Realtime - Suppress socket.io errors in logs

.github/nginx-reverse-proxy-test.conf

@@ -0,0 +1,33 @@
+server {
+    listen 80;
+    server_name localhost;
+
+    # Test basic reverse proxy to changedetection.io
+    location / {
+        proxy_pass http://changedet-app:5000;
+        proxy_set_header Host $host;
+        proxy_set_header X-Real-IP $remote_addr;
+        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+        proxy_set_header X-Forwarded-Proto $scheme;
+
+        # WebSocket support
+        proxy_http_version 1.1;
+        proxy_set_header Upgrade $http_upgrade;
+        proxy_set_header Connection "upgrade";
+    }
+
+    # Test subpath deployment with X-Forwarded-Prefix
+    location /changedet-sub/ {
+        proxy_pass http://changedet-app:5000/;
+        proxy_set_header X-Forwarded-Prefix /changedet-sub;
+        proxy_set_header Host $host;
+        proxy_set_header X-Real-IP $remote_addr;
+        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+        proxy_set_header X-Forwarded-Proto $scheme;
+
+        # WebSocket support
+        proxy_http_version 1.1;
+        proxy_set_header Upgrade $http_upgrade;
+        proxy_set_header Connection "upgrade";
+    }
+}

.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/pypi-release.yml

@@ -21,7 +21,7 @@ jobs:
     - name: Build a binary wheel and a source tarball
       run: python3 -m build
     - name: Store the distribution packages
-      uses: actions/upload-artifact@v6
+      uses: actions/upload-artifact@v7
       with:
         name: python-package-distributions
         path: dist/
@@ -34,7 +34,7 @@ jobs:
     - build
     steps:
     - name: Download all the dists
-      uses: actions/download-artifact@v7
+      uses: actions/download-artifact@v8
       with:
         name: python-package-distributions
         path: dist/
@@ -93,7 +93,7 @@ jobs:
 
     steps:
     - name: Download all the dists
-      uses: actions/download-artifact@v7
+      uses: actions/download-artifact@v8
       with:
         name: python-package-distributions
         path: dist/

.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
@@ -71,7 +71,7 @@ jobs:
           docker save test-changedetectionio -o /tmp/test-changedetectionio.tar
 
       - name: Upload Docker image artifact
-        uses: actions/upload-artifact@v6
+        uses: actions/upload-artifact@v7
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp/test-changedetectionio.tar
@@ -88,7 +88,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -116,7 +116,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -165,14 +165,14 @@ jobs:
 
       - name: Store test artifacts
         if: always()
-        uses: actions/upload-artifact@v6
+        uses: actions/upload-artifact@v7
         with:
           name: test-cdio-basic-tests-output-py${{ env.PYTHON_VERSION }}
           path: output-logs
 
       - name: Store CLI test output
         if: always()
-        uses: actions/upload-artifact@v6
+        uses: actions/upload-artifact@v7
         with:
           name: test-cdio-cli-opts-output-py${{ env.PYTHON_VERSION }}
           path: cli-opts-output.txt
@@ -188,7 +188,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -230,7 +230,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -270,7 +270,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -306,7 +306,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -324,6 +324,175 @@ jobs:
         run: |
           docker run --rm --network changedet-network test-changedetectionio bash -c 'cd changedetectionio;pytest tests/smtp/test_notification_smtp.py'
 
+  nginx-reverse-proxy:
+    runs-on: ubuntu-latest
+    needs: build
+    timeout-minutes: 10
+    env:
+      PYTHON_VERSION: ${{ inputs.python-version }}
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Download Docker image artifact
+        uses: actions/download-artifact@v8
+        with:
+          name: test-changedetectionio-${{ env.PYTHON_VERSION }}
+          path: /tmp
+
+      - name: Load Docker image
+        run: |
+          docker load -i /tmp/test-changedetectionio.tar
+
+      - name: Spin up services
+        run: |
+          docker network create changedet-network
+
+          # Start changedetection.io container with X-Forwarded headers support
+          docker run --name changedet-app --hostname changedet-app --network changedet-network \
+            -e USE_X_SETTINGS=true \
+            -d test-changedetectionio
+          sleep 3
+
+      - name: Start nginx reverse proxy
+        run: |
+          # Start nginx with our test configuration
+          docker run --name nginx-proxy --network changedet-network -d -p 8080:80 --rm \
+            -v ${{ github.workspace }}/.github/nginx-reverse-proxy-test.conf:/etc/nginx/conf.d/default.conf:ro \
+            nginx:alpine
+          sleep 2
+
+      - name: Test reverse proxy - root path
+        run: |
+          echo "=== Testing nginx reverse proxy at root path ==="
+          curl --retry-connrefused --retry 6 -s http://localhost:8080/ > /tmp/nginx-test-root.html
+
+          # Check for changedetection.io UI elements
+          if grep -q "checkbox-uuid" /tmp/nginx-test-root.html; then
+            echo "✓ Found checkbox-uuid in response"
+          else
+            echo "ERROR: checkbox-uuid not found in response"
+            cat /tmp/nginx-test-root.html
+            exit 1
+          fi
+
+          # Check for watchlist content
+          if grep -q -i "watch" /tmp/nginx-test-root.html; then
+            echo "✓ Found watch/watchlist content in response"
+          else
+            echo "ERROR: watchlist content not found"
+            cat /tmp/nginx-test-root.html
+            exit 1
+          fi
+
+          echo "✓ Root path reverse proxy working correctly"
+
+      - name: Test reverse proxy - subpath with X-Forwarded-Prefix
+        run: |
+          echo "=== Testing nginx reverse proxy at subpath /changedet-sub/ ==="
+          curl --retry-connrefused --retry 6 -s http://localhost:8080/changedet-sub/ > /tmp/nginx-test-subpath.html
+
+          # Check for changedetection.io UI elements
+          if grep -q "checkbox-uuid" /tmp/nginx-test-subpath.html; then
+            echo "✓ Found checkbox-uuid in subpath response"
+          else
+            echo "ERROR: checkbox-uuid not found in subpath response"
+            cat /tmp/nginx-test-subpath.html
+            exit 1
+          fi
+
+          echo "✓ Subpath reverse proxy working correctly"
+
+      - name: Test API through reverse proxy subpath
+        run: |
+          echo "=== Testing API endpoints through nginx subpath /changedet-sub/ ==="
+
+          # Extract API key from the changedetection.io datastore
+          API_KEY=$(docker exec changedet-app cat /datastore/changedetection.json | grep -o '"api_access_token": *"[^"]*"' | cut -d'"' -f4)
+
+          if [ -z "$API_KEY" ]; then
+            echo "ERROR: Could not extract API key from datastore"
+            docker exec changedet-app cat /datastore/changedetection.json
+            exit 1
+          fi
+
+          echo "✓ Extracted API key: ${API_KEY:0:8}..."
+
+          # Create a watch via API through nginx proxy subpath
+          echo "Creating watch via POST to /changedet-sub/api/v1/watch"
+          RESPONSE=$(curl -s -w "\n%{http_code}" -X POST "http://localhost:8080/changedet-sub/api/v1/watch" \
+            -H "x-api-key: ${API_KEY}" \
+            -H "Content-Type: application/json" \
+            -d '{
+              "url": "https://example.com/test-nginx-proxy",
+              "tag": "nginx-test"
+            }')
+
+          HTTP_CODE=$(echo "$RESPONSE" | tail -n1)
+          BODY=$(echo "$RESPONSE" | head -n-1)
+
+          if [ "$HTTP_CODE" != "201" ]; then
+            echo "ERROR: Expected HTTP 201, got $HTTP_CODE"
+            echo "Response: $BODY"
+            exit 1
+          fi
+
+          echo "✓ Watch created successfully (HTTP 201)"
+
+          # Extract the watch UUID from response
+          WATCH_UUID=$(echo "$BODY" | grep -o '"uuid": *"[^"]*"' | cut -d'"' -f4)
+          echo "✓ Watch UUID: $WATCH_UUID"
+
+          # Update the watch via PUT through nginx proxy subpath
+          echo "Updating watch via PUT to /changedet-sub/api/v1/watch/${WATCH_UUID}"
+          RESPONSE=$(curl -s -w "\n%{http_code}" -X PUT "http://localhost:8080/changedet-sub/api/v1/watch/${WATCH_UUID}" \
+            -H "x-api-key: ${API_KEY}" \
+            -H "Content-Type: application/json" \
+            -d '{
+              "paused": true
+            }')
+
+          HTTP_CODE=$(echo "$RESPONSE" | tail -n1)
+          BODY=$(echo "$RESPONSE" | head -n-1)
+
+          if [ "$HTTP_CODE" != "200" ]; then
+            echo "ERROR: Expected HTTP 200, got $HTTP_CODE"
+            echo "Response: $BODY"
+            exit 1
+          fi
+
+          if echo "$BODY" | grep -q 'OK'; then
+            echo "✓ Watch updated successfully (HTTP 200, response: OK)"
+          else
+            echo "ERROR: Expected response 'OK', got: $BODY"
+            echo "Response: $BODY"
+            exit 1
+          fi
+
+          # Verify the watch is paused via GET
+          echo "Verifying watch is paused via GET"
+          RESPONSE=$(curl -s "http://localhost:8080/changedet-sub/api/v1/watch/${WATCH_UUID}" \
+            -H "x-api-key: ${API_KEY}")
+
+          if echo "$RESPONSE" | grep -q '"paused": *true'; then
+            echo "✓ Watch is paused as expected"
+          else
+            echo "ERROR: Watch paused state not confirmed"
+            echo "Response: $RESPONSE"
+            exit 1
+          fi
+
+          echo "✓ API tests through nginx subpath completed successfully"
+
+      - name: Cleanup nginx test
+        if: always()
+        run: |
+          docker logs nginx-proxy || true
+          docker logs changedet-app || true
+          docker stop nginx-proxy changedet-app || true
+          docker rm nginx-proxy changedet-app || true
+
+
+
   # Proxy tests
   proxy-tests:
     runs-on: ubuntu-latest
@@ -335,7 +504,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -375,7 +544,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -405,7 +574,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -429,7 +598,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -474,7 +643,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download Docker image artifact
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v8
         with:
           name: test-changedetectionio-${{ env.PYTHON_VERSION }}
           path: /tmp
@@ -537,7 +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
@@ -546,7 +727,7 @@ jobs:
           pip install 'pyOpenSSL>=23.2.0'
 
           echo "=== Running version 0.49.1 to create datastore ==="
-          python3 ./changedetection.py -C -d /tmp/data &
+          ALLOW_IANA_RESTRICTED_ADDRESSES=true python3 ./changedetection.py -C -d /tmp/data &
           APP_PID=$!
 
           # Wait for app to be ready
@@ -594,7 +775,7 @@ jobs:
           pip install -r requirements.txt
 
           echo "=== Running current version (commit ${{ github.sha }}) with old datastore (testing mode) ==="
-          TESTING_SHUTDOWN_AFTER_DATASTORE_LOAD=1 python3 ./changedetection.py -d /tmp/data > /tmp/upgrade-test.log 2>&1
+          ALLOW_IANA_RESTRICTED_ADDRESSES=true TESTING_SHUTDOWN_AFTER_DATASTORE_LOAD=1 python3 ./changedetection.py -d /tmp/data > /tmp/upgrade-test.log 2>&1
 
           echo "=== Upgrade test output ==="
           cat /tmp/upgrade-test.log
@@ -602,7 +783,7 @@ jobs:
 
           # Now start the current version normally to verify the tag survived
           echo "=== Starting current version to verify tag exists after upgrade ==="
-          timeout 20 python3 ./changedetection.py -d /tmp/data > /tmp/ui-test.log 2>&1 &
+          ALLOW_IANA_RESTRICTED_ADDRESSES=true timeout 20 python3 ./changedetection.py -d /tmp/data > /tmp/ui-test.log 2>&1 &
           APP_PID=$!
 
           # Wait for app to be ready and fetch UI
@@ -651,7 +832,7 @@ jobs:
 
       - name: Upload upgrade test logs
         if: always()
-        uses: actions/upload-artifact@v6
+        uses: actions/upload-artifact@v7
         with:
           name: upgrade-test-logs-py${{ env.PYTHON_VERSION }}
           path: /tmp/upgrade-test.log

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__ = '0.52.9'
+__version__ = '0.54.6'
 
 from changedetectionio.strtobool import strtobool
 from json.decoder import JSONDecodeError
@@ -61,8 +61,22 @@ 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:
@@ -610,7 +624,7 @@ def main():
 
     @app.context_processor
     def inject_template_globals():
-        return dict(right_sticky="v{}".format(datastore.data['version_tag']),
+        return dict(right_sticky="v"+__version__,
                     new_version_available=app.config['NEW_VERSION_AVAILABLE'],
                     has_password=datastore.data['settings']['application']['password'] != False,
                     socket_io_enabled=datastore.data['settings']['application'].get('ui', {}).get('socket_io_enabled', True),

changedetectionio/api/Spec.py

@@ -0,0 +1,21 @@
+import functools
+from flask import make_response
+from flask_restful import Resource
+
+
+@functools.cache
+def _get_spec_yaml():
+    """Build and cache the merged spec as a YAML string (only serialized once per process)."""
+    import yaml
+    from changedetectionio.api import build_merged_spec_dict
+    return yaml.dump(build_merged_spec_dict(), default_flow_style=False, allow_unicode=True)
+
+
+class Spec(Resource):
+    def get(self):
+        """Return the merged OpenAPI spec including all registered processor extensions."""
+        return make_response(
+            _get_spec_yaml(),
+            200,
+            {'Content-Type': 'application/yaml'}
+        )

changedetectionio/api/Tags.py

@@ -17,7 +17,7 @@ class Tag(Resource):
         self.update_q = kwargs['update_q']
 
     # Get information about a single tag
-    # curl http://localhost:5000/api/v1/tag/<string:uuid>
+    # curl http://localhost:5000/api/v1/tag/<uuid_str:uuid>
     @auth.check_token
     @validate_openapi_request('getTag')
     def get(self, uuid):
@@ -79,7 +79,7 @@ class Tag(Resource):
             'browser_steps_last_error_step', 'check_count', 'consecutive_filter_failures',
             'content-type', 'fetch_time', 'last_changed', 'last_checked', 'last_error',
             'last_notification_error', 'last_viewed', 'notification_alert_count',
-            'page_title', 'previous_md5', 'previous_md5_before_filters', 'remote_server_reply'
+            'page_title', 'previous_md5', 'remote_server_reply'
         }
 
         # Create clean tag dict without Watch-specific fields
@@ -97,17 +97,6 @@ class Tag(Resource):
         # Delete the tag, and any tag reference
         del self.datastore.data['settings']['application']['tags'][uuid]
 
-        # Delete tag.json file if it exists
-        import os
-        tag_dir = os.path.join(self.datastore.datastore_path, uuid)
-        tag_json = os.path.join(tag_dir, "tag.json")
-        if os.path.exists(tag_json):
-            try:
-                os.unlink(tag_json)
-                logger.info(f"Deleted tag.json for tag {uuid}")
-            except Exception as e:
-                logger.error(f"Failed to delete tag.json for tag {uuid}: {e}")
-
         # Remove tag from all watches
         for watch_uuid, watch in self.datastore.data['watching'].items():
             if watch.get('tags') and uuid in watch['tags']:
@@ -160,6 +149,11 @@ class Tag(Resource):
         tag.update(json_data)
         tag.commit()
 
+        # Clear checksums for all watches using this tag to force reprocessing
+        # Tag changes affect inherited configuration
+        cleared_count = self.datastore.clear_checksums_for_tag(uuid)
+        logger.info(f"Tag {uuid} updated via API, cleared {cleared_count} watch checksums")
+
         return "OK", 200
 
 
@@ -183,6 +177,13 @@ 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

@@ -57,7 +57,7 @@ class Watch(Resource):
         self.update_q = kwargs['update_q']
 
     # Get information about a single watch, excluding the history list (can be large)
-    # curl http://localhost:5000/api/v1/watch/<string:uuid>
+    # curl http://localhost:5000/api/v1/watch/<uuid_str:uuid>
     # @todo - version2 - ?muted and ?paused should be able to be called together, return the watch struct not "OK"
     # ?recheck=true
     @auth.check_token
@@ -217,7 +217,7 @@ class WatchHistory(Resource):
         self.datastore = kwargs['datastore']
 
     # Get a list of available history for a watch by UUID
-    # curl http://localhost:5000/api/v1/watch/<string:uuid>/history
+    # curl http://localhost:5000/api/v1/watch/<uuid_str:uuid>/history
     @auth.check_token
     @validate_openapi_request('getWatchHistory')
     def get(self, uuid):
@@ -338,7 +338,7 @@ 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 +349,7 @@ 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 +567,4 @@ class CreateWatch(Resource):
 
                 return {'status': f'OK, queueing {len(watches_to_queue)} watches in background'}, 202
 
-        return list, 200
+        return list, 200

changedetectionio/api/__init__.py

@@ -3,29 +3,18 @@ from flask import request, abort
 from loguru import logger
 
 @functools.cache
-def get_openapi_spec():
+def build_merged_spec_dict():
-    """Lazy load OpenAPI spec and dependencies only when validation is needed."""
-    import os
-    import yaml  # Lazy import - only loaded when API validation is actually used
-    from openapi_core import OpenAPI  # Lazy import - saves ~10.7 MB on startup
-
-    spec_path = os.path.join(os.path.dirname(__file__), '../../docs/api-spec.yaml')
-    if not os.path.exists(spec_path):
-        # Possibly for pip3 packages
-        spec_path = os.path.join(os.path.dirname(__file__), '../docs/api-spec.yaml')
-
-    with open(spec_path, 'r', encoding='utf-8') as f:
-        spec_dict = yaml.safe_load(f)
-    _openapi_spec = OpenAPI.from_dict(spec_dict)
-    return _openapi_spec
-
-@functools.cache
-def get_openapi_schema_dict():
     """
-    Get the raw OpenAPI spec dictionary for schema access.
+    Load the base OpenAPI spec and merge in any per-processor api.yaml extensions.
 
-    Used by Import endpoint to validate and convert query parameters.
+    Each processor can provide an api.yaml file alongside its __init__.py that defines
-    Returns the YAML dict directly (not the OpenAPI object).
+    additional schemas (e.g., processor_config_restock_diff). These are merged into
+    WatchBase.properties so the spec accurately reflects what the API accepts.
+
+    Plugin processors (via pluggy) are also supported - they just need an api.yaml
+    next to their processor module.
+
+    Returns the merged dict (cached - do not mutate the returned value).
     """
     import os
     import yaml
@@ -35,7 +24,59 @@ def get_openapi_schema_dict():
         spec_path = os.path.join(os.path.dirname(__file__), '../docs/api-spec.yaml')
 
     with open(spec_path, 'r', encoding='utf-8') as f:
-        return yaml.safe_load(f)
+        spec_dict = yaml.safe_load(f)
+
+    try:
+        from changedetectionio.processors import find_processors, get_parent_module
+        for module, proc_name in find_processors():
+            parent = get_parent_module(module)
+            if not parent or not hasattr(parent, '__file__'):
+                continue
+            api_yaml_path = os.path.join(os.path.dirname(parent.__file__), 'api.yaml')
+            if not os.path.exists(api_yaml_path):
+                continue
+            with open(api_yaml_path, 'r', encoding='utf-8') as f:
+                proc_spec = yaml.safe_load(f)
+            # Merge schemas
+            proc_schemas = proc_spec.get('components', {}).get('schemas', {})
+            spec_dict['components']['schemas'].update(proc_schemas)
+            # Inject processor_config_{name} into WatchBase if the schema is defined
+            schema_key = f'processor_config_{proc_name}'
+            if schema_key in proc_schemas:
+                spec_dict['components']['schemas']['WatchBase']['properties'][schema_key] = {
+                    '$ref': f'#/components/schemas/{schema_key}'
+                }
+            # Append x-code-samples from processor paths into existing path operations
+            for path, path_item in proc_spec.get('paths', {}).items():
+                if path not in spec_dict.get('paths', {}):
+                    continue
+                for method, operation in path_item.items():
+                    if method not in spec_dict['paths'][path]:
+                        continue
+                    if 'x-code-samples' in operation:
+                        existing = spec_dict['paths'][path][method].get('x-code-samples', [])
+                        spec_dict['paths'][path][method]['x-code-samples'] = existing + operation['x-code-samples']
+    except Exception as e:
+        logger.warning(f"Failed to merge processor API specs: {e}")
+
+    return spec_dict
+
+
+@functools.cache
+def get_openapi_spec():
+    """Lazy load OpenAPI spec and dependencies only when validation is needed."""
+    from openapi_core import OpenAPI  # Lazy import - saves ~10.7 MB on startup
+    return OpenAPI.from_dict(build_merged_spec_dict())
+
+@functools.cache
+def get_openapi_schema_dict():
+    """
+    Get the raw OpenAPI spec dictionary for schema access.
+
+    Used by Import endpoint to validate and convert query parameters.
+    Returns the merged YAML dict (not the OpenAPI object).
+    """
+    return build_merged_spec_dict()
 
 @functools.cache
 def _resolve_schema_properties(schema_name):
@@ -70,46 +111,6 @@ def _resolve_schema_properties(schema_name):
 
     return properties
 
-@functools.cache
-def _resolve_readonly_fields(schema_name):
-    """
-    Generic helper to resolve readOnly fields, including allOf inheritance.
-
-    Args:
-        schema_name: Name of the schema (e.g., 'Watch', 'Tag')
-
-    Returns:
-        frozenset: All readOnly field names including inherited ones
-    """
-    spec_dict = get_openapi_schema_dict()
-    schema = spec_dict['components']['schemas'].get(schema_name, {})
-
-    readonly_fields = set()
-
-    # Handle allOf (schema inheritance)
-    if 'allOf' in schema:
-        for item in schema['allOf']:
-            # Resolve $ref to parent schema
-            if '$ref' in item:
-                ref_path = item['$ref'].split('/')[-1]
-                ref_schema = spec_dict['components']['schemas'].get(ref_path, {})
-                if 'properties' in ref_schema:
-                    for field_name, field_def in ref_schema['properties'].items():
-                        if field_def.get('readOnly') is True:
-                            readonly_fields.add(field_name)
-            # Check schema-specific properties
-            if 'properties' in item:
-                for field_name, field_def in item['properties'].items():
-                    if field_def.get('readOnly') is True:
-                        readonly_fields.add(field_name)
-    else:
-        # Direct properties (no inheritance)
-        if 'properties' in schema:
-            for field_name, field_def in schema['properties'].items():
-                if field_def.get('readOnly') is True:
-                    readonly_fields.add(field_name)
-
-    return frozenset(readonly_fields)
 
 @functools.cache
 def get_watch_schema_properties():
@@ -120,14 +121,8 @@ def get_watch_schema_properties():
     """
     return _resolve_schema_properties('WatchBase')
 
-@functools.cache
+# Import readonly field utilities from shared module (avoids circular dependencies with model layer)
-def get_readonly_watch_fields():
+from changedetectionio.model.schema_utils import get_readonly_watch_fields, get_readonly_tag_fields
-    """
-    Extract readOnly field names from Watch schema in OpenAPI spec.
-
-    Returns readOnly fields from WatchBase (uuid, date_created) + Watch-specific readOnly fields.
-    """
-    return _resolve_readonly_fields('Watch')
 
 @functools.cache
 def get_tag_schema_properties():
@@ -138,15 +133,6 @@ def get_tag_schema_properties():
     """
     return _resolve_schema_properties('Tag')
 
-@functools.cache
-def get_readonly_tag_fields():
-    """
-    Extract readOnly field names from Tag schema in OpenAPI spec.
-
-    Returns readOnly fields from WatchBase (uuid, date_created) + Tag-specific readOnly fields.
-    """
-    return _resolve_readonly_fields('Tag')
-
 def validate_openapi_request(operation_id):
     """Decorator to validate incoming requests against OpenAPI spec."""
     def decorator(f):
@@ -158,6 +144,7 @@ def validate_openapi_request(operation_id):
                 if request.method.upper() != 'GET':
                     # Lazy import - only loaded when actually validating a request
                     from openapi_core.contrib.flask import FlaskOpenAPIRequest
+                    from openapi_core.templating.paths.exceptions import ServerNotFound, PathNotFound, PathError
 
                     spec = get_openapi_spec()
                     openapi_request = FlaskOpenAPIRequest(request)
@@ -165,6 +152,16 @@ def validate_openapi_request(operation_id):
                     if result.errors:
                         error_details = []
                         for error in result.errors:
+                            # Skip path/server validation errors for reverse proxy compatibility
+                            # Flask routing already validates that endpoints exist (returns 404 if not).
+                            # OpenAPI validation here is primarily for request body schema validation.
+                            # When behind nginx/reverse proxy, URLs may have path prefixes that don't
+                            # match the OpenAPI server definitions, causing false positives.
+                            if isinstance(error, PathError):
+                                logger.debug(f"API Call - Skipping path/server validation (delegated to Flask): {error}")
+                                continue
+
+                            error_str = str(error)
                             # Extract detailed schema errors from __cause__
                             if hasattr(error, '__cause__') and hasattr(error.__cause__, 'schema_errors'):
                                 for schema_error in error.__cause__.schema_errors:
@@ -172,9 +169,12 @@ def validate_openapi_request(operation_id):
                                     msg = schema_error.message if hasattr(schema_error, 'message') else str(schema_error)
                                     error_details.append(f"{field}: {msg}")
                             else:
-                                error_details.append(str(error))
+                                error_details.append(error_str)
+
+                        # Only raise if we have actual validation errors (not path/server issues)
+                        if error_details:
                             logger.error(f"API Call - Validation failed: {'; '.join(error_details)}")
-                        raise BadRequest(f"Validation failed: {'; '.join(error_details)}")
+                            raise BadRequest(f"Validation failed: {'; '.join(error_details)}")
             except BadRequest:
                 # Re-raise BadRequest exceptions (validation failures)
                 raise
@@ -191,5 +191,6 @@ from .Watch import Watch, WatchHistory, WatchSingleHistory, WatchHistoryDiff, Cr
 from .Tags import Tags, Tag
 from .Import import Import
 from .SystemInfo import SystemInfo
+from .Spec import Spec
 from .Notifications import Notifications
 

changedetectionio/blueprint/backups/__init__.py

@@ -13,7 +13,7 @@ from loguru import logger
 BACKUP_FILENAME_FORMAT = "changedetection-backup-{}.zip"
 
 
-def create_backup(datastore_path, watches: dict):
+def create_backup(datastore_path, watches: dict, tags: dict = None):
     logger.debug("Creating backup...")
     import zipfile
     from pathlib import Path
@@ -40,10 +40,14 @@ def create_backup(datastore_path, watches: dict):
             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)
+        # Add tag data directories (each tag has its own {uuid}/tag.json)
-        secret_file = os.path.join(datastore_path, "secret.txt")
+        for uuid, tag in (tags or {}).items():
-        if os.path.isfile(secret_file):
+            for f in Path(tag.data_dir).glob('*'):
-            zipObj.write(secret_file, arcname="secret.txt")
+                zipObj.write(f,
+                             arcname=os.path.join(f.parts[-2], f.parts[-1]),
+                             compress_type=zipfile.ZIP_DEFLATED,
+                             compresslevel=8)
+            logger.debug(f"Added tag '{tag.get('title')}' ({uuid}) to backup")
 
         # Add any data in the watch data directory.
         for uuid, w in watches.items():
@@ -88,7 +92,10 @@ def create_backup(datastore_path, watches: dict):
 
 
 def construct_blueprint(datastore: ChangeDetectionStore):
+    from .restore import construct_restore_blueprint
+
     backups_blueprint = Blueprint('backups', __name__, template_folder="templates")
+    backups_blueprint.register_blueprint(construct_restore_blueprint(datastore))
     backup_threads = []
 
     @login_optionally_required
@@ -96,16 +103,17 @@ def construct_blueprint(datastore: ChangeDetectionStore):
     def request_backup():
         if any(thread.is_alive() for thread in backup_threads):
             flash(gettext("A backup is already running, check back in a few minutes"), "error")
-            return redirect(url_for('backups.index'))
+            return redirect(url_for('backups.create'))
 
         if len(find_backups()) > int(os.getenv("MAX_NUMBER_BACKUPS", 100)):
             flash(gettext("Maximum number of backups reached, please remove some"), "error")
-            return redirect(url_for('backups.index'))
+            return redirect(url_for('backups.create'))
 
         # With immediate persistence, all data is already saved
         zip_thread = threading.Thread(
             target=create_backup,
             args=(datastore.datastore_path, datastore.data.get("watching")),
+            kwargs={'tags': datastore.data['settings']['application'].get('tags', {})},
             daemon=True,
             name="BackupCreator"
         )
@@ -113,7 +121,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         backup_threads.append(zip_thread)
         flash(gettext("Backup building in background, check back in a few minutes."))
 
-        return redirect(url_for('backups.index'))
+        return redirect(url_for('backups.create'))
 
     def find_backups():
         backup_filepath = os.path.join(datastore.datastore_path, BACKUP_FILENAME_FORMAT.format("*"))
@@ -138,31 +146,34 @@ def construct_blueprint(datastore: ChangeDetectionStore):
     def download_backup(filename):
         import re
         filename = filename.strip()
-        backup_filename_regex = BACKUP_FILENAME_FORMAT.format("\d+")
+        backup_filename_regex = BACKUP_FILENAME_FORMAT.format(r"\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)
 
+        # 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)
 
     @login_optionally_required
-    @backups_blueprint.route("", methods=['GET'])
+    @backups_blueprint.route("/", methods=['GET'])
-    def index():
+    @backups_blueprint.route("/create", methods=['GET'])
+    def create():
         backups = find_backups()
-        output = render_template("overview.html",
+        output = render_template("backup_create.html",
                                  available_backups=backups,
                                  backup_running=any(thread.is_alive() for thread in backup_threads)
                                  )
-
         return output
 
     @login_optionally_required
@@ -176,6 +187,6 @@ def construct_blueprint(datastore: ChangeDetectionStore):
 
         flash(gettext("Backups were deleted."))
 
-        return redirect(url_for('backups.index'))
+        return redirect(url_for('backups.create'))
 
     return backups_blueprint

changedetectionio/blueprint/backups/restore.py

@@ -0,0 +1,248 @@
+import io
+import json
+import os
+import re
+import shutil
+import tempfile
+import threading
+import zipfile
+
+from flask import Blueprint, render_template, flash, url_for, redirect, request
+from flask_babel import gettext, lazy_gettext as _l
+from wtforms import Form, BooleanField, SubmitField
+from flask_wtf.file import FileField, FileAllowed
+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=[
+        FileAllowed(['zip'], _l('Must be a .zip backup file!'))
+    ])
+    include_groups = BooleanField(_l('Include groups'), default=True)
+    include_groups_replace_existing = BooleanField(_l('Replace existing groups of the same UUID'), default=True)
+    include_watches = BooleanField(_l('Include watches'), default=True)
+    include_watches_replace_existing = BooleanField(_l('Replace existing watches of the same UUID'), default=True)
+    submit = SubmitField(_l('Restore backup'))
+
+
+def import_from_zip(zip_stream, datastore, include_groups, include_groups_replace, include_watches, include_watches_replace):
+    """
+    Extract and import watches and groups from a backup zip stream.
+
+    Mirrors the store's _load_watches / _load_tags loading pattern:
+      - UUID dirs with tag.json  → Tag.model + tag_obj.commit()
+      - UUID dirs with watch.json → rehydrate_entity + watch_obj.commit()
+
+    Returns a dict with counts: restored_groups, skipped_groups, restored_watches, skipped_watches.
+    Raises zipfile.BadZipFile if the stream is not a valid zip.
+    """
+    from changedetectionio.model import Tag
+
+    restored_groups = 0
+    skipped_groups = 0
+    restored_watches = 0
+    skipped_watches = 0
+
+    current_tags = datastore.data['settings']['application'].get('tags', {})
+    current_watches = datastore.data['watching']
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        logger.debug(f"Restore: extracting zip to {tmpdir}")
+        with zipfile.ZipFile(zip_stream, 'r') as zf:
+            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):
+            if not entry.is_dir():
+                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')
+
+            # --- Tags (groups) ---
+            if include_groups and os.path.exists(tag_json_path):
+                if uuid in current_tags and not include_groups_replace:
+                    logger.debug(f"Restore: skipping existing group {uuid} (replace not requested)")
+                    skipped_groups += 1
+                    continue
+
+                try:
+                    with open(tag_json_path, 'r', encoding='utf-8') as f:
+                        tag_data = json.load(f)
+                except (json.JSONDecodeError, IOError) as e:
+                    logger.error(f"Restore: failed to read tag.json for {uuid}: {e}")
+                    continue
+
+                title = tag_data.get('title', uuid)
+                logger.debug(f"Restore: importing group '{title}' ({uuid})")
+
+                # Mirror _load_tags: set uuid and force processor
+                tag_data['uuid'] = uuid
+                tag_data['processor'] = 'restock_diff'
+
+                # Copy the UUID directory so data_dir exists for commit()
+                dst_dir = os.path.join(datastore.datastore_path, uuid)
+                if os.path.exists(dst_dir):
+                    shutil.rmtree(dst_dir)
+                shutil.copytree(entry.path, dst_dir)
+
+                tag_obj = Tag.model(
+                    datastore_path=datastore.datastore_path,
+                    __datastore=datastore.data,
+                    default=tag_data
+                )
+                current_tags[uuid] = tag_obj
+                tag_obj.commit()
+                restored_groups += 1
+                logger.success(f"Restore: group '{title}' ({uuid}) restored")
+
+            # --- Watches ---
+            elif include_watches and os.path.exists(watch_json_path):
+                if uuid in current_watches and not include_watches_replace:
+                    logger.debug(f"Restore: skipping existing watch {uuid} (replace not requested)")
+                    skipped_watches += 1
+                    continue
+
+                try:
+                    with open(watch_json_path, 'r', encoding='utf-8') as f:
+                        watch_data = json.load(f)
+                except (json.JSONDecodeError, IOError) as e:
+                    logger.error(f"Restore: failed to read watch.json for {uuid}: {e}")
+                    continue
+
+                url = watch_data.get('url', uuid)
+                logger.debug(f"Restore: importing watch '{url}' ({uuid})")
+
+                # Copy UUID directory first so data_dir and history files exist
+                dst_dir = os.path.join(datastore.datastore_path, uuid)
+                if os.path.exists(dst_dir):
+                    shutil.rmtree(dst_dir)
+                shutil.copytree(entry.path, dst_dir)
+
+                # Mirror _load_watches / rehydrate_entity
+                watch_data['uuid'] = uuid
+                watch_obj = datastore.rehydrate_entity(uuid, watch_data)
+                current_watches[uuid] = watch_obj
+                watch_obj.commit()
+                restored_watches += 1
+                logger.success(f"Restore: watch '{url}' ({uuid}) restored")
+
+        logger.debug(f"Restore: scan complete - groups {restored_groups} restored / {skipped_groups} skipped, "
+                     f"watches {restored_watches} restored / {skipped_watches} skipped")
+
+    # Persist changedetection.json (includes the updated tags dict)
+    logger.debug("Restore: committing datastore settings")
+    datastore.commit()
+
+    return {
+        'restored_groups': restored_groups,
+        'skipped_groups': skipped_groups,
+        'restored_watches': restored_watches,
+        'skipped_watches': skipped_watches,
+    }
+
+
+
+def construct_restore_blueprint(datastore):
+    restore_blueprint = Blueprint('restore', __name__, template_folder="templates")
+    restore_threads = []
+
+    @login_optionally_required
+    @restore_blueprint.route("/restore", methods=['GET'])
+    def restore():
+        form = RestoreForm()
+        return render_template("backup_restore.html",
+                               form=form,
+                               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'])
+    def backups_restore_start():
+        if any(t.is_alive() for t in restore_threads):
+            flash(gettext("A restore is already running, check back in a few minutes"), "error")
+            return redirect(url_for('backups.restore.restore'))
+
+        zip_file = request.files.get('zip_file')
+        if not zip_file or not zip_file.filename:
+            flash(gettext("No file uploaded"), "error")
+            return redirect(url_for('backups.restore.restore'))
+
+        if not zip_file.filename.lower().endswith('.zip'):
+            flash(gettext("File must be a .zip backup file"), "error")
+            return redirect(url_for('backups.restore.restore'))
+
+        # 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:
+            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")
+            return redirect(url_for('backups.restore.restore'))
+
+        include_groups = request.form.get('include_groups') == 'y'
+        include_groups_replace = request.form.get('include_groups_replace_existing') == 'y'
+        include_watches = request.form.get('include_watches') == 'y'
+        include_watches_replace = request.form.get('include_watches_replace_existing') == 'y'
+
+        restore_thread = threading.Thread(
+            target=import_from_zip,
+            kwargs={
+                'zip_stream': zip_bytes,
+                'datastore': datastore,
+                'include_groups': include_groups,
+                'include_groups_replace': include_groups_replace,
+                'include_watches': include_watches,
+                'include_watches_replace': include_watches_replace,
+            },
+            daemon=True,
+            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'))
+
+    return restore_blueprint

changedetectionio/blueprint/backups/templates/backup_create.html

@@ -0,0 +1,49 @@
+{% extends 'base.html' %}
+{% block content %}
+    {% from '_helpers.html' import render_simple_field, render_field %}
+
+    <div class="edit-form">
+        <div class="tabs collapsable">
+            <ul>
+                <li class="tab active" id=""><a href="{{ url_for('backups.create') }}">{{ _('Create') }}</a></li>
+                <li class="tab"><a href="{{ url_for('backups.restore.restore') }}">{{ _('Restore') }}</a></li>
+            </ul>
+        </div>
+        <div class="box-wrap inner">
+            <div id="general">
+                {% if backup_running %}
+                    <p>
+                        <span class="spinner"></span>&nbsp;<strong>{{ _('A backup is running!') }}</strong>
+                    </p>
+                {% endif %}
+
+                <p>
+                    {{ _('Here you can download and request a new backup, when a backup is completed you will see it listed below.') }}
+                </p>
+                <br>
+                {% if available_backups %}
+                    <ul>
+                        {% for backup in available_backups %}
+                            <li>
+                                <a href="{{ url_for('backups.download_backup', filename=backup["filename"]) }}">{{ backup["filename"] }}</a> {{ backup["filesize"] }} {{ _('Mb') }}
+                            </li>
+                        {% endfor %}
+                    </ul>
+                {% else %}
+                    <p>
+                        <strong>{{ _('No backups found.') }}</strong>
+                    </p>
+                {% endif %}
+
+                <a class="pure-button pure-button-primary"
+                   href="{{ url_for('backups.request_backup') }}">{{ _('Create backup') }}</a>
+                {% if available_backups %}
+                    <a class="pure-button button-small button-error "
+                       href="{{ url_for('backups.remove_backups') }}">{{ _('Remove backups') }}</a>
+                {% endif %}
+
+            </div>
+
+        </div>
+    </div>
+{% endblock %}

changedetectionio/blueprint/backups/templates/backup_restore.html

@@ -0,0 +1,62 @@
+{% extends 'base.html' %}
+{% block content %}
+    {% from '_helpers.html' import render_field, render_checkbox_field %}
+
+    <div class="edit-form">
+        <div class="tabs collapsable">
+            <ul>
+                <li class="tab"><a href="{{ url_for('backups.create') }}">{{ _('Create') }}</a></li>
+                <li class="tab active"><a href="{{ url_for('backups.restore.restore') }}">{{ _('Restore') }}</a></li>
+            </ul>
+        </div>
+        <div class="box-wrap inner">
+            <div id="general">
+                {% if restore_running %}
+                    <p>
+                        <span class="spinner"></span>&nbsp;<strong>{{ _('A restore is running!') }}</strong>
+                    </p>
+                {% endif %}
+
+                <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') }}"
+                      method="POST"
+                      enctype="multipart/form-data">
+                    <input type="hidden" name="csrf_token" value="{{ csrf_token() }}">
+
+                    <div class="pure-control-group">
+                        {{ render_checkbox_field(form.include_groups) }}
+                        <span class="pure-form-message-inline">{{ _('Include all groups found in backup?') }}</span>
+                    </div>
+                    <div class="pure-control-group">
+                        {{ render_checkbox_field(form.include_groups_replace_existing) }}
+                        <span class="pure-form-message-inline">{{ _('Replace any existing groups of the same UUID?') }}</span>
+                    </div>
+
+                    <div class="pure-control-group">
+                        {{ render_checkbox_field(form.include_watches) }}
+                        <span class="pure-form-message-inline">{{ _('Include all watches found in backup?') }}</span>
+                    </div>
+                    <div class="pure-control-group">
+                        {{ render_checkbox_field(form.include_watches_replace_existing) }}
+                        <span class="pure-form-message-inline">{{ _('Replace any existing watches of the same UUID?') }}</span>
+                    </div>
+
+                    <div class="pure-control-group">
+                        {{ render_field(form.zip_file) }}
+                    </div>
+
+                    <div class="pure-controls">
+                        <button type="submit" class="pure-button pure-button-primary">{{ _('Restore backup') }}</button>
+                    </div>
+                </form>
+            </div>
+        </div>
+    </div>
+{% endblock %}

changedetectionio/blueprint/backups/templates/overview.html

@@ -1,36 +0,0 @@
-{% extends 'base.html' %}
-{% block content %}
-    {% from '_helpers.html' import render_simple_field, render_field %}
-    <div class="edit-form">
-        <div class="box-wrap inner">
-            <h2>{{ _('Backups') }}</h2>
-            {% if backup_running %}
-                <p>
-                    <span class="spinner"></span>&nbsp;<strong>{{ _('A backup is running!') }}</strong>
-                </p>
-            {% endif %}
-            <p>
-                {{ _('Here you can download and request a new backup, when a backup is completed you will see it listed below.') }}
-            </p>
-            <br>
-                {% if available_backups %}
-                    <ul>
-                    {% for backup in available_backups %}
-                        <li><a href="{{ url_for('backups.download_backup', filename=backup["filename"]) }}">{{ backup["filename"] }}</a> {{  backup["filesize"] }} {{ _('Mb') }}</li>
-                    {% endfor %}
-                    </ul>
-                {% else %}
-                    <p>
-                    <strong>{{ _('No backups found.') }}</strong>
-                    </p>
-                {% endif %}
-
-            <a class="pure-button pure-button-primary" href="{{ url_for('backups.request_backup') }}">{{ _('Create backup') }}</a>
-            {% if available_backups %}
-                <a class="pure-button button-small button-error " href="{{ url_for('backups.remove_backups') }}">{{ _('Remove backups') }}</a>
-            {% endif %}
-        </div>
-    </div>
-
-
-{% endblock %}

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
+        try:
-        browserstepper = session_data.get('browserstepper')
+            run_async_in_browser_loop(_close_session_resources(session_data, label=f" for session {session_id}"))
-        if browserstepper:
+        except Exception as e:
-            try:
+            logger.error(f"Error cleaning up session {session_id}: {e}")
-                run_async_in_browser_loop(browserstepper.cleanup())
-            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')
+        try:
-        if browserstepper:
+            run_async_in_browser_loop(_close_session_resources(session_data, label=f" for watch {watch_uuid}"))
-            try:
+        except Exception as e:
-                run_async_in_browser_loop(browserstepper.cleanup())
+            logger.error(f"Error cleaning up session {session_id} for watch {watch_uuid}: {e}")
-            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]
@@ -174,63 +198,73 @@ def construct_blueprint(datastore: ChangeDetectionStore):
     browser_steps_blueprint = Blueprint('browser_steps', __name__, template_folder="templates")
 
     async def start_browsersteps_session(watch_uuid):
-        from . import browser_steps
+        from changedetectionio.browser_steps import browser_steps
         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
+        # Build proxy dict first — needed by both the CDP path and fetcher-specific launchers
-        playwright_instance = async_playwright()
-        playwright_context = await playwright_instance.start()
-
-        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}"
-
-        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')
+            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
+        # 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_name = watch.get_fetch_backend or 'system'
+        if fetcher_name == 'system':
+            fetcher_name = datastore.data['settings']['application'].get('fetch_backend', 'html_requests')
+        fetcher_class = getattr(content_fetchers, fetcher_name, None)
+
+        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_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}")
+
+        browsersteps_start_session['browser'] = browser
+        browsersteps_start_session['playwright_context'] = playwright_context
+
         browserstepper = browser_steps.browsersteps_live_ui(
             playwright_browser=browser,
             proxy=proxy,
-            start_url=datastore.data['watching'][watch_uuid].link,
+            start_url=watch.link,
-            headers=datastore.data['watching'][watch_uuid].get('headers')
+            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
 
 
@@ -238,7 +272,6 @@ def construct_blueprint(datastore: ChangeDetectionStore):
     @browser_steps_blueprint.route("/browsersteps_start_session", methods=['GET'])
     def browsersteps_start_session():
         # A new session was requested, return sessionID
-        import asyncio
         import uuid
         browsersteps_session_id = str(uuid.uuid4())
         watch_uuid = request.args.get('uuid')
@@ -301,11 +334,10 @@ def construct_blueprint(datastore: ChangeDetectionStore):
     @browser_steps_blueprint.route("/browsersteps_update", methods=['POST'])
     def browsersteps_ui_update():
         import base64
-        import playwright._impl._errors
-        from changedetectionio.blueprint.browser_steps import browser_steps
 
-        remaining =0
+        remaining = 0
         uuid = request.args.get('uuid')
+        goto_website_url_first_step = request.args.get('goto_website_url_first_step')
 
         browsersteps_session_id = request.args.get('browsersteps_session_id')
 
@@ -316,33 +348,33 @@ def construct_blueprint(datastore: ChangeDetectionStore):
             return make_response('No session exists under that ID', 500)
 
         is_last_step = False
-        # Actions - step/apply/etc, do the thing and return state
+
-        if request.method == 'POST':
+        # @todo - should always be an existing session
-            # @todo - should always be an existing session
+        if goto_website_url_first_step:
+            logger.debug("Going to site (requested automatically before stepping)..")
+            step_operation = "Goto site"
+            step_selector = None
+            step_optional_value = None
+        else:
             step_operation = request.form.get('operation')
             step_selector = request.form.get('selector')
             step_optional_value = request.form.get('optional_value')
             is_last_step = strtobool(request.form.get('is_last_step'))
 
-            try:
+        try:
-                # Run the async call_action method in the dedicated browser steps event loop
+            # Run the async call_action method in the dedicated browser steps event loop
-                run_async_in_browser_loop(
+            run_async_in_browser_loop(
-                    browsersteps_sessions[browsersteps_session_id]['browserstepper'].call_action(
+                browsersteps_sessions[browsersteps_session_id]['browserstepper'].call_action(
-                        action_name=step_operation,
+                    action_name=step_operation,
-                        selector=step_selector,
+                    selector=step_selector,
-                        optional_value=step_optional_value
+                    optional_value=step_optional_value
-                    )
                 )
+            )
 
-            except Exception as e:
+        except Exception as e:
-                logger.error(f"Exception when calling step operation {step_operation} {str(e)}")
+            logger.error(f"Exception when calling step operation {step_operation} {str(e)}")
-                # Try to find something of value to give back to the user
+            # Try to find something of value to give back to the user
-                return make_response(str(e).splitlines()[0], 401)
+            return make_response(str(e).splitlines()[0], 401)
-
-
-#        if not this_session.page:
-#            cleanup_playwright_session()
-#            return make_response('Browser session ran out of time :( Please reload this page.', 401)
 
         # Screenshots and other info only needed on requesting a step (POST)
         try:
@@ -350,7 +382,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
             (screenshot, xpath_data) = run_async_in_browser_loop(
                 browsersteps_sessions[browsersteps_session_id]['browserstepper'].get_current_state()
             )
-                
+
             if is_last_step:
                 watch = datastore.data['watching'].get(uuid)
                 u = browsersteps_sessions[browsersteps_session_id]['browserstepper'].page.url

changedetectionio/blueprint/check_proxies/__init__.py

@@ -94,13 +94,13 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         return results
 
     @login_required
-    @check_proxies_blueprint.route("/<string:uuid>/status", methods=['GET'])
+    @check_proxies_blueprint.route("/<uuid_str:uuid>/status", methods=['GET'])
     def get_recheck_status(uuid):
         results = _recalc_check_status(uuid=uuid)
         return results
 
     @login_required
-    @check_proxies_blueprint.route("/<string:uuid>/start", methods=['GET'])
+    @check_proxies_blueprint.route("/<uuid_str:uuid>/start", methods=['GET'])
     def start_check(uuid):
 
         if not datastore.proxy_list:

changedetectionio/blueprint/imports/templates/import.html

@@ -16,6 +16,11 @@
         <form class="pure-form" action="{{url_for('imports.import_page')}}" method="POST" enctype="multipart/form-data">
             <input type="hidden" name="csrf_token" value="{{ csrf_token() }}">
             <div class="tab-pane-inner" id="url-list">
+
+                <p>
+                {{ _('Restoring changedetection.io backups is in the') }}<a href="{{ url_for('backups.restore.restore') }}"> {{ _('backups section') }}</a>.
+                <br>
+                </p>
                 <div class="pure-control-group">
                         {{ _('Enter one URL per line, and optionally add tags for each URL after a space, delineated by comma (,):') }}
                         <br>
@@ -37,9 +42,6 @@
             </div>
 
             <div class="tab-pane-inner" id="distill-io">
-
-
-
                     <div class="pure-control-group">
                         {{ _('Copy and Paste your Distill.io watch \'export\' file, this should be a JSON file.') }}<br>
                         {{ _('This is') }} <i>{{ _('experimental') }}</i>, {{ _('supported fields are') }} <code>name</code>, <code>uri</code>, <code>tags</code>, <code>config:selections</code>, {{ _('the rest (including') }} <code>schedule</code>) {{ _('are ignored.') }}
@@ -49,8 +51,6 @@
                         {{ _('Be sure to set your default fetcher to Chrome if required.') }}<br>
                         </p>
                     </div>
-
-
                     <textarea name="distill-io" class="pure-input-1-2" style="width: 100%;
                                 font-family:monospace;
                                 white-space: pre;
@@ -114,6 +114,7 @@
                 </div>
             </div>
             <button type="submit" class="pure-button pure-input-1-2 pure-button-primary">{{ _('Import') }}</button>
+
         </form>
 
     </div>

changedetectionio/blueprint/price_data_follower/__init__.py

@@ -15,7 +15,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q: PriorityQueue
     price_data_follower_blueprint = Blueprint('price_data_follower', __name__)
 
     @login_required
-    @price_data_follower_blueprint.route("/<string:uuid>/accept", methods=['GET'])
+    @price_data_follower_blueprint.route("/<uuid_str:uuid>/accept", methods=['GET'])
     def accept(uuid):
         datastore.data['watching'][uuid]['track_ldjson_price_data'] = PRICE_DATA_TRACK_ACCEPT
         datastore.data['watching'][uuid]['processor'] = 'restock_diff'
@@ -25,7 +25,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q: PriorityQueue
         return redirect(url_for("watchlist.index"))
 
     @login_required
-    @price_data_follower_blueprint.route("/<string:uuid>/reject", methods=['GET'])
+    @price_data_follower_blueprint.route("/<uuid_str:uuid>/reject", methods=['GET'])
     def reject(uuid):
         datastore.data['watching'][uuid]['track_ldjson_price_data'] = PRICE_DATA_TRACK_REJECT
         datastore.data['watching'][uuid].commit()

changedetectionio/blueprint/rss/single_watch.py

@@ -9,11 +9,12 @@ def construct_single_watch_routes(rss_blueprint, datastore):
         datastore: The ChangeDetectionStore instance
     """
 
-    @rss_blueprint.route("/watch/<string:uuid>", methods=['GET'])
+    @rss_blueprint.route("/watch/<uuid_str:uuid>", methods=['GET'])
     def rss_single_watch(uuid):
         import time
 
-        from flask import make_response, request
+        from flask import make_response, request, Response
+        from flask_babel import lazy_gettext as _l
         from feedgen.feed import FeedGenerator
         from loguru import logger
 
@@ -42,12 +43,12 @@ def construct_single_watch_routes(rss_blueprint, datastore):
         # Get the watch by UUID
         watch = datastore.data['watching'].get(uuid)
         if not watch:
-            return f"Watch with UUID {uuid} not found", 404
+            return Response(_l("Watch with UUID %(uuid)s not found", uuid=uuid), status=404, mimetype='text/plain')
 
         # Check if watch has at least 2 history snapshots
         dates = list(watch.history.keys())
         if len(dates) < 2:
-            return f"Watch {uuid} does not have enough history snapshots to show changes (need at least 2)", 400
+            return Response(_l("Watch %(uuid)s does not have enough history snapshots to show changes (need at least 2)", uuid=uuid), status=400, mimetype='text/plain')
 
         # Get the number of diffs to include (default: 5)
         rss_diff_length = datastore.data['settings']['application'].get('rss_diff_length', 5)

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

@@ -83,6 +83,10 @@ def construct_blueprint(datastore: ChangeDetectionStore):
                 datastore.data['settings']['requests'].update(form.data['requests'])
                 datastore.commit()
 
+                # Clear all checksums to force reprocessing with new settings
+                # Global settings can affect watch behavior (filters, rendering, etc.)
+                datastore.clear_all_last_checksums()
+
                 # Adjust worker count if it changed
                 if new_worker_count != old_worker_count:
                     from changedetectionio import worker_pool

changedetectionio/blueprint/settings/templates/settings.html

@@ -25,7 +25,7 @@
             <li class="tab"><a href="#ui-options">{{ _('UI Options') }}</a></li>
             <li class="tab"><a href="#api">{{ _('API') }}</a></li>
             <li class="tab"><a href="#rss">{{ _('RSS') }}</a></li>
-            <li class="tab"><a href="{{ url_for('backups.index') }}">{{ _('Backups') }}</a></li>
+            <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>
             {% if plugin_tabs %}
@@ -154,9 +154,8 @@
                     </span>
                 </div>
                 <div class="pure-control-group">
-                                        <br>
+                <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>
+                    {{ _('Tip:') }} <a href="{{ url_for('settings.settings_page')}}#proxies">{{ _('Connect using Bright Data proxies, find out more here.') }}</a>
-
                 </div>
             </div>
 
@@ -352,7 +351,7 @@ 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/tags/__init__.py

@@ -54,7 +54,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
 
         return redirect(url_for('tags.tags_overview_page'))
 
-    @tags_blueprint.route("/mute/<string:uuid>", methods=['GET'])
+    @tags_blueprint.route("/mute/<uuid_str:uuid>", methods=['GET'])
     @login_optionally_required
     def mute(uuid):
         tag = datastore.data['settings']['application']['tags'].get(uuid)
@@ -63,24 +63,13 @@ def construct_blueprint(datastore: ChangeDetectionStore):
             tag.commit()
         return redirect(url_for('tags.tags_overview_page'))
 
-    @tags_blueprint.route("/delete/<string:uuid>", methods=['GET'])
+    @tags_blueprint.route("/delete/<uuid_str:uuid>", methods=['GET'])
     @login_optionally_required
     def delete(uuid):
         # Delete the tag from settings immediately
         if datastore.data['settings']['application']['tags'].get(uuid):
             del datastore.data['settings']['application']['tags'][uuid]
 
-        # Delete tag.json file if it exists
-        import os
-        tag_dir = os.path.join(datastore.datastore_path, uuid)
-        tag_json = os.path.join(tag_dir, "tag.json")
-        if os.path.exists(tag_json):
-            try:
-                os.unlink(tag_json)
-                logger.info(f"Deleted tag.json for tag {uuid}")
-            except Exception as e:
-                logger.error(f"Failed to delete tag.json for tag {uuid}: {e}")
-
         # Remove tag from all watches in background thread to avoid blocking
         def remove_tag_background(tag_uuid):
             """Background thread to remove tag from watches - discarded after completion."""
@@ -101,7 +90,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         flash(gettext("Tag deleted, removing from watches in background"))
         return redirect(url_for('tags.tags_overview_page'))
 
-    @tags_blueprint.route("/unlink/<string:uuid>", methods=['GET'])
+    @tags_blueprint.route("/unlink/<uuid_str:uuid>", methods=['GET'])
     @login_optionally_required
     def unlink(uuid):
         # Unlink tag from all watches in background thread to avoid blocking
@@ -127,19 +116,11 @@ def construct_blueprint(datastore: ChangeDetectionStore):
     @tags_blueprint.route("/delete_all", methods=['GET'])
     @login_optionally_required
     def delete_all():
-        # Delete all tag.json files
-        import os
-        for tag_uuid in list(datastore.data['settings']['application']['tags'].keys()):
-            tag_dir = os.path.join(datastore.datastore_path, tag_uuid)
-            tag_json = os.path.join(tag_dir, "tag.json")
-            if os.path.exists(tag_json):
-                try:
-                    os.unlink(tag_json)
-                except Exception as e:
-                    logger.error(f"Failed to delete tag.json for tag {tag_uuid}: {e}")
 
-        # Clear all tags from settings immediately
+        for tag_uuid in list(datastore.data['settings']['application']['tags'].keys()):
-        datastore.data['settings']['application']['tags'] = {}
+# TagsDict 'del' handler will remove the dir
+            del datastore.data['settings']['application']['tags'][tag_uuid]
+
 
         # Clear tags from all watches in background thread to avoid blocking
         def clear_all_tags_background():
@@ -160,7 +141,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         flash(gettext("All tags deleted, clearing from watches in background"))
         return redirect(url_for('tags.tags_overview_page'))
 
-    @tags_blueprint.route("/edit/<string:uuid>", methods=['GET'])
+    @tags_blueprint.route("/edit/<uuid_str:uuid>", methods=['GET'])
     @login_optionally_required
     def form_tag_edit(uuid):
         from changedetectionio.blueprint.tags.form import group_restock_settings_form
@@ -179,6 +160,21 @@ def construct_blueprint(datastore: ChangeDetectionStore):
                                        default_system_settings = datastore.data['settings'],
                                        )
 
+        # Bridge API-stored processor_config_* values into the form's FormField sub-forms.
+        # The API stores processor_config_restock_diff in the tag dict; find the matching
+        # FormField by checking which one's sub-fields cover the config keys.
+        from wtforms.fields.form import FormField as WTFormField
+        for key, value in default.items():
+            if not key.startswith('processor_config_') or not isinstance(value, dict):
+                continue
+            for form_field in form:
+                if isinstance(form_field, WTFormField) and all(k in form_field.form._fields for k in value):
+                    for sub_key, sub_value in value.items():
+                        sub_field = form_field.form._fields.get(sub_key)
+                        if sub_field is not None:
+                            sub_field.data = sub_value
+                    break
+
         template_args = {
             'data': default,
             'form': form,
@@ -222,7 +218,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         return output
 
 
-    @tags_blueprint.route("/edit/<string:uuid>", methods=['POST'])
+    @tags_blueprint.route("/edit/<uuid_str:uuid>", methods=['POST'])
     @login_optionally_required
     def form_tag_edit_submit(uuid):
         from changedetectionio.blueprint.tags.form import group_restock_settings_form
@@ -244,12 +240,15 @@ def construct_blueprint(datastore: ChangeDetectionStore):
         tag.update(form.data)
         tag['processor'] = 'restock_diff'
         tag.commit()
+
+        # Clear checksums for all watches using this tag to force reprocessing
+        # Tag changes affect inherited configuration
+        cleared_count = datastore.clear_checksums_for_tag(uuid)
+        logger.info(f"Tag {uuid} updated, cleared {cleared_count} watch checksums")
+
         flash(gettext("Updated"))
 
         return redirect(url_for('tags.tags_overview_page'))
 
 
-    @tags_blueprint.route("/delete/<string:uuid>", methods=['GET'])
-    def form_tag_delete(uuid):
-        return redirect(url_for('tags.tags_overview_page'))
     return tags_blueprint

changedetectionio/blueprint/ui/__init__.py

@@ -141,7 +141,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, worker_pool,
     # Import the login decorator
     from changedetectionio.auth_decorator import login_optionally_required
 
-    @ui_blueprint.route("/clear_history/<string:uuid>", methods=['GET'])
+    @ui_blueprint.route("/clear_history/<uuid_str:uuid>", methods=['GET'])
     @login_optionally_required
     def clear_watch_history(uuid):
         try:
@@ -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
@@ -194,9 +194,9 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, worker_pool,
         tag_limit = request.args.get('tag')
         now = int(time.time())
 
-        # Mark watches as viewed in background thread to avoid blocking
+        # Mark watches as viewed - use background thread only for large watch counts
-        def mark_viewed_background():
+        def mark_viewed_impl():
-            """Background thread to mark watches as viewed - discarded after completion."""
+            """Mark watches as viewed - can run synchronously or in background thread."""
             marked_count = 0
             try:
                 for watch_uuid, watch in datastore.data['watching'].items():
@@ -209,15 +209,21 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, worker_pool,
                     datastore.set_last_viewed(watch_uuid, now)
                     marked_count += 1
 
-                logger.info(f"Background marking complete: {marked_count} watches marked as viewed")
+                logger.info(f"Marking complete: {marked_count} watches marked as viewed")
             except Exception as e:
-                logger.error(f"Error in background mark as viewed: {e}")
+                logger.error(f"Error marking as viewed: {e}")
 
-        # Start background thread and return immediately
+        # For small watch counts (< 10), run synchronously to avoid race conditions in tests
-        thread = threading.Thread(target=mark_viewed_background, daemon=True)
+        # For larger counts, use background thread to avoid blocking the UI
-        thread.start()
+        watch_count = len(datastore.data['watching'])
+        if watch_count < 10:
+            # Run synchronously for small watch counts
+            mark_viewed_impl()
+        else:
+            # Start background thread for large watch counts
+            thread = threading.Thread(target=mark_viewed_impl, daemon=True)
+            thread.start()
 
-        flash(gettext("Marking watches as viewed in background..."))
         return redirect(url_for('watchlist.index', tag=tag_limit))
 
     @ui_blueprint.route("/delete", methods=['GET'])
@@ -360,7 +366,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, worker_pool,
         return redirect(url_for('watchlist.index'))
 
 
-    @ui_blueprint.route("/share-url/<string:uuid>", methods=['GET'])
+    @ui_blueprint.route("/share-url/<uuid_str:uuid>", methods=['GET'])
     @login_optionally_required
     def form_share_put_watch(uuid):
         """Given a watch UUID, upload the info and return a share-link

changedetectionio/blueprint/ui/diff.py

@@ -66,7 +66,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
 
         return Markup(result)
 
-    @diff_blueprint.route("/diff/<string:uuid>", methods=['GET'])
+    @diff_blueprint.route("/diff/<uuid_str:uuid>", methods=['GET'])
     @login_optionally_required
     def diff_history_page(uuid):
         """
@@ -128,7 +128,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
             redirect=redirect
         )
 
-    @diff_blueprint.route("/diff/<string:uuid>/extract", methods=['GET'])
+    @diff_blueprint.route("/diff/<uuid_str:uuid>/extract", methods=['GET'])
     @login_optionally_required
     def diff_history_page_extract_GET(uuid):
         """
@@ -182,7 +182,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
             redirect=redirect
         )
 
-    @diff_blueprint.route("/diff/<string:uuid>/extract", methods=['POST'])
+    @diff_blueprint.route("/diff/<uuid_str:uuid>/extract", methods=['POST'])
     @login_optionally_required
     def diff_history_page_extract_POST(uuid):
         """
@@ -238,7 +238,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
             redirect=redirect
         )
 
-    @diff_blueprint.route("/diff/<string:uuid>/processor-asset/<string:asset_name>", methods=['GET'])
+    @diff_blueprint.route("/diff/<uuid_str:uuid>/processor-asset/<string:asset_name>", methods=['GET'])
     @login_optionally_required
     def processor_asset(uuid, asset_name):
         """

changedetectionio/blueprint/ui/edit.py

@@ -20,13 +20,13 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
             if tag_uuid in watch.get('tags', []) and (tag.get('include_filters') or tag.get('subtractive_selectors')):
                 return True
 
-    @edit_blueprint.route("/edit/<string:uuid>", methods=['GET', 'POST'])
+    @edit_blueprint.route("/edit/<uuid_str:uuid>", methods=['GET', 'POST'])
     @login_optionally_required
     # https://stackoverflow.com/questions/42984453/wtforms-populate-form-with-data-if-data-exists
     # https://wtforms.readthedocs.io/en/3.0.x/forms/#wtforms.form.Form.populate_obj ?
     def edit_page(uuid):
         from changedetectionio import forms
-        from changedetectionio.blueprint.browser_steps.browser_steps import browser_step_ui_config
+        from changedetectionio.browser_steps.browser_steps import browser_step_ui_config
         from changedetectionio import processors
         import importlib
 
@@ -117,12 +117,25 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
                 processor_config = processor_instance.get_extra_watch_config(config_filename)
 
                 if processor_config:
+                    from wtforms.fields.form import FormField
                     # Populate processor-config-* fields from JSON
                     for config_key, config_value in processor_config.items():
-                        field_name = f'processor_config_{config_key}'
+                        if not isinstance(config_value, dict):
-                        if hasattr(form, field_name):
+                            continue
-                            getattr(form, field_name).data = config_value
+                        # Try exact API-named field first (e.g., processor_config_restock_diff)
-                            logger.debug(f"Loaded processor config from {config_filename}: {field_name} = {config_value}")
+                        target_field = getattr(form, f'processor_config_{config_key}', None)
+                        # Fallback: find any FormField sub-form whose fields cover config_value keys
+                        if target_field is None:
+                            for form_field in form:
+                                if isinstance(form_field, FormField) and all(k in form_field.form._fields for k in config_value):
+                                    target_field = form_field
+                                    break
+                        if target_field is not None:
+                            for sub_key, sub_value in config_value.items():
+                                sub_field = target_field.form._fields.get(sub_key)
+                                if sub_field is not None:
+                                    sub_field.data = sub_value
+                                    logger.debug(f"Loaded processor config from {config_filename}: {sub_key} = {sub_value}")
             except Exception as e:
                 logger.warning(f"Failed to load processor config: {e}")
 
@@ -327,7 +340,7 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
 
         return output
 
-    @edit_blueprint.route("/edit/<string:uuid>/get-html", methods=['GET'])
+    @edit_blueprint.route("/edit/<uuid_str:uuid>/get-html", methods=['GET'])
     @login_optionally_required
     def watch_get_latest_html(uuid):
         from io import BytesIO
@@ -354,8 +367,58 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
         # Return a 500 error
         abort(500)
 
+    @edit_blueprint.route("/edit/<uuid_str:uuid>/get-data-package", methods=['GET'])
+    @login_optionally_required
+    def watch_get_data_package(uuid):
+        """Download all data for a single watch as a zip file"""
+        from io import BytesIO
+        from flask import send_file
+        import zipfile
+        from pathlib import Path
+        import datetime
+
+        watch = datastore.data['watching'].get(uuid)
+        if not watch:
+            abort(404)
+
+        # Create zip in memory
+        memory_file = BytesIO()
+
+        with zipfile.ZipFile(memory_file, 'w',
+                           compression=zipfile.ZIP_DEFLATED,
+                           compresslevel=8) as zipObj:
+
+            # Add the watch's JSON file if it exists
+            watch_json_path = os.path.join(watch.data_dir, 'watch.json')
+            if os.path.isfile(watch_json_path):
+                zipObj.write(watch_json_path,
+                           arcname=os.path.join(uuid, 'watch.json'),
+                           compress_type=zipfile.ZIP_DEFLATED,
+                           compresslevel=8)
+
+            # Add all files in the watch data directory
+            if os.path.isdir(watch.data_dir):
+                for f in Path(watch.data_dir).glob('*'):
+                    if f.is_file() and f.name != 'watch.json':  # Skip watch.json since we already added it
+                        zipObj.write(f,
+                                   arcname=os.path.join(uuid, f.name),
+                                   compress_type=zipfile.ZIP_DEFLATED,
+                                   compresslevel=8)
+
+        # Seek to beginning of file
+        memory_file.seek(0)
+
+        # Generate filename with timestamp
+        timestamp = datetime.datetime.now().strftime("%Y%m%d%H%M%S")
+        filename = f"watch-data-{uuid[:8]}-{timestamp}.zip"
+
+        return send_file(memory_file,
+                        as_attachment=True,
+                        download_name=filename,
+                        mimetype='application/zip')
+
     # Ajax callback
-    @edit_blueprint.route("/edit/<string:uuid>/preview-rendered", methods=['POST'])
+    @edit_blueprint.route("/edit/<uuid_str:uuid>/preview-rendered", methods=['POST'])
     @login_optionally_required
     def watch_get_preview_rendered(uuid):
         '''For when viewing the "preview" of the rendered text from inside of Edit'''

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/<string: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 = watch.fetcher_supports_screenshots
-        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
 
         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:
@@ -125,7 +124,7 @@ def construct_blueprint(datastore: ChangeDetectionStore):
 
         return output
 
-    @preview_blueprint.route("/preview/<string:uuid>/processor-asset/<string:asset_name>", methods=['GET'])
+    @preview_blueprint.route("/preview/<uuid_str:uuid>/processor-asset/<string:asset_name>", methods=['GET'])
     @login_optionally_required
     def processor_asset(uuid, asset_name):
         """

changedetectionio/blueprint/ui/templates/edit.html

@@ -488,6 +488,7 @@ Math: {{ 1 + 1 }}") }}
                     {% if watch.history_n %}
                         <p>
                              <a href="{{url_for('ui.ui_edit.watch_get_latest_html', uuid=uuid)}}" class="pure-button button-small">{{ _('Download latest HTML snapshot') }}</a>
+                             <a href="{{url_for('ui.ui_edit.watch_get_data_package', uuid=uuid)}}" class="pure-button button-small">{{ _('Download watch data package') }}</a>
                         </p>
                     {% endif %}
 

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,
@@ -110,6 +111,16 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
             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>
@@ -304,11 +305,20 @@ html[data-darkmode="true"] .watch-tag-list.tag-{{ class_name }} {
                             </span>
                         {%- endif -%}
 
-                        {%- if watch.get('restock') and watch['restock']['price'] != None -%}
+                        {%- if watch.get('restock') and watch['restock'].get('price') -%}
-                            {%- if watch['restock']['price'] != None -%}
+                            {%- set restock = watch['restock'] -%}
-                                <span class="restock-label price" title="{{ _('Price') }}">
+                            {%- set price = restock.get('price') -%}
-                                {{ watch['restock']['price']|format_number_locale if watch['restock'].get('price') else '' }} {{ watch['restock'].get('currency','') }}
+                            {%- set cur = restock.get('currency','') -%}
-                                </span>
+
+                            {%- 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>

changedetectionio/browser_steps/browser_steps.py

@@ -8,6 +8,17 @@ from changedetectionio.content_fetchers import SCREENSHOT_MAX_HEIGHT_DEFAULT
 from changedetectionio.content_fetchers.base import manage_user_agent
 from changedetectionio.jinja2_custom import render as jinja_render
 
+def browser_steps_get_valid_steps(browser_steps: list):
+    if browser_steps is not None and len(browser_steps):
+        valid_steps = list(filter(
+            lambda s: (s['operation'] and len(s['operation']) and s['operation'] != 'Choose one'),browser_steps))
+
+        # Just incase they selected Goto site by accident with older JS
+        if valid_steps and valid_steps[0]['operation'] == 'Goto site':
+            del(valid_steps[0])
+
+        return valid_steps
+    return []
 
 
 # Two flags, tell the JS which of the "Selector" or "Value" field should be enabled in the front end

changedetectionio/content_fetchers/base.py

@@ -38,7 +38,6 @@ def manage_user_agent(headers, current_ua=''):
 
     return None
 
-
 class Fetcher():
     browser_connection_is_custom = None
     browser_connection_url = None
@@ -163,30 +162,16 @@ class Fetcher():
         """
         return {k.lower(): v for k, v in self.headers.items()}
 
-    def browser_steps_get_valid_steps(self):
-        if self.browser_steps is not None and len(self.browser_steps):
-            valid_steps = list(filter(
-                lambda s: (s['operation'] and len(s['operation']) and s['operation'] != 'Choose one'),
-                self.browser_steps))
-
-            # Just incase they selected Goto site by accident with older JS
-            if valid_steps and valid_steps[0]['operation'] == 'Goto site':
-                del(valid_steps[0])
-
-            return valid_steps
-
-        return None
-
     async def iterate_browser_steps(self, start_url=None):
-        from changedetectionio.blueprint.browser_steps.browser_steps import steppable_browser_interface
+        from changedetectionio.browser_steps.browser_steps import steppable_browser_interface, browser_steps_get_valid_steps
         from playwright._impl._errors import TimeoutError, Error
         from changedetectionio.jinja2_custom import render as jinja_render
         step_n = 0
 
-        if self.browser_steps is not None and len(self.browser_steps):
+        if self.browser_steps:
             interface = steppable_browser_interface(start_url=start_url)
             interface.page = self.page
-            valid_steps = self.browser_steps_get_valid_steps()
+            valid_steps = browser_steps_get_valid_steps(self.browser_steps)
 
             for step in valid_steps:
                 step_n += 1

changedetectionio/content_fetchers/playwright.py

@@ -295,7 +295,7 @@ class fetcher(Fetcher):
             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.blueprint.browser_steps.browser_steps import steppable_browser_interface
+            from changedetectionio.browser_steps.browser_steps import steppable_browser_interface
             browsersteps_interface = steppable_browser_interface(start_url=url)
             browsersteps_interface.page = self.page
 
@@ -362,7 +362,7 @@ class fetcher(Fetcher):
             # Wrap remaining operations in try/finally to ensure cleanup
             try:
                 # Run Browser Steps here
-                if self.browser_steps_get_valid_steps():
+                if self.browser_steps:
                     try:
                         await self.iterate_browser_steps(start_url=url)
                     except BrowserStepsStepException:

changedetectionio/content_fetchers/puppeteer.py

@@ -86,8 +86,8 @@ async def capture_full_page(page, screenshot_format='JPEG', watch_uuid=None, loc
         # better than scrollTo incase they override it in the page
         await page.evaluate(
             """(y) => {
-                document.documentElement.scrollTop = y;
+                const el = document.scrollingElement;
-                document.body.scrollTop = y;
+                if (el) el.scrollTop = y;
             }""",
             y
         )
@@ -305,6 +305,8 @@ class fetcher(Fetcher):
                 await asyncio.wait_for(self.browser.close(), timeout=3.0)
             except Exception as cleanup_error:
                 logger.error(f"[{watch_uuid}] Failed to cleanup browser after page creation failure: {cleanup_error}")
+            finally:
+                self.browser = None
             raise
         
         # Add console handler to capture console.log from favicon fetcher
@@ -456,7 +458,7 @@ class fetcher(Fetcher):
 
         # Run Browser Steps here
         # @todo not yet supported, we switch to playwright in this case
-        #            if self.browser_steps_get_valid_steps():
+        #            if self.browser_steps:
         #                self.iterate_browser_steps()
 
 
@@ -532,6 +534,14 @@ class fetcher(Fetcher):
             )
         except asyncio.TimeoutError:
             raise (BrowserFetchTimedOut(msg=f"Browser connected but was unable to process the page in {max_time} seconds."))
+        finally:
+            # Internal cleanup on any exception/timeout - call quit() immediately
+            # This prevents connection leaks during exception bursts
+            # Worker.py's quit() call becomes a redundant safety net (idempotent)
+            try:
+                await self.quit(watch={'uuid': watch_uuid} if watch_uuid else None)
+            except Exception as cleanup_error:
+                logger.error(f"[{watch_uuid}] Error during internal quit() cleanup: {cleanup_error}")
 
 
 # Plugin registration for built-in fetcher

changedetectionio/content_fetchers/requests.py

@@ -1,12 +1,14 @@
 from loguru import logger
+from urllib.parse import urljoin, urlparse
 import hashlib
 import os
 import re
 import asyncio
-from functools import partial
+
 from changedetectionio import strtobool
 from changedetectionio.content_fetchers.exceptions import BrowserStepsInUnsupportedFetcher, EmptyReply, Non200ErrorCodeReceived
 from changedetectionio.content_fetchers.base import Fetcher
+from changedetectionio.validate_url import is_private_hostname
 
 
 # "html_requests" is listed as the default fetcher in store.py!
@@ -36,7 +38,7 @@ class fetcher(Fetcher):
         import requests
         from requests.exceptions import ProxyError, ConnectionError, RequestException
 
-        if self.browser_steps_get_valid_steps():
+        if self.browser_steps:
             raise BrowserStepsInUnsupportedFetcher(url=url)
 
         proxies = {}
@@ -79,14 +81,48 @@ class fetcher(Fetcher):
         if strtobool(os.getenv('ALLOW_FILE_URI', 'false')) and url.startswith('file://'):
             from requests_file import FileAdapter
             session.mount('file://', FileAdapter())
+
+        allow_iana_restricted = strtobool(os.getenv('ALLOW_IANA_RESTRICTED_ADDRESSES', 'false'))
+
         try:
+            # Fresh DNS check at fetch time — catches DNS rebinding regardless of add-time cache.
+            if not allow_iana_restricted:
+                parsed_initial = urlparse(url)
+                if parsed_initial.hostname and is_private_hostname(parsed_initial.hostname):
+                    raise Exception(f"Fetch blocked: '{url}' resolves to a private/reserved IP address. "
+                                    f"Set ALLOW_IANA_RESTRICTED_ADDRESSES=true to allow.")
+
             r = session.request(method=request_method,
                                 data=request_body.encode('utf-8') if type(request_body) is str else request_body,
                                 url=url,
                                 headers=request_headers,
                                 timeout=timeout,
                                 proxies=proxies,
-                                verify=False)
+                                verify=False,
+                                allow_redirects=False)
+
+            # Manually follow redirects so each hop's resolved IP can be validated,
+            # preventing SSRF via an open redirect on a public host.
+            current_url = url
+            for _ in range(10):
+                if not r.is_redirect:
+                    break
+                location = r.headers.get('Location', '')
+                redirect_url = urljoin(current_url, location)
+                if not allow_iana_restricted:
+                    parsed_redirect = urlparse(redirect_url)
+                    if parsed_redirect.hostname and is_private_hostname(parsed_redirect.hostname):
+                        raise Exception(f"Redirect blocked: '{redirect_url}' resolves to a private/reserved IP address.")
+                current_url = redirect_url
+                r = session.request('GET', redirect_url,
+                                    headers=request_headers,
+                                    timeout=timeout,
+                                    proxies=proxies,
+                                    verify=False,
+                                    allow_redirects=False)
+            else:
+                raise Exception("Too many redirects")
+
         except Exception as e:
             msg = str(e)
             if proxies and 'SOCKSHTTPSConnectionPool' in msg:
@@ -112,10 +148,32 @@ class fetcher(Fetcher):
                         # Default to UTF-8 for XML if no encoding found
                         r.encoding = 'utf-8'
                 else:
-                    # For other content types, use chardet
+                    # No charset in HTTP header - sniff encoding in priority order matching browsers
-                    encoding = chardet.detect(r.content)['encoding']
+                    # (WHATWG encoding sniffing algorithm):
-                    if encoding:
+                    # 1. BOM - highest confidence, check before anything else
-                        r.encoding = encoding
+                    # 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
 
@@ -184,7 +242,6 @@ class fetcher(Fetcher):
         )
 
     async def quit(self, watch=None):
-
         # In case they switched to `requests` fetcher from something else
         # Then the screenshot could be old, in any case, it's not used here.
         # REMOVE_REQUESTS_OLD_SCREENSHOTS - Mainly used for testing

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
@@ -27,7 +28,6 @@ from flask import (
     session,
     url_for,
 )
-from flask_compress import Compress as FlaskCompress
 from flask_restful import abort, Api
 from flask_cors import CORS
 
@@ -40,7 +40,7 @@ from loguru import logger
 
 from changedetectionio import __version__
 from changedetectionio import queuedWatchMetaData
-from changedetectionio.api import Watch, WatchHistory, WatchSingleHistory, WatchHistoryDiff, CreateWatch, Import, SystemInfo, Tag, Tags, Notifications, WatchFavicon
+from changedetectionio.api import Watch, WatchHistory, WatchSingleHistory, WatchHistoryDiff, CreateWatch, Import, SystemInfo, Tag, Tags, Notifications, WatchFavicon, Spec
 from changedetectionio.api.Search import Search
 from .time_handler import is_within_schedule
 from changedetectionio.languages import get_available_languages, get_language_codes, get_flag_for_locale, get_timeago_locale
@@ -69,15 +69,43 @@ socketio_server = None
 
 # Enable CORS, especially useful for the Chrome extension to operate from anywhere
 CORS(app)
+from werkzeug.routing import BaseConverter, ValidationError
+from uuid import UUID
+
+class StrictUUIDConverter(BaseConverter):
+    # Special sentinel values allowed in addition to strict UUIDs
+    _ALLOWED_SENTINELS = frozenset({'first'})
+
+    def to_python(self, value: str) -> str:
+        if value in self._ALLOWED_SENTINELS:
+            return value
+        try:
+            u = UUID(value)
+        except ValueError as e:
+            raise ValidationError() from e
+        # Reject non-standard formats (braces, URNs, no-hyphens)
+        if str(u) != value.lower():
+            raise ValidationError()
+        return str(u)
+
+    def to_url(self, value) -> str:
+        return str(value)
+
+# app setup (once)
+app.url_map.converters["uuid_str"] = StrictUUIDConverter
+
+# Flask-Compress handles HTTP compression, Socket.IO compression disabled to prevent memory leak.
+# There's also a bug between flask compress and socketio that causes some kind of slow memory leak
+# It's better to use compression on your reverse proxy (nginx etc) instead.
+if strtobool(os.getenv("FLASK_ENABLE_COMPRESSION")):
+    from flask_compress import Compress as FlaskCompress
+    app.config['COMPRESS_MIN_SIZE'] = 2096
+    app.config['COMPRESS_MIMETYPES'] = ['text/html', 'text/css', 'text/javascript', 'application/json', 'application/javascript', 'image/svg+xml']
+    # Use gzip only - smaller memory footprint than zstd/brotli (4-8KB vs 200-500KB contexts)
+    app.config['COMPRESS_ALGORITHM'] = ['gzip']
+    compress = FlaskCompress()
+    compress.init_app(app)
 
-# Super handy for compressing large BrowserSteps responses and others
-# Flask-Compress handles HTTP compression, Socket.IO compression disabled to prevent memory leak
-compress = FlaskCompress()
-app.config['COMPRESS_MIN_SIZE'] = 2096
-app.config['COMPRESS_MIMETYPES'] = ['text/html', 'text/css', 'text/javascript', 'application/json', 'application/javascript', 'image/svg+xml']
-# Use gzip only - smaller memory footprint than zstd/brotli (4-8KB vs 200-500KB contexts)
-app.config['COMPRESS_ALGORITHM'] = ['gzip']
-compress.init_app(app)
 app.config['TEMPLATES_AUTO_RELOAD'] = False
 
 
@@ -190,9 +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'])
@@ -356,6 +388,8 @@ def _jinja2_filter_fetcher_status_icons(fetcher_name):
 
     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):
     """Sanitize a tag title to create a valid CSS class name.
@@ -367,9 +401,8 @@ 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
@@ -457,28 +490,21 @@ def changedetection_app(config=None, datastore_o=None):
     available_languages = get_available_languages()
     language_codes = get_language_codes()
 
-    def get_locale():
+    _locale_aliases = {
-        # Locale aliases: map browser language codes to translation directory names
+        'zh-TW': 'zh_Hant_TW',  # Traditional Chinese: browser sends zh-TW, we use zh_Hant_TW
-        # This handles cases where browsers send standard codes (e.g., zh-TW)
+        'zh_TW': 'zh_Hant_TW',  # Also handle underscore variant
-        # but our translations use more specific codes (e.g., zh_Hant_TW)
+    }
-        locale_aliases = {
+    _locale_match_list = language_codes + list(_locale_aliases.keys())
-            '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
-        }
 
+    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(_locale_match_list)
-        browser_locale = request.accept_languages.best_match(language_codes + list(locale_aliases.keys()))
+        # 3. Map browser locale to our internal locale if needed
-
+        return _locale_aliases.get(browser_locale, browser_locale)
-        # 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
 
     # Initialize Babel with locale selector
     babel = Babel(app, locale_selector=get_locale)
@@ -530,22 +556,22 @@ def changedetection_app(config=None, datastore_o=None):
 
 
     watch_api.add_resource(WatchHistoryDiff,
-                           '/api/v1/watch/<string:uuid>/difference/<string:from_timestamp>/<string:to_timestamp>',
+                           '/api/v1/watch/<uuid_str:uuid>/difference/<string:from_timestamp>/<string:to_timestamp>',
                            resource_class_kwargs={'datastore': datastore})
     watch_api.add_resource(WatchSingleHistory,
-                           '/api/v1/watch/<string:uuid>/history/<string:timestamp>',
+                           '/api/v1/watch/<uuid_str:uuid>/history/<string:timestamp>',
                            resource_class_kwargs={'datastore': datastore, 'update_q': update_q})
     watch_api.add_resource(WatchFavicon,
-                           '/api/v1/watch/<string:uuid>/favicon',
+                           '/api/v1/watch/<uuid_str:uuid>/favicon',
                            resource_class_kwargs={'datastore': datastore})
     watch_api.add_resource(WatchHistory,
-                           '/api/v1/watch/<string:uuid>/history',
+                           '/api/v1/watch/<uuid_str:uuid>/history',
                            resource_class_kwargs={'datastore': datastore})
 
     watch_api.add_resource(CreateWatch, '/api/v1/watch',
                            resource_class_kwargs={'datastore': datastore, 'update_q': update_q})
 
-    watch_api.add_resource(Watch, '/api/v1/watch/<string:uuid>',
+    watch_api.add_resource(Watch, '/api/v1/watch/<uuid_str:uuid>',
                            resource_class_kwargs={'datastore': datastore, 'update_q': update_q})
 
     watch_api.add_resource(SystemInfo, '/api/v1/systeminfo',
@@ -558,7 +584,7 @@ def changedetection_app(config=None, datastore_o=None):
     watch_api.add_resource(Tags, '/api/v1/tags',
                            resource_class_kwargs={'datastore': datastore})
 
-    watch_api.add_resource(Tag, '/api/v1/tag', '/api/v1/tag/<string:uuid>',
+    watch_api.add_resource(Tag, '/api/v1/tag', '/api/v1/tag/<uuid_str:uuid>',
                            resource_class_kwargs={'datastore': datastore, 'update_q': update_q})
                            
     watch_api.add_resource(Search, '/api/v1/search',
@@ -567,6 +593,8 @@ def changedetection_app(config=None, datastore_o=None):
     watch_api.add_resource(Notifications, '/api/v1/notifications',
                            resource_class_kwargs={'datastore': datastore})
 
+    watch_api.add_resource(Spec, '/api/v1/full-spec')
+
     @login_manager.user_loader
     def user_loader(email):
         user = User()
@@ -708,8 +736,14 @@ def changedetection_app(config=None, datastore_o=None):
     def static_content(group, filename):
         from flask import make_response
         import re
-        group = re.sub(r'[^\w.-]+', '', group.lower())
+
-        filename = re.sub(r'[^\w.-]+', '', filename.lower())
+        # Strict sanitization: only allow a-z, 0-9, and underscore (blocks .. and other traversal)
+        group = re.sub(r'[^a-z0-9_-]+', '', group.lower())
+        filename = filename
+
+        # Additional safety: reject if sanitization resulted in empty strings
+        if not group or not filename:
+            abort(404)
 
         if group == 'screenshot':
             # Could be sensitive, follow password requirements
@@ -983,15 +1017,16 @@ 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

@@ -7,8 +7,6 @@ from flask_babel import lazy_gettext as _l, gettext
 from changedetectionio.blueprint.rss import RSS_FORMAT_TYPES, RSS_TEMPLATE_TYPE_OPTIONS, RSS_TEMPLATE_HTML_DEFAULT
 from changedetectionio.conditions.form import ConditionFormRow
 from changedetectionio.notification_service import NotificationContextData
-from changedetectionio.processors.image_ssim_diff import SCREENSHOT_COMPARISON_THRESHOLD_OPTIONS, \
-    SCREENSHOT_COMPARISON_THRESHOLD_OPTIONS_DEFAULT
 from changedetectionio.strtobool import strtobool
 from changedetectionio import processors
 
@@ -37,7 +35,7 @@ from changedetectionio.widgets import TernaryNoneBooleanField
 
 # default
 # each select <option data-enabled="enabled-0-0"
-from changedetectionio.blueprint.browser_steps.browser_steps import browser_step_ui_config
+from changedetectionio.browser_steps.browser_steps import browser_step_ui_config
 
 from changedetectionio import html_tools, content_fetchers
 
@@ -494,7 +492,6 @@ class ValidateJinja2Template(object):
     Validates that a {token} is from a valid set
     """
     def __call__(self, form, field):
-        from changedetectionio import notification
         from changedetectionio.jinja2_custom import create_jinja_env
         from jinja2 import BaseLoader, TemplateSyntaxError, UndefinedError
         from jinja2.meta import find_undeclared_variables
@@ -611,13 +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 changedetectionio.html_tools import SafeXPath3Parser
-                from elementpath.xpath3 import XPath3Parser
                 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)))
@@ -820,8 +816,7 @@ class processor_text_json_diff_form(commonSettingsForm):
     filter_text_removed = BooleanField(_l('Removed lines'), default=True)
 
     trigger_text = StringListField(_l('Keyword triggers - Trigger/wait for text'), [validators.Optional(), ValidateListRegex()])
-    if os.getenv("PLAYWRIGHT_DRIVER_URL"):
+    browser_steps = FieldList(FormField(SingleBrowserStep), min_entries=10)
-        browser_steps = FieldList(FormField(SingleBrowserStep), min_entries=10)
     text_should_not_be_present = StringListField(_l('Block change-detection while text matches'), [validators.Optional(), ValidateListRegex()])
     webdriver_js_execute_code = TextAreaField(_l('Execute JavaScript before change detection'), render_kw={"rows": "5"}, validators=[validators.Optional()])
 

changedetectionio/html_tools.py

@@ -23,6 +23,53 @@ 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',
+    '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 +230,6 @@ 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 +255,7 @@ 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 +280,9 @@ 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
@@ -439,13 +487,25 @@ 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
+        # Check for JSONP wrapper: someCallback({...}) or some.namespace({...})
-        try:
+        # Server may claim application/json but actually return JSONP
-            stripped_text_from_html = extract_json_blob_from_html(content=content,
+        jsonp_match = re.match(r'^\w[\w.]*\s*\((.+)\)\s*;?\s*$', content.lstrip("\ufeff").strip(), re.DOTALL)
-                                                                  ensure_is_ldjson_info_type=ensure_is_ldjson_info_type,
+        if jsonp_match:
-                                                                  json_filter=json_filter                                                                  )
+            try:
-        except json.JSONDecodeError as e:
+                inner = jsonp_match.group(1).strip()
-            logger.warning(f"Error processing JSON while extracting JSON from HTML blob {content[:20]}...{str(e)})")
+                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
@@ -561,10 +621,33 @@ def html_to_text(html_content: str, render_anchor_tag_content=False, is_rss=Fals
         )
     else:
         parser_config = None
-
     if is_rss:
         html_content = re.sub(r'<title([\s>])', r'<h1\1', html_content)
         html_content = re.sub(r'</title>', r'</h1>', html_content)
+    else:
+        # Use BS4 html.parser to strip bloat — SPA's often dump 10MB+ of CSS/JS into <head>,
+        # causing inscriptis to silently give up. Regex-based stripping is unsafe because tags
+        # can appear inside JSON data attributes with JS-escaped closing tags (e.g. <\/script>),
+        # causing the regex to scan past the intended close and eat real page content.
+        from bs4 import BeautifulSoup
+        soup = BeautifulSoup(html_content, 'html.parser')
+        # Strip tags that inscriptis cannot render as meaningful text and which can be very large.
+        # svg/math: produce path-data/MathML garbage; canvas/iframe/template: no inscriptis handlers.
+        # video/audio/picture are kept — they may contain meaningful fallback text or captions.
+        for tag in soup.find_all(['head', 'script', 'style', 'noscript', 'svg',
+                                  'math', 'canvas', 'iframe', 'template']):
+            tag.decompose()
+
+        # SPAs often use <body style="display:none"> to hide content until JS loads.
+        # inscriptis respects CSS display rules, so strip hiding styles from the body tag.
+        body_tag = soup.find('body')
+        if body_tag and body_tag.get('style'):
+            style = body_tag['style']
+            if re.search(r'\b(?:display\s*:\s*none|visibility\s*:\s*hidden)\b', style, re.IGNORECASE):
+                logger.debug(f"html_to_text: Removing hiding styles from body tag (found: '{style}')")
+                del body_tag['style']
+
+        html_content = str(soup)
 
     text_content = get_text(html_content, config=parser_config)
     return text_content

changedetectionio/languages.py

@@ -37,6 +37,7 @@ def get_timeago_locale(flask_locale):
         'no': 'nb_NO',      # Norwegian Bokmål
         'hi': 'in_HI',      # Hindi
         'cs': 'en',         # Czech not supported by timeago, fallback to English
+        'uk': 'uk',         # Ukrainian
         'en_GB': 'en',      # British English - timeago uses 'en'
         'en_US': 'en',      # American English - timeago uses 'en'
     }
@@ -67,6 +68,7 @@ LANGUAGE_DATA = {
     'tr': {'flag': 'fi fi-tr fis', 'name': 'Türkçe'},
     'ar': {'flag': 'fi fi-sa fis', 'name': 'العربية'},
     'hi': {'flag': 'fi fi-in fis', 'name': 'हिन्दी'},
+    'uk': {'flag': 'fi fi-ua fis', 'name': 'Українська'},
 }
 
 

changedetectionio/model/App.py

@@ -2,6 +2,7 @@ from os import getenv
 from copy import deepcopy
 
 from changedetectionio.blueprint.rss import RSS_FORMAT_TYPES, RSS_CONTENT_FORMAT_DEFAULT
+from changedetectionio.model.Tags import TagsDict
 
 from changedetectionio.notification import (
     default_notification_body,
@@ -68,7 +69,7 @@ class model(dict):
                     'schema_version' : 0,
                     'shared_diff_access': False,
                     'strip_ignored_lines': False,
-                    'tags': {}, #@todo use Tag.model initialisers
+                    'tags': None,  # Initialized in __init__ with real datastore_path
                     'webdriver_delay': None , # Extra delay in seconds before extracting text
                     'ui': {
                         'use_page_title_in_list': True,
@@ -80,10 +81,16 @@ class model(dict):
             }
         }
 
-    def __init__(self, *arg, **kw):
+    def __init__(self, *arg, datastore_path=None, **kw):
         super(model, self).__init__(*arg, **kw)
+        # Capture any tags data passed in before base_config overwrites the structure
+        existing_tags = self.get('settings', {}).get('application', {}).get('tags') or {}
         # CRITICAL: deepcopy to avoid sharing mutable objects between instances
         self.update(deepcopy(self.base_config))
+        # TagsDict requires the real datastore_path at runtime (cannot be set at class-definition time)
+        if datastore_path is None:
+            raise ValueError("App.model() requires 'datastore_path' keyword argument")
+        self['settings']['application']['tags'] = TagsDict(existing_tags, datastore_path=datastore_path)
 
 
 def parse_headers_from_text_file(filepath):

changedetectionio/model/Tags.py

@@ -0,0 +1,39 @@
+import os
+import shutil
+from pathlib import Path
+from loguru import logger
+
+_SENTINEL = object()
+
+
+class TagsDict(dict):
+    """Dict subclass that removes the corresponding tag.json file when a tag is deleted."""
+
+    def __init__(self, *args, datastore_path: str | os.PathLike, **kwargs) -> None:
+        self._datastore_path = Path(datastore_path)
+        super().__init__(*args, **kwargs)
+
+    def __delitem__(self, key: str) -> None:
+        super().__delitem__(key)
+        tag_dir = self._datastore_path / key
+        tag_json_file = tag_dir / "tag.json"
+        if not os.path.exists(tag_json_file):
+            logger.critical(f"Aborting deletion of directory '{tag_dir}' because '{tag_json_file}' does not exist.")
+            return
+        try:
+            shutil.rmtree(tag_dir)
+            logger.info(f"Deleted tag directory for tag {key!r}")
+        except FileNotFoundError:
+            pass
+        except OSError as e:
+            logger.error(f"Failed to delete tag directory for tag {key!r}: {e}")
+
+    def pop(self, key: str, default=_SENTINEL):
+        """Remove and return tag, deleting its tag.json file. Raises KeyError if missing and no default given."""
+        if key in self:
+            value = self[key]
+            del self[key]
+            return value
+        if default is _SENTINEL:
+            raise KeyError(key)
+        return default

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}
 
@@ -335,7 +340,6 @@ class model(EntityPersistenceMixin, watch_base):
             'last_notification_error': False,
             'last_viewed': 0,
             'previous_md5': False,
-            'previous_md5_before_filters': False,
             'remote_server_reply': None,
             'track_ldjson_price_data': None
         })
@@ -384,12 +388,37 @@ class model(EntityPersistenceMixin, watch_base):
 
         return self.get('fetch_backend')
 
+    @property
+    def fetcher_supports_screenshots(self):
+        """Return True if the fetcher configured for this watch supports screenshots.
+
+        Resolves 'system' via self._datastore, then checks supports_screenshots on
+        the actual fetcher class. Works for built-in and plugin fetchers alike.
+        """
+        from changedetectionio import content_fetchers
+
+        fetcher_name = self.get_fetch_backend  # already handles is_pdf → html_requests
+        if not fetcher_name or fetcher_name == 'system':
+            fetcher_name = self._datastore['settings']['application'].get('fetch_backend', 'html_requests')
+
+        fetcher_class = getattr(content_fetchers, fetcher_name, None)
+        if fetcher_class is None:
+            return False
+
+        return bool(getattr(fetcher_class, 'supports_screenshots', False))
+
     @property
     def is_pdf(self):
-        # content_type field is set in the future
+        url = str(self.get("url") or "").lower()
-        # https://github.com/dgtlmoon/changedetection.io/issues/1392
+        content_type = str(self.get("content-type") or "").lower()
-        # Not sure the best logic here
+
-        return self.get('url', '').lower().endswith('.pdf') or 'pdf' in self.get('content_type', '').lower()
+        if content_type in ("none", "null", ""):
+            content_type = ""
+
+        return (
+                url.endswith(".pdf")
+                or content_type.split(";")[0].strip() == "application/pdf"
+        )
 
     @property
     def label(self):
@@ -801,9 +830,8 @@ class model(EntityPersistenceMixin, watch_base):
                     with open(fname, 'wb') as f:
                         f.write(decoded)
 
-                    # Invalidate favicon filename cache
+                    # Invalidate module-level favicon filename cache for this watch
-                    if hasattr(self, '_favicon_filename_cache'):
+                    _FAVICON_FILENAME_CACHE.pop(self.data_dir, None)
-                        delattr(self, '_favicon_filename_cache')
 
                     # A signal that could trigger the socket server to update the browser also
                     watch_check_update = signal('watch_favicon_bump')
@@ -818,35 +846,23 @@ class model(EntityPersistenceMixin, watch_base):
 
     def get_favicon_filename(self) -> str | None:
         """
-        Find any favicon.* file in the current working directory
+        Find any favicon.* file in the watch data directory.
-        and return the contents of the newest one.
 
-        MEMORY LEAK FIX: Cache the result to avoid repeated glob.glob() operations.
+        Uses a module-level cache keyed by data_dir to survive Watch object recreation,
-        glob.glob() causes millions of fnmatch allocations when called for every watch on page load.
+        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)
+        if self.data_dir in _FAVICON_FILENAME_CACHE:
-        cache_key = '_favicon_filename_cache'
+            return _FAVICON_FILENAME_CACHE[self.data_dir]
-        if hasattr(self, cache_key):
-            return getattr(self, cache_key)
 
         import glob
-
-        # Search for all favicon.* files
         files = glob.glob(os.path.join(self.data_dir, "favicon.*"))
-
+        fname = os.path.basename(files[0]) if files else None
-        if not files:
+        _FAVICON_FILENAME_CACHE[self.data_dir] = fname
-            result = None
+        return fname
-        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
 
     def get_screenshot_as_thumbnail(self, max_age=3200):
         """Return path to a square thumbnail of the most recent screenshot.
@@ -1177,18 +1193,13 @@ 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:
+        has_app_context = has_request_context()
-            url_for('settings.settings_page')
-        except Exception as e:
-            has_app_context = False
-        else:
-            has_app_context = True
 
         # has app+request context, we can use url_for()
         if has_app_context:

changedetectionio/model/__init__.py

@@ -6,6 +6,8 @@ from .persistence import EntityPersistenceMixin, _determine_entity_type
 
 __all__ = ['EntityPersistenceMixin', 'watch_base']
 
+from ..browser_steps.browser_steps import browser_steps_get_valid_steps
+
 USE_SYSTEM_DEFAULT_NOTIFICATION_FORMAT_FOR_WATCH = 'System default'
 CONDITIONS_MATCH_LOGIC_DEFAULT = 'ALL'
 
@@ -129,7 +131,6 @@ class watch_base(dict):
         fetch_time (float): Duration of last fetch in seconds
         consecutive_filter_failures (int): Counter for consecutive filter match failures
         previous_md5 (str|bool): MD5 hash of previous content
-        previous_md5_before_filters (str|bool): MD5 hash before filters applied
         history_snapshot_max_length (int|None): Max history snapshots to keep (None = use global)
 
     Conditions:
@@ -166,6 +167,10 @@ class watch_base(dict):
         if kw.get('datastore_path'):
             del kw['datastore_path']
 
+        # IMPORTANT: Don't initialize __watch_was_edited yet!
+        # We'll initialize it AFTER the initial update() call below
+        # This prevents marking the watch as edited during initialization
+
         self.update({
             # Custom notification content
             # Re #110, so then if this is set to None, we know to use the default value instead
@@ -211,7 +216,6 @@ class watch_base(dict):
             'page_title': None, # <title> from the page
             'paused': False,
             'previous_md5': False,
-            'previous_md5_before_filters': False,  # Used for skipping changedetection entirely
             'processor': 'text_json_diff',  # could be restock_diff or others from .processors
             'price_change_threshold_percent': None,
             'proxy': None,  # Preferred proxy connection
@@ -297,9 +301,121 @@ class watch_base(dict):
 
         super(watch_base, self).__init__(*arg, **kw)
 
+        # Check if we're being initialized from an existing watch object
+        # that has was_edited=True, so we can preserve the flag
+        preserve_edited_flag = False
         if self.get('default'):
+            # When creating a new watch object from an existing one (e.g., changing processor),
+            # preserve the was_edited flag if it was True
+            default_watch = self.get('default')
+            if hasattr(default_watch, 'was_edited') and default_watch.was_edited:
+                preserve_edited_flag = True
             del self['default']
 
+        # NOW initialize the edited flag after all initial setup is complete
+        # This ensures initialization doesn't trigger the edited flag
+        # But preserve it if the source watch had it set to True
+        self.__watch_was_edited = preserve_edited_flag
+
+    def _mark_field_as_edited(self, key):
+        """
+        Helper to mark a field as edited if it's writable.
+
+        Internal method used by __setitem__, update(), pop(), etc.
+        """
+        # Don't track edits during initial load or if already edited
+        if not hasattr(self, '_watch_base__watch_was_edited'):
+            return
+        if self.__watch_was_edited:
+            return  # Already marked as edited
+
+        # Import from shared schema utilities (no circular dependency)
+        from .schema_utils import get_readonly_watch_fields
+        readonly_fields = get_readonly_watch_fields()
+
+        # Additional system-managed fields not in OpenAPI spec (yet)
+        # These are set by processors/workers and should not trigger edited flag
+        additional_system_fields = {
+            'last_check_status',  # Set by processors
+            'restock',  # Set by restock processor
+            'last_viewed',  # Set by mark_all_viewed endpoint
+        }
+
+        # Only mark as edited if this is a user-writable field
+        if key not in readonly_fields and key not in additional_system_fields:
+            self.__watch_was_edited = True
+
+    def __setitem__(self, key, value):
+        """
+        Override dict.__setitem__ to track when writable watch fields are modified.
+
+        This enables skipping reprocessing when:
+        1. HTML content is unchanged (checksumFromPreviousCheckWasTheSame)
+        2. AND watch configuration was not edited
+
+        Only sets the edited flag when field is NOT in readonly_fields (from OpenAPI spec).
+        """
+        # Set the value first (always)
+        super().__setitem__(key, value)
+        # Mark as edited if writable field
+        self._mark_field_as_edited(key)
+
+    def __delitem__(self, key):
+        """Override dict.__delitem__ to track deletions of writable fields."""
+        super().__delitem__(key)
+        self._mark_field_as_edited(key)
+
+    def update(self, *args, **kwargs):
+
+        if args and args[0].get('browser_steps'):
+            args[0]['browser_steps'] = browser_steps_get_valid_steps(args[0].get('browser_steps'))
+
+        """Override dict.update() to track modifications to writable fields."""
+        # Call parent update first
+        super().update(*args, **kwargs)
+
+        # Mark as edited for any writable fields that were updated
+        # Handle both update(dict) and update(key=value) forms
+        if args:
+            for key in args[0].keys():
+                self._mark_field_as_edited(key)
+        for key in kwargs.keys():
+            self._mark_field_as_edited(key)
+
+
+    def pop(self, key, *args):
+        """Override dict.pop() to track removal of writable fields."""
+        result = super().pop(key, *args)
+        self._mark_field_as_edited(key)
+        return result
+
+    def setdefault(self, key, default=None):
+        """Override dict.setdefault() to track modifications to writable fields."""
+        # Only marks as edited if key didn't exist (i.e., a new value was set)
+        existed = key in self
+        result = super().setdefault(key, default)
+        if not existed:
+            self._mark_field_as_edited(key)
+        return result
+
+    @property
+    def was_edited(self):
+        """
+        Check if watch configuration was edited since last processing.
+
+        Returns:
+            bool: True if writable fields were modified, False otherwise
+        """
+        return getattr(self, '_watch_base__watch_was_edited', False)
+
+    def reset_watch_edited_flag(self):
+        """
+        Reset the watch edited flag after successful processing.
+
+        Call this after processing completes to allow future content-only change detection.
+        """
+        self.__watch_was_edited = False
+
     @classmethod
     def get_property_names(cls):
         """

changedetectionio/model/schema_utils.py

@@ -0,0 +1,92 @@
+"""
+Schema utilities for Watch and Tag models.
+
+Provides functions to extract readonly fields and properties from OpenAPI spec.
+Shared by both the model layer and API layer to avoid circular dependencies.
+"""
+
+import functools
+
+
+@functools.cache
+def get_openapi_schema_dict():
+    """
+    Get the raw OpenAPI spec dictionary for schema access.
+
+    Returns the YAML dict directly (not the OpenAPI object).
+    """
+    import os
+    import yaml
+
+    spec_path = os.path.join(os.path.dirname(__file__), '../../docs/api-spec.yaml')
+    if not os.path.exists(spec_path):
+        spec_path = os.path.join(os.path.dirname(__file__), '../docs/api-spec.yaml')
+
+    with open(spec_path, 'r', encoding='utf-8') as f:
+        return yaml.safe_load(f)
+
+
+@functools.cache
+def _resolve_readonly_fields(schema_name):
+    """
+    Generic helper to resolve readOnly fields, including allOf inheritance.
+
+    Args:
+        schema_name: Name of the schema (e.g., 'Watch', 'Tag')
+
+    Returns:
+        frozenset: All readOnly field names including inherited ones
+    """
+    spec_dict = get_openapi_schema_dict()
+    schema = spec_dict['components']['schemas'].get(schema_name, {})
+
+    readonly_fields = set()
+
+    # Handle allOf (schema inheritance)
+    if 'allOf' in schema:
+        for item in schema['allOf']:
+            # Resolve $ref to parent schema
+            if '$ref' in item:
+                ref_path = item['$ref'].split('/')[-1]
+                ref_schema = spec_dict['components']['schemas'].get(ref_path, {})
+                if 'properties' in ref_schema:
+                    for field_name, field_def in ref_schema['properties'].items():
+                        if field_def.get('readOnly') is True:
+                            readonly_fields.add(field_name)
+            # Check schema-specific properties
+            if 'properties' in item:
+                for field_name, field_def in item['properties'].items():
+                    if field_def.get('readOnly') is True:
+                        readonly_fields.add(field_name)
+    else:
+        # Direct properties (no inheritance)
+        if 'properties' in schema:
+            for field_name, field_def in schema['properties'].items():
+                if field_def.get('readOnly') is True:
+                    readonly_fields.add(field_name)
+
+    return frozenset(readonly_fields)
+
+
+@functools.cache
+def get_readonly_watch_fields():
+    """
+    Extract readOnly field names from Watch schema in OpenAPI spec.
+
+    Returns readOnly fields from WatchBase (uuid, date_created) + Watch-specific readOnly fields.
+
+    Used by:
+    - model/watch_base.py: Track when writable fields are edited
+    - api/Watch.py: Filter readonly fields from PUT requests
+    """
+    return _resolve_readonly_fields('Watch')
+
+
+@functools.cache
+def get_readonly_tag_fields():
+    """
+    Extract readOnly field names from Tag schema in OpenAPI spec.
+
+    Returns readOnly fields from WatchBase (uuid, date_created) + Tag-specific readOnly fields.
+    """
+    return _resolve_readonly_fields('Tag')

changedetectionio/notification_service.py

@@ -54,34 +54,128 @@ def _check_cascading_vars(datastore, var_name, watch):
     return None
 
 
+class FormattableTimestamp(str):
+    """
+    A str subclass representing a formatted datetime. As a plain string it renders
+    with the default format, but can also be called with a custom format argument
+    in Jinja2 templates:
+
+        {{ change_datetime }}                        → '2024-01-15 10:30:00 UTC'
+        {{ change_datetime(format='%Y') }}           → '2024'
+        {{ change_datetime(format='%A') }}           → 'Monday'
+        {{ change_datetime(format='%Y-%m-%d') }}     → '2024-01-15'
+
+    Being a str subclass means it is natively JSON serializable.
+    """
+    _DEFAULT_FORMAT = '%Y-%m-%d %H:%M:%S %Z'
+
+    def __new__(cls, timestamp):
+        dt = datetime.datetime.fromtimestamp(int(timestamp), tz=pytz.UTC)
+        local_tz = datetime.datetime.now().astimezone().tzinfo
+        dt_local = dt.astimezone(local_tz)
+        try:
+            formatted = dt_local.strftime(cls._DEFAULT_FORMAT)
+        except Exception:
+            formatted = dt_local.isoformat()
+        instance = super().__new__(cls, formatted)
+        instance._dt = dt_local
+        return instance
+
+    def __call__(self, format=_DEFAULT_FORMAT):
+        try:
+            return self._dt.strftime(format)
+        except Exception:
+            return self._dt.isoformat()
+
+
+class FormattableDiff(str):
+    """
+    A str subclass representing a rendered diff. As a plain string it renders
+    with the default options for that variant, but can be called with custom
+    arguments in Jinja2 templates:
+
+        {{ diff }}                                    → default diff output
+        {{ diff(lines=5) }}                           → truncate to 5 lines
+        {{ diff(added_only=true) }}                   → only show added lines
+        {{ diff(removed_only=true) }}                 → only show removed lines
+        {{ diff(context=3) }}                         → 3 lines of context around changes
+        {{ diff(word_diff=false) }}                   → line-level diff instead of word-level
+        {{ diff(lines=10, added_only=true) }}         → combine args
+        {{ diff_added(lines=5) }}                     → works on any diff_* variant too
+
+    Being a str subclass means it is natively JSON serializable.
+    """
+    def __new__(cls, prev_snapshot, current_snapshot, **base_kwargs):
+        if prev_snapshot or current_snapshot:
+            from changedetectionio import diff as diff_module
+            rendered = diff_module.render_diff(prev_snapshot, current_snapshot, **base_kwargs)
+        else:
+            rendered = ''
+        instance = super().__new__(cls, rendered)
+        instance._prev = prev_snapshot
+        instance._current = current_snapshot
+        instance._base_kwargs = base_kwargs
+        return instance
+
+    def __call__(self, lines=None, added_only=False, removed_only=False, context=0,
+                 word_diff=None, case_insensitive=False, ignore_junk=False):
+        from changedetectionio import diff as diff_module
+        kwargs = dict(self._base_kwargs)
+
+        if added_only:
+            kwargs['include_removed'] = False
+        if removed_only:
+            kwargs['include_added'] = False
+        if context:
+            kwargs['context_lines'] = int(context)
+        if word_diff is not None:
+            kwargs['word_diff'] = bool(word_diff)
+        if case_insensitive:
+            kwargs['case_insensitive'] = True
+        if ignore_junk:
+            kwargs['ignore_junk'] = True
+
+        result = diff_module.render_diff(self._prev or '', self._current or '', **kwargs)
+
+        if lines is not None:
+            result = '\n'.join(result.splitlines()[:int(lines)])
+
+        return result
+
+
+
 # What is passed around as notification context, also used as the complete list of valid {{ tokens }}
 class NotificationContextData(dict):
     def __init__(self, initial_data=None, **kwargs):
+        # ValidateJinja2Template() validates against the keynames of this dict to check for valid tokens in the body (user submission)
         super().__init__({
             'base_url': None,
+            'change_datetime': FormattableTimestamp(time.time()),
             'current_snapshot': None,
-            'diff': None,
+            'diff': FormattableDiff('', ''),
-            'diff_clean': None,
+            'diff_clean': FormattableDiff('', '', include_change_type_prefix=False),
-            'diff_added': None,
+            'diff_added': FormattableDiff('', '', include_removed=False),
-            'diff_added_clean': None,
+            'diff_added_clean': FormattableDiff('', '', include_removed=False, include_change_type_prefix=False),
-            'diff_full': None,
+            'diff_full': FormattableDiff('', '', include_equal=True),
-            'diff_full_clean': None,
+            'diff_full_clean': FormattableDiff('', '', include_equal=True, include_change_type_prefix=False),
-            'diff_patch': None,
+            'diff_patch': FormattableDiff('', '', patch_format=True),
-            'diff_removed': None,
+            'diff_removed': FormattableDiff('', '', include_added=False),
-            'diff_removed_clean': None,
+            'diff_removed_clean': FormattableDiff('', '', include_added=False, include_change_type_prefix=False),
             'diff_url': None,
             'markup_text_links_to_html_links': False, # If automatic conversion of plaintext to HTML should happen
             'notification_timestamp': time.time(),
+            'prev_snapshot': None,
             'preview_url': None,
             'screenshot': None,
-            'triggered_text': None,
             'timestamp_from': None,
             'timestamp_to': None,
+            'triggered_text': None,
             'uuid': 'XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX',  # Converted to 'watch_uuid' in create_notification_parameters
             'watch_mime_type': None,
             'watch_tag': None,
             'watch_title': None,
             'watch_url': 'https://WATCH-PLACE-HOLDER/',
+            'watch_uuid': 'XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX',  # Converted to 'watch_uuid' in create_notification_parameters
         })
 
         # Apply any initial data passed in
@@ -103,7 +197,7 @@ class NotificationContextData(dict):
         So we can test the output in the notification body
         """
         for key in self.keys():
-            if key in ['uuid', 'time', 'watch_uuid']:
+            if key in ['uuid', 'time', 'watch_uuid', 'change_datetime'] or key.startswith('diff'):
                 continue
             rand_str = 'RANDOM-PLACEHOLDER-'+''.join(random.choices(string.ascii_letters + string.digits, k=12))
             self[key] = rand_str
@@ -115,24 +209,6 @@ class NotificationContextData(dict):
 
         super().__setitem__(key, value)
 
-def timestamp_to_localtime(timestamp):
-    # Format the date using locale-aware formatting with timezone
-    dt = datetime.datetime.fromtimestamp(int(timestamp))
-    dt = dt.replace(tzinfo=pytz.UTC)
-
-    # Get local timezone-aware datetime
-    local_tz = datetime.datetime.now().astimezone().tzinfo
-    local_dt = dt.astimezone(local_tz)
-
-    # Format date with timezone - using strftime for locale awareness
-    try:
-        formatted_date = local_dt.strftime('%Y-%m-%d %H:%M:%S %Z')
-    except:
-        # Fallback if locale issues
-        formatted_date = local_dt.isoformat()
-
-    return formatted_date
-
 def add_rendered_diff_to_notification_vars(notification_scan_text:str, prev_snapshot:str, current_snapshot:str, word_diff:bool):
     """
     Efficiently renders only the diff placeholders that are actually used in the notification text.
@@ -150,13 +226,12 @@ def add_rendered_diff_to_notification_vars(notification_scan_text:str, prev_snap
     Returns:
         dict: Only the diff placeholders that were found in notification_scan_text, with rendered content
     """
-    from changedetectionio import diff
     import re
-    from functools import lru_cache
 
     now = time.time()
 
-    # Define specifications for each diff variant
+    # Define base kwargs for each diff variant — these become the stored defaults
+    # on the FormattableDiff object, so {{ diff(lines=5) }} overrides on top of them
     diff_specs = {
         'diff': {'word_diff': word_diff},
         'diff_clean': {'word_diff': word_diff, 'include_change_type_prefix': False},
@@ -169,22 +244,15 @@ def add_rendered_diff_to_notification_vars(notification_scan_text:str, prev_snap
         'diff_removed_clean': {'word_diff': word_diff, 'include_added': False, 'include_change_type_prefix': False},
     }
 
-    # Memoize render_diff to avoid duplicate renders with same kwargs
-    @lru_cache(maxsize=4)
-    def cached_render(kwargs_tuple):
-        return diff.render_diff(prev_snapshot, current_snapshot, **dict(kwargs_tuple))
-
     ret = {}
     rendered_count = 0
-    # Only check and render diff keys that exist in NotificationContextData
+    # Only create FormattableDiff objects for diff keys actually used in the notification text
     for key in NotificationContextData().keys():
         if key.startswith('diff') and key in diff_specs:
             # Check if this placeholder is actually used in the notification text
             pattern = rf"(?<![A-Za-z0-9_]){re.escape(key)}(?![A-Za-z0-9_])"
             if re.search(pattern, notification_scan_text, re.IGNORECASE):
-                kwargs = diff_specs[key]
+                ret[key] = FormattableDiff(prev_snapshot, current_snapshot, **diff_specs[key])
-                # Convert dict to sorted tuple for cache key (handles duplicate kwarg combinations)
-                ret[key] = cached_render(tuple(sorted(kwargs.items())))
                 rendered_count += 1
 
     if rendered_count:
@@ -198,7 +266,7 @@ def set_basic_notification_vars(current_snapshot, prev_snapshot, watch, triggere
         'current_snapshot': current_snapshot,
         'prev_snapshot': prev_snapshot,
         'screenshot': watch.get_screenshot() if watch and watch.get('notification_screenshot') else None,
-        'change_datetime': timestamp_to_localtime(timestamp_changed) if timestamp_changed else None,
+        'change_datetime': FormattableTimestamp(timestamp_changed) if timestamp_changed else None,
         'triggered_text': triggered_text,
         'uuid': watch.get('uuid') if watch else None,
         'watch_url': watch.get('url') if watch else None,

changedetectionio/pluggy_interface.py

@@ -129,6 +129,51 @@ class ChangeDetectionSpec:
         """
         pass
 
+    @hookspec
+    def update_handler_alter(update_handler, watch, datastore):
+        """Modify or wrap the update_handler before it processes a watch.
+
+        This hook is called after the update_handler (perform_site_check instance) is created
+        but before it calls call_browser() and run_changedetection(). Plugins can use this to:
+        - Wrap the handler to add logging/metrics
+        - Modify handler configuration
+        - Add custom preprocessing logic
+
+        Args:
+            update_handler: The perform_site_check instance that will process the watch
+            watch: The watch dict being processed
+            datastore: The application datastore
+
+        Returns:
+            object or None: Return a modified/wrapped handler, or None to keep the original.
+                           If multiple plugins return handlers, they are chained in registration order.
+        """
+        pass
+
+    @hookspec
+    def update_finalize(update_handler, watch, datastore, processing_exception):
+        """Called after watch processing completes (success or failure).
+
+        This hook is called in the finally block after all processing is complete,
+        allowing plugins to perform cleanup, update metrics, or log final status.
+
+        The plugin can access update_handler.last_logging_insert_id if it was stored
+        during update_handler_alter, and use processing_exception to determine if
+        the processing succeeded or failed.
+
+        Args:
+            update_handler: The perform_site_check instance (may be None if creation failed)
+            watch: The watch dict that was processed (may be None if not loaded)
+            datastore: The application datastore
+            processing_exception: The exception from the main processing block, or None if successful.
+                                 This does NOT include cleanup exceptions - only exceptions from
+                                 the actual watch processing (fetch, diff, etc).
+
+        Returns:
+            None: This hook doesn't return a value
+        """
+        pass
+
 
 # Set up Plugin Manager
 plugin_manager = pluggy.PluginManager(PLUGIN_NAMESPACE)
@@ -499,4 +544,66 @@ def get_plugin_template_paths():
                 template_paths.append(templates_dir)
                 logger.debug(f"Added plugin template path: {templates_dir}")
 
-    return template_paths
+    return template_paths
+
+
+def apply_update_handler_alter(update_handler, watch, datastore):
+    """Apply update_handler_alter hooks from all plugins.
+
+    Allows plugins to wrap or modify the update_handler before it processes a watch.
+    Multiple plugins can chain modifications - each plugin receives the result from
+    the previous plugin.
+
+    Args:
+        update_handler: The perform_site_check instance to potentially modify
+        watch: The watch dict being processed
+        datastore: The application datastore
+
+    Returns:
+        object: The (potentially modified/wrapped) update_handler
+    """
+    # Get all plugins that implement the update_handler_alter hook
+    results = plugin_manager.hook.update_handler_alter(
+        update_handler=update_handler,
+        watch=watch,
+        datastore=datastore
+    )
+
+    # Chain results - each plugin gets the result from the previous one
+    current_handler = update_handler
+    if results:
+        for result in results:
+            if result is not None:
+                logger.debug(f"Plugin modified update_handler for watch {watch.get('uuid')}")
+                current_handler = result
+
+    return current_handler
+
+
+def apply_update_finalize(update_handler, watch, datastore, processing_exception):
+    """Apply update_finalize hooks from all plugins.
+
+    Called in the finally block after watch processing completes, allowing plugins
+    to perform cleanup, update metrics, or log final status.
+
+    Args:
+        update_handler: The perform_site_check instance (may be None)
+        watch: The watch dict that was processed (may be None)
+        datastore: The application datastore
+        processing_exception: The exception from processing, or None if successful
+
+    Returns:
+        None
+    """
+    try:
+        # Call all plugins that implement the update_finalize hook
+        plugin_manager.hook.update_finalize(
+            update_handler=update_handler,
+            watch=watch,
+            datastore=datastore,
+            processing_exception=processing_exception
+        )
+    except Exception as e:
+        # Don't let plugin errors crash the worker
+        logger.error(f"Error in update_finalize hook: {e}")
+        logger.exception(f"update_finalize hook exception details:")

changedetectionio/processors/README.md

@@ -9,6 +9,15 @@ Some suggestions for the future
 
 - `graphical` 
 
+## API schema extension (`api.yaml`)
+
+A processor can extend the Watch/Tag API schema by placing an `api.yaml` alongside its `__init__.py`.
+Define a `components.schemas.processor_config_<name>` entry and it will be merged into `WatchBase` at startup,
+making `processor_config_<name>` a valid field on all watch create/update API calls.
+The fully merged spec is served live at `/api/v1/full-spec`.
+
+See `restock_diff/api.yaml` for a working example.
+
 ## Todo
 
 - Make each processor return a extra list of sub-processed (so you could configure a single processor in different ways)

changedetectionio/processors/__init__.py

@@ -1,6 +1,6 @@
 from functools import lru_cache
 from loguru import logger
-from flask_babel import gettext
+from flask_babel import gettext, get_locale
 import importlib
 import inspect
 import os
@@ -190,14 +190,15 @@ def get_plugin_processor_metadata():
         logger.warning(f"Error getting plugin processor metadata: {e}")
     return metadata
 
-
+@lru_cache(maxsize=32)
-def available_processors():
+def _available_processors_cached(locale_str):
-    """
-    Get a list of processors by name and description for the UI elements.
-    Can be filtered via DISABLED_PROCESSORS environment variable (comma-separated list).
-    :return: A list :)
     """
+    Internal cached function that includes locale in cache key.
+    This ensures translations are cached per-language instead of globally.
 
+    :param locale_str: The locale string (e.g., 'en', 'it', 'zh')
+    :return: A list of tuples (processor_name, translated_description, weight)
+    """
     processor_classes = find_processors()
 
     # Check if DISABLED_PROCESSORS env var is set
@@ -256,6 +257,22 @@ def available_processors():
     # Return as tuples without weight (for backwards compatibility)
     return [(name, desc) for name, desc, weight in available]
 
+def available_processors():
+    """
+    Get a list of processors by name and description for the UI elements.
+    Can be filtered via DISABLED_PROCESSORS environment variable (comma-separated list).
+
+    This function delegates to a locale-aware cached version to ensure translations
+    are cached per-language instead of globally.
+
+    :return: A list of tuples (processor_name, translated_description)
+    """
+    # Get current locale and use it as cache key
+    # Convert Babel Locale object to string for use as cache key
+    locale = get_locale()
+    locale_str = str(locale) if locale else 'en'
+    return _available_processors_cached(locale_str)
+
 
 def get_default_processor():
     """

changedetectionio/processors/base.py

@@ -1,10 +1,15 @@
+import asyncio
 import re
 import hashlib
+
+from changedetectionio.browser_steps.browser_steps import browser_steps_get_valid_steps
 from changedetectionio.content_fetchers.base import Fetcher
 from changedetectionio.strtobool import strtobool
+from changedetectionio.validate_url import is_private_hostname
 from copy import deepcopy
 from abc import abstractmethod
 import os
+from urllib.parse import urlparse
 from loguru import logger
 
 SCREENSHOT_FORMAT_JPEG = 'JPEG'
@@ -19,6 +24,7 @@ class difference_detection_processor():
     xpath_data = None
     preferred_proxy = None
     screenshot_format = SCREENSHOT_FORMAT_JPEG
+    last_raw_content_checksum = None
 
     def __init__(self, datastore, watch_uuid):
         self.datastore = datastore
@@ -34,6 +40,81 @@ class difference_detection_processor():
         # Generic fetcher that should be extended (requests, playwright etc)
         self.fetcher = Fetcher()
 
+        # Load the last raw content checksum from file
+        self.read_last_raw_content_checksum()
+
+    def update_last_raw_content_checksum(self, checksum):
+        """
+        Save the raw content MD5 checksum to file.
+        This is used for skip logic - avoid reprocessing if raw HTML unchanged.
+        """
+        if not checksum:
+            return
+
+        watch = self.datastore.data['watching'].get(self.watch_uuid)
+        if not watch:
+            return
+
+        data_dir = watch.data_dir
+        if not data_dir:
+            return
+
+        watch.ensure_data_dir_exists()
+        checksum_file = os.path.join(data_dir, 'last-checksum.txt')
+
+        try:
+            with open(checksum_file, 'w', encoding='utf-8') as f:
+                f.write(checksum)
+            self.last_raw_content_checksum = checksum
+        except IOError as e:
+            logger.warning(f"Failed to write checksum file for {self.watch_uuid}: {e}")
+
+    def read_last_raw_content_checksum(self):
+        """
+        Read the last raw content MD5 checksum from file.
+        Returns None if file doesn't exist (first run) or can't be read.
+        """
+        watch = self.datastore.data['watching'].get(self.watch_uuid)
+        if not watch:
+            self.last_raw_content_checksum = None
+            return
+
+        data_dir = watch.data_dir
+        if not data_dir:
+            self.last_raw_content_checksum = None
+            return
+
+        checksum_file = os.path.join(data_dir, 'last-checksum.txt')
+
+        if not os.path.isfile(checksum_file):
+            self.last_raw_content_checksum = None
+            return
+
+        try:
+            with open(checksum_file, 'r', encoding='utf-8') as f:
+                self.last_raw_content_checksum = f.read().strip()
+        except IOError as e:
+            logger.warning(f"Failed to read checksum file for {self.watch_uuid}: {e}")
+            self.last_raw_content_checksum = None
+
+
+    async def validate_iana_url(self):
+        """Pre-flight SSRF check — runs DNS lookup in executor to avoid blocking the event loop.
+        Covers all fetchers (requests, playwright, puppeteer, plugins) since every fetch goes
+        through call_browser().
+        """
+        if strtobool(os.getenv('ALLOW_IANA_RESTRICTED_ADDRESSES', 'false')):
+            return
+        parsed = urlparse(self.watch.link)
+        if not parsed.hostname:
+            return
+        loop = asyncio.get_running_loop()
+        if await loop.run_in_executor(None, is_private_hostname, parsed.hostname):
+            raise Exception(
+                f"Fetch blocked: '{self.watch.link}' resolves to a private/reserved IP address. "
+                f"Set ALLOW_IANA_RESTRICTED_ADDRESSES=true to allow."
+            )
+
     async def call_browser(self, preferred_proxy_id=None):
 
         from requests.structures import CaseInsensitiveDict
@@ -47,6 +128,8 @@ class difference_detection_processor():
                     "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')
 
@@ -110,7 +193,7 @@ class difference_detection_processor():
                                    )
 
         if self.watch.has_browser_steps:
-            self.fetcher.browser_steps = self.watch.get('browser_steps', [])
+            self.fetcher.browser_steps = browser_steps_get_valid_steps(self.watch.get('browser_steps', []))
             self.fetcher.browser_steps_screenshot_path = os.path.join(self.datastore.datastore_path, self.watch.get('uuid'))
 
         # Tweak the base config with the per-watch ones
@@ -177,6 +260,16 @@ 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)
 
+        # 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
 
     def get_extra_watch_config(self, filename):
@@ -257,8 +350,16 @@ class difference_detection_processor():
         except IOError as e:
             logger.error(f"Failed to write extra watch config {filename}: {e}")
 
+    def get_raw_document_checksum(self):
+        checksum = None
+
+        if self.fetcher.content:
+            checksum = hashlib.md5(self.fetcher.content.encode('utf-8')).hexdigest()
+
+        return checksum
+
     @abstractmethod
-    def run_changedetection(self, watch):
+    def run_changedetection(self, watch, force_reprocess=False):
         update_obj = {'last_notification_error': False, 'last_error': False}
         some_data = 'xxxxx'
         update_obj["previous_md5"] = hashlib.md5(some_data.encode('utf-8')).hexdigest()

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 = watch.fetcher_supports_screenshots
-    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
 
     password_enabled_and_share_is_off = False
     if datastore.data['settings']['application'].get('password') or os.getenv("SALTED_PASS", False):

changedetectionio/processors/image_ssim_diff/processor.py

@@ -30,7 +30,7 @@ class perform_site_check(difference_detection_processor):
     # Override to use PNG format for better image comparison (JPEG compression creates noise)
     screenshot_format = SCREENSHOT_FORMAT_PNG
 
-    def run_changedetection(self, watch):
+    def run_changedetection(self, watch, force_reprocess=False):
         """
         Perform screenshot comparison using OpenCV subprocess handler.
 

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

@@ -31,6 +31,7 @@ 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
@@ -67,10 +68,6 @@ class Watch(BaseWatch):
         super().__init__(*arg, **kw)
         self['restock'] = Restock(kw['default']['restock']) if kw.get('default') and kw['default'].get('restock') else Restock()
 
-        self['restock_settings'] = kw['default']['restock_settings'] if kw.get('default',{}).get('restock_settings') else {
-            'follow_price_changes': True,
-            'in_stock_processing' : 'in_stock_only'
-        } #@todo update
 
     def clear_watch(self):
         super().clear_watch()

changedetectionio/processors/restock_diff/api.yaml

@@ -0,0 +1,149 @@
+components:
+  schemas:
+    processor_config_restock_diff:
+      type: object
+      description: Configuration for the restock_diff processor (restock and price tracking)
+      properties:
+        in_stock_processing:
+          type: string
+          enum: [in_stock_only, all_changes, 'off']
+          default: in_stock_only
+          description: |
+            When to trigger on stock changes:
+            - `in_stock_only`: Only trigger on Out Of Stock -> In Stock transitions
+            - `all_changes`: Trigger on any availability change
+            - `off`: Disable stock/availability tracking
+        follow_price_changes:
+          type: boolean
+          default: true
+          description: Monitor and track price changes
+        price_change_min:
+          type: [number, 'null']
+          description: Trigger a notification when the price drops below this value
+        price_change_max:
+          type: [number, 'null']
+          description: Trigger a notification when the price rises above this value
+        price_change_threshold_percent:
+          type: [number, 'null']
+          minimum: 0
+          maximum: 100
+          description: Minimum price change percentage since the original price to trigger a notification
+
+paths:
+  /watch:
+    post:
+      x-code-samples:
+        - lang: 'curl'
+          label: 'Restock & price tracking'
+          source: |
+            curl -X POST "http://localhost:5000/api/v1/watch" \
+              -H "x-api-key: YOUR_API_KEY" \
+              -H "Content-Type: application/json" \
+              -d '{
+                "url": "https://example.com/product",
+                "processor": "restock_diff",
+                "processor_config_restock_diff": {
+                  "in_stock_processing": "in_stock_only",
+                  "follow_price_changes": true,
+                  "price_change_threshold_percent": 5
+                }
+              }'
+        - lang: 'Python'
+          label: 'Restock & price tracking'
+          source: |
+            import requests
+
+            headers = {
+                'x-api-key': 'YOUR_API_KEY',
+                'Content-Type': 'application/json'
+            }
+            data = {
+                'url': 'https://example.com/product',
+                'processor': 'restock_diff',
+                'processor_config_restock_diff': {
+                    'in_stock_processing': 'in_stock_only',
+                    'follow_price_changes': True,
+                    'price_change_threshold_percent': 5,
+                }
+            }
+            response = requests.post('http://localhost:5000/api/v1/watch',
+                                     headers=headers, json=data)
+            print(response.json())
+
+  /watch/{uuid}:
+    put:
+      x-code-samples:
+        - lang: 'curl'
+          label: 'Update restock config'
+          source: |
+            curl -X PUT "http://localhost:5000/api/v1/watch/YOUR-UUID" \
+              -H "x-api-key: YOUR_API_KEY" \
+              -H "Content-Type: application/json" \
+              -d '{
+                "processor_config_restock_diff": {
+                  "in_stock_processing": "all_changes",
+                  "follow_price_changes": true,
+                  "price_change_min": 10.00,
+                  "price_change_max": 500.00
+                }
+              }'
+        - lang: 'Python'
+          label: 'Update restock config'
+          source: |
+            import requests
+
+            headers = {
+                'x-api-key': 'YOUR_API_KEY',
+                'Content-Type': 'application/json'
+            }
+            uuid = 'YOUR-UUID'
+            data = {
+                'processor_config_restock_diff': {
+                    'in_stock_processing': 'all_changes',
+                    'follow_price_changes': True,
+                    'price_change_min': 10.00,
+                    'price_change_max': 500.00,
+                }
+            }
+            response = requests.put(f'http://localhost:5000/api/v1/watch/{uuid}',
+                                    headers=headers, json=data)
+            print(response.text)
+
+  /tag/{uuid}:
+    put:
+      x-code-samples:
+        - lang: 'curl'
+          label: 'Set restock config on group/tag'
+          source: |
+            curl -X PUT "http://localhost:5000/api/v1/tag/YOUR-TAG-UUID" \
+              -H "x-api-key: YOUR_API_KEY" \
+              -H "Content-Type: application/json" \
+              -d '{
+                "overrides_watch": true,
+                "processor_config_restock_diff": {
+                  "in_stock_processing": "in_stock_only",
+                  "follow_price_changes": true,
+                  "price_change_threshold_percent": 10
+                }
+              }'
+        - lang: 'Python'
+          label: 'Set restock config on group/tag'
+          source: |
+            import requests
+
+            headers = {
+                'x-api-key': 'YOUR_API_KEY',
+                'Content-Type': 'application/json'
+            }
+            tag_uuid = 'YOUR-TAG-UUID'
+            data = {
+                'overrides_watch': True,
+                'processor_config_restock_diff': {
+                    'in_stock_processing': 'in_stock_only',
+                    'follow_price_changes': True,
+                    'price_change_threshold_percent': 10,
+                }
+            }
+            response = requests.put(f'http://localhost:5000/api/v1/tag/{tag_uuid}',
+                                    headers=headers, json=data)
+            print(response.text)

changedetectionio/processors/restock_diff/forms.py

@@ -31,7 +31,7 @@ class RestockSettingsForm(Form):
     follow_price_changes = BooleanField(_l('Follow price changes'), default=True)
 
 class processor_settings_form(processor_text_json_diff_form):
-    restock_settings = FormField(RestockSettingsForm)
+    processor_config_restock_diff = FormField(RestockSettingsForm)
 
     def extra_tab_content(self):
         return _l('Restock & Price Detection')
@@ -48,34 +48,34 @@ class processor_settings_form(processor_text_json_diff_form):
 
         output += """
         {% from '_helpers.html' import render_field, render_checkbox_field, render_button %}
-        <script>        
+        <script>
             $(document).ready(function () {
-                toggleOpacity('#restock_settings-follow_price_changes', '.price-change-minmax', true);
+                toggleOpacity('#processor_config_restock_diff-follow_price_changes', '.price-change-minmax', true);
             });
         </script>
 
         <fieldset id="restock-fieldset-price-group">
             <div class="pure-control-group">
                 <fieldset class="pure-group inline-radio">
-                    {{ render_field(form.restock_settings.in_stock_processing) }}
+                    {{ render_field(form.processor_config_restock_diff.in_stock_processing) }}
                 </fieldset>
                 <fieldset class="pure-group">
-                    {{ render_checkbox_field(form.restock_settings.follow_price_changes) }}
+                    {{ render_checkbox_field(form.processor_config_restock_diff.follow_price_changes) }}
                     <span class="pure-form-message-inline">Changes in price should trigger a notification</span>
                 </fieldset>
-                <fieldset class="pure-group price-change-minmax">               
+                <fieldset class="pure-group price-change-minmax">
-                    {{ render_field(form.restock_settings.price_change_min, placeholder=watch.get('restock', {}).get('price')) }}
+                    {{ render_field(form.processor_config_restock_diff.price_change_min, placeholder=watch.get('restock', {}).get('price')) }}
                     <span class="pure-form-message-inline">Minimum amount, Trigger a change/notification when the price drops <i>below</i> this value.</span>
                 </fieldset>
                 <fieldset class="pure-group price-change-minmax">
-                    {{ render_field(form.restock_settings.price_change_max, placeholder=watch.get('restock', {}).get('price')) }}
+                    {{ render_field(form.processor_config_restock_diff.price_change_max, placeholder=watch.get('restock', {}).get('price')) }}
                     <span class="pure-form-message-inline">Maximum amount, Trigger a change/notification when the price rises <i>above</i> this value.</span>
                 </fieldset>
                 <fieldset class="pure-group price-change-minmax">
-                    {{ render_field(form.restock_settings.price_change_threshold_percent) }}
+                    {{ render_field(form.processor_config_restock_diff.price_change_threshold_percent) }}
                     <span class="pure-form-message-inline">Price must change more than this % to trigger a change since the first check.</span><br>
                     <span class="pure-form-message-inline">For example, If the product is $1,000 USD originally, <strong>2%</strong> would mean it has to change more than $20 since the first check.</span><br>
-                </fieldset>                
+                </fieldset>
             </div>
         </fieldset>
         """

changedetectionio/processors/restock_diff/processor.py

@@ -2,6 +2,7 @@ from ..base import difference_detection_processor
 from ..exceptions import ProcessorException
 from . import Restock
 from loguru import logger
+from changedetectionio.content_fetchers.exceptions import checksumFromPreviousCheckWasTheSame
 
 import urllib3
 import time
@@ -403,44 +404,65 @@ class perform_site_check(difference_detection_processor):
     screenshot = None
     xpath_data = None
 
-    def run_changedetection(self, watch):
+    def run_changedetection(self, watch, force_reprocess=False):
         import hashlib
 
         if not watch:
             raise Exception("Watch no longer exists.")
 
+        current_raw_document_checksum = self.get_raw_document_checksum()
+        # Skip processing only if BOTH conditions are true:
+        # 1. HTML content unchanged (checksum matches last saved checksum)
+        # 2. Watch configuration was not edited (including trigger_text, filters, etc.)
+        # The was_edited flag handles all watch configuration changes, so we don't need
+        # separate checks for trigger_text or other processing rules.
+        if (not force_reprocess and
+            not watch.was_edited and
+            self.last_raw_content_checksum and
+            self.last_raw_content_checksum == current_raw_document_checksum):
+            raise checksumFromPreviousCheckWasTheSame()
+
         # Unset any existing notification error
         update_obj = {'last_notification_error': False, 'last_error': False, 'restock':  Restock()}
 
         self.screenshot = self.fetcher.screenshot
         self.xpath_data = self.fetcher.xpath_data
 
-        # Track the content type
+        # Track the content type (readonly field, doesn't trigger was_edited)
-        update_obj['content_type'] = self.fetcher.headers.get('Content-Type', '')
+        update_obj['content-type'] = self.fetcher.headers.get('Content-Type', '')  # Use hyphen (matches OpenAPI spec)
         update_obj["last_check_status"] = self.fetcher.get_last_status_code()
 
+        # Save the raw content checksum to file (processor implementation detail, not watch config)
+        self.update_last_raw_content_checksum(current_raw_document_checksum)
+
         # 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
+#useless
-        text = html_to_text(self.fetcher.content)
+#        from ...html_tools import html_to_text
-        logger.debug(f"Length of text after conversion: {len(text)}")
+#        text = html_to_text(self.fetcher.content)
-        if not len(text):
+#        logger.debug(f"Length of text after conversion: {len(text)}")
-            from ...content_fetchers.exceptions import ReplyWithContentButNoText
+#        if not len(text):
-            raise ReplyWithContentButNoText(url=watch.link,
+#            from ...content_fetchers.exceptions import ReplyWithContentButNoText
-                                            status_code=self.fetcher.get_last_status_code(),
+#            raise ReplyWithContentButNoText(url=watch.link,
-                                            screenshot=self.fetcher.screenshot,
+#                                            status_code=self.fetcher.get_last_status_code(),
-                                            html_content=self.fetcher.content,
+#                                            screenshot=self.fetcher.screenshot,
-                                            xpath_data=self.fetcher.xpath_data
+#                                            html_content=self.fetcher.content,
-                                            )
+#                                            xpath_data=self.fetcher.xpath_data
+#                                            )
 
         # Which restock settings to compare against?
-        restock_settings = watch.get('restock_settings', {})
+        # Settings are stored in restock_diff.json (migrated from watch.json by update_30).
+        _extra_config = self.get_extra_watch_config('restock_diff.json')
+        restock_settings = _extra_config.get('restock_diff') or {
+            'follow_price_changes': True,
+            'in_stock_processing': 'in_stock_only',
+        }
 
         # See if any tags have 'activate for individual watches in this tag/group?' enabled and use the first we find
         for tag_uuid in watch.get('tags'):
             tag = self.datastore.data['settings']['application']['tags'].get(tag_uuid, {})
             if tag.get('overrides_watch'):
-                restock_settings = tag.get('restock_settings', {})
+                restock_settings = tag.get('processor_config_restock_diff') or {}
                 logger.info(f"Watch {watch.get('uuid')} - Tag '{tag.get('title')}' selected for restock settings override")
                 break
 

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/__init__.py

@@ -17,7 +17,8 @@ def _task(watch, update_handler):
 
     try:
         # The slow process (we run 2 of these in parallel)
-        changed_detected, update_obj, text_after_filter = update_handler.run_changedetection(watch=watch)
+        # Always force reprocess for preview - we want to show the filtered content regardless of checksums
+        changed_detected, update_obj, text_after_filter = update_handler.run_changedetection(watch=watch, force_reprocess=True)
     except FilterNotFoundInResponse as e:
         text_after_filter = f"Filter not found in HTML: {str(e)}"
     except ReplyWithContentButNoText as e:
@@ -35,7 +36,7 @@ def _task(watch, update_handler):
 
 
 def prepare_filter_prevew(datastore, watch_uuid, form_data):
-    '''Used by @app.route("/edit/<string:uuid>/preview-rendered", methods=['POST'])'''
+    '''Used by @app.route("/edit/<uuid_str:uuid>/preview-rendered", methods=['POST'])'''
     from changedetectionio import forms, html_tools
     from changedetectionio.model.Watch import model as watch_model
     from concurrent.futures import ThreadPoolExecutor

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 = watch.fetcher_supports_screenshots
-
-    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
 
     password_enabled_and_share_is_off = False
     if datastore.data['settings']['application'].get('password') or os.getenv("SALTED_PASS", False):

changedetectionio/processors/text_json_diff/processor.py

@@ -7,6 +7,7 @@ import re
 import urllib3
 
 from changedetectionio.conditions import execute_ruleset_against_all_plugins
+from changedetectionio.content_fetchers.exceptions import checksumFromPreviousCheckWasTheSame
 from ..base import difference_detection_processor
 from changedetectionio.html_tools import PERL_STYLE_REGEX, cdata_in_document_to_text, TRANSLATE_WHITESPACE_TABLE
 from changedetectionio import html_tools, content_fetchers
@@ -346,6 +347,7 @@ class ContentProcessor:
     def extract_text_from_html(self, html_content, stream_content_type):
         """Convert HTML to plain text."""
         do_anchor = self.datastore.data["settings"]["application"].get("render_anchor_tag_content", False)
+
         return html_tools.html_to_text(
             html_content=html_content,
             render_anchor_tag_content=do_anchor,
@@ -368,12 +370,24 @@ class ChecksumCalculator:
 # (set_proxy_from_list)
 class perform_site_check(difference_detection_processor):
 
-    def run_changedetection(self, watch):
+    def run_changedetection(self, watch, force_reprocess=False):
         changed_detected = False
 
         if not watch:
             raise Exception("Watch no longer exists.")
 
+        current_raw_document_checksum = self.get_raw_document_checksum()
+        # Skip processing only if BOTH conditions are true:
+        # 1. HTML content unchanged (checksum matches last saved checksum)
+        # 2. Watch configuration was not edited (including trigger_text, filters, etc.)
+        # The was_edited flag handles all watch configuration changes, so we don't need
+        # separate checks for trigger_text or other processing rules.
+        if (not force_reprocess and
+            not watch.was_edited and
+            self.last_raw_content_checksum and
+            self.last_raw_content_checksum == current_raw_document_checksum):
+            raise checksumFromPreviousCheckWasTheSame()
+
         # Initialize components
         filter_config = FilterConfig(watch, self.datastore)
         content_processor = ContentProcessor(self.fetcher, watch, filter_config, self.datastore)
@@ -391,9 +405,11 @@ class perform_site_check(difference_detection_processor):
         self.screenshot = self.fetcher.screenshot
         self.xpath_data = self.fetcher.xpath_data
 
-        # Track the content type and checksum before filters
+        # Track the content type (readonly field, doesn't trigger was_edited)
-        update_obj['content_type'] = ctype_header
+        update_obj['content-type'] = ctype_header  # Use hyphen (matches OpenAPI spec and watch_base default)
-        update_obj['previous_md5_before_filters'] = hashlib.md5(self.fetcher.content.encode('utf-8')).hexdigest()
+
+        # Save the raw content checksum to file (processor implementation detail, not watch config)
+        self.update_last_raw_content_checksum(current_raw_document_checksum)
 
         # === CONTENT PREPROCESSING ===
         # Avoid creating unnecessary intermediate string copies by reassigning only when needed

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
 
@@ -345,4 +368,4 @@ def init_socketio(app, datastore):
 
     logger.info("Socket.IO initialized and attached to main Flask app")
     logger.info(f"Socket.IO: Registered event handlers: {socketio.handlers if hasattr(socketio, 'handlers') else 'No handlers found'}")
-    return socketio
+    return socketio

changedetectionio/run_basic_tests.sh

@@ -44,12 +44,12 @@ data_sanity_test () {
   cd ..
   TMPDIR=$(mktemp -d)
   PORT_N=$((5000 + RANDOM % (6501 - 5000)))
-  ./changedetection.py -p $PORT_N -d $TMPDIR -u "https://localhost?test-url-is-sanity=1" &
+  ALLOW_IANA_RESTRICTED_ADDRESSES=true ./changedetection.py -p $PORT_N -d $TMPDIR -u "https://localhost?test-url-is-sanity=1" &
   PID=$!
   sleep 5
   kill $PID
   sleep 2
-  ./changedetection.py -p $PORT_N -d $TMPDIR &
+  ALLOW_IANA_RESTRICTED_ADDRESSES=true ./changedetection.py -p $PORT_N -d $TMPDIR &
   PID=$!
   sleep 5
   # On a restart the URL should still be there

changedetectionio/static/js/browser-steps.js

@@ -17,8 +17,6 @@ $(document).ready(function () {
         set_scale();
     });
     // Should always be disabled
-    $('#browser_steps-0-operation option[value="Goto site"]').prop("selected", "selected");
-    $('#browser_steps-0-operation').attr('disabled', 'disabled');
 
     $('#browsersteps-click-start').click(function () {
         $("#browsersteps-click-start").fadeOut();
@@ -45,12 +43,6 @@ $(document).ready(function () {
         browsersteps_session_id = false;
         apply_buttons_disabled = false;
         ctx.clearRect(0, 0, c.width, c.height);
-        set_first_gotosite_disabled();
-    }
-
-    function set_first_gotosite_disabled() {
-        $('#browser_steps >li:first-child select').val('Goto site').attr('disabled', 'disabled');
-        $('#browser_steps >li:first-child').css('opacity', '0.5');
     }
 
     // Show seconds remaining until the browser interface needs to restart the session
@@ -243,14 +235,54 @@ $(document).ready(function () {
         ctx.fill();
     }
 
+    // Reusable AJAX function for browser step operations
+    function executeBrowserStep(url, data = {}) {
+        $('#browser-steps-ui .loader .spinner').fadeIn();
+        apply_buttons_disabled = true;
+        $('ul#browser_steps li .control .apply').css('opacity', 0.5);
+        $("#browsersteps-img").css('opacity', 0.65);
+
+        return $.ajax({
+            method: "POST",
+            url: url,
+            data: data,
+            statusCode: {
+                400: function () {
+                    alert("There was a problem processing the request, please reload the page.");
+                    $("#loading-status-text").hide();
+                    $('#browser-steps-ui .loader .spinner').fadeOut();
+                },
+                401: function (data) {
+                    alert(data.responseText);
+                    $("#loading-status-text").hide();
+                    $('#browser-steps-ui .loader .spinner').fadeOut();
+                }
+            }
+        }).done(function (data) {
+            xpath_data = data.xpath_data;
+            $('#browsersteps-img').attr('src', data.screenshot);
+            $('#browser-steps-ui .loader .spinner').fadeOut();
+            apply_buttons_disabled = false;
+            $("#browsersteps-img").css('opacity', 1);
+            $('ul#browser_steps li .control .apply').css('opacity', 1);
+            $("#loading-status-text").hide();
+        }).fail(function (data) {
+            console.log(data);
+            if (data.responseText && data.responseText.includes("Browser session expired")) {
+                disable_browsersteps_ui();
+            }
+            apply_buttons_disabled = false;
+            $("#loading-status-text").hide();
+            $('ul#browser_steps li .control .apply').css('opacity', 1);
+            $("#browsersteps-img").css('opacity', 1);
+        });
+    }
+
     function start() {
         console.log("Starting browser-steps UI");
         browsersteps_session_id = false;
-        // @todo This setting of the first one should be done at the datalayer but wtforms doesnt wanna play nice
-        $('#browser_steps >li:first-child').removeClass('empty');
-        set_first_gotosite_disabled();
         $('#browser-steps-ui .loader .spinner').show();
-        $('.clear,.remove', $('#browser_steps >li:first-child')).hide();
+        // Request a new session
         $.ajax({
             type: "GET",
             url: browser_steps_start_url,
@@ -267,11 +299,12 @@ $(document).ready(function () {
         }).done(function (data) {
             $("#loading-status-text").fadeIn();
             browsersteps_session_id = data.browsersteps_session_id;
-            // This should trigger 'Goto site'
-            console.log("Got startup response, requesting Goto-Site (first) step fake click");
-            $('#browser_steps >li:first-child .apply').click();
             browser_interface_seconds_remaining = 500;
-            set_first_gotosite_disabled();
+            // Request goto_site operation
+            executeBrowserStep(
+                browser_steps_sync_url + "&browsersteps_session_id=" + browsersteps_session_id + "&goto_website_url_first_step=true"
+            );
+
         }).fail(function (data) {
             console.log(data);
             alert('There was an error communicating with the server.');
@@ -280,7 +313,6 @@ $(document).ready(function () {
     }
 
     function disable_browsersteps_ui() {
-        set_first_gotosite_disabled();
         $("#browser-steps-ui").css('opacity', '0.3');
         $('#browsersteps-selector-canvas').off("mousemove mousedown click");
     }
@@ -328,16 +360,13 @@ $(document).ready(function () {
     // Add the extra buttons to the steps
     $('ul#browser_steps li').each(function (i) {
             var s = '<div class="control">' + '<a data-step-index=' + i + ' class="pure-button button-secondary button-green button-xsmall apply" >Apply</a>&nbsp;';
-            if (i > 0) {
+            s += `<a data-step-index="${i}" class="pure-button button-secondary button-xsmall clear" >Clear</a>&nbsp;` +
-                // The first step never gets these (Goto-site)
+                `<a data-step-index="${i}" class="pure-button button-secondary button-red button-xsmall remove" >Remove</a>`;
-                s += `<a data-step-index="${i}" class="pure-button button-secondary button-xsmall clear" >Clear</a>&nbsp;` +
-                    `<a data-step-index="${i}" class="pure-button button-secondary button-red button-xsmall remove" >Remove</a>`;
 
-                // if a screenshot is available
+            // if a screenshot is available
-                if (browser_steps_available_screenshots.includes(i.toString())) {
+            if (browser_steps_available_screenshots.includes(i.toString())) {
-                    var d = (browser_steps_last_error_step === i+1) ? 'before' : 'after';
+                var d = (browser_steps_last_error_step === i+1) ? 'before' : 'after';
-                    s += `&nbsp;<a data-step-index="${i}" class="pure-button button-secondary button-xsmall show-screenshot" title="Show screenshot from last run" data-type="${d}">Pic</a>&nbsp;`;
+                s += `&nbsp;<a data-step-index="${i}" class="pure-button button-secondary button-xsmall show-screenshot" title="Show screenshot from last run" data-type="${d}">Pic</a>&nbsp;`;
-                }
             }
             s += '</div>';
             $(this).append(s)
@@ -376,80 +405,35 @@ $(document).ready(function () {
     });
 
     $('ul#browser_steps li .control .apply').click(function (event) {
-        // sequential requests @todo refactor
         if (apply_buttons_disabled) {
             return;
         }
 
         var current_data = $(event.currentTarget).closest('li');
-        $('#browser-steps-ui .loader .spinner').fadeIn();
-        apply_buttons_disabled = true;
-        $('ul#browser_steps li .control .apply').css('opacity', 0.5);
-        $("#browsersteps-img").css('opacity', 0.65);
-
-        var is_last_step = 0;
         var step_n = $(event.currentTarget).data('step-index');
 
-        // On the last step, we should also be getting data ready for the visual selector
+        // Determine if this is the last configured step
+        var is_last_step = 0;
         $('ul#browser_steps li select').each(function (i) {
             if ($(this).val() !== 'Choose one') {
                 is_last_step += 1;
             }
         });
-
+        is_last_step = (is_last_step == (step_n + 1));
-        if (is_last_step == (step_n + 1)) {
-            is_last_step = true;
-        } else {
-            is_last_step = false;
-        }
 
         console.log("Requesting step via POST " + $("select[id$='operation']", current_data).first().val());
-        // POST the currently clicked step form widget back and await response, redraw
+
-        $.ajax({
+        // Execute the browser step
-            method: "POST",
+        executeBrowserStep(
-            url: browser_steps_sync_url + "&browsersteps_session_id=" + browsersteps_session_id,
+            browser_steps_sync_url + "&browsersteps_session_id=" + browsersteps_session_id,
-            data: {
+            {
                 'operation': $("select[id$='operation']", current_data).first().val(),
                 'selector': $("input[id$='selector']", current_data).first().val(),
                 'optional_value': $("input[id$='optional_value']", current_data).first().val(),
                 'step_n': step_n,
                 'is_last_step': is_last_step
-            },
-            statusCode: {
-                400: function () {
-                    // More than likely the CSRF token was lost when the server restarted
-                    alert("There was a problem processing the request, please reload the page.");
-                    $("#loading-status-text").hide();
-                    $('#browser-steps-ui .loader .spinner').fadeOut();
-                },
-                401: function (data) {
-                    // More than likely the CSRF token was lost when the server restarted
-                    alert(data.responseText);
-                    $("#loading-status-text").hide();
-                    $('#browser-steps-ui .loader .spinner').fadeOut();
-                }
             }
-        }).done(function (data) {
+        );
-            // it should return the new state (selectors available and screenshot)
-            xpath_data = data.xpath_data;
-            $('#browsersteps-img').attr('src', data.screenshot);
-            $('#browser-steps-ui .loader .spinner').fadeOut();
-            apply_buttons_disabled = false;
-            $("#browsersteps-img").css('opacity', 1);
-            $('ul#browser_steps li .control .apply').css('opacity', 1);
-            $("#loading-status-text").hide();
-            set_first_gotosite_disabled();
-        }).fail(function (data) {
-            console.log(data);
-            if (data.responseText.includes("Browser session expired")) {
-                disable_browsersteps_ui();
-            }
-            apply_buttons_disabled = false;
-            $("#loading-status-text").hide();
-            $('ul#browser_steps li .control .apply').css('opacity', 1);
-            $("#browsersteps-img").css('opacity', 1);
-        });
-
     });
 
     $('ul#browser_steps li .control .show-screenshot').click(function (element) {

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/search-modal.js

@@ -102,7 +102,9 @@
         }
 
         // Navigate to search results (always redirect to watchlist home)
-        window.location.href = '/?' + params.toString();
+        // Use base_path if available (for sub-path deployments like /enlighten-richerx)
+        const basePath = typeof base_path !== 'undefined' ? base_path : '';
+        window.location.href = basePath + '/?' + params.toString();
       });
     }
   });

changedetectionio/static/styles/diff.css

@@ -1 +1 @@
-#diff-form{background:rgba(0,0,0,.05);padding:1em;border-radius:10px;margin-bottom:1em;color:#fff;font-size:.9rem;text-align:center}#diff-form label.from-to-label{width:4rem;text-decoration:none;padding:.5rem}#diff-form label.from-to-label#change-from{color:#b30000;background:#fadad7}#diff-form label.from-to-label#change-to{background:#eaf2c2;color:#406619}#diff-form #diff-style>span{display:inline-block;padding:.3em}#diff-form #diff-style>span label{font-weight:normal}#diff-form *{vertical-align:middle}body.difference-page section.content{padding-top:40px}#diff-ui{background:var(--color-background);padding:1rem;border-radius:5px}@media(min-width: 767px){#diff-ui{min-width:50%}}#diff-ui #text{font-size:11px}#diff-ui pre{white-space:break-spaces}#diff-ui h1{display:inline;font-size:100%}#diff-ui #result{white-space:pre-wrap;word-break:break-word;overflow-wrap:break-word}#diff-ui .source{position:absolute;right:1%;top:.2em}@-moz-document url-prefix(){#diff-ui body{height:99%}}#diff-ui td#diff-col div{text-align:justify;white-space:pre-wrap}#diff-ui .ignored{background-color:#ccc;opacity:.7}#diff-ui .triggered{background-color:#1b98f8}#diff-ui .ignored.triggered{background-color:red}#diff-ui .tab-pane-inner#screenshot{text-align:center}#diff-ui .tab-pane-inner#screenshot img{max-width:99%}#diff-ui .pure-form button.reset-margin{margin:0px}#diff-ui .diff-fieldset{display:flex;align-items:center;gap:4px;flex-wrap:wrap}#diff-ui ul#highlightSnippetActions{list-style-type:none;display:flex;align-items:center;justify-content:center;gap:1.5rem;flex-wrap:wrap;padding:0;margin:0}#diff-ui ul#highlightSnippetActions li{display:flex;flex-direction:column;align-items:center;text-align:center;padding:.5rem;gap:.3rem}#diff-ui ul#highlightSnippetActions li button,#diff-ui ul#highlightSnippetActions li a{white-space:nowrap}#diff-ui ul#highlightSnippetActions span{font-size:.8rem;color:var(--color-text-input-description)}#diff-ui #cell-diff-jump-visualiser{display:flex;flex-direction:row;gap:1px;background:var(--color-background);border-radius:3px;overflow-x:hidden;position:sticky;top:0;z-index:10;padding-top:1rem;padding-bottom:1rem;justify-content:center}#diff-ui #cell-diff-jump-visualiser>div{flex:1;min-width:1px;max-width:10px;height:10px;background:var(--color-background-button-cancel);opacity:.3;border-radius:1px;transition:opacity .2s;position:relative}#diff-ui #cell-diff-jump-visualiser>div.deletion{background:#b30000;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.insertion{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.note{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.mixed{background:linear-gradient(to right, #b30000 50%, #406619 50%);opacity:1}#diff-ui #cell-diff-jump-visualiser>div.current-position::after{content:"";position:absolute;bottom:-6px;left:50%;transform:translateX(-50%);width:0;height:0;border-left:4px solid rgba(0,0,0,0);border-right:4px solid rgba(0,0,0,0);border-bottom:4px solid var(--color-text)}#diff-ui #cell-diff-jump-visualiser>div:hover{opacity:.8;cursor:pointer}#text-diff-heading-area .snapshot-age{padding:4px;margin:.5rem 0;background-color:var(--color-background-snapshot-age);border-radius:3px;font-weight:bold;margin-bottom:4px}#text-diff-heading-area .snapshot-age.error{background-color:var(--color-error-background-snapshot-age);color:var(--color-error-text-snapshot-age)}#text-diff-heading-area .snapshot-age>*{padding-right:1rem}
+#diff-form{background:rgba(0,0,0,.05);padding:1em;border-radius:10px;margin-bottom:1em;color:#fff;font-size:.9rem;text-align:center}#diff-form label.from-to-label{width:4rem;text-decoration:none;padding:.5rem}#diff-form label.from-to-label#change-from{color:#b30000;background:#fadad7}#diff-form label.from-to-label#change-to{background:#eaf2c2;color:#406619}#diff-form #diff-style>span{display:inline-block;padding:.3em}#diff-form #diff-style>span label{font-weight:normal}#diff-form *{vertical-align:middle}body.difference-page section.content{padding-top:40px}#diff-ui{background:var(--color-background);padding:1rem;border-radius:5px}@media(min-width: 767px){#diff-ui{min-width:50%}}#diff-ui #text{font-size:11px}#diff-ui pre{white-space:break-spaces;overflow-wrap:anywhere}#diff-ui h1{display:inline;font-size:100%}#diff-ui #result{white-space:pre-wrap;word-break:break-word;overflow-wrap:break-word}#diff-ui .source{position:absolute;right:1%;top:.2em}@-moz-document url-prefix(){#diff-ui body{height:99%}}#diff-ui td#diff-col div{text-align:justify;white-space:pre-wrap}#diff-ui .ignored{background-color:#ccc;opacity:.7}#diff-ui .triggered{background-color:#1b98f8}#diff-ui .ignored.triggered{background-color:red}#diff-ui .tab-pane-inner#screenshot{text-align:center}#diff-ui .tab-pane-inner#screenshot img{max-width:99%}#diff-ui .pure-form button.reset-margin{margin:0px}#diff-ui .diff-fieldset{display:flex;align-items:center;gap:4px;flex-wrap:wrap}#diff-ui ul#highlightSnippetActions{list-style-type:none;display:flex;align-items:center;justify-content:center;gap:1.5rem;flex-wrap:wrap;padding:0;margin:0}#diff-ui ul#highlightSnippetActions li{display:flex;flex-direction:column;align-items:center;text-align:center;padding:.5rem;gap:.3rem}#diff-ui ul#highlightSnippetActions li button,#diff-ui ul#highlightSnippetActions li a{white-space:nowrap}#diff-ui ul#highlightSnippetActions span{font-size:.8rem;color:var(--color-text-input-description)}#diff-ui #cell-diff-jump-visualiser{display:flex;flex-direction:row;gap:1px;background:var(--color-background);border-radius:3px;overflow-x:hidden;position:sticky;top:0;z-index:10;padding-top:1rem;padding-bottom:1rem;justify-content:center}#diff-ui #cell-diff-jump-visualiser>div{flex:1;min-width:1px;max-width:10px;height:10px;background:var(--color-background-button-cancel);opacity:.3;border-radius:1px;transition:opacity .2s;position:relative}#diff-ui #cell-diff-jump-visualiser>div.deletion{background:#b30000;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.insertion{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.note{background:#406619;opacity:1}#diff-ui #cell-diff-jump-visualiser>div.mixed{background:linear-gradient(to right, #b30000 50%, #406619 50%);opacity:1}#diff-ui #cell-diff-jump-visualiser>div.current-position::after{content:"";position:absolute;bottom:-6px;left:50%;transform:translateX(-50%);width:0;height:0;border-left:4px solid rgba(0,0,0,0);border-right:4px solid rgba(0,0,0,0);border-bottom:4px solid var(--color-text)}#diff-ui #cell-diff-jump-visualiser>div:hover{opacity:.8;cursor:pointer}#text-diff-heading-area .snapshot-age{padding:4px;margin:.5rem 0;background-color:var(--color-background-snapshot-age);border-radius:3px;font-weight:bold;margin-bottom:4px}#text-diff-heading-area .snapshot-age.error{background-color:var(--color-error-background-snapshot-age);color:var(--color-error-text-snapshot-age)}#text-diff-heading-area .snapshot-age>*{padding-right:1rem}

changedetectionio/static/styles/scss/diff.scss

@@ -62,6 +62,7 @@ body.difference-page {
 
   pre {
     white-space: break-spaces;
+    overflow-wrap: anywhere;
   }
 
 

changedetectionio/store/__init__.py

@@ -22,6 +22,8 @@ import uuid as uuid_builder
 from loguru import logger
 from blinker import signal
 
+from ..model.Tags import TagsDict
+
 # Try to import orjson for faster JSON serialization
 try:
     import orjson
@@ -121,6 +123,11 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
             if 'application' in settings_data['settings']:
                 self.__data['settings']['application'].update(settings_data['settings']['application'])
 
+                # Use our Tags dict with cleanup helpers etc
+                # @todo Same for Watches
+                existing_tags = settings_data.get('settings', {}).get('application', {}).get('tags') or {}
+                self.__data['settings']['application']['tags'] = TagsDict(existing_tags, datastore_path=self.datastore_path)
+
         # More or less for the old format which had this data in the one url-watches.json
         # cant hurt to leave it here,
         if 'watching' in settings_data:
@@ -196,7 +203,7 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
         self.datastore_path = datastore_path
 
         # Initialize data structure
-        self.__data = App.model()
+        self.__data = App.model(datastore_path=datastore_path)
         self.json_store_path = os.path.join(self.datastore_path, "changedetection.json")
 
         # Base definition for all watchers (deepcopy part of #569)
@@ -235,6 +242,8 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
             # No datastore yet - check if this is a fresh install or legacy migration
             self.init_fresh_install(include_default_watches=include_default_watches,
                                     version_tag=version_tag)
+            # Maybe they copied a bunch of watch subdirs across too
+            self._load_state()
 
     def init_fresh_install(self, include_default_watches, version_tag):
       # Generate app_guid FIRST (required for all operations)
@@ -353,6 +362,9 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
         # Deep copy settings to avoid modifying the original
         settings_copy = copy.deepcopy(self.__data['settings'])
 
+        # Is saved as {uuid}/tag.json
+        settings_copy['application']['tags'] = {}
+
         return {
             'note': 'Settings file - watches are in {uuid}/watch.json, tags are in {uuid}/tag.json',
             'app_guid': self.__data.get('app_guid'),
@@ -456,6 +468,63 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
         self.__data['settings']['application']['password'] = False
         self.commit()
 
+    def clear_all_last_checksums(self):
+        """
+        Delete all last-checksum.txt files to force reprocessing of all watches.
+
+        This should be called when global settings change, since watches inherit
+        configuration and need to reprocess even if their individual watch dict
+        hasn't been modified.
+
+        Note: We delete the checksum file rather than setting was_edited=True because:
+        - was_edited is not persisted across restarts
+        - File deletion ensures reprocessing works across app restarts
+        """
+        deleted_count = 0
+        for uuid in self.__data['watching'].keys():
+            watch = self.__data['watching'][uuid]
+            if watch.data_dir:
+                checksum_file = os.path.join(watch.data_dir, 'last-checksum.txt')
+                if os.path.isfile(checksum_file):
+                    try:
+                        os.remove(checksum_file)
+                        deleted_count += 1
+                        logger.debug(f"Cleared checksum for watch {uuid}")
+                    except OSError as e:
+                        logger.warning(f"Failed to delete checksum file for {uuid}: {e}")
+
+        logger.info(f"Cleared {deleted_count} checksum files to force reprocessing")
+        return deleted_count
+
+    def clear_checksums_for_tag(self, tag_uuid):
+        """
+        Delete last-checksum.txt files for all watches using a specific tag.
+
+        This should be called when a tag configuration is edited, since watches
+        inherit tag settings and need to reprocess.
+
+        Args:
+            tag_uuid: UUID of the tag that was modified
+
+        Returns:
+            int: Number of checksum files deleted
+        """
+        deleted_count = 0
+        for uuid, watch in self.__data['watching'].items():
+            if watch.get('tags') and tag_uuid in watch['tags']:
+                if watch.data_dir:
+                    checksum_file = os.path.join(watch.data_dir, 'last-checksum.txt')
+                    if os.path.isfile(checksum_file):
+                        try:
+                            os.remove(checksum_file)
+                            deleted_count += 1
+                            logger.debug(f"Cleared checksum for watch {uuid} (tag {tag_uuid})")
+                        except OSError as e:
+                            logger.warning(f"Failed to delete checksum file for {uuid}: {e}")
+
+        logger.info(f"Cleared {deleted_count} checksum files for tag {tag_uuid}")
+        return deleted_count
+
     def commit(self):
         """
         Save settings immediately to disk using atomic write.
@@ -659,8 +728,11 @@ class ChangeDetectionStore(DatastoreUpdatesMixin, FileSavingDataStore):
                 return False
 
         if not is_safe_valid_url(url):
-            flash(gettext('Watch protocol is not permitted or invalid URL format'), 'error')
+            from flask import has_request_context
-
+            if has_request_context():
+                flash(gettext('Watch protocol is not permitted or invalid URL format'), 'error')
+            else:
+                logger.error(f"add_watch: URL '{url}' is not permitted or invalid, skipping.")
             return None
 
         # Check PAGE_WATCH_LIMIT if set

changedetectionio/store/updates.py

@@ -669,7 +669,9 @@ class DatastoreUpdatesMixin:
     def update_26(self):
         self.migrate_legacy_db_format()
 
-    def update_28(self):
+    # Re-run tag to JSON migration
+    def update_29(self):
+
         """
         Migrate tags to individual tag.json files.
 
@@ -682,8 +684,6 @@ class DatastoreUpdatesMixin:
         - Enables independent tag versioning/backup
         - Maintains backwards compatibility (tags stay in settings too)
         """
-        # Force save as tag.json (not watch.json) even if object is corrupted
-
         logger.critical("=" * 80)
         logger.critical("Running migration: Individual tag persistence (update_28)")
         logger.critical("Creating individual tag.json files")
@@ -702,6 +702,9 @@ class DatastoreUpdatesMixin:
         failed_count = 0
 
         for uuid, tag_data in tags.items():
+            if os.path.isfile(os.path.join(self.datastore_path, uuid, "tag.json")):
+                logger.debug(f"Tag {uuid} tag.json exists, skipping")
+                continue
             try:
                 tag_data.commit()
                 saved_count += 1
@@ -723,3 +726,52 @@ class DatastoreUpdatesMixin:
         logger.info("Future tag edits will update both locations (dual storage)")
         logger.critical("=" * 80)
 
+        # write it to disk, it will be saved without ['tags'] in the JSON db because we find it from disk glob
+        # (left this out by accident in previous update, added tags={} in the changedetection.json save_to_disk)
+        self._save_settings()
+
+    def update_30(self):
+        """Migrate restock_settings out of watch.json into restock_diff.json processor config file.
+
+        Previously, restock_diff processor settings (in_stock_processing, follow_price_changes, etc.)
+        were stored directly in the watch dict (watch.json). They now belong in a separate per-watch
+        processor config file (restock_diff.json) consistent with the processor_config_* API system.
+
+        For tags: restock_settings key is renamed to processor_config_restock_diff in the tag dict,
+        matching what the API writes when updating a tag.
+
+        Safe to re-run: skips watches that already have a restock_diff.json, skips tags that already
+        have processor_config_restock_diff set.
+        """
+        import json
+
+        # --- Watches ---
+        for uuid, watch in self.data['watching'].items():
+            if watch.get('processor') != 'restock_diff':
+                continue
+            restock_settings = watch.get('restock_settings')
+            if not restock_settings:
+                continue
+
+            data_dir = watch.data_dir
+            if data_dir:
+                watch.ensure_data_dir_exists()
+                filepath = os.path.join(data_dir, 'restock_diff.json')
+                if not os.path.isfile(filepath):
+                    with open(filepath, 'w', encoding='utf-8') as f:
+                        json.dump({'restock_diff': restock_settings}, f, indent=2)
+                    logger.info(f"update_30: migrated restock_settings → {filepath}")
+
+            del self.data['watching'][uuid]['restock_settings']
+            watch.commit()
+
+        # --- Tags ---
+        for tag_uuid, tag in self.data['settings']['application']['tags'].items():
+            restock_settings = tag.get('restock_settings')
+            if not restock_settings or tag.get('processor_config_restock_diff'):
+                continue
+            tag['processor_config_restock_diff'] = restock_settings
+            del tag['restock_settings']
+            tag.commit()
+            logger.info(f"update_30: migrated tag {tag_uuid} restock_settings → processor_config_restock_diff")
+

changedetectionio/templates/_common_fields.html

@@ -44,13 +44,26 @@
                 <td><code>{{ '{{preview_url}}' }}</code></td>
                 <td>{{ _('The URL of the preview page generated by changedetection.io.') }}</td>
             </tr>
+            <tr>
+                <td><code>{{ '{{change_datetime}}' }}</code></td>
+                <td>{{ _('Date/time of the change, accepts format=, change_datetime(format=\'%A\')\', default is \'%Y-%m-%d %H:%M:%S %Z\'') }}</td>
+            </tr>
+            <tr>
+                <td><code>{{ '{{diff_url}}' }}</code></td>
+                <td>{{ _('The URL of the diff output for the watch.') }}</td>
+            </tr>
             <tr>
                 <td><code>{{ '{{diff_url}}' }}</code></td>
                 <td>{{ _('The URL of the diff output for the watch.') }}</td>
             </tr>
             <tr>
                 <td><code>{{ '{{diff}}' }}</code></td>
-                <td>{{ _('The diff output - only changes, additions, and removals') }}</td>
+                <td>{{ _('The diff output - only changes, additions, and removals') }}<br>
+                    <small>
+                        {{ _('All diff variants accept') }} <code>lines=</code>, <code>context=</code>, <code>word_diff=</code>, <code>ignore_junk=</code> {{ _('args, e.g.') }}
+                        <code>{{ '{{diff(lines=10)}}' }}</code>, <code>{{ '{{diff_added(lines=5, context=2)}}' }}</code>
+                    </small>
+                </td>
             </tr>
             <tr>
                 <td><code>{{ '{{diff_clean}}' }}</code></td>

changedetectionio/templates/edit/text-options.html

@@ -10,6 +10,7 @@
                         <li>{{ _('Trigger text is processed from the result-text that comes out of any CSS/JSON Filters for this monitor') }}</li>
                         <li>{{ _('Each line is processed separately (think of each line as "OR")') }}</li>
                         <li>{{ _('Note: Wrap in forward slash / to use regex example:') }} <code>/foo\d/</code></li>
+                        <li>{{ _('You can also use')}} <a href="#conditions">{{ _('conditions')}}</a> - {{ _('"Page text" - with Contains, Starts With, Not Contains and many more' ) }} <code>/foo\d/</code></li>
                     </ul>
                         </span>
                     </div>

changedetectionio/tests/conftest.py

@@ -2,6 +2,7 @@
 import psutil
 import time
 from threading import Thread
+import multiprocessing
 
 import pytest
 import arrow
@@ -13,6 +14,10 @@ import sys
 # When test server is slow/unresponsive, workers fail fast instead of holding UUIDs for 45s
 # This prevents exponential priority growth from repeated deferrals (priority × 10 each defer)
 os.environ['DEFAULT_SETTINGS_REQUESTS_TIMEOUT'] = '5'
+# Test server runs on localhost (127.0.0.1) which is a private IP.
+# Allow it globally so all existing tests keep working; test_ssrf_protection
+# uses monkeypatch to temporarily override this for its own assertions.
+os.environ['ALLOW_IANA_RESTRICTED_ADDRESSES'] = 'true'
 
 from changedetectionio.flask_app import init_app_secret, changedetection_app
 from changedetectionio.tests.util import live_server_setup, new_live_server_setup
@@ -187,6 +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.
 
@@ -331,6 +364,7 @@ def prepare_test_function(live_server, datastore_path):
     # Cleanup: Clear watches and queue after test
     try:
         from changedetectionio.flask_app import update_q
+        from pathlib import Path
 
         # Clear the queue to prevent leakage to next test
         while not update_q.empty():
@@ -340,6 +374,18 @@ def prepare_test_function(live_server, datastore_path):
                 break
 
         datastore.data['watching'] = {}
+
+        # Delete any old watch metadata JSON files
+        base_path = Path(datastore.datastore_path).resolve()
+        max_depth = 2
+
+        for file in base_path.rglob("*.json"):
+            # Calculate depth relative to base path
+            depth = len(file.relative_to(base_path).parts) - 1
+
+            if depth <= max_depth and file.is_file():
+                file.unlink()
+
     except Exception as e:
         logger.warning(f"Error during datastore cleanup: {e}")
 

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(
@@ -807,6 +815,88 @@ def test_api_import_large_background(client, live_server, measure_memory_usage,
     print(f"\n✓ Successfully created {num_urls} watches in background (took {elapsed}s)")
 
 
+def test_api_restock_processor_config(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that processor_config_restock_diff is accepted by the API for watches using
+    restock_diff processor, that its schema is validated (enum values, types), and that
+    genuinely unknown fields are rejected with an error that originates from the
+    OpenAPI spec validation layer.
+    """
+    api_key = live_server.app.config['DATASTORE'].data['settings']['application'].get('api_access_token')
+    test_url = url_for('test_endpoint', _external=True)
+
+    # Create a watch in restock_diff mode WITH processor_config in the POST body (matches the API docs example)
+    res = client.post(
+        url_for("createwatch"),
+        data=json.dumps({
+            "url": test_url,
+            "processor": "restock_diff",
+            "title": "Restock test",
+            "processor_config_restock_diff": {
+                "in_stock_processing": "in_stock_only",
+                "follow_price_changes": True,
+                "price_change_min": 8888888.0,
+            }
+        }),
+        headers={'content-type': 'application/json', 'x-api-key': api_key},
+        follow_redirects=True
+    )
+    assert res.status_code == 201
+    watch_uuid = res.json.get('uuid')
+    assert is_valid_uuid(watch_uuid)
+
+    # Verify the value set on POST is reflected in the UI edit page (not just via PUT)
+    res = client.get(url_for("ui.ui_edit.edit_page", uuid=watch_uuid))
+    assert res.status_code == 200
+    assert b'8888888' in res.data, "price_change_min set via POST should appear in the UI edit form"
+
+    # Valid processor_config_restock_diff update via PUT should also be accepted
+    res = client.put(
+        url_for("watch", uuid=watch_uuid),
+        headers={'x-api-key': api_key, 'content-type': 'application/json'},
+        data=json.dumps({
+            "processor_config_restock_diff": {
+                "in_stock_processing": "all_changes",
+                "follow_price_changes": False,
+                "price_change_min": 8888888.0,
+                "price_change_max": 9999999.0,
+            }
+        }),
+    )
+    assert res.status_code == 200, f"Valid processor_config_restock_diff should be accepted, got: {res.data}"
+
+    # Verify the updated value is still reflected in the UI edit page
+    res = client.get(url_for("ui.ui_edit.edit_page", uuid=watch_uuid))
+    assert res.status_code == 200
+    assert b'8888888' in res.data, "price_change_min set via PUT should appear in the UI edit form"
+
+    # An invalid enum value inside processor_config_restock_diff should be rejected by the spec
+    res = client.put(
+        url_for("watch", uuid=watch_uuid),
+        headers={'x-api-key': api_key, 'content-type': 'application/json'},
+        data=json.dumps({
+            "processor_config_restock_diff": {
+                "in_stock_processing": "not_a_valid_enum_value"
+            }
+        }),
+    )
+    assert res.status_code == 400, "Invalid enum value in processor config should be rejected"
+    assert b'Validation failed' in res.data, "Rejection should come from OpenAPI spec validation layer"
+
+    # A completely unknown field should be rejected (either by OpenAPI spec validation or
+    # the application-level field filter — both are acceptable gatekeepers)
+    res = client.put(
+        url_for("watch", uuid=watch_uuid),
+        headers={'x-api-key': api_key, 'content-type': 'application/json'},
+        data=json.dumps({"field_that_is_not_in_the_spec_at_all": "some value"}),
+    )
+    assert res.status_code == 400, "Unknown fields should be rejected"
+    assert (b'Validation failed' in res.data or b'Unknown field' in res.data), \
+        "Rejection should come from either the OpenAPI spec validation layer or application field filter"
+
+    delete_all_watches(client)
+
+
 def test_api_conflict_UI_password(client, live_server, measure_memory_usage, datastore_path):
 
 

changedetectionio/tests/test_api_openapi.py

@@ -9,7 +9,51 @@ by testing various scenarios that should trigger validation errors.
 import time
 import json
 from flask import url_for
-from .util import live_server_setup, wait_for_all_checks
+from .util import live_server_setup, wait_for_all_checks, delete_all_watches
+
+
+def test_openapi_merged_spec_contains_restock_fields():
+    """
+    Unit test: verify that build_merged_spec_dict() correctly merges the
+    restock_diff processor api.yaml into the base spec so that
+    WatchBase.properties includes processor_config_restock_diff with all
+    expected sub-fields.  No live server required.
+    """
+    from changedetectionio.api import build_merged_spec_dict
+
+    spec = build_merged_spec_dict()
+    schemas = spec['components']['schemas']
+
+    # The merged schema for processor_config_restock_diff should exist
+    assert 'processor_config_restock_diff' in schemas, \
+        "processor_config_restock_diff schema missing from merged spec"
+
+    restock_schema = schemas['processor_config_restock_diff']
+    props = restock_schema.get('properties', {})
+
+    expected_fields = {
+        'in_stock_processing',
+        'follow_price_changes',
+        'price_change_min',
+        'price_change_max',
+        'price_change_threshold_percent',
+    }
+    missing = expected_fields - set(props.keys())
+    assert not missing, f"Missing fields in processor_config_restock_diff schema: {missing}"
+
+    # in_stock_processing must be an enum with the three valid values
+    enum_values = set(props['in_stock_processing'].get('enum', []))
+    assert enum_values == {'in_stock_only', 'all_changes', 'off'}, \
+        f"Unexpected enum values for in_stock_processing: {enum_values}"
+
+    # WatchBase.properties must carry a $ref to the restock schema so the
+    # validation middleware can enforce it on every POST/PUT to /watch
+    watchbase_props = schemas['WatchBase']['properties']
+    assert 'processor_config_restock_diff' in watchbase_props, \
+        "processor_config_restock_diff not wired into WatchBase.properties"
+    ref = watchbase_props['processor_config_restock_diff'].get('$ref', '')
+    assert 'processor_config_restock_diff' in ref, \
+        f"WatchBase.processor_config_restock_diff should $ref the schema, got: {ref}"
 
 
 def test_openapi_validation_invalid_content_type_on_create_watch(client, live_server, measure_memory_usage, datastore_path):
@@ -27,6 +71,7 @@ def test_openapi_validation_invalid_content_type_on_create_watch(client, live_se
     # Should get 400 error due to OpenAPI validation failure
     assert res.status_code == 400, f"Expected 400 but got {res.status_code}"
     assert b"Validation failed" in res.data, "Should contain validation error message"
+    delete_all_watches(client)
 
 
 def test_openapi_validation_missing_required_field_create_watch(client, live_server, measure_memory_usage, datastore_path):
@@ -44,6 +89,7 @@ def test_openapi_validation_missing_required_field_create_watch(client, live_ser
     # Should get 400 error due to missing required field
     assert res.status_code == 400, f"Expected 400 but got {res.status_code}"
     assert b"Validation failed" in res.data, "Should contain validation error message"
+    delete_all_watches(client)
 
 
 def test_openapi_validation_invalid_field_in_request_body(client, live_server, measure_memory_usage, datastore_path):
@@ -83,6 +129,7 @@ def test_openapi_validation_invalid_field_in_request_body(client, live_server, m
     # Backend validation now returns "Unknown field(s):" message
     assert b"Unknown field" in res.data, \
             "Should contain validation error about unknown fields"
+    delete_all_watches(client)
 
 
 def test_openapi_validation_import_wrong_content_type(client, live_server, measure_memory_usage, datastore_path):
@@ -100,6 +147,7 @@ def test_openapi_validation_import_wrong_content_type(client, live_server, measu
     # Should get 400 error due to content-type mismatch
     assert res.status_code == 400, f"Expected 400 but got {res.status_code}"
     assert b"Validation failed" in res.data, "Should contain validation error message"
+    delete_all_watches(client)
 
 
 def test_openapi_validation_import_correct_content_type_succeeds(client, live_server, measure_memory_usage, datastore_path):
@@ -117,6 +165,7 @@ def test_openapi_validation_import_correct_content_type_succeeds(client, live_se
     # Should succeed
     assert res.status_code == 200, f"Expected 200 but got {res.status_code}"
     assert len(res.json) == 2, "Should import 2 URLs"
+    delete_all_watches(client)
 
 
 def test_openapi_validation_get_requests_bypass_validation(client, live_server, measure_memory_usage, datastore_path):
@@ -141,6 +190,7 @@ def test_openapi_validation_get_requests_bypass_validation(client, live_server,
 
     # Should return JSON with watch list (empty in this case)
     assert isinstance(res.json, dict), "Should return JSON dictionary for watch list"
+    delete_all_watches(client)
 
 
 def test_openapi_validation_create_tag_missing_required_title(client, live_server, measure_memory_usage, datastore_path):
@@ -158,10 +208,13 @@ def test_openapi_validation_create_tag_missing_required_title(client, live_serve
     # Should get 400 error due to missing required field
     assert res.status_code == 400, f"Expected 400 but got {res.status_code}"
     assert b"Validation failed" in res.data, "Should contain validation error message"
+    delete_all_watches(client)
 
 
 def test_openapi_validation_watch_update_allows_partial_updates(client, live_server, measure_memory_usage, datastore_path):
+
     """Test that watch updates allow partial updates without requiring all fields (positive test)."""
+#xxx
     api_key = live_server.app.config['DATASTORE'].data['settings']['application'].get('api_access_token')
 
     # First create a valid watch
@@ -198,4 +251,5 @@ def test_openapi_validation_watch_update_allows_partial_updates(client, live_ser
     )
     assert res.status_code == 200
     assert res.json.get('title') == 'Updated Title Only', "Title should be updated"
-    assert res.json.get('url') == 'https://example.com', "URL should remain unchanged"
+    assert res.json.get('url') == 'https://example.com', "URL should remain unchanged"
+    delete_all_watches(client)

changedetectionio/tests/test_api_tags.py

@@ -176,6 +176,97 @@ def test_api_tags_listing(client, live_server, measure_memory_usage, datastore_p
     assert res.status_code == 204
 
 
+def test_api_tag_restock_processor_config(client, live_server, measure_memory_usage, datastore_path):
+    """
+    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 with processor_config_restock_diff in a single POST (issue #3966)
+    res = client.post(
+        url_for("tag"),
+        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, f"POST tag with restock config failed: {res.data}"
+    tag_uuid = res.json.get('uuid')
+
+    # 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'},
+        data=json.dumps({
+            "overrides_watch": True,
+            "processor_config_restock_diff": {
+                "in_stock_processing": "in_stock_only",
+                "follow_price_changes": True,
+                "price_change_min": 8888888
+            }
+        })
+    )
+    assert res.status_code == 200, f"PUT tag with restock config failed: {res.data}"
+
+    # Verify the config was stored via API
+    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
+    assert tag_data.get('processor_config_restock_diff', {}).get('in_stock_processing') == 'in_stock_only'
+    assert tag_data.get('processor_config_restock_diff', {}).get('price_change_min') == 8888888
+
+    # Verify the value is also reflected in the UI tag edit page
+    res = client.get(url_for("tags.form_tag_edit", uuid=tag_uuid))
+    assert res.status_code == 200
+    assert b'8888888' in res.data, "price_change_min set via API should appear in the UI tag edit form"
+
+    # Invalid enum value should be rejected by OpenAPI spec validation
+    res = client.put(
+        url_for("tag", uuid=tag_uuid),
+        headers={'x-api-key': api_key, 'content-type': 'application/json'},
+        data=json.dumps({
+            "processor_config_restock_diff": {
+                "in_stock_processing": "not_a_valid_value"
+            }
+        })
+    )
+    assert res.status_code == 400
+    assert b'Validation failed' in res.data
+
+    # Clean up
+    res = client.delete(
+        url_for("tag", uuid=tag_uuid),
+        headers={'x-api-key': api_key}
+    )
+    assert res.status_code == 204
+
+
 def test_roundtrip_API(client, live_server, measure_memory_usage, datastore_path):
     """
     Test the full round trip, this way we test the default Model fits back into OpenAPI spec

changedetectionio/tests/test_backend.py

@@ -6,8 +6,6 @@ from flask import url_for
 from .util import set_original_response, set_modified_response, live_server_setup, wait_for_all_checks, extract_rss_token_from_UI, \
     extract_UUID_from_client, delete_all_watches
 
-sleep_time_for_fetch_thread = 3
-
 
 # Basic test to check inscriptus is not adding return line chars, basically works etc
 def test_inscriptus():
@@ -50,6 +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)
 

changedetectionio/tests/test_backup.py

@@ -3,14 +3,13 @@
 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
 
 
 def test_backup(client, live_server, measure_memory_usage, datastore_path):
-   #  live_server_setup(live_server) # Setup on conftest per function
-
     set_original_response(datastore_path=datastore_path)
 
 
@@ -32,7 +31,7 @@ def test_backup(client, live_server, measure_memory_usage, datastore_path):
     time.sleep(4)
 
     res = client.get(
-        url_for("backups.index"),
+        url_for("backups.create"),
         follow_redirects=True
     )
     # Can see the download link to the backup
@@ -54,11 +53,11 @@ def test_backup(client, live_server, measure_memory_usage, datastore_path):
     backup = ZipFile(io.BytesIO(res.data))
     l = backup.namelist()
 
-    # Check for UUID-based txt files (history and snapshot)
+    # Check for UUID-based txt files (history, snapshot, and last-checksum)
     uuid4hex_txt = re.compile('^[a-f0-9]{8}-?[a-f0-9]{4}-?4[a-f0-9]{3}-?[89ab][a-f0-9]{3}-?[a-f0-9]{12}.*txt', re.I)
     txt_files = list(filter(uuid4hex_txt.match, l))
-    # Should be two txt files in the archive (history and the snapshot)
+    # Should be three txt files in the archive (history, snapshot, and last-checksum)
-    assert len(txt_files) == 2
+    assert len(txt_files) == 3
 
     # Check for watch.json files (new format)
     uuid4hex_json = re.compile('^[a-f0-9]{8}-?[a-f0-9]{4}-?4[a-f0-9]{3}-?[89ab][a-f0-9]{3}-?[a-f0-9]{12}/watch\.json$', re.I)
@@ -69,10 +68,194 @@ 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"),
         follow_redirects=True
     )
 
-    assert b'No backups found.' in res.data
+    assert b'No backups found.' in res.data
+
+
+def test_watch_data_package_download(client, live_server, measure_memory_usage, datastore_path):
+    """Test downloading a single watch's data as a zip package"""
+
+    set_original_response(datastore_path=datastore_path)
+
+    uuid = client.application.config.get('DATASTORE').add_watch(url=url_for('test_endpoint', _external=True))
+    tag_uuid = client.application.config.get('DATASTORE').add_tag(title="Tasty backup tag")
+    tag_uuid2 = client.application.config.get('DATASTORE').add_tag(title="Tasty backup tag number two")
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+
+    wait_for_all_checks(client)
+
+    # Download the watch data package
+    res = client.get(url_for("ui.ui_edit.watch_get_data_package", uuid=uuid))
+
+    # Should get the right zip content type
+    assert res.content_type == "application/zip"
+
+    # Should be PK/ZIP stream (PKzip header)
+    assert res.data[:2] == b'PK', "File should start with PK (PKzip header)"
+    assert res.data.count(b'PK') >= 2, "Should have multiple PK markers (zip file structure)"
+
+    # Verify zip contents
+    backup = ZipFile(io.BytesIO(res.data))
+    files = backup.namelist()
+
+    # Should have files in a UUID directory
+    assert any(uuid in f for f in files), f"Files should be in UUID directory: {files}"
+
+    # Should contain watch.json
+    watch_json_path = f"{uuid}/watch.json"
+    assert watch_json_path in files, f"Should contain watch.json, got: {files}"
+
+    # Should contain history/snapshot files
+    uuid4hex_txt = re.compile(f'^{re.escape(uuid)}/.*\\.txt', re.I)
+    txt_files = list(filter(uuid4hex_txt.match, files))
+    assert len(txt_files) > 0, f"Should have at least one .txt file (history/snapshot), got: {files}"
+
+
+def test_backup_restore(client, live_server, measure_memory_usage, datastore_path):
+    """Test that a full backup zip can be restored — watches and tags survive a round-trip."""
+
+    set_original_response(datastore_path=datastore_path)
+
+    datastore = live_server.app.config['DATASTORE']
+    watch_url = url_for('test_endpoint', _external=True)
+
+    # Set up: one watch and two tags
+    uuid = datastore.add_watch(url=watch_url)
+    tag_uuid = datastore.add_tag(title="Tasty backup tag")
+    tag_uuid2 = datastore.add_tag(title="Tasty backup tag number two")
+
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Create a full backup
+    client.get(url_for("backups.request_backup"), follow_redirects=True)
+    time.sleep(4)
+
+    # Download the latest backup zip
+    res = client.get(url_for("backups.download_backup", filename="latest"), follow_redirects=True)
+    assert res.content_type == "application/zip"
+    zip_data = res.data
+
+    # Confirm the zip contains both watch.json and tag.json entries
+    backup = ZipFile(io.BytesIO(zip_data))
+    names = backup.namelist()
+    assert f"{uuid}/watch.json" in names, f"watch.json missing from backup: {names}"
+    assert f"{tag_uuid}/tag.json" in names, f"tag.json for tag 1 missing from backup: {names}"
+    assert f"{tag_uuid2}/tag.json" in names, f"tag.json for tag 2 missing from backup: {names}"
+
+    # --- Wipe everything ---
+    datastore.delete('all')
+    client.get(url_for("tags.delete_all"), follow_redirects=True)
+
+    assert uuid not in datastore.data['watching'], "Watch should be gone after delete"
+    assert tag_uuid not in datastore.data['settings']['application']['tags'], "Tag 1 should be gone after delete"
+    assert tag_uuid2 not in datastore.data['settings']['application']['tags'], "Tag 2 should be gone after delete"
+
+    # --- Restore from the backup zip ---
+    res = client.post(
+        url_for("backups.restore.backups_restore_start"),
+        data={
+            'zip_file': (io.BytesIO(zip_data), 'backup.zip'),
+            'include_groups': 'y',
+            'include_groups_replace_existing': 'y',
+            'include_watches': 'y',
+            'include_watches_replace_existing': 'y',
+        },
+        content_type='multipart/form-data',
+        follow_redirects=True
+    )
+    assert res.status_code == 200
+
+    # Wait for the thread to finish
+    time.sleep(2)
+
+    # --- Watch checks ---
+    restored_watch = datastore.data['watching'].get(uuid)
+    assert restored_watch is not None, f"Watch {uuid} not found after restore"
+    assert restored_watch['url'] == watch_url, "Restored watch URL does not match"
+    assert isinstance(restored_watch, Watch.model), \
+        f"Watch not properly rehydrated, got {type(restored_watch)}"
+    assert restored_watch.history_n >= 1, \
+        f"Restored watch should have at least 1 history entry, got {restored_watch.history_n}"
+
+    # --- Tag checks ---
+    restored_tags = datastore.data['settings']['application']['tags']
+
+    restored_tag = restored_tags.get(tag_uuid)
+    assert restored_tag is not None, f"Tag {tag_uuid} not found after restore"
+    assert restored_tag['title'] == "Tasty backup tag", "Restored tag 1 title does not match"
+    assert isinstance(restored_tag, Tag.model), \
+        f"Tag 1 not properly rehydrated, got {type(restored_tag)}"
+
+    restored_tag2 = restored_tags.get(tag_uuid2)
+    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)}"
+
+
+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_css_selector.py

@@ -71,22 +71,19 @@ def test_include_filters_output():
 
 # Tests the whole stack works with the CSS Filter
 def test_check_markup_include_filters_restriction(client, live_server, measure_memory_usage, datastore_path):
-    sleep_time_for_fetch_thread = 3
 
     include_filters = "#sametext"
 
     set_original_response(datastore_path=datastore_path)
 
-    # Give the endpoint time to spin up
-    time.sleep(1)
 
     # Add our URL to the import page
     test_url = url_for('test_endpoint', _external=True)
     uuid = client.application.config.get('DATASTORE').add_watch(url=test_url)
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
 
-    # Give the thread time to pick it up
+    wait_for_all_checks(client)
-    time.sleep(sleep_time_for_fetch_thread)
+
 
     # Goto the edit page, add our ignore text
     # Add our URL to the import page
@@ -103,15 +100,15 @@ def test_check_markup_include_filters_restriction(client, live_server, measure_m
     )
     assert bytes(include_filters.encode('utf-8')) in res.data
 
-    # Give the thread time to pick it up
+    wait_for_all_checks(client)
-    time.sleep(sleep_time_for_fetch_thread)
+
     #  Make a change
     set_modified_response(datastore_path=datastore_path)
 
     # Trigger a check
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
-    # Give the thread time to pick it up
+    wait_for_all_checks(client)
-    time.sleep(sleep_time_for_fetch_thread)
+
 
     # It should have 'has-unread-changes' still
     # Because it should be looking at only that 'sametext' id

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_extract_csv.py

@@ -6,10 +6,6 @@ from urllib.request import urlopen
 from .util import set_original_response, set_modified_response, live_server_setup, wait_for_all_checks
 import os
 
-sleep_time_for_fetch_thread = 3
-
-
-
 def test_check_extract_text_from_diff(client, live_server, measure_memory_usage, datastore_path):
     import time
     with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:

changedetectionio/tests/test_history_consistency.py

@@ -106,7 +106,7 @@ def test_consistent_history(client, live_server, measure_memory_usage, datastore
 
         # Find the snapshot one
         for fname in files_in_watch_dir:
-            if fname != 'history.txt' and fname != 'watch.json' and 'html' not in fname:
+            if fname != 'history.txt' and fname != 'watch.json' and fname != 'last-checksum.txt' and 'html' not in fname:
                 if strtobool(os.getenv("TEST_WITH_BROTLI")):
                     assert fname.endswith('.br'), "Forced TEST_WITH_BROTLI then it should be a .br filename"
 
@@ -123,11 +123,18 @@ def test_consistent_history(client, live_server, measure_memory_usage, datastore
                 assert json_obj['watching'][w]['title'], "Watch should have a title set"
                 assert contents.startswith(watch_title + "x"), f"Snapshot contents in file {fname} should start with '{watch_title}x', got '{contents}'"
 
-        # With new format, we also have watch.json, so 4 files total
+        # With new format, we have watch.json, so 4 files minimum
+        # Note: last-checksum.txt may or may not exist - it gets cleared by settings changes,
+        # and this test changes settings before checking files
+        # This assertion should be AFTER the loop, not inside it
         if os.path.exists(changedetection_json):
-            assert len(files_in_watch_dir) == 4, "Should be four files in the dir with new format: watch.json, html.br snapshot, history.txt and the extracted text snapshot"
+            # 4 required files: watch.json, html.br, history.txt, extracted text snapshot
+            # last-checksum.txt is optional (cleared by settings changes in this test)
+            assert len(files_in_watch_dir) >= 4 and len(files_in_watch_dir) <= 5, f"Should be 4-5 files in the dir with new format (last-checksum.txt is optional). Found {len(files_in_watch_dir)}: {files_in_watch_dir}"
         else:
-            assert len(files_in_watch_dir) == 3, "Should be just three files in the dir with legacy format: html.br snapshot, history.txt and the extracted text snapshot"
+            # 3 required files: html.br, history.txt, extracted text snapshot
+            # last-checksum.txt is optional
+            assert len(files_in_watch_dir) >= 3 and len(files_in_watch_dir) <= 4, f"Should be 3-4 files in the dir with legacy format (last-checksum.txt is optional). Found {len(files_in_watch_dir)}: {files_in_watch_dir}"
 
     # Check that 'default' Watch vars aren't accidentally being saved
     if os.path.exists(changedetection_json):

changedetectionio/tests/test_i18n.py

@@ -624,3 +624,76 @@ def test_session_locale_overrides_accept_language(client, live_server, measure_m
     assert "분".encode() in res.data, "Expected Korean '분' for Minutes"
     assert "小時".encode() not in res.data, "Should not have Traditional Chinese '小時' when Korean is set"
     assert "分鐘".encode() not in res.data, "Should not have Traditional Chinese '分鐘' when Korean is set"
+
+
+def test_clear_history_translated_confirmation(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that clearing snapshot history works with translated confirmation text.
+
+    Issue #3865: When the app language is set to German, the clear history
+    confirmation dialog shows the translated word (e.g. 'loschen') but the
+    backend only accepted the English word 'clear', making it impossible
+    to clear snapshots in non-English languages.
+    """
+    from flask import url_for
+
+    test_url = url_for('test_endpoint', _external=True)
+
+    # Add a watch so there is history to clear
+    res = client.post(
+        url_for("imports.import_page"),
+        data={"urls": test_url},
+        follow_redirects=True
+    )
+    assert b"1 Imported" in res.data
+    wait_for_all_checks(client)
+
+    # Set language to German
+    res = client.get(
+        url_for("set_language", locale="de"),
+        follow_redirects=True
+    )
+    assert res.status_code == 200
+
+    # Verify the clear history page shows the German confirmation word
+    res = client.get(
+        url_for("ui.clear_all_history"),
+        follow_redirects=True
+    )
+    assert res.status_code == 200
+    assert "löschen".encode() in res.data, "Expected German word 'loschen' on clear history page"
+
+    # Submit the form with the German translated word
+    res = client.post(
+        url_for("ui.clear_all_history"),
+        data={"confirmtext": "löschen"},
+        follow_redirects=True
+    )
+    assert res.status_code == 200
+    # Should NOT show error message
+    assert b"Incorrect confirmation text" not in res.data, \
+        "German confirmation word 'loschen' should be accepted (issue #3865)"
+
+    # Switch back to English and verify English word still works
+    res = client.get(
+        url_for("set_language", locale="en_US"),
+        follow_redirects=True
+    )
+
+    res = client.post(
+        url_for("ui.clear_all_history"),
+        data={"confirmtext": "clear"},
+        follow_redirects=True
+    )
+    assert res.status_code == 200
+    assert b"Incorrect confirmation text" not in res.data, \
+        "English confirmation word 'clear' should still be accepted"
+
+    # Verify that missing/empty confirmtext does not crash the server
+    res = client.post(
+        url_for("ui.clear_all_history"),
+        data={},
+        follow_redirects=True
+    )
+    assert res.status_code == 200, \
+        "Missing confirmtext should not crash the server"

changedetectionio/tests/test_ignorehyperlinks.py

@@ -41,7 +41,6 @@ def set_modified_ignore_response(datastore_path):
 def test_render_anchor_tag_content_true(client, live_server, measure_memory_usage, datastore_path):
     """Testing that the link changes are detected when
     render_anchor_tag_content setting is set to true"""
-    sleep_time_for_fetch_thread = 3
 
     # Give the endpoint time to spin up
     time.sleep(1)

changedetectionio/tests/test_ignorestatuscode.py

@@ -100,7 +100,6 @@ def test_normal_page_check_works_with_ignore_status_code(client, live_server, me
 
 # Tests the whole stack works with staus codes ignored
 def test_403_page_check_works_with_ignore_status_code(client, live_server, measure_memory_usage, datastore_path):
-    sleep_time_for_fetch_thread = 3
 
     set_original_response(datastore_path=datastore_path)
 
@@ -112,8 +111,7 @@ def test_403_page_check_works_with_ignore_status_code(client, live_server, measu
     uuid = client.application.config.get('DATASTORE').add_watch(url=test_url)
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
 
-    # Give the thread time to pick it up
+    wait_for_all_checks(client)
-    time.sleep(sleep_time_for_fetch_thread)
 
     # Goto the edit page, check our ignore option
     # Add our URL to the import page

changedetectionio/tests/test_ignorewhitespace.py

@@ -2,10 +2,9 @@
 
 import time
 from flask import url_for
-from . util import live_server_setup
 import os
 
-
+from .util import live_server_setup, delete_all_watches, wait_for_all_checks
 
 
 # Should be the same as set_original_ignore_response(datastore_path=datastore_path) but with a little more whitespacing
@@ -50,10 +49,7 @@ def set_original_ignore_response(datastore_path):
 
 # If there was only a change in the whitespacing, then we shouldnt have a change detected
 def test_check_ignore_whitespace(client, live_server, measure_memory_usage, datastore_path):
-    sleep_time_for_fetch_thread = 3
 
-    # Give the endpoint time to spin up
-    time.sleep(1)
 
     set_original_ignore_response(datastore_path=datastore_path)
 
@@ -74,17 +70,17 @@ def test_check_ignore_whitespace(client, live_server, measure_memory_usage, data
     uuid = client.application.config.get('DATASTORE').add_watch(url=test_url)
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
 
-    time.sleep(sleep_time_for_fetch_thread)
+    wait_for_all_checks(client)
     # Trigger a check
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
 
     set_original_ignore_response_but_with_whitespace(datastore_path)
-    time.sleep(sleep_time_for_fetch_thread)
+    wait_for_all_checks(client)
     # Trigger a check
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
 
     # Give the thread time to pick it up
-    time.sleep(sleep_time_for_fetch_thread)
+    wait_for_all_checks(client)
 
     # It should report nothing found (no new 'has-unread-changes' class)
     res = client.get(url_for("watchlist.index"))

changedetectionio/tests/test_jsonpath_jq_selector.py

@@ -16,6 +16,51 @@ except ModuleNotFoundError:
 
 
 
+def test_jsonp_treated_as_plaintext():
+    from ..processors.magic import guess_stream_type
+
+    # JSONP content (server wrongly claims application/json) should be detected as plaintext
+    # Callback names are arbitrary identifiers, not always 'cb'
+    jsonp_content = 'jQuery123456({ "version": "8.0.41", "url": "https://example.com/app.apk" })'
+    result = guess_stream_type(http_content_header="application/json", content=jsonp_content)
+    assert result.is_json is False
+    assert result.is_plaintext is True
+
+    # Variation with dotted callback name e.g. jQuery.cb(...)
+    jsonp_dotted = 'some.callback({ "version": "1.0" })'
+    result = guess_stream_type(http_content_header="application/json", content=jsonp_dotted)
+    assert result.is_json is False
+    assert result.is_plaintext is True
+
+    # Real JSON should still be detected as JSON
+    json_content = '{ "version": "8.0.41", "url": "https://example.com/app.apk" }'
+    result = guess_stream_type(http_content_header="application/json", content=json_content)
+    assert result.is_json is True
+    assert result.is_plaintext is False
+
+
+def test_jsonp_json_filter_extraction():
+    from .. import html_tools
+
+    # Tough case: dotted namespace callback, trailing semicolon, deeply nested content with arrays
+    jsonp_content = 'weixin.update.callback({"platforms": {"android": {"variants": [{"arch": "arm64", "versionName": "8.0.68", "url": "https://example.com/app-arm64.apk"}, {"arch": "arm32", "versionName": "8.0.41", "url": "https://example.com/app-arm32.apk"}]}}});'
+
+    # Deep nested jsonpath filter into array element
+    text = html_tools.extract_json_as_string(jsonp_content, "json:$.platforms.android.variants[0].versionName")
+    assert text == '"8.0.68"'
+
+    # Filter that selects the second array element
+    text = html_tools.extract_json_as_string(jsonp_content, "json:$.platforms.android.variants[1].arch")
+    assert text == '"arm32"'
+
+    if jq_support:
+        text = html_tools.extract_json_as_string(jsonp_content, "jq:.platforms.android.variants[0].versionName")
+        assert text == '"8.0.68"'
+
+        text = html_tools.extract_json_as_string(jsonp_content, "jqraw:.platforms.android.variants[1].url")
+        assert text == "https://example.com/app-arm32.apk"
+
+
 def test_unittest_inline_html_extract():
     # So lets pretend that the JSON we want is inside some HTML
     content="""

changedetectionio/tests/test_notification.py

@@ -17,6 +17,7 @@ from changedetectionio.notification import (
 )
 from ..diff import HTML_CHANGED_STYLE
 from ..model import USE_SYSTEM_DEFAULT_NOTIFICATION_FORMAT_FOR_WATCH
+from ..notification_service import FormattableTimestamp
 
 
 # Hard to just add more live server URLs when one test is already running (I think)
@@ -107,7 +108,11 @@ def test_check_notification(client, live_server, measure_memory_usage, datastore
                                                    "Diff Added: {{diff_added}}\n"
                                                    "Diff Removed: {{diff_removed}}\n"
                                                    "Diff Full: {{diff_full}}\n"
+                                                   "Diff with args: {{diff(context=3)}}"
                                                    "Diff as Patch: {{diff_patch}}\n"
+                                                   "Change datetime: {{change_datetime}}\n"
+                                                   "Change datetime format: Weekday {{change_datetime(format='%A')}}\n"
+                                                   "Change datetime format: {{change_datetime(format='%Y-%m-%dT%H:%M:%S%z')}}\n"
                                                    ":-)",
                               "notification_screenshot": True,
                               "notification_format": 'text'}
@@ -135,8 +140,6 @@ def test_check_notification(client, live_server, measure_memory_usage, datastore
     assert bytes(notification_url.encode('utf-8')) in res.data
     assert bytes("New ChangeDetection.io Notification".encode('utf-8')) in res.data
 
-
-
     ## Now recheck, and it should have sent the notification
     wait_for_all_checks(client)
     set_modified_response(datastore_path=datastore_path)
@@ -172,11 +175,23 @@ def test_check_notification(client, live_server, measure_memory_usage, datastore
     assert ":-)" in notification_submission
     assert "New ChangeDetection.io Notification - {}".format(test_url) in notification_submission
     assert test_url in notification_submission
+
     assert ':-)' in notification_submission
     # Check the attachment was added, and that it is a JPEG from the original PNG
     notification_submission_object = json.loads(notification_submission)
     assert notification_submission_object
 
+    import time
+    # Could be from a few seconds ago (when the notification was fired vs in this test checking), so check for any
+    times_possible = [str(FormattableTimestamp(int(time.time()) - i)) for i in range(15)]
+    assert any(t in notification_submission for t in times_possible)
+
+    txt = f"Weekday {FormattableTimestamp(int(time.time()))(format='%A')}"
+    assert txt in notification_submission
+
+
+
+
     # We keep PNG screenshots for now
     # IF THIS FAILS YOU SHOULD BE TESTING WITH ENV VAR REMOVE_REQUESTS_OLD_SCREENSHOTS=False
     assert notification_submission_object['attachments'][0]['filename'] == 'last-screenshot.png'

changedetectionio/tests/test_restock_itemprop.py

@@ -109,7 +109,7 @@ def test_itemprop_price_change(client, live_server, measure_memory_usage, datast
     set_original_response(props_markup=instock_props[0], price='120.45', datastore_path=datastore_path)
     res = client.post(
         url_for("ui.ui_edit.edit_page", uuid="first"),
-        data={"restock_settings-follow_price_changes": "", "url": test_url, "tags": "", "headers": "", 'fetch_backend': "html_requests", "time_between_check_use_default": "y"},
+        data={"processor_config_restock_diff-follow_price_changes": "", "url": test_url, "tags": "", "headers": "", 'fetch_backend': "html_requests", "time_between_check_use_default": "y"},
         follow_redirects=True
     )
     assert b"Updated watch." in res.data
@@ -204,9 +204,9 @@ def _run_test_minmax_limit(client, extra_watch_edit_form, datastore_path):
 def test_restock_itemprop_minmax(client, live_server, measure_memory_usage, datastore_path):
     
     extras = {
-        "restock_settings-follow_price_changes": "y",
+        "processor_config_restock_diff-follow_price_changes": "y",
-        "restock_settings-price_change_min": 900.0,
+        "processor_config_restock_diff-price_change_min": 900.0,
-        "restock_settings-price_change_max": 1100.10
+        "processor_config_restock_diff-price_change_max": 1100.10
     }
     _run_test_minmax_limit(client, extra_watch_edit_form=extras, datastore_path=datastore_path)
 
@@ -223,9 +223,9 @@ def test_restock_itemprop_with_tag(client, live_server, measure_memory_usage, da
     res = client.post(
         url_for("tags.form_tag_edit_submit", uuid="first"),
         data={"name": "test-tag",
-              "restock_settings-follow_price_changes": "y",
+              "processor_config_restock_diff-follow_price_changes": "y",
-              "restock_settings-price_change_min": 900.0,
+              "processor_config_restock_diff-price_change_min": 900.0,
-              "restock_settings-price_change_max": 1100.10,
+              "processor_config_restock_diff-price_change_max": 1100.10,
               "overrides_watch": "y", #overrides_watch should be restock_overrides_watch
               },
         follow_redirects=True
@@ -258,8 +258,8 @@ def test_itemprop_percent_threshold(client, live_server, measure_memory_usage, d
 
     res = client.post(
         url_for("ui.ui_edit.edit_page", uuid="first"),
-        data={"restock_settings-follow_price_changes": "y",
+        data={"processor_config_restock_diff-follow_price_changes": "y",
-              "restock_settings-price_change_threshold_percent": 5.0,
+              "processor_config_restock_diff-price_change_threshold_percent": 5.0,
               "url": test_url,
               "tags": "",
               "headers": "",
@@ -305,8 +305,8 @@ def test_itemprop_percent_threshold(client, live_server, measure_memory_usage, d
 
     res = client.post(
         url_for("ui.ui_edit.edit_page", uuid=uuid),
-        data={"restock_settings-follow_price_changes": "y",
+        data={"processor_config_restock_diff-follow_price_changes": "y",
-              "restock_settings-price_change_threshold_percent": 5.05,
+              "processor_config_restock_diff-price_change_threshold_percent": 5.05,
               "processor": "text_json_diff",
               "url": test_url,
               'fetch_backend': "html_requests",
@@ -467,3 +467,38 @@ def test_special_prop_examples(client, live_server, measure_memory_usage, datast
             assert b'155.55' in res.data
 
     delete_all_watches(client)
+
+
+def test_itemprop_as_str(client, live_server, measure_memory_usage, datastore_path):
+
+    test_return_data = f"""<html>
+       <body>
+     Some initial text<br>
+     <p>Which is across multiple lines</p>
+<span itemprop="offers" itemscope itemtype="http://schema.org/Offer">
+<meta content="767.55" itemprop="price"/>
+<meta content="EUR" itemprop="priceCurrency"/>
+<meta content="InStock" itemprop="availability"/>
+<meta content="https://www.123-test.dk" itemprop="url"/>
+</span>
+     </body>
+     </html>
+    """
+
+    with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:
+        f.write(test_return_data)
+
+
+    test_url = url_for('test_endpoint', _external=True)
+
+    client.post(
+        url_for("ui.ui_views.form_quick_watch_add"),
+        data={"url": test_url, "tags": 'restock tests', 'processor': 'restock_diff'},
+        follow_redirects=True
+    )
+
+    client.get(url_for("ui.form_watch_checknow"))
+    wait_for_all_checks(client)
+
+    res = client.get(url_for("watchlist.index"))
+    assert b'767.55' in res.data

changedetectionio/tests/test_security.py

@@ -1,4 +1,5 @@
 import os
+import pytest
 
 from flask import url_for
 
@@ -24,6 +25,31 @@ def set_original_response(datastore_path):
         f.write(test_return_data)
     return None
 
+
+def test_favicon(client, live_server, measure_memory_usage, datastore_path):
+    # Attempt to fetch it, make sure that works
+    SVG_BASE64 = 'PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHZpZXdCb3g9IjAgMCAxIDEiLz4='
+    uuid = client.application.config.get('DATASTORE').add_watch(url='https://localhost')
+    live_server.app.config['DATASTORE'].data['watching'][uuid].bump_favicon(url="favicon-set-type.svg",
+                                                                            favicon_base_64=SVG_BASE64
+                                                                            )
+
+
+    res = client.get(url_for('static_content', group='favicon', filename=uuid))
+    assert res.status_code == 200
+    assert len(res.data) > 10
+
+    res = client.get(url_for('static_content', group='..', filename='__init__.py'))
+    assert res.status_code != 200
+
+
+    res = client.get(url_for('static_content', group='.', filename='../__init__.py'))
+    assert res.status_code != 200
+
+    # Traverse by filename protection
+    res = client.get(url_for('static_content', group='js', filename='../styles/styles.css'))
+    assert res.status_code != 200
+
 def test_bad_access(client, live_server, measure_memory_usage, datastore_path):
 
     res = client.post(
@@ -478,3 +504,243 @@ def test_logout_with_redirect(client, live_server, measure_memory_usage, datasto
     # Cleanup
     del client.application.config['DATASTORE'].data['settings']['application']['password']
 
+
+def test_static_directory_traversal(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that the static file serving route properly blocks directory traversal attempts.
+    This tests the fix for GHSA-9jj8-v89v-xjvw (CVE pending).
+
+    The vulnerability was in /static/<group>/<filename> where the sanitization regex
+    allowed dots, enabling "../" traversal to read application source files.
+
+    The fix changed the regex from r'[^\w.-]+' to r'[^a-z0-9_]+' which blocks dots.
+    """
+
+    # Test 1: Direct .. traversal attempt (URL-encoded)
+    res = client.get(
+        "/static/%2e%2e/flask_app.py",
+        follow_redirects=False
+    )
+    # Should be blocked (404 or 403)
+    assert res.status_code in [404, 403], f"Expected 404/403, got {res.status_code}"
+    # Should NOT contain application source code
+    assert b"def static_content" not in res.data
+    assert b"changedetection_app" not in res.data
+
+    # Test 2: Direct .. traversal attempt (unencoded)
+    res = client.get(
+        "/static/../flask_app.py",
+        follow_redirects=False
+    )
+    assert res.status_code in [404, 403], f"Expected 404/403, got {res.status_code}"
+    assert b"def static_content" not in res.data
+
+    # Test 3: Multiple dots traversal
+    res = client.get(
+        "/static/..../flask_app.py",
+        follow_redirects=False
+    )
+    assert res.status_code in [404, 403], f"Expected 404/403, got {res.status_code}"
+    assert b"def static_content" not in res.data
+
+    # Test 4: Try to access other application files
+    for filename in ["__init__.py", "datastore.py", "store.py"]:
+        res = client.get(
+            f"/static/%2e%2e/{filename}",
+            follow_redirects=False
+        )
+        assert res.status_code in [404, 403], f"File {filename} should be blocked"
+        # Should not contain Python code indicators
+        assert b"import" not in res.data or b"# Test" in res.data  # Allow "1 Imported" etc
+
+    # Test 5: Verify legitimate static files still work
+    # Note: We can't test actual files without knowing what exists,
+    # but we can verify the sanitization doesn't break valid groups
+    res = client.get(
+        "/static/images/test.png",  # Will 404 if file doesn't exist, but won't traverse
+        follow_redirects=False
+    )
+    # Should get 404 (file not found) not 403 (blocked)
+    # This confirms the group name "images" is valid
+    assert res.status_code == 404
+
+    # Test 6: Ensure hyphens and dots are blocked in group names
+    res = client.get(
+        "/static/../../../etc/passwd",
+        follow_redirects=False
+    )
+    assert res.status_code in [404, 403]
+    assert b"root:" not in res.data
+
+    # Test 7: Test that underscores still work (they're allowed)
+    res = client.get(
+        "/static/visual_selector_data/test.json",
+        follow_redirects=False
+    )
+    # visual_selector_data is a real group, but requires auth
+    # Should get 403 (not authenticated) or 404 (file not found), not a path traversal
+    assert res.status_code in [403, 404]
+
+
+def test_ssrf_private_ip_blocked(client, live_server, monkeypatch, measure_memory_usage, datastore_path):
+    """
+    SSRF protection: IANA-reserved/private IP addresses are blocked at fetch-time, not add-time.
+
+    Watches targeting private/reserved IPs can be *added* freely; the block happens when the
+    fetcher actually tries to reach the URL (via validate_iana_url() in call_browser()).
+
+    Covers:
+    1. is_private_hostname() correctly classifies all reserved ranges
+    2. is_safe_valid_url() ALLOWS private-IP URLs at add-time (IANA check moved to fetch-time)
+    3. ALLOW_IANA_RESTRICTED_ADDRESSES has no effect on add-time; it only controls fetch-time
+    4. UI form accepts private-IP URLs at add-time without error
+    5. Requests fetcher blocks fetch-time DNS rebinding (fresh check on every fetch)
+    6. Requests fetcher blocks redirects that lead to a private IP (open-redirect bypass)
+
+    conftest.py sets ALLOW_IANA_RESTRICTED_ADDRESSES=true globally so the test
+    server (localhost) keeps working for all other tests.  monkeypatch temporarily
+    overrides it to 'false' here, and is automatically restored after the test.
+    """
+    from unittest.mock import patch, MagicMock
+    from changedetectionio.validate_url import is_safe_valid_url, is_private_hostname
+
+    monkeypatch.setenv('ALLOW_IANA_RESTRICTED_ADDRESSES', 'false')
+
+    # ------------------------------------------------------------------
+    # 1. is_private_hostname() — unit tests across all reserved ranges
+    # ------------------------------------------------------------------
+    private_hosts = [
+        '127.0.0.1',          # loopback
+        '10.0.0.1',           # RFC 1918
+        '172.16.0.1',         # RFC 1918
+        '192.168.1.1',        # RFC 1918
+        '169.254.169.254',    # link-local / AWS metadata endpoint
+        '::1',                # IPv6 loopback
+        'fc00::1',            # IPv6 unique local
+        'fe80::1',            # IPv6 link-local
+    ]
+    for host in private_hosts:
+        assert is_private_hostname(host), f"{host} should be identified as private/reserved"
+
+    for host in ['8.8.8.8', '1.1.1.1']:
+        assert not is_private_hostname(host), f"{host} should be identified as public"
+
+    # ------------------------------------------------------------------
+    # 2. is_safe_valid_url() ALLOWS private-IP URLs at add-time
+    #    IANA check is no longer done here — it moved to fetch-time validate_iana_url()
+    # ------------------------------------------------------------------
+    private_ip_urls = [
+        'http://127.0.0.1/',
+        'http://10.0.0.1/',
+        'http://172.16.0.1/',
+        'http://192.168.1.1/',
+        'http://169.254.169.254/',
+        'http://169.254.169.254/latest/meta-data/iam/security-credentials/',
+        'http://[::1]/',
+        'http://[fc00::1]/',
+        'http://[fe80::1]/',
+    ]
+    for url in private_ip_urls:
+        assert is_safe_valid_url(url), f"{url} should be allowed by is_safe_valid_url (IANA check is at fetch-time)"
+
+    # ------------------------------------------------------------------
+    # 3. ALLOW_IANA_RESTRICTED_ADDRESSES does not affect add-time validation
+    #    It only controls fetch-time blocking inside validate_iana_url()
+    # ------------------------------------------------------------------
+    monkeypatch.setenv('ALLOW_IANA_RESTRICTED_ADDRESSES', 'true')
+    assert is_safe_valid_url('http://127.0.0.1/'), \
+        "Private IP should be allowed at add-time regardless of ALLOW_IANA_RESTRICTED_ADDRESSES"
+
+    monkeypatch.setenv('ALLOW_IANA_RESTRICTED_ADDRESSES', 'false')
+    assert is_safe_valid_url('http://127.0.0.1/'), \
+        "Private IP should be allowed at add-time regardless of ALLOW_IANA_RESTRICTED_ADDRESSES"
+
+    # ------------------------------------------------------------------
+    # 4. UI form accepts private-IP URLs at add-time
+    #    The watch is created; the SSRF block fires later at fetch-time
+    # ------------------------------------------------------------------
+    for url in ['http://127.0.0.1/', 'http://169.254.169.254/latest/meta-data/']:
+        res = client.post(
+            url_for('ui.ui_views.form_quick_watch_add'),
+            data={'url': url, 'tags': ''},
+            follow_redirects=True
+        )
+        assert b'Watch protocol is not permitted or invalid URL format' not in res.data, \
+            f"UI should accept {url} at add-time (SSRF is blocked at fetch-time)"
+
+    # ------------------------------------------------------------------
+    # 5. Fetch-time DNS-rebinding check in the requests fetcher
+    #    Simulates: URL passed add-time validation with a public IP, but
+    #    by fetch time DNS has been rebound to a private IP.
+    # ------------------------------------------------------------------
+    from changedetectionio.content_fetchers.requests import fetcher as RequestsFetcher
+
+    f = RequestsFetcher()
+
+    with patch('changedetectionio.content_fetchers.requests.is_private_hostname', return_value=True):
+        with pytest.raises(Exception, match='private/reserved'):
+            f._run_sync(
+                url='http://example.com/',
+                timeout=5,
+                request_headers={},
+                request_body=None,
+                request_method='GET',
+            )
+
+    # ------------------------------------------------------------------
+    # 6. Redirect-to-private-IP blocked (open-redirect SSRF bypass)
+    #    Public host returns a 302 pointing at an IANA-reserved address.
+    # ------------------------------------------------------------------
+    mock_redirect = MagicMock()
+    mock_redirect.is_redirect = True
+    mock_redirect.status_code = 302
+    mock_redirect.headers = {'Location': 'http://169.254.169.254/latest/meta-data/'}
+
+    def _private_only_for_redirect(hostname):
+        # Initial host is "public"; the redirect target is private
+        return hostname in {'169.254.169.254', '10.0.0.1', '172.16.0.1',
+                            '192.168.0.1', '127.0.0.1', '::1'}
+
+    with patch('changedetectionio.content_fetchers.requests.is_private_hostname',
+               side_effect=_private_only_for_redirect):
+        with patch('requests.Session.request', return_value=mock_redirect):
+            with pytest.raises(Exception, match='Redirect blocked'):
+                f._run_sync(
+                    url='http://example.com/',
+                    timeout=5,
+                    request_headers={},
+                    request_body=None,
+                    request_method='GET',
+                )
+
+
+def test_unresolvable_hostname_is_allowed(client, live_server, monkeypatch):
+    """
+    Unresolvable hostnames must NOT be blocked at add-time when ALLOW_IANA_RESTRICTED_ADDRESSES=false.
+
+    DNS failure (gaierror) at add-time does not mean the URL resolves to a private IP —
+    the domain may simply be offline or not yet live. Blocking it would be a false positive.
+    The real DNS-rebinding protection happens at fetch-time in call_browser().
+    """
+    from changedetectionio.validate_url import is_safe_valid_url
+
+    monkeypatch.setenv('ALLOW_IANA_RESTRICTED_ADDRESSES', 'false')
+
+    url = 'http://this-host-does-not-exist-xyz987.invalid/some/path'
+
+    # Should pass URL validation despite being unresolvable
+    assert is_safe_valid_url(url), \
+        "Unresolvable hostname should pass is_safe_valid_url — DNS failure is not a private-IP signal"
+
+    # Should be accepted via the UI form and appear in the watch list
+    res = client.post(
+        url_for('ui.ui_views.form_quick_watch_add'),
+        data={'url': url, 'tags': ''},
+        follow_redirects=True
+    )
+    assert b'Watch protocol is not permitted or invalid URL format' not in res.data, \
+        "UI should not reject a URL just because its hostname is unresolvable"
+
+    res = client.get(url_for('watchlist.index'))
+    assert b'this-host-does-not-exist-xyz987.invalid' in res.data, \
+        "Unresolvable hostname watch should appear in the watch overview list"

changedetectionio/tests/test_settings_tag_force_reprocess.py

@@ -0,0 +1,208 @@
+#!/usr/bin/env python3
+"""
+Test that changing global settings or tag configurations forces reprocessing.
+
+When settings or tag configurations change, all affected watches need to
+reprocess even if their content hasn't changed, because configuration affects
+the processing result.
+"""
+
+import os
+import time
+from flask import url_for
+from .util import wait_for_all_checks
+
+
+def test_settings_change_forces_reprocess(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that changing global settings clears all checksums to force reprocessing.
+    """
+
+    # Setup test content
+    test_html = """<html>
+     <body>
+     <p>Test content that stays the same</p>
+     </body>
+     </html>
+    """
+    with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:
+        f.write(test_html)
+
+    test_url = url_for('test_endpoint', _external=True)
+
+    # Add two watches
+    datastore = client.application.config.get('DATASTORE')
+    uuid1 = datastore.add_watch(url=test_url, extras={'title': 'Watch 1'})
+    uuid2 = datastore.add_watch(url=test_url, extras={'title': 'Watch 2'})
+
+    # Unpause watches
+    datastore.data['watching'][uuid1]['paused'] = False
+    datastore.data['watching'][uuid2]['paused'] = False
+
+    # First check - establishes baseline
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Verify checksum files were created
+    checksum1 = os.path.join(datastore_path, uuid1, 'last-checksum.txt')
+    checksum2 = os.path.join(datastore_path, uuid2, 'last-checksum.txt')
+    assert os.path.isfile(checksum1), "First check should create checksum file for watch 1"
+    assert os.path.isfile(checksum2), "First check should create checksum file for watch 2"
+
+    # Change global settings (any setting will do)
+    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_requests"
+        },
+        follow_redirects=True
+    )
+    assert b"Settings updated." in res.data
+
+    # Give it a moment to process
+    time.sleep(0.5)
+
+    # Verify ALL checksum files were deleted
+    assert not os.path.isfile(checksum1), "Settings change should delete checksum for watch 1"
+    assert not os.path.isfile(checksum2), "Settings change should delete checksum for watch 2"
+
+    # Next check should reprocess (not skip) and recreate checksums
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Verify checksum files were recreated
+    assert os.path.isfile(checksum1), "Reprocessing should recreate checksum file for watch 1"
+    assert os.path.isfile(checksum2), "Reprocessing should recreate checksum file for watch 2"
+
+    print("✓ Settings change forces reprocessing of all watches")
+
+
+def test_tag_change_forces_reprocess(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that changing a tag configuration clears checksums only for watches with that tag.
+    """
+
+    # Setup test content
+    test_html = """<html>
+     <body>
+     <p>Test content that stays the same</p>
+     </body>
+     </html>
+    """
+    with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:
+        f.write(test_html)
+
+    test_url = url_for('test_endpoint', _external=True)
+
+    # Create a tag
+    datastore = client.application.config.get('DATASTORE')
+    tag_uuid = datastore.add_tag('Test Tag')
+
+    # Add watches - one with tag, one without
+    uuid_with_tag = datastore.add_watch(url=test_url, extras={'title': 'Watch With Tag', 'tags': [tag_uuid]})
+    uuid_without_tag = datastore.add_watch(url=test_url, extras={'title': 'Watch Without Tag'})
+
+    # Unpause watches
+    datastore.data['watching'][uuid_with_tag]['paused'] = False
+    datastore.data['watching'][uuid_without_tag]['paused'] = False
+
+    # First check - establishes baseline
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Verify checksum files were created
+    checksum_with = os.path.join(datastore_path, uuid_with_tag, 'last-checksum.txt')
+    checksum_without = os.path.join(datastore_path, uuid_without_tag, 'last-checksum.txt')
+    assert os.path.isfile(checksum_with), "First check should create checksum for tagged watch"
+    assert os.path.isfile(checksum_without), "First check should create checksum for untagged watch"
+
+    # Edit the tag (change notification_muted as an example)
+    tag = datastore.data['settings']['application']['tags'][tag_uuid]
+    res = client.post(
+        url_for("tags.form_tag_edit_submit", uuid=tag_uuid),
+        data={
+            'title': 'Test Tag',
+            'notification_muted': 'y',
+            'overrides_watch': 'n'
+        },
+        follow_redirects=True
+    )
+    assert b"Updated" in res.data
+
+    # Give it a moment to process
+    time.sleep(0.5)
+
+    # Verify ONLY the tagged watch's checksum was deleted
+    assert not os.path.isfile(checksum_with), "Tag change should delete checksum for watch WITH tag"
+    assert os.path.isfile(checksum_without), "Tag change should NOT delete checksum for watch WITHOUT tag"
+
+    # Next check should reprocess tagged watch and recreate its checksum
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Verify tagged watch's checksum was recreated
+    assert os.path.isfile(checksum_with), "Reprocessing should recreate checksum for tagged watch"
+    assert os.path.isfile(checksum_without), "Untagged watch should still have its checksum"
+
+    print("✓ Tag change forces reprocessing only for watches with that tag")
+
+
+def test_tag_change_via_api_forces_reprocess(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that updating a tag via API also clears checksums for affected watches.
+    """
+
+    # Setup test content
+    test_html = """<html>
+     <body>
+     <p>Test content</p>
+     </body>
+     </html>
+    """
+    with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:
+        f.write(test_html)
+
+    test_url = url_for('test_endpoint', _external=True)
+
+    # Create a tag
+    datastore = client.application.config.get('DATASTORE')
+    tag_uuid = datastore.add_tag('API Test Tag')
+
+    # Add watch with tag
+    uuid_with_tag = datastore.add_watch(url=test_url, extras={'title': 'API Watch'})
+    datastore.data['watching'][uuid_with_tag]['paused'] = False
+    datastore.data['watching'][uuid_with_tag]['tags'] = [tag_uuid]
+    datastore.data['watching'][uuid_with_tag].commit()
+
+    # First check
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Verify checksum exists
+    checksum_file = os.path.join(datastore_path, uuid_with_tag, 'last-checksum.txt')
+    assert os.path.isfile(checksum_file), "First check should create checksum file"
+
+    # Update tag via API
+    res = client.put(
+        f'/api/v1/tag/{tag_uuid}',
+        json={'notification_muted': True},
+        headers={'x-api-key': datastore.data['settings']['application']['api_access_token']}
+    )
+    assert res.status_code == 200, f"API call failed with status {res.status_code}: {res.data}"
+
+    # Give it more time for async operations
+    time.sleep(1.0)
+
+    # Debug: Check if checksum still exists
+    if os.path.isfile(checksum_file):
+        # Read checksum to see if it changed
+        with open(checksum_file, 'r') as f:
+            checksum_content = f.read()
+            print(f"Checksum still exists: {checksum_content}")
+
+    # Verify checksum was deleted
+    assert not os.path.isfile(checksum_file), "API tag update should delete checksum"
+
+    print("✓ Tag update via API forces reprocessing")

changedetectionio/tests/test_share_watch.py

@@ -6,9 +6,6 @@ from urllib.request import urlopen
 from .util import set_original_response, set_modified_response, live_server_setup, delete_all_watches
 import re
 
-sleep_time_for_fetch_thread = 3
-
-
 def test_share_watch(client, live_server, measure_memory_usage, datastore_path):
     set_original_response(datastore_path=datastore_path)
 

changedetectionio/tests/test_source.py

@@ -6,7 +6,6 @@ from urllib.request import urlopen
 from .util import set_original_response, set_modified_response, live_server_setup, wait_for_all_checks
 from ..diff import ADDED_STYLE
 
-sleep_time_for_fetch_thread = 3
 
 def test_check_basic_change_detection_functionality_source(client, live_server, measure_memory_usage, datastore_path):
     set_original_response(datastore_path=datastore_path)
@@ -72,7 +71,10 @@ def test_check_ignore_elements(client, live_server, measure_memory_usage, datast
         follow_redirects=True
     )
 
-    time.sleep(sleep_time_for_fetch_thread)
+    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"),

changedetectionio/tests/test_trigger_regex_with_filter.py

@@ -2,7 +2,8 @@
 
 import time
 from flask import url_for
-from . util import live_server_setup, delete_all_watches
+
+from .util import live_server_setup, delete_all_watches, wait_for_all_checks
 import os
 
 
@@ -25,9 +26,6 @@ def set_original_ignore_response(datastore_path):
 
 def test_trigger_regex_functionality_with_filter(client, live_server, measure_memory_usage, datastore_path):
 
-   #  live_server_setup(live_server) # Setup on conftest per function
-    sleep_time_for_fetch_thread = 3
-
     set_original_ignore_response(datastore_path=datastore_path)
 
     # Give the endpoint time to spin up
@@ -38,8 +36,7 @@ def test_trigger_regex_functionality_with_filter(client, live_server, measure_me
     uuid = client.application.config.get('DATASTORE').add_watch(url=test_url)
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
 
-    # it needs time to save the original version
+    wait_for_all_checks(client)
-    time.sleep(sleep_time_for_fetch_thread)
 
     ### test regex with filter
     res = client.post(
@@ -52,8 +49,9 @@ def test_trigger_regex_functionality_with_filter(client, live_server, measure_me
         follow_redirects=True
     )
 
-    # Give the thread time to pick it up
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
-    time.sleep(sleep_time_for_fetch_thread)
+
+    wait_for_all_checks(client)
 
     client.get(url_for("ui.ui_diff.diff_history_page", uuid="first"))
 
@@ -62,7 +60,8 @@ def test_trigger_regex_functionality_with_filter(client, live_server, measure_me
         f.write("<html>some new noise with cool stuff2 ok</html>")
 
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
-    time.sleep(sleep_time_for_fetch_thread)
+
+    wait_for_all_checks(client)
 
     # It should report nothing found (nothing should match the regex and filter)
     res = client.get(url_for("watchlist.index"))
@@ -73,7 +72,8 @@ def test_trigger_regex_functionality_with_filter(client, live_server, measure_me
         f.write("<html>some new noise with <span id=in-here>cool stuff6</span> ok</html>")
 
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
-    time.sleep(sleep_time_for_fetch_thread)
+
+    wait_for_all_checks(client)
     res = client.get(url_for("watchlist.index"))
     assert b'has-unread-changes' in res.data
 

changedetectionio/tests/test_watch_edited_flag.py

@@ -0,0 +1,246 @@
+#!/usr/bin/env python3
+"""
+Test the watch edited flag functionality.
+
+This tests the private __watch_was_edited flag that tracks when writable
+watch fields are modified, which prevents skipping reprocessing when the
+watch configuration has changed.
+"""
+
+import os
+import time
+from flask import url_for
+from .util import live_server_setup, wait_for_all_checks
+
+
+def set_test_content(datastore_path):
+    """Write test HTML content to endpoint-content.txt for test server."""
+    test_html = """<html>
+     <body>
+     <p>Test content for watch edited flag tests</p>
+     <p>This content stays the same across checks</p>
+     </body>
+     </html>
+    """
+    with open(os.path.join(datastore_path, "endpoint-content.txt"), "w") as f:
+        f.write(test_html)
+
+
+def test_watch_edited_flag_lifecycle(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test the full lifecycle of the was_edited flag:
+    1. Flag starts False when watch is created
+    2. Flag becomes True when writable fields are modified
+    3. Flag is reset False after worker processing
+    4. Flag stays False when readonly fields are modified
+    """
+
+    # Setup - Add a watch
+    test_url = url_for('test_endpoint', _external=True)
+    res = client.post(
+        url_for("ui.ui_views.form_quick_watch_add"),
+        data={"url": test_url, "tags": "", "edit_and_watch_submit_button": "Edit > Watch"},
+        follow_redirects=True
+    )
+    assert b"Watch added" in res.data or b"Updated watch" in res.data
+
+    # Get the watch UUID
+    datastore = client.application.config.get('DATASTORE')
+    uuid = list(datastore.data['watching'].keys())[0]
+    watch = datastore.data['watching'][uuid]
+
+    # Reset flag after initial form submission (form sets fields which trigger the flag)
+    watch.reset_watch_edited_flag()
+
+    # Test 1: Flag should be False after reset
+    assert not watch.was_edited, "Flag should be False after reset"
+
+    # Test 2: Modify a writable field (title) - flag should become True
+    watch['title'] = 'New Title'
+    assert watch.was_edited, "Flag should be True after modifying writable field 'title'"
+
+    # Test 3: Reset flag manually (simulating what worker does)
+    watch.reset_watch_edited_flag()
+    assert not watch.was_edited, "Flag should be False after reset"
+
+    # Test 4: Modify another writable field (url) - flag should become True again
+    watch['url'] = 'https://example.com'
+    assert watch.was_edited, "Flag should be True after modifying writable field 'url'"
+
+    # Test 5: Reset and modify a readonly field - flag should stay False
+    watch.reset_watch_edited_flag()
+    assert not watch.was_edited, "Flag should be False after reset"
+
+    # Modify readonly field (uuid) - should not set flag
+    old_uuid = watch['uuid']
+    watch['uuid'] = 'readonly-test-uuid'
+    assert not watch.was_edited, "Flag should stay False when modifying readonly field 'uuid'"
+    watch['uuid'] = old_uuid  # Restore original
+
+    # Note: Worker reset behavior is tested in test_check_removed_line_contains_trigger
+    # and test_watch_edited_flag_prevents_skip
+
+    print("✓ All watch edited flag lifecycle tests passed")
+
+
+def test_watch_edited_flag_dict_methods(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that the flag is set correctly by various dict methods:
+    - __setitem__ (watch['key'] = value)
+    - update() (watch.update({'key': value}))
+    - setdefault() (watch.setdefault('key', default))
+    - pop() (watch.pop('key'))
+    - __delitem__ (del watch['key'])
+    """
+
+    # Setup - Add a watch
+    test_url = url_for('test_endpoint', _external=True)
+    res = client.post(
+        url_for("ui.ui_views.form_quick_watch_add"),
+        data={"url": test_url, "tags": "", "edit_and_watch_submit_button": "Edit > Watch"},
+        follow_redirects=True
+    )
+
+    datastore = client.application.config.get('DATASTORE')
+    uuid = list(datastore.data['watching'].keys())[0]
+    watch = datastore.data['watching'][uuid]
+
+    # Test __setitem__
+    watch.reset_watch_edited_flag()
+    watch['title'] = 'Test via setitem'
+    assert watch.was_edited, "Flag should be True after __setitem__ on writable field"
+
+    # Test update() with dict
+    watch.reset_watch_edited_flag()
+    watch.update({'title': 'Test via update dict'})
+    assert watch.was_edited, "Flag should be True after update() with writable field"
+
+    # Test update() with kwargs
+    watch.reset_watch_edited_flag()
+    watch.update(title='Test via update kwargs')
+    assert watch.was_edited, "Flag should be True after update() kwargs with writable field"
+
+    # Test setdefault() on new key
+    watch.reset_watch_edited_flag()
+    watch.setdefault('title', 'Should not be set')  # Key exists, no change
+    assert not watch.was_edited, "Flag should stay False when setdefault() doesn't change existing key"
+
+    watch.setdefault('custom_field', 'New value')  # New key
+    assert watch.was_edited, "Flag should be True after setdefault() creates new writable field"
+
+    # Test pop() on writable field
+    watch.reset_watch_edited_flag()
+    watch.pop('custom_field', None)
+    assert watch.was_edited, "Flag should be True after pop() on writable field"
+
+    # Test __delitem__ on writable field
+    watch.reset_watch_edited_flag()
+    watch['temp_field'] = 'temp'
+    watch.reset_watch_edited_flag()  # Reset after adding
+    del watch['temp_field']
+    assert watch.was_edited, "Flag should be True after __delitem__ on writable field"
+
+    print("✓ All dict methods correctly set the flag")
+
+
+def test_watch_edited_flag_prevents_skip(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that the was_edited flag prevents skipping reprocessing.
+    When watch configuration is edited, it should reprocess even if content unchanged.
+    After worker processing, flag should be reset and subsequent checks can skip.
+    """
+
+    # Setup test content
+    set_test_content(datastore_path)
+
+    # Setup - Add a watch
+    test_url = url_for('test_endpoint', _external=True)
+    res = client.post(
+        url_for("ui.ui_views.form_quick_watch_add"),
+        data={"url": test_url, "tags": "", "edit_and_watch_submit_button": "Edit > Watch"},
+        follow_redirects=True
+    )
+    assert b"Watch added" in res.data or b"Updated watch" in res.data
+
+    datastore = client.application.config.get('DATASTORE')
+    uuid = list(datastore.data['watching'].keys())[0]
+    watch = datastore.data['watching'][uuid]
+
+    # Unpause the watch (watches are paused by default in tests)
+    watch['paused'] = False
+
+    # Run first check to establish baseline
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Verify first check completed successfully - checksum file should exist
+    checksum_file = os.path.join(datastore_path, uuid, 'last-checksum.txt')
+    assert os.path.isfile(checksum_file), "First check should create last-checksum.txt file"
+
+    # Reset the was_edited flag (simulating clean state after processing)
+    watch.reset_watch_edited_flag()
+    assert not watch.was_edited, "Flag should be False after reset"
+
+    # Run second check without any changes - should skip via checksumFromPreviousCheckWasTheSame
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # Verify it was skipped (last_check_status should indicate skip)
+    # Note: The actual skip is tested in test_check_removed_line_contains_trigger
+    # Here we're focused on the was_edited flag interaction
+
+    # Now modify the watch - flag should become True
+    watch['title'] = 'Modified Title'
+    assert watch.was_edited, "Flag should be True after modifying watch"
+
+    # Run third check - should NOT skip because was_edited=True even though content unchanged
+    client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    wait_for_all_checks(client)
+
+    # After worker processing, the flag should be reset by the worker
+    # This reset happens in the processor's run() method after processing completes
+    assert not watch.was_edited, "Flag should be False after worker processing"
+
+    print("✓ was_edited flag correctly prevents skip and is reset by worker")
+
+
+def test_watch_edited_flag_system_fields(client, live_server, measure_memory_usage, datastore_path):
+    """
+    Test that system fields (readonly + additional system fields) don't trigger the flag.
+    """
+
+    # Setup - Add a watch
+    test_url = url_for('test_endpoint', _external=True)
+    res = client.post(
+        url_for("ui.ui_views.form_quick_watch_add"),
+        data={"url": test_url, "tags": "", "edit_and_watch_submit_button": "Edit > Watch"},
+        follow_redirects=True
+    )
+
+    datastore = client.application.config.get('DATASTORE')
+    uuid = list(datastore.data['watching'].keys())[0]
+    watch = datastore.data['watching'][uuid]
+
+    # Test readonly fields from OpenAPI spec
+    readonly_fields = ['uuid', 'date_created', 'last_viewed']
+    for field in readonly_fields:
+        watch.reset_watch_edited_flag()
+        if field in watch:
+            old_value = watch[field]
+            watch[field] = 'modified-readonly-value'
+            assert not watch.was_edited, f"Flag should stay False when modifying readonly field '{field}'"
+            watch[field] = old_value  # Restore
+
+    # Test additional system fields not in OpenAPI spec yet
+    system_fields = ['last_check_status']
+    for field in system_fields:
+        watch.reset_watch_edited_flag()
+        watch[field] = 'system-value'
+        assert not watch.was_edited, f"Flag should stay False when modifying system field '{field}'"
+
+    # Test that content-type (readonly per OpenAPI) doesn't trigger flag
+    watch.reset_watch_edited_flag()
+    watch['content-type'] = 'text/html'
+    assert not watch.was_edited, "Flag should stay False when modifying 'content-type' (readonly)"
+
+    print("✓ System fields correctly don't trigger the flag")