Small safety catch

MANIFEST.in

@@ -1,9 +1,9 @@
 recursive-include changedetectionio/api *
-recursive-include changedetectionio/apprise_plugin *
 recursive-include changedetectionio/blueprint *
 recursive-include changedetectionio/content_fetchers *
 recursive-include changedetectionio/conditions *
 recursive-include changedetectionio/model *
+recursive-include changedetectionio/notification *
 recursive-include changedetectionio/processors *
 recursive-include changedetectionio/static *
 recursive-include changedetectionio/templates *

README.md

@@ -105,6 +105,12 @@ We [recommend and use Bright Data](https://brightdata.grsm.io/n0r16zf7eivq) glob
 
 Please :star: star :star: this project and help it grow! https://github.com/dgtlmoon/changedetection.io/
 
+### Conditional web page changes
+
+Easily [configure conditional actions](https://changedetection.io/tutorial/conditional-actions-web-page-changes), for example, only trigger when a price is above or below a preset amount, or [when a web page includes (or does not include) a keyword](https://changedetection.io/tutorial/how-monitor-keywords-any-website)
+
+<img src="./docs/web-page-change-conditions.png" style="max-width:80%;" alt="Conditional web page changes"  title="Conditional web page changes"  />
+
 ### Schedule web page watches in any timezone, limit by day of week and time.
 
 Easily set a re-check schedule, for example you could limit the web page change detection to only operate during business hours.

changedetectionio/__init__.py

@@ -2,7 +2,7 @@
 
 # Read more https://github.com/dgtlmoon/changedetection.io/wiki
 
-__version__ = '0.49.12'
+__version__ = '0.49.13'
 
 from changedetectionio.strtobool import strtobool
 from json.decoder import JSONDecodeError

changedetectionio/api/Notifications.py

@@ -0,0 +1,145 @@
+from flask_expects_json import expects_json
+from flask_restful import Resource
+from . import auth
+from flask_restful import abort, Resource
+from flask import request
+from . import auth
+from . import schema_create_notification_urls, schema_delete_notification_urls
+
+class Notifications(Resource):
+    def __init__(self, **kwargs):
+        # datastore is a black box dependency
+        self.datastore = kwargs['datastore']
+
+    @auth.check_token
+    def get(self):
+        """
+        @api {get} /api/v1/notifications Return Notification URL List
+        @apiDescription Return the Notification URL List from the configuration
+        @apiExample {curl} Example usage:
+            curl http://localhost:5000/api/v1/notifications -H"x-api-key:813031b16330fe25e3780cf0325daa45"
+            HTTP/1.0 200
+            {
+                'notification_urls': ["notification-urls-list"]
+            }
+        @apiName Get
+        @apiGroup Notifications
+        """
+
+        notification_urls = self.datastore.data.get('settings', {}).get('application', {}).get('notification_urls', [])        
+
+        return {
+                'notification_urls': notification_urls,
+               }, 200
+    
+    @auth.check_token
+    @expects_json(schema_create_notification_urls)
+    def post(self):
+        """
+        @api {post} /api/v1/notifications Create Notification URLs
+        @apiDescription Add one or more notification URLs from the configuration
+        @apiExample {curl} Example usage:
+            curl http://localhost:5000/api/v1/notifications/batch -H"x-api-key:813031b16330fe25e3780cf0325daa45" -H "Content-Type: application/json" -d '{"notification_urls": ["url1", "url2"]}'
+        @apiName CreateBatch
+        @apiGroup Notifications
+        @apiSuccess (201) {Object[]} notification_urls List of added notification URLs
+        @apiError (400) {String} Invalid input
+        """
+
+        json_data = request.get_json()
+        notification_urls = json_data.get("notification_urls", [])
+
+        from wtforms import ValidationError
+        try:
+            validate_notification_urls(notification_urls)
+        except ValidationError as e:
+            return str(e), 400
+
+        added_urls = []
+
+        for url in notification_urls:
+            clean_url = url.strip()
+            added_url = self.datastore.add_notification_url(clean_url)
+            if added_url:
+                added_urls.append(added_url)
+
+        if not added_urls:
+            return "No valid notification URLs were added", 400
+
+        return {'notification_urls': added_urls}, 201
+    
+    @auth.check_token
+    @expects_json(schema_create_notification_urls)
+    def put(self):
+        """
+        @api {put} /api/v1/notifications Replace Notification URLs
+        @apiDescription Replace all notification URLs with the provided list (can be empty)
+        @apiExample {curl} Example usage:
+            curl -X PUT http://localhost:5000/api/v1/notifications -H"x-api-key:813031b16330fe25e3780cf0325daa45" -H "Content-Type: application/json" -d '{"notification_urls": ["url1", "url2"]}'
+        @apiName Replace
+        @apiGroup Notifications
+        @apiSuccess (200) {Object[]} notification_urls List of current notification URLs
+        @apiError (400) {String} Invalid input
+        """
+        json_data = request.get_json()
+        notification_urls = json_data.get("notification_urls", [])
+
+        from wtforms import ValidationError
+        try:
+            validate_notification_urls(notification_urls)
+        except ValidationError as e:
+            return str(e), 400
+        
+        if not isinstance(notification_urls, list):
+            return "Invalid input format", 400
+
+        clean_urls = [url.strip() for url in notification_urls if isinstance(url, str)]
+        self.datastore.data['settings']['application']['notification_urls'] = clean_urls
+        self.datastore.needs_write = True
+
+        return {'notification_urls': clean_urls}, 200
+        
+    @auth.check_token
+    @expects_json(schema_delete_notification_urls)
+    def delete(self):
+        """
+        @api {delete} /api/v1/notifications Delete Notification URLs
+        @apiDescription Deletes one or more notification URLs from the configuration
+        @apiExample {curl} Example usage:
+            curl http://localhost:5000/api/v1/notifications -X DELETE -H"x-api-key:813031b16330fe25e3780cf0325daa45" -H "Content-Type: application/json" -d '{"notification_urls": ["url1", "url2"]}'
+        @apiParam {String[]} notification_urls The notification URLs to delete.
+        @apiName Delete
+        @apiGroup Notifications
+        @apiSuccess (204) {String} OK Deleted
+        @apiError (400) {String} No matching notification URLs found.
+        """
+
+        json_data = request.get_json()
+        urls_to_delete = json_data.get("notification_urls", [])
+        if not isinstance(urls_to_delete, list):
+            abort(400, message="Expected a list of notification URLs.")
+
+        notification_urls = self.datastore.data['settings']['application'].get('notification_urls', [])
+        deleted = []
+
+        for url in urls_to_delete:
+            clean_url = url.strip()
+            if clean_url in notification_urls:
+                notification_urls.remove(clean_url)
+                deleted.append(clean_url)
+
+        if not deleted:
+            abort(400, message="No matching notification URLs found.")
+
+        self.datastore.data['settings']['application']['notification_urls'] = notification_urls
+        self.datastore.needs_write = True
+
+        return 'OK', 204
+    
+def validate_notification_urls(notification_urls):
+    from changedetectionio.forms import ValidateAppRiseServers
+    validator = ValidateAppRiseServers()
+    class DummyForm: pass
+    dummy_form = DummyForm()
+    field = type("Field", (object,), {"data": notification_urls, "gettext": lambda self, x: x})()
+    validator(dummy_form, field)

changedetectionio/api/__init__.py

@@ -19,8 +19,15 @@ schema_create_tag['required'] = ['title']
 schema_update_tag = copy.deepcopy(schema_tag)
 schema_update_tag['additionalProperties'] = False
 
+schema_notification_urls = copy.deepcopy(schema)
+schema_create_notification_urls = copy.deepcopy(schema_notification_urls)
+schema_create_notification_urls['required'] = ['notification_urls']
+schema_delete_notification_urls = copy.deepcopy(schema_notification_urls)
+schema_delete_notification_urls['required'] = ['notification_urls']
+
 # Import all API resources
 from .Watch import Watch, WatchHistory, WatchSingleHistory, CreateWatch
 from .Tags import Tags, Tag
 from .Import import Import
 from .SystemInfo import SystemInfo
+from .Notifications import Notifications

changedetectionio/api/api_schema.py

@@ -1,5 +1,7 @@
 # Responsible for building the storage dict into a set of rules ("JSON Schema") acceptable via the API
 # Probably other ways to solve this when the backend switches to some ORM
+from changedetectionio.notification import valid_notification_formats
+
 
 def build_time_between_check_json_schema():
     # Setup time between check schema
@@ -98,8 +100,6 @@ def build_watch_json_schema(d):
         }
     }
 
-    from changedetectionio.notification import valid_notification_formats
-
     schema['properties']['notification_format'] = {'type': 'string',
                                                    'enum': list(valid_notification_formats.keys())
                                                    }

changedetectionio/blueprint/browser_steps/browser_steps.py

@@ -4,7 +4,7 @@ import re
 from random import randint
 from loguru import logger
 
-from changedetectionio.content_fetchers.helpers import capture_stitched_together_full_page, SCREENSHOT_SIZE_STITCH_THRESHOLD
+from changedetectionio.content_fetchers import SCREENSHOT_MAX_HEIGHT_DEFAULT
 from changedetectionio.content_fetchers.base import manage_user_agent
 from changedetectionio.safe_jinja import render as jinja_render
 
@@ -293,19 +293,16 @@ class browsersteps_live_ui(steppable_browser_interface):
     def get_current_state(self):
         """Return the screenshot and interactive elements mapping, generally always called after action_()"""
         import importlib.resources
+        import json
+        # because we for now only run browser steps in playwright mode (not puppeteer mode)
+        from changedetectionio.content_fetchers.playwright import capture_full_page
+
         xpath_element_js = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('xpath_element_scraper.js').read_text()
 
         now = time.time()
         self.page.wait_for_timeout(1 * 1000)
 
-
-        full_height = self.page.evaluate("document.documentElement.scrollHeight")
-
-        if full_height >= SCREENSHOT_SIZE_STITCH_THRESHOLD:
-            logger.warning(f"Page full Height: {full_height}px longer than {SCREENSHOT_SIZE_STITCH_THRESHOLD}px, using 'stitched screenshot method'.")
-            screenshot = capture_stitched_together_full_page(self.page)
-        else:
-            screenshot = self.page.screenshot(type='jpeg', full_page=True, quality=40)
+        screenshot = capture_full_page(page=self.page)
 
         logger.debug(f"Time to get screenshot from browser {time.time() - now:.2f}s")
 
@@ -313,13 +310,21 @@ class browsersteps_live_ui(steppable_browser_interface):
         self.page.evaluate("var include_filters=''")
         # Go find the interactive elements
         # @todo in the future, something smarter that can scan for elements with .click/focus etc event handlers?
-        elements = 'a,button,input,select,textarea,i,th,td,p,li,h1,h2,h3,h4,div,span'
-        xpath_element_js = xpath_element_js.replace('%ELEMENTS%', elements)
 
-        xpath_data = self.page.evaluate("async () => {" + xpath_element_js + "}")
+        self.page.request_gc()
+
+        scan_elements = 'a,button,input,select,textarea,i,th,td,p,li,h1,h2,h3,h4,div,span'
+
+        MAX_TOTAL_HEIGHT = int(os.getenv("SCREENSHOT_MAX_HEIGHT", SCREENSHOT_MAX_HEIGHT_DEFAULT))
+        xpath_data = json.loads(self.page.evaluate(xpath_element_js, {
+            "visualselector_xpath_selectors": scan_elements,
+            "max_height": MAX_TOTAL_HEIGHT
+        }))
+        self.page.request_gc()
+
         # So the JS will find the smallest one first
         xpath_data['size_pos'] = sorted(xpath_data['size_pos'], key=lambda k: k['width'] * k['height'], reverse=True)
-        logger.debug(f"Time to scrape xpath element data in browser {time.time()-now:.2f}s")
+        logger.debug(f"Time to scrape xPath element data in browser {time.time()-now:.2f}s")
 
         # playwright._impl._api_types.Error: Browser closed.
         # @todo show some countdown timer?

changedetectionio/blueprint/settings/templates/settings.html

@@ -22,6 +22,7 @@
             <li class="tab"><a href="#notifications">Notifications</a></li>
             <li class="tab"><a href="#fetching">Fetching</a></li>
             <li class="tab"><a href="#filters">Global Filters</a></li>
+            <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="#timedate">Time &amp Date</a></li>
             <li class="tab"><a href="#proxies">CAPTCHA &amp; Proxies</a></li>
@@ -240,6 +241,12 @@ nav
                     </p>
                 </div>
             </div>
+            <div class="tab-pane-inner" id="ui-options">
+                <div class="pure-control-group">
+                    {{ render_checkbox_field(form.application.form.ui.form.open_diff_in_new_tab, class="open_diff_in_new_tab") }}
+                    <span class="pure-form-message-inline">Enable this setting to open the diff page in a new tab. If disabled, the diff page will open in the current tab.</span>
+                </div>
+            </div>
             <div class="tab-pane-inner" id="proxies">
                 <div id="recommended-proxy">
                     <div>

changedetectionio/blueprint/ui/edit.py

@@ -19,6 +19,20 @@ 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
 
+    def levenshtein_ratio_recent_history(watch):
+        try:
+            from Levenshtein import ratio, distance
+            k = list(watch.history.keys())
+            if len(k) >= 2:
+                a = watch.get_history_snapshot(timestamp=k[0])
+                b = watch.get_history_snapshot(timestamp=k[1])
+                distance = distance(a, b)
+                return distance
+        except Exception as e:
+            logger.warning("Unable to calc similarity", e)
+            return "Unable to calc similarity"
+        return ''
+
     @edit_blueprint.route("/edit/<string:uuid>", methods=['GET', 'POST'])
     @login_optionally_required
     # https://stackoverflow.com/questions/42984453/wtforms-populate-form-with-data-if-data-exists
@@ -247,14 +261,15 @@ def construct_blueprint(datastore: ChangeDetectionStore, update_q, queuedWatchMe
                 'has_default_notification_urls': True if len(datastore.data['settings']['application']['notification_urls']) else False,
                 'has_extra_headers_file': len(datastore.get_all_headers_in_textfile_for_watch(uuid=uuid)) > 0,
                 'has_special_tag_options': _watch_has_tag_options_set(watch=watch),
-                'watch_uses_webdriver': watch_uses_webdriver,
                 'jq_support': jq_support,
+                'lev_info': levenshtein_ratio_recent_history(watch),
                 'playwright_enabled': os.getenv('PLAYWRIGHT_DRIVER_URL', False),
                 'settings_application': datastore.data['settings']['application'],
                 'timezone_default_config': datastore.data['settings']['application'].get('timezone'),
                 'using_global_webdriver_wait': not default['webdriver_delay'],
                 'uuid': uuid,
-                'watch': watch
+                'watch': watch,
+                'watch_uses_webdriver': watch_uses_webdriver,
             }
 
             included_content = None

changedetectionio/blueprint/ui/notification.py

@@ -4,7 +4,6 @@ from loguru import logger
 
 from changedetectionio.store import ChangeDetectionStore
 from changedetectionio.auth_decorator import login_optionally_required
-from changedetectionio.notification import process_notification
 
 def construct_blueprint(datastore: ChangeDetectionStore):
     notification_blueprint = Blueprint('ui_notification', __name__, template_folder="../ui/templates")
@@ -18,8 +17,11 @@ def construct_blueprint(datastore: ChangeDetectionStore):
 
         # Watch_uuid could be unset in the case it`s used in tag editor, global settings
         import apprise
-        from ...apprise_plugin.assets import apprise_asset
-        from ...apprise_plugin.custom_handlers import apprise_http_custom_handler  # noqa: F401
+        from changedetectionio.notification.handler import process_notification
+        from changedetectionio.notification.apprise_plugin.assets import apprise_asset
+
+        from changedetectionio.notification.apprise_plugin.custom_handlers import apprise_http_custom_handler
+
         apobj = apprise.Apprise(asset=apprise_asset)
 
         is_global_settings_form = request.args.get('mode', '') == 'global-settings'

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

@@ -209,15 +209,18 @@
                     <a href="{{ url_for('ui.ui_edit.edit_page', uuid=watch.uuid, tag=active_tag_uuid)}}#general" class="pure-button pure-button-primary">Edit</a>
                     {% if watch.history_n >= 2 %}
 
+                        {% set open_diff_in_new_tab = datastore.data['settings']['application']['ui'].get('open_diff_in_new_tab') %}
+                        {% set target_attr = ' target="' ~ watch.uuid ~ '"' if open_diff_in_new_tab else '' %}
+
                         {%  if is_unviewed %}
-                           <a href="{{ url_for('ui.ui_views.diff_history_page', uuid=watch.uuid, from_version=watch.get_from_version_based_on_last_viewed) }}" target="{{watch.uuid}}" class="pure-button pure-button-primary diff-link">History</a>
+                           <a href="{{ url_for('ui.ui_views.diff_history_page', uuid=watch.uuid, from_version=watch.get_from_version_based_on_last_viewed) }}" {{target_attr}} class="pure-button pure-button-primary diff-link">History</a>
                         {% else %}
-                           <a href="{{ url_for('ui.ui_views.diff_history_page', uuid=watch.uuid)}}" target="{{watch.uuid}}" class="pure-button pure-button-primary diff-link">History</a>
+                           <a href="{{ url_for('ui.ui_views.diff_history_page', uuid=watch.uuid)}}" {{target_attr}} class="pure-button pure-button-primary diff-link">History</a>
                         {% endif %}
 
                     {% else %}
                         {% if watch.history_n == 1 or (watch.history_n ==0 and watch.error_text_ctime )%}
-                            <a href="{{ url_for('ui.ui_views.preview_page', uuid=watch.uuid)}}" target="{{watch.uuid}}" class="pure-button pure-button-primary">Preview</a>
+                            <a href="{{ url_for('ui.ui_views.preview_page', uuid=watch.uuid)}}" {{target_attr}} class="pure-button pure-button-primary">Preview</a>
                         {% endif %}
                     {% endif %}
                 </td>

changedetectionio/conditions/__init__.py

@@ -96,7 +96,7 @@ def execute_ruleset_against_all_plugins(current_watch_uuid: str, application_dat
     
     ruleset_settings = application_datastruct['watching'].get(current_watch_uuid)
 
-    if ruleset_settings.get("conditions"):
+    if ruleset_settings and ruleset_settings.get("conditions"):
         logic_operator = "and" if ruleset_settings.get("conditions_match_logic", "ALL") == "ALL" else "or"
         complete_rules = filter_complete_rules(ruleset_settings['conditions'])
         if complete_rules:

changedetectionio/content_fetchers/__init__.py

@@ -7,11 +7,29 @@ import os
 # Visual Selector scraper - 'Button' is there because some sites have <button>OUT OF STOCK</button>.
 visualselector_xpath_selectors = 'div,span,form,table,tbody,tr,td,a,p,ul,li,h1,h2,h3,h4,header,footer,section,article,aside,details,main,nav,section,summary,button'
 
+SCREENSHOT_MAX_HEIGHT_DEFAULT = 20000
+SCREENSHOT_DEFAULT_QUALITY = 40
+
+# Maximum total height for the final image (When in stitch mode).
+# We limit this to 16000px due to the huge amount of RAM that was being used
+# Example: 16000 × 1400 × 3 = 67,200,000 bytes ≈ 64.1 MB (not including buffers in PIL etc)
+SCREENSHOT_MAX_TOTAL_HEIGHT = int(os.getenv("SCREENSHOT_MAX_HEIGHT", SCREENSHOT_MAX_HEIGHT_DEFAULT))
+
+# The size at which we will switch to stitching method, when below this (and
+# MAX_TOTAL_HEIGHT which can be set by a user) we will use the default
+# screenshot method.
+SCREENSHOT_SIZE_STITCH_THRESHOLD = 8000
 
 # available_fetchers() will scan this implementation looking for anything starting with html_
 # this information is used in the form selections
 from changedetectionio.content_fetchers.requests import fetcher as html_requests
 
+
+import importlib.resources
+XPATH_ELEMENT_JS = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('xpath_element_scraper.js').read_text(encoding='utf-8')
+INSTOCK_DATA_JS = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('stock-not-in-stock.js').read_text(encoding='utf-8')
+
+
 def available_fetchers():
     # See the if statement at the bottom of this file for how we switch between playwright and webdriver
     import inspect

changedetectionio/content_fetchers/base.py

@@ -63,11 +63,6 @@ class Fetcher():
     # Time ONTOP of the system defined env minimum time
     render_extract_delay = 0
 
-    def __init__(self):
-        import importlib.resources
-        self.xpath_element_js = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('xpath_element_scraper.js').read_text(encoding='utf-8')
-        self.instock_data_js = importlib.resources.files("changedetectionio.content_fetchers.res").joinpath('stock-not-in-stock.js').read_text(encoding='utf-8')
-
     @abstractmethod
     def get_error(self):
         return self.error
@@ -87,7 +82,7 @@ class Fetcher():
         pass
 
     @abstractmethod
-    def quit(self):
+    def quit(self, watch=None):
         return
 
     @abstractmethod
@@ -143,6 +138,7 @@ class Fetcher():
                 logger.debug(f">> Iterating check - browser Step n {step_n} - {step['operation']}...")
                 self.screenshot_step("before-" + str(step_n))
                 self.save_step_html("before-" + str(step_n))
+
                 try:
                     optional_value = step['optional_value']
                     selector = step['selector']

changedetectionio/content_fetchers/helpers.py

@@ -1,104 +0,0 @@
-
-# Pages with a vertical height longer than this will use the 'stitch together' method.
-
-# - Many GPUs have a max texture size of 16384x16384px (or lower on older devices).
-# - If a page is taller than ~8000–10000px, it risks exceeding GPU memory limits.
-# - This is especially important on headless Chromium, where Playwright may fail to allocate a massive full-page buffer.
-
-
-# The size at which we will switch to stitching method
-SCREENSHOT_SIZE_STITCH_THRESHOLD=8000
-
-from loguru import logger
-
-def capture_stitched_together_full_page(page):
-    import io
-    import os
-    import time
-    from PIL import Image, ImageDraw, ImageFont
-
-    MAX_TOTAL_HEIGHT = SCREENSHOT_SIZE_STITCH_THRESHOLD*4  # Maximum total height for the final image (When in stitch mode)
-    MAX_CHUNK_HEIGHT = 4000  # Height per screenshot chunk
-    WARNING_TEXT_HEIGHT = 20  # Height of the warning text overlay
-
-    # Save the original viewport size
-    original_viewport = page.viewport_size
-    now = time.time()
-
-    try:
-        viewport = page.viewport_size
-        page_height = page.evaluate("document.documentElement.scrollHeight")
-
-        # Limit the total capture height
-        capture_height = min(page_height, MAX_TOTAL_HEIGHT)
-
-        images = []
-        total_captured_height = 0
-
-        for offset in range(0, capture_height, MAX_CHUNK_HEIGHT):
-            # Ensure we do not exceed the total height limit
-            chunk_height = min(MAX_CHUNK_HEIGHT, MAX_TOTAL_HEIGHT - total_captured_height)
-
-            # Adjust viewport size for this chunk
-            page.set_viewport_size({"width": viewport["width"], "height": chunk_height})
-
-            # Scroll to the correct position
-            page.evaluate(f"window.scrollTo(0, {offset})")
-
-            # Capture screenshot chunk
-            screenshot_bytes = page.screenshot(type='jpeg', quality=int(os.getenv("SCREENSHOT_QUALITY", 30)))
-            images.append(Image.open(io.BytesIO(screenshot_bytes)))
-
-            total_captured_height += chunk_height
-
-            # Stop if we reached the maximum total height
-            if total_captured_height >= MAX_TOTAL_HEIGHT:
-                break
-
-        # Create the final stitched image
-        stitched_image = Image.new('RGB', (viewport["width"], total_captured_height))
-        y_offset = 0
-
-        # Stitch the screenshot chunks together
-        for img in images:
-            stitched_image.paste(img, (0, y_offset))
-            y_offset += img.height
-
-        logger.debug(f"Screenshot stitched together in {time.time()-now:.2f}s")
-
-        # Overlay warning text if the screenshot was trimmed
-        if page_height > MAX_TOTAL_HEIGHT:
-            draw = ImageDraw.Draw(stitched_image)
-            warning_text = f"WARNING: Screenshot was {page_height}px but trimmed to {MAX_TOTAL_HEIGHT}px because it was too long"
-
-            # Load font (default system font if Arial is unavailable)
-            try:
-                font = ImageFont.truetype("arial.ttf", WARNING_TEXT_HEIGHT)  # Arial (Windows/Mac)
-            except IOError:
-                font = ImageFont.load_default()  # Default font if Arial not found
-
-            # Get text bounding box (correct method for newer Pillow versions)
-            text_bbox = draw.textbbox((0, 0), warning_text, font=font)
-            text_width = text_bbox[2] - text_bbox[0]  # Calculate text width
-            text_height = text_bbox[3] - text_bbox[1]  # Calculate text height
-
-            # Define background rectangle (top of the image)
-            draw.rectangle([(0, 0), (viewport["width"], WARNING_TEXT_HEIGHT)], fill="white")
-
-            # Center text horizontally within the warning area
-            text_x = (viewport["width"] - text_width) // 2
-            text_y = (WARNING_TEXT_HEIGHT - text_height) // 2
-
-            # Draw the warning text in red
-            draw.text((text_x, text_y), warning_text, fill="red", font=font)
-
-        # Save or return the final image
-        output = io.BytesIO()
-        stitched_image.save(output, format="JPEG", quality=int(os.getenv("SCREENSHOT_QUALITY", 30)))
-        screenshot = output.getvalue()
-
-    finally:
-        # Restore the original viewport size
-        page.set_viewport_size(original_viewport)
-
-    return screenshot

changedetectionio/content_fetchers/playwright.py

@@ -4,10 +4,76 @@ from urllib.parse import urlparse
 
 from loguru import logger
 
-from changedetectionio.content_fetchers.helpers import capture_stitched_together_full_page, SCREENSHOT_SIZE_STITCH_THRESHOLD
+from changedetectionio.content_fetchers import SCREENSHOT_MAX_HEIGHT_DEFAULT, visualselector_xpath_selectors, \
+    SCREENSHOT_SIZE_STITCH_THRESHOLD, SCREENSHOT_MAX_TOTAL_HEIGHT, XPATH_ELEMENT_JS, INSTOCK_DATA_JS
 from changedetectionio.content_fetchers.base import Fetcher, manage_user_agent
 from changedetectionio.content_fetchers.exceptions import PageUnloadable, Non200ErrorCodeReceived, EmptyReply, ScreenshotUnavailable
 
+def capture_full_page(page):
+    import os
+    import time
+    from multiprocessing import Process, Pipe
+
+    start = time.time()
+
+    page_height = page.evaluate("document.documentElement.scrollHeight")
+    page_width = page.evaluate("document.documentElement.scrollWidth")
+    original_viewport = page.viewport_size
+
+    logger.debug(f"Playwright viewport size {page.viewport_size} page height {page_height} page width {page_width}")
+
+    # Use an approach similar to puppeteer: set a larger viewport and take screenshots in chunks
+    step_size = SCREENSHOT_SIZE_STITCH_THRESHOLD # Size that won't cause GPU to overflow
+    screenshot_chunks = []
+    y = 0
+
+    if page_height > page.viewport_size['height']:
+        if page_height < step_size:
+            step_size = page_height # Incase page is bigger than default viewport but smaller than proposed step size
+        logger.debug(f"Setting bigger viewport to step through large page width W{page.viewport_size['width']}xH{step_size} because page_height > viewport_size")
+        # Set viewport to a larger size to capture more content at once
+        page.set_viewport_size({'width': page.viewport_size['width'], 'height': step_size})
+
+    # Capture screenshots in chunks up to the max total height
+    while y < min(page_height, SCREENSHOT_MAX_TOTAL_HEIGHT):
+        page.request_gc()
+        page.evaluate(f"window.scrollTo(0, {y})")
+        page.request_gc()
+        screenshot_chunks.append(page.screenshot(
+            type="jpeg",
+            full_page=False,
+            quality=int(os.getenv("SCREENSHOT_QUALITY", 72))
+        ))
+        y += step_size
+        page.request_gc()
+
+    # Restore original viewport size
+    page.set_viewport_size({'width': original_viewport['width'], 'height': original_viewport['height']})
+
+    # If we have multiple chunks, stitch them together
+    if len(screenshot_chunks) > 1:
+        from changedetectionio.content_fetchers.screenshot_handler import stitch_images_worker
+        logger.debug(f"Screenshot stitching {len(screenshot_chunks)} chunks together")
+        parent_conn, child_conn = Pipe()
+        p = Process(target=stitch_images_worker, args=(child_conn, screenshot_chunks, page_height, SCREENSHOT_MAX_TOTAL_HEIGHT))
+        p.start()
+        screenshot = parent_conn.recv_bytes()
+        p.join()
+        logger.debug(
+            f"Screenshot (chunked/stitched) - Page height: {page_height} Capture height: {SCREENSHOT_MAX_TOTAL_HEIGHT} - Stitched together in {time.time() - start:.2f}s")
+        # Explicit cleanup
+        del screenshot_chunks
+        del p
+        del parent_conn, child_conn
+        screenshot_chunks = None
+        return screenshot
+
+    logger.debug(
+        f"Screenshot Page height: {page_height} Capture height: {SCREENSHOT_MAX_TOTAL_HEIGHT} - Stitched together in {time.time() - start:.2f}s")
+
+    return screenshot_chunks[0]
+
+
 class fetcher(Fetcher):
     fetcher_description = "Playwright {}/Javascript".format(
         os.getenv("PLAYWRIGHT_BROWSER_TYPE", 'chromium').capitalize()
@@ -60,7 +126,8 @@ class fetcher(Fetcher):
 
     def screenshot_step(self, step_n=''):
         super().screenshot_step(step_n=step_n)
-        screenshot = self.page.screenshot(type='jpeg', full_page=True, quality=int(os.getenv("SCREENSHOT_QUALITY", 72)))
+        screenshot = capture_full_page(page=self.page)
+
 
         if self.browser_steps_screenshot_path is not None:
             destination = os.path.join(self.browser_steps_screenshot_path, 'step_{}.jpeg'.format(step_n))
@@ -89,7 +156,6 @@ class fetcher(Fetcher):
 
         from playwright.sync_api import sync_playwright
         import playwright._impl._errors
-        from changedetectionio.content_fetchers import visualselector_xpath_selectors
         import time
         self.delete_browser_steps_screenshots()
         response = None
@@ -164,9 +230,7 @@ class fetcher(Fetcher):
                 raise PageUnloadable(url=url, status_code=None, message=str(e))
 
             if self.status_code != 200 and not ignore_status_codes:
-                screenshot = self.page.screenshot(type='jpeg', full_page=True,
-                                                  quality=int(os.getenv("SCREENSHOT_QUALITY", 72)))
-
+                screenshot = capture_full_page(self.page)
                 raise Non200ErrorCodeReceived(url=url, status_code=self.status_code, screenshot=screenshot)
 
             if not empty_pages_are_a_change and len(self.page.content().strip()) == 0:
@@ -187,13 +251,23 @@ class fetcher(Fetcher):
                 self.page.evaluate("var include_filters={}".format(json.dumps(current_include_filters)))
             else:
                 self.page.evaluate("var include_filters=''")
+            self.page.request_gc()
 
-            self.xpath_data = self.page.evaluate(
-                "async () => {" + self.xpath_element_js.replace('%ELEMENTS%', visualselector_xpath_selectors) + "}")
-            self.instock_data = self.page.evaluate("async () => {" + self.instock_data_js + "}")
+            # request_gc before and after evaluate to free up memory
+            # @todo browsersteps etc
+            MAX_TOTAL_HEIGHT = int(os.getenv("SCREENSHOT_MAX_HEIGHT", SCREENSHOT_MAX_HEIGHT_DEFAULT))
+            self.xpath_data = self.page.evaluate(XPATH_ELEMENT_JS, {
+                "visualselector_xpath_selectors": visualselector_xpath_selectors,
+                "max_height": MAX_TOTAL_HEIGHT
+            })
+            self.page.request_gc()
+
+            self.instock_data = self.page.evaluate(INSTOCK_DATA_JS)
+            self.page.request_gc()
 
             self.content = self.page.content()
-            logger.debug(f"Time to scrape xpath element data in browser {time.time() - now:.2f}s")
+            self.page.request_gc()
+            logger.debug(f"Scrape xPath element data in browser done in {time.time() - now:.2f}s")
 
             # Bug 3 in Playwright screenshot handling
             # Some bug where it gives the wrong screenshot size, but making a request with the clip set first seems to solve it
@@ -204,18 +278,41 @@ class fetcher(Fetcher):
             # acceptable screenshot quality here
             try:
                 # The actual screenshot - this always base64 and needs decoding! horrible! huge CPU usage
-                full_height = self.page.evaluate("document.documentElement.scrollHeight")
-
-                if full_height >= SCREENSHOT_SIZE_STITCH_THRESHOLD:
-                    logger.warning(
-                        f"Page full Height: {full_height}px longer than {SCREENSHOT_SIZE_STITCH_THRESHOLD}px, using 'stitched screenshot method'.")
-                    self.screenshot = capture_stitched_together_full_page(self.page)
-                else:
-                    self.screenshot = self.page.screenshot(type='jpeg', full_page=True, quality=int(os.getenv("SCREENSHOT_QUALITY", 30)))
+                self.screenshot = capture_full_page(page=self.page)
 
             except Exception as e:
                 # It's likely the screenshot was too long/big and something crashed
                 raise ScreenshotUnavailable(url=url, status_code=self.status_code)
             finally:
-                context.close()
-                browser.close()
+                # Request garbage collection one more time before closing
+                try:
+                    self.page.request_gc()
+                except:
+                    pass
+                
+                # Clean up resources properly
+                try:
+                    self.page.request_gc()
+                except:
+                    pass
+
+                try:
+                    self.page.close()
+                except:
+                    pass
+                self.page = None
+
+                try:
+                    context.close()
+                except:
+                    pass
+                context = None
+
+                try:
+                    browser.close()
+                except:
+                    pass
+                browser = None
+
+
+

changedetectionio/content_fetchers/puppeteer.py

@@ -6,8 +6,77 @@ from urllib.parse import urlparse
 
 from loguru import logger
 
+from changedetectionio.content_fetchers import SCREENSHOT_MAX_HEIGHT_DEFAULT, visualselector_xpath_selectors, \
+    SCREENSHOT_SIZE_STITCH_THRESHOLD, SCREENSHOT_DEFAULT_QUALITY, XPATH_ELEMENT_JS, INSTOCK_DATA_JS, \
+    SCREENSHOT_MAX_TOTAL_HEIGHT
 from changedetectionio.content_fetchers.base import Fetcher, manage_user_agent
-from changedetectionio.content_fetchers.exceptions import PageUnloadable, Non200ErrorCodeReceived, EmptyReply, BrowserFetchTimedOut, BrowserConnectError
+from changedetectionio.content_fetchers.exceptions import PageUnloadable, Non200ErrorCodeReceived, EmptyReply, BrowserFetchTimedOut, \
+    BrowserConnectError
+
+
+# Bug 3 in Playwright screenshot handling
+# Some bug where it gives the wrong screenshot size, but making a request with the clip set first seems to solve it
+
+# Screenshots also travel via the ws:// (websocket) meaning that the binary data is base64 encoded
+# which will significantly increase the IO size between the server and client, it's recommended to use the lowest
+# acceptable screenshot quality here
+async def capture_full_page(page):
+    import os
+    import time
+    from multiprocessing import Process, Pipe
+
+    start = time.time()
+
+    page_height = await page.evaluate("document.documentElement.scrollHeight")
+    page_width = await page.evaluate("document.documentElement.scrollWidth")
+    original_viewport = page.viewport
+
+    logger.debug(f"Puppeteer viewport size {page.viewport} page height {page_height} page width {page_width}")
+
+    # Bug 3 in Playwright screenshot handling
+    # Some bug where it gives the wrong screenshot size, but making a request with the clip set first seems to solve it
+    # JPEG is better here because the screenshots can be very very large
+
+    # Screenshots also travel via the ws:// (websocket) meaning that the binary data is base64 encoded
+    # which will significantly increase the IO size between the server and client, it's recommended to use the lowest
+    # acceptable screenshot quality here
+
+
+    step_size = SCREENSHOT_SIZE_STITCH_THRESHOLD # Something that will not cause the GPU to overflow when taking the screenshot
+    screenshot_chunks = []
+    y = 0
+    if page_height > page.viewport['height']:
+        if page_height < step_size:
+            step_size = page_height # Incase page is bigger than default viewport but smaller than proposed step size
+        await page.setViewport({'width': page.viewport['width'], 'height': step_size})
+
+    while y < min(page_height, SCREENSHOT_MAX_TOTAL_HEIGHT):
+        await page.evaluate(f"window.scrollTo(0, {y})")
+        screenshot_chunks.append(await page.screenshot(type_='jpeg',
+                                                       fullPage=False,
+                                                       quality=int(os.getenv("SCREENSHOT_QUALITY", 72))))
+        y += step_size
+
+    await page.setViewport({'width': original_viewport['width'], 'height': original_viewport['height']})
+
+    if len(screenshot_chunks) > 1:
+        from changedetectionio.content_fetchers.screenshot_handler import stitch_images_worker
+        logger.debug(f"Screenshot stitching {len(screenshot_chunks)} chunks together")
+        parent_conn, child_conn = Pipe()
+        p = Process(target=stitch_images_worker, args=(child_conn, screenshot_chunks, page_height, SCREENSHOT_MAX_TOTAL_HEIGHT))
+        p.start()
+        screenshot = parent_conn.recv_bytes()
+        p.join()
+        logger.debug(
+            f"Screenshot (chunked/stitched) - Page height: {page_height} Capture height: {SCREENSHOT_MAX_TOTAL_HEIGHT} - Stitched together in {time.time() - start:.2f}s")
+
+        screenshot_chunks = None
+        return screenshot
+
+    logger.debug(
+        f"Screenshot Page height: {page_height} Capture height: {SCREENSHOT_MAX_TOTAL_HEIGHT} - Stitched together in {time.time() - start:.2f}s")
+    return screenshot_chunks[0]
+
 
 class fetcher(Fetcher):
     fetcher_description = "Puppeteer/direct {}/Javascript".format(
@@ -79,7 +148,6 @@ class fetcher(Fetcher):
                          empty_pages_are_a_change
                          ):
 
-        from changedetectionio.content_fetchers import visualselector_xpath_selectors
         self.delete_browser_steps_screenshots()
         extra_wait = int(os.getenv("WEBDRIVER_DELAY_BEFORE_CONTENT_READY", 5)) + self.render_extract_delay
 
@@ -181,11 +249,10 @@ class fetcher(Fetcher):
             raise PageUnloadable(url=url, status_code=None, message=str(e))
 
         if self.status_code != 200 and not ignore_status_codes:
-            screenshot = await self.page.screenshot(type_='jpeg',
-                                                    fullPage=True,
-                                                    quality=int(os.getenv("SCREENSHOT_QUALITY", 72)))
+            screenshot = await capture_full_page(page=self.page)
 
             raise Non200ErrorCodeReceived(url=url, status_code=self.status_code, screenshot=screenshot)
+
         content = await self.page.content
 
         if not empty_pages_are_a_change and len(content.strip()) == 0:
@@ -203,46 +270,31 @@ class fetcher(Fetcher):
 
         # So we can find an element on the page where its selector was entered manually (maybe not xPath etc)
         # Setup the xPath/VisualSelector scraper
-        if current_include_filters is not None:
+        if current_include_filters:
             js = json.dumps(current_include_filters)
             await self.page.evaluate(f"var include_filters={js}")
         else:
             await self.page.evaluate(f"var include_filters=''")
 
-        self.xpath_data = await self.page.evaluate(
-            "async () => {" + self.xpath_element_js.replace('%ELEMENTS%', visualselector_xpath_selectors) + "}")
-        self.instock_data = await self.page.evaluate("async () => {" + self.instock_data_js + "}")
+        MAX_TOTAL_HEIGHT = int(os.getenv("SCREENSHOT_MAX_HEIGHT", SCREENSHOT_MAX_HEIGHT_DEFAULT))
+        self.xpath_data = await self.page.evaluate(XPATH_ELEMENT_JS, {
+            "visualselector_xpath_selectors": visualselector_xpath_selectors,
+            "max_height": MAX_TOTAL_HEIGHT
+        })
+        if not self.xpath_data:
+            raise Exception(f"Content Fetcher > xPath scraper failed. Please report this URL so we can fix it :)")
+
+        self.instock_data = await self.page.evaluate(INSTOCK_DATA_JS)
 
         self.content = await self.page.content
-        # Bug 3 in Playwright screenshot handling
-        # Some bug where it gives the wrong screenshot size, but making a request with the clip set first seems to solve it
-        # JPEG is better here because the screenshots can be very very large
 
-        # Screenshots also travel via the ws:// (websocket) meaning that the binary data is base64 encoded
-        # which will significantly increase the IO size between the server and client, it's recommended to use the lowest
-        # acceptable screenshot quality here
-        try:
-            self.screenshot = await self.page.screenshot(type_='jpeg',
-                                                         fullPage=True,
-                                                         quality=int(os.getenv("SCREENSHOT_QUALITY", 72)))
-        except Exception as e:
-            logger.error("Error fetching screenshot")
-            # // May fail on very large pages with 'WARNING: tile memory limits exceeded, some content may not draw'
-            # // @ todo after text extract, we can place some overlay text with red background to say 'croppped'
-            logger.error('ERROR: content-fetcher page was maybe too large for a screenshot, reverting to viewport only screenshot')
-            try:
-                self.screenshot = await self.page.screenshot(type_='jpeg',
-                                                             fullPage=False,
-                                                             quality=int(os.getenv("SCREENSHOT_QUALITY", 72)))
-            except Exception as e:
-                logger.error('ERROR: Failed to get viewport-only reduced screenshot :(')
-                pass
-        finally:
-            # It's good to log here in the case that the browser crashes on shutting down but we still get the data we need
-            logger.success(f"Fetching '{url}' complete, closing page")
-            await self.page.close()
-            logger.success(f"Fetching '{url}' complete, closing browser")
-            await browser.close()
+        self.screenshot = await capture_full_page(page=self.page)
+
+        # It's good to log here in the case that the browser crashes on shutting down but we still get the data we need
+        logger.success(f"Fetching '{url}' complete, closing page")
+        await self.page.close()
+        logger.success(f"Fetching '{url}' complete, closing browser")
+        await browser.close()
         logger.success(f"Fetching '{url}' complete, exiting puppeteer fetch.")
 
     async def main(self, **kwargs):

changedetectionio/content_fetchers/requests.py

@@ -96,3 +96,17 @@ class fetcher(Fetcher):
 
 
         self.raw_content = r.content
+
+    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
+        if strtobool(os.getenv("REMOVE_REQUESTS_OLD_SCREENSHOTS", 'true')):
+            screenshot = watch.get_screenshot()
+            if screenshot:
+                try:
+                    os.unlink(screenshot)
+                except Exception as e:
+                    logger.warning(f"Failed to unlink screenshot: {screenshot} - {e}")
+

changedetectionio/content_fetchers/res/puppeteer_fetch.js

@@ -1,190 +0,0 @@
-module.exports = async ({page, context}) => {
-
-    var {
-        url,
-        execute_js,
-        user_agent,
-        extra_wait_ms,
-        req_headers,
-        include_filters,
-        xpath_element_js,
-        screenshot_quality,
-        proxy_username,
-        proxy_password,
-        disk_cache_dir,
-        no_cache_list,
-        block_url_list,
-    } = context;
-
-    await page.setBypassCSP(true)
-    await page.setExtraHTTPHeaders(req_headers);
-
-    if (user_agent) {
-        await page.setUserAgent(user_agent);
-    }
-    // https://ourcodeworld.com/articles/read/1106/how-to-solve-puppeteer-timeouterror-navigation-timeout-of-30000-ms-exceeded
-
-    await page.setDefaultNavigationTimeout(0);
-
-    if (proxy_username) {
-        // Setting Proxy-Authentication header is deprecated, and doing so can trigger header change errors from Puppeteer
-        // https://github.com/puppeteer/puppeteer/issues/676 ?
-        // https://help.brightdata.com/hc/en-us/articles/12632549957649-Proxy-Manager-How-to-Guides#h_01HAKWR4Q0AFS8RZTNYWRDFJC2
-        // https://cri.dev/posts/2020-03-30-How-to-solve-Puppeteer-Chrome-Error-ERR_INVALID_ARGUMENT/
-        await page.authenticate({
-            username: proxy_username,
-            password: proxy_password
-        });
-    }
-
-    await page.setViewport({
-        width: 1024,
-        height: 768,
-        deviceScaleFactor: 1,
-    });
-
-    await page.setRequestInterception(true);
-    if (disk_cache_dir) {
-        console.log(">>>>>>>>>>>>>>> LOCAL DISK CACHE ENABLED <<<<<<<<<<<<<<<<<<<<<");
-    }
-    const fs = require('fs');
-    const crypto = require('crypto');
-
-    function file_is_expired(file_path) {
-        if (!fs.existsSync(file_path)) {
-            return true;
-        }
-        var stats = fs.statSync(file_path);
-        const now_date = new Date();
-        const expire_seconds = 300;
-        if ((now_date / 1000) - (stats.mtime.getTime() / 1000) > expire_seconds) {
-            console.log("CACHE EXPIRED: " + file_path);
-            return true;
-        }
-        return false;
-
-    }
-
-    page.on('request', async (request) => {
-        // General blocking of requests that waste traffic
-        if (block_url_list.some(substring => request.url().toLowerCase().includes(substring))) return request.abort();
-
-        if (disk_cache_dir) {
-            const url = request.url();
-            const key = crypto.createHash('md5').update(url).digest("hex");
-            const dir_path = disk_cache_dir + key.slice(0, 1) + '/' + key.slice(1, 2) + '/' + key.slice(2, 3) + '/';
-
-            // https://stackoverflow.com/questions/4482686/check-synchronously-if-file-directory-exists-in-node-js
-
-            if (fs.existsSync(dir_path + key)) {
-                console.log("* CACHE HIT , using - " + dir_path + key + " - " + url);
-                const cached_data = fs.readFileSync(dir_path + key);
-                // @todo headers can come from dir_path+key+".meta" json file
-                request.respond({
-                    status: 200,
-                    //contentType: 'text/html', //@todo
-                    body: cached_data
-                });
-                return;
-            }
-        }
-        request.continue();
-    });
-
-
-    if (disk_cache_dir) {
-        page.on('response', async (response) => {
-            const url = response.url();
-            // Basic filtering for sane responses
-            if (response.request().method() != 'GET' || response.request().resourceType() == 'xhr' || response.request().resourceType() == 'document' || response.status() != 200) {
-                console.log("Skipping (not useful) - Status:" + response.status() + " Method:" + response.request().method() + " ResourceType:" + response.request().resourceType() + " " + url);
-                return;
-            }
-            if (no_cache_list.some(substring => url.toLowerCase().includes(substring))) {
-                console.log("Skipping (no_cache_list) - " + url);
-                return;
-            }
-            if (url.toLowerCase().includes('data:')) {
-                console.log("Skipping (embedded-data) - " + url);
-                return;
-            }
-            response.buffer().then(buffer => {
-                if (buffer.length > 100) {
-                    console.log("Cache - Saving " + response.request().method() + " - " + url + " - " + response.request().resourceType());
-
-                    const key = crypto.createHash('md5').update(url).digest("hex");
-                    const dir_path = disk_cache_dir + key.slice(0, 1) + '/' + key.slice(1, 2) + '/' + key.slice(2, 3) + '/';
-
-                    if (!fs.existsSync(dir_path)) {
-                        fs.mkdirSync(dir_path, {recursive: true})
-                    }
-
-                    if (fs.existsSync(dir_path + key)) {
-                        if (file_is_expired(dir_path + key)) {
-                            fs.writeFileSync(dir_path + key, buffer);
-                        }
-                    } else {
-                        fs.writeFileSync(dir_path + key, buffer);
-                    }
-                }
-            });
-        });
-    }
-
-    const r = await page.goto(url, {
-        waitUntil: 'load'
-    });
-
-    await page.waitForTimeout(1000);
-    await page.waitForTimeout(extra_wait_ms);
-
-    if (execute_js) {
-        await page.evaluate(execute_js);
-        await page.waitForTimeout(200);
-    }
-
-    var xpath_data;
-    var instock_data;
-    try {
-        // Not sure the best way here, in the future this should be a new package added to npm then run in evaluatedCode
-        // (Once the old playwright is removed)
-        xpath_data = await page.evaluate((include_filters) => {%xpath_scrape_code%}, include_filters);
-        instock_data = await page.evaluate(() => {%instock_scrape_code%});
-    } catch (e) {
-        console.log(e);
-    }
-
-    // Protocol error (Page.captureScreenshot): Cannot take screenshot with 0 width can come from a proxy auth failure
-    // Wrap it here (for now)
-
-    var b64s = false;
-    try {
-        b64s = await page.screenshot({encoding: "base64", fullPage: true, quality: screenshot_quality, type: 'jpeg'});
-    } catch (e) {
-        console.log(e);
-    }
-
-    // May fail on very large pages with 'WARNING: tile memory limits exceeded, some content may not draw'
-    if (!b64s) {
-        // @todo after text extract, we can place some overlay text with red background to say 'croppped'
-        console.error('ERROR: content-fetcher page was maybe too large for a screenshot, reverting to viewport only screenshot');
-        try {
-            b64s = await page.screenshot({encoding: "base64", quality: screenshot_quality, type: 'jpeg'});
-        } catch (e) {
-            console.log(e);
-        }
-    }
-
-    var html = await page.content();
-    return {
-        data: {
-            'content': html,
-            'headers': r.headers(),
-            'instock_data': instock_data,
-            'screenshot': b64s,
-            'status_code': r.status(),
-            'xpath_data': xpath_data
-        },
-        type: 'application/json',
-    };
-};

changedetectionio/content_fetchers/res/stock-not-in-stock.js

@@ -1,229 +1,220 @@
-// Restock Detector
-// (c) Leigh Morresi dgtlmoon@gmail.com
-//
-// Assumes the product is in stock to begin with, unless the following appears above the fold ;
-// - outOfStockTexts appears above the fold (out of stock)
-// - negateOutOfStockRegex (really is in stock)
+async () => {
 
-function isItemInStock() {
-    // @todo Pass these in so the same list can be used in non-JS fetchers
-    const outOfStockTexts = [
-        ' أخبرني عندما يتوفر',
-        '0 in stock',
-        'actuellement indisponible',
-        'agotado',
-        'article épuisé',
-        'artikel zurzeit vergriffen',
-        'as soon as stock is available',
-        'ausverkauft', // sold out
-        'available for back order',
-        'awaiting stock',
-        'back in stock soon',
-        'back-order or out of stock',
-        'backordered',
-        'benachrichtigt mich', // notify me
-        'brak na stanie',
-        'brak w magazynie',
-        'coming soon',
-        'currently have any tickets for this',
-        'currently unavailable',
-        'dieser artikel ist bald wieder verfügbar',
-        'dostępne wkrótce',
-        'en rupture',
-        'en rupture de stock',
-        'épuisé',
-        'esgotado',
-        'indisponible',
-        'indisponível',
-        'isn\'t in stock right now',
-        'isnt in stock right now',
-        'isn’t in stock right now',
-        'item is no longer available',
-        'let me know when it\'s available',
-        'mail me when available',
-        'message if back in stock',
-        'mevcut değil',
-        'nachricht bei',
-        'nicht auf lager',
-        'nicht lagernd',
-        'nicht lieferbar',
-        'nicht verfügbar',
-        'nicht vorrätig',
-        'nicht zur verfügung',
-        'nie znaleziono produktów',
-        'niet beschikbaar',
-        'niet leverbaar',
-        'niet op voorraad',
-        'no disponible',
-        'non disponibile',
-        'non disponible',
-        'no longer in stock',
-        'no tickets available',
-        'not available',
-        'not currently available',
-        'not in stock',
-        'notify me when available',
-        'notify me',
-        'notify when available',
-        'não disponível',
-        'não estamos a aceitar encomendas',
-        'out of stock',
-        'out-of-stock',
-        'plus disponible',
-        'prodotto esaurito',
-        'produkt niedostępny',
-        'rupture',
-        'sold out',
-        'sold-out',
-        'stok habis',
-        'stok kosong',
-        'stok varian ini habis',
-        'stokta yok',
-        'temporarily out of stock',
-        'temporarily unavailable',
-        'there were no search results for',
-        'this item is currently unavailable',
-        'tickets unavailable',
-        'tidak dijual',
-        'tidak tersedia',
-        'tijdelijk uitverkocht',
-        'tiket tidak tersedia',
-        'tükendi',
-        'unavailable nearby',
-        'unavailable tickets',
-        'vergriffen',
-        'vorbestellen',
-        'vorbestellung ist bald möglich',
-        'we don\'t currently have any',
-        'we couldn\'t find any products that match',
-        'we do not currently have an estimate of when this product will be back in stock.',
-        'we don\'t know when or if this item will be back in stock.',
-        'we were not able to find a match',
-        'when this arrives in stock',
-        'zur zeit nicht an lager',
-        '品切れ',
-        '已售',
-        '已售完',
-        '품절'
-    ];
+    function isItemInStock() {
+        // @todo Pass these in so the same list can be used in non-JS fetchers
+        const outOfStockTexts = [
+            ' أخبرني عندما يتوفر',
+            '0 in stock',
+            'actuellement indisponible',
+            'agotado',
+            'article épuisé',
+            'artikel zurzeit vergriffen',
+            'as soon as stock is available',
+            'ausverkauft', // sold out
+            'available for back order',
+            'awaiting stock',
+            'back in stock soon',
+            'back-order or out of stock',
+            'backordered',
+            'benachrichtigt mich', // notify me
+            'brak na stanie',
+            'brak w magazynie',
+            'coming soon',
+            'currently have any tickets for this',
+            'currently unavailable',
+            'dieser artikel ist bald wieder verfügbar',
+            'dostępne wkrótce',
+            'en rupture',
+            'en rupture de stock',
+            'épuisé',
+            'esgotado',
+            'indisponible',
+            'indisponível',
+            'isn\'t in stock right now',
+            'isnt in stock right now',
+            'isn’t in stock right now',
+            'item is no longer available',
+            'let me know when it\'s available',
+            'mail me when available',
+            'message if back in stock',
+            'mevcut değil',
+            'nachricht bei',
+            'nicht auf lager',
+            'nicht lagernd',
+            'nicht lieferbar',
+            'nicht verfügbar',
+            'nicht vorrätig',
+            'nicht zur verfügung',
+            'nie znaleziono produktów',
+            'niet beschikbaar',
+            'niet leverbaar',
+            'niet op voorraad',
+            'no disponible',
+            'non disponibile',
+            'non disponible',
+            'no longer in stock',
+            'no tickets available',
+            'not available',
+            'not currently available',
+            'not in stock',
+            'notify me when available',
+            'notify me',
+            'notify when available',
+            'não disponível',
+            'não estamos a aceitar encomendas',
+            'out of stock',
+            'out-of-stock',
+            'plus disponible',
+            'prodotto esaurito',
+            'produkt niedostępny',
+            'rupture',
+            'sold out',
+            'sold-out',
+            'stok habis',
+            'stok kosong',
+            'stok varian ini habis',
+            'stokta yok',
+            'temporarily out of stock',
+            'temporarily unavailable',
+            'there were no search results for',
+            'this item is currently unavailable',
+            'tickets unavailable',
+            'tidak dijual',
+            'tidak tersedia',
+            'tijdelijk uitverkocht',
+            'tiket tidak tersedia',
+            'tükendi',
+            'unavailable nearby',
+            'unavailable tickets',
+            'vergriffen',
+            'vorbestellen',
+            'vorbestellung ist bald möglich',
+            'we don\'t currently have any',
+            'we couldn\'t find any products that match',
+            'we do not currently have an estimate of when this product will be back in stock.',
+            'we don\'t know when or if this item will be back in stock.',
+            'we were not able to find a match',
+            'when this arrives in stock',
+            'zur zeit nicht an lager',
+            '品切れ',
+            '已售',
+            '已售完',
+            '품절'
+        ];
 
 
-    const vh = Math.max(document.documentElement.clientHeight || 0, window.innerHeight || 0);
+        const vh = Math.max(document.documentElement.clientHeight || 0, window.innerHeight || 0);
 
-    function getElementBaseText(element) {
-        // .textContent can include text from children which may give the wrong results
-        // scan only immediate TEXT_NODEs, which will be a child of the element
-        var text = "";
-        for (var i = 0; i < element.childNodes.length; ++i)
-            if (element.childNodes[i].nodeType === Node.TEXT_NODE)
-                text += element.childNodes[i].textContent;
-        return text.toLowerCase().trim();
-    }
+        function getElementBaseText(element) {
+            // .textContent can include text from children which may give the wrong results
+            // scan only immediate TEXT_NODEs, which will be a child of the element
+            var text = "";
+            for (var i = 0; i < element.childNodes.length; ++i)
+                if (element.childNodes[i].nodeType === Node.TEXT_NODE)
+                    text += element.childNodes[i].textContent;
+            return text.toLowerCase().trim();
+        }
 
-    const negateOutOfStockRegex = new RegExp('^([0-9] in stock|add to cart|in stock)', 'ig');
+        const negateOutOfStockRegex = new RegExp('^([0-9] in stock|add to cart|in stock)', 'ig');
 
-    // The out-of-stock or in-stock-text is generally always above-the-fold
-    // and often below-the-fold is a list of related products that may or may not contain trigger text
-    // so it's good to filter to just the 'above the fold' elements
-    // and it should be atleast 100px from the top to ignore items in the toolbar, sometimes menu items like "Coming soon" exist
+        // The out-of-stock or in-stock-text is generally always above-the-fold
+        // and often below-the-fold is a list of related products that may or may not contain trigger text
+        // so it's good to filter to just the 'above the fold' elements
+        // and it should be atleast 100px from the top to ignore items in the toolbar, sometimes menu items like "Coming soon" exist
 
 
 // @todo - if it's SVG or IMG, go into image diff mode
-// %ELEMENTS% replaced at injection time because different interfaces use it with different settings
 
-    console.log("Scanning %ELEMENTS%");
+        function collectVisibleElements(parent, visibleElements) {
+            if (!parent) return; // Base case: if parent is null or undefined, return
 
-    function collectVisibleElements(parent, visibleElements) {
-        if (!parent) return; // Base case: if parent is null or undefined, return
+            // Add the parent itself to the visible elements array if it's of the specified types
+            visibleElements.push(parent);
 
-        // Add the parent itself to the visible elements array if it's of the specified types
-        visibleElements.push(parent);
-
-        // Iterate over the parent's children
-        const children = parent.children;
-        for (let i = 0; i < children.length; i++) {
-            const child = children[i];
-            if (
-                child.nodeType === Node.ELEMENT_NODE &&
-                window.getComputedStyle(child).display !== 'none' &&
-                window.getComputedStyle(child).visibility !== 'hidden' &&
-                child.offsetWidth >= 0 &&
-                child.offsetHeight >= 0 &&
-                window.getComputedStyle(child).contentVisibility !== 'hidden'
-            ) {
-                // If the child is an element and is visible, recursively collect visible elements
-                collectVisibleElements(child, visibleElements);
+            // Iterate over the parent's children
+            const children = parent.children;
+            for (let i = 0; i < children.length; i++) {
+                const child = children[i];
+                if (
+                    child.nodeType === Node.ELEMENT_NODE &&
+                    window.getComputedStyle(child).display !== 'none' &&
+                    window.getComputedStyle(child).visibility !== 'hidden' &&
+                    child.offsetWidth >= 0 &&
+                    child.offsetHeight >= 0 &&
+                    window.getComputedStyle(child).contentVisibility !== 'hidden'
+                ) {
+                    // If the child is an element and is visible, recursively collect visible elements
+                    collectVisibleElements(child, visibleElements);
+                }
             }
         }
-    }
 
-    const elementsToScan = [];
-    collectVisibleElements(document.body, elementsToScan);
+        const elementsToScan = [];
+        collectVisibleElements(document.body, elementsToScan);
 
-    var elementText = "";
+        var elementText = "";
 
-    // REGEXS THAT REALLY MEAN IT'S IN STOCK
-    for (let i = elementsToScan.length - 1; i >= 0; i--) {
-        const element = elementsToScan[i];
+        // REGEXS THAT REALLY MEAN IT'S IN STOCK
+        for (let i = elementsToScan.length - 1; i >= 0; i--) {
+            const element = elementsToScan[i];
 
-        // outside the 'fold' or some weird text in the heading area
-        // .getBoundingClientRect() was causing a crash in chrome 119, can only be run on contentVisibility != hidden
-        if (element.getBoundingClientRect().top + window.scrollY >= vh || element.getBoundingClientRect().top + window.scrollY <= 100) {
-            continue
+            // outside the 'fold' or some weird text in the heading area
+            // .getBoundingClientRect() was causing a crash in chrome 119, can only be run on contentVisibility != hidden
+            if (element.getBoundingClientRect().top + window.scrollY >= vh || element.getBoundingClientRect().top + window.scrollY <= 100) {
+                continue
+            }
+
+            elementText = "";
+            try {
+                if (element.tagName.toLowerCase() === "input") {
+                    elementText = element.value.toLowerCase().trim();
+                } else {
+                    elementText = getElementBaseText(element);
+                }
+            } catch (e) {
+                console.warn('stock-not-in-stock.js scraper - handling element for gettext failed', e);
+            }
+
+            if (elementText.length) {
+                // try which ones could mean its in stock
+                if (negateOutOfStockRegex.test(elementText) && !elementText.includes('(0 products)')) {
+                    console.log(`Negating/overriding 'Out of Stock' back to "Possibly in stock" found "${elementText}"`)
+                    return 'Possibly in stock';
+                }
+            }
         }
 
-        elementText = "";
-        try {
+        // OTHER STUFF THAT COULD BE THAT IT'S OUT OF STOCK
+        for (let i = elementsToScan.length - 1; i >= 0; i--) {
+            const element = elementsToScan[i];
+            // outside the 'fold' or some weird text in the heading area
+            // .getBoundingClientRect() was causing a crash in chrome 119, can only be run on contentVisibility != hidden
+            // Note: theres also an automated test that places the 'out of stock' text fairly low down
+            if (element.getBoundingClientRect().top + window.scrollY >= vh + 250 || element.getBoundingClientRect().top + window.scrollY <= 100) {
+                continue
+            }
+            elementText = "";
             if (element.tagName.toLowerCase() === "input") {
                 elementText = element.value.toLowerCase().trim();
             } else {
                 elementText = getElementBaseText(element);
             }
-        } catch (e) {
-            console.warn('stock-not-in-stock.js scraper - handling element for gettext failed', e);
-        }
 
-        if (elementText.length) {
-            // try which ones could mean its in stock
-            if (negateOutOfStockRegex.test(elementText) && !elementText.includes('(0 products)')) {
-                console.log(`Negating/overriding 'Out of Stock' back to "Possibly in stock" found "${elementText}"`)
-                return 'Possibly in stock';
-            }
-        }
-    }
-
-    // OTHER STUFF THAT COULD BE THAT IT'S OUT OF STOCK
-    for (let i = elementsToScan.length - 1; i >= 0; i--) {
-        const element = elementsToScan[i];
-        // outside the 'fold' or some weird text in the heading area
-        // .getBoundingClientRect() was causing a crash in chrome 119, can only be run on contentVisibility != hidden
-        // Note: theres also an automated test that places the 'out of stock' text fairly low down
-        if (element.getBoundingClientRect().top + window.scrollY >= vh + 250 || element.getBoundingClientRect().top + window.scrollY <= 100) {
-            continue
-        }
-        elementText = "";
-        if (element.tagName.toLowerCase() === "input") {
-            elementText = element.value.toLowerCase().trim();
-        } else {
-            elementText = getElementBaseText(element);
-        }
-
-        if (elementText.length) {
-            // and these mean its out of stock
-            for (const outOfStockText of outOfStockTexts) {
-                if (elementText.includes(outOfStockText)) {
-                    console.log(`Selected 'Out of Stock' - found text "${outOfStockText}" - "${elementText}" - offset top ${element.getBoundingClientRect().top}, page height is ${vh}`)
-                    return outOfStockText; // item is out of stock
+            if (elementText.length) {
+                // and these mean its out of stock
+                for (const outOfStockText of outOfStockTexts) {
+                    if (elementText.includes(outOfStockText)) {
+                        console.log(`Selected 'Out of Stock' - found text "${outOfStockText}" - "${elementText}" - offset top ${element.getBoundingClientRect().top}, page height is ${vh}`)
+                        return outOfStockText; // item is out of stock
+                    }
                 }
             }
         }
+
+        console.log(`Returning 'Possibly in stock' - cant' find any useful matching text`)
+        return 'Possibly in stock'; // possibly in stock, cant decide otherwise.
     }
 
-    console.log(`Returning 'Possibly in stock' - cant' find any useful matching text`)
-    return 'Possibly in stock'; // possibly in stock, cant decide otherwise.
-}
-
 // returns the element text that makes it think it's out of stock
-return isItemInStock().trim()
-
-
+    return isItemInStock().trim()
+}

changedetectionio/content_fetchers/res/xpath_element_scraper.js

@@ -1,285 +1,285 @@
-// Copyright (C) 2021 Leigh Morresi (dgtlmoon@gmail.com)
-// All rights reserved.
+async (options) => {
 
-// @file Scrape the page looking for elements of concern (%ELEMENTS%)
-// http://matatk.agrip.org.uk/tests/position-and-width/
-// https://stackoverflow.com/questions/26813480/when-is-element-getboundingclientrect-guaranteed-to-be-updated-accurate
-//
-// Some pages like https://www.londonstockexchange.com/stock/NCCL/ncondezi-energy-limited/analysis
-// will automatically force a scroll somewhere, so include the position offset
-// Lets hope the position doesnt change while we iterate the bbox's, but this is better than nothing
-var scroll_y = 0;
-try {
-    scroll_y = +document.documentElement.scrollTop || document.body.scrollTop
-} catch (e) {
-    console.log(e);
-}
+    let visualselector_xpath_selectors = options.visualselector_xpath_selectors
+    let max_height = options.max_height
 
-
-// Include the getXpath script directly, easier than fetching
-function getxpath(e) {
-    var n = e;
-    if (n && n.id) return '//*[@id="' + n.id + '"]';
-    for (var o = []; n && Node.ELEMENT_NODE === n.nodeType;) {
-        for (var i = 0, r = !1, d = n.previousSibling; d;) d.nodeType !== Node.DOCUMENT_TYPE_NODE && d.nodeName === n.nodeName && i++, d = d.previousSibling;
-        for (d = n.nextSibling; d;) {
-            if (d.nodeName === n.nodeName) {
-                r = !0;
-                break
-            }
-            d = d.nextSibling
-        }
-        o.push((n.prefix ? n.prefix + ":" : "") + n.localName + (i || r ? "[" + (i + 1) + "]" : "")), n = n.parentNode
-    }
-    return o.length ? "/" + o.reverse().join("/") : ""
-}
-
-const findUpTag = (el) => {
-    let r = el
-    chained_css = [];
-    depth = 0;
-
-    //  Strategy 1: If it's an input, with name, and there's only one, prefer that
-    if (el.name !== undefined && el.name.length) {
-        var proposed = el.tagName + "[name=\"" + CSS.escape(el.name) + "\"]";
-        var proposed_element = window.document.querySelectorAll(proposed);
-        if (proposed_element.length) {
-            if (proposed_element.length === 1) {
-                return proposed;
-            } else {
-                // Some sites change ID but name= stays the same, we can hit it if we know the index
-                // Find all the elements that match and work out the input[n]
-                var n = Array.from(proposed_element).indexOf(el);
-                // Return a Playwright selector for nthinput[name=zipcode]
-                return proposed + " >> nth=" + n;
-            }
-        }
-    }
-
-    // Strategy 2: Keep going up until we hit an ID tag, imagine it's like  #list-widget div h4
-    while (r.parentNode) {
-        if (depth === 5) {
-            break;
-        }
-        if ('' !== r.id) {
-            chained_css.unshift("#" + CSS.escape(r.id));
-            final_selector = chained_css.join(' > ');
-            // Be sure theres only one, some sites have multiples of the same ID tag :-(
-            if (window.document.querySelectorAll(final_selector).length === 1) {
-                return final_selector;
-            }
-            return null;
-        } else {
-            chained_css.unshift(r.tagName.toLowerCase());
-        }
-        r = r.parentNode;
-        depth += 1;
-    }
-    return null;
-}
-
-
-// @todo - if it's SVG or IMG, go into image diff mode
-// %ELEMENTS% replaced at injection time because different interfaces use it with different settings
-
-var size_pos = [];
-// after page fetch, inject this JS
-// build a map of all elements and their positions (maybe that only include text?)
-var bbox;
-console.log("Scanning %ELEMENTS%");
-
-function collectVisibleElements(parent, visibleElements) {
-    if (!parent) return; // Base case: if parent is null or undefined, return
-
-
-    // Add the parent itself to the visible elements array if it's of the specified types
-    const tagName = parent.tagName.toLowerCase();
-    if ("%ELEMENTS%".split(',').includes(tagName)) {
-        visibleElements.push(parent);
-    }
-
-    // Iterate over the parent's children
-    const children = parent.children;
-    for (let i = 0; i < children.length; i++) {
-        const child = children[i];
-        const computedStyle = window.getComputedStyle(child);
-
-        if (
-            child.nodeType === Node.ELEMENT_NODE &&
-            computedStyle.display !== 'none' &&
-            computedStyle.visibility !== 'hidden' &&
-            child.offsetWidth >= 0 &&
-            child.offsetHeight >= 0 &&
-            computedStyle.contentVisibility !== 'hidden'
-        ) {
-            // If the child is an element and is visible, recursively collect visible elements
-            collectVisibleElements(child, visibleElements);
-        }
-    }
-}
-
-// Create an array to hold the visible elements
-const visibleElementsArray = [];
-
-// Call collectVisibleElements with the starting parent element
-collectVisibleElements(document.body, visibleElementsArray);
-
-
-visibleElementsArray.forEach(function (element) {
-
-    bbox = element.getBoundingClientRect();
-
-    // Skip really small ones, and where width or height ==0
-    if (bbox['width'] * bbox['height'] < 10) {
-        return
-    }
-
-    // Don't include elements that are offset from canvas
-    if (bbox['top'] + scroll_y < 0 || bbox['left'] < 0) {
-        return
-    }
-
-    // @todo the getXpath kind of sucks, it doesnt know when there is for example just one ID sometimes
-    // it should not traverse when we know we can anchor off just an ID one level up etc..
-    // maybe, get current class or id, keep traversing up looking for only class or id until there is just one match
-
-    // 1st primitive - if it has class, try joining it all and select, if theres only one.. well thats us.
-    xpath_result = false;
+    var scroll_y = 0;
     try {
-        var d = findUpTag(element);
-        if (d) {
-            xpath_result = d;
-        }
+        scroll_y = +document.documentElement.scrollTop || document.body.scrollTop
     } catch (e) {
         console.log(e);
     }
-    // You could swap it and default to getXpath and then try the smarter one
-    // default back to the less intelligent one
-    if (!xpath_result) {
-        try {
-            // I've seen on FB and eBay that this doesnt work
-            // ReferenceError: getXPath is not defined at eval (eval at evaluate (:152:29), <anonymous>:67:20) at UtilityScript.evaluate (<anonymous>:159:18) at UtilityScript.<anonymous> (<anonymous>:1:44)
-            xpath_result = getxpath(element);
-        } catch (e) {
-            console.log(e);
-            return
+
+// Include the getXpath script directly, easier than fetching
+    function getxpath(e) {
+        var n = e;
+        if (n && n.id) return '//*[@id="' + n.id + '"]';
+        for (var o = []; n && Node.ELEMENT_NODE === n.nodeType;) {
+            for (var i = 0, r = !1, d = n.previousSibling; d;) d.nodeType !== Node.DOCUMENT_TYPE_NODE && d.nodeName === n.nodeName && i++, d = d.previousSibling;
+            for (d = n.nextSibling; d;) {
+                if (d.nodeName === n.nodeName) {
+                    r = !0;
+                    break
+                }
+                d = d.nextSibling
+            }
+            o.push((n.prefix ? n.prefix + ":" : "") + n.localName + (i || r ? "[" + (i + 1) + "]" : "")), n = n.parentNode
+        }
+        return o.length ? "/" + o.reverse().join("/") : ""
+    }
+
+    const findUpTag = (el) => {
+        let r = el
+        chained_css = [];
+        depth = 0;
+
+        //  Strategy 1: If it's an input, with name, and there's only one, prefer that
+        if (el.name !== undefined && el.name.length) {
+            var proposed = el.tagName + "[name=\"" + CSS.escape(el.name) + "\"]";
+            var proposed_element = window.document.querySelectorAll(proposed);
+            if (proposed_element.length) {
+                if (proposed_element.length === 1) {
+                    return proposed;
+                } else {
+                    // Some sites change ID but name= stays the same, we can hit it if we know the index
+                    // Find all the elements that match and work out the input[n]
+                    var n = Array.from(proposed_element).indexOf(el);
+                    // Return a Playwright selector for nthinput[name=zipcode]
+                    return proposed + " >> nth=" + n;
+                }
+            }
+        }
+
+        // Strategy 2: Keep going up until we hit an ID tag, imagine it's like  #list-widget div h4
+        while (r.parentNode) {
+            if (depth === 5) {
+                break;
+            }
+            if ('' !== r.id) {
+                chained_css.unshift("#" + CSS.escape(r.id));
+                final_selector = chained_css.join(' > ');
+                // Be sure theres only one, some sites have multiples of the same ID tag :-(
+                if (window.document.querySelectorAll(final_selector).length === 1) {
+                    return final_selector;
+                }
+                return null;
+            } else {
+                chained_css.unshift(r.tagName.toLowerCase());
+            }
+            r = r.parentNode;
+            depth += 1;
+        }
+        return null;
+    }
+
+
+// @todo - if it's SVG or IMG, go into image diff mode
+
+    var size_pos = [];
+// after page fetch, inject this JS
+// build a map of all elements and their positions (maybe that only include text?)
+    var bbox;
+    console.log(`Scanning for "${visualselector_xpath_selectors}"`);
+
+    function collectVisibleElements(parent, visibleElements) {
+        if (!parent) return; // Base case: if parent is null or undefined, return
+
+
+        // Add the parent itself to the visible elements array if it's of the specified types
+        const tagName = parent.tagName.toLowerCase();
+        if (visualselector_xpath_selectors.split(',').includes(tagName)) {
+            visibleElements.push(parent);
+        }
+
+        // Iterate over the parent's children
+        const children = parent.children;
+        for (let i = 0; i < children.length; i++) {
+            const child = children[i];
+            const computedStyle = window.getComputedStyle(child);
+
+            if (
+                child.nodeType === Node.ELEMENT_NODE &&
+                computedStyle.display !== 'none' &&
+                computedStyle.visibility !== 'hidden' &&
+                child.offsetWidth >= 0 &&
+                child.offsetHeight >= 0 &&
+                computedStyle.contentVisibility !== 'hidden'
+            ) {
+                // If the child is an element and is visible, recursively collect visible elements
+                collectVisibleElements(child, visibleElements);
+            }
         }
     }
 
-    let label = "not-interesting" // A placeholder, the actual labels for training are done by hand for now
+// Create an array to hold the visible elements
+    const visibleElementsArray = [];
 
-    let text = element.textContent.trim().slice(0, 30).trim();
-    while (/\n{2,}|\t{2,}/.test(text)) {
-        text = text.replace(/\n{2,}/g, '\n').replace(/\t{2,}/g, '\t')
-    }
+// Call collectVisibleElements with the starting parent element
+    collectVisibleElements(document.body, visibleElementsArray);
 
-    // Try to identify any possible currency amounts "Sale: 4000" or "Sale now 3000 Kc", can help with the training.
-    const hasDigitCurrency = (/\d/.test(text.slice(0, 6)) || /\d/.test(text.slice(-6)) ) &&  /([€£$¥₩₹]|USD|AUD|EUR|Kč|kr|SEK|,–)/.test(text) ;
-    const computedStyle = window.getComputedStyle(element);
 
-    size_pos.push({
-        xpath: xpath_result,
-        width: Math.round(bbox['width']),
-        height: Math.round(bbox['height']),
-        left: Math.floor(bbox['left']),
-        top: Math.floor(bbox['top']) + scroll_y,
-        // tagName used by Browser Steps
-        tagName: (element.tagName) ? element.tagName.toLowerCase() : '',
-        // tagtype used by Browser Steps
-        tagtype: (element.tagName.toLowerCase() === 'input' && element.type) ? element.type.toLowerCase() : '',
-        isClickable: computedStyle.cursor === "pointer",
-        // Used by the keras trainer
-        fontSize: computedStyle.getPropertyValue('font-size'),
-        fontWeight: computedStyle.getPropertyValue('font-weight'),
-        hasDigitCurrency: hasDigitCurrency,
-        label: label,
+    visibleElementsArray.forEach(function (element) {
+
+        bbox = element.getBoundingClientRect();
+
+        // Skip really small ones, and where width or height ==0
+        if (bbox['width'] * bbox['height'] < 10) {
+            return
+        }
+
+        // Don't include elements that are offset from canvas
+        if (bbox['top'] + scroll_y < 0 || bbox['left'] < 0) {
+            return
+        }
+
+        // @todo the getXpath kind of sucks, it doesnt know when there is for example just one ID sometimes
+        // it should not traverse when we know we can anchor off just an ID one level up etc..
+        // maybe, get current class or id, keep traversing up looking for only class or id until there is just one match
+
+        // 1st primitive - if it has class, try joining it all and select, if theres only one.. well thats us.
+        xpath_result = false;
+        try {
+            var d = findUpTag(element);
+            if (d) {
+                xpath_result = d;
+            }
+        } catch (e) {
+            console.log(e);
+        }
+        // You could swap it and default to getXpath and then try the smarter one
+        // default back to the less intelligent one
+        if (!xpath_result) {
+            try {
+                // I've seen on FB and eBay that this doesnt work
+                // ReferenceError: getXPath is not defined at eval (eval at evaluate (:152:29), <anonymous>:67:20) at UtilityScript.evaluate (<anonymous>:159:18) at UtilityScript.<anonymous> (<anonymous>:1:44)
+                xpath_result = getxpath(element);
+            } catch (e) {
+                console.log(e);
+                return
+            }
+        }
+
+        let label = "not-interesting" // A placeholder, the actual labels for training are done by hand for now
+
+        let text = element.textContent.trim().slice(0, 30).trim();
+        while (/\n{2,}|\t{2,}/.test(text)) {
+            text = text.replace(/\n{2,}/g, '\n').replace(/\t{2,}/g, '\t')
+        }
+
+        // Try to identify any possible currency amounts "Sale: 4000" or "Sale now 3000 Kc", can help with the training.
+        const hasDigitCurrency = (/\d/.test(text.slice(0, 6)) || /\d/.test(text.slice(-6))) && /([€£$¥₩₹]|USD|AUD|EUR|Kč|kr|SEK|,–)/.test(text);
+        const computedStyle = window.getComputedStyle(element);
+
+        if (Math.floor(bbox['top']) + scroll_y > max_height) {
+            return
+        }
+
+        size_pos.push({
+            xpath: xpath_result,
+            width: Math.round(bbox['width']),
+            height: Math.round(bbox['height']),
+            left: Math.floor(bbox['left']),
+            top: Math.floor(bbox['top']) + scroll_y,
+            // tagName used by Browser Steps
+            tagName: (element.tagName) ? element.tagName.toLowerCase() : '',
+            // tagtype used by Browser Steps
+            tagtype: (element.tagName.toLowerCase() === 'input' && element.type) ? element.type.toLowerCase() : '',
+            isClickable: computedStyle.cursor === "pointer",
+            // Used by the keras trainer
+            fontSize: computedStyle.getPropertyValue('font-size'),
+            fontWeight: computedStyle.getPropertyValue('font-weight'),
+            hasDigitCurrency: hasDigitCurrency,
+            label: label,
+        });
+
     });
 
-});
-
 
 // Inject the current one set in the include_filters, which may be a CSS rule
 // used for displaying the current one in VisualSelector, where its not one we generated.
-if (include_filters.length) {
-    let results;
-    // Foreach filter, go and find it on the page and add it to the results so we can visualise it again
-    for (const f of include_filters) {
-        bbox = false;
-        q = false;
+    if (include_filters.length) {
+        let results;
+        // Foreach filter, go and find it on the page and add it to the results so we can visualise it again
+        for (const f of include_filters) {
+            bbox = false;
+            q = false;
 
-        if (!f.length) {
-            console.log("xpath_element_scraper: Empty filter, skipping");
-            continue;
-        }
-
-        try {
-            // is it xpath?
-            if (f.startsWith('/') || f.startsWith('xpath')) {
-                var qry_f = f.replace(/xpath(:|\d:)/, '')
-                console.log("[xpath] Scanning for included filter " + qry_f)
-                let xpathResult = document.evaluate(qry_f, document, null, XPathResult.ORDERED_NODE_SNAPSHOT_TYPE, null);
-                results = [];
-                for (let i = 0; i < xpathResult.snapshotLength; i++) {
-                    results.push(xpathResult.snapshotItem(i));
-                }
-            } else {
-                console.log("[css] Scanning for included filter " + f)
-                console.log("[css] Scanning for included filter " + f);
-                results = document.querySelectorAll(f);
+            if (!f.length) {
+                console.log("xpath_element_scraper: Empty filter, skipping");
+                continue;
             }
-        } catch (e) {
-            // Maybe catch DOMException and alert?
-            console.log("xpath_element_scraper: Exception selecting element from filter " + f);
-            console.log(e);
-        }
 
-        if (results != null && results.length) {
-
-            // Iterate over the results
-            results.forEach(node => {
-                // Try to resolve //something/text() back to its /something so we can atleast get the bounding box
-                try {
-                    if (typeof node.nodeName == 'string' && node.nodeName === '#text') {
-                        node = node.parentElement
+            try {
+                // is it xpath?
+                if (f.startsWith('/') || f.startsWith('xpath')) {
+                    var qry_f = f.replace(/xpath(:|\d:)/, '')
+                    console.log("[xpath] Scanning for included filter " + qry_f)
+                    let xpathResult = document.evaluate(qry_f, document, null, XPathResult.ORDERED_NODE_SNAPSHOT_TYPE, null);
+                    results = [];
+                    for (let i = 0; i < xpathResult.snapshotLength; i++) {
+                        results.push(xpathResult.snapshotItem(i));
                     }
-                } catch (e) {
-                    console.log(e)
-                    console.log("xpath_element_scraper: #text resolver")
-                }
-
-                // #1231 - IN the case XPath attribute filter is applied, we will have to traverse up and find the element.
-                if (typeof node.getBoundingClientRect == 'function') {
-                    bbox = node.getBoundingClientRect();
-                    console.log("xpath_element_scraper: Got filter element, scroll from top was " + scroll_y)
                 } else {
+                    console.log("[css] Scanning for included filter " + f)
+                    console.log("[css] Scanning for included filter " + f);
+                    results = document.querySelectorAll(f);
+                }
+            } catch (e) {
+                // Maybe catch DOMException and alert?
+                console.log("xpath_element_scraper: Exception selecting element from filter " + f);
+                console.log(e);
+            }
+
+            if (results != null && results.length) {
+
+                // Iterate over the results
+                results.forEach(node => {
+                    // Try to resolve //something/text() back to its /something so we can atleast get the bounding box
                     try {
-                        // Try and see we can find its ownerElement
-                        bbox = node.ownerElement.getBoundingClientRect();
-                        console.log("xpath_element_scraper: Got filter by ownerElement element, scroll from top was " + scroll_y)
+                        if (typeof node.nodeName == 'string' && node.nodeName === '#text') {
+                            node = node.parentElement
+                        }
                     } catch (e) {
                         console.log(e)
-                        console.log("xpath_element_scraper: error looking up q.ownerElement")
+                        console.log("xpath_element_scraper: #text resolver")
                     }
-                }
 
-                if (bbox && bbox['width'] > 0 && bbox['height'] > 0) {
-                    size_pos.push({
-                        xpath: f,
-                        width: parseInt(bbox['width']),
-                        height: parseInt(bbox['height']),
-                        left: parseInt(bbox['left']),
-                        top: parseInt(bbox['top']) + scroll_y,
-                        highlight_as_custom_filter: true
-                    });
-                }
-            });
+                    // #1231 - IN the case XPath attribute filter is applied, we will have to traverse up and find the element.
+                    if (typeof node.getBoundingClientRect == 'function') {
+                        bbox = node.getBoundingClientRect();
+                        console.log("xpath_element_scraper: Got filter element, scroll from top was " + scroll_y)
+                    } else {
+                        try {
+                            // Try and see we can find its ownerElement
+                            bbox = node.ownerElement.getBoundingClientRect();
+                            console.log("xpath_element_scraper: Got filter by ownerElement element, scroll from top was " + scroll_y)
+                        } catch (e) {
+                            console.log(e)
+                            console.log("xpath_element_scraper: error looking up q.ownerElement")
+                        }
+                    }
+
+                    if (bbox && bbox['width'] > 0 && bbox['height'] > 0) {
+                        size_pos.push({
+                            xpath: f,
+                            width: parseInt(bbox['width']),
+                            height: parseInt(bbox['height']),
+                            left: parseInt(bbox['left']),
+                            top: parseInt(bbox['top']) + scroll_y,
+                            highlight_as_custom_filter: true
+                        });
+                    }
+                });
+            }
         }
     }
-}
 
 // Sort the elements so we find the smallest one first, in other words, we find the smallest one matching in that area
 // so that we dont select the wrapping element by mistake and be unable to select what we want
-size_pos.sort((a, b) => (a.width * a.height > b.width * b.height) ? 1 : -1)
+    size_pos.sort((a, b) => (a.width * a.height > b.width * b.height) ? 1 : -1)
+
+// browser_width required for proper scaling in the frontend
+    // Return as a string to save playwright for juggling thousands of objects
+    return JSON.stringify({'size_pos': size_pos, 'browser_width': window.innerWidth});
+}
 
-// Window.width required for proper scaling in the frontend
-return {'size_pos': size_pos, 'browser_width': window.innerWidth};

changedetectionio/content_fetchers/screenshot_handler.py

@@ -0,0 +1,73 @@
+# Pages with a vertical height longer than this will use the 'stitch together' method.
+
+# - Many GPUs have a max texture size of 16384x16384px (or lower on older devices).
+# - If a page is taller than ~8000–10000px, it risks exceeding GPU memory limits.
+# - This is especially important on headless Chromium, where Playwright may fail to allocate a massive full-page buffer.
+
+from loguru import logger
+
+from changedetectionio.content_fetchers import SCREENSHOT_MAX_HEIGHT_DEFAULT, SCREENSHOT_DEFAULT_QUALITY
+
+
+def stitch_images_worker(pipe_conn, chunks_bytes, original_page_height, capture_height):
+    import os
+    import io
+    from PIL import Image, ImageDraw, ImageFont
+
+    try:
+
+        # Load images from byte chunks
+        images = [Image.open(io.BytesIO(b)) for b in chunks_bytes]
+        total_height = sum(im.height for im in images)
+        max_width = max(im.width for im in images)
+
+        # Create stitched image
+        stitched = Image.new('RGB', (max_width, total_height))
+        y_offset = 0
+        for im in images:
+            stitched.paste(im, (0, y_offset))
+            y_offset += im.height
+
+        # Draw caption on top (overlaid, not extending canvas)
+        draw = ImageDraw.Draw(stitched)
+
+
+        caption_text = f"WARNING: Screenshot was {original_page_height}px but trimmed to {capture_height}px because it was too long"
+        padding = 10
+        font_size = 35
+        font_color = (255, 0, 0)
+        background_color = (255, 255, 255)
+
+
+        # Try to load a proper font
+        try:
+            font = ImageFont.truetype("arial.ttf", font_size)
+        except IOError:
+            font = ImageFont.load_default()
+
+        bbox = draw.textbbox((0, 0), caption_text, font=font)
+        text_width = bbox[2] - bbox[0]
+        text_height = bbox[3] - bbox[1]
+
+        # Draw white rectangle background behind text
+        rect_top = 0
+        rect_bottom = text_height + 2 * padding
+        draw.rectangle([(0, rect_top), (max_width, rect_bottom)], fill=background_color)
+
+        # Draw text centered horizontally, 10px padding from top of the rectangle
+        text_x = (max_width - text_width) // 2
+        text_y = padding
+        draw.text((text_x, text_y), caption_text, font=font, fill=font_color)
+
+        # Encode and send image
+        output = io.BytesIO()
+        stitched.save(output, format="JPEG", quality=int(os.getenv("SCREENSHOT_QUALITY", SCREENSHOT_DEFAULT_QUALITY)))
+        pipe_conn.send_bytes(output.getvalue())
+
+        stitched.close()
+    except Exception as e:
+        pipe_conn.send(f"error:{e}")
+    finally:
+        pipe_conn.close()
+
+

changedetectionio/content_fetchers/webdriver_selenium.py

@@ -65,6 +65,7 @@ class fetcher(Fetcher):
         # request_body, request_method unused for now, until some magic in the future happens.
 
         options = ChromeOptions()
+        options.add_argument("--headless")
         if self.proxy:
             options.proxy = self.proxy
 
@@ -112,9 +113,9 @@ class fetcher(Fetcher):
         self.quit()
         return True
 
-    def quit(self):
+    def quit(self, watch=None):
         if self.driver:
             try:
                 self.driver.quit()
             except Exception as e:
-                logger.debug(f"Content Fetcher > Exception in chrome shutdown/quit {str(e)}")
+                logger.debug(f"Content Fetcher > Exception in chrome shutdown/quit {str(e)}")

changedetectionio/flask_app.py

@@ -33,7 +33,7 @@ from loguru import logger
 
 from changedetectionio import __version__
 from changedetectionio import queuedWatchMetaData
-from changedetectionio.api import Watch, WatchHistory, WatchSingleHistory, CreateWatch, Import, SystemInfo, Tag, Tags
+from changedetectionio.api import Watch, WatchHistory, WatchSingleHistory, CreateWatch, Import, SystemInfo, Tag, Tags, Notifications
 from changedetectionio.api.Search import Search
 from .time_handler import is_within_schedule
 
@@ -285,7 +285,8 @@ def changedetection_app(config=None, datastore_o=None):
     watch_api.add_resource(Search, '/api/v1/search',
                            resource_class_kwargs={'datastore': datastore})
 
-
+    watch_api.add_resource(Notifications, '/api/v1/notifications',
+                           resource_class_kwargs={'datastore': datastore})
 
     @login_manager.user_loader
     def user_loader(email):
@@ -394,7 +395,7 @@ def changedetection_app(config=None, datastore_o=None):
                     response.headers['Content-Type'] = 'application/json'
                     response.headers['Content-Encoding'] = 'deflate'
                 else:
-                    logger.error(f'Request elements.deflate at "{watch_directory}" but was notfound.')
+                    logger.error(f'Request elements.deflate at "{watch_directory}" but was not found.')
                     abort(404)
 
                 if response:
@@ -514,7 +515,8 @@ def notification_runner():
             sent_obj = None
 
             try:
-                from changedetectionio import notification
+                from changedetectionio.notification.handler import process_notification
+
                 # Fallback to system config if not set
                 if not n_object.get('notification_body') and datastore.data['settings']['application'].get('notification_body'):
                     n_object['notification_body'] = datastore.data['settings']['application'].get('notification_body')
@@ -524,8 +526,8 @@ def notification_runner():
 
                 if not n_object.get('notification_format') and datastore.data['settings']['application'].get('notification_format'):
                     n_object['notification_format'] = datastore.data['settings']['application'].get('notification_format')
-
-                sent_obj = notification.process_notification(n_object, datastore)
+                if n_object.get('notification_urls', {}):
+                    sent_obj = process_notification(n_object, datastore)
 
             except Exception as e:
                 logger.error(f"Watch URL: {n_object['watch_url']}  Error {str(e)}")

changedetectionio/forms.py

@@ -306,8 +306,8 @@ class ValidateAppRiseServers(object):
 
     def __call__(self, form, field):
         import apprise
-        from .apprise_plugin.assets import apprise_asset
-        from .apprise_plugin.custom_handlers import apprise_http_custom_handler  # noqa: F401
+        from .notification.apprise_plugin.assets import apprise_asset
+        from .notification.apprise_plugin.custom_handlers import apprise_http_custom_handler  # noqa: F401
 
         apobj = apprise.Apprise(asset=apprise_asset)
 
@@ -721,6 +721,8 @@ class globalSettingsRequestForm(Form):
                     self.extra_proxies.errors.append('Both a name, and a Proxy URL is required.')
                     return False
 
+class globalSettingsApplicationUIForm(Form):
+    open_diff_in_new_tab = BooleanField('Open diff page in a new tab', default=True, validators=[validators.Optional()])
 
 # datastore.data['settings']['application']..
 class globalSettingsApplicationForm(commonSettingsForm):
@@ -752,6 +754,7 @@ class globalSettingsApplicationForm(commonSettingsForm):
                                                                   render_kw={"style": "width: 5em;"},
                                                                   validators=[validators.NumberRange(min=0,
                                                                                                      message="Should contain zero or more attempts")])
+    ui = FormField(globalSettingsApplicationUIForm)
 
 
 class globalSettingsForm(Form):

changedetectionio/html_tools.py

@@ -366,22 +366,41 @@ def extract_json_as_string(content, json_filter, ensure_is_ldjson_info_type=None
 # wordlist - list of regex's (str) or words (str)
 # Preserves all linefeeds and other whitespacing, its not the job of this to remove that
 def strip_ignore_text(content, wordlist, mode="content"):
-    i = 0
-    output = []
     ignore_text = []
     ignore_regex = []
-    ignored_line_numbers = []
+    ignore_regex_multiline = []
+    ignored_lines = []
 
     for k in wordlist:
         # Is it a regex?
         res = re.search(PERL_STYLE_REGEX, k, re.IGNORECASE)
         if res:
-            ignore_regex.append(re.compile(perl_style_slash_enclosed_regex_to_options(k)))
+            res = re.compile(perl_style_slash_enclosed_regex_to_options(k))
+            if res.flags & re.DOTALL or res.flags & re.MULTILINE:
+                ignore_regex_multiline.append(res)
+            else:
+                ignore_regex.append(res)
         else:
             ignore_text.append(k.strip())
 
-    for line in content.splitlines(keepends=True):
-        i += 1
+    for r in ignore_regex_multiline:
+        for match in r.finditer(content):
+            content_lines = content[:match.end()].splitlines(keepends=True)
+            match_lines = content[match.start():match.end()].splitlines(keepends=True)
+
+            end_line = len(content_lines)
+            start_line = end_line - len(match_lines)
+
+            if end_line - start_line <= 1:
+                # Match is empty or in the middle of the line
+                ignored_lines.append(start_line)
+            else:
+                for i in range(start_line, end_line):
+                    ignored_lines.append(i)
+
+    line_index = 0
+    lines = content.splitlines(keepends=True)
+    for line in lines:
         # Always ignore blank lines in this mode. (when this function gets called)
         got_match = False
         for l in ignore_text:
@@ -393,17 +412,19 @@ def strip_ignore_text(content, wordlist, mode="content"):
                 if r.search(line):
                     got_match = True
 
-        if not got_match:
-            # Not ignored, and should preserve "keepends"
-            output.append(line)
-        else:
-            ignored_line_numbers.append(i)
+        if got_match:
+            ignored_lines.append(line_index)
+
+        line_index += 1
+
+    ignored_lines = set([i for i in ignored_lines if i >= 0 and i < len(lines)])
 
     # Used for finding out what to highlight
     if mode == "line numbers":
-        return ignored_line_numbers
+        return [i + 1 for i in ignored_lines]
 
-    return ''.join(output)
+    output_lines = set(range(len(lines))) - ignored_lines
+    return ''.join([lines[i] for i in output_lines])
 
 def cdata_in_document_to_text(html_content: str, render_anchor_tag_content=False) -> str:
     from xml.sax.saxutils import escape as xml_escape
@@ -414,7 +435,9 @@ def cdata_in_document_to_text(html_content: str, render_anchor_tag_content=False
 
     return re.sub(pattern, repl, html_content)
 
-def html_to_text(html_content: str, render_anchor_tag_content=False, is_rss=False) -> str:
+
+def html_to_text_sub_worker(conn, html_content: str, render_anchor_tag_content=False, is_rss=False):
+
     from inscriptis import get_text
     from inscriptis.model.config import ParserConfig
 
@@ -449,15 +472,27 @@ def html_to_text(html_content: str, render_anchor_tag_content=False, is_rss=Fals
         html_content = re.sub(r'</title>', r'</h1>', html_content)
 
     text_content = get_text(html_content, config=parser_config)
+    conn.send(text_content)
+    conn.close()
 
-    return text_content
+# NOTE!! ANYTHING LIBXML, HTML5LIB ETC WILL CAUSE SOME SMALL MEMORY LEAK IN THE LOCAL "LIB" IMPLEMENTATION OUTSIDE PYTHON
+def html_to_text(html_content: str, render_anchor_tag_content=False, is_rss=False):
+    from multiprocessing import Process, Pipe
 
+    parent_conn, child_conn = Pipe()
+    p = Process(target=html_to_text_sub_worker, args=(child_conn, html_content, render_anchor_tag_content, is_rss))
+    p.start()
+    text = parent_conn.recv()
+    p.join()
+    return text
 
 # Does LD+JSON exist with a @type=='product' and a .price set anywhere?
 def has_ldjson_product_info(content):
     try:
-        lc = content.lower()
-        if 'application/ld+json' in lc and lc.count('"price"') == 1 and '"pricecurrency"' in lc:
+        # Better than .lower() which can use a lot of ram
+        if (re.search(r'application/ld\+json', content, re.IGNORECASE) and
+            re.search(r'"price"', content, re.IGNORECASE) and
+            re.search(r'"pricecurrency"', content, re.IGNORECASE)):
             return True
 
 #       On some pages this is really terribly expensive when they dont really need it

changedetectionio/model/App.py

@@ -60,6 +60,9 @@ class model(dict):
                     'webdriver_delay': None , # Extra delay in seconds before extracting text
                     'tags': {}, #@todo use Tag.model initialisers
                     'timezone': None, # Default IANA timezone name
+                    'ui': {
+                        'open_diff_in_new_tab': True,
+                    },
                 }
             }
         }

changedetectionio/model/Watch.py

@@ -553,7 +553,10 @@ class model(watch_base):
         self.ensure_data_dir_exists()
 
         with open(target_path, 'wb') as f:
-            f.write(zlib.compress(json.dumps(data).encode()))
+            if not isinstance(data, str):
+                f.write(zlib.compress(json.dumps(data).encode()))
+            else:
+                f.write(zlib.compress(data.encode()))
             f.close()
 
     # Save as PNG, PNG is larger but better for doing visual diff in the future

changedetectionio/model/__init__.py

@@ -2,7 +2,7 @@ import os
 import uuid
 
 from changedetectionio import strtobool
-from changedetectionio.notification import default_notification_format_for_watch
+default_notification_format_for_watch = 'System default'
 
 class watch_base(dict):
 

changedetectionio/notification/__init__.py

@@ -0,0 +1,35 @@
+from changedetectionio.model import default_notification_format_for_watch
+
+ult_notification_format_for_watch = 'System default'
+default_notification_format = 'HTML Color'
+default_notification_body = '{{watch_url}} had a change.\n---\n{{diff}}\n---\n'
+default_notification_title = 'ChangeDetection.io Notification - {{watch_url}}'
+
+# The values (markdown etc) are from apprise NotifyFormat,
+# But to avoid importing the whole heavy module just use the same strings here.
+valid_notification_formats = {
+    'Text': 'text',
+    'Markdown': 'markdown',
+    'HTML': 'html',
+    'HTML Color': 'htmlcolor',
+    # Used only for editing a watch (not for global)
+    default_notification_format_for_watch: default_notification_format_for_watch
+}
+
+
+valid_tokens = {
+    'base_url': '',
+    'current_snapshot': '',
+    'diff': '',
+    'diff_added': '',
+    'diff_full': '',
+    'diff_patch': '',
+    'diff_removed': '',
+    'diff_url': '',
+    'preview_url': '',
+    'triggered_text': '',
+    'watch_tag': '',
+    'watch_title': '',
+    'watch_url': '',
+    'watch_uuid': '',
+}

changedetectionio/notification/handler.py

@@ -1,47 +1,17 @@
 
 import time
-from apprise import NotifyFormat
 import apprise
 from loguru import logger
 
-from .apprise_plugin.assets import APPRISE_AVATAR_URL
-from .apprise_plugin.custom_handlers import apprise_http_custom_handler  # noqa: F401
-from .safe_jinja import render as jinja_render
-
-valid_tokens = {
-    'base_url': '',
-    'current_snapshot': '',
-    'diff': '',
-    'diff_added': '',
-    'diff_full': '',
-    'diff_patch': '',
-    'diff_removed': '',
-    'diff_url': '',
-    'preview_url': '',
-    'triggered_text': '',
-    'watch_tag': '',
-    'watch_title': '',
-    'watch_url': '',
-    'watch_uuid': '',
-}
-
-default_notification_format_for_watch = 'System default'
-default_notification_format = 'HTML Color'
-default_notification_body = '{{watch_url}} had a change.\n---\n{{diff}}\n---\n'
-default_notification_title = 'ChangeDetection.io Notification - {{watch_url}}'
-
-valid_notification_formats = {
-    'Text': NotifyFormat.TEXT,
-    'Markdown': NotifyFormat.MARKDOWN,
-    'HTML': NotifyFormat.HTML,
-    'HTML Color': 'htmlcolor',
-    # Used only for editing a watch (not for global)
-    default_notification_format_for_watch: default_notification_format_for_watch
-}
-
+from .apprise_plugin.assets import apprise_asset, APPRISE_AVATAR_URL
 
 
 def process_notification(n_object, datastore):
+    from changedetectionio.safe_jinja import render as jinja_render
+    from . import default_notification_format_for_watch, default_notification_format, valid_notification_formats
+    # be sure its registered
+    from .apprise_plugin.custom_handlers import apprise_http_custom_handler
+
     now = time.time()
     if n_object.get('notification_timestamp'):
         logger.trace(f"Time since queued {now-n_object['notification_timestamp']:.3f}s")
@@ -58,14 +28,13 @@ def process_notification(n_object, datastore):
         # Initially text or whatever
         n_format = datastore.data['settings']['application'].get('notification_format', valid_notification_formats[default_notification_format])
 
-    logger.trace(f"Complete notification body including Jinja and placeholders calculated in  {time.time() - now:.3f}s")
+    logger.trace(f"Complete notification body including Jinja and placeholders calculated in  {time.time() - now:.2f}s")
 
     # https://github.com/caronc/apprise/wiki/Development_LogCapture
     # Anything higher than or equal to WARNING (which covers things like Connection errors)
     # raise it as an exception
 
     sent_objs = []
-    from .apprise_plugin.assets import apprise_asset
 
     if 'as_async' in n_object:
         apprise_asset.async_mode = n_object.get('as_async')
@@ -176,6 +145,7 @@ def process_notification(n_object, datastore):
 # ( Where we prepare the tokens in the notification to be replaced with actual values )
 def create_notification_parameters(n_object, datastore):
     from copy import deepcopy
+    from . import valid_tokens
 
     # in the case we send a test notification from the main settings, there is no UUID.
     uuid = n_object['uuid'] if 'uuid' in n_object else ''

changedetectionio/processors/__init__.py

@@ -159,7 +159,7 @@ class difference_detection_processor():
                          )
 
         #@todo .quit here could go on close object, so we can run JS if change-detected
-        self.fetcher.quit()
+        self.fetcher.quit(watch=self.watch)
 
         # After init, call run_changedetection() which will do the actual change-detection
 

changedetectionio/run_basic_tests.sh

@@ -14,7 +14,8 @@ SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
 find tests/test_*py -type f|while read test_name
 do
   echo "TEST RUNNING $test_name"
-  pytest $test_name
+  # REMOVE_REQUESTS_OLD_SCREENSHOTS disabled so that we can write a screenshot and send it in test_notifications.py without a real browser
+  REMOVE_REQUESTS_OLD_SCREENSHOTS=false pytest $test_name
 done
 
 echo "RUNNING WITH BASE_URL SET"
@@ -22,7 +23,7 @@ echo "RUNNING WITH BASE_URL SET"
 # Now re-run some tests with BASE_URL enabled
 # Re #65 - Ability to include a link back to the installation, in the notification.
 export BASE_URL="https://really-unique-domain.io"
-pytest tests/test_notification.py
+REMOVE_REQUESTS_OLD_SCREENSHOTS=false pytest tests/test_notification.py
 
 
 # Re-run with HIDE_REFERER set - could affect login
@@ -32,7 +33,7 @@ pytest tests/test_access_control.py
 # Re-run a few tests that will trigger brotli based storage
 export SNAPSHOT_BROTLI_COMPRESSION_THRESHOLD=5
 pytest tests/test_access_control.py
-pytest tests/test_notification.py
+REMOVE_REQUESTS_OLD_SCREENSHOTS=false pytest tests/test_notification.py
 pytest tests/test_backend.py
 pytest tests/test_rss.py
 pytest tests/test_unique_lines.py

changedetectionio/store.py

@@ -964,3 +964,25 @@ class ChangeDetectionStore:
                         f_d.write(zlib.compress(f_j.read()))
                         os.unlink(json_path)
 
+    def add_notification_url(self, notification_url):
+        
+        logger.debug(f">>> Adding new notification_url - '{notification_url}'")
+
+        notification_urls = self.data['settings']['application'].get('notification_urls', [])
+
+        if notification_url in notification_urls:
+            return notification_url
+
+        with self.lock:
+            notification_urls = self.__data['settings']['application'].get('notification_urls', [])
+
+            if notification_url in notification_urls:
+                return notification_url
+
+            # Append and update the datastore
+            notification_urls.append(notification_url)
+            self.__data['settings']['application']['notification_urls'] = notification_urls
+            self.needs_write = True
+
+        return notification_url
+

changedetectionio/templates/edit.html

@@ -443,6 +443,10 @@ Math: {{ 1 + 1 }}") }}
                         </tr>
                         </tbody>
                     </table>
+
+                    <h4>Text similarity</h4>
+                    <p><strong>Levenshtein Distance</strong> - Last 2 snapshots: {{ lev_info }}</p>
+                    <p style="max-width: 80%; font-size: 80%"><strong>Levenshtein Distance</strong> Calculates the minimum number of insertions, deletions, and substitutions required to change one text into the other.</p>
                     {% 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>

changedetectionio/tests/test_api_notifications.py

@@ -0,0 +1,108 @@
+#!/usr/bin/env python3
+
+from flask import url_for
+from .util import live_server_setup
+import json
+
+def test_api_notifications_crud(client, live_server):
+    live_server_setup(live_server)
+    api_key = live_server.app.config['DATASTORE'].data['settings']['application'].get('api_access_token')
+
+    # Confirm notifications are initially empty
+    res = client.get(
+        url_for("notifications"),
+        headers={'x-api-key': api_key}
+    )
+    assert res.status_code == 200
+    assert res.json == {"notification_urls": []}
+
+    # Add notification URLs
+    test_urls = ["posts://example.com/notify1", "posts://example.com/notify2"]
+    res = client.post(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": test_urls}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 201
+    for url in test_urls:
+        assert url in res.json["notification_urls"]
+
+    # Confirm the notification URLs were added
+    res = client.get(
+        url_for("notifications"),
+        headers={'x-api-key': api_key}
+    )
+    assert res.status_code == 200
+    for url in test_urls:
+        assert url in res.json["notification_urls"]
+
+    # Delete one notification URL
+    res = client.delete(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": [test_urls[0]]}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 204
+
+    # Confirm it was removed and the other remains
+    res = client.get(
+        url_for("notifications"),
+        headers={'x-api-key': api_key}
+    )
+    assert res.status_code == 200
+    assert test_urls[0] not in res.json["notification_urls"]
+    assert test_urls[1] in res.json["notification_urls"]
+
+    # Try deleting a non-existent URL
+    res = client.delete(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": ["posts://nonexistent.com"]}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 400
+
+    res = client.post(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": test_urls}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 201
+
+    # Replace with a new list
+    replacement_urls = ["posts://new.example.com"]
+    res = client.put(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": replacement_urls}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 200
+    assert res.json["notification_urls"] == replacement_urls
+
+    # Replace with an empty list
+    res = client.put(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": []}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 200
+    assert res.json["notification_urls"] == []
+
+    # Provide an invalid AppRise URL to trigger validation error
+    invalid_urls = ["ftp://not-app-rise"]
+    res = client.post(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": invalid_urls}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 400
+    assert "is not a valid AppRise URL." in res.data.decode()
+
+    res = client.put(
+        url_for("notifications"),
+        data=json.dumps({"notification_urls": invalid_urls}),
+        headers={'content-type': 'application/json', 'x-api-key': api_key}
+    )
+    assert res.status_code == 400
+    assert "is not a valid AppRise URL." in res.data.decode()
+
+    

changedetectionio/tests/test_backend.py

@@ -74,6 +74,11 @@ def test_check_basic_change_detection_functionality(client, live_server, measure
     res = client.get(url_for("ui.ui_edit.watch_get_latest_html", uuid=uuid))
     assert b'which has this one new line' in res.data
 
+    # Check the 'levenshtein' distance calc showed something useful
+    res = client.get(url_for("ui.ui_edit.edit_page", uuid=uuid))
+    assert b'Last 2 snapshots: 17' in res.data
+
+
     # Now something should be ready, indicated by having a 'unviewed' class
     res = client.get(url_for("watchlist.index"))
     assert b'unviewed' in res.data

changedetectionio/tests/test_ignore_regex_text.py

@@ -32,7 +32,6 @@ def test_strip_regex_text_func():
     ]
 
     stripped_content = html_tools.strip_ignore_text(test_content, ignore_lines)
-
     assert "but 1 lines" in stripped_content
     assert "igNORe-cAse text" not in stripped_content
     assert "but 1234 lines" not in stripped_content
@@ -42,6 +41,46 @@ def test_strip_regex_text_func():
     # Check line number reporting
     stripped_content = html_tools.strip_ignore_text(test_content, ignore_lines, mode="line numbers")
     assert stripped_content == [2, 5, 6, 7, 8, 10]
+    
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/but 1.+5 lines/s'])
+    assert "but 1 lines" not in stripped_content
+    assert "skip 5 lines" not in stripped_content
+    
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/but 1.+5 lines/s'], mode="line numbers")
+    assert stripped_content == [4, 5]
+    
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/.+/s'])
+    assert stripped_content == ""
+    
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/.+/s'], mode="line numbers")
+    assert stripped_content == [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]
+
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/^.+but.+\\n.+lines$/m'])
+    assert "but 1 lines" not in stripped_content
+    assert "skip 5 lines" not in stripped_content
+
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/^.+but.+\\n.+lines$/m'], mode="line numbers")
+    assert stripped_content == [4, 5]
+
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/^.+?\.$/m'])
+    assert "but sometimes we want to remove the lines." not in stripped_content
+    assert "but not always." not in stripped_content
+
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/^.+?\.$/m'], mode="line numbers")
+    assert stripped_content == [2, 11]
+
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/but.+?but/ms'])
+    assert "but sometimes we want to remove the lines." not in stripped_content
+    assert "but 1 lines" not in stripped_content
+    assert "but 1234 lines" not in stripped_content
+    assert "igNORe-cAse text we dont want to keep" not in stripped_content
+    assert "but not always." not in stripped_content
+
+    stripped_content = html_tools.strip_ignore_text(test_content, ['/but.+?but/ms'], mode="line numbers")
+    assert stripped_content == [2, 3, 4, 9, 10, 11]
+
+    stripped_content = html_tools.strip_ignore_text("\n\ntext\n\ntext\n\n", ['/^$/ms'], mode="line numbers")
+    assert stripped_content == [1, 2, 4, 6]
 
     # Check that linefeeds are preserved when there are is no matching ignores
     content = "some text\n\nand other text\n"

changedetectionio/tests/test_notification.py

@@ -167,7 +167,10 @@ def test_check_notification(client, live_server, measure_memory_usage):
     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
+
     # 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'
     assert len(notification_submission_object['attachments'][0]['base64'])
     assert notification_submission_object['attachments'][0]['mimetype'] == 'image/png'

changedetectionio/tests/test_ui.py

@@ -0,0 +1,80 @@
+#!/usr/bin/env python3
+
+from flask import url_for
+from .util import set_original_response, set_modified_response, live_server_setup, wait_for_all_checks
+
+def test_checkbox_open_diff_in_new_tab(client, live_server):
+    
+    set_original_response()
+    live_server_setup(live_server)
+
+    # Add our URL to the import page
+    res = client.post(
+        url_for("imports.import_page"),
+        data={"urls": url_for('test_endpoint', _external=True)},
+        follow_redirects=True
+    )
+
+    assert b"1 Imported" in res.data
+    wait_for_all_checks(client)
+
+    # Make a change
+    set_modified_response()
+
+    # Test case 1 - checkbox is enabled in settings
+    res = client.post(
+        url_for("settings.settings_page"),
+        data={"application-ui-open_diff_in_new_tab": "1"},
+        follow_redirects=True
+    )
+    assert b'Settings updated' in res.data
+
+    # Force recheck
+    res = client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    assert b'Queued 1 watch for rechecking.' in res.data
+
+    wait_for_all_checks(client)
+    
+    res = client.get(url_for("watchlist.index"))
+    lines = res.data.decode().split("\n")
+
+    # Find link to diff page
+    target_line = None
+    for line in lines:
+        if '/diff' in line:
+            target_line = line.strip()
+            break
+
+    assert target_line != None
+    assert 'target=' in target_line
+
+    # Test case 2 - checkbox is disabled in settings
+    res = client.post(
+        url_for("settings.settings_page"),
+        data={"application-ui-open_diff_in_new_tab": ""},
+        follow_redirects=True
+    )
+    assert b'Settings updated' in res.data
+
+    # Force recheck
+    res = client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
+    assert b'Queued 1 watch for rechecking.' in res.data
+
+    wait_for_all_checks(client)
+    
+    res = client.get(url_for("watchlist.index"))
+    lines = res.data.decode().split("\n")
+
+    # Find link to diff page
+    target_line = None
+    for line in lines:
+        if '/diff' in line:
+            target_line = line.strip()
+            break
+
+    assert target_line != None
+    assert 'target=' not in target_line
+
+    # Cleanup everything
+    res = client.get(url_for("ui.form_delete", uuid="all"), follow_redirects=True)
+    assert b'Deleted' in res.data

changedetectionio/update_worker.py

@@ -109,7 +109,6 @@ class update_worker(threading.Thread):
             default_notification_title
         )
 
-
         # Would be better if this was some kind of Object where Watch can reference the parent datastore etc
         v = watch.get(var_name)
         if v and not watch.get('notification_muted'):
@@ -592,6 +591,7 @@ class update_worker(threading.Thread):
 
                 self.current_uuid = None  # Done
                 self.q.task_done()
+                update_handler = None
                 logger.debug(f"Watch {uuid} done in {time.time()-fetch_start_time:.2f}s")
 
                 # Give the CPU time to interrupt

docker-compose.yml

@@ -63,6 +63,10 @@ services:
   #
   #        A valid timezone name to run as (for scheduling watch checking) see https://en.wikipedia.org/wiki/List_of_tz_database_time_zones
   #      - TZ=America/Los_Angeles
+  #
+  #        Maximum height of screenshots, default is 16000 px, screenshots will be clipped to this if exceeded.
+  #        RAM usage will be higher if you increase this.
+  #      - SCREENSHOT_MAX_HEIGHT=16000
   
       # Comment out ports: when using behind a reverse proxy , enable networks: etc.
       ports:

requirements.txt

@@ -68,6 +68,8 @@ openpyxl
 jq~=1.3; python_version >= "3.8" and sys_platform == "darwin"
 jq~=1.3; python_version >= "3.8" and sys_platform == "linux"
 
+levenshtein
+
 # playwright is installed at Dockerfile build time because it's not available on all platforms
 
 pyppeteer-ng==2.0.0rc9