshot_quality = 100 if screenshot_type == 'png' else int(os.getenv("SCREENSHOT_QUALITY", 72))
-
-    while y < min(page_height, SCREENSHOT_MAX_TOTAL_HEIGHT):
-        # Only scroll if not at the top (y > 0)
-        if y > 0:
-            await page.evaluate(f"window.scrollTo(0, {y})")
-
-        # Request GC only before screenshot (not 3x per chunk)
-        await page.request_gc()
-
-        screenshot_kwargs = {
-            'type': screenshot_type,
-            'full_page': False
-        }
-        # Only pass quality parameter for jpeg (PNG doesn't support it in Playwright)
-        if screenshot_type == 'jpeg':
-            screenshot_kwargs['quality'] = screenshot_quality
-
-        chunk_start = time.time()
-        screenshot_chunks.append(await page.screenshot(**screenshot_kwargs))
-        chunk_time = time.time() - chunk_start
-        chunk_times.append(chunk_time)
-        logger.debug(f"{watch_info}Chunk {len(screenshot_chunks)} captured in {chunk_time:.2f}s")
-        y += step_size
-
-    # Restore original viewport size
-    await page.set_viewport_size({'width': original_viewport['width'], 'height': original_viewport['height']})
-
-    # Unlock element dimensions if they were locked
-    if elements_locked:
-        unlock_elements_js_path = os.path.join(os.path.dirname(__file__), 'res', 'unlock-elements-sizing.js')
-        with open(unlock_elements_js_path, 'r') as f:
-            unlock_elements_js = f.read()
-        await page.evaluate(unlock_elements_js)
-        logger.debug(f"{watch_info}Element dimensions unlocked after screenshot capture")
-
-    capture_time = time.time() - capture_start
-    total_capture_time = sum(chunk_times)
-    logger.debug(f"{watch_info}All {len(screenshot_chunks)} chunks captured in {capture_time:.2f}s (total chunk time: {total_capture_time:.2f}s)")
-
-    # If we have multiple chunks, stitch them together
-    if len(screenshot_chunks) > 1:
-        stitch_start = time.time()
-        logger.debug(f"{watch_info}Starting stitching of {len(screenshot_chunks)} chunks")
-
-        # Always use spawn subprocess for ANY stitching (2+ chunks)
-        # PIL allocates at C level and Python GC never releases it - subprocess exit forces OS to reclaim
-        # Trade-off: 35MB resource_tracker vs 500MB+ PIL leak in main process
-        from changedetectionio.content_fetchers.screenshot_handler import stitch_images_worker_raw_bytes
-        import multiprocessing
-        import struct
-
-        ctx = multiprocessing.get_context('spawn')
-        parent_conn, child_conn = ctx.Pipe()
-        p = ctx.Process(target=stitch_images_worker_raw_bytes, args=(child_conn, page_height, SCREENSHOT_MAX_TOTAL_HEIGHT))
-        p.start()
-
-        # Send via raw bytes (no pickle)
-        parent_conn.send_bytes(struct.pack('I', len(screenshot_chunks)))
-        for chunk in screenshot_chunks:
-            parent_conn.send_bytes(chunk)
-
-        screenshot = parent_conn.recv_bytes()
-        p.join()
-
-        parent_conn.close()
-        child_conn.close()
-        del p, parent_conn, child_conn
-
-        stitch_time = time.time() - stitch_start
-        total_time = time.time() - start
-        setup_time = total_time - capture_time - stitch_time
-        logger.debug(
-            f"{watch_info}Screenshot complete - Page height: {page_height}px, Capture height: {SCREENSHOT_MAX_TOTAL_HEIGHT}px | "
-            f"Setup: {setup_time:.2f}s, Capture: {capture_time:.2f}s, Stitching: {stitch_time:.2f}s, Total: {total_time:.2f}s")
-        return screenshot
-
-    total_time = time.time() - start
-    setup_time = total_time - capture_time
-    logger.debug(
-        f"{watch_info}Screenshot complete - Page height: {page_height}px, Capture height: {SCREENSHOT_MAX_TOTAL_HEIGHT}px | "
-        f"Setup: {setup_time:.2f}s, Single chunk: {capture_time:.2f}s, Total: {total_time:.2f}s")
-
-    return screenshot_chunks[0]
-
-class fetcher(Fetcher):
-    fetcher_description = "Playwright {}/Javascript".format(
-        os.getenv("PLAYWRIGHT_BROWSER_TYPE", 'chromium').capitalize()
-    )
-    if os.getenv("PLAYWRIGHT_DRIVER_URL"):
-        fetcher_description += " via '{}'".format(os.getenv("PLAYWRIGHT_DRIVER_URL"))
-
-    browser_type = ''
-    command_executor = ''
-
-    # Configs for Proxy setup
-    # In the ENV vars, is prefixed with "playwright_proxy_", so it is for example "playwright_proxy_server"
-    playwright_proxy_settings_mappings = ['bypass', 'server', 'username', 'password']
-
-    proxy = None
-
-    # Capability flags
-    supports_browser_steps = True
-    supports_screenshots = True
-    supports_xpath_element_data = True
-
-    @classmethod
-    def get_status_icon_data(cls):
-        """Return Chrome browser icon data for Playwright fetcher."""
-        return {
-            'filename': 'google-chrome-icon.png',
-            'alt': 'Using a Chrome browser',
-            'title': 'Using a Chrome browser'
-        }
-
-    def __init__(self, proxy_override=None, custom_browser_connection_url=None, **kwargs):
-        super().__init__(**kwargs)
-
-        self.browser_type = os.getenv("PLAYWRIGHT_BROWSER_TYPE", 'chromium').strip('"')
-
-        if custom_browser_connection_url:
-            self.browser_connection_is_custom = True
-            self.browser_connection_url = custom_browser_connection_url
-        else:
-            # Fallback to fetching from system
-            # .strip('"') is going to save someone a lot of time when they accidently wrap the env value
-            self.browser_connection_url = os.getenv("PLAYWRIGHT_DRIVER_URL", 'ws://playwright-chrome:3000').strip('"')
-
-        # If any proxy settings are enabled, then we should setup the proxy object
-        proxy_args = {}
-        for k in self.playwright_proxy_settings_mappings:
-            v = os.getenv('playwright_proxy_' + k, False)
-            if v:
-                proxy_args[k] = v.strip('"')
-
-        if proxy_args:
-            self.proxy = proxy_args
-
-        # allow per-watch proxy selection override
-        if proxy_override:
-            self.proxy = {'server': proxy_override}
-
-        if self.proxy:
-            # Playwright needs separate username and password values
-            parsed = urlparse(self.proxy.get('server'))
-            if parsed.username:
-                self.proxy['username'] = parsed.username
-                self.proxy['password'] = parsed.password
-
-    async def screenshot_step(self, step_n=''):
-        super().screenshot_step(step_n=step_n)
-        watch_uuid = getattr(self, 'watch_uuid', None)
-        screenshot = await capture_full_page_async(page=self.page, screenshot_format=self.screenshot_format, watch_uuid=watch_uuid, lock_viewport_elements=self.lock_viewport_elements)
-
-        # Request GC immediately after screenshot to free memory
-        # Screenshots can be large and browser steps take many of them
-        await self.page.request_gc()
-
-        if self.browser_steps_screenshot_path is not None:
-            destination = os.path.join(self.browser_steps_screenshot_path, 'step_{}.jpeg'.format(step_n))
-            logger.debug(f"Saving step screenshot to {destination}")
-            with open(destination, 'wb') as f:
-                f.write(screenshot)
-            # Clear local reference to allow screenshot bytes to be collected
-            del screenshot
-            gc.collect()
-
-    async def save_step_html(self, step_n):
-        super().save_step_html(step_n=step_n)
-        content = await self.page.content()
-
-        # Request GC after getting page content
-        await self.page.request_gc()
-
-        destination = os.path.join(self.browser_steps_screenshot_path, 'step_{}.html'.format(step_n))
-        logger.debug(f"Saving step HTML to {destination}")
-        with open(destination, 'w', encoding='utf-8') as f:
-            f.write(content)
-        # Clear local reference
-        del content
-        gc.collect()
-
-    async def run(self,
-                  fetch_favicon=True,
-                  current_include_filters=None,
-                  empty_pages_are_a_change=False,
-                  ignore_status_codes=False,
-                  is_binary=False,
-                  request_body=None,
-                  request_headers=None,
-                  request_method=None,
-                  screenshot_format=None,
-                  timeout=None,
-                  url=None,
-                  watch_uuid=None,
-                  ):
-
-        from playwright.async_api import async_playwright
-        import playwright._impl._errors
-        import time
-        self.delete_browser_steps_screenshots()
-        self.watch_uuid = watch_uuid  # Store for use in screenshot_step
-        response = None
-
-        async with async_playwright() as p:
-            browser_type = getattr(p, self.browser_type)
-
-            # Seemed to cause a connection Exception even tho I can see it connect
-            # self.browser = browser_type.connect(self.command_executor, timeout=timeout*1000)
-            # 60,000 connection timeout only
-            browser = await browser_type.connect_over_cdp(self.browser_connection_url, timeout=60000)
-
-            # SOCKS5 with authentication is not supported (yet)
-            # https://github.com/microsoft/playwright/issues/10567
-
-            # Set user agent to prevent Cloudflare from blocking the browser
-            # Use the default one configured in the App.py model that's passed from fetch_site_status.py
-            context = await browser.new_context(
-                accept_downloads=False,  # Should never be needed
-                bypass_csp=True,  # This is needed to enable JavaScript execution on GitHub and others
-                extra_http_headers=request_headers,
-                ignore_https_errors=True,
-                proxy=self.proxy,
-                service_workers=os.getenv('PLAYWRIGHT_SERVICE_WORKERS', 'allow'), # Should be `allow` or `block` - sites like YouTube can transmit large amounts of data via Service Workers
-                user_agent=manage_user_agent(headers=request_headers),
-            )
-
-            self.page = await context.new_page()
-
-            # Listen for all console events and handle errors
-            self.page.on("console", lambda msg: logger.debug(f"Playwright console: Watch URL: {url} {msg.type}: {msg.text} {msg.args}"))
-
-            # Re-use as much code from browser steps as possible so its the same
-            from changedetectionio.browser_steps.browser_steps import steppable_browser_interface
-            browsersteps_interface = steppable_browser_interface(start_url=url)
-            browsersteps_interface.page = self.page
-
-            response = await browsersteps_interface.action_goto_url(value=url)
-
-            if response is None:
-                await context.close()
-                await browser.close()
-                logger.debug("Content Fetcher > Response object from the browser communication was none")
-                raise EmptyReply(url=url, status_code=None)
-
-            # In async_playwright, all_headers() returns a coroutine
-            try:
-                self.headers = await response.all_headers()
-            except TypeError:
-                # Fallback for sync version
-                self.headers = response.all_headers()
-
-            try:
-                if self.webdriver_js_execute_code is not None and len(self.webdriver_js_execute_code):
-                    await browsersteps_interface.action_execute_js(value=self.webdriver_js_execute_code, selector=None)
-            except playwright._impl._errors.TimeoutError as e:
-                await context.close()
-                await browser.close()
-                # This can be ok, we will try to grab what we could retrieve
-                pass
-            except Exception as e:
-                logger.debug(f"Content Fetcher > Other exception when executing custom JS code {str(e)}")
-                await context.close()
-                await browser.close()
-                raise PageUnloadable(url=url, status_code=None, message=str(e))
-
-            extra_wait = int(os.getenv("WEBDRIVER_DELAY_BEFORE_CONTENT_READY", 5)) + self.render_extract_delay
-            await self.page.wait_for_timeout(extra_wait * 1000)
-
-            try:
-                self.status_code = response.status
-            except Exception as e:
-                # https://github.com/dgtlmoon/changedetection.io/discussions/2122#discussioncomment-8241962
-                logger.critical(f"Response from the browser/Playwright did not have a status_code! Response follows.")
-                logger.critical(response)
-                await context.close()
-                await browser.close()
-                raise PageUnloadable(url=url, status_code=None, message=str(e))
-
-            if fetch_favicon:
-                try:
-                    self.favicon_blob = await self.page.evaluate(FAVICON_FETCHER_JS)
-                    await self.page.request_gc()
-                except Exception as e:
-                    logger.error(f"Error fetching FavIcon info {str(e)}, continuing.")
-
-            if self.status_code != 200 and not ignore_status_codes:
-                screenshot = await capture_full_page_async(self.page, screenshot_format=self.screenshot_format, watch_uuid=watch_uuid, lock_viewport_elements=self.lock_viewport_elements)
-                # Finally block will handle cleanup
-                raise Non200ErrorCodeReceived(url=url, status_code=self.status_code, screenshot=screenshot)
-
-            if not empty_pages_are_a_change and len((await self.page.content()).strip()) == 0:
-                logger.debug("Content Fetcher > Content was empty, empty_pages_are_a_change = False")
-                await context.close()
-                await browser.close()
-                raise EmptyReply(url=url, status_code=response.status)
-
-            # Wrap remaining operations in try/finally to ensure cleanup
-            try:
-                # Run Browser Steps here
-                if self.browser_steps:
-                    try:
-                        await self.iterate_browser_steps(start_url=url)
-                    except BrowserStepsStepException:
-                        # Finally block will handle cleanup
-                        raise
-
-                    await self.page.wait_for_timeout(extra_wait * 1000)
-
-                now = time.time()
-                # So we can find an element on the page where its selector was entered manually (maybe not xPath etc)
-                if current_include_filters is not None:
-                    await self.page.evaluate("var include_filters={}".format(json.dumps(current_include_filters)))
-                else:
-                    await self.page.evaluate("var include_filters=''")
-                await self.page.request_gc()
-
-                # request_gc before and after evaluate to free up memory
-                # @todo browsersteps etc
-                MAX_TOTAL_HEIGHT = int(os.getenv("SCREENSHOT_MAX_HEIGHT", SCREENSHOT_MAX_HEIGHT_DEFAULT))
-                self.xpath_data = await self.page.evaluate(XPATH_ELEMENT_JS, {
-                    "visualselector_xpath_selectors": visualselector_xpath_selectors,
-                    "max_height": MAX_TOTAL_HEIGHT
-                })
-                await self.page.request_gc()
-
-                self.instock_data = await self.page.evaluate(INSTOCK_DATA_JS)
-                await self.page.request_gc()
-
-                self.content = await self.page.content()
-                await self.page.request_gc()
-                logger.debug(f"Scrape xPath element data in browser done in {time.time() - now:.2f}s")
-
-
-                # Bug 3 in Playwright screenshot handling
-                # Some bug where it gives the wrong screenshot size, but making a request with the clip set first seems to solve it
-                # JPEG is better here because the screenshots can be very very large
-
-                # Screenshots also travel via the ws:// (websocket) meaning that the binary data is base64 encoded
-                # which will significantly increase the IO size between the server and client, it's recommended to use the lowest
-                # acceptable screenshot quality here
-                # The actual screenshot - this always base64 and needs decoding! horrible! huge CPU usage
-                self.screenshot = await capture_full_page_async(page=self.page, screenshot_format=self.screenshot_format, watch_uuid=watch_uuid, lock_viewport_elements=self.lock_viewport_elements)
-
-                # Force aggressive memory cleanup - screenshots are large and base64 decode creates temporary buffers
-                await self.page.request_gc()
-                gc.collect()
-
-            except ScreenshotUnavailable:
-                # Re-raise screenshot unavailable exceptions
-                raise ScreenshotUnavailable(url=url, status_code=self.status_code)
-
-            finally:
-                # Clean up resources properly with timeouts to prevent hanging
-                try:
-                    if hasattr(self, 'page') and self.page:
-                        await self.page.request_gc()
-                        await asyncio.wait_for(self.page.close(), timeout=5.0)
-                        logger.debug(f"Successfully closed page for {url}")
-                except asyncio.TimeoutError:
-                    logger.warning(f"Timed out closing page for {url} (5s)")
-                except Exception as e:
-                    logger.warning(f"Error closing page for {url}: {e}")
-                finally:
-                    self.page = None
-
-                try:
-                    if context:
-                        await asyncio.wait_for(context.close(), timeout=5.0)
-                        logger.debug(f"Successfully closed context for {url}")
-                except asyncio.TimeoutError:
-                    logger.warning(f"Timed out closing context for {url} (5s)")
-                except Exception as e:
-                    logger.warning(f"Error closing context for {url}: {e}")
-                finally:
-                    context = None
-
-                try:
-                    if browser:
-                        await asyncio.wait_for(browser.close(), timeout=5.0)
-                        logger.debug(f"Successfully closed browser connection for {url}")
-                except asyncio.TimeoutError:
-                    logger.warning(f"Timed out closing browser connection for {url} (5s)")
-                except Exception as e:
-                    logger.warning(f"Error closing browser for {url}: {e}")
-                finally:
-                    browser = None
-
-                # Force Python GC to release Playwright resources immediately
-                # Playwright objects can have circular references that delay cleanup
-                gc.collect()
-
-
-# Plugin registration for built-in fetcher
-class PlaywrightFetcherPlugin:
-    """Plugin class that registers the Playwright fetcher as a built-in plugin."""
-
-    def register_content_fetcher(self):
-        """Register the Playwright fetcher"""
-        return ('html_webdriver', fetcher)
-
-
-# Create module-level instance for plugin registration
-playwright_plugin = PlaywrightFetcherPlugin()
-
-
-

changedetectionio/content_fetchers/playwright/CDP.py

@@ -0,0 +1,41 @@
+"""
+Playwright CDP fetcher — connects to a remote browser via Chrome DevTools Protocol.
+
+browser_connection_url must be supplied via the resolved BrowserProfile
+(set by preconfigure_browser_profiles_based_on_env at startup or edited in the UI).
+"""
+from loguru import logger
+from changedetectionio.pluggy_interface import hookimpl
+from changedetectionio.content_fetchers.playwright import PlaywrightBaseFetcher
+
+
+class fetcher(PlaywrightBaseFetcher):
+    fetcher_description = "Playwright Chrome (CDP/Remote)"
+    requires_connection_url = True
+
+    def __init__(self, proxy_override=None, custom_browser_connection_url=None, **kwargs):
+        super().__init__(proxy_override=proxy_override, custom_browser_connection_url=custom_browser_connection_url, **kwargs)
+
+        if custom_browser_connection_url:
+            self.browser_connection_is_custom = True
+            self.browser_connection_url = custom_browser_connection_url
+        else:
+            logger.critical("Playwright CDP fetcher has no browser_connection_url — browser profile was not configured. "
+                            "Set PLAYWRIGHT_DRIVER_URL or configure a browser profile in Settings.")
+            self.browser_connection_url = None
+
+        # CDP always connects to Chromium
+        self.browser_type = 'chromium'
+
+    async def _connect_browser(self, p):
+        browser_type = getattr(p, self.browser_type)
+        return await browser_type.connect_over_cdp(self.browser_connection_url, timeout=60_000)
+
+
+class PlaywrightCDPPlugin:
+    @hookimpl
+    def register_content_fetcher(self):
+        return ('playwright_cdp', fetcher)
+
+
+cdp_plugin = PlaywrightCDPPlugin()

changedetectionio/content_fetchers/playwright/__init__.py

@@ -0,0 +1,403 @@
+"""
+Playwright-based content fetchers.
+
+Submodules:
+  cdp     — connect to a remote browser via Chrome DevTools Protocol (CDP/WebSocket)
+  chrome  — launch a local Chromium browser
+  firefox — launch a local Firefox browser
+  webkit  — launch a local WebKit (Safari-engine) browser
+"""
+
+import asyncio
+import gc
+import json
+import os
+import re
+from urllib.parse import urlparse
+
+from loguru import logger
+
+from changedetectionio.content_fetchers import (
+    SCREENSHOT_MAX_HEIGHT_DEFAULT,
+    SCREENSHOT_MAX_TOTAL_HEIGHT,
+    SCREENSHOT_SIZE_STITCH_THRESHOLD,
+    FAVICON_FETCHER_JS,
+    INSTOCK_DATA_JS,
+    XPATH_ELEMENT_JS,
+    visualselector_xpath_selectors,
+)
+from changedetectionio.content_fetchers.base import Fetcher, manage_user_agent
+from changedetectionio.content_fetchers.exceptions import (
+    BrowserStepsStepException,
+    EmptyReply,
+    Non200ErrorCodeReceived,
+    PageUnloadable,
+    ScreenshotUnavailable,
+)
+
+
+async def capture_full_page_async(page, screenshot_format='JPEG', watch_uuid=None, lock_viewport_elements=False):
+    import time
+
+    start = time.time()
+    watch_info = f"[{watch_uuid}] " if watch_uuid else ""
+
+    setup_start = time.time()
+    page_height = await page.evaluate("document.documentElement.scrollHeight")
+    page_width = await page.evaluate("document.documentElement.scrollWidth")
+    original_viewport = page.viewport_size
+    dimensions_time = time.time() - setup_start
+
+    logger.debug(f"{watch_info}Playwright viewport size {page.viewport_size} page height {page_height} page width {page_width} (got dimensions in {dimensions_time:.2f}s)")
+
+    step_size = SCREENSHOT_SIZE_STITCH_THRESHOLD
+    screenshot_chunks = []
+    y = 0
+    elements_locked = False
+
+    if lock_viewport_elements and page_height > page.viewport_size['height']:
+        lock_start = time.time()
+        lock_elements_js_path = os.path.join(os.path.dirname(__file__), '..', 'res', 'lock-elements-sizing.js')
+        with open(lock_elements_js_path, 'r') as f:
+            lock_elements_js = f.read()
+        await page.evaluate(lock_elements_js)
+        elements_locked = True
+        logger.debug(f"{watch_info}Viewport element locking enabled (took {time.time() - lock_start:.2f}s)")
+
+    if page_height > page.viewport_size['height']:
+        if page_height < step_size:
+            step_size = page_height
+        await page.set_viewport_size({'width': page.viewport_size['width'], 'height': step_size})
+
+    capture_start = time.time()
+    chunk_times = []
+    screenshot_type = screenshot_format.lower() if screenshot_format else 'jpeg'
+    screenshot_quality = 100 if screenshot_type == 'png' else int(os.getenv("SCREENSHOT_QUALITY", 72))
+
+    while y < min(page_height, SCREENSHOT_MAX_TOTAL_HEIGHT):
+        if y > 0:
+            await page.evaluate(f"window.scrollTo(0, {y})")
+
+        await _safe_request_gc(page)
+
+        screenshot_kwargs = {'type': screenshot_type, 'full_page': False}
+        if screenshot_type == 'jpeg':
+            screenshot_kwargs['quality'] = screenshot_quality
+
+        chunk_start = time.time()
+        screenshot_chunks.append(await page.screenshot(**screenshot_kwargs))
+        chunk_time = time.time() - chunk_start
+        chunk_times.append(chunk_time)
+        logger.debug(f"{watch_info}Chunk {len(screenshot_chunks)} captured in {chunk_time:.2f}s")
+        y += step_size
+
+    await page.set_viewport_size({'width': original_viewport['width'], 'height': original_viewport['height']})
+
+    if elements_locked:
+        unlock_elements_js_path = os.path.join(os.path.dirname(__file__), '..', 'res', 'unlock-elements-sizing.js')
+        with open(unlock_elements_js_path, 'r') as f:
+            unlock_elements_js = f.read()
+        await page.evaluate(unlock_elements_js)
+
+    capture_time = time.time() - capture_start
+
+    if len(screenshot_chunks) > 1:
+        stitch_start = time.time()
+        from changedetectionio.content_fetchers.screenshot_handler import stitch_images_worker_raw_bytes
+        import multiprocessing
+        import struct
+
+        ctx = multiprocessing.get_context('spawn')
+        parent_conn, child_conn = ctx.Pipe()
+        p = ctx.Process(target=stitch_images_worker_raw_bytes, args=(child_conn, page_height, SCREENSHOT_MAX_TOTAL_HEIGHT))
+        p.start()
+
+        parent_conn.send_bytes(struct.pack('I', len(screenshot_chunks)))
+        for chunk in screenshot_chunks:
+            parent_conn.send_bytes(chunk)
+
+        screenshot = parent_conn.recv_bytes()
+        p.join()
+        parent_conn.close()
+        child_conn.close()
+        del p, parent_conn, child_conn
+
+        stitch_time = time.time() - stitch_start
+        total_time = time.time() - start
+        setup_time = total_time - capture_time - stitch_time
+        logger.debug(
+            f"{watch_info}Screenshot complete - Page height: {page_height}px | "
+            f"Setup: {setup_time:.2f}s, Capture: {capture_time:.2f}s, Stitching: {stitch_time:.2f}s, Total: {total_time:.2f}s")
+        return screenshot
+
+    total_time = time.time() - start
+    logger.debug(
+        f"{watch_info}Screenshot complete - Page height: {page_height}px | "
+        f"Setup: {total_time - capture_time:.2f}s, Single chunk: {capture_time:.2f}s, Total: {total_time:.2f}s")
+    return screenshot_chunks[0]
+
+
+async def _safe_request_gc(page):
+    """Request browser GC — Chromium-specific, silently ignored on Firefox/WebKit."""
+    try:
+        await page.request_gc()
+    except Exception:
+        pass
+
+
+class PlaywrightBaseFetcher(Fetcher):
+    """
+    Shared base for all Playwright fetchers.
+
+    Subclasses implement ``_connect_browser(playwright_instance)`` to return a
+    connected-or-launched browser object.  Everything else — context creation,
+    page interaction, screenshot capture, browser-steps execution — lives here.
+    """
+
+    playwright_proxy_settings_mappings = ['bypass', 'server', 'username', 'password']
+
+    proxy = None
+
+    # Capability flags
+    supports_browser_steps = True
+    supports_screenshots = True
+    supports_xpath_element_data = True
+
+    status_icon = {'filename': 'google-chrome-icon.png', 'alt': 'Using a Chrome browser', 'title': 'Using a Chrome browser'}
+
+    def __init__(self, proxy_override=None, custom_browser_connection_url=None, **kwargs):
+        super().__init__(**kwargs)
+
+        # Subclasses may use this (e.g. CDP); others ignore it
+        self._custom_browser_connection_url = custom_browser_connection_url
+
+        proxy_args = {}
+        for k in self.playwright_proxy_settings_mappings:
+            v = os.getenv('playwright_proxy_' + k, False)
+            if v:
+                proxy_args[k] = v.strip('"')
+
+        if proxy_args:
+            self.proxy = proxy_args
+
+        if proxy_override:
+            self.proxy = {'server': proxy_override}
+
+        if self.proxy:
+            parsed = urlparse(self.proxy.get('server', ''))
+            if parsed.username:
+                self.proxy['username'] = parsed.username
+                self.proxy['password'] = parsed.password
+
+    def disk_cleanup_after_fetch(self):
+        """Delete browser-step screenshots written during this fetch."""
+        self.delete_browser_steps_screenshots()
+
+    async def _connect_browser(self, playwright_instance):
+        """Return an open browser object.  Must be overridden by each subclass."""
+        raise NotImplementedError(f"{type(self).__name__} must implement _connect_browser()")
+
+    async def screenshot_step(self, step_n=''):
+        super().screenshot_step(step_n=step_n)
+        watch_uuid = getattr(self, 'watch_uuid', None)
+        screenshot = await capture_full_page_async(
+            page=self.page,
+            screenshot_format=self.screenshot_format,
+            watch_uuid=watch_uuid,
+            lock_viewport_elements=self.lock_viewport_elements,
+        )
+        await _safe_request_gc(self.page)
+
+        if self.browser_steps_screenshot_path is not None:
+            destination = os.path.join(self.browser_steps_screenshot_path, 'step_{}.jpeg'.format(step_n))
+            logger.debug(f"Saving step screenshot to {destination}")
+            with open(destination, 'wb') as f:
+                f.write(screenshot)
+            del screenshot
+            gc.collect()
+
+    async def save_step_html(self, step_n):
+        super().save_step_html(step_n=step_n)
+        content = await self.page.content()
+        await _safe_request_gc(self.page)
+
+        destination = os.path.join(self.browser_steps_screenshot_path, 'step_{}.html'.format(step_n))
+        logger.debug(f"Saving step HTML to {destination}")
+        with open(destination, 'w', encoding='utf-8') as f:
+            f.write(content)
+        del content
+        gc.collect()
+
+    async def run(self,
+                  fetch_favicon=True,
+                  current_include_filters=None,
+                  empty_pages_are_a_change=False,
+                  ignore_status_codes=False,
+                  is_binary=False,
+                  request_body=None,
+                  request_headers=None,
+                  request_method=None,
+                  screenshot_format=None,
+                  timeout=None,
+                  url=None,
+                  watch_uuid=None,
+                  ):
+        from playwright.async_api import async_playwright
+        import playwright._impl._errors
+        import time
+
+        self.delete_browser_steps_screenshots()
+        self.watch_uuid = watch_uuid
+        response = None
+
+        async with async_playwright() as p:
+            browser = await self._connect_browser(p)
+
+            ua = manage_user_agent(headers=request_headers) or self.profile_user_agent or None
+
+            context_kwargs = dict(
+                accept_downloads=False,
+                bypass_csp=True,
+                extra_http_headers=request_headers,
+                ignore_https_errors=self.ignore_https_errors,
+                proxy=self.proxy,
+                service_workers=self.service_workers,
+                user_agent=ua,
+                viewport={'width': self.viewport_width, 'height': self.viewport_height},
+            )
+            if self.locale:
+                context_kwargs['locale'] = self.locale
+
+            context = await browser.new_context(**context_kwargs)
+
+            if self.block_images:
+                await context.route(
+                    re.compile(r'\.(png|jpe?g|gif|svg|ico|webp|avif|bmp)(\?.*)?$', re.IGNORECASE),
+                    lambda route: route.abort()
+                )
+            if self.block_fonts:
+                await context.route(
+                    re.compile(r'\.(woff2?|ttf|otf|eot)(\?.*)?$', re.IGNORECASE),
+                    lambda route: route.abort()
+                )
+
+            self.page = await context.new_page()
+            self.page.on("console", lambda msg: logger.debug(f"Playwright console: {url} {msg.type}: {msg.text}"))
+
+            from changedetectionio.browser_steps.browser_steps import steppable_browser_interface
+            browsersteps_interface = steppable_browser_interface(start_url=url)
+            browsersteps_interface.page = self.page
+
+            response = await browsersteps_interface.action_goto_url(value=url)
+
+            if response is None:
+                await context.close()
+                await browser.close()
+                raise EmptyReply(url=url, status_code=None)
+
+            try:
+                self.headers = await response.all_headers()
+            except TypeError:
+                self.headers = response.all_headers()
+
+            try:
+                if self.webdriver_js_execute_code is not None and len(self.webdriver_js_execute_code):
+                    await browsersteps_interface.action_execute_js(value=self.webdriver_js_execute_code, selector=None)
+            except playwright._impl._errors.TimeoutError:
+                await context.close()
+                await browser.close()
+                pass
+            except Exception as e:
+                await context.close()
+                await browser.close()
+                raise PageUnloadable(url=url, status_code=None, message=str(e))
+
+            extra_wait = self.extra_delay + self.render_extract_delay
+            await self.page.wait_for_timeout(extra_wait * 1000)
+
+            try:
+                self.status_code = response.status
+            except Exception as e:
+                await context.close()
+                await browser.close()
+                raise PageUnloadable(url=url, status_code=None, message=str(e))
+
+            if fetch_favicon:
+                try:
+                    self.favicon_blob = await self.page.evaluate(FAVICON_FETCHER_JS)
+                    await _safe_request_gc(self.page)
+                except Exception as e:
+                    logger.error(f"Error fetching favicon: {e}")
+
+            if self.status_code != 200 and not ignore_status_codes:
+                screenshot = await capture_full_page_async(self.page, screenshot_format=self.screenshot_format, watch_uuid=watch_uuid, lock_viewport_elements=self.lock_viewport_elements)
+                try:
+                    page_html = await self.page.content()
+                except Exception as e:
+                    logger.warning(f"Got non-200 status {self.status_code} but failed to fetch page content: {e}")
+                    page_html = None
+                raise Non200ErrorCodeReceived(url=url, status_code=self.status_code, screenshot=screenshot, page_html=page_html)
+
+            if not empty_pages_are_a_change and len((await self.page.content()).strip()) == 0:
+                await context.close()
+                await browser.close()
+                raise EmptyReply(url=url, status_code=response.status)
+
+            try:
+                if self.browser_steps:
+                    try:
+                        await self.iterate_browser_steps(start_url=url)
+                    except BrowserStepsStepException:
+                        raise
+                    await self.page.wait_for_timeout(extra_wait * 1000)
+
+                now = time.time()
+                if current_include_filters is not None:
+                    await self.page.evaluate("var include_filters={}".format(json.dumps(current_include_filters)))
+                else:
+                    await self.page.evaluate("var include_filters=''")
+                await _safe_request_gc(self.page)
+
+                MAX_TOTAL_HEIGHT = int(os.getenv("SCREENSHOT_MAX_HEIGHT", SCREENSHOT_MAX_HEIGHT_DEFAULT))
+                self.xpath_data = await self.page.evaluate(XPATH_ELEMENT_JS, {
+                    "visualselector_xpath_selectors": visualselector_xpath_selectors,
+                    "max_height": MAX_TOTAL_HEIGHT
+                })
+                await _safe_request_gc(self.page)
+
+                self.instock_data = await self.page.evaluate(INSTOCK_DATA_JS)
+                await _safe_request_gc(self.page)
+
+                self.content = await self.page.content()
+                await _safe_request_gc(self.page)
+                logger.debug(f"Scrape xPath element data done in {time.time() - now:.2f}s")
+
+                self.screenshot = await capture_full_page_async(
+                    page=self.page,
+                    screenshot_format=self.screenshot_format,
+                    watch_uuid=watch_uuid,
+                    lock_viewport_elements=self.lock_viewport_elements,
+                )
+                await _safe_request_gc(self.page)
+                gc.collect()
+
+            except ScreenshotUnavailable:
+                raise ScreenshotUnavailable(url=url, status_code=self.status_code)
+
+            finally:
+                for obj, name, close_coro in [
+                    (self.page if hasattr(self, 'page') and self.page else None, 'page', lambda: self.page.close() if self.page else asyncio.sleep(0)),
+                    (context, 'context', lambda: context.close() if context else asyncio.sleep(0)),
+                    (browser, 'browser', lambda: browser.close() if browser else asyncio.sleep(0)),
+                ]:
+                    try:
+                        await asyncio.wait_for(close_coro(), timeout=5.0)
+                    except asyncio.TimeoutError:
+                        logger.warning(f"Timed out closing {name} for {url}")
+                    except Exception as e:
+                        logger.warning(f"Error closing {name} for {url}: {e}")
+
+                self.page = None
+                context = None
+                browser = None
+                gc.collect()

changedetectionio/content_fetchers/playwright/chrome.py

@@ -0,0 +1,27 @@
+"""
+Playwright Chrome fetcher — launches a local Chromium browser directly.
+
+No external browser container is required.  Playwright must be installed
+with Chromium browsers: ``playwright install chromium``.
+"""
+from changedetectionio.pluggy_interface import hookimpl
+from changedetectionio.content_fetchers.playwright import PlaywrightBaseFetcher
+
+
+class fetcher(PlaywrightBaseFetcher):
+    fetcher_description = "Playwright Chrome (local)"
+
+    async def _connect_browser(self, p):
+        launch_kwargs = {'headless': True}
+        if self.proxy:
+            launch_kwargs['proxy'] = self.proxy
+        return await p.chromium.launch(**launch_kwargs)
+
+
+class PlaywrightChromePlugin:
+    @hookimpl
+    def register_content_fetcher(self):
+        return ('playwright_chrome', fetcher)
+
+
+chrome_plugin = PlaywrightChromePlugin()

changedetectionio/content_fetchers/playwright/firefox.py

@@ -0,0 +1,33 @@
+"""
+Playwright Firefox fetcher — launches a local Firefox browser directly.
+
+No external browser container is required.  Playwright must be installed
+with Firefox browsers: ``playwright install firefox``.
+
+Note: ``page.request_gc()`` is Chromium-specific and is silently skipped
+on Firefox — this is handled transparently by ``_safe_request_gc()`` in
+the base package.
+"""
+from changedetectionio.pluggy_interface import hookimpl
+from changedetectionio.content_fetchers.playwright import PlaywrightBaseFetcher
+
+
+class fetcher(PlaywrightBaseFetcher):
+    fetcher_description = "Playwright Firefox (local)"
+
+    status_icon = {'filename': 'firefox-icon.svg', 'alt': 'Using Firefox', 'title': 'Using Firefox'}
+
+    async def _connect_browser(self, p):
+        launch_kwargs = {'headless': True}
+        if self.proxy:
+            launch_kwargs['proxy'] = self.proxy
+        return await p.firefox.launch(**launch_kwargs)
+
+
+class PlaywrightFirefoxPlugin:
+    @hookimpl
+    def register_content_fetcher(self):
+        return ('playwright_firefox', fetcher)
+
+
+firefox_plugin = PlaywrightFirefoxPlugin()

changedetectionio/content_fetchers/playwright/webkit.py

@@ -0,0 +1,30 @@
+"""
+Playwright WebKit fetcher — launches a local WebKit (Safari-engine) browser.
+
+No external browser container is required.  Playwright must be installed
+with WebKit browsers: ``playwright install webkit``.
+
+Note: ``page.request_gc()`` is Chromium-specific and is silently skipped
+on WebKit — handled transparently by ``_safe_request_gc()`` in the base package.
+"""
+from changedetectionio.pluggy_interface import hookimpl
+from changedetectionio.content_fetchers.playwright import PlaywrightBaseFetcher
+
+
+class fetcher(PlaywrightBaseFetcher):
+    fetcher_description = "Playwright WebKit/Safari (local)"
+
+    async def _connect_browser(self, p):
+        launch_kwargs = {'headless': True}
+        if self.proxy:
+            launch_kwargs['proxy'] = self.proxy
+        return await p.webkit.launch(**launch_kwargs)
+
+
+class PlaywrightWebKitPlugin:
+    @hookimpl
+    def register_content_fetcher(self):
+        return ('playwright_webkit', fetcher)
+
+
+webkit_plugin = PlaywrightWebKitPlugin()

changedetectionio/content_fetchers/puppeteer.py

@@ -7,6 +7,7 @@ from urllib.parse import urlparse
 
 from loguru import logger
 
+from changedetectionio.pluggy_interface import hookimpl
 from changedetectionio.content_fetchers import SCREENSHOT_MAX_HEIGHT_DEFAULT, visualselector_xpath_selectors, \
     SCREENSHOT_SIZE_STITCH_THRESHOLD, SCREENSHOT_DEFAULT_QUALITY, XPATH_ELEMENT_JS, INSTOCK_DATA_JS, \
     SCREENSHOT_MAX_TOTAL_HEIGHT, FAVICON_FETCHER_JS
@@ -166,11 +167,8 @@ async def capture_full_page(page, screenshot_format='JPEG', watch_uuid=None, loc
 
 
 class fetcher(Fetcher):
-    fetcher_description = "Puppeteer/direct {}/Javascript".format(
-        os.getenv("PLAYWRIGHT_BROWSER_TYPE", 'chromium').capitalize()
-    )
-    if os.getenv("PLAYWRIGHT_DRIVER_URL"):
-        fetcher_description += " via '{}'".format(os.getenv("PLAYWRIGHT_DRIVER_URL"))
+    fetcher_description = "Puppeteer Chromium"
+    requires_connection_url = True
 
     browser = None
     browser_type = ''
@@ -182,14 +180,10 @@ class fetcher(Fetcher):
     supports_screenshots = True
     supports_xpath_element_data = True
 
-    @classmethod
-    def get_status_icon_data(cls):
-        """Return Chrome browser icon data for Puppeteer fetcher."""
-        return {
-            'filename': 'google-chrome-icon.png',
-            'alt': 'Using a Chrome browser',
-            'title': 'Using a Chrome browser'
-        }
+    status_icon = {'filename': 'google-chrome-icon.png', 'alt': 'Using a Chrome browser', 'title': 'Using a Chrome browser'}
+
+    def disk_cleanup_after_fetch(self):
+        self.delete_browser_steps_screenshots()
 
     def __init__(self, proxy_override=None, custom_browser_connection_url=None, **kwargs):
         super().__init__(**kwargs)
@@ -198,9 +192,10 @@ class fetcher(Fetcher):
             self.browser_connection_is_custom = True
             self.browser_connection_url = custom_browser_connection_url
         else:
-            # Fallback to fetching from system
-            # .strip('"') is going to save someone a lot of time when they accidently wrap the env value
-            self.browser_connection_url = os.getenv("PLAYWRIGHT_DRIVER_URL", 'ws://playwright-chrome:3000').strip('"')
+            from loguru import logger
+            logger.critical("Puppeteer fetcher has no browser_connection_url — browser profile was not configured. "
+                            "Set PLAYWRIGHT_DRIVER_URL or configure a browser profile in Settings.")
+            self.browser_connection_url = None
 
         # allow per-watch proxy selection override
         # @todo check global too?
@@ -270,7 +265,7 @@ class fetcher(Fetcher):
         import re
         self.delete_browser_steps_screenshots()
 
-        n = int(os.getenv("WEBDRIVER_DELAY_BEFORE_CONTENT_READY", 12)) + self.render_extract_delay
+        n = self.extra_delay + self.render_extract_delay
         extra_wait = min(n, 15)
 
         logger.debug(f"Extra wait set to {extra_wait}s, requested was {n}s.")
@@ -447,8 +442,12 @@ class fetcher(Fetcher):
 
         if self.status_code != 200 and not ignore_status_codes:
             screenshot = await capture_full_page(page=self.page, screenshot_format=self.screenshot_format, watch_uuid=watch_uuid, lock_viewport_elements=self.lock_viewport_elements)
-
-            raise Non200ErrorCodeReceived(url=url, status_code=self.status_code, screenshot=screenshot)
+            try:
+                page_html = await self.page.content
+            except Exception as e:
+                logger.warning(f"Got non-200 status {self.status_code} but failed to fetch page content: {e}")
+                page_html = None
+            raise Non200ErrorCodeReceived(url=url, status_code=self.status_code, screenshot=screenshot, page_html=page_html)
 
         content = await self.page.content
 
@@ -548,9 +547,10 @@ class fetcher(Fetcher):
 class PuppeteerFetcherPlugin:
     """Plugin class that registers the Puppeteer fetcher as a built-in plugin."""
 
+    @hookimpl
     def register_content_fetcher(self):
         """Register the Puppeteer fetcher"""
-        return ('html_webdriver', fetcher)
+        return ('puppeteer', fetcher)
 
 
 # Create module-level instance for plugin registration

changedetectionio/content_fetchers/requests.py

@@ -8,6 +8,7 @@ import asyncio
 from changedetectionio import strtobool
 from changedetectionio.content_fetchers.exceptions import BrowserStepsInUnsupportedFetcher, EmptyReply, Non200ErrorCodeReceived
 from changedetectionio.content_fetchers.base import Fetcher
+from changedetectionio.pluggy_interface import hookimpl
 from changedetectionio.validate_url import is_private_hostname
 
 
@@ -148,10 +149,32 @@ class fetcher(Fetcher):
                         # Default to UTF-8 for XML if no encoding found
                         r.encoding = 'utf-8'
                 else:
-                    # For other content types, use chardet
-                    encoding = chardet.detect(r.content)['encoding']
-                    if encoding:
-                        r.encoding = encoding
+                    # No charset in HTTP header - sniff encoding in priority order matching browsers
+                    # (WHATWG encoding sniffing algorithm):
+                    # 1. BOM - highest confidence, check before anything else
+                    # 2. <meta charset> in first 2kb
+                    # 3. chardet statistical detection - last resort
+                    # See: https://github.com/dgtlmoon/changedetection.io/issues/3952
+                    boms = [
+                        (b'\xef\xbb\xbf', 'utf-8-sig'),
+                        (b'\xff\xfe', 'utf-16-le'),
+                        (b'\xfe\xff', 'utf-16-be'),
+                    ]
+                    bom_encoding = next((enc for bom, enc in boms if r.content.startswith(bom)), None)
+                    if bom_encoding:
+                        logger.info(f"URL: {url} Using encoding '{bom_encoding}' detected from BOM")
+                        r.encoding = bom_encoding
+                    else:
+                        meta_charset_match = re.search(rb'<meta[^>]+charset\s*=\s*["\']?\s*([^"\'\s;>]+)', r.content[:2000], re.IGNORECASE)
+                        if meta_charset_match:
+                            encoding = meta_charset_match.group(1).decode('ascii', errors='ignore')
+                            logger.info(f"URL: {url} No content-type encoding in HTTP headers - Using encoding '{encoding}' from HTML meta charset tag")
+                            r.encoding = encoding
+                        else:
+                            encoding = chardet.detect(r.content)['encoding']
+                            logger.warning(f"URL: {url} No charset in headers or meta tag, guessed encoding as '{encoding}' via chardet")
+                            if encoding:
+                                r.encoding = encoding
 
         self.headers = r.headers
 
@@ -236,9 +259,10 @@ class fetcher(Fetcher):
 class RequestsFetcherPlugin:
     """Plugin class that registers the requests fetcher as a built-in plugin."""
 
+    @hookimpl
     def register_content_fetcher(self):
         """Register the requests fetcher"""
-        return ('html_requests', fetcher)
+        return ('requests', fetcher)
 
 
 # Create module-level instance for plugin registration

changedetectionio/content_fetchers/webdriver_selenium.py

@@ -3,13 +3,13 @@ import time
 
 from loguru import logger
 from changedetectionio.content_fetchers.base import Fetcher
+from changedetectionio.content_fetchers.exceptions import Non200ErrorCodeReceived
+from changedetectionio.pluggy_interface import hookimpl
 
 
 class fetcher(Fetcher):
-    if os.getenv("WEBDRIVER_URL"):
-        fetcher_description = f"WebDriver Chrome/Javascript via \"{os.getenv('WEBDRIVER_URL', '')}\""
-    else:
-        fetcher_description = "WebDriver Chrome/Javascript"
+    fetcher_description = "Selenium WebDriver Chrome"
+    requires_connection_url = True
 
     proxy = None
     proxy_url = None
@@ -19,26 +19,21 @@ class fetcher(Fetcher):
     supports_screenshots = True
     supports_xpath_element_data = True
 
-    @classmethod
-    def get_status_icon_data(cls):
-        """Return Chrome browser icon data for WebDriver fetcher."""
-        return {
-            'filename': 'google-chrome-icon.png',
-            'alt': 'Using a Chrome browser',
-            'title': 'Using a Chrome browser'
-        }
+    status_icon = {'filename': 'google-chrome-icon.png', 'alt': 'Using a Chrome browser', 'title': 'Using a Chrome browser'}
 
     def __init__(self, proxy_override=None, custom_browser_connection_url=None, **kwargs):
         super().__init__(**kwargs)
         from urllib.parse import urlparse
         from selenium.webdriver.common.proxy import Proxy
 
-        # .strip('"') is going to save someone a lot of time when they accidently wrap the env value
-        if not custom_browser_connection_url:
-            self.browser_connection_url = os.getenv("WEBDRIVER_URL", 'http://browser-chrome:4444/wd/hub').strip('"')
-        else:
+        if custom_browser_connection_url:
             self.browser_connection_is_custom = True
             self.browser_connection_url = custom_browser_connection_url
+        else:
+            from loguru import logger
+            logger.critical("Selenium WebDriver fetcher has no browser_connection_url — browser profile was not configured. "
+                            "Set WEBDRIVER_URL or configure a browser profile in Settings.")
+            self.browser_connection_url = None
 
         ##### PROXY SETUP #####
 
@@ -130,22 +125,28 @@ class fetcher(Fetcher):
                 if not "--window-size" in os.getenv("CHROME_OPTIONS", ""):
                     driver.set_window_size(1280, 1024)
 
-                driver.implicitly_wait(int(os.getenv("WEBDRIVER_DELAY_BEFORE_CONTENT_READY", 5)))
+                driver.implicitly_wait(self.extra_delay)
 
                 if self.webdriver_js_execute_code is not None:
                     driver.execute_script(self.webdriver_js_execute_code)
                     # Selenium doesn't automatically wait for actions as good as Playwright, so wait again
-                    driver.implicitly_wait(int(os.getenv("WEBDRIVER_DELAY_BEFORE_CONTENT_READY", 5)))
-
-                # @todo - how to check this? is it possible?
-                self.status_code = 200
-                # @todo somehow we should try to get this working for WebDriver
-                # raise EmptyReply(url=url, status_code=r.status_code)
+                    driver.implicitly_wait(self.extra_delay)
 
                 # @todo - dom wait loaded?
                 import time
-                time.sleep(int(os.getenv("WEBDRIVER_DELAY_BEFORE_CONTENT_READY", 5)) + self.render_extract_delay)
+                time.sleep(self.extra_delay + self.render_extract_delay)
                 self.content = driver.page_source
+
+                # Use Navigation Timing API to get the real HTTP status code (Chrome 102+)
+                # Read after the sleep so the page is fully settled
+                try:
+                    nav_status = driver.execute_script(
+                        "return window.performance.getEntriesByType('navigation')[0]?.responseStatus"
+                    )
+                    # Guard against 0 (file://, blocked requests) which should not raise Non200
+                    self.status_code = int(nav_status) if nav_status and int(nav_status) > 0 else 200
+                except Exception:
+                    self.status_code = 200
                 self.headers = {}
 
                 # Selenium always captures as PNG, convert to JPEG if needed
@@ -175,6 +176,10 @@ class fetcher(Fetcher):
                     img.close()
                 else:
                     self.screenshot = screenshot_png
+
+                if self.status_code != 200 and not ignore_status_codes:
+                    raise Non200ErrorCodeReceived(url=url, status_code=self.status_code, screenshot=self.screenshot, page_html=self.content)
+
             except Exception as e:
                 driver.quit()
                 raise e
@@ -190,9 +195,10 @@ class fetcher(Fetcher):
 class WebDriverSeleniumFetcherPlugin:
     """Plugin class that registers the WebDriver Selenium fetcher as a built-in plugin."""
 
+    @hookimpl
     def register_content_fetcher(self):
         """Register the WebDriver Selenium fetcher"""
-        return ('html_webdriver', fetcher)
+        return ('selenium', fetcher)
 
 
 # Create module-level instance for plugin registration

changedetectionio/flask_app.py

@@ -4,6 +4,7 @@ import flask_login
 import locale
 import os
 import queue
+import re
 import sys
 import threading
 import time
@@ -217,9 +218,13 @@ def _jinja2_filter_format_number_locale(value: float) -> str:
     "Formats for example 4000.10 to the local locale default of 4,000.10"
     # Format the number with two decimal places (locale format string will return 6 decimal)
     formatted_value = locale.format_string("%.2f", value, grouping=True)
-
     return formatted_value
 
+@app.template_filter('regex_search')
+def _jinja2_filter_regex_search(value, pattern):
+    import re
+    return re.search(pattern, str(value)) is not None
+
 @app.template_global('is_checking_now')
 def _watch_is_checking_now(watch_obj, format="%Y-%m-%d %H:%M:%S"):
     return worker_pool.is_watch_running(watch_obj['uuid'])
@@ -336,52 +341,38 @@ def _jinja2_filter_format_duration(seconds):
 
 @app.template_filter('fetcher_status_icons')
 def _jinja2_filter_fetcher_status_icons(fetcher_name):
-    """Get status icon HTML for a given fetcher.
+    """Return status icon HTML for a fetcher, or empty string if none.
 
-    This filter checks both built-in fetchers and plugin fetchers for status icons.
-
-    Args:
-        fetcher_name: The fetcher name (e.g., 'html_webdriver', 'html_js_zyte')
-
-    Returns:
-        str: HTML string containing status icon elements
+    Built-in fetchers declare their icon via the ``status_icon`` class attribute
+    on their ``Fetcher`` subclass.  Plugin fetchers may still use the pluggy
+    ``collect_fetcher_status_icons`` hook as a fallback.
     """
     from changedetectionio import content_fetchers
-    from changedetectionio.pluggy_interface import collect_fetcher_status_icons
     from markupsafe import Markup
     from flask import url_for
 
     icon_data = None
 
-    # First check if it's a plugin fetcher (plugins have priority)
-    plugin_icon_data = collect_fetcher_status_icons(fetcher_name)
-    if plugin_icon_data:
-        icon_data = plugin_icon_data
-    # Check if it's a built-in fetcher
-    elif hasattr(content_fetchers, fetcher_name):
-        fetcher_class = getattr(content_fetchers, fetcher_name)
-        if hasattr(fetcher_class, 'get_status_icon_data'):
+    fetcher_class = content_fetchers.get_fetcher(fetcher_name)
+    if fetcher_class is not None:
+        icon_data = getattr(fetcher_class, 'status_icon', None)
+        if not icon_data and callable(getattr(fetcher_class, 'get_status_icon_data', None)):
             icon_data = fetcher_class.get_status_icon_data()
 
-    # Build HTML from icon data
-    if icon_data and isinstance(icon_data, dict):
-        # Use 'group' from icon_data if specified, otherwise default to 'images'
-        group = icon_data.get('group', 'images')
+    # Fallback: pluggy hook for plugins that implement fetcher_status_icon
+    if not icon_data:
+        from changedetectionio.pluggy_interface import collect_fetcher_status_icons
+        icon_data = collect_fetcher_status_icons(fetcher_name)
 
-        # Try to use url_for, but fall back to manual URL building if endpoint not registered yet
-        try:
-            icon_url = url_for('static_content', group=group, filename=icon_data['filename'])
-        except:
-            # Fallback: build URL manually respecting APPLICATION_ROOT
-            from flask import request
-            app_root = request.script_root if hasattr(request, 'script_root') else ''
-            icon_url = f"{app_root}/static/{group}/{icon_data['filename']}"
+    if not icon_data:
+        return ''
 
-        style_attr = f' style="{icon_data["style"]}"' if icon_data.get('style') else ''
-        html = f'<img class="status-icon" src="{icon_url}" alt="{icon_data["alt"]}" title="{icon_data["title"]}"{style_attr}>'
-        return Markup(html)
+    group = icon_data.get('group', 'images')
+    icon_url = url_for('static_content', group=group, filename=icon_data['filename'])
+    style_attr = f' style="{icon_data["style"]}"' if icon_data.get('style') else ''
+    return Markup(f'<img class="status-icon" src="{icon_url}" alt="{icon_data["alt"]}" title="{icon_data["title"]}"{style_attr}>')
 
-    return ''
+_RE_SANITIZE_TAG = re.compile(r'[^a-zA-Z0-9]')
 
 @app.template_filter('sanitize_tag_class')
 def _jinja2_filter_sanitize_tag_class(tag_title):
@@ -394,9 +385,8 @@ def _jinja2_filter_sanitize_tag_class(tag_title):
     Returns:
         str: A sanitized string suitable for use as a CSS class name
     """
-    import re
     # Remove all non-alphanumeric characters and convert to lowercase
-    sanitized = re.sub(r'[^a-zA-Z0-9]', '', tag_title).lower()
+    sanitized = _RE_SANITIZE_TAG.sub('', tag_title).lower()
     # Ensure it starts with a letter (CSS requirement)
     if sanitized and not sanitized[0].isalpha():
         sanitized = 'tag' + sanitized
@@ -484,28 +474,21 @@ def changedetection_app(config=None, datastore_o=None):
     available_languages = get_available_languages()
     language_codes = get_language_codes()
 
-    def get_locale():
-        # Locale aliases: map browser language codes to translation directory names
-        # This handles cases where browsers send standard codes (e.g., zh-TW)
-        # but our translations use more specific codes (e.g., zh_Hant_TW)
-        locale_aliases = {
-            'zh-TW': 'zh_Hant_TW',  # Traditional Chinese: browser sends zh-TW, we use zh_Hant_TW
-            'zh_TW': 'zh_Hant_TW',  # Also handle underscore variant
-        }
+    _locale_aliases = {
+        'zh-TW': 'zh_Hant_TW',  # Traditional Chinese: browser sends zh-TW, we use zh_Hant_TW
+        'zh_TW': 'zh_Hant_TW',  # Also handle underscore variant
+    }
+    _locale_match_list = language_codes + list(_locale_aliases.keys())
 
+    def get_locale():
         # 1. Try to get locale from session (user explicitly selected)
         if 'locale' in session:
             return session['locale']
 
         # 2. Fall back to Accept-Language header
-        # Get the best match from browser's Accept-Language header
-        browser_locale = request.accept_languages.best_match(language_codes + list(locale_aliases.keys()))
-
-        # 3. Check if we need to map the browser locale to our internal locale
-        if browser_locale in locale_aliases:
-            return locale_aliases[browser_locale]
-
-        return browser_locale
+        browser_locale = request.accept_languages.best_match(_locale_match_list)
+        # 3. Map browser locale to our internal locale if needed
+        return _locale_aliases.get(browser_locale, browser_locale)
 
     # Initialize Babel with locale selector
     babel = Babel(app, locale_selector=get_locale)
@@ -1018,15 +1001,16 @@ def check_for_new_version():
     import urllib3
     urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
 
+    session = requests.Session()
+    session.verify = False
+
     while not app.config.exit.is_set():
         try:
-            r = requests.post("https://changedetection.io/check-ver.php",
+            r = session.post("https://changedetection.io/check-ver.php",
                               data={'version': __version__,
                                     'app_guid': datastore.data['app_guid'],
                                     'watch_count': len(datastore.data['watching'])
-                                    },
-
-                              verify=False)
+                                    })
         except:
             pass
 

changedetectionio/forms.py

@@ -608,13 +608,12 @@ class ValidateCSSJSONXPATHInput(object):
                     raise ValidationError("XPath not permitted in this field!")
                 from lxml import etree, html
                 import elementpath
-                # xpath 2.0-3.1
-                from elementpath.xpath3 import XPath3Parser
+                from changedetectionio.html_tools import SafeXPath3Parser
                 tree = html.fromstring("<html></html>")
                 line = line.replace('xpath:', '')
 
                 try:
-                    elementpath.select(tree, line.strip(), parser=XPath3Parser)
+                    elementpath.select(tree, line.strip(), parser=SafeXPath3Parser)
                 except elementpath.ElementPathError as e:
                     message = field.gettext('\'%s\' is not a valid XPath expression. (%s)')
                     raise ValidationError(message % (line, str(e)))
@@ -668,9 +667,11 @@ class ValidateCSSJSONXPATHInput(object):
                     # `jq` requires full compilation in windows and so isn't generally available
                     raise ValidationError("jq not support not found")
 
+                from changedetectionio.html_tools import validate_jq_expression
                 input = line.replace('jq:', '')
 
                 try:
+                    validate_jq_expression(input)
                     jq.compile(input)
                 except (ValueError) as e:
                     message = field.gettext('\'%s\' is not a valid jq expression. (%s)')
@@ -741,7 +742,6 @@ class commonSettingsForm(Form):
         self.notification_title.extra_notification_tokens = kwargs.get('extra_notification_tokens', {})
         self.notification_urls.extra_notification_tokens = kwargs.get('extra_notification_tokens', {})
 
-    fetch_backend = RadioField(_l('Fetch Method'), choices=content_fetchers.available_fetchers(), validators=[ValidateContentFetcherIsReady()])
     notification_body = TextAreaField(_l('Notification Body'), default='{{ watch_url }} had a change.', validators=[validators.Optional(), ValidateJinja2Template()])
     notification_format = SelectField(_l('Notification format'), choices=list(valid_notification_formats.items()))
     notification_title = StringField(_l('Notification Title'), default='ChangeDetection.io Notification - {{ watch_url }}', validators=[validators.Optional(), ValidateJinja2Template()])
@@ -778,6 +778,7 @@ class SingleBrowserStep(Form):
 
 class processor_text_json_diff_form(commonSettingsForm):
 
+    browser_profile = RadioField(_l('Browser / Fetch method'), choices=[])  # populated at runtime in edit.py
     url = fields.URLField('Web Page URL', validators=[validateURL()])
     tags = StringTagUUID('Group Tag', [validators.Optional()], default='')
 
@@ -939,10 +940,66 @@ class SingleExtraBrowser(Form):
         ValidateSimpleURL()
     ], render_kw={"placeholder": "wss://brightdata... wss://oxylabs etc", "size":50})
 
-class DefaultUAInputForm(Form):
-    html_requests = StringField(_l('Plaintext requests'), validators=[validators.Optional()], render_kw={"placeholder": "<default>"})
-    if os.getenv("PLAYWRIGHT_DRIVER_URL") or os.getenv("WEBDRIVER_URL"):
-        html_webdriver = StringField(_l('Chrome requests'), validators=[validators.Optional()], render_kw={"placeholder": "<default>"})
+
+class BrowserProfileForm(Form):
+    """Create or edit a named BrowserProfile stored in settings.application.browser_profiles."""
+
+    name = StringField(
+        _l('Profile name'),
+        [validators.DataRequired(), validators.Length(max=100)],
+        render_kw={"placeholder": _l("e.g. Mobile Chrome, Bright Data CDP"), "maxlength": "100"}
+    )
+    fetch_backend = SelectField(
+        _l('Fetch method'),
+        choices=[],  # populated at runtime from available_fetchers()
+    )
+    browser_connection_url = StringField(
+        _l('Browser connection URL'),
+        [
+            validators.Optional(),
+            ValidateStartsWithRegex(
+                regex=r'^(wss?|ws|http|https)://',
+                flags=re.IGNORECASE,
+                message=_l('Browser connection URL must start with ws://, wss://, http://, https://')
+            ),
+            ValidateSimpleURL(),
+        ],
+        render_kw={"placeholder": "ws://my-chrome:3000", "size": 50}
+    )
+    viewport_width = IntegerField(
+        _l('Viewport width (px)'),
+        [validators.Optional(), validators.NumberRange(min=100, max=7680)],
+        default=1280,
+        render_kw={"style": "width:5em;"}
+    )
+    viewport_height = IntegerField(
+        _l('Viewport height (px)'),
+        [validators.Optional(), validators.NumberRange(min=100, max=4320)],
+        default=1000,
+        render_kw={"style": "width:5em;"}
+    )
+    block_images = BooleanField(_l('Block images (faster loads)'), default=False)
+    block_fonts = BooleanField(_l('Block web fonts'), default=False)
+    ignore_https_errors = BooleanField(_l('Ignore HTTPS/TLS errors'), default=False)
+    user_agent = StringField(
+        _l('User-Agent override'),
+        [validators.Optional(), validators.Length(max=500)],
+        render_kw={"placeholder": _l("Leave blank to use fetcher default"), "size": 60}
+    )
+    locale = StringField(
+        _l('Locale'),
+        [validators.Optional(), validators.Length(max=20)],
+        render_kw={"placeholder": "en-US, de-DE, fr-FR …", "size": 15}
+    )
+    custom_headers = TextAreaField(
+        _l('Custom headers'),
+        [validators.Optional()],
+        render_kw={
+            "placeholder": "Header-Name: value\nAnother-Header: value",
+            "rows": 4, "cols": 60,
+            "style": "font-family:monospace;"
+        }
+    )
 
 # datastore.data['settings']['requests']..
 class globalSettingsRequestForm(Form):
@@ -966,8 +1023,6 @@ class globalSettingsRequestForm(Form):
     extra_proxies = FieldList(FormField(SingleExtraProxy), min_entries=5)
     extra_browsers = FieldList(FormField(SingleExtraBrowser), min_entries=5)
 
-    default_ua = FormField(DefaultUAInputForm, label=_l("Default User-Agent overrides"))
-
     def validate_extra_proxies(self, extra_validators=None):
         for e in self.data['extra_proxies']:
             if e.get('proxy_name') or e.get('proxy_url'):
@@ -990,7 +1045,6 @@ class globalSettingsApplicationForm(commonSettingsForm):
                            render_kw={"placeholder": os.getenv('BASE_URL', 'Not set')}
                            )
     empty_pages_are_a_change =  BooleanField(_l('Treat empty pages as a change?'), default=False)
-    fetch_backend = RadioField(_l('Fetch Method'), default="html_requests", choices=content_fetchers.available_fetchers(), validators=[ValidateContentFetcherIsReady()])
     global_ignore_text = StringListField(_l('Ignore Text'), [ValidateListRegex()])
     global_subtractive_selectors = StringListField(_l('Remove elements'), [ValidateCSSJSONXPATHInput(allow_json=False)])
     ignore_whitespace = BooleanField(_l('Ignore whitespace'))
@@ -1006,7 +1060,7 @@ class globalSettingsApplicationForm(commonSettingsForm):
         render_kw={"placeholder": "0.1", "style": "width: 8em;"}
     )
 
-    password = SaltyPasswordField(_l('Password'))
+    password = SaltyPasswordField(_l('Password'), render_kw={"autocomplete": "new-password"})
     pager_size = IntegerField(_l('Pager size'),
                               render_kw={"style": "width: 5em;"},
                               validators=[validators.NumberRange(min=0,

changedetectionio/html_tools.py

@@ -4,6 +4,7 @@ from loguru import logger
 from typing import List
 import html
 import json
+import os
 import re
 
 # HTML added to be sure each result matching a filter (.example) gets converted to a new line by Inscriptis
@@ -13,6 +14,45 @@ PERL_STYLE_REGEX = r'^/(.*?)/([a-z]*)?$'
 
 TITLE_RE = re.compile(r"<title[^>]*>(.*?)</title>", re.I | re.S)
 META_CS  = re.compile(r'<meta[^>]+charset=["\']?\s*([a-z0-9_\-:+.]+)', re.I)
+
+# jq builtins that can leak sensitive data or cause harm when user-supplied expressions are executed.
+# env/$ENV reads all process environment variables (passwords, API keys, etc.)
+# include/import can read arbitrary files from disk
+# input/inputs reads beyond the supplied JSON data
+# debug/stderr leaks data to stderr
+# halt/halt_error terminates the process (DoS)
+_JQ_BLOCKED_PATTERNS = [
+    (re.compile(r'\benv\b'),                    'env (reads environment variables)'),
+    (re.compile(r'\$ENV\b'),                    '$ENV (reads environment variables)'),
+    (re.compile(r'\binclude\b'),                'include (reads files from disk)'),
+    (re.compile(r'\bimport\b'),                 'import (reads files from disk)'),
+    (re.compile(r'\binputs?\b'),                'input/inputs (reads beyond provided data)'),
+    (re.compile(r'\bdebug\b'),                  'debug (leaks data to stderr)'),
+    (re.compile(r'\bstderr\b'),                 'stderr (leaks data to stderr)'),
+    (re.compile(r'\bhalt(?:_error)?\b'),        'halt/halt_error (terminates the process)'),
+    (re.compile(r'\$__loc__\b'),                '$__loc__ (leaks file path information)'),
+    (re.compile(r'\bbuiltins\b'),               'builtins (enumerates available functions)'),
+    (re.compile(r'\bmodulemeta\b'),             'modulemeta (leaks module information)'),
+    (re.compile(r'\$JQ_BUILD_CONFIGURATION\b'), '$JQ_BUILD_CONFIGURATION (leaks build information)'),
+]
+
+def validate_jq_expression(expression: str) -> None:
+    """Raise ValueError if the jq expression uses any dangerous builtin.
+
+    User-supplied jq expressions are executed server-side. Without this check,
+    builtins like `env` expose every process environment variable (SALTED_PASS,
+    proxy credentials, API keys, etc.) as watch output.
+    """
+    from changedetectionio.strtobool import strtobool
+    if strtobool(os.getenv('JQ_ALLOW_RISKY_EXPRESSIONS', 'false')):
+        return
+
+    for pattern, description in _JQ_BLOCKED_PATTERNS:
+        if pattern.search(expression):
+            msg = f"jq expression uses disallowed builtin: {description}"
+            logger.critical(f"Security: blocked jq expression containing '{description}' - expression: {expression!r}")
+            raise ValueError(msg)
+
 META_CT  = re.compile(r'<meta[^>]+http-equiv=["\']?content-type["\']?[^>]*content=["\'][^>]*charset=([a-z0-9_\-:+.]+)', re.I)
 
 # 'price' , 'lowPrice', 'highPrice' are usually under here
@@ -23,6 +63,59 @@ class JSONNotFound(ValueError):
     def __init__(self, msg):
         ValueError.__init__(self, msg)
 
+
+_DEFAULT_UNSAFE_XPATH3_FUNCTIONS = [
+    'unparsed-text',
+    'unparsed-text-lines',
+    'unparsed-text-available',
+    'doc',
+    'doc-available',
+    'json-doc',
+    'json-doc-available',
+    'collection',           # XPath 2.0+: loads XML node collections from arbitrary URIs
+    'uri-collection',       # XPath 3.0+: enumerates URIs from resource collections
+    'transform',            # XPath 3.1: XSLT transformation (currently raises, block proactively)
+    'load-xquery-module',   # XPath 3.1: loads XQuery modules (currently raises, block proactively)
+    'environment-variable',
+    'available-environment-variables',
+]
+
+
+def _build_safe_xpath3_parser():
+    """Return an XPath3Parser subclass with filesystem/environment access functions removed.
+
+    XPath 3.0 includes functions that can read arbitrary files or environment variables:
+      - unparsed-text / unparsed-text-lines / unparsed-text-available  (file read)
+      - doc / doc-available                                             (XML fetch from URI)
+      - environment-variable / available-environment-variables         (env var leakage)
+
+    Subclassing gives us an independent symbol_table copy (not shared with the parent class),
+    so removing entries here does not affect XPath3Parser itself.
+
+    Override the blocked list via the XPATH_BLOCKED_FUNCTIONS environment variable
+    (comma-separated, e.g. "unparsed-text,doc,environment-variable").
+    """
+    import os
+    from elementpath.xpath3 import XPath3Parser
+
+    class SafeXPath3Parser(XPath3Parser):
+        pass
+
+    env_override = os.getenv('XPATH_BLOCKED_FUNCTIONS')
+    if env_override is not None:
+        blocked = [f.strip() for f in env_override.split(',') if f.strip()]
+    else:
+        blocked = _DEFAULT_UNSAFE_XPATH3_FUNCTIONS
+
+    for _fn in blocked:
+        SafeXPath3Parser.symbol_table.pop(_fn, None)
+
+    return SafeXPath3Parser
+
+
+# Module-level singleton — built once, reused everywhere.
+SafeXPath3Parser = _build_safe_xpath3_parser()
+
 # Doesn't look like python supports forward slash auto enclosure in re.findall
 # So convert it to inline flag "(?i)foobar" type configuration
 @lru_cache(maxsize=100)
@@ -183,8 +276,6 @@ def xpath_filter(xpath_filter, html_content, append_pretty_line_formatting=False
     """
     from lxml import etree, html
     import elementpath
-    # xpath 2.0-3.1
-    from elementpath.xpath3 import XPath3Parser
 
     parser = etree.HTMLParser()
     tree = None
@@ -210,7 +301,7 @@ def xpath_filter(xpath_filter, html_content, append_pretty_line_formatting=False
             # This allows //title to match elements in the default namespace
             namespaces[''] = tree.nsmap[None]
 
-        r = elementpath.select(tree, xpath_filter.strip(), namespaces=namespaces, parser=XPath3Parser)
+        r = elementpath.select(tree, xpath_filter.strip(), namespaces=namespaces, parser=SafeXPath3Parser)
         #@note: //title/text() now works with default namespaces (fixed by registering '' prefix)
         #@note: //title/text() wont work where <title>CDATA.. (use cdata_in_document_to_text first)
 
@@ -235,6 +326,9 @@ def xpath_filter(xpath_filter, html_content, append_pretty_line_formatting=False
             else:
                 html_block += elementpath_tostring(element)
 
+        # Drop element references before the finally block so tree.clear() can release
+        # the libxml2 document immediately (elements pin the C-level doc via refcount).
+        del r
         return html_block
     finally:
         # Explicitly clear the tree to free memory
@@ -330,12 +424,16 @@ def _parse_json(json_data, json_filter):
             raise Exception("jq not support not found")
 
         if json_filter.startswith("jq:"):
-            jq_expression = jq.compile(json_filter.removeprefix("jq:"))
+            expr = json_filter.removeprefix("jq:")
+            validate_jq_expression(expr)
+            jq_expression = jq.compile(expr)
             match = jq_expression.input(json_data).all()
             return _get_stripped_text_from_json_match(match)
 
         if json_filter.startswith("jqraw:"):
-            jq_expression = jq.compile(json_filter.removeprefix("jqraw:"))
+            expr = json_filter.removeprefix("jqraw:")
+            validate_jq_expression(expr)
+            jq_expression = jq.compile(expr)
             match = jq_expression.input(json_data).all()
             return '\n'.join(str(item) for item in match)
 
@@ -439,13 +537,25 @@ def extract_json_as_string(content, json_filter, ensure_is_ldjson_info_type=None
         except json.JSONDecodeError as e:
             logger.warning(f"Error processing JSON {content[:20]}...{str(e)})")
     else:
-        # Probably something else, go fish inside for it
-        try:
-            stripped_text_from_html = extract_json_blob_from_html(content=content,
-                                                                  ensure_is_ldjson_info_type=ensure_is_ldjson_info_type,
-                                                                  json_filter=json_filter                                                                  )
-        except json.JSONDecodeError as e:
-            logger.warning(f"Error processing JSON while extracting JSON from HTML blob {content[:20]}...{str(e)})")
+        # Check for JSONP wrapper: someCallback({...}) or some.namespace({...})
+        # Server may claim application/json but actually return JSONP
+        jsonp_match = re.match(r'^\w[\w.]*\s*\((.+)\)\s*;?\s*$', content.lstrip("\ufeff").strip(), re.DOTALL)
+        if jsonp_match:
+            try:
+                inner = jsonp_match.group(1).strip()
+                logger.warning(f"Content looks like JSONP, attempting to extract inner JSON for filter '{json_filter}'")
+                stripped_text_from_html = _parse_json(json.loads(inner), json_filter)
+            except json.JSONDecodeError as e:
+                logger.warning(f"Error processing JSONP inner content {content[:20]}...{str(e)})")
+
+        if not stripped_text_from_html:
+            # Probably something else, go fish inside for it
+            try:
+                stripped_text_from_html = extract_json_blob_from_html(content=content,
+                                                                      ensure_is_ldjson_info_type=ensure_is_ldjson_info_type,
+                                                                      json_filter=json_filter)
+            except json.JSONDecodeError as e:
+                logger.warning(f"Error processing JSON while extracting JSON from HTML blob {content[:20]}...{str(e)})")
 
     if not stripped_text_from_html:
         # Re 265 - Just return an empty string when filter not found

changedetectionio/model/App.py

@@ -12,7 +12,6 @@ from changedetectionio.notification import (
 
 # Equal to or greater than this number of FilterNotFoundInResponse exceptions will trigger a filter-not-found notification
 _FILTER_FAILURE_THRESHOLD_ATTEMPTS_DEFAULT = 6
-DEFAULT_SETTINGS_HEADERS_USERAGENT='Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/87.0.4280.66 Safari/537.36'
 
 
 
@@ -31,10 +30,6 @@ class model(dict):
                     'time_between_check': {'weeks': None, 'days': None, 'hours': 3, 'minutes': None, 'seconds': None},
                     'timeout': int(getenv("DEFAULT_SETTINGS_REQUESTS_TIMEOUT", "45")),  # Default 45 seconds
                     'workers': int(getenv("DEFAULT_SETTINGS_REQUESTS_WORKERS", "5")),  # Number of threads, lower is better for slow connections
-                    'default_ua': {
-                        'html_requests': getenv("DEFAULT_SETTINGS_HEADERS_USERAGENT", DEFAULT_SETTINGS_HEADERS_USERAGENT),
-                        'html_webdriver': None,
-                    }
                 },
                 'application': {
                     # Custom notification content
@@ -43,7 +38,9 @@ class model(dict):
                     'api_access_token_enabled': True,
                     'base_url' : None,
                     'empty_pages_are_a_change': False,
-                    'fetch_backend': getenv("DEFAULT_FETCH_BACKEND", "html_requests"),
+                    'browser_profile': None,           # machine-name of the system-default BrowserProfile
+                    'browser_profiles': {},            # user-defined profiles keyed by machine name
+                    'fetch_backend': getenv("DEFAULT_FETCH_BACKEND", "requests"),
                     'filter_failure_notification_threshold_attempts': _FILTER_FAILURE_THRESHOLD_ATTEMPTS_DEFAULT,
                     'global_ignore_text': [], # List of text to ignore when calculating the comparison checksum
                     'global_subtractive_selectors': [],

changedetectionio/model/Watch.py

@@ -43,6 +43,11 @@ from ..html_tools import TRANSLATE_WHITESPACE_TABLE
 FAVICON_RESAVE_THRESHOLD_SECONDS=86400
 BROTLI_COMPRESS_SIZE_THRESHOLD = int(os.getenv('SNAPSHOT_BROTLI_COMPRESSION_THRESHOLD', 1024*20))
 
+# Module-level favicon filename cache: data_dir → basename (or None)
+# Keyed by data_dir so it survives Watch object recreation, deepcopy, and concurrent requests.
+# Invalidated explicitly in bump_favicon() when a new favicon is saved.
+_FAVICON_FILENAME_CACHE: dict = {}
+
 minimum_seconds_recheck_time = int(os.getenv('MINIMUM_SECONDS_RECHECK_TIME', 3))
 mtable = {'seconds': 1, 'minutes': 60, 'hours': 3600, 'days': 86400, 'weeks': 86400 * 7}
 
@@ -348,41 +353,41 @@ class model(EntityPersistenceMixin, watch_base):
     def is_source_type_url(self):
         return self.get('url', '').startswith('source:')
 
+    @property
+    def effective_browser_profile(self):
+        """Resolve the effective BrowserProfile for this watch.
+
+        Walks the chain: watch → tag (overrides_watch=True) → global settings → built-in fallback.
+        Never raises. Returns a BrowserProfile instance.
+        """
+        from changedetectionio.model.browser_profile import resolve_browser_profile, BUILTIN_REQUESTS
+        if not self._datastore:
+            return BUILTIN_REQUESTS
+        try:
+            return resolve_browser_profile(self, self._datastore)
+        except Exception:
+            return BUILTIN_REQUESTS
+
     @property
     def get_fetch_backend(self):
+        """Legacy property — prefer effective_browser_profile.fetch_backend for new code.
+
+        Returns the raw fetch_backend stored on this watch (or 'requests' for PDFs).
+        Does NOT walk the tag/global resolution chain.
         """
-        Get the fetch backend for this watch with special case handling.
-
-        CHAIN RESOLUTION OPPORTUNITY:
-        Currently returns watch.fetch_backend directly, but doesn't implement
-        Watch → Tag → Global resolution chain. With Pydantic:
-
-        @computed_field
-        def resolved_fetch_backend(self) -> str:
-            # Special case: PDFs always use html_requests
-            if self.is_pdf:
-                return 'html_requests'
-
-            # Watch override
-            if self.fetch_backend and self.fetch_backend != 'system':
-                return self.fetch_backend
-
-            # Tag override (first tag with overrides_watch=True wins)
-            for tag_uuid in self.tags:
-                tag = self._datastore.get_tag(tag_uuid)
-                if tag.overrides_watch and tag.fetch_backend:
-                    return tag.fetch_backend
-
-            # Global default
-            return self._datastore.settings.fetch_backend
-        """
-        # Maybe also if is_image etc?
-        # This is because chrome/playwright wont render the PDF in the browser and we will just fetch it and use pdf2html to see the text.
         if self.is_pdf:
-            return 'html_requests'
-
+            return 'requests'
         return self.get('fetch_backend')
 
+    @property
+    def fetcher_supports_screenshots(self):
+        """Return True if the resolved fetcher for this watch supports screenshots."""
+        from changedetectionio import content_fetchers
+        fetcher_class = content_fetchers.get_fetcher(self.effective_browser_profile.fetch_backend)
+        if fetcher_class is None:
+            return False
+        return bool(getattr(fetcher_class, 'supports_screenshots', False))
+
     @property
     def is_pdf(self):
         url = str(self.get("url") or "").lower()
@@ -806,9 +811,8 @@ class model(EntityPersistenceMixin, watch_base):
                     with open(fname, 'wb') as f:
                         f.write(decoded)
 
-                    # Invalidate favicon filename cache
-                    if hasattr(self, '_favicon_filename_cache'):
-                        delattr(self, '_favicon_filename_cache')
+                    # Invalidate module-level favicon filename cache for this watch
+                    _FAVICON_FILENAME_CACHE.pop(self.data_dir, None)
 
                     # A signal that could trigger the socket server to update the browser also
                     watch_check_update = signal('watch_favicon_bump')
@@ -823,35 +827,23 @@ class model(EntityPersistenceMixin, watch_base):
 
     def get_favicon_filename(self) -> str | None:
         """
-        Find any favicon.* file in the current working directory
-        and return the contents of the newest one.
+        Find any favicon.* file in the watch data directory.
 
-        MEMORY LEAK FIX: Cache the result to avoid repeated glob.glob() operations.
-        glob.glob() causes millions of fnmatch allocations when called for every watch on page load.
+        Uses a module-level cache keyed by data_dir to survive Watch object recreation,
+        deepcopy (which drops instance attrs), and concurrent request races.
+        Invalidated by bump_favicon() when a new favicon is saved.
 
         Returns:
-            str: Basename of the newest favicon file, or None if not found.
+            str: Basename of the favicon file, or None if not found.
         """
-        # Check cache first (prevents 26M+ allocations from repeated glob operations)
-        cache_key = '_favicon_filename_cache'
-        if hasattr(self, cache_key):
-            return getattr(self, cache_key)
+        if self.data_dir in _FAVICON_FILENAME_CACHE:
+            return _FAVICON_FILENAME_CACHE[self.data_dir]
 
         import glob
-
-        # Search for all favicon.* files
         files = glob.glob(os.path.join(self.data_dir, "favicon.*"))
-
-        if not files:
-            result = None
-        else:
-            # Find the newest by modification time
-            newest_file = max(files, key=os.path.getmtime)
-            result = os.path.basename(newest_file)
-
-        # Cache the result
-        setattr(self, cache_key, result)
-        return result
+        fname = os.path.basename(files[0]) if files else None
+        _FAVICON_FILENAME_CACHE[self.data_dir] = fname
+        return fname
 
     def get_screenshot_as_thumbnail(self, max_age=3200):
         """Return path to a square thumbnail of the most recent screenshot.
@@ -1182,18 +1174,13 @@ class model(EntityPersistenceMixin, watch_base):
     def compile_error_texts(self, has_proxies=None):
         """Compile error texts for this watch.
         Accepts has_proxies parameter to ensure it works even outside app context"""
-        from flask import url_for
+        from flask import url_for, has_request_context
         from markupsafe import Markup
 
         output = []  # Initialize as list since we're using append
         last_error = self.get('last_error','')
 
-        try:
-            url_for('settings.settings_page')
-        except Exception as e:
-            has_app_context = False
-        else:
-            has_app_context = True
+        has_app_context = has_request_context()
 
         # has app+request context, we can use url_for()
         if has_app_context:

changedetectionio/model/__init__.py

@@ -187,6 +187,7 @@ class watch_base(dict):
             'content-type': None,
             'date_created': None,
             'extract_text': [],  # Extract text by regex after filters
+            'browser_profile': 'system',  # machine-name key of a BrowserProfile; 'system' → resolve via chain
             'fetch_backend': 'system',  # plaintext, playwright etc
             'fetch_time': 0.0,
             'filter_failure_notification_send': strtobool(os.getenv('FILTER_FAILURE_NOTIFICATION_SEND_DEFAULT', 'True')),
@@ -589,7 +590,9 @@ class watch_base(dict):
             return None
 
         try:
-            value = self._datastore['settings']
+            # _datastore is a ChangeDetectionStore (has .data) or a plain dict (unit tests)
+            store_data = self._datastore.data if hasattr(self._datastore, 'data') else self._datastore
+            value = store_data['settings']
             for key in path:
                 value = value[key]
             return value

changedetectionio/model/browser_profile.py

@@ -0,0 +1,380 @@
+"""
+BrowserProfile — named, reusable browser/fetcher configuration.
+
+Storage key
+-----------
+Profiles are stored in ``settings.application.browser_profiles`` as a plain dict
+keyed by *machine name* — a lowercase, underscore-separated slug derived from the
+human-readable ``name`` field:
+
+    'My Blocking Chrome'        →  'my_blocking_chrome'
+    'Custom CDP — Mobile (375px)' →  'custom_cdp_mobile_375px'
+
+Using the machine name as the key means that deleting a profile and recreating
+it with the same name restores the original key, so all watches that referenced
+it continue to work without any manual re-linking.
+
+Resolution chain
+----------------
+``resolve_browser_profile(watch, datastore)`` walks:
+
+    watch.browser_profile  →  first tag with overrides_watch=True  →
+    settings.application.browser_profile  →  built-in fallback
+
+It never raises.  Stale / missing machine-name references are logged and the
+resolver falls through to the next level.
+
+Built-in profiles
+-----------------
+``BUILTIN_REQUESTS`` and ``BUILTIN_BROWSER`` are always available and cannot be
+deleted from the UI (``is_builtin=True``).  Their machine names are stored in
+``RESERVED_MACHINE_NAMES`` to block user profiles from shadowing them.
+
+Migration
+---------
+``store/updates.py::update_31`` converts the legacy ``fetch_backend`` field on
+watches, tags and global settings into ``browser_profile`` machine-name
+references.  After that migration no legacy paths are needed here.
+"""
+
+from __future__ import annotations
+
+import os
+import re
+from typing import Optional
+
+from loguru import logger
+from pydantic import BaseModel, field_validator
+
+# Default User-Agent for the built-in plaintext requests profile.
+# Overridable via environment variable for deployments that need a custom UA.
+_DEFAULT_REQUESTS_UA = os.getenv(
+    "DEFAULT_SETTINGS_HEADERS_USERAGENT",
+    'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/87.0.4280.66 Safari/537.36'
+)
+
+# ---------------------------------------------------------------------------
+# Constants
+# ---------------------------------------------------------------------------
+
+NAME_MAX_LEN = 100
+
+
+# ---------------------------------------------------------------------------
+# Model
+# ---------------------------------------------------------------------------
+
+class BrowserProfile(BaseModel):
+    """
+    A named, reusable configuration for how a watch fetches its target URL.
+
+    The *machine name* (see ``get_machine_name()``) is the stable storage key.
+    Updating ``name`` changes the machine name; any watch that referenced the
+    old machine name will then fall back through the resolution chain until it
+    is explicitly re-pointed.  To replace a profile without breaking watches,
+    delete it and recreate it with the *same* name.
+    """
+
+    name: str
+    """Human-readable label shown in the UI.  Max 100 characters."""
+
+    fetch_backend: str = 'requests'
+    """
+    Which fetch engine to use.  This is the *clean* fetcher name without the
+    ``html_`` module prefix (e.g. ``'requests'``, ``'webdriver'``,
+    ``'playwright'``, ``'puppeteer'``, ``'cloakbrowser'``).
+
+    The module-level ``html_`` prefix (``html_requests``, ``html_webdriver``,
+    …) is an implementation detail of ``content_fetchers/``.  Use
+    ``get_fetcher_class_name()`` to obtain the full module attribute name when
+    you need to look up the class.
+
+    Must be non-empty and contain only ``[a-z0-9_]`` characters.
+    """
+
+    is_builtin: bool = False
+    """Built-in profiles are always present and cannot be deleted from the UI."""
+
+    # ------------------------------------------------------------------
+    # Browser-specific settings (silently ignored by html_requests)
+    # ------------------------------------------------------------------
+
+    browser_connection_url: Optional[str] = None
+    """
+    Custom CDP / WebSocket endpoint, e.g. ``ws://my-chrome:3000``.
+    Overrides the system-wide ``PLAYWRIGHT_DRIVER_URL`` for this profile.
+    Only meaningful for ``html_webdriver`` profiles.
+    """
+
+    viewport_width: int = 1280
+    """
+    Browser viewport width in pixels.
+    Common presets: 375 (iPhone), 768 (tablet), 1280 (desktop).
+    """
+
+    viewport_height: int = 1000
+    """
+    Browser viewport height in pixels.
+    Common presets: 812 (iPhone), 1024 (tablet), 1000 (desktop).
+    """
+
+    block_images: bool = False
+    """
+    Block all image requests.  Typically cuts page-load time by 40-70 % on
+    image-heavy sites with no impact on text-based change detection.
+    """
+
+    block_fonts: bool = False
+    """Block web-font requests.  Modest speed gain; rarely affects detection."""
+
+    user_agent: Optional[str] = None
+    """
+    Override the browser User-Agent string.
+    ``None`` keeps the fetcher's built-in default, which already strips
+    obvious headless markers such as ``HeadlessChrome``.
+    """
+
+    ignore_https_errors: bool = False
+    """
+    Proceed even when the server's TLS certificate is invalid or self-signed.
+    Useful for staging / development environments.
+    """
+
+    locale: Optional[str] = None
+    """
+    Browser locale (e.g. ``en-US``, ``de-DE``).
+    Sets the ``Accept-Language`` header and ``navigator.language``.
+    Some sites serve different prices or copy based on locale.
+    """
+
+    custom_headers: str = ''
+    """
+    Extra HTTP headers sent with every request using this profile, in ``Key: Value`` format
+    (one per line, ``#`` lines are ignored).  Applied before per-watch headers so
+    individual watches can override them.
+    """
+
+    service_workers: str = 'allow'
+    """
+    Whether to allow Service Workers in the browser context.
+    Playwright accepts ``'allow'`` or ``'block'``.
+    Block to avoid large Service Worker data transfers (e.g. YouTube).
+    """
+
+    extra_delay: int = 0
+    """
+    Extra seconds to wait after page load before extracting content
+    (on top of the per-watch ``render_extract_delay``).
+    Sourced from ``WEBDRIVER_DELAY_BEFORE_CONTENT_READY`` at startup.
+    """
+
+    model_config = {"frozen": False}
+
+    # ------------------------------------------------------------------
+    # Validators
+    # ------------------------------------------------------------------
+
+    @field_validator('fetch_backend')
+    @classmethod
+    def _validate_fetch_backend(cls, v: str) -> str:
+        v = v.strip()
+        if not v:
+            raise ValueError('fetch_backend cannot be empty')
+        if not re.fullmatch(r'[a-z0-9_]+', v):
+            raise ValueError(
+                f"fetch_backend must contain only lowercase letters, digits and underscores, got {v!r}"
+            )
+        if v.startswith('html_'):
+            raise ValueError(
+                f"fetch_backend should be the clean fetcher name without the 'html_' prefix "
+                f"(e.g. 'requests', 'webdriver', 'playwright'). Got {v!r}. "
+                f"Use get_fetcher_class_name() to obtain the full module attribute name."
+            )
+        return v
+
+    @field_validator('name')
+    @classmethod
+    def _validate_name(cls, v: str) -> str:
+        v = v.strip()
+        if not v:
+            raise ValueError('Name cannot be empty')
+        if len(v) > NAME_MAX_LEN:
+            raise ValueError(f'Name must be {NAME_MAX_LEN} characters or less')
+        return v
+
+    # ------------------------------------------------------------------
+    # Machine-name helpers
+    # ------------------------------------------------------------------
+
+    @staticmethod
+    def machine_name_from_str(name: str) -> str:
+        """
+        Convert a human name to a machine-safe storage key.
+
+        Transformation rules (applied in order):
+
+        1. Strip surrounding whitespace; lower-case.
+        2. Replace runs of whitespace or hyphens with a single ``_``.
+        3. Drop every character that is not ``[a-z0-9_]``.
+        4. Collapse consecutive underscores.
+        5. Strip leading / trailing underscores.
+        6. Truncate to ``NAME_MAX_LEN`` characters.
+
+        Examples::
+
+            'My Blocking Browser Chrome'  →  'my_blocking_browser_chrome'
+            'Custom CDP — Mobile (375px)' →  'custom_cdp_mobile_375px'
+            '  Weird   ---   Name  '      →  'weird_name'
+        """
+        s = name.strip().lower()
+        s = re.sub(r'[\s\-]+', '_', s)    # whitespace / hyphens → underscore
+        s = re.sub(r'[^a-z0-9_]', '', s)  # drop everything else
+        s = re.sub(r'_+', '_', s)         # collapse repeated underscores
+        s = s.strip('_')                   # drop leading / trailing underscores
+        return s[:NAME_MAX_LEN]
+
+    def get_machine_name(self) -> str:
+        """Return the machine-safe storage key derived from this profile's ``name``."""
+        return self.machine_name_from_str(self.name)
+
+    def get_fetcher_class_name(self) -> str:
+        """Return the clean fetcher name for this profile (same as ``fetch_backend``).
+
+        Use with ``content_fetchers.get_fetcher()``::
+
+            from changedetectionio import content_fetchers
+            fetcher_cls = content_fetchers.get_fetcher(profile.get_fetcher_class_name())
+        """
+        return self.fetch_backend
+
+
+# ---------------------------------------------------------------------------
+# Built-in profiles (always present, cannot be deleted)
+# ---------------------------------------------------------------------------
+
+BUILTIN_REQUESTS = BrowserProfile(
+    name='Direct HTTP (requests)',
+    fetch_backend='requests',
+    is_builtin=True,
+    user_agent=_DEFAULT_REQUESTS_UA,
+)
+
+BUILTIN_PLAYWRIGHT = BrowserProfile(
+    name='Browser (Chrome/Playwright)',
+    fetch_backend='playwright_cdp',
+    is_builtin=True,
+)
+
+BUILTIN_SELENIUM = BrowserProfile(
+    name='Browser (Chrome/Selenium)',
+    fetch_backend='selenium',
+    is_builtin=True,
+)
+
+BUILTIN_PUPPETEER = BrowserProfile(
+    name='Browser (Chrome/Puppeteer)',
+    fetch_backend='puppeteer',
+    is_builtin=True,
+)
+
+# Backwards-compatible alias — code that imported BUILTIN_BROWSER keeps working.
+BUILTIN_BROWSER = BUILTIN_PLAYWRIGHT
+
+# Keyed by machine name for O(1) lookup.
+_BUILTINS: dict[str, BrowserProfile] = {
+    b.get_machine_name(): b
+    for b in (BUILTIN_REQUESTS, BUILTIN_PLAYWRIGHT, BUILTIN_SELENIUM, BUILTIN_PUPPETEER)
+}
+
+# Machine names that cannot be used by user-created profiles.
+RESERVED_MACHINE_NAMES: frozenset[str] = frozenset(_BUILTINS.keys())
+
+
+def get_default_browser_builtin() -> BrowserProfile:
+    """Final fallback when no profile can be resolved through the chain.
+
+    ``preconfigure_browser_profiles_based_on_env()`` sets
+    ``settings.application.browser_profile`` explicitly at startup, so this
+    fallback is only reached for watches with stale / missing machine-name
+    references.  Safe default is always direct HTTP requests.
+    """
+    return BUILTIN_REQUESTS
+
+
+# ---------------------------------------------------------------------------
+# Lookup helpers
+# ---------------------------------------------------------------------------
+
+def get_builtin_profiles() -> dict[str, BrowserProfile]:
+    """Return a shallow copy of the built-in profiles dict (keyed by machine name)."""
+    return dict(_BUILTINS)
+
+
+def get_profile(machine_name: str, store_profiles: dict) -> Optional[BrowserProfile]:
+    """
+    Look up a ``BrowserProfile`` by machine name.
+
+    Stored profiles are checked first so that env-configured built-ins (written
+    by ``preconfigure_browser_profiles_based_on_env``) take priority over the
+    bare module-level defaults.  Falls back to ``_BUILTINS`` when no stored
+    version exists.
+
+    Returns ``None`` when the machine name is unknown or the stored data is
+    corrupt (a warning is logged in the latter case).
+    """
+    raw = store_profiles.get(machine_name)
+    if raw is not None:
+        if isinstance(raw, BrowserProfile):
+            return raw
+        try:
+            return BrowserProfile(**raw)
+        except Exception as exc:
+            logger.warning(f"BrowserProfile '{machine_name}': failed to deserialize — {exc}")
+            # Fall through to built-in
+
+    if machine_name in _BUILTINS:
+        return _BUILTINS[machine_name]
+
+    return None
+
+
+# ---------------------------------------------------------------------------
+# Resolution
+# ---------------------------------------------------------------------------
+
+def resolve_browser_profile(watch, datastore) -> BrowserProfile:
+    """
+    Resolve the effective ``BrowserProfile`` for *watch*.
+
+    Resolution chain
+    ~~~~~~~~~~~~~~~~
+    1. ``watch['browser_profile']`` — explicit machine name set on the watch.
+    2. First tag with ``overrides_watch=True`` that has ``browser_profile`` set.
+    3. ``settings.application['browser_profile']`` — system-wide default.
+    4. Built-in fallback: ``BUILTIN_REQUESTS`` (requests is always the safe default).
+
+    Never raises.  A stale / missing machine-name reference produces a
+    ``logger.warning`` and the resolver continues down the chain.
+    """
+    from changedetectionio.model.resolver import resolve_setting
+
+    store_profiles: dict = datastore.data['settings']['application'].get('browser_profiles', {})
+
+    machine_name = resolve_setting(
+        watch, datastore,
+        field_name='browser_profile',
+        sentinel_values={'system', 'default', ''},
+        default=None,
+        require_tag_override=True,
+    )
+
+    if machine_name:
+        profile = get_profile(machine_name, store_profiles)
+        if profile:
+            return profile
+        logger.warning(
+            f"Watch {watch.get('uuid')!r}: browser_profile {machine_name!r} not found, "
+            f"falling back through the chain"
+        )
+
+    return get_default_browser_builtin()

changedetectionio/model/resolver.py

@@ -0,0 +1,63 @@
+"""
+Unified Watch → Tag → Global settings cascade resolver.
+
+All settings resolution follows the same priority order:
+  1. Watch-level setting (if set and not a sentinel "use parent" value)
+  2. First tag with overrides_watch=True that has the field set
+  3. Global application settings
+  4. Caller-supplied default
+
+This replaces the previously scattered manual resolution loops found in
+notification_service.py, processors/base.py, and the restock processor.
+"""
+
+
+def resolve_setting(watch, datastore, field_name, *,
+                    sentinel_values=None,
+                    default=None,
+                    require_tag_override=True):
+    """
+    Resolve a single setting value by walking the Watch → Tag → Global chain.
+
+    Args:
+        watch:               Watch dict / model object.
+        datastore:           App datastore (must have get_all_tags_for_watch() and
+                             data['settings']['application']).
+        field_name:          The setting key to look up at each level.
+        sentinel_values:     Set of values that mean "not configured here, keep looking".
+                             For example {'system'} for fetch_backend.
+        default:             Value returned when nothing is found in the chain.
+        require_tag_override: If True (default), only tags where overrides_watch=True
+                             contribute to the cascade.  Set to False when every tag
+                             that carries the field should be considered (e.g. for
+                             fields that make sense to merge/override at any tag level).
+
+    Returns:
+        The first non-sentinel, non-empty value found, or *default*.
+    """
+    _sentinels = set(sentinel_values) if sentinel_values else set()
+
+    def _is_unset(v):
+        return v is None or v == '' or v in _sentinels
+
+    # 1. Watch level
+    v = watch.get(field_name)
+    if not _is_unset(v):
+        return v
+
+    # 2. Tag level
+    tags = datastore.get_all_tags_for_watch(uuid=watch.get('uuid'))
+    if tags:
+        for tag in tags.values():
+            if require_tag_override and not tag.get('overrides_watch'):
+                continue
+            v = tag.get(field_name)
+            if not _is_unset(v):
+                return v
+
+    # 3. Global application settings
+    v = datastore.data['settings']['application'].get(field_name)
+    if not _is_unset(v):
+        return v
+
+    return default

changedetectionio/notification_profiles/__init__.py

@@ -0,0 +1,3 @@
+from .registry import registry, NotificationProfileType, AppriseProfileType
+
+__all__ = ['registry', 'NotificationProfileType', 'AppriseProfileType']

changedetectionio/notification_profiles/log.py

@@ -0,0 +1,73 @@
+"""
+Per-profile notification log.
+
+Each profile gets its own log file at:
+  {datastore_path}/notification-logs/{profile_uuid}.log
+
+Entries are stored as JSON-lines (one JSON object per line).
+The file is capped at MAX_ENTRIES lines (oldest pruned first).
+"""
+
+import json
+import os
+from datetime import datetime, timezone
+
+MAX_ENTRIES = 100
+_LOG_DIR = 'notification-logs'
+
+
+def _log_file(datastore_path: str, profile_uuid: str) -> str:
+    return os.path.join(datastore_path, _LOG_DIR, f'{profile_uuid}.log')
+
+
+def write_profile_log(datastore_path: str, profile_uuid: str, *,
+                      watch_url: str = '',
+                      watch_uuid: str = '',
+                      status: str,        # 'ok' | 'error' | 'test'
+                      message: str = ''):
+    """Append one log entry; prune to MAX_ENTRIES."""
+    log_dir = os.path.join(datastore_path, _LOG_DIR)
+    os.makedirs(log_dir, exist_ok=True)
+
+    entry = json.dumps({
+        'ts':         datetime.now(tz=timezone.utc).strftime('%Y-%m-%d %H:%M:%S UTC'),
+        'watch_url':  watch_url[:200],
+        'watch_uuid': watch_uuid,
+        'status':     status,
+        'message':    message[:500],
+    }, ensure_ascii=False)
+
+    path = _log_file(datastore_path, profile_uuid)
+    try:
+        with open(path, 'r', encoding='utf-8') as fh:
+            lines = [l for l in fh.read().splitlines() if l.strip()]
+    except FileNotFoundError:
+        lines = []
+
+    lines.append(entry)
+    lines = lines[-MAX_ENTRIES:]
+
+    with open(path, 'w', encoding='utf-8') as fh:
+        fh.write('\n'.join(lines) + '\n')
+
+
+def read_profile_log(datastore_path: str, profile_uuid: str) -> list:
+    """Return log entries as a list of dicts, newest first."""
+    path = _log_file(datastore_path, profile_uuid)
+    try:
+        with open(path, 'r', encoding='utf-8') as fh:
+            lines = [l.strip() for l in fh if l.strip()]
+    except FileNotFoundError:
+        return []
+
+    entries = []
+    for line in reversed(lines):
+        try:
+            entries.append(json.loads(line))
+        except (json.JSONDecodeError, ValueError):
+            pass
+    return entries
+
+
+def has_log(datastore_path: str, profile_uuid: str) -> bool:
+    return os.path.exists(_log_file(datastore_path, profile_uuid))

changedetectionio/notification_profiles/registry.py

@@ -0,0 +1,111 @@
+"""
+Notification Profile Type plugin registry.
+
+NotificationProfileType is the abstract base — the only contract is send().
+Plugins are free to use any delivery mechanism (Apprise, direct HTTP, SDK, etc.).
+
+Built-in: AppriseProfileType (raw Apprise URL list).
+
+Third-party plugins register additional types:
+
+    from changedetectionio.notification_profiles.registry import registry, NotificationProfileType
+
+    @registry.register
+    class MyProfileType(NotificationProfileType):
+        type_id      = "mytype"
+        display_name = "My Service"
+        icon         = "bell"
+        template     = "my_plugin/notification_profiles/types/mytype.html"
+
+        def send(self, config: dict, n_object: dict, datastore) -> bool:
+            requests.post(config['webhook_url'], json={"text": n_object['notification_body']})
+            return True
+"""
+
+from abc import ABC, abstractmethod
+
+
+class NotificationProfileType(ABC):
+    type_id:      str = NotImplemented
+    display_name: str = NotImplemented
+    icon:         str = "bell"          # feather icon name
+    template:     str = NotImplemented  # Jinja2 partial rendered in the profile edit form
+
+    @abstractmethod
+    def send(self, config: dict, n_object: dict, datastore) -> bool:
+        """
+        Deliver the notification.
+
+        Args:
+            config:    The profile's config dict (type-specific fields).
+            n_object:  Fully-rendered NotificationContextData (title, body, format, etc.).
+            datastore: App datastore for any extra lookups.
+
+        Returns True on success, False on failure (do not raise — log instead).
+        """
+
+    def validate(self, config: dict) -> None:
+        """Raise ValueError with a user-readable message on invalid config."""
+        pass
+
+    def get_url_hint(self, config: dict) -> str:
+        """Short display string shown in the selector chip tooltip / dropdown row."""
+        return ''
+
+
+class AppriseProfileType(NotificationProfileType):
+    """Delivers notifications via Apprise using a raw URL list."""
+
+    type_id      = "apprise"
+    display_name = "Apprise"
+    icon         = "bell"
+    template     = "notification_profiles/types/apprise.html"
+
+    def get_apprise_urls(self, config: dict) -> list:
+        return config.get('notification_urls') or []
+
+    def send(self, config: dict, n_object, datastore) -> bool:
+        from changedetectionio.notification.handler import process_notification
+        from changedetectionio.notification_service import NotificationContextData
+        urls = self.get_apprise_urls(config)
+        if not urls:
+            return False
+        if not isinstance(n_object, NotificationContextData):
+            n_object = NotificationContextData(n_object)
+        n_object['notification_urls']   = urls
+        n_object['notification_title']  = config.get('notification_title') or n_object.get('notification_title')
+        n_object['notification_body']   = config.get('notification_body')  or n_object.get('notification_body')
+        n_object['notification_format'] = config.get('notification_format') or n_object.get('notification_format')
+        process_notification(n_object, datastore)
+        return True
+
+    def get_url_hint(self, config: dict) -> str:
+        urls = config.get('notification_urls') or []
+        if urls:
+            u = urls[0]
+            return (u[:60] + '…') if len(u) > 60 else u
+        return ''
+
+
+class _Registry:
+    def __init__(self):
+        self._types: dict = {}
+
+    def register(self, cls):
+        """Register a NotificationProfileType subclass. Usable as a decorator."""
+        instance = cls()
+        self._types[instance.type_id] = instance
+        return cls
+
+    def get(self, type_id: str) -> NotificationProfileType:
+        return self._types.get(type_id, self._types.get('apprise'))
+
+    def all(self) -> list:
+        return list(self._types.values())
+
+    def choices(self) -> list:
+        return [(t.type_id, t.display_name) for t in self._types.values()]
+
+
+registry = _Registry()
+registry.register(AppriseProfileType)

changedetectionio/notification_profiles/resolver.py

@@ -0,0 +1,49 @@
+"""
+Resolve the full set of NotificationProfile objects that should fire for a given watch.
+
+Merges profile UUIDs from: Watch → Tags → System (union, deduplicated).
+Mute cascade is checked separately via resolve_setting() before calling this.
+"""
+
+from loguru import logger
+
+
+def resolve_notification_profiles(watch, datastore) -> list:
+    """
+    Return list of (profile_dict, NotificationProfileType) tuples to fire for *watch*.
+
+    Profiles are deduplicated by UUID — if the same UUID appears at multiple levels
+    it fires once, not multiple times.
+    """
+    from changedetectionio.notification_profiles.registry import registry
+
+    all_profiles = datastore.data['settings']['application'].get('notification_profile_data', {})
+
+    seen = set()
+    result = []
+
+    def _add(uuids):
+        for uid in (uuids or []):
+            if uid in seen:
+                continue
+            profile = all_profiles.get(uid)
+            if not profile:
+                logger.warning(f"Notification profile UUID {uid!r} not found, skipping")
+                continue
+            seen.add(uid)
+            type_handler = registry.get(profile.get('type', 'apprise'))
+            result.append((profile, type_handler))
+
+    # 1. Watch-level
+    _add(watch.get('notification_profiles', []))
+
+    # 2. Tag/group level
+    tags = datastore.get_all_tags_for_watch(uuid=watch.get('uuid'))
+    if tags:
+        for tag in tags.values():
+            _add(tag.get('notification_profiles', []))
+
+    # 3. System level
+    _add(datastore.data['settings']['application'].get('notification_profiles', []))
+
+    return result

changedetectionio/pluggy_interface.py

@@ -237,14 +237,23 @@ def register_builtin_fetchers():
     This is called from content_fetchers/__init__.py after all fetchers are imported
     to avoid circular import issues.
     """
-    from changedetectionio.content_fetchers import requests, playwright, puppeteer, webdriver_selenium
+    from changedetectionio.content_fetchers import requests, puppeteer, webdriver_selenium
+    from changedetectionio.content_fetchers.playwright import CDP, chrome, firefox, webkit
 
-    # Register each built-in fetcher plugin
     if hasattr(requests, 'requests_plugin'):
         plugin_manager.register(requests.requests_plugin, 'builtin_requests')
 
-    if hasattr(playwright, 'playwright_plugin'):
-        plugin_manager.register(playwright.playwright_plugin, 'builtin_playwright')
+    if hasattr(CDP, 'cdp_plugin'):
+        plugin_manager.register(CDP.cdp_plugin, 'builtin_playwright_cdp')
+
+    if hasattr(chrome, 'chrome_plugin'):
+        plugin_manager.register(chrome.chrome_plugin, 'builtin_playwright_chrome')
+
+    if hasattr(firefox, 'firefox_plugin'):
+        plugin_manager.register(firefox.firefox_plugin, 'builtin_playwright_firefox')
+
+    if hasattr(webkit, 'webkit_plugin'):
+        plugin_manager.register(webkit.webkit_plugin, 'builtin_playwright_webkit')
 
     if hasattr(puppeteer, 'puppeteer_plugin'):
         plugin_manager.register(puppeteer.puppeteer_plugin, 'builtin_puppeteer')
@@ -360,57 +369,28 @@ def get_active_plugins():
 
 
 def get_fetcher_capabilities(watch, datastore):
-    """Get capability flags for a watch's fetcher.
+    """Get capability flags for a watch's resolved fetcher.
 
-    Args:
-        watch: The watch object/dict
-        datastore: The datastore to resolve 'system' fetcher
+    Uses the BrowserProfile resolution chain (watch → tag → global → built-in)
+    to determine the actual fetcher class, then reads its capability flags.
 
     Returns:
-        dict: Dictionary with capability flags:
-            {
-                'supports_browser_steps': bool,
-                'supports_screenshots': bool,
-                'supports_xpath_element_data': bool
-            }
+        dict: {'supports_browser_steps': bool, 'supports_screenshots': bool,
+               'supports_xpath_element_data': bool}
     """
-    # Get the fetcher name from watch
-    fetcher_name = watch.get('fetch_backend', 'system')
-
-    # Resolve 'system' to actual fetcher
-    if fetcher_name == 'system':
-        fetcher_name = datastore.data['settings']['application'].get('fetch_backend', 'html_requests')
-
-    # Get the fetcher class
+    from changedetectionio.model.browser_profile import resolve_browser_profile
     from changedetectionio import content_fetchers
 
-    # Try to get from built-in fetchers first
-    if hasattr(content_fetchers, fetcher_name):
-        fetcher_class = getattr(content_fetchers, fetcher_name)
-        return {
-            'supports_browser_steps': getattr(fetcher_class, 'supports_browser_steps', False),
-            'supports_screenshots': getattr(fetcher_class, 'supports_screenshots', False),
-            'supports_xpath_element_data': getattr(fetcher_class, 'supports_xpath_element_data', False)
-        }
+    profile = resolve_browser_profile(watch, datastore)
+    fetcher_class = content_fetchers.get_fetcher(profile.fetch_backend)
 
-    # Try to get from plugin-provided fetchers
-    # Query all plugins for registered fetchers
-    plugin_fetchers = plugin_manager.hook.register_content_fetcher()
-    for fetcher_registration in plugin_fetchers:
-        if fetcher_registration:
-            name, fetcher_class = fetcher_registration
-            if name == fetcher_name:
-                return {
-                    'supports_browser_steps': getattr(fetcher_class, 'supports_browser_steps', False),
-                    'supports_screenshots': getattr(fetcher_class, 'supports_screenshots', False),
-                    'supports_xpath_element_data': getattr(fetcher_class, 'supports_xpath_element_data', False)
-                }
+    if fetcher_class is None:
+        return {'supports_browser_steps': False, 'supports_screenshots': False, 'supports_xpath_element_data': False}
 
-    # Default: no capabilities
     return {
-        'supports_browser_steps': False,
-        'supports_screenshots': False,
-        'supports_xpath_element_data': False
+        'supports_browser_steps': getattr(fetcher_class, 'supports_browser_steps', False),
+        'supports_screenshots': getattr(fetcher_class, 'supports_screenshots', False),
+        'supports_xpath_element_data': getattr(fetcher_class, 'supports_xpath_element_data', False),
     }
 
 

changedetectionio/processors/base.py

@@ -23,6 +23,7 @@ class difference_detection_processor():
     watch = None
     xpath_data = None
     preferred_proxy = None
+    preferred_proxy_override = None   # Set externally to force a specific proxy (e.g. proxy checker)
     screenshot_format = SCREENSHOT_FORMAT_JPEG
     last_raw_content_checksum = None
 
@@ -36,6 +37,8 @@ class difference_detection_processor():
         # 2. Preserves Watch object with properties (.link, .is_pdf, etc.) - can't use dict()
         # 3. Safe now: Watch.__deepcopy__() shares datastore ref (no memory leak) but copies dict data
         self.watch = deepcopy(self.datastore.data['watching'].get(watch_uuid))
+        if self.watch is None:
+            raise KeyError(f"Watch UUID {watch_uuid} not found in datastore (deleted before processing?)")
 
         # Generic fetcher that should be extended (requests, playwright etc)
         self.fetcher = Fetcher()
@@ -115,82 +118,65 @@ class difference_detection_processor():
                 f"Set ALLOW_IANA_RESTRICTED_ADDRESSES=true to allow."
             )
 
-    async def call_browser(self, preferred_proxy_id=None):
+    async def call_browser(self):
 
         from requests.structures import CaseInsensitiveDict
+        from changedetectionio.model.browser_profile import resolve_browser_profile, BUILTIN_REQUESTS
 
         url = self.watch.link
 
-        # Protect against file:, file:/, file:// access, check the real "link" without any meta "source:" etc prepended.
+        # Protect against file:, file:/, file:// access
         if re.search(r'^file:', url.strip(), re.IGNORECASE):
             if not strtobool(os.getenv('ALLOW_FILE_URI', 'false')):
-                raise Exception(
-                    "file:// type access is denied for security reasons."
-                )
+                raise Exception("file:// type access is denied for security reasons.")
 
         await self.validate_iana_url()
 
-        # Requests, playwright, other browser via wss:// etc, fetch_extra_something
-        prefer_fetch_backend = self.watch.get('fetch_backend', 'system')
+        # Resolve the full browser profile for this watch (watch → tag → global → built-in)
+        profile = resolve_browser_profile(self.watch, self.datastore)
 
-        # Proxy ID "key"
-        preferred_proxy_id = preferred_proxy_id if preferred_proxy_id else self.datastore.get_preferred_proxy_for_watch(
-            uuid=self.watch.get('uuid'))
-
-        # Pluggable content self.fetcher
-        if not prefer_fetch_backend or prefer_fetch_backend == 'system':
-            prefer_fetch_backend = self.datastore.data['settings']['application'].get('fetch_backend')
-
-        # In the case that the preferred fetcher was a browser config with custom connection URL..
-        # @todo - on save watch, if its extra_browser_ then it should be obvious it will use playwright (like if its requests now..)
-        custom_browser_connection_url = None
-        if prefer_fetch_backend.startswith('extra_browser_'):
-            (t, key) = prefer_fetch_backend.split('extra_browser_')
-            connection = list(
-                filter(lambda s: (s['browser_name'] == key), self.datastore.data['settings']['requests'].get('extra_browsers', [])))
-            if connection:
-                prefer_fetch_backend = 'html_webdriver'
-                custom_browser_connection_url = connection[0].get('browser_connection_url')
-
-        # PDF should be html_requests because playwright will serve it up (so far) in a embedded page
+        # PDFs always use the requests fetcher — browsers render them in an embedded viewer
         # @todo https://github.com/dgtlmoon/changedetection.io/issues/2019
-        # @todo needs test to or a fix
         if self.watch.is_pdf:
-            prefer_fetch_backend = "html_requests"
+            profile = BUILTIN_REQUESTS
 
-        # Grab the right kind of 'fetcher', (playwright, requests, etc)
+        # Resolve proxy for the target URL fetch.
+        # Note: browser_connection_url is the WebSocket endpoint to reach the remote browser,
+        # which is separate from the proxy used by the browser to fetch target pages.
+        proxy_url = self.datastore.get_proxy_url_for_watch(self.watch.get('uuid'), override_id=self.preferred_proxy_override)
+        if proxy_url:
+            logger.debug(f"Proxy '{proxy_url}' for {url}")
+
+        logger.debug(f"BrowserProfile '{profile.get_machine_name()}' (fetcher={profile.fetch_backend}) for watch {self.watch['uuid']}")
+
+        # Select the fetcher class
         from changedetectionio import content_fetchers
-        if hasattr(content_fetchers, prefer_fetch_backend):
-            # @todo TEMPORARY HACK - SWITCH BACK TO PLAYWRIGHT FOR BROWSERSTEPS
-            if prefer_fetch_backend == 'html_webdriver' and self.watch.has_browser_steps:
-                # This is never supported in selenium anyway
-                logger.warning(
-                    "Using playwright fetcher override for possible puppeteer request in browsersteps, because puppetteer:browser steps is incomplete.")
-                from changedetectionio.content_fetchers.playwright import fetcher as playwright_fetcher
-                fetcher_obj = playwright_fetcher
-            else:
-                fetcher_obj = getattr(content_fetchers, prefer_fetch_backend)
-        else:
-            # What it referenced doesnt exist, Just use a default
-            fetcher_obj = getattr(content_fetchers, "html_requests")
+        fetcher_class_name = profile.get_fetcher_class_name()
 
-        proxy_url = None
-        if preferred_proxy_id:
-            # Custom browser endpoints should NOT have a proxy added
-            if not prefer_fetch_backend.startswith('extra_browser_'):
-                proxy_url = self.datastore.proxy_list.get(preferred_proxy_id).get('url')
-                logger.debug(f"Selected proxy key '{preferred_proxy_id}' as proxy URL '{proxy_url}' for {url}")
-            else:
-                logger.debug("Skipping adding proxy data when custom Browser endpoint is specified. ")
+        fetcher_obj = content_fetchers.get_fetcher(fetcher_class_name)
+        if fetcher_obj is None:
+            logger.warning(f"Fetcher '{fetcher_class_name}' not found, falling back to requests")
+            fetcher_obj = content_fetchers.get_fetcher('requests')
+        elif self.watch.has_browser_steps and not getattr(fetcher_obj, 'supports_browser_steps', False):
+            # Browser steps require Playwright — override if the resolved fetcher doesn't support them
+            logger.warning(f"Fetcher '{fetcher_class_name}' does not support browser steps, overriding to Playwright")
+            fetcher_obj = content_fetchers.get_fetcher('playwright')
 
-        logger.debug(f"Using proxy '{proxy_url}' for {self.watch['uuid']}")
-
-        # Now call the fetcher (playwright/requests/etc) with arguments that only a fetcher would need.
-        # When browser_connection_url is None, it method should default to working out whats the best defaults (os env vars etc)
-        self.fetcher = fetcher_obj(proxy_override=proxy_url,
-                                   custom_browser_connection_url=custom_browser_connection_url,
-                                   screenshot_format=self.screenshot_format
-                                   )
+        self.fetcher = fetcher_obj(
+            proxy_override=proxy_url,
+            custom_browser_connection_url=profile.browser_connection_url,
+            screenshot_format=self.screenshot_format,
+            # BrowserProfile fields — browser fetchers use these; html_requests ignores them
+            viewport_width=profile.viewport_width,
+            viewport_height=profile.viewport_height,
+            block_images=profile.block_images,
+            block_fonts=profile.block_fonts,
+            profile_user_agent=profile.user_agent,
+            ignore_https_errors=profile.ignore_https_errors,
+            locale=profile.locale,
+            service_workers=profile.service_workers,
+            extra_delay=profile.extra_delay,
+        )
 
         if self.watch.has_browser_steps:
             self.fetcher.browser_steps = browser_steps_get_valid_steps(self.watch.get('browser_steps', []))
@@ -200,9 +186,17 @@ class difference_detection_processor():
         from changedetectionio.jinja2_custom import render as jinja_render
         request_headers = CaseInsensitiveDict()
 
-        ua = self.datastore.data['settings']['requests'].get('default_ua')
-        if ua and ua.get(prefer_fetch_backend):
-            request_headers.update({'User-Agent': ua.get(prefer_fetch_backend)})
+        # Browser profile: UA override (lowest priority — watch headers override this)
+        if profile.user_agent:
+            request_headers['User-Agent'] = profile.user_agent
+
+        # Browser profile: custom headers (override profile UA, but watch headers override these)
+        if profile.custom_headers:
+            for line in profile.custom_headers.splitlines():
+                line = line.strip()
+                if not line.startswith('#') and ':' in line:
+                    k, v = line.split(':', 1)
+                    request_headers[k.strip()] = v.strip()
 
         request_headers.update(self.watch.get('headers', {}))
         request_headers.update(self.datastore.get_all_base_headers())
@@ -259,6 +253,17 @@ class difference_detection_processor():
 
         # @todo .quit here could go on close object, so we can run JS if change-detected
         await self.fetcher.quit(watch=self.watch)
+        self.fetcher.disk_cleanup_after_fetch()
+
+        # Sanitize lone surrogates - these can appear when servers return malformed/mixed-encoding
+        # content that gets decoded into surrogate characters (e.g. \udcad). Without this,
+        # encode('utf-8') raises UnicodeEncodeError downstream in checksums, diffs, file writes, etc.
+        # Covers all fetchers (requests, playwright, puppeteer, selenium) in one place.
+        # Also note: By this point we SHOULD know the original encoding so it can safely convert to utf-8 for the rest of the app.
+        # See: https://github.com/dgtlmoon/changedetection.io/issues/3952
+
+        if self.fetcher.content and isinstance(self.fetcher.content, str):
+            self.fetcher.content = self.fetcher.content.encode('utf-8', errors='replace').decode('utf-8')
 
         # After init, call run_changedetection() which will do the actual change-detection
 

changedetectionio/processors/extract.py

@@ -42,10 +42,7 @@ def render_form(watch, datastore, request, url_for, render_template, flash, redi
     # Get error information for the template
     screenshot_url = watch.get_screenshot()
 
-    system_uses_webdriver = datastore.data['settings']['application']['fetch_backend'] == 'html_webdriver'
-    is_html_webdriver = False
-    if (watch.get('fetch_backend') == 'system' and system_uses_webdriver) or watch.get('fetch_backend') == 'html_webdriver' or watch.get('fetch_backend', '').startswith('extra_browser_'):
-        is_html_webdriver = True
+    fetcher_supports_screenshots = watch.fetcher_supports_screenshots
 
     password_enabled_and_share_is_off = False
     if datastore.data['settings']['application'].get('password') or os.getenv("SALTED_PASS", False):
@@ -62,7 +59,7 @@ def render_form(watch, datastore, request, url_for, render_template, flash, redi
         last_error_screenshot=watch.get_error_snapshot(),
         last_error_text=watch.get_error_text(),
         screenshot=screenshot_url,
-        is_html_webdriver=is_html_webdriver,
+        fetcher_supports_screenshots=fetcher_supports_screenshots,
         password_enabled_and_share_is_off=password_enabled_and_share_is_off,
         extra_title=f" - {watch.label} - Extract Data",
         extra_stylesheets=[url_for('static_content', group='styles', filename='diff.css')],

changedetectionio/processors/magic.py

@@ -100,7 +100,13 @@ class guess_stream_type():
         if any(s in http_content_header for s in RSS_XML_CONTENT_TYPES):
             self.is_rss = True
         elif any(s in http_content_header for s in JSON_CONTENT_TYPES):
-            self.is_json = True
+            # JSONP detection: server claims application/json but content is actually JSONP (e.g. cb({...}))
+            # A JSONP response starts with an identifier followed by '(' - not valid JSON
+            if re.match(r'^\w[\w.]*\s*\(', test_content):
+                logger.warning(f"Content-Type header claims JSON but content looks like JSONP (starts with identifier+parenthesis) - treating as plaintext")
+                self.is_plaintext = True
+            else:
+                self.is_json = True
         elif 'pdf' in magic_content_header:
             self.is_pdf = True
         # magic will call a rss document 'xml'

changedetectionio/processors/restock_diff/__init__.py

@@ -1,6 +1,7 @@
 
 from babel.numbers import parse_decimal
 from changedetectionio.model.Watch import model as BaseWatch
+from decimal import Decimal, InvalidOperation
 from typing import Union
 import re
 
@@ -10,6 +11,8 @@ supports_browser_steps = True
 supports_text_filters_and_triggers = True
 supports_text_filters_and_triggers_elements = True
 supports_request_type = True
+_price_re = re.compile(r"Price:\s*(\d+(?:\.\d+)?)", re.IGNORECASE)
+
 
 class Restock(dict):
 
@@ -31,6 +34,7 @@ class Restock(dict):
 
         if standardized_value:
             # Convert to float
+            # @todo locale needs to be the locale of the webpage
             return float(parse_decimal(standardized_value, locale='en'))
 
         return None
@@ -62,6 +66,17 @@ class Restock(dict):
 
         super().__setitem__(key, value)
 
+def get_price_from_history_str(history_str):
+    m = _price_re.search(history_str)
+    if not m:
+        return None
+
+    try:
+        return str(Decimal(m.group(1)))
+    except InvalidOperation:
+        return None
+
+
 class Watch(BaseWatch):
     def __init__(self, *arg, **kw):
         super().__init__(*arg, **kw)
@@ -75,13 +90,27 @@ class Watch(BaseWatch):
     def extra_notification_token_values(self):
         values = super().extra_notification_token_values()
         values['restock'] = self.get('restock', {})
+
+        values['restock']['previous_price'] = None
+        if self.history_n >= 2:
+            history = self.history
+            if history and len(history) >=2:
+                """Unfortunately for now timestamp is stored as string key"""
+                sorted_keys = sorted(list(history), key=lambda x: int(x))
+                sorted_keys.reverse()
+
+                price_str = self.get_history_snapshot(timestamp=sorted_keys[-1])
+                if price_str:
+                    values['restock']['previous_price'] = get_price_from_history_str(price_str)
         return values
 
     def extra_notification_token_placeholder_info(self):
         values = super().extra_notification_token_placeholder_info()
 
         values.append(('restock.price', "Price detected"))
+        values.append(('restock.in_stock', "In stock status"))
         values.append(('restock.original_price', "Original price at first check"))
+        values.append(('restock.previous_price', "Previous price in history"))
 
         return values
 

changedetectionio/processors/restock_diff/processor.py

@@ -437,17 +437,18 @@ class perform_site_check(difference_detection_processor):
 
         # Only try to process restock information (like scraping for keywords) if the page was actually rendered correctly.
         # Otherwise it will assume "in stock" because nothing suggesting the opposite was found
-        from ...html_tools import html_to_text
-        text = html_to_text(self.fetcher.content)
-        logger.debug(f"Length of text after conversion: {len(text)}")
-        if not len(text):
-            from ...content_fetchers.exceptions import ReplyWithContentButNoText
-            raise ReplyWithContentButNoText(url=watch.link,
-                                            status_code=self.fetcher.get_last_status_code(),
-                                            screenshot=self.fetcher.screenshot,
-                                            html_content=self.fetcher.content,
-                                            xpath_data=self.fetcher.xpath_data
-                                            )
+#useless
+#        from ...html_tools import html_to_text
+#        text = html_to_text(self.fetcher.content)
+#        logger.debug(f"Length of text after conversion: {len(text)}")
+#        if not len(text):
+#            from ...content_fetchers.exceptions import ReplyWithContentButNoText
+#            raise ReplyWithContentButNoText(url=watch.link,
+#                                            status_code=self.fetcher.get_last_status_code(),
+#                                            screenshot=self.fetcher.screenshot,
+#                                            html_content=self.fetcher.content,
+#                                            xpath_data=self.fetcher.xpath_data
+#                                            )
 
         # Which restock settings to compare against?
         # Settings are stored in restock_diff.json (migrated from watch.json by update_30).
@@ -488,19 +489,9 @@ class perform_site_check(difference_detection_processor):
         # @TODO !!! some setting like "Use as fallback" or "always use", "t
         if not (has_price and has_availability) or True:
             from changedetectionio.pluggy_interface import get_itemprop_availability_from_plugin
-            fetcher_name = watch.get('fetch_backend', 'html_requests')
-
-            # Resolve 'system' to the actual fetcher being used
-            # This allows plugins to work even when watch uses "system settings default"
-            if fetcher_name == 'system':
-                # Get the actual fetcher that was used (from self.fetcher)
-                # Fetcher class name gives us the actual backend (e.g., 'html_requests', 'html_webdriver')
-                actual_fetcher = type(self.fetcher).__name__
-                if 'html_requests' in actual_fetcher.lower():
-                    fetcher_name = 'html_requests'
-                elif 'webdriver' in actual_fetcher.lower() or 'playwright' in actual_fetcher.lower():
-                    fetcher_name = 'html_webdriver'
-                logger.debug(f"Resolved 'system' fetcher to actual fetcher: {fetcher_name}")
+            # Use the actual resolved fetcher name from the fetcher instance
+            fetcher_name = self.watch.effective_browser_profile.fetch_backend
+            logger.debug(f"Resolved effective fetcher: {fetcher_name}")
 
             # Try plugin override - plugins can decide if they support this fetcher
             if fetcher_name:

changedetectionio/processors/restock_diff/pure_python_extractor.py

@@ -283,4 +283,7 @@ def query_price_availability(extracted_data):
     if not result.get('availability') and 'availability' in microdata:
         result['availability'] = microdata['availability']
 
+    # result['price'] could be float or str here, depending on the website, for example it might contain "1,00" commas, etc.
+    # using something like babel you need to know the locale of the website and even then it can be problematic
+    # we dont really do anything with the price data so far.. so just accept it the way it comes.
     return result

changedetectionio/processors/text_json_diff/difference.py

@@ -154,11 +154,7 @@ def render(watch, datastore, request, url_for, render_template, flash, redirect,
 
     screenshot_url = watch.get_screenshot()
 
-    system_uses_webdriver = datastore.data['settings']['application']['fetch_backend'] == 'html_webdriver'
-
-    is_html_webdriver = False
-    if (watch.get('fetch_backend') == 'system' and system_uses_webdriver) or watch.get('fetch_backend') == 'html_webdriver' or watch.get('fetch_backend', '').startswith('extra_browser_'):
-        is_html_webdriver = True
+    fetcher_supports_screenshots = watch.fetcher_supports_screenshots
 
     password_enabled_and_share_is_off = False
     if datastore.data['settings']['application'].get('password') or os.getenv("SALTED_PASS", False):
@@ -214,7 +210,7 @@ def render(watch, datastore, request, url_for, render_template, flash, redirect,
                              extra_title=f" - {watch.label} - History",
                              extract_form=extract_form,
                              from_version=str(from_version),
-                             is_html_webdriver=is_html_webdriver,
+                             fetcher_supports_screenshots=fetcher_supports_screenshots,
                              last_error=watch['last_error'],
                              last_error_screenshot=watch.get_error_snapshot(),
                              last_error_text=watch.get_error_text(),

changedetectionio/realtime/events.py

@@ -29,9 +29,11 @@ def register_watch_operation_handlers(socketio, datastore):
             # Perform the operation
             if op == 'pause':
                 watch.toggle_pause()
+                watch.commit()
                 logger.info(f"Socket.IO: Toggled pause for watch {uuid}")
             elif op == 'mute':
                 watch.toggle_mute()
+                watch.commit()
                 logger.info(f"Socket.IO: Toggled mute for watch {uuid}")
             elif op == 'recheck':
                 # Import here to avoid circular imports

changedetectionio/realtime/socket_server.py

@@ -199,8 +199,31 @@ def handle_watch_update(socketio, **kwargs):
         logger.error(f"Socket.IO error in handle_watch_update: {str(e)}")
 
 
+def _suppress_werkzeug_ws_abrupt_disconnect_noise():
+    """Patch BaseWSGIServer.log to suppress the AssertionError traceback that fires when
+    a browser closes a WebSocket connection mid-handshake (e.g. closing a tab).
+    The exception is caught inside run_wsgi and routed to self.server.log() — it never
+    propagates out, so wrapping run_wsgi doesn't help. Patching the log method is the
+    only reliable intercept point. The error is cosmetic: Socket.IO already handles the
+    disconnect correctly via its own disconnect handler and timeout logic."""
+    try:
+        from werkzeug.serving import BaseWSGIServer
+        _original_log = BaseWSGIServer.log
+
+        def _filtered_log(self, type, message, *args):
+            if type == 'error' and 'write() before start_response' in message:
+                return
+            _original_log(self, type, message, *args)
+
+        BaseWSGIServer.log = _filtered_log
+    except Exception:
+        pass
+
+
 def init_socketio(app, datastore):
     """Initialize SocketIO with the main Flask app"""
+    _suppress_werkzeug_ws_abrupt_disconnect_noise()
+
     import platform
     import sys
 

changedetectionio/static/js/realtime.js

@@ -116,6 +116,14 @@ $(document).ready(function () {
                 $('#realtime-conn-error').show();
             });
 
+            // Tell the server we're leaving cleanly so it can release the connection
+            // immediately rather than waiting for a timeout.
+            // Note: this only fires for voluntary closes (tab/window close, navigation away).
+            // Hard kills, crashes and network drops will still timeout normally on the server.
+            window.addEventListener('beforeunload', function () {
+                socket.disconnect();
+            });
+
             socket.on('queue_size', function (data) {
                 console.log(`${data.event_timestamp} - Queue size update: ${data.q_length}`);
                 if(queueSizePagerInfoText) {

changedetectionio/static/js/stepper.js

@@ -4,7 +4,7 @@ $(document).ready(function(){
 });
 
 var checkUserVal = function(){
-    if($('#fetch_backend input:checked').val()=='html_requests') {
+    if($('#fetch_backend input:checked').val()=='requests') {
       $('#request-override').show();
       $('#webdriver-stepper').hide();
     } else {

changedetectionio/static/js/vis.js

@@ -3,21 +3,40 @@ $(document).ready(function () {
     // Lazy Hide/Show elements mechanism
     $('[data-visible-for]').hide();
     function show_related_elem(e) {
-        var n = $(e).attr('name') + "=" + $(e).val();
-        if (n === 'fetch_backend=system') {
+        var name = $(e).attr('name');
+        var val = $(e).val();
+        var n = name + "=" + val;
+
+        // Resolve browser_profile select → underlying fetch_backend class name
+        // browserProfileFetcherMap is injected by the page as {machine_name: 'playwright', ...}
+        if (name && name.endsWith('browser_profile') && typeof browserProfileFetcherMap !== 'undefined') {
+            var fetcherClass = val === 'system'
+                ? (typeof default_system_fetch_backend !== 'undefined' ? default_system_fetch_backend : null)
+                : browserProfileFetcherMap[val];
+            if (fetcherClass) {
+                n = 'fetch_backend=' + fetcherClass;
+            }
+        } else if (n === 'fetch_backend=system') {
             n = "fetch_backend=" + default_system_fetch_backend;
         }
         $(`[data-visible-for~="${n}"]`).show();
     }
-    $(':radio').on('keyup keypress blur change click', function (e) {
+
+    $('select, :radio').on('change', function (e) {
+        $(`[data-visible-for]`).hide();
+        $('.advanced-options').hide();
+        show_related_elem(this);
+    });
+    // Retain original click/keyup handling for radio buttons
+    $(':radio').on('keyup keypress blur click', function (e) {
         $(`[data-visible-for]`).hide();
         $('.advanced-options').hide();
         show_related_elem(this);
     });
 
-    $(':radio:checked').each(function (e) {
+    $(':radio:checked, select').each(function (e) {
        show_related_elem(this);
-    })
+    });
 
 
     // Show advanced
@@ -26,4 +45,4 @@ $(document).ready(function () {
             $(this).toggle();
         })
     });
-});
+});

changedetectionio/store/__init__.py

@@ -143,7 +143,7 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
 
             self.__data['settings']['application']['tags'][uuid] = Tag.model(
                 datastore_path=self.datastore_path,
-                __datastore=self.__data,
+                __datastore=self,
                 default=tag
             )
             logger.info(f"Tag: {uuid} {tag['title']}")
@@ -207,7 +207,7 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
         self.json_store_path = os.path.join(self.datastore_path, "changedetection.json")
 
         # Base definition for all watchers (deepcopy part of #569)
-        self.generic_definition = deepcopy(Watch.model(datastore_path=datastore_path, __datastore=self.__data, default={}))
+        self.generic_definition = deepcopy(Watch.model(datastore_path=datastore_path, __datastore=self, default={}))
 
         # Load build SHA if available (Docker deployments)
         if path.isfile('changedetectionio/source.txt'):
@@ -245,6 +245,10 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
             # Maybe they copied a bunch of watch subdirs across too
             self._load_state()
 
+        # Apply env-var browser config after state is fully loaded so we can safely
+        # read existing settings without risk of being overwritten.
+        self.preconfigure_browser_profiles_based_on_env()
+
     def init_fresh_install(self, include_default_watches, version_tag):
       # Generate app_guid FIRST (required for all operations)
         if "pytest" in sys.modules or "PYTEST_CURRENT_TEST" in os.environ:
@@ -268,13 +272,11 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
         if include_default_watches:
             self.add_watch(
                 url='https://news.ycombinator.com/',
-                tag='Tech news',
-                extras={'fetch_backend': 'html_requests'}
+                tag='Tech news'
             )
             self.add_watch(
                 url='https://changedetection.io/CHANGELOG.txt',
-                tag='changedetection.io',
-                extras={'fetch_backend': 'html_requests'}
+                tag='changedetection.io'
             )
 
         # Create changedetection.json immediately
@@ -331,9 +333,64 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
         if entity.get('processor') != 'text_json_diff':
             logger.trace(f"Loading Watch object '{watch_class.__module__}.{watch_class.__name__}' for UUID {uuid}")
 
-        entity = watch_class(datastore_path=self.datastore_path, __datastore=self.__data, default=entity)
+        entity = watch_class(datastore_path=self.datastore_path, __datastore=self, default=entity)
         return entity
 
+    def preconfigure_browser_profiles_based_on_env(self):
+        """Instantiate browser profiles from environment variables and store them.
+
+        Always runs at the end of reload_state() — covers fresh installs,
+        existing datastores, and server restarts.  Env vars always win so that
+        changing PLAYWRIGHT_DRIVER_URL and restarting is reflected immediately.
+
+        Creates BrowserProfile instances from env vars and stores them in
+        ``settings.application.browser_profiles`` under their machine names,
+        then sets ``settings.application.browser_profile`` to that profile as
+        the system-wide default.
+        """
+        from changedetectionio.model import browser_profile as bp
+        from changedetectionio.strtobool import strtobool
+
+        store_profiles = self.__data['settings']['application'].setdefault('browser_profiles', {})
+        service_workers = os.getenv('PLAYWRIGHT_SERVICE_WORKERS', 'allow')
+        extra_delay = int(os.getenv('WEBDRIVER_DELAY_BEFORE_CONTENT_READY', 0))
+        configured_profile = None
+
+        playwright_url = os.getenv('PLAYWRIGHT_DRIVER_URL')
+        if playwright_url:
+            playwright_url = playwright_url.strip('"')
+            builtin = bp.BUILTIN_PUPPETEER if strtobool(os.getenv('FAST_PUPPETEER_CHROME_FETCHER', 'False')) else bp.BUILTIN_PLAYWRIGHT
+            profile = bp.BrowserProfile(
+                name=builtin.name,
+                fetch_backend=builtin.fetch_backend,
+                browser_connection_url=playwright_url,
+                service_workers=service_workers,
+                extra_delay=extra_delay,
+                is_builtin=True,
+            )
+            logger.debug(f"Configuring browser profile '{profile.get_machine_name()}' from env")
+            store_profiles[profile.get_machine_name()] = profile.model_dump()
+            configured_profile = profile
+
+        webdriver_url = os.getenv('WEBDRIVER_URL')
+        if webdriver_url:
+            profile = bp.BrowserProfile(
+                name=bp.BUILTIN_SELENIUM.name,
+                fetch_backend=bp.BUILTIN_SELENIUM.fetch_backend,
+                browser_connection_url=webdriver_url.strip('"'),
+                extra_delay=extra_delay,
+                is_builtin=True,
+            )
+            logger.debug(f"Configuring browser profile '{profile.get_machine_name()}' from env")
+            store_profiles[profile.get_machine_name()] = profile.model_dump()
+            if not configured_profile:
+                configured_profile = profile
+
+        if configured_profile:
+            logger.debug(f"Setting system default browser profile to '{configured_profile.get_machine_name()}'")
+            self.__data['settings']['application']['browser_profile'] = configured_profile.get_machine_name()
+
+
     # ============================================================================
     # FileSavingDataStore Abstract Method Implementations
     # ============================================================================
@@ -365,6 +422,14 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
         # Is saved as {uuid}/tag.json
         settings_copy['application']['tags'] = {}
 
+        # Serialize BrowserProfile Pydantic instances to plain dicts for JSON storage
+        raw_profiles = settings_copy['application'].get('browser_profiles', {})
+        from changedetectionio.model.browser_profile import BrowserProfile
+        settings_copy['application']['browser_profiles'] = {
+            k: v.model_dump() if isinstance(v, BrowserProfile) else v
+            for k, v in raw_profiles.items()
+        }
+
         return {
             'note': 'Settings file - watches are in {uuid}/watch.json, tags are in {uuid}/tag.json',
             'app_guid': self.__data.get('app_guid'),
@@ -421,7 +486,7 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
 
             return Tag.model(
                 datastore_path=self.datastore_path,
-                __datastore=self.__data,
+                __datastore=self,
                 default=entity_dict
             )
 
@@ -767,7 +832,7 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
 
         # If the processor also has its own Watch implementation
         watch_class = get_custom_watch_obj_for_processor(apply_extras.get('processor'))
-        new_watch = watch_class(datastore_path=self.datastore_path, __datastore=self.__data, url=url)
+        new_watch = watch_class(datastore_path=self.datastore_path, __datastore=self, url=url)
 
         new_uuid = new_watch.get('uuid')
 
@@ -852,6 +917,16 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
 
         return proxy_list if len(proxy_list) else None
 
+    def get_proxy_url_for_watch(self, uuid, override_id=None):
+        """
+        Returns the resolved proxy URL string for a watch, or None.
+        override_id forces a specific proxy (e.g. proxy checker bypass).
+        """
+        proxy_id = override_id or self.get_preferred_proxy_for_watch(uuid)
+        if proxy_id:
+            return self.proxy_list.get(proxy_id, {}).get('url')
+        return None
+
     def get_preferred_proxy_for_watch(self, uuid):
         """
         Returns the preferred proxy by ID key
@@ -885,6 +960,71 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
 
         return None
 
+    # ------------------------------------------------------------------
+    # BrowserProfile helpers
+    # ------------------------------------------------------------------
+
+    def get_browser_profile(self, machine_name: str):
+        """Return a BrowserProfile by machine name, or None if not found.
+
+        Built-in profiles (direct_http_requests, browser_chromeplaywright) are
+        always available and checked first.
+        """
+        from changedetectionio.model.browser_profile import get_profile
+        store_profiles = self.data['settings']['application'].get('browser_profiles', {})
+        return get_profile(machine_name, store_profiles)
+
+    def delete_browser_profile(self, machine_name: str):
+        """Delete a user-defined BrowserProfile by machine name.
+
+        Rules enforced:
+        - Built-in profiles cannot be deleted.
+        - The profile cannot be the current system default
+          (settings.application.browser_profile); caller must change the
+          default first.
+        - Any watch or tag that referenced this profile is reset to None
+          (falls back through the chain on next fetch).
+
+        Returns the number of watches/tags that were reset.
+        """
+        from changedetectionio.model.browser_profile import RESERVED_MACHINE_NAMES
+
+        if machine_name in RESERVED_MACHINE_NAMES:
+            raise ValueError(f"Built-in profile '{machine_name}' cannot be deleted")
+
+        system_default = self.data['settings']['application'].get('browser_profile')
+        if system_default == machine_name:
+            raise ValueError(
+                f"Profile '{machine_name}' is the system default. "
+                f"Change the system default before deleting it."
+            )
+
+        store_profiles = self.data['settings']['application'].get('browser_profiles', {})
+        if machine_name not in store_profiles:
+            return 0
+
+        del store_profiles[machine_name]
+
+        reset_count = 0
+
+        # Reset watches that reference this profile
+        for uuid, watch in self.data['watching'].items():
+            if watch.get('browser_profile') == machine_name:
+                watch['browser_profile'] = None
+                watch.commit()
+                reset_count += 1
+
+        # Reset tags that reference this profile
+        for tag_uuid, tag in self.data['settings']['application'].get('tags', {}).items():
+            if tag.get('browser_profile') == machine_name:
+                tag['browser_profile'] = None
+                tag.commit()
+                reset_count += 1
+
+        self._save_settings()
+        logger.info(f"Deleted BrowserProfile '{machine_name}', reset {reset_count} watches/tags")
+        return reset_count
+
     @property
     def has_extra_headers_file(self):
         filepath = os.path.join(self.datastore_path, 'headers.txt')
@@ -962,7 +1102,7 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
             from ..model import Tag
             new_tag = Tag.model(
                 datastore_path=self.datastore_path,
-                __datastore=self.__data,
+                __datastore=self,
                 default={
                     'title': title.strip(),
                     'date_created': int(time.time())

changedetectionio/store/updates.py

@@ -15,6 +15,7 @@ import tarfile
 import time
 from loguru import logger
 from copy import deepcopy
+from typing import Optional
 
 
 # Try to import orjson for faster JSON serialization
@@ -730,6 +731,144 @@ class DatastoreUpdatesMixin:
         # (left this out by accident in previous update, added tags={} in the changedetection.json save_to_disk)
         self._save_settings()
 
+    def update_31(self):
+        """
+        Migrate legacy ``fetch_backend`` strings to the new ``browser_profile``
+        machine-name system.
+
+        What this migration does
+        ------------------------
+        1. ``settings.requests.extra_browsers`` entries are converted into
+           ``BrowserProfile`` objects and stored in
+           ``settings.application.browser_profiles`` keyed by machine name.
+
+        2. ``settings.application.fetch_backend`` (the system-wide default) is
+           translated to a machine name and written to
+           ``settings.application.browser_profile``.
+
+        3. Every watch that has an explicit ``fetch_backend`` (not ``'system'``)
+           gets a corresponding ``browser_profile`` machine name set, then
+           ``fetch_backend`` is reset to ``'system'``.
+
+        4. The same translation is applied to tags with ``overrides_watch=True``
+           that carry an explicit ``fetch_backend``.
+
+        Legacy mapping
+        ~~~~~~~~~~~~~~
+        * ``'html_requests'``        → built-in ``'direct_http_requests'``
+        * ``'html_webdriver'``       → built-in ``'browser_chromeplaywright'``
+        * ``'extra_browser_<name>'`` → machine name of the migrated custom profile
+        * ``'system'`` / missing     → ``None``  (continue to use chain resolution)
+
+        Safe to re-run: skips watches / tags that already have ``browser_profile``
+        set, and skips extra_browser entries that have already been migrated.
+        """
+        from ..model.browser_profile import (
+            BrowserProfile,
+            BUILTIN_REQUESTS,
+            BUILTIN_BROWSER,
+        )
+
+        app_settings = self.data['settings']['application']
+
+        # ------------------------------------------------------------------
+        # 1. Migrate extra_browsers → browser_profiles
+        # ------------------------------------------------------------------
+        extra_browsers = self.data['settings']['requests'].get('extra_browsers', [])
+        browser_profiles: dict = app_settings.setdefault('browser_profiles', {})
+
+        extra_browser_name_to_machine: dict[str, str] = {}
+
+        for entry in extra_browsers:
+            browser_name = entry.get('browser_name', '').strip()
+            connection_url = entry.get('browser_connection_url', '').strip()
+            if not browser_name:
+                continue
+
+            profile = BrowserProfile(
+                name=browser_name,
+                fetch_backend='playwright_cdp',
+                browser_connection_url=connection_url or None,
+            )
+            machine_name = profile.get_machine_name()
+
+            if machine_name not in browser_profiles:
+                browser_profiles[machine_name] = profile.model_dump()
+                logger.info(f"update_31: migrated extra_browser '{browser_name}' → profile '{machine_name}'")
+
+            extra_browser_name_to_machine[browser_name] = machine_name
+
+        # ------------------------------------------------------------------
+        # Helper: translate a fetch_backend string to a machine name
+        # ------------------------------------------------------------------
+        builtin_requests_name = BUILTIN_REQUESTS.get_machine_name()
+        builtin_browser_name  = BUILTIN_BROWSER.get_machine_name()
+
+        def _to_machine_name(fetch_backend: str) -> Optional[str]:
+            if not fetch_backend or fetch_backend in ('system', 'default', ''):
+                return None
+            if fetch_backend.startswith('extra_browser_'):
+                key = fetch_backend[len('extra_browser_'):]
+                return extra_browser_name_to_machine.get(key)
+            # Strip legacy html_ prefix then query the fetcher registry
+            from changedetectionio import content_fetchers as cf
+            clean = fetch_backend[5:] if fetch_backend.startswith('html_') else fetch_backend
+            fetcher_cls = cf.get_fetcher(clean)
+            if fetcher_cls is None:
+                logger.warning(f"update_31: unknown fetch_backend value {fetch_backend!r}, skipping")
+                return None
+            if fetcher_cls.supports_screenshots:
+                return builtin_browser_name
+            return builtin_requests_name
+
+        # ------------------------------------------------------------------
+        # 2. Migrate system-wide default
+        # ------------------------------------------------------------------
+        system_fetch_backend = app_settings.get('fetch_backend', 'requests')
+        if not app_settings.get('browser_profile'):
+            machine = _to_machine_name(system_fetch_backend)
+            app_settings['browser_profile'] = machine
+            logger.info(
+                f"update_31: system fetch_backend '{system_fetch_backend}' → browser_profile '{machine}'"
+            )
+
+        # ------------------------------------------------------------------
+        # 3. Migrate watches
+        # ------------------------------------------------------------------
+        for uuid, watch in self.data['watching'].items():
+            if watch.get('browser_profile'):
+                continue  # already migrated
+
+            fetch_backend = watch.get('fetch_backend', 'system')
+            machine = _to_machine_name(fetch_backend)
+            watch['browser_profile'] = machine
+            watch['fetch_backend'] = 'system'  # clear legacy value
+            watch.commit()
+            if machine:
+                logger.info(
+                    f"update_31: watch {uuid} fetch_backend '{fetch_backend}' → browser_profile '{machine}'"
+                )
+
+        # ------------------------------------------------------------------
+        # 4. Migrate tags
+        # ------------------------------------------------------------------
+        for tag_uuid, tag in app_settings.get('tags', {}).items():
+            if tag.get('browser_profile'):
+                continue  # already migrated
+
+            fetch_backend = tag.get('fetch_backend', 'system')
+            machine = _to_machine_name(fetch_backend)
+            if machine:
+                tag['browser_profile'] = machine
+                tag['fetch_backend'] = 'system'
+                tag.commit()
+                logger.info(
+                    f"update_31: tag {tag_uuid} fetch_backend '{fetch_backend}' → browser_profile '{machine}'"
+                )
+
+        self._save_settings()
+        logger.success("update_31: fetch_backend → browser_profile migration complete")
+
     def update_30(self):
         """Migrate restock_settings out of watch.json into restock_diff.json processor config file.
 

changedetectionio/templates/_notification_profiles_selector.html

@@ -0,0 +1,208 @@
+{#
+  Notification Profile Selector widget.
+
+  Usage:
+    {% from '_notification_profiles_selector.html' import render_notification_profile_selector %}
+    {{ render_notification_profile_selector(
+        own_profiles=watch.get('notification_profiles', []),
+        inherited_profiles=inherited_notification_profiles,
+        all_profile_data=settings_application.get('notification_profile_data', {}),
+        registry=registry
+    ) }}
+
+  own_profiles       — list of UUIDs directly linked to this watch/group
+  inherited_profiles — list of (uuid, origin_label) tuples from parent groups/system
+  all_profile_data   — dict of uuid→profile from settings.application.notification_profile_data
+  registry           — notification_profiles.registry instance
+#}
+
+{% macro render_notification_profile_selector(own_profiles, inherited_profiles, all_profile_data, registry) %}
+<div class="notification-profile-selector" id="notification-profile-selector">
+
+  {# Hidden inputs — one per selected UUID, submitted with the form #}
+  <div id="np-hidden-inputs">
+    {% for uid in own_profiles %}
+    <input type="hidden" name="notification_profiles" value="{{ uid }}">
+    {% endfor %}
+  </div>
+
+  <div class="np-chips" id="np-chips">
+
+    {# Own profiles — solid chips, removable #}
+    {% for uid in own_profiles %}
+      {% set profile = all_profile_data.get(uid) %}
+      {% if profile %}
+        {% set handler = registry.get(profile.get('type', 'apprise')) %}
+        <span class="np-chip np-chip-own" data-uuid="{{ uid }}"
+              title="{{ handler.get_url_hint(profile.get('config', {})) }}">
+          <i data-feather="{{ handler.icon }}" class="np-chip-icon"></i>
+          <span class="np-chip-name">{{ profile.get('name', uid) }}</span>
+          <span class="np-chip-remove" data-uuid="{{ uid }}" title="{{ _('Remove') }}">×</span>
+        </span>
+      {% endif %}
+    {% endfor %}
+
+    {# Inherited profiles — dimmed, read-only, show origin #}
+    {% for uid, origin_label in (inherited_profiles or []) %}
+      {% if uid not in own_profiles %}
+        {% set profile = all_profile_data.get(uid) %}
+        {% if profile %}
+          {% set handler = registry.get(profile.get('type', 'apprise')) %}
+          <span class="np-chip np-chip-inherited"
+                title="{{ _('Inherited from') }}: {{ origin_label }} — {{ handler.get_url_hint(profile.get('config', {})) }}">
+            <i data-feather="{{ handler.icon }}" class="np-chip-icon"></i>
+            <span class="np-chip-name">{{ profile.get('name', uid) }}</span>
+            <i data-feather="lock" class="np-chip-lock"></i>
+          </span>
+        {% endif %}
+      {% endif %}
+    {% endfor %}
+
+    {# Add button + dropdown #}
+    <div class="np-add-wrapper" id="np-add-wrapper">
+      <button type="button" class="np-add-btn pure-button button-xsmall" id="np-add-btn">
+        <i data-feather="plus"></i> {{ _('Add profile') }}
+      </button>
+
+      <div class="np-dropdown" id="np-dropdown" style="display:none;">
+        <input type="text" class="np-search" id="np-search" placeholder="{{ _('Search profiles…') }}" autocomplete="off">
+        <div class="np-options" id="np-options">
+          {% set has_options = [] %}
+          {% for uid, profile in all_profile_data.items() %}
+            {% if uid not in own_profiles %}
+              {% set handler = registry.get(profile.get('type', 'apprise')) %}
+              {% set hint = handler.get_url_hint(profile.get('config', {})) %}
+              <div class="np-option" data-uuid="{{ uid }}"
+                   data-name="{{ profile.get('name', '') }}"
+                   data-icon="{{ handler.icon }}"
+                   data-hint="{{ hint }}">
+                <i data-feather="{{ handler.icon }}" class="np-option-icon"></i>
+                <span class="np-option-text">
+                  <strong class="np-option-name">{{ profile.get('name', uid) }}</strong>
+                  {% if hint %}<small class="np-option-hint">{{ hint }}</small>{% endif %}
+                </span>
+              </div>
+              {% if has_options.append(1) %}{% endif %}
+            {% endif %}
+          {% endfor %}
+          {% if not has_options %}
+          <div class="np-option np-no-results" style="pointer-events:none; color: var(--color-grey-600);">
+            {{ _('No other profiles available') }}
+          </div>
+          {% endif %}
+          <div class="np-no-match" style="display:none; padding: 8px 12px; color: var(--color-grey-600); font-size: 0.85em;">
+            {{ _('No profiles match') }}
+          </div>
+        </div>
+        <a href="{{ url_for('notification_profiles.edit') }}" class="np-create-new">
+          <i data-feather="plus-circle"></i> {{ _('Create new profile') }}
+        </a>
+      </div>
+    </div>
+
+  </div>{# .np-chips #}
+
+  {% if not own_profiles and not inherited_profiles %}
+  <p class="pure-form-message-inline" style="margin: 4px 0 0 0; color: var(--color-grey-600);">
+    {{ _('No notification profiles linked. Notifications will not be sent for this watch.') }}
+  </p>
+  {% endif %}
+
+</div>{# .notification-profile-selector #}
+
+<script>
+(function() {
+    var selector = document.getElementById('notification-profile-selector');
+    if (!selector) return;
+
+    var addBtn     = selector.querySelector('#np-add-btn');
+    var dropdown   = selector.querySelector('#np-dropdown');
+    var search     = selector.querySelector('#np-search');
+    var chips      = selector.querySelector('#np-chips');
+    var hiddenWrap = selector.querySelector('#np-hidden-inputs');
+    var noMatch    = selector.querySelector('.np-no-match');
+
+    // Toggle dropdown
+    addBtn.addEventListener('click', function(e) {
+        e.stopPropagation();
+        var open = dropdown.style.display !== 'none';
+        dropdown.style.display = open ? 'none' : 'block';
+        if (!open) { search.value = ''; filterOptions(''); search.focus(); }
+    });
+
+    // Close on outside click
+    document.addEventListener('click', function(e) {
+        if (!selector.contains(e.target)) dropdown.style.display = 'none';
+    });
+
+    // Search filter
+    search.addEventListener('input', function() { filterOptions(this.value.toLowerCase()); });
+
+    function filterOptions(q) {
+        var opts = selector.querySelectorAll('.np-option:not(.np-no-results)');
+        var visible = 0;
+        opts.forEach(function(opt) {
+            var match = !q || opt.dataset.name.toLowerCase().indexOf(q) !== -1
+                           || (opt.dataset.hint || '').toLowerCase().indexOf(q) !== -1;
+            opt.style.display = match ? '' : 'none';
+            if (match) visible++;
+        });
+        noMatch.style.display = (visible === 0 && q) ? 'block' : 'none';
+    }
+
+    // Add profile
+    selector.querySelectorAll('.np-option:not(.np-no-results)').forEach(function(opt) {
+        opt.addEventListener('click', function() {
+            var uuid = this.dataset.uuid;
+            var name = this.dataset.name;
+            var icon = this.dataset.icon;
+            var hint = this.dataset.hint;
+
+            // Add hidden input
+            var inp = document.createElement('input');
+            inp.type = 'hidden'; inp.name = 'notification_profiles'; inp.value = uuid;
+            hiddenWrap.appendChild(inp);
+
+            // Add chip (before the add-wrapper)
+            var chip = document.createElement('span');
+            chip.className = 'np-chip np-chip-own';
+            chip.dataset.uuid = uuid;
+            chip.title = hint || '';
+            chip.innerHTML = '<i data-feather="' + icon + '" class="np-chip-icon"></i>'
+                           + '<span class="np-chip-name">' + escHtml(name) + '</span>'
+                           + '<span class="np-chip-remove" data-uuid="' + uuid + '" title="{{ _("Remove") }}">×</span>';
+            chips.insertBefore(chip, selector.querySelector('#np-add-wrapper'));
+            chip.querySelector('.np-chip-remove').addEventListener('click', removeChip);
+
+            // Hide this option in dropdown
+            this.style.display = 'none';
+
+            dropdown.style.display = 'none';
+            if (window.feather) feather.replace();
+        });
+    });
+
+    // Remove chip
+    selector.querySelectorAll('.np-chip-remove').forEach(function(btn) {
+        btn.addEventListener('click', removeChip);
+    });
+
+    function removeChip() {
+        var uuid = this.dataset.uuid;
+        var chip = selector.querySelector('.np-chip-own[data-uuid="' + uuid + '"]');
+        if (chip) chip.remove();
+        var inp = hiddenWrap.querySelector('input[value="' + uuid + '"]');
+        if (inp) inp.remove();
+        // Re-show in dropdown
+        var opt = selector.querySelector('.np-option[data-uuid="' + uuid + '"]');
+        if (opt) opt.style.display = '';
+    }
+
+    function escHtml(s) {
+        return s.replace(/&/g,'&amp;').replace(/</g,'&lt;').replace(/>/g,'&gt;').replace(/"/g,'&quot;');
+    }
+
+    if (window.feather) feather.replace();
+})();
+</script>
+{% endmacro %}

changedetectionio/tests/conftest.py

@@ -2,6 +2,7 @@
 import psutil
 import time
 from threading import Thread
+import multiprocessing
 
 import pytest
 import arrow
@@ -191,6 +192,34 @@ def cleanup(datastore_path):
             if os.path.isfile(f):
                 os.unlink(f)
 
+def pytest_configure(config):
+    """Configure pytest environment before tests run.
+
+    CRITICAL: Set multiprocessing start method to 'fork' for Python 3.14+ compatibility.
+
+    Python 3.14 changed the default start method from 'fork' to 'forkserver' on Linux.
+    The forkserver method requires all objects to be picklable, but pytest-flask's
+    LiveServer uses nested functions that can't be pickled.
+
+    Setting 'fork' explicitly:
+    - Maintains compatibility with Python 3.10-3.13 (where 'fork' was already default)
+    - Fixes Python 3.14 pickling errors
+    - Only affects Unix-like systems (Windows uses 'spawn' regardless)
+
+    See: https://github.com/python/cpython/issues/126831
+    See: https://docs.python.org/3/whatsnew/3.14.html
+    """
+    # Only set if not already set (respects existing configuration)
+    if multiprocessing.get_start_method(allow_none=True) is None:
+        try:
+            # 'fork' is available on Unix-like systems (Linux, macOS)
+            # On Windows, this will have no effect as 'spawn' is the only option
+            multiprocessing.set_start_method('fork', force=False)
+            logger.debug("Set multiprocessing start method to 'fork' for Python 3.14+ compatibility")
+        except (ValueError, RuntimeError):
+            # Already set, not available on this platform, or context already created
+            pass
+
 def pytest_addoption(parser):
     """Add custom command-line options for pytest.
 

changedetectionio/tests/custom_browser_url/test_custom_browser_url.py

@@ -4,30 +4,54 @@ import os
 from flask import url_for
 from ..util import live_server_setup, wait_for_all_checks
 
-def do_test(client, live_server, make_test_use_extra_browser=False):
+CUSTOM_PROFILE_NAME = 'Custom Browser URL'
+CUSTOM_PROFILE_MACHINE_NAME = 'custom_browser_url'
+CUSTOM_BROWSER_WS = 'ws://sockpuppetbrowser-custom-url:3000'
 
-    # Grep for this string in the logs?
-    test_url = "https://changedetection.io/ci-test.html?non-custom-default=true"
-    # "non-custom-default" should not appear in the custom browser connection
-    custom_browser_name = 'custom browser URL'
+
+def create_custom_browser_profile(client):
+    """Create a browser profile that uses the custom sockpuppet container."""
+    res = client.post(
+        url_for("settings.settings_browsers.save"),
+        data={
+            "name": CUSTOM_PROFILE_NAME,
+            "fetch_backend": "playwright_cdp",
+            "browser_connection_url": CUSTOM_BROWSER_WS,
+            "viewport_width": 1280,
+            "viewport_height": 1000,
+            "block_images": "",
+            "block_fonts": "",
+            "ignore_https_errors": "",
+            "user_agent": "",
+            "locale": "",
+            "original_machine_name": "",
+        },
+        follow_redirects=True
+    )
+    assert b"saved." in res.data, f"Expected profile save confirmation, got: {res.data[:500]}"
+
+
+def do_test(client, live_server, make_test_use_extra_browser=False):
 
     # needs to be set and something like 'ws://127.0.0.1:3000'
     assert os.getenv('PLAYWRIGHT_DRIVER_URL'), "Needs PLAYWRIGHT_DRIVER_URL set for this test"
 
-    #####################
+    test_url = "https://changedetection.io/ci-test.html?non-custom-default=true"
+
+    # preconfigure_browser_profiles_based_on_env() already set the correct system default
     res = client.post(
         url_for("settings.settings_page"),
-        data={"application-empty_pages_are_a_change": "",
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_webdriver",
-              'requests-extra_browsers-0-browser_connection_url': 'ws://sockpuppetbrowser-custom-url:3000',
-              'requests-extra_browsers-0-browser_name': custom_browser_name
-              },
+        data={
+            "application-empty_pages_are_a_change": "",
+            "requests-time_between_check-minutes": 180,
+        },
         follow_redirects=True
     )
-
     assert b"Settings updated." in res.data
 
+    # Create the custom browser profile
+    create_custom_browser_profile(client)
+
     # Add our URL to the import page
     uuid = client.application.config.get('DATASTORE').add_watch(url=test_url)
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
@@ -35,23 +59,24 @@ def do_test(client, live_server, make_test_use_extra_browser=False):
 
     if make_test_use_extra_browser:
 
-        # So the name should appear in the edit page under "Request" > "Fetch Method"
+        # The custom profile name should appear in the edit page under "Request" tab
         res = client.get(
             url_for("ui.ui_edit.edit_page", uuid="first"),
             follow_redirects=True
         )
-        assert b'custom browser URL' in res.data
+        assert CUSTOM_PROFILE_NAME.encode() in res.data, \
+            f"Expected '{CUSTOM_PROFILE_NAME}' in edit page fetch method choices"
 
         res = client.post(
             url_for("ui.ui_edit.edit_page", uuid="first"),
             data={
-                # 'run_customer_browser_url_tests.sh' will search for this string to know if we hit the right browser container or not
-                  "url": "https://changedetection.io/ci-test.html?custom-browser-search-string=1",
-                  "tags": "",
-                  "headers": "",
-                  'fetch_backend': f"extra_browser_{custom_browser_name}",
-                  'webdriver_js_execute_code': '',
-                  "time_between_check_use_default": "y"
+                # 'run_custom_browser_url_tests.sh' will grep for this string in the custom container logs
+                "url": "https://changedetection.io/ci-test.html?custom-browser-search-string=1",
+                "tags": "",
+                "headers": "",
+                "browser_profile": CUSTOM_PROFILE_MACHINE_NAME,
+                "webdriver_js_execute_code": "",
+                "time_between_check_use_default": "y"
             },
             follow_redirects=True
         )
@@ -74,12 +99,10 @@ def do_test(client, live_server, make_test_use_extra_browser=False):
 
 # Requires playwright to be installed
 def test_request_via_custom_browser_url(client, live_server, measure_memory_usage, datastore_path):
-   #  live_server_setup(live_server) # Setup on conftest per function
     # We do this so we can grep the logs of the custom container and see if the request actually went through that container
     do_test(client, live_server, make_test_use_extra_browser=True)
 
 
 def test_request_not_via_custom_browser_url(client, live_server, measure_memory_usage, datastore_path):
-   #  live_server_setup(live_server) # Setup on conftest per function
     # We do this so we can grep the logs of the custom container and see if the request actually went through that container
     do_test(client, live_server, make_test_use_extra_browser=False)

changedetectionio/tests/fetchers/test_content.py

@@ -12,12 +12,13 @@ def test_fetch_webdriver_content(client, live_server, measure_memory_usage, data
     #  live_server_setup(live_server) # Setup on conftest per function
 
     #####################
+    # preconfigure_browser_profiles_based_on_env() already set the correct system default
+    # (playwright or puppeteer depending on FAST_PUPPETEER_CHROME_FETCHER) — no need to override it.
     res = client.post(
         url_for("settings.settings_page"),
         data={
             "application-empty_pages_are_a_change": "",
             "requests-time_between_check-minutes": 180,
-            'application-fetch_backend': "html_webdriver",
             'application-ui-favicons_enabled': "y",
         },
         follow_redirects=True

changedetectionio/tests/fetchers/test_custom_js_before_content.py

@@ -25,7 +25,6 @@ def test_execute_custom_js(client, live_server, measure_memory_usage, datastore_
         data={
             "url": test_url,
             "tags": "",
-            'fetch_backend': "html_webdriver",
             'webdriver_js_execute_code': 'document.querySelector("button[name=test-button]").click();',
             'headers': "testheader: yes\buser-agent: MyCustomAgent",
             "time_between_check_use_default": "y",

changedetectionio/tests/proxy_list/test_multiple_proxy.py

@@ -22,7 +22,7 @@ def test_preferred_proxy(client, live_server, measure_memory_usage, datastore_pa
         url_for("ui.ui_edit.edit_page", uuid="first", unpause_on_save=1),
         data={
                 "include_filters": "",
-                "fetch_backend": 'html_webdriver' if os.getenv('PLAYWRIGHT_DRIVER_URL') else 'html_requests',
+                "browser_profile": "system",
                 "headers": "",
                 "proxy": "proxy-two",
                 "tags": "",

changedetectionio/tests/proxy_list/test_noproxy.py

@@ -22,7 +22,6 @@ def test_noproxy_option(client, live_server, measure_memory_usage, datastore_pat
         data={
             "requests-time_between_check-minutes": 180,
             "application-ignore_whitespace": "y",
-            "application-fetch_backend": "html_requests",
             "requests-extra_proxies-0-proxy_name": "custom-one-proxy",
             "requests-extra_proxies-0-proxy_url": "http://test:awesome@squid-one:3128",
             "requests-extra_proxies-1-proxy_name": "custom-two-proxy",
@@ -57,7 +56,6 @@ def test_noproxy_option(client, live_server, measure_memory_usage, datastore_pat
         url_for("ui.ui_edit.edit_page", uuid=uuid, unpause_on_save=1),
         data={
                 "include_filters": "",
-                "fetch_backend": "html_requests",
                 "headers": "",
                 "proxy": "no-proxy",
                 "tags": "",

changedetectionio/tests/proxy_list/test_proxy_noconnect.py

@@ -21,7 +21,6 @@ def test_proxy_noconnect_custom(client, live_server, measure_memory_usage, datas
         data={
             "requests-time_between_check-minutes": 180,
             "application-ignore_whitespace": "y",
-            "application-fetch_backend": 'html_webdriver' if os.getenv('PLAYWRIGHT_DRIVER_URL') or os.getenv("WEBDRIVER_URL") else 'html_requests',
             "requests-extra_proxies-0-proxy_name": "custom-test-proxy",
             # test:awesome is set in tests/proxy_list/squid-passwords.txt
             "requests-extra_proxies-0-proxy_url": "http://127.0.0.1:3128",
@@ -42,7 +41,7 @@ def test_proxy_noconnect_custom(client, live_server, measure_memory_usage, datas
 
     options = {
         "url": test_url,
-        "fetch_backend": "html_webdriver" if os.getenv('PLAYWRIGHT_DRIVER_URL') or os.getenv("WEBDRIVER_URL") else "html_requests",
+        "browser_profile": "system",
         "proxy": "ui-0custom-test-proxy",
         "time_between_check_use_default": "y",
     }

changedetectionio/tests/proxy_list/test_select_custom_proxy.py

@@ -15,7 +15,6 @@ def test_select_custom(client, live_server, measure_memory_usage, datastore_path
         data={
             "requests-time_between_check-minutes": 180,
             "application-ignore_whitespace": "y",
-            "application-fetch_backend": 'html_webdriver' if os.getenv('PLAYWRIGHT_DRIVER_URL') else 'html_requests',
             "requests-extra_proxies-0-proxy_name": "custom-test-proxy",
             # test:awesome is set in tests/proxy_list/squid-passwords.txt
             "requests-extra_proxies-0-proxy_url": "http://test:awesome@squid-custom:3128",
@@ -59,7 +58,6 @@ def test_custom_proxy_validation(client, live_server, measure_memory_usage, data
         data={
             "requests-time_between_check-minutes": 180,
             "application-ignore_whitespace": "y",
-            "application-fetch_backend": 'html_requests',
             "requests-extra_proxies-0-proxy_name": "custom-test-proxy",
             "requests-extra_proxies-0-proxy_url": "xxxxhtt/333??p://test:awesome@squid-custom:3128",
         },
@@ -75,7 +73,6 @@ def test_custom_proxy_validation(client, live_server, measure_memory_usage, data
         data={
             "requests-time_between_check-minutes": 180,
             "application-ignore_whitespace": "y",
-            "application-fetch_backend": 'html_requests',
             "requests-extra_proxies-0-proxy_name": "custom-test-proxy",
             "requests-extra_proxies-0-proxy_url": "https://",
         },

changedetectionio/tests/proxy_socks5/test_socks5_proxy.py

@@ -29,7 +29,6 @@ def test_socks5(client, live_server, measure_memory_usage, datastore_path):
         data={
             "requests-time_between_check-minutes": 180,
             "application-ignore_whitespace": "y",
-            "application-fetch_backend": "html_requests",
             # set in .github/workflows/test-only.yml
             "requests-extra_proxies-0-proxy_url": "socks5://proxy_user123:proxy_pass123@socks5proxy:1080",
             "requests-extra_proxies-0-proxy_name": "socks5proxy",
@@ -61,7 +60,7 @@ def test_socks5(client, live_server, measure_memory_usage, datastore_path):
         url_for("ui.ui_edit.edit_page", uuid="first", unpause_on_save=1),
         data={
             "include_filters": "",
-            "fetch_backend": 'html_webdriver' if os.getenv('PLAYWRIGHT_DRIVER_URL') else 'html_requests',
+            "browser_profile": "system",
             "headers": "",
             "proxy": "ui-0socks5proxy",
             "tags": "",

changedetectionio/tests/proxy_socks5/test_socks5_proxy_sources.py

@@ -48,7 +48,7 @@ def test_socks5_from_proxiesjson_file(client, live_server, measure_memory_usage,
         url_for("ui.ui_edit.edit_page", uuid="first", unpause_on_save=1),
         data={
             "include_filters": "",
-            "fetch_backend": 'html_webdriver' if os.getenv('PLAYWRIGHT_DRIVER_URL') else 'html_requests',
+            "browser_profile": "system",
             "headers": "",
             "proxy": "socks5proxy",
             "tags": "",

changedetectionio/tests/restock/test_restock.py

@@ -60,15 +60,14 @@ def test_restock_detection(client, live_server, measure_memory_usage, datastore_
 
 
     #####################
-    # Set this up for when we remove the notification from the watch, it should fallback with these details
+    # preconfigure_browser_profiles_based_on_env() already set the correct system default
     res = client.post(
         url_for("settings.settings_page"),
         data={"application-notification_urls": notification_url,
               "application-notification_title": "fallback-title "+default_notification_title,
               "application-notification_body": "fallback-body "+default_notification_body,
               "application-notification_format": default_notification_format,
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_webdriver"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
     # Add our URL to the import page, because the docker container (playwright/selenium) wont be able to connect to our usual test url

changedetectionio/tests/smtp/test_notification_smtp.py

@@ -56,8 +56,7 @@ def test_check_notification_email_formats_default_HTML(client, live_server, meas
               "application-notification_title": "fallback-title " + default_notification_title,
               "application-notification_body": "some text\nfallback-body<br> " + default_notification_body,
               "application-notification_format": 'html',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
     assert b"Settings updated." in res.data
@@ -126,8 +125,7 @@ def test_check_notification_plaintext_format(client, live_server, measure_memory
               "application-notification_title": "fallback-title {{watch_title}}  {{ diff_added.splitlines()[0] if diff_added else 'diff added didnt split' }}  " + default_notification_title,
               "application-notification_body": f"some text\n" + default_notification_body + f"\nMore output test\n{ALL_MARKUP_TOKENS}",
               "application-notification_format": 'text',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 
@@ -188,8 +186,7 @@ def test_check_notification_html_color_format(client, live_server, measure_memor
               "application-notification_title": "fallback-title {{watch_title}} - diff_added_lines_test : '{{ diff_added.splitlines()[0] if diff_added else 'diff added didnt split' }}' " + default_notification_title,
               "application-notification_body": f"some text\n{default_notification_body}\nMore output test\n{ALL_MARKUP_TOKENS}",
               "application-notification_format": 'htmlcolor',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 
@@ -273,8 +270,7 @@ def test_check_notification_markdown_format(client, live_server, measure_memory_
               "application-notification_title": "fallback-title  diff_added_lines_test : '{{ diff_added.splitlines()[0] if diff_added else 'diff added didnt split' }}' " + default_notification_title,
               "application-notification_body": "*header*\n\nsome text\n" + default_notification_body,
               "application-notification_format": 'markdown',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 
@@ -369,8 +365,7 @@ def test_check_notification_email_formats_default_Text_override_HTML(client, liv
               "application-notification_title": "fallback-title " + default_notification_title,
               "application-notification_body": notification_body,
               "application-notification_format": 'text',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
     assert b"Settings updated." in res.data
@@ -420,7 +415,7 @@ def test_check_notification_email_formats_default_Text_override_HTML(client, liv
         data={
             "url": test_url,
             "notification_format": 'html',
-            'fetch_backend': "html_requests",
+            'browser_profile': "direct_http_requests",
             "time_between_check_use_default": "y"},
         follow_redirects=True
     )
@@ -480,8 +475,7 @@ def test_check_plaintext_document_plaintext_notification_smtp(client, live_serve
               "application-notification_title": "fallback-title " + default_notification_title,
               "application-notification_body": f"{notification_body}\nMore output test\n{ALL_MARKUP_TOKENS}",
               "application-notification_format": 'text',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
     assert b"Settings updated." in res.data
@@ -533,8 +527,7 @@ def test_check_plaintext_document_html_notifications(client, live_server, measur
               "application-notification_title": "fallback-title " + default_notification_title,
               "application-notification_body": f"{notification_body}\nMore output test\n{ALL_MARKUP_TOKENS}",
               "application-notification_format": 'html',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
     assert b"Settings updated." in res.data
@@ -613,8 +606,7 @@ def test_check_plaintext_document_html_color_notifications(client, live_server,
               "application-notification_title": "fallback-title " + default_notification_title,
               "application-notification_body": f"{notification_body}\nMore output test\n{ALL_MARKUP_TOKENS}",
               "application-notification_format": 'htmlcolor',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 
@@ -686,8 +678,7 @@ def test_check_html_document_plaintext_notification(client, live_server, measure
               "application-notification_title": "fallback-title " + default_notification_title,
               "application-notification_body": f"{notification_body}\nMore output test\n{ALL_MARKUP_TOKENS}",
               "application-notification_format": 'text',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 
@@ -740,8 +731,7 @@ def test_check_html_notification_with_apprise_format_is_html(client, live_server
               "application-notification_title": "fallback-title " + default_notification_title,
               "application-notification_body": "some text\nfallback-body<br> " + default_notification_body,
               "application-notification_format": 'html',
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
     assert b"Settings updated." in res.data

changedetectionio/tests/test_access_control.py

@@ -32,8 +32,7 @@ def test_check_access_control(app, client, live_server, measure_memory_usage, da
             url_for("settings.settings_page"),
             data={"application-password": "foobar",
                   "application-shared_diff_access": "True",
-                  "requests-time_between_check-minutes": 180,
-                  'application-fetch_backend': "html_requests"},
+                  "requests-time_between_check-minutes": 180},
             follow_redirects=True
         )
 
@@ -91,8 +90,7 @@ def test_check_access_control(app, client, live_server, measure_memory_usage, da
         res = c.post(
             url_for("settings.settings_page"),
             data={
-                  "requests-time_between_check-minutes": 180,
-                  'application-fetch_backend': "html_requests"},
+                  "requests-time_between_check-minutes": 180},
             follow_redirects=True
         )
 
@@ -127,16 +125,16 @@ def test_check_access_control(app, client, live_server, measure_memory_usage, da
         assert b"IMPORT" in res.data
         assert b"LOG OUT" in res.data
         assert b"time_between_check-minutes" in res.data
-        assert b"fetch_backend" in res.data
+
 
         ##################################################
         # Remove password button, and check that it worked
         ##################################################
+        # preconfigure_browser_profiles_based_on_env() already set the correct system default
         res = c.post(
             url_for("settings.settings_page"),
             data={
                 "requests-time_between_check-minutes": 180,
-                "application-fetch_backend": "html_webdriver",
                 "application-removepassword_button": "Remove password"
             },
             follow_redirects=True,
@@ -150,8 +148,7 @@ def test_check_access_control(app, client, live_server, measure_memory_usage, da
         res = c.post(
             url_for("settings.settings_page"),
             data={"application-password": "",
-                  "requests-time_between_check-minutes": 180,
-                  'application-fetch_backend': "html_requests"},
+                  "requests-time_between_check-minutes": 180},
             follow_redirects=True
         )
 
@@ -164,8 +161,7 @@ def test_check_access_control(app, client, live_server, measure_memory_usage, da
             data={"application-password": "foobar",
                   # Should be disabled
                   "application-shared_diff_access": "",
-                  "requests-time_between_check-minutes": 180,
-                  'application-fetch_backend': "html_requests"},
+                  "requests-time_between_check-minutes": 180},
             follow_redirects=True
         )
 

changedetectionio/tests/test_add_replace_remove_filter.py

@@ -60,7 +60,6 @@ def test_check_removed_line_contains_trigger(client, live_server, measure_memory
         url_for("ui.ui_edit.edit_page", uuid="first"),
         data={"trigger_text": 'The golden line',
               "url": test_url,
-              'fetch_backend': "html_requests",
               'filter_text_removed': 'y',
               "time_between_check_use_default": "y"},
         follow_redirects=True
@@ -127,8 +126,7 @@ def test_check_add_line_contains_trigger(client, live_server, measure_memory_usa
               # https://github.com/caronc/apprise/wiki/Notify_Custom_JSON#get-parameter-manipulation
               "application-notification_urls": test_notification_url,
               "application-notification_format": 'text',
-              "application-minutes_between_check": 180,
-              "application-fetch_backend": "html_requests"
+              "application-minutes_between_check": 180
               },
         follow_redirects=True
     )
@@ -149,7 +147,6 @@ def test_check_add_line_contains_trigger(client, live_server, measure_memory_usa
         data={"trigger_text": 'Oh yes please',
               "url": test_url,
               'processor': 'text_json_diff',
-              'fetch_backend': "html_requests",
               'filter_text_removed': '',
               'filter_text_added': 'y',
               "time_between_check_use_default": "y"},

changedetectionio/tests/test_api.py

@@ -170,6 +170,14 @@ def test_api_simple(client, live_server, measure_memory_usage, datastore_path):
         headers={'x-api-key': api_key},
     )
     assert b'(changed) Which is across' in res.data
+    assert b'Some text thats the same' in res.data
+
+    # Fetch the difference between two versions (default text format)
+    res = client.get(
+        url_for("watchhistorydiff", uuid=watch_uuid, from_timestamp='previous', to_timestamp='latest')+"?changesOnly=true",
+        headers={'x-api-key': api_key},
+    )
+    assert b'Some text thats the same' not in res.data
 
     # Test htmlcolor format
     res = client.get(
@@ -408,7 +416,6 @@ def test_access_denied(client, live_server, measure_memory_usage, datastore_path
         url_for("settings.settings_page"),
         data={
             "requests-time_between_check-minutes": 180,
-            "application-fetch_backend": "html_requests",
             "application-api_access_token_enabled": ""
         },
         follow_redirects=True
@@ -428,7 +435,6 @@ def test_access_denied(client, live_server, measure_memory_usage, datastore_path
         url_for("settings.settings_page"),
         data={
             "requests-time_between_check-minutes": 180,
-            "application-fetch_backend": "html_requests",
             "application-api_access_token_enabled": "y"
         },
         follow_redirects=True
@@ -899,8 +905,7 @@ def test_api_conflict_UI_password(client, live_server, measure_memory_usage, dat
         url_for("settings.settings_page"),
         data={"application-password": "foobar", # password is now set! API should still work!
               "application-api_access_token_enabled": "y",
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 

changedetectionio/tests/test_api_openapi.py

@@ -177,7 +177,6 @@ def test_openapi_validation_get_requests_bypass_validation(client, live_server,
         url_for("settings.settings_page"),
         data={
             "requests-time_between_check-minutes": 180,
-            "application-fetch_backend": "html_requests",
             "application-api_access_token_enabled": ""
         },
         follow_redirects=True

changedetectionio/tests/test_api_tags.py

@@ -178,23 +178,44 @@ def test_api_tags_listing(client, live_server, measure_memory_usage, datastore_p
 
 def test_api_tag_restock_processor_config(client, live_server, measure_memory_usage, datastore_path):
     """
-    Test that a tag/group can be updated with processor_config_restock_diff via the API.
+    Test that a tag/group can be created and updated with processor_config_restock_diff via the API.
     Since Tag extends WatchBase, processor config fields injected into WatchBase are also valid for tags.
     """
     api_key = live_server.app.config['DATASTORE'].data['settings']['application'].get('api_access_token')
 
     set_original_response(datastore_path=datastore_path)
 
-    # Create a tag
+    # Create a tag with processor_config_restock_diff in a single POST (issue #3966)
     res = client.post(
         url_for("tag"),
-        data=json.dumps({"title": "Restock Group"}),
+        data=json.dumps({
+            "title": "Restock Group",
+            "overrides_watch": True,
+            "processor_config_restock_diff": {
+                "in_stock_processing": "in_stock_only",
+                "follow_price_changes": True,
+                "price_change_min": 7777777
+            }
+        }),
         headers={'content-type': 'application/json', 'x-api-key': api_key}
     )
-    assert res.status_code == 201
+    assert res.status_code == 201, f"POST tag with restock config failed: {res.data}"
     tag_uuid = res.json.get('uuid')
 
-    # Update tag with valid processor_config_restock_diff
+    # Verify processor config was saved during creation (the bug: these were discarded)
+    res = client.get(
+        url_for("tag", uuid=tag_uuid),
+        headers={'x-api-key': api_key}
+    )
+    assert res.status_code == 200
+    tag_data = res.json
+    assert tag_data.get('overrides_watch') == True, "overrides_watch should be saved on POST"
+    assert tag_data.get('processor_config_restock_diff', {}).get('in_stock_processing') == 'in_stock_only', \
+        "processor_config_restock_diff should be saved on POST"
+    assert tag_data.get('processor_config_restock_diff', {}).get('price_change_min') == 7777777, \
+        "price_change_min should be saved on POST"
+
+    # Update tag with valid processor_config_restock_diff via PUT
     res = client.put(
         url_for("tag", uuid=tag_uuid),
         headers={'x-api-key': api_key, 'content-type': 'application/json'},

changedetectionio/tests/test_auth.py

@@ -19,7 +19,7 @@ def test_basic_auth(client, live_server, measure_memory_usage, datastore_path):
     # Check form validation
     res = client.post(
         url_for("ui.ui_edit.edit_page", uuid="first"),
-        data={"include_filters": "", "url": test_url, "tags": "", "headers": "", 'fetch_backend': "html_requests", "time_between_check_use_default": "y"},
+        data={"include_filters": "", "url": test_url, "tags": "", "headers": "", "time_between_check_use_default": "y"},
         follow_redirects=True
     )
     assert b"Updated watch." in res.data

changedetectionio/tests/test_backend.py

@@ -48,6 +48,15 @@ def test_check_basic_change_detection_functionality(client, live_server, measure
     # Check this class does not appear (that we didnt see the actual source)
     assert b'foobar-detection' not in res.data
 
+    # Check POST preview
+    res = client.post(
+        url_for("ui.ui_preview.preview_page", uuid="first"),
+        follow_redirects=True
+    )
+    # Check this class does not appear (that we didnt see the actual source)
+    assert b'foobar-detection' not in res.data
+
+
     # Make a change
     set_modified_response(datastore_path=datastore_path)
 
@@ -163,8 +172,7 @@ def test_title_scraper(client, live_server, measure_memory_usage, datastore_path
     res = client.post(
         url_for("settings.settings_page"),
         data={"application-ui-use_page_title_in_list": "",
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 
@@ -206,8 +214,7 @@ def test_requests_timeout(client, live_server, measure_memory_usage, datastore_p
         url_for("settings.settings_page"),
         data={"application-ui-use_page_title_in_list": "",
               "requests-time_between_check-minutes": 180,
-              "requests-timeout": delay - 1,
-              'application-fetch_backend': "html_requests"},
+              "requests-timeout": delay - 1},
         follow_redirects=True
     )
 
@@ -225,8 +232,7 @@ def test_requests_timeout(client, live_server, measure_memory_usage, datastore_p
         url_for("settings.settings_page"),
         data={"application-ui-use_page_title_in_list": "",
               "requests-time_between_check-minutes": 180,
-              "requests-timeout": delay + 1, # timeout should be a second more than the reply time
-              'application-fetch_backend': "html_requests"},
+              "requests-timeout": delay + 1}, # timeout should be a second more than the reply time
         follow_redirects=True
     )
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
@@ -413,3 +419,28 @@ def test_plaintext_even_if_xml_content_and_can_apply_filters(client, live_server
     assert b'<foobar' not in res.data
 
     res = delete_all_watches(client)
+
+
+def test_last_error_cleared_on_same_checksum(client, live_server, datastore_path):
+    """last_error should be cleared even when content is unchanged (checksumFromPreviousCheckWasTheSame path)"""
+    set_original_response(datastore_path=datastore_path)
+
+    uuid = client.application.config.get('DATASTORE').add_watch(url=url_for('test_endpoint', _external=True))
+
+    # First check - establishes baseline checksum
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Inject a stale last_error directly (simulates a prior failed check)
+    datastore = client.application.config.get('DATASTORE')
+    datastore.update_watch(uuid=uuid, update_obj={'last_error': 'Some previous error'})
+    assert datastore.data['watching'][uuid].get('last_error') == 'Some previous error'
+
+    # Second check - same content, so checksumFromPreviousCheckWasTheSame will fire
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # last_error must be cleared even though no change was detected
+    assert datastore.data['watching'][uuid].get('last_error') == False
+
+    delete_all_watches(client)

changedetectionio/tests/test_backup.py

@@ -3,7 +3,7 @@
 from .util import set_original_response, live_server_setup, wait_for_all_checks
 from flask import url_for
 import io
-from zipfile import ZipFile
+from zipfile import ZipFile, ZIP_DEFLATED
 import re
 import time
 from changedetectionio.model import Watch, Tag
@@ -68,6 +68,9 @@ def test_backup(client, live_server, measure_memory_usage, datastore_path):
     # Check for changedetection.json (settings file)
     assert 'changedetection.json' in l, "changedetection.json should be in backup"
 
+    # secret.txt must never be included — it contains the Flask session key
+    assert 'secret.txt' not in l, "secret.txt (Flask session key) must not be included in backup"
+
     # Get the latest one
     res = client.get(
         url_for("backups.remove_backups"),
@@ -196,4 +199,63 @@ def test_backup_restore(client, live_server, measure_memory_usage, datastore_pat
     assert restored_tag2 is not None, f"Tag {tag_uuid2} not found after restore"
     assert restored_tag2['title'] == "Tasty backup tag number two", "Restored tag 2 title does not match"
     assert isinstance(restored_tag2, Tag.model), \
-        f"Tag 2 not properly rehydrated, got {type(restored_tag2)}"
+        f"Tag 2 not properly rehydrated, got {type(restored_tag2)}"
+
+
+def test_backup_restore_zip_slip_rejected(client, live_server, measure_memory_usage, datastore_path):
+    """Zip Slip path traversal entries in a restore zip must be rejected."""
+    import pytest
+    from changedetectionio.blueprint.backups.restore import import_from_zip
+
+    # Build a zip with a path traversal entry that would escape the extraction dir
+    malicious_zip = io.BytesIO()
+    with ZipFile(malicious_zip, 'w') as zf:
+        zf.writestr("../escaped.txt", "ATTACKER-CONTROLLED")
+    malicious_zip.seek(0)
+
+    datastore = live_server.app.config['DATASTORE']
+
+    with pytest.raises(ValueError, match="Zip Slip"):
+        import_from_zip(
+            zip_stream=malicious_zip,
+            datastore=datastore,
+            include_groups=True,
+            include_groups_replace=True,
+            include_watches=True,
+            include_watches_replace=True,
+        )
+
+
+def test_backup_restore_zip_bomb_rejected(client, live_server, measure_memory_usage, datastore_path):
+    """A zip whose total uncompressed size exceeds the limit must be rejected.
+
+    The guard reads file_size from the zip central-directory metadata — no
+    actual decompression happens, so this test is fast and uses minimal RAM.
+    100 KB of zeros compresses to ~100 bytes; monkeypatching the limit to
+    50 KB is enough to trigger the check without creating any large files.
+    """
+    import pytest
+    import changedetectionio.blueprint.backups.restore as restore_mod
+    from changedetectionio.blueprint.backups.restore import import_from_zip
+
+    # ~100 KB of zeros → deflate compresses to ~100 bytes, but file_size metadata = 100 KB
+    bomb_zip = io.BytesIO()
+    with ZipFile(bomb_zip, 'w', compression=ZIP_DEFLATED) as zf:
+        zf.writestr("data.txt", b"\x00" * (100 * 1024))
+    bomb_zip.seek(0)
+
+    datastore = live_server.app.config['DATASTORE']
+    original_limit = restore_mod._MAX_DECOMPRESSED_BYTES
+    try:
+        restore_mod._MAX_DECOMPRESSED_BYTES = 50 * 1024  # 50 KB limit for this test
+        with pytest.raises(ValueError, match="decompressed size"):
+            import_from_zip(
+                zip_stream=bomb_zip,
+                datastore=datastore,
+                include_groups=True,
+                include_groups_replace=True,
+                include_watches=True,
+                include_watches_replace=True,
+            )
+    finally:
+        restore_mod._MAX_DECOMPRESSED_BYTES = original_limit

changedetectionio/tests/test_basic_socketio.py

@@ -118,8 +118,7 @@ def test_everything(live_server, client, measure_memory_usage, datastore_path):
     res = client.post(
         url_for("settings.settings_page"),
         data={"application-password": "foobar",
-              "requests-time_between_check-minutes": 180,
-              'application-fetch_backend': "html_requests"},
+              "requests-time_between_check-minutes": 180},
         follow_redirects=True
     )
 

changedetectionio/tests/test_block_while_text_present.py

@@ -83,7 +83,6 @@ def test_check_block_changedetection_text_NOT_present(client, live_server, measu
         url_for("ui.ui_edit.edit_page", uuid=uuid),
         data={"text_should_not_be_present": ignore_text,
               "url": test_url,
-              'fetch_backend': "html_requests",
               "time_between_check_use_default": "y"
               },
         follow_redirects=True

changedetectionio/tests/test_browser_profile_status_icon.py

@@ -0,0 +1,281 @@
+#!/usr/bin/env python3
+"""
+Tests that the watchlist shows/hides the browser status icon based on the
+effective browser profile, covering the full inheritance chain:
+
+  watch browser_profile → system default browser_profile → direct_http_requests
+"""
+
+import pytest
+from flask import url_for
+
+
+def set_system_default_profile(client, profile_machine_name):
+    res = client.post(
+        url_for('settings.settings_browsers.set_default'),
+        data={'machine_name': profile_machine_name},
+        follow_redirects=True,
+    )
+    assert res.status_code == 200
+
+
+def create_custom_browser_profile(client, name='My Custom Chrome'):
+    """Create a custom browser profile using playwright_cdp and return its machine name."""
+    res = client.post(
+        url_for('settings.settings_browsers.save'),
+        data={
+            'name': name,
+            'fetch_backend': 'playwright_cdp',
+            'browser_connection_url': 'ws://localhost:3000',
+            'viewport_width': 1280,
+            'viewport_height': 1000,
+            'block_images': '',
+            'block_fonts': '',
+            'ignore_https_errors': '',
+            'user_agent': '',
+            'locale': '',
+            'custom_headers': '',
+            'original_machine_name': '',
+        },
+        follow_redirects=True,
+    )
+    assert b'saved.' in res.data
+    from changedetectionio.model.browser_profile import BrowserProfile
+    return BrowserProfile(name=name, fetch_backend='playwright_cdp').get_machine_name()
+
+
+def create_requests_browser_profile(client, name, user_agent='', custom_headers=''):
+    """Create a requests-type browser profile with optional UA and custom headers."""
+    res = client.post(
+        url_for('settings.settings_browsers.save'),
+        data={
+            'name': name,
+            'fetch_backend': 'requests',
+            'browser_connection_url': '',
+            'viewport_width': 1280,
+            'viewport_height': 1000,
+            'block_images': '',
+            'block_fonts': '',
+            'ignore_https_errors': '',
+            'user_agent': user_agent,
+            'locale': '',
+            'custom_headers': custom_headers,
+            'original_machine_name': '',
+        },
+        follow_redirects=True,
+    )
+    assert b'saved.' in res.data
+    from changedetectionio.model.browser_profile import BrowserProfile
+    return BrowserProfile(name=name, fetch_backend='requests').get_machine_name()
+
+
+# ---------------------------------------------------------------------------
+# Unit tests — status_icon attribute on fetcher classes
+# ---------------------------------------------------------------------------
+
+def test_status_icon_on_browser_fetchers():
+    """Browser fetcher classes must declare a status_icon dict."""
+    from changedetectionio.content_fetchers.playwright.CDP import fetcher as playwright_fetcher
+    from changedetectionio.content_fetchers.puppeteer import fetcher as puppeteer_fetcher
+    from changedetectionio.content_fetchers.webdriver_selenium import fetcher as selenium_fetcher
+
+    for cls in (playwright_fetcher, puppeteer_fetcher, selenium_fetcher):
+        assert cls.status_icon is not None, f"{cls} should have status_icon set"
+        assert 'filename' in cls.status_icon
+        assert 'alt' in cls.status_icon
+        assert 'title' in cls.status_icon
+
+
+def test_no_status_icon_on_requests_fetcher():
+    """The plain requests fetcher must have status_icon = None."""
+    from changedetectionio.content_fetchers.requests import fetcher as requests_fetcher
+    assert requests_fetcher.status_icon is None
+
+
+def test_fetcher_status_icons_filter_uses_status_icon(monkeypatch):
+    """fetcher_status_icons filter returns icon HTML for a class with status_icon set."""
+    from changedetectionio import content_fetchers
+
+    class FakeBrowserFetcher:
+        status_icon = {'filename': 'test-icon.png', 'alt': 'Test browser', 'title': 'Test browser'}
+        supports_screenshots = True
+
+    monkeypatch.setitem(content_fetchers.FETCHERS, 'fake_browser', FakeBrowserFetcher)
+
+    from changedetectionio.flask_app import app
+    with app.test_request_context('/'):
+        from changedetectionio.flask_app import _jinja2_filter_fetcher_status_icons
+        result = _jinja2_filter_fetcher_status_icons('fake_browser')
+        assert 'test-icon.png' in result
+        assert 'Test browser' in result
+
+    # Requests fetcher → empty string
+    with app.test_request_context('/'):
+        result = _jinja2_filter_fetcher_status_icons('requests')
+        assert result == ''
+
+
+# ---------------------------------------------------------------------------
+# Integration tests — inheritance chain
+# ---------------------------------------------------------------------------
+
+def test_watch_explicit_browser_profile_shows_icon(client, live_server, measure_memory_usage, datastore_path):
+    """Watch explicitly assigned a browser profile shows the chrome icon,
+    even when the system default is requests."""
+    datastore = client.application.config.get('DATASTORE')
+    set_system_default_profile(client, 'direct_http_requests')
+
+    machine_name = create_custom_browser_profile(client)
+    uuid = datastore.add_watch(url='http://example.com', extras={'browser_profile': machine_name, 'paused': True})
+    res = client.get(url_for('watchlist.index'), follow_redirects=True)
+    assert b'Using a Chrome browser' in res.data, \
+        "Chrome icon should appear when watch is explicitly set to a browser profile"
+
+    datastore.delete(uuid)
+    client.get(url_for('settings.settings_browsers.delete', machine_name=machine_name), follow_redirects=True)
+
+
+def test_watch_explicit_requests_profile_no_icon(client, live_server, measure_memory_usage, datastore_path):
+    """Watch explicitly set to direct_http_requests never shows the chrome icon,
+    even when the system default is a browser."""
+    datastore = client.application.config.get('DATASTORE')
+
+    machine_name = create_custom_browser_profile(client)
+    set_system_default_profile(client, machine_name)
+
+    uuid = datastore.add_watch(url='http://example.com', extras={'browser_profile': 'direct_http_requests', 'paused': True})
+    res = client.get(url_for('watchlist.index'), follow_redirects=True)
+    assert b'Using a Chrome browser' not in res.data, \
+        "Chrome icon should NOT appear when watch is explicitly set to direct_http_requests"
+
+    datastore.delete(uuid)
+    set_system_default_profile(client, 'direct_http_requests')
+    client.get(url_for('settings.settings_browsers.delete', machine_name=machine_name), follow_redirects=True)
+
+
+def test_system_default_requests_inherited_by_watch(client, live_server, measure_memory_usage, datastore_path):
+    """Watch using system default inherits requests → no icon."""
+    datastore = client.application.config.get('DATASTORE')
+    set_system_default_profile(client, 'direct_http_requests')
+
+    uuid = datastore.add_watch(url='http://example.com', extras={'paused': True})
+    res = client.get(url_for('watchlist.index'), follow_redirects=True)
+    assert b'Using a Chrome browser' not in res.data, \
+        "Chrome icon should NOT appear when system default is requests and watch uses system default"
+
+    datastore.delete(uuid)
+
+
+def test_system_default_browser_inherited_by_watch(client, live_server, measure_memory_usage, datastore_path):
+    """Watch using system default inherits a browser profile → icon shown."""
+    datastore = client.application.config.get('DATASTORE')
+
+    machine_name = create_custom_browser_profile(client)
+    set_system_default_profile(client, machine_name)
+
+    uuid = datastore.add_watch(url='http://example.com', extras={'paused': True})
+    res = client.get(url_for('watchlist.index'), follow_redirects=True)
+    assert b'Using a Chrome browser' in res.data, \
+        "Chrome icon should appear when system default is a browser profile and watch uses system default"
+
+    datastore.delete(uuid)
+    set_system_default_profile(client, 'direct_http_requests')
+    client.get(url_for('settings.settings_browsers.delete', machine_name=machine_name), follow_redirects=True)
+
+# ---------------------------------------------------------------------------
+# Integration tests — BrowserProfile UA and custom_headers applied to requests
+# ---------------------------------------------------------------------------
+
+def test_browser_profile_user_agent_applied(client, live_server, measure_memory_usage, datastore_path):
+    """User-Agent set on a BrowserProfile appears in the fetched request;
+    a per-watch User-Agent header overrides it."""
+    from changedetectionio.tests.util import wait_for_all_checks
+
+    datastore = client.application.config.get('DATASTORE')
+    test_url = url_for('test_headers', _external=True)
+
+    machine_name = create_requests_browser_profile(
+        client, name='UA Profile Test', user_agent='profile-ua/2.0'
+    )
+
+    uuid = datastore.add_watch(url=test_url, extras={'browser_profile': machine_name})
+    client.get(url_for('ui.form_watch_checknow'), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    res = client.get(url_for('ui.ui_preview.preview_page', uuid='first'), follow_redirects=True)
+    assert b'profile-ua/2.0' in res.data, "Profile UA should appear in the echoed request headers"
+
+    # Per-watch User-Agent header overrides the profile UA
+    client.post(
+        url_for('ui.ui_edit.edit_page', uuid='first'),
+        data={
+            'url': test_url,
+            'tags': '',
+            'browser_profile': machine_name,
+            'headers': 'User-Agent: watch-ua/3.0',
+            'time_between_check_use_default': 'y',
+        },
+        follow_redirects=True,
+    )
+    client.get(url_for('ui.form_watch_checknow'), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    res = client.get(url_for('ui.ui_preview.preview_page', uuid='first'), follow_redirects=True)
+    assert b'watch-ua/3.0' in res.data, "Watch-level UA should override profile UA"
+    assert b'profile-ua/2.0' not in res.data, "Profile UA should be superseded by watch-level header"
+
+    datastore.delete(uuid)
+    client.get(url_for('settings.settings_browsers.delete', machine_name=machine_name), follow_redirects=True)
+
+
+def test_browser_profile_custom_headers_applied(client, live_server, measure_memory_usage, datastore_path):
+    """Custom headers set on a BrowserProfile are sent with every request using that profile;
+    per-watch headers override them when the same header name is used."""
+    from changedetectionio.tests.util import wait_for_all_checks
+
+    datastore = client.application.config.get('DATASTORE')
+    test_url = url_for('test_headers', _external=True)
+
+    machine_name = create_requests_browser_profile(
+        client,
+        name='Headers Profile Test',
+        custom_headers='X-Profile-Header: profile-value\nX-Shared-Header: from-profile',
+    )
+
+    uuid = datastore.add_watch(url=test_url, extras={'browser_profile': machine_name})
+    client.get(url_for('ui.form_watch_checknow'), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    res = client.get(url_for('ui.ui_preview.preview_page', uuid='first'), follow_redirects=True)
+    assert b'X-Profile-Header:profile-value' in res.data, \
+        "Profile custom header should appear in the echoed request"
+    assert b'X-Shared-Header:from-profile' in res.data, \
+        "Second profile custom header should appear"
+
+    # Per-watch header for the same key overrides the profile header
+    client.post(
+        url_for('ui.ui_edit.edit_page', uuid='first'),
+        data={
+            'url': test_url,
+            'tags': '',
+            'browser_profile': machine_name,
+            'headers': 'X-Shared-Header: from-watch\nX-Watch-Only: watch-value',
+            'time_between_check_use_default': 'y',
+        },
+        follow_redirects=True,
+    )
+    client.get(url_for('ui.form_watch_checknow'), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    res = client.get(url_for('ui.ui_preview.preview_page', uuid='first'), follow_redirects=True)
+    assert b'X-Profile-Header:profile-value' in res.data, \
+        "Unrelated profile header should still be present"
+    assert b'X-Shared-Header:from-watch' in res.data, \
+        "Watch-level header should override the same-named profile header"
+    assert b'X-Shared-Header:from-profile' not in res.data, \
+        "Profile value for overridden header should be gone"
+    assert b'X-Watch-Only:watch-value' in res.data, \
+        "Watch-only header should appear"
+
+    datastore.delete(uuid)
+    client.get(url_for('settings.settings_browsers.delete', machine_name=machine_name), follow_redirects=True)

changedetectionio/tests/test_commit_persistence.py

@@ -464,7 +464,7 @@ def test_settings_persist_after_update(client, live_server):
 
     # Update settings directly (bypass form validation issues)
     datastore.data['settings']['application']['empty_pages_are_a_change'] = True
-    datastore.data['settings']['application']['fetch_backend'] = 'html_requests'
+    datastore.data['settings']['application']['browser_profile'] = 'direct_http_requests'
     datastore.data['settings']['requests']['time_between_check']['minutes'] = 120
     datastore.commit()
 
@@ -478,7 +478,7 @@ def test_settings_persist_after_update(client, live_server):
 
     # Verify settings survived
     assert datastore2.data['settings']['application']['empty_pages_are_a_change'] == True, "empty_pages_are_a_change should persist"
-    assert datastore2.data['settings']['application']['fetch_backend'] == 'html_requests', "fetch_backend should persist"
+    assert datastore2.data['settings']['application']['browser_profile'] == 'direct_http_requests', "browser_profile should persist"
     assert datastore2.data['settings']['requests']['time_between_check']['minutes'] == 120, "time_between_check should persist"
 
 
@@ -634,7 +634,7 @@ def test_ui_watch_edit_persists_all_fields(client, live_server):
             'time_between_check-hours': '2',
             'time_between_check-minutes': '30',
             'include_filters': '#content',
-            'fetch_backend': 'html_requests',
+            'browser_profile': 'direct_http_requests',
             'method': 'POST',
             'ignore_text': 'Advertisement\nTracking'
         },
@@ -657,5 +657,5 @@ def test_ui_watch_edit_persists_all_fields(client, live_server):
     assert watch['title'] == 'Updated Watch Title'
     assert watch['time_between_check']['hours'] == 2
     assert watch['time_between_check']['minutes'] == 30
-    assert watch['fetch_backend'] == 'html_requests'
+    assert watch['browser_profile'] == 'direct_http_requests'
     assert watch['method'] == 'POST'

changedetectionio/tests/test_conditions.py

@@ -72,7 +72,6 @@ def test_conditions_with_text_and_number(client, live_server, measure_memory_usa
         url_for("ui.ui_edit.edit_page", uuid=uuid),
         data={
             "url": test_url,
-            "fetch_backend": "html_requests",
             "include_filters": ".number-container",
             "title": "Number AND Text Condition Test",
             "conditions_match_logic": CONDITIONS_MATCH_LOGIC_DEFAULT,  # ALL = AND logic
@@ -258,7 +257,6 @@ def test_lev_conditions_plugin(client, live_server, measure_memory_usage, datast
         url_for("ui.ui_edit.edit_page", uuid=uuid, unpause_on_save=1),
         data={
             "url": test_url,
-            "fetch_backend": "html_requests",
             "conditions_match_logic": CONDITIONS_MATCH_LOGIC_DEFAULT,  # ALL = AND logic
             "conditions-0-field": "levenshtein_ratio",
             "conditions-0-operator": "<",

changedetectionio/tests/test_css_selector.py

@@ -89,7 +89,7 @@ def test_check_markup_include_filters_restriction(client, live_server, measure_m
     # Add our URL to the import page
     res = client.post(
         url_for("ui.ui_edit.edit_page", uuid="first"),
-        data={"include_filters": include_filters, "url": test_url, "tags": "", "headers": "", 'fetch_backend': "html_requests", "time_between_check_use_default": "y"},
+        data={"include_filters": include_filters, "url": test_url, "tags": "", "headers": "", 'browser_profile': "direct_http_requests", "time_between_check_use_default": "y"},
         follow_redirects=True
     )
     assert b"Updated watch." in res.data
@@ -144,7 +144,7 @@ def test_check_multiple_filters(client, live_server, measure_memory_usage, datas
               "url": test_url,
               "tags": "",
               "headers": "",
-              'fetch_backend': "html_requests",
+              'browser_profile': "direct_http_requests",
               "time_between_check_use_default": "y"},
         follow_redirects=True
     )
@@ -195,7 +195,7 @@ def test_filter_is_empty_help_suggestion(client, live_server, measure_memory_usa
               "url": test_url,
               "tags": "",
               "headers": "",
-              'fetch_backend': "html_requests",
+              'browser_profile': "direct_http_requests",
               "time_between_check_use_default": "y"},
         follow_redirects=True
     )

changedetectionio/tests/test_element_removal.py

@@ -171,7 +171,7 @@ def test_element_removal_full(client, live_server, measure_memory_usage, datasto
             "url": test_url,
             "tags": "",
             "headers": "",
-            "fetch_backend": "html_requests",
+            "browser_profile": "direct_http_requests",
             "time_between_check_use_default": "y",
         },
         follow_redirects=True,

changedetectionio/tests/test_encoding.py

@@ -1,6 +1,7 @@
 #!/usr/bin/env python3
 # coding=utf-8
 
+import hashlib
 import time
 from flask import url_for
 from .util import live_server_setup, wait_for_all_checks, extract_UUID_from_client
@@ -11,6 +12,69 @@ import os
 
 
 
+def test_surrogate_characters_in_content_are_sanitized():
+    """Lone surrogates can appear in requests' r.text when a server returns malformed/mixed-encoding
+    content. Without sanitization, encoding to UTF-8 raises UnicodeEncodeError.
+    See: https://github.com/dgtlmoon/changedetection.io/issues/3952
+    """
+    content_with_surrogate = '<html><body>Hello \udcad World</body></html>'
+
+    # Confirm the raw problem exists
+    with pytest.raises(UnicodeEncodeError):
+        content_with_surrogate.encode('utf-8')
+
+    # Our fix: sanitize after fetcher.run() in processors/base.py call_browser()
+    sanitized = content_with_surrogate.encode('utf-8', errors='replace').decode('utf-8')
+    assert 'Hello' in sanitized
+    assert 'World' in sanitized
+    assert '\udcad' not in sanitized
+
+    # Checksum computation (processors/base.py get_raw_document_checksum) must not crash
+    hashlib.md5(sanitized.encode('utf-8')).hexdigest()
+
+
+def test_utf8_content_without_charset_header(client, live_server, datastore_path):
+    """Server returns UTF-8 content but no charset in Content-Type header.
+    chardet can misdetect such pages as UTF-7 (Python 3.14 then produces surrogates).
+    Our fix tries UTF-8 first before falling back to chardet.
+    See: https://github.com/dgtlmoon/changedetection.io/issues/3952
+    """
+    from .util import write_test_file_and_sync
+    # UTF-8 encoded content with non-ASCII chars - no charset will be in the header
+    html = '<html><body><p>Español</p><p>Français</p><p>日本語</p></body></html>'
+    write_test_file_and_sync(os.path.join(datastore_path, "endpoint-content.txt"), html.encode('utf-8'), mode='wb')
+
+    test_url = url_for('test_endpoint', content_type="text/html", _external=True)
+    client.application.config.get('DATASTORE').add_watch(url=test_url)
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    res = client.get(url_for("ui.ui_preview.preview_page", uuid="first"), follow_redirects=True)
+    # Should decode correctly as UTF-8, not produce mojibake (Español) or replacement chars
+    assert 'Español'.encode('utf-8') in res.data
+    assert 'Français'.encode('utf-8') in res.data
+    assert '日本語'.encode('utf-8') in res.data
+
+
+def test_shiftjis_with_meta_charset(client, live_server, datastore_path):
+    """Server returns Shift-JIS content with no charset in HTTP header, but the HTML
+    declares <meta charset="Shift-JIS">. We should use the meta tag, not chardet.
+    Real-world case: https://github.com/dgtlmoon/changedetection.io/issues/3952
+    """
+    from .util import write_test_file_and_sync
+    japanese_text = '日本語のページ'
+    html = f'<html><head><meta http-equiv="Content-Type" content="text/html;charset=Shift-JIS"></head><body><p>{japanese_text}</p></body></html>'
+    write_test_file_and_sync(os.path.join(datastore_path, "endpoint-content.txt"), html.encode('shift_jis'), mode='wb')
+
+    test_url = url_for('test_endpoint', content_type="text/html", _external=True)
+    client.application.config.get('DATASTORE').add_watch(url=test_url)
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    res = client.get(url_for("ui.ui_preview.preview_page", uuid="first"), follow_redirects=True)
+    assert japanese_text.encode('utf-8') in res.data
+
+
 def set_html_response(datastore_path):
     test_return_data = """
 <html><body><span class="nav_second_img_text">

changedetectionio/tests/test_errorhandling.py

@@ -10,6 +10,8 @@ from .util import live_server_setup, wait_for_all_checks, delete_all_watches
 
 
 def _runner_test_http_errors(client, live_server, http_code, expected_text, datastore_path):
+    from loguru import logger
+    logger.debug(f"_runner_test_http_errors - testing text '{expected_text}' for code {http_code}")
 
     with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:
         f.write("Now you going to get a {} error code\n".format(http_code))
@@ -20,6 +22,11 @@ def _runner_test_http_errors(client, live_server, http_code, expected_text, data
                        status_code=http_code,
                        _external=True)
 
+    if  os.getenv("PLAYWRIGHT_DRIVER_URL") or os.getenv('WEBDRIVER_URL'):
+        logger.warning("!!! Looks like we're running test with playwright or selenium, so FORCE a connection back to our container 'cdio'")
+        test_url = test_url.replace('localhost.localdomain', 'changedet')
+        test_url = test_url.replace('localhost', 'changedet')
+
     uuid = client.application.config.get('DATASTORE').add_watch(url=test_url)
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
 
@@ -76,7 +83,8 @@ def test_DNS_errors(client, live_server, measure_memory_usage, datastore_path):
         b"nodename nor servname provided" in res.data or
         b"Temporary failure in name resolution" in res.data or
         b"Failed to establish a new connection" in res.data or
-        b"Connection error occurred" in res.data
+        b"Connection error occurred" in res.data or
+        b"net::ERR_NAME_NOT_RESOLVED" in res.data
     )
     assert found_name_resolution_error
     # Should always record that we tried
@@ -108,7 +116,8 @@ def test_low_level_errors_clear_correctly(client, live_server, measure_memory_us
         b"nodename nor servname provided" in res.data or
         b"Temporary failure in name resolution" in res.data or
         b"Failed to establish a new connection" in res.data or
-        b"Connection error occurred" in res.data
+        b"Connection error occurred" in res.data or
+        b"net::ERR_NAME_NOT_RESOLVED" in res.data
     )
     assert found_name_resolution_error
 
@@ -117,7 +126,7 @@ def test_low_level_errors_clear_correctly(client, live_server, measure_memory_us
         url_for("ui.ui_edit.edit_page", uuid="first"),
         data={
             "url": test_url,
-            "fetch_backend": "html_requests",
+            "browser_profile": "direct_http_requests",
             "time_between_check_use_default": "y"},
         follow_redirects=True
     )
@@ -131,7 +140,8 @@ def test_low_level_errors_clear_correctly(client, live_server, measure_memory_us
         b"nodename nor servname provided" in res.data or
         b"Temporary failure in name resolution" in res.data or
         b"Failed to establish a new connection" in res.data or
-        b"Connection error occurred" in res.data
+        b"Connection error occurred" in res.data or
+        b"net::ERR_NAME_NOT_RESOLVED" in res.data
     )
     assert not found_name_resolution_error
 

changedetectionio/tests/test_extract_regex.py

@@ -92,7 +92,7 @@ def test_check_filter_multiline(client, live_server, measure_memory_usage, datas
               "url": test_url,
               "tags": "",
               "headers": "",
-              'fetch_backend': "html_requests",
+              'browser_profile': "direct_http_requests",
               "time_between_check_use_default": "y"
               },
         follow_redirects=True
@@ -143,7 +143,7 @@ def test_check_filter_and_regex_extract(client, live_server, measure_memory_usag
               "url": test_url,
               "tags": "",
               "headers": "",
-              'fetch_backend': "html_requests",
+              'browser_profile': "direct_http_requests",
               "time_between_check_use_default": "y"
               },
         follow_redirects=True
@@ -212,7 +212,7 @@ def test_regex_error_handling(client, live_server, measure_memory_usage, datasto
         url_for("ui.ui_edit.edit_page", uuid=uuid),
         data={"extract_text": '/something bad\d{3/XYZ',
               "url": test_url,
-              "fetch_backend": "html_requests",
+              "browser_profile": "direct_http_requests",
               "time_between_check_use_default": "y"},
         follow_redirects=True
     )

changedetectionio/tests/test_filter_exist_changes.py

@@ -96,7 +96,7 @@ def test_filter_doesnt_exist_then_exists_should_get_notification(client, live_se
         # preprended with extra filter that intentionally doesn't match any entry,
         # notification should still be sent even if first filter does not match (PR#3516)
         "include_filters": ".non-matching-selector\n.ticket-available",
-        "fetch_backend": "html_requests",
+        "browser_profile": "direct_http_requests",
         "time_between_check_use_default": "y"})
 
     res = client.post(

changedetectionio/tests/test_filter_failure_notification.py

@@ -70,7 +70,7 @@ def run_filter_test(client, live_server, content_filter, app_notification_format
                                        "Diff as Patch: {{diff_patch}}\n"
                                        ":-)",
                   "notification_format": 'text',
-                  "fetch_backend": "html_requests",
+                  "browser_profile": "direct_http_requests",
                   "filter_failure_notification_send": 'y',
                   "time_between_check_use_default": "y",
                   "headers": "",

changedetectionio/tests/test_group.py

@@ -417,7 +417,7 @@ def test_order_of_filters_tag_filter_and_watch_filter(client, live_server, measu
             "url": test_url,
             "tags": "test-tag-keep-order",
             "headers": "",
-            'fetch_backend': "html_requests",
+            'browser_profile': "direct_http_requests",
             "time_between_check_use_default": "y"},
         follow_redirects=True
     )