Update selenium library

Co-authored-by: silversub <silversub@gmail.com>

changedetectionio/conditions/__init__.py

@@ -5,7 +5,7 @@ from json_logic.builtins import BUILTINS
 from .exceptions import EmptyConditionRuleRowNotUsable
 from .pluggy_interface import plugin_manager  # Import the pluggy plugin manager
 from . import default_plugin
-
+from loguru import logger
 # List of all supported JSON Logic operators
 operator_choices = [
     (None, "Choose one - Operator"),
@@ -113,12 +113,14 @@ def execute_ruleset_against_all_plugins(current_watch_uuid: str, application_dat
                             application_datastruct=application_datastruct,
                             ephemeral_data=ephemeral_data
                         )
-                        
+                        logger.debug(f"Trying plugin {plugin}....")
+
                         # Set a timeout of 10 seconds
                         try:
                             new_execute_data = future.result(timeout=10)
                             if new_execute_data and isinstance(new_execute_data, dict):
                                 EXECUTE_DATA.update(new_execute_data)
+
                         except concurrent.futures.TimeoutError:
                             # The plugin took too long, abort processing for this watch
                             raise Exception(f"Plugin {plugin.__class__.__name__} took more than 10 seconds to run.")

changedetectionio/conditions/plugins/levenshtein_plugin.py

@@ -9,15 +9,20 @@ def levenshtein_ratio_recent_history(watch, incoming_text=None):
     try:
         from Levenshtein import ratio, distance
         k = list(watch.history.keys())
-        if len(k) >= 2:
-            # When called from ui_edit_stats_extras, we don't have incoming_text
-            if incoming_text is None:
-                a = watch.get_history_snapshot(timestamp=k[-1])  # Latest snapshot
-                b = watch.get_history_snapshot(timestamp=k[-2])  # Previous snapshot
-            else:
-                a = watch.get_history_snapshot(timestamp=k[-2]) # Second newest, incoming_text will be "newest"
-                b = incoming_text
-            
+        a = None
+        b = None
+
+        # When called from ui_edit_stats_extras, we don't have incoming_text
+        if incoming_text is None:
+            a = watch.get_history_snapshot(timestamp=k[-1])  # Latest snapshot
+            b = watch.get_history_snapshot(timestamp=k[-2])  # Previous snapshot
+
+        # Needs atleast one snapshot
+        elif len(k) >= 1: # Should be atleast one snapshot to compare against
+            a = watch.get_history_snapshot(timestamp=k[-1]) # Latest saved snapshot
+            b = incoming_text if incoming_text else k[-2]
+
+        if a and b:
             distance_value = distance(a, b)
             ratio_value = ratio(a, b)
             return {
@@ -53,7 +58,7 @@ def add_data(current_watch_uuid, application_datastruct, ephemeral_data):
     # ephemeral_data['text'] will be the current text after filters, they may have edited filters but not saved them yet etc
 
     if watch and 'text' in ephemeral_data:
-        lev_data = levenshtein_ratio_recent_history(watch, ephemeral_data['text'])
+        lev_data = levenshtein_ratio_recent_history(watch, ephemeral_data.get('text',''))
         if isinstance(lev_data, dict):
             res['levenshtein_ratio'] = lev_data.get('ratio', 0)
             res['levenshtein_similarity'] = lev_data.get('percent_similar', 0)

changedetectionio/content_fetchers/puppeteer.py

@@ -147,7 +147,7 @@ class fetcher(Fetcher):
                          is_binary,
                          empty_pages_are_a_change
                          ):
-
+        import re
         self.delete_browser_steps_screenshots()
         extra_wait = int(os.getenv("WEBDRIVER_DELAY_BEFORE_CONTENT_READY", 5)) + self.render_extract_delay
 
@@ -172,6 +172,17 @@ class fetcher(Fetcher):
         # headless - ask a new page
         self.page = (pages := await browser.pages) and len(pages) or await browser.newPage()
 
+        if '--window-size' in self.browser_connection_url:
+            # Be sure the viewport is always the window-size, this is often not the same thing
+            match = re.search(r'--window-size=(\d+),(\d+)', self.browser_connection_url)
+            if match:
+                logger.debug(f"Setting viewport to same as --window-size in browser connection URL {int(match.group(1))},{int(match.group(2))}")
+                await self.page.setViewport({
+                    "width": int(match.group(1)),
+                    "height": int(match.group(2))
+                })
+                logger.debug(f"Puppeteer viewport size {self.page.viewport}")
+
         try:
             from pyppeteerstealth import inject_evasions_into_page
         except ImportError:
@@ -218,7 +229,6 @@ class fetcher(Fetcher):
 
         response = await self.page.goto(url, waitUntil="load")
 
-
         if response is None:
             await self.page.close()
             await browser.close()

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

@@ -51,6 +51,7 @@ async () => {
             'niet op voorraad',
             'no disponible',
             'no featured offers available',
+            'no longer available',
             'no longer in stock',
             'no tickets available',
             'non disponibile',
@@ -125,6 +126,20 @@ async () => {
         // 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
 
+        function elementIsInEyeBallRange(element) {
+            // 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
+            // Skip text that could be in the header area
+            if (element.getBoundingClientRect().bottom + window.scrollY <= 300 ) {
+                return false;
+            }
+            // Skip text that could be much further down (like a list of "you may like" products that have 'sold out' in there
+            if (element.getBoundingClientRect().bottom + window.scrollY >= 1300 ) {
+                return false;
+            }
+            return true;
+        }
 
 // @todo - if it's SVG or IMG, go into image diff mode
 
@@ -161,9 +176,7 @@ async () => {
         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) {
+            if (!elementIsInEyeBallRange(element)) {
                 continue
             }
 
@@ -177,11 +190,11 @@ async () => {
             } 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 "${elementText}"`)
+                    element.style.border = "2px solid green"; // highlight the element that was detected as in stock
                     return 'Possibly in stock';
                 }
             }
@@ -190,10 +203,8 @@ async () => {
         // OTHER STUFF THAT COULD BE THAT IT'S OUT OF STOCK
         for (let i = elementsToScan.length - 1; i >= 0; i--) {
             const element = elementsToScan[i];
-            // outside the 'fold' or some weird text in the heading area
-            // .getBoundingClientRect() was causing a crash in chrome 119, can only be run on contentVisibility != hidden
-            // Note: theres also an automated test that places the 'out of stock' text fairly low down
-            if (element.getBoundingClientRect().top + window.scrollY >= vh + 250 || element.getBoundingClientRect().top + window.scrollY <= 100) {
+
+            if (!elementIsInEyeBallRange(element)) {
                 continue
             }
             elementText = "";
@@ -208,6 +219,7 @@ async () => {
                 for (const outOfStockText of outOfStockTexts) {
                     if (elementText.includes(outOfStockText)) {
                         console.log(`Selected 'Out of Stock' - found text "${outOfStockText}" - "${elementText}" - offset top ${element.getBoundingClientRect().top}, page height is ${vh}`)
+                        element.style.border = "2px solid red"; // highlight the element that was detected as out of stock
                         return outOfStockText; // item is out of stock
                     }
                 }

changedetectionio/content_fetchers/res/xpath_element_scraper.js

@@ -202,7 +202,6 @@ async (options) => {
         // 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");
@@ -255,7 +254,7 @@ async (options) => {
                             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")
+                            console.log("xpath_element_scraper: error looking up node.ownerElement")
                         }
                     }
 

changedetectionio/tests/restock/test_restock.py

@@ -14,6 +14,8 @@ from changedetectionio.notification import (
 def set_original_response():
     test_return_data = """<html>
        <body>
+       <section id=header style="padding: 50px; height: 350px">This is the header which should be ignored always - <span>add to cart</span></section>
+       <!-- stock-not-in-stock.js will ignore text in the first 300px, see elementIsInEyeBallRange(), sometimes "add to cart" and other junk is here -->
      Some initial text<br>
      <p>Which is across multiple lines</p>
      <br>
@@ -52,8 +54,6 @@ def test_restock_detection(client, live_server, measure_memory_usage):
 
     set_original_response()
     #assert os.getenv('PLAYWRIGHT_DRIVER_URL'), "Needs PLAYWRIGHT_DRIVER_URL set for this test"
-
-    time.sleep(1)
     live_server_setup(live_server)
     #####################
     notification_url = url_for('test_notification_endpoint', _external=True).replace('http://localhost', 'http://changedet').replace('http', 'json')
@@ -84,7 +84,8 @@ def test_restock_detection(client, live_server, measure_memory_usage):
     # Is it correctly show as NOT in stock?
     wait_for_all_checks(client)
     res = client.get(url_for("watchlist.index"))
-    assert b'not-in-stock' in res.data
+    assert b'processor-restock_diff' in res.data # Should have saved in restock mode
+    assert b'not-in-stock' in res.data # should be out of stock
 
     # Is it correctly shown as in stock
     set_back_in_stock_response()

changedetectionio/tests/test_conditions.py

@@ -196,7 +196,11 @@ def test_condition_validate_rule_row(client, live_server):
     )
     assert res.status_code == 200
     assert b'false' in res.data
-
+    # cleanup for the next
+    client.get(
+        url_for("ui.form_delete", uuid="all"),
+        follow_redirects=True
+    )
 
 
 
@@ -235,4 +239,107 @@ def test_wordcount_conditions_plugin(client, live_server, measure_memory_usage):
     )
 
     # Assert the word count is counted correctly
-    assert b'<td>13</td>' in res.data
+    assert b'<td>13</td>' in res.data
+
+    # cleanup for the next
+    client.get(
+        url_for("ui.form_delete", uuid="all"),
+        follow_redirects=True
+    )
+
+# If there was only a change in the whitespacing, then we shouldnt have a change detected
+def test_lev_conditions_plugin(client, live_server, measure_memory_usage):
+    #live_server_setup(live_server)
+
+    with open("test-datastore/endpoint-content.txt", "w") as f:
+        f.write("""<html>
+       <body>
+     Some initial text<br>
+     <p>Which is across multiple lines</p>
+     <br>
+     So let's see what happens.  <br>
+     </body>
+     </html>
+    """)
+
+    # Add our URL to the import page
+    test_url = url_for('test_endpoint', _external=True)
+    res = client.post(
+        url_for("ui.ui_views.form_quick_watch_add"),
+        data={"url": test_url, "tags": '', 'edit_and_watch_submit_button': 'Edit > Watch'},
+        follow_redirects=True
+    )
+    assert b"Watch added in Paused state, saving will unpause" in res.data
+
+    uuid = next(iter(live_server.app.config['DATASTORE'].data['watching']))
+    # Give the thread time to pick it up
+    wait_for_all_checks(client)
+    res = client.post(
+        url_for("ui.ui_edit.edit_page", uuid=uuid, unpause_on_save=1),
+        data={
+            "url": test_url,
+            "fetch_backend": "html_requests",
+            "conditions_match_logic": "ALL",  # ALL = AND logic
+            "conditions-0-field": "levenshtein_ratio",
+            "conditions-0-operator": "<",
+            "conditions-0-value": "0.8" # needs to be more of a diff to trigger a change
+        },
+        follow_redirects=True
+    )
+
+    assert b"unpaused" in res.data
+
+    wait_for_all_checks(client)
+    res = client.get(url_for("watchlist.index"))
+    assert b'unviewed' not in res.data
+
+    # Check the content saved initially, even tho a condition was set - this is the first snapshot so shouldnt be affected by conditions
+    res = client.get(
+        url_for("ui.ui_views.preview_page", uuid=uuid),
+        follow_redirects=True
+    )
+    assert b'Which is across multiple lines' in res.data
+
+
+    ############### Now change it a LITTLE bit...
+    with open("test-datastore/endpoint-content.txt", "w") as f:
+        f.write("""<html>
+       <body>
+     Some initial text<br>
+     <p>Which is across multiple lines</p>
+     <br>
+     So let's see what happenxxxxxxxxx.  <br>
+     </body>
+     </html>
+    """)
+
+    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"))
+    assert b'unviewed' not in res.data #because this will be like 0.90 not 0.8 threshold
+
+    ############### Now change it a MORE THAN 50%
+    test_return_data = """<html>
+       <body>
+     Some sxxxx<br>
+     <p>Which is across a lines</p>
+     <br>
+     ok.  <br>
+     </body>
+     </html>
+    """
+
+    with open("test-datastore/endpoint-content.txt", "w") as f:
+        f.write(test_return_data)
+    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"))
+    assert b'unviewed' in res.data
+    # cleanup for the next
+    client.get(
+        url_for("ui.form_delete", uuid="all"),
+        follow_redirects=True
+    )

changedetectionio/tests/test_ignore_text.py

@@ -32,13 +32,14 @@ def test_strip_text_func():
     stripped_content = html_tools.strip_ignore_text(test_content, ignore)
     assert stripped_content == "Some initial text\n\nWhich is across multiple lines\n\n\n\nSo let's see what happens."
 
-def set_original_ignore_response():
-    test_return_data = """<html>
+def set_original_ignore_response(ver_stamp="123"):
+    test_return_data = f"""<html>
        <body>
      Some initial text<br>
      <p>Which is across multiple lines</p>
      <br>
      So let's see what happens.  <br>
+     <link href="https://www.somesite/wp-content/themes/cooltheme/style2.css?v={ver_stamp}" rel="stylesheet"/>
      </body>
      </html>
 
@@ -48,13 +49,14 @@ def set_original_ignore_response():
         f.write(test_return_data)
 
 
-def set_modified_original_ignore_response():
-    test_return_data = """<html>
+def set_modified_original_ignore_response(ver_stamp="123"):
+    test_return_data = f"""<html>
        <body>
      Some NEW nice initial text<br>
      <p>Which is across multiple lines</p>
      <br>
      So let's see what happens.  <br>
+     <link href="https://www.somesite/wp-content/themes/cooltheme/style2.css?v={ver_stamp}" rel="stylesheet"/>
      <p>new ignore stuff</p>
      <p>blah</p>
      </body>
@@ -67,14 +69,15 @@ def set_modified_original_ignore_response():
 
 
 # Is the same but includes ZZZZZ, 'ZZZZZ' is the last line in ignore_text
-def set_modified_ignore_response():
-    test_return_data = """<html>
+def set_modified_ignore_response(ver_stamp="123"):
+    test_return_data = f"""<html>
        <body>
      Some initial text<br>
      <p>Which is across multiple lines</p>
      <P>ZZZZz</P>
      <br>
      So let's see what happens.  <br>
+     <link href="https://www.somesite/wp-content/themes/cooltheme/style2.css?v={ver_stamp}" rel="stylesheet"/>
      </body>
      </html>
 
@@ -165,9 +168,9 @@ def test_check_ignore_text_functionality(client, live_server, measure_memory_usa
     assert b'Deleted' in res.data
 
 # When adding some ignore text, it should not trigger a change, even if something else on that line changes
-def test_check_global_ignore_text_functionality(client, live_server, measure_memory_usage):
-    #live_server_setup(live_server)
-    ignore_text = "XXXXX\r\nYYYYY\r\nZZZZZ"
+def _run_test_global_ignore(client, as_source=False, extra_ignore=""):
+    ignore_text = "XXXXX\r\nYYYYY\r\nZZZZZ\r\n"+extra_ignore
+
     set_original_ignore_response()
 
     # Goto the settings page, add our ignore text
@@ -186,6 +189,10 @@ def test_check_global_ignore_text_functionality(client, live_server, measure_mem
 
     # Add our URL to the import page
     test_url = url_for('test_endpoint', _external=True)
+    if as_source:
+        # Switch to source mode so we can test that too!
+        test_url = "source:"+test_url
+
     res = client.post(
         url_for("imports.import_page"),
         data={"urls": test_url},
@@ -203,12 +210,15 @@ def test_check_global_ignore_text_functionality(client, live_server, measure_mem
         follow_redirects=True
     )
     assert b"Updated watch." in res.data
-
+    wait_for_all_checks(client)
     # Check it saved
     res = client.get(
         url_for("settings.settings_page"),
     )
-    assert bytes(ignore_text.encode('utf-8')) in res.data
+
+    for i in ignore_text.splitlines():
+        assert bytes(i.encode('utf-8')) in res.data
+
 
     # Trigger a check
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
@@ -221,7 +231,8 @@ def test_check_global_ignore_text_functionality(client, live_server, measure_mem
 
     # Make a change which includes the ignore text, it should be ignored and no 'change' triggered
     # It adds text with "ZZZZzzzz" and "ZZZZ" is in the ignore list
-    set_modified_ignore_response()
+    # And tweaks the ver_stamp which should be picked up by global regex ignore
+    set_modified_ignore_response(ver_stamp=time.time())
 
     # Trigger a check
     client.get(url_for("ui.form_watch_checknow"), follow_redirects=True)
@@ -243,3 +254,11 @@ def test_check_global_ignore_text_functionality(client, live_server, measure_mem
 
     res = client.get(url_for("ui.form_delete", uuid="all"), follow_redirects=True)
     assert b'Deleted' in res.data
+
+def test_check_global_ignore_text_functionality(client, live_server):
+    #live_server_setup(live_server)
+    _run_test_global_ignore(client, as_source=False)
+
+def test_check_global_ignore_text_functionality_as_source(client, live_server):
+    #live_server_setup(live_server)
+    _run_test_global_ignore(client, as_source=True, extra_ignore='/\?v=\d/')

docker-compose.yml

@@ -72,7 +72,7 @@ services:
   
       # Comment out ports: when using behind a reverse proxy , enable networks: etc.
       ports:
-        - 5000:5000
+        - 127.0.0.1:5000:5000
       restart: unless-stopped
 
      # Used for fetching pages via WebDriver+Chrome where you need Javascript support.
@@ -82,7 +82,7 @@ services:
      # If WEBDRIVER or PLAYWRIGHT are enabled, changedetection container depends on that
      # and must wait before starting (substitute "browser-chrome" with "playwright-chrome" if last one is used)
 #      depends_on:
-#          sockpuppetbrowser:
+#          browser-sockpuppet-chrome:
 #              condition: service_started
 
 

requirements.txt

@@ -53,7 +53,7 @@ lxml >=4.8.0,<6,!=5.2.0,!=5.2.1
 # XPath 2.0-3.1 support - 4.2.0 broke something?
 elementpath==4.1.5
 
-selenium~=4.14.0
+selenium~=4.31.0
 
 # https://github.com/pallets/werkzeug/issues/2985
 # Maybe related to pytest?
@@ -90,6 +90,8 @@ extruct
 # For cleaning up unknown currency formats
 babel
 
+levenshtein
+
 # Needed for > 3.10, https://github.com/microsoft/playwright-python/issues/2096
 greenlet >= 3.0.3