PlaywrightCapture 1.25.10__py3-none-any.whl → 1.25.11__py3-none-any.whl

playwrightcapture/capture.py

@@ -601,6 +601,8 @@ class Capture():
             elif await page.get_by_test_id("uc-accept-all-button").is_visible():
                 self.logger.info('Consent window found, clicking through.')
                 await page.get_by_test_id("uc-accept-all-button").click(timeout=2000)
+            elif await page.locator('#axeptio_btn_acceptAll').is_visible():
+                await page.locator('#axeptio_btn_acceptAll').click(timeout=2000)
             else:
                 self.logger.info('Consent window found (dialog), but no button to click through.')
         await page.add_locator_handler(
@@ -646,6 +648,24 @@ class Capture():
         )
         self.logger.info('Piwik handler added')
 
+    async def __frame_consent(self, frame: Frame) -> bool:
+        """Search & Click content in iframes. Cannot easily use the locator handler for this without having many many handlers.
+        And the iframes don't have a title or a role to easily identify them so we just try with generic locators that vary by language."""
+        got_button: bool = False
+        if await frame.get_by_label("Alle akzeptieren").is_visible():
+            got_button = True
+            await frame.get_by_label("Alle akzeptieren").click(timeout=2000)
+        elif await frame.get_by_label("Accept & continue").is_visible():
+            got_button = True
+            await frame.get_by_label("Accept & continue").click(timeout=2000)
+        elif await frame.get_by_label("Accepter et continuer").is_visible():
+            got_button = True
+            await frame.get_by_label("Accepter et continuer").click(timeout=2000)
+        elif await frame.get_by_label("Accepteer").is_visible():
+            got_button = True
+            await frame.get_by_label("Accepteer").click(timeout=2000)
+        return got_button
+
     async def capture_page(self, url: str, *, max_depth_capture_time: int,
                            referer: str | None=None,
                            page: Page | None=None, depth: int=0,
@@ -911,8 +931,19 @@ class Capture():
                 #    self.logger.warning('Unable to move time forward.')
 
                 self.logger.debug('Done with instrumentation, waiting for network idle.')
+                if allow_tracking:
+                    self.logger.debug('Check iFrames for button')
+                    for frame in page.frames:
+                        frame_title = await frame.title()
+                        self.logger.debug(f'Check button on {frame_title}')
+                        if await self.__frame_consent(frame):
+                            self.logger.debug(f'Got button on {frame_title}')
+                            await self._wait_for_random_timeout(page, 10)  # Wait 10 sec after click
+                    self.logger.debug('Done with iFrames.')
+
                 await self._wait_for_random_timeout(page, 5)  # Wait 5 sec after instrumentation
                 await self._safe_wait(page)
+
                 self.logger.debug('Done with instrumentation, done with waiting.')
 
                 if content := await self._failsafe_get_content(page):

{playwrightcapture-1.25.10.dist-info → playwrightcapture-1.25.11.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PlaywrightCapture
-Version: 1.25.10
+Version: 1.25.11
 Summary: A simple library to capture websites using playwright
 Home-page: https://github.com/Lookyloo/PlaywrightCapture
 License: BSD-3-Clause
@@ -22,13 +22,13 @@ Classifier: Topic :: Security
 Provides-Extra: recaptcha
 Requires-Dist: SpeechRecognition (>=3.10.4,<4.0.0) ; extra == "recaptcha"
 Requires-Dist: aiohttp-socks (>=0.9,<0.10)
-Requires-Dist: aiohttp[speedups] (>=3.10.1,<4.0.0)
+Requires-Dist: aiohttp[speedups] (>=3.10.3,<4.0.0)
 Requires-Dist: async-timeout (>=4.0.3,<5.0.0) ; python_version < "3.11"
 Requires-Dist: beautifulsoup4[charset-normalizer,lxml] (>=4.12.3,<5.0.0)
 Requires-Dist: dateparser (>=1.2.0,<2.0.0)
-Requires-Dist: playwright (>=1.45.1,<2.0.0)
+Requires-Dist: playwright (>=1.46.0,<2.0.0)
 Requires-Dist: playwright-stealth (>=1.0.6,<2.0.0)
-Requires-Dist: puremagic (>=1.26,<2.0)
+Requires-Dist: puremagic (>=1.27,<2.0)
 Requires-Dist: pydub (>=0.25.1,<0.26.0) ; extra == "recaptcha"
 Requires-Dist: pytz (>=2024.1,<2025.0) ; python_version < "3.9"
 Requires-Dist: setuptools (>=72.1.0,<73.0.0)

{playwrightcapture-1.25.10.dist-info → playwrightcapture-1.25.11.dist-info}/RECORD

@@ -1,9 +1,9 @@
 playwrightcapture/__init__.py,sha256=F90Y8wYS13tDjgsfjuFrCfmzQfdnH44G-ovuilJfLEE,511
-playwrightcapture/capture.py,sha256=AayQPmsjzcN2TIBi4uuwDmV0kqIVLDdAfEaNPUg8TXY,72606
+playwrightcapture/capture.py,sha256=Rmo_EVRlR9btsgE2H99OtGPRZwIe8RVq-JCc2GzUWiI,74446
 playwrightcapture/exceptions.py,sha256=LhGJQCGHzEu7Sx2Dfl28OFeDg1OmrwufFjAWXlxQnEA,366
 playwrightcapture/helpers.py,sha256=SXQLEuxMs8-bcWykMiUVosHzzxBKuS-QC0gBV3OmKmo,1764
 playwrightcapture/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-playwrightcapture-1.25.10.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
-playwrightcapture-1.25.10.dist-info/METADATA,sha256=3nVVIzM7kcR62_XYA1mRUxfoflPZwfLS5B8KDPy3Hks,3172
-playwrightcapture-1.25.10.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-playwrightcapture-1.25.10.dist-info/RECORD,,
+playwrightcapture-1.25.11.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
+playwrightcapture-1.25.11.dist-info/METADATA,sha256=nGuO6TAlz2lKM15HiIgZJ4iERLBO_AXNBBpgqo8nfhM,3172
+playwrightcapture-1.25.11.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+playwrightcapture-1.25.11.dist-info/RECORD,,