PlaywrightCapture 1.28.2__py3-none-any.whl → 1.28.4__py3-none-any.whl

playwrightcapture/capture.py

@@ -736,7 +736,7 @@ class Capture():
         got_button: bool = False
         try:
             try:
-                async with timeout(5):
+                async with timeout(3):
                     if await frame.locator("button.button__acceptAll").is_visible():
                         self.logger.info('Consent window found, clicking through.')
                         got_button = True
@@ -746,7 +746,7 @@ class Capture():
 
             for label in labels_to_click:
                 try:
-                    async with timeout(5):
+                    async with timeout(3):
                         if await frame.get_by_label(label).is_visible():
                             got_button = True
                             self.logger.debug(f'Got button by label on frame: {label}')
@@ -756,7 +756,7 @@ class Capture():
                     self.logger.warning(f'Consent timeout (label {label}) : {e}')
 
                 try:
-                    async with timeout(5):
+                    async with timeout(3):
                         if await frame.get_by_role("button", name=label).is_visible():
                             got_button = True
                             self.logger.debug(f'Got button by role on frame: {label}')
@@ -780,7 +780,15 @@ class Capture():
         except Exception as e:
             self.logger.info(f'Error while moving time forward: {e}')
 
-    async def __instrumentation(self, page: Page, url: str, allow_tracking: bool, clock_set: bool) -> None:
+    async def __instrumentation(self, page: Page, url: str, allow_tracking: bool) -> None:
+        try:
+            # NOTE: the clock must be installed after the page is loaded, otherwise it sometimes cause the complete capture to hang.
+            await page.clock.install()
+            clock_set = True
+        except Error as e:
+            self.logger.warning(f'Unable to install the clock: {e}')
+            clock_set = False
+
         # page instrumentation
         await self._wait_for_random_timeout(page, 5)  # Wait 5 sec after document loaded
         self.logger.debug('Start instrumentation.')
@@ -920,9 +928,9 @@ class Capture():
                            referer: str | None=None,
                            page: Page | None=None, depth: int=0,
                            rendered_hostname_only: bool=True,
+                           with_screenshot: bool=True,
                            with_favicon: bool=False,
                            allow_tracking: bool=False,
-                           clock_set: bool=False
                            ) -> CaptureResponse:
 
         to_return: CaptureResponse = {}
@@ -990,13 +998,6 @@ class Capture():
                 self.should_retry = True
                 return to_return
 
-            try:
-                await page.clock.install()
-                clock_set = True
-            except Error as e:
-                self.logger.warning(f'Unable to install the clock: {e}')
-                clock_set = False
-
             if allow_tracking:
                 # Add authorization clickthroughs
                 await self.__dialog_didomi_clickthrough(page)
@@ -1019,8 +1020,8 @@ class Capture():
 
         try:
             try:
-                await page.goto(url, wait_until='domcontentloaded', referer=referer if referer else '')
                 page.on("download", handle_download)
+                await page.goto(url, wait_until='domcontentloaded', referer=referer if referer else '')
             except Error as initial_error:
                 self._update_exceptions(initial_error)
                 # So this one is really annoying: chromium raises a net::ERR_ABORTED when it hits a download
@@ -1065,7 +1066,7 @@ class Capture():
 
                 try:
                     if self.headless:
-                        await self.__instrumentation(page, url, allow_tracking, clock_set)
+                        await self.__instrumentation(page, url, allow_tracking)
                     else:
                         self.logger.debug('Headed mode, skipping instrumentation.')
                         await self._wait_for_random_timeout(page, self._capture_timeout - 5)
@@ -1099,7 +1100,9 @@ class Capture():
                         self.logger.warning(f'Unable to get favicons: {e}')
 
                 to_return['last_redirected_url'] = page.url
-                to_return['png'] = await self._failsafe_get_screenshot(page)
+
+                if with_screenshot:
+                    to_return['png'] = await self._failsafe_get_screenshot(page)
 
                 self._already_captured.add(url)
                 if depth > 0 and to_return.get('html') and to_return['html']:
@@ -1131,7 +1134,7 @@ class Capture():
                                         page=page, depth=depth,
                                         rendered_hostname_only=rendered_hostname_only,
                                         max_depth_capture_time=max_capture_time,
-                                        clock_set=clock_set)
+                                        with_screenshot=with_screenshot)
                                     to_return['children'].append(child_capture)  # type: ignore[union-attr]
                             except (TimeoutError, asyncio.TimeoutError):
                                 self.logger.info(f'Timeout error, took more than {max_capture_time}s. Unable to capture {url}.')
@@ -1197,12 +1200,12 @@ class Capture():
             self.logger.debug('Finishing up capture.')
             if not capturing_sub:
                 try:
-                    to_return['storage'] = await self.context.storage_state(indexed_db=True)
-                    to_return['cookies'] = await self.context.cookies()
-                    self.logger.debug('Done with cookies.')
+                    to_return['storage'] = await self._failsafe_get_storage()
+                    to_return['cookies'] = await self._failsafe_get_cookies()
+                    self.logger.debug('Done with cookies and storage.')
                 except Exception as e:
                     if 'error' not in to_return:
-                        to_return['error'] = f'Unable to get the cookies: {e}'
+                        to_return['error'] = f'Unable to get the storage: {e}'
                 # frames_tree = self.make_frame_tree(page.main_frame)
                 try:
                     async with timeout(60):
@@ -1224,6 +1227,22 @@ class Capture():
         self.logger.debug('Capture done')
         return to_return
 
+    async def _failsafe_get_cookies(self) -> list[Cookie] | None:
+        try:
+            async with timeout(15):
+                return await self.context.cookies()
+        except (TimeoutError, asyncio.TimeoutError):
+            self.logger.warning("Unable to get cookies (timeout).")
+        return None
+
+    async def _failsafe_get_storage(self) -> StorageState | None:
+        try:
+            async with timeout(15):
+                return await self.context.storage_state(indexed_db=True)
+        except (TimeoutError, asyncio.TimeoutError):
+            self.logger.warning("Unable to get storage (timeout).")
+        return None
+
     async def _failsafe_get_screenshot(self, page: Page) -> bytes:
         self.logger.debug("Capturing a screenshot of the full page.")
         try:

{playwrightcapture-1.28.2.dist-info → playwrightcapture-1.28.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: PlaywrightCapture
-Version: 1.28.2
+Version: 1.28.4
 Summary: A simple library to capture websites using playwright
 License: BSD-3-Clause
 Author: Raphaël Vinot
@@ -18,9 +18,9 @@ Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Internet
 Classifier: Topic :: Security
 Provides-Extra: recaptcha
-Requires-Dist: SpeechRecognition (>=3.14.1) ; extra == "recaptcha"
+Requires-Dist: SpeechRecognition (>=3.14.2) ; extra == "recaptcha"
 Requires-Dist: aiohttp-socks (>=0.10.1)
-Requires-Dist: aiohttp[speedups] (>=3.11.14)
+Requires-Dist: aiohttp[speedups] (>=3.11.16)
 Requires-Dist: async-timeout (>=5.0.1) ; python_version < "3.11"
 Requires-Dist: beautifulsoup4[charset-normalizer,lxml] (>=4.13.3)
 Requires-Dist: dateparser (>=1.2.1)
@@ -28,9 +28,9 @@ Requires-Dist: playwright (>=1.51.0)
 Requires-Dist: playwright-stealth (>=1.0.6)
 Requires-Dist: puremagic (>=1.28)
 Requires-Dist: pydub (>=0.25.1) ; extra == "recaptcha"
-Requires-Dist: setuptools (>=77.0.1)
+Requires-Dist: setuptools (>=78.1.0)
 Requires-Dist: typing-extensions (>=4.12.2,<5.0.0) ; python_version < "3.12"
-Requires-Dist: tzdata (>=2025.1)
+Requires-Dist: tzdata (>=2025.2)
 Requires-Dist: w3lib (>=2.3.1)
 Project-URL: Issues, https://github.com/Lookyloo/PlaywrightCapture/issues
 Project-URL: Repository, https://github.com/Lookyloo/PlaywrightCapture

playwrightcapture-1.28.4.dist-info/RECORD

@@ -0,0 +1,9 @@
+playwrightcapture/__init__.py,sha256=F90Y8wYS13tDjgsfjuFrCfmzQfdnH44G-ovuilJfLEE,511
+playwrightcapture/capture.py,sha256=Iicc_nNjlztCMGIJ9wSB6UhKoIcVJCh_00BssV68XDU,82297
+playwrightcapture/exceptions.py,sha256=LhGJQCGHzEu7Sx2Dfl28OFeDg1OmrwufFjAWXlxQnEA,366
+playwrightcapture/helpers.py,sha256=Xqs09zHhzAWnpBtQ0A9YAxg80P3Lj7aBj5M2WuEr0so,1843
+playwrightcapture/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+playwrightcapture-1.28.4.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
+playwrightcapture-1.28.4.dist-info/METADATA,sha256=o32IMwzDiGFMVmlmaHJF0JTg0p2r5_kZ9KLXuJlVI9M,3075
+playwrightcapture-1.28.4.dist-info/WHEEL,sha256=fGIA9gx4Qxk2KDKeNJCbOEwSrmLtjWCwzBz351GyrPQ,88
+playwrightcapture-1.28.4.dist-info/RECORD,,

{playwrightcapture-1.28.2.dist-info → playwrightcapture-1.28.4.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: poetry-core 2.1.1
+Generator: poetry-core 2.1.2
 Root-Is-Purelib: true
 Tag: py3-none-any

playwrightcapture-1.28.2.dist-info/RECORD

@@ -1,9 +0,0 @@
-playwrightcapture/__init__.py,sha256=F90Y8wYS13tDjgsfjuFrCfmzQfdnH44G-ovuilJfLEE,511
-playwrightcapture/capture.py,sha256=UbZ8kA_JLoWdu2-vxS0-ScV41DEp-WyTk1bYRY5YANA,81515
-playwrightcapture/exceptions.py,sha256=LhGJQCGHzEu7Sx2Dfl28OFeDg1OmrwufFjAWXlxQnEA,366
-playwrightcapture/helpers.py,sha256=Xqs09zHhzAWnpBtQ0A9YAxg80P3Lj7aBj5M2WuEr0so,1843
-playwrightcapture/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-playwrightcapture-1.28.2.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
-playwrightcapture-1.28.2.dist-info/METADATA,sha256=a_Jpi4OyZcmYkV82L-cHdPTuC_woBOlSnj-VTorJZKI,3075
-playwrightcapture-1.28.2.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
-playwrightcapture-1.28.2.dist-info/RECORD,,