PlaywrightCapture 1.25.12__py3-none-any.whl → 1.25.14__py3-none-any.whl

playwrightcapture/capture.py

@@ -636,6 +636,28 @@ class Capture():
         )
         self.logger.info('Yahoo handler added')
 
+    async def __dialog_tarteaucitron_clickthrough(self, page: Page) -> None:
+        # https://github.com/AmauriC/tarteaucitron.js/
+        async def handler() -> None:
+            if await page.locator('#tarteaucitronAlertBig').locator('button.tarteaucitronAllow').is_visible():
+                self.logger.debug('Got TarteAuCitron big , clicking through.')
+                await page.locator('#tarteaucitronAlertBig').locator("button.tarteaucitronAllow").click(timeout=2000)
+            elif await page.locator('#tarteaucitronAlertSmall').locator('button.tarteaucitronAllow').is_visible():
+                self.logger.debug('Got TarteAuCitron small, clicking through.')
+                await page.locator('#tarteaucitronAlertSmall').locator("button.tarteaucitronAllow").click(timeout=2000)
+
+        await page.add_locator_handler(
+            page.locator('#tarteaucitronAlertBig'),
+            handler,
+            times=1, no_wait_after=True
+        )
+        await page.add_locator_handler(
+            page.locator('#tarteaucitronAlertSmall'),
+            handler,
+            times=1, no_wait_after=True
+        )
+        self.logger.info('TarteAuCitron handler added')
+
     async def __dialog_ppms_clickthrough(self, page: Page) -> None:
         async def handler() -> None:
             if await page.locator('.ppms_cm_popup_overlay').locator("button.ppms_cm_agree-to-all").is_visible():
@@ -661,9 +683,12 @@ class Capture():
             "Tout accepter",
             "Accepter",
             "Accepter les cookies",
+            "Autoriser",
             # English
             "Accept & continue",
             "Accept all",
+            "Accept",
+            "Agree and close",
             # Dutch
             "Accepteer",
             # Spanish
@@ -680,21 +705,35 @@ class Capture():
 
         got_button: bool = False
         try:
-            if await frame.locator("button.button__acceptAll").is_visible():
-                self.logger.info('Consent window found, clicking through.')
-                got_button = True
-                await frame.locator("button.button__acceptAll").click(timeout=2000)
+            try:
+                async with timeout(5):
+                    if await frame.locator("button.button__acceptAll").is_visible():
+                        self.logger.info('Consent window found, clicking through.')
+                        got_button = True
+                        await frame.locator("button.button__acceptAll").click(timeout=2000)
+            except (TimeoutError, asyncio.TimeoutError) as e:
+                self.logger.warning(f'Frame consent timeout: {e}')
+
             for label in labels_to_click:
-                if await frame.get_by_label(label).is_visible():
-                    got_button = True
-                    self.logger.debug(f'Got button by label on frame: {label}')
-                    await frame.get_by_label(label).click(timeout=2000)
-                    break
-                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}')
-                    await frame.get_by_role("button", name=label).click(timeout=2000)
-                    break
+                try:
+                    async with timeout(5):
+                        if await frame.get_by_label(label).is_visible():
+                            got_button = True
+                            self.logger.debug(f'Got button by label on frame: {label}')
+                            await frame.get_by_label(label).click(timeout=2000)
+                            break
+                except (TimeoutError, asyncio.TimeoutError) as e:
+                    self.logger.warning(f'Frame consent timeout: {e}')
+
+                try:
+                    async with timeout(5):
+                        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}')
+                            await frame.get_by_role("button", name=label).click(timeout=2000)
+                            break
+                except (TimeoutError, asyncio.TimeoutError) as e:
+                    self.logger.warning(f'Frame consent timeout: {e}')
         except Exception as e:
             self.logger.info(f'Issue with frame consent: {e}')
         return got_button
@@ -794,6 +833,7 @@ class Capture():
                 await self.__dialog_ppms_clickthrough(page)
                 await self.__dialog_alert_dialog_clickthrough(page)
                 await self.__dialog_clickthrough(page)
+                await self.__dialog_tarteaucitron_clickthrough(page)
 
             await stealth_async(page, PCStealthConfig())
 
@@ -884,8 +924,13 @@ class Capture():
                     self.logger.debug('Done with captcha.')
 
                     # move mouse
-                    await page.mouse.move(x=random.uniform(300, 800), y=random.uniform(200, 500))
-                    self.logger.debug('Moved mouse.')
+                    try:
+                        async with timeout(5):
+                            await page.mouse.move(x=random.uniform(300, 800), y=random.uniform(200, 500))
+                            self.logger.debug('Moved mouse.')
+                    except (asyncio.TimeoutError, TimeoutError):
+                        self.logger.debug('Moving the mouse caused a timeout.')
+
                     await self._wait_for_random_timeout(page, 5)
                     self.logger.debug('Keep going after moving mouse.')
 
@@ -955,11 +1000,14 @@ class Capture():
                     self.logger.debug('Keep going after moving on page.')
 
                     try:
-                        await page.keyboard.press('PageUp')
-                        self.logger.debug('PageUp on keyboard')
-                        await self._wait_for_random_timeout(page, 3)
-                        await page.keyboard.press('PageDown')
-                        self.logger.debug('PageDown on keyboard')
+                        async with timeout(5):
+                            await page.keyboard.press('PageUp')
+                            self.logger.debug('PageUp on keyboard')
+                            await self._wait_for_random_timeout(page, 3)
+                            await page.keyboard.press('PageDown')
+                            self.logger.debug('PageDown on keyboard')
+                    except (asyncio.TimeoutError, TimeoutError):
+                        self.logger.debug('Using keyboard caused a timeout.')
                     except Error as e:
                         self.logger.debug(f'Unable to use keyboard: {e}')
                 if self.wait_for_download > 0:

{playwrightcapture-1.25.12.dist-info → playwrightcapture-1.25.14.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PlaywrightCapture
-Version: 1.25.12
+Version: 1.25.14
 Summary: A simple library to capture websites using playwright
 Home-page: https://github.com/Lookyloo/PlaywrightCapture
 License: BSD-3-Clause
@@ -22,7 +22,7 @@ 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] (>=3.10.3,<4.0.0)
+Requires-Dist: aiohttp[speedups] (>=3.10.5,<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)
@@ -31,7 +31,7 @@ Requires-Dist: playwright-stealth (>=1.0.6,<2.0.0)
 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 (>=72.2.0,<73.0.0)
+Requires-Dist: setuptools (>=74.0.0,<75.0.0)
 Requires-Dist: tzdata (>=2024.1,<2025.0)
 Requires-Dist: w3lib (>=2.2.1,<3.0.0)
 Project-URL: Repository, https://github.com/Lookyloo/PlaywrightCapture

{playwrightcapture-1.25.12.dist-info → playwrightcapture-1.25.14.dist-info}/RECORD

@@ -1,9 +1,9 @@
 playwrightcapture/__init__.py,sha256=F90Y8wYS13tDjgsfjuFrCfmzQfdnH44G-ovuilJfLEE,511
-playwrightcapture/capture.py,sha256=R8ayCKpD1Q78ewaQ8-4tuMP8XVE8YGOwhwVYZDs4a8g,75867
+playwrightcapture/capture.py,sha256=9ESwyexAnBtGc2QV3u4WvOKZqisMVdqHL4rko1RxJXQ,78337
 playwrightcapture/exceptions.py,sha256=LhGJQCGHzEu7Sx2Dfl28OFeDg1OmrwufFjAWXlxQnEA,366
 playwrightcapture/helpers.py,sha256=SXQLEuxMs8-bcWykMiUVosHzzxBKuS-QC0gBV3OmKmo,1764
 playwrightcapture/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-playwrightcapture-1.25.12.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
-playwrightcapture-1.25.12.dist-info/METADATA,sha256=Kw9CVpRSN9vLL1XSf_ZGIKAzX11VY4MJuXrO0OVCmHw,3172
-playwrightcapture-1.25.12.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-playwrightcapture-1.25.12.dist-info/RECORD,,
+playwrightcapture-1.25.14.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
+playwrightcapture-1.25.14.dist-info/METADATA,sha256=45rFgcxqSi2TAU8KwsE0dxdWja1u7tvHchj0x3H1dCM,3172
+playwrightcapture-1.25.14.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+playwrightcapture-1.25.14.dist-info/RECORD,,