PlaywrightCapture 1.24.3__py3-none-any.whl → 1.24.5__py3-none-any.whl

playwrightcapture/capture.py

@@ -14,8 +14,9 @@ import time
 
 from base64 import b64decode
 from io import BytesIO
+from logging import LoggerAdapter, Logger
 from tempfile import NamedTemporaryFile
-from typing import Any, TypedDict, Literal, TYPE_CHECKING
+from typing import Any, TypedDict, Literal, TYPE_CHECKING, MutableMapping
 from urllib.parse import urlparse, unquote, urljoin
 from zipfile import ZipFile
 
@@ -79,6 +80,16 @@ class CaptureResponse(TypedDict, total=False):
     potential_favicons: set[bytes] | None
 
 
+class PlaywrightCaptureLogAdapter(LoggerAdapter):  # type: ignore[type-arg]
+    """
+    Prepend log entry with the UUID of the capture
+    """
+    def process(self, msg: str, kwargs: MutableMapping[str, Any]) -> tuple[str, MutableMapping[str, Any]]:
+        if self.extra:
+            return '[{}] {}'.format(self.extra['uuid'], msg), kwargs
+        return msg, kwargs
+
+
 class Capture():
 
     _browsers: list[BROWSER] = ['chromium', 'firefox', 'webkit']
@@ -90,7 +101,8 @@ class Capture():
 
     def __init__(self, browser: BROWSER | None=None, device_name: str | None=None,
                  proxy: str | dict[str, str] | None=None,
-                 general_timeout_in_sec: int | None = None, loglevel: str | int='INFO'):
+                 general_timeout_in_sec: int | None = None, loglevel: str | int='INFO',
+                 uuid: str | None=None):
         """Captures a page with Playwright.
 
         :param browser: The browser to use for the capture.
@@ -98,9 +110,15 @@ class Capture():
         :param proxy: The external proxy to use for the capture.
         :param general_timeout_in_sec: The general timeout for the capture, including children.
         :param loglevel: Python loglevel
+        :param uuid: The UUID of the capture.
         """
-        self.logger = logging.getLogger('playwrightcapture')
-        self.logger.setLevel(loglevel)
+        master_logger = logging.getLogger('playwrightcapture')
+        master_logger.setLevel(loglevel)
+        self.logger: Logger | PlaywrightCaptureLogAdapter
+        if uuid is not None:
+            self.logger = PlaywrightCaptureLogAdapter(master_logger, {'uuid': uuid})
+        else:
+            self.logger = master_logger
         self.browser_name: BROWSER = browser if browser else 'chromium'
 
         if general_timeout_in_sec is None:
@@ -681,7 +699,7 @@ class Capture():
                             self.should_retry = True
                         except Exception:
                             raise e
-                elif self._exception_is_network_error(initial_error):
+                else:
                     raise initial_error
             else:
                 await page.bring_to_front()
@@ -718,8 +736,8 @@ class Capture():
                     if allow_tracking:
                         await self._wait_for_random_timeout(page, 2)
                         # This event is required trigger the add_locator_handler
-                        if await page.locator("body").is_visible():
-                            await page.locator("body").click(button="right", timeout=2000)
+                        if await page.locator("body").first.is_visible():
+                            await page.locator("body").first.click(button="right", timeout=2000)
 
                     # move mouse
                     await page.mouse.move(x=random.uniform(300, 800), y=random.uniform(200, 500))
@@ -864,7 +882,13 @@ class Capture():
                             'Connection closed',
                             'Navigation interrupted by another one',
                             'Navigation failed because page was closed!',
-                            'Protocol error (Page.bringToFront): Not attached to an active page']:
+                            'Protocol error (Page.bringToFront): Not attached to an active page',
+                            'Peer failed to perform TLS handshake: The TLS connection was non-properly terminated.',
+                            'Peer failed to perform TLS handshake: Error sending data: Connection reset by peer',
+                            'Peer sent fatal TLS alert: The server name sent was not recognized',
+                            'Load cannot follow more than 20 redirections',
+                            'Page crashed',
+                            'Error receiving data: Connection reset by peer']:
                 # Other errors, let's give it another shot
                 self.logger.info(f'Issue with {url} (retrying): {e.message}')
                 self.should_retry = True
@@ -872,6 +896,9 @@ class Capture():
                 # The browser barfed, let's try again
                 self.logger.info(f'Browser barfed on {url} (retrying): {e.message}')
                 self.should_retry = True
+            elif e.name in ['net::ERR_INVALID_AUTH_CREDENTIALS']:
+                # No need to retry, the credentials are wrong/missing.
+                pass
             else:
                 # Unexpected ones
                 self.logger.exception(f'Something went poorly with {url}: {e.message}')
@@ -932,8 +959,8 @@ class Capture():
                 max_wait = force_max_wait_in_sec
             else:
                 max_wait = self._capture_timeout / self.__network_not_idle
-            max_wait *= 1000
             self.logger.debug(f'Waiting for network idle, max wait: {max_wait}s')
+            max_wait *= 1000
             # If we don't have networkidle relatively quick, it's probably because we're playing a video.
             await page.wait_for_load_state('networkidle', timeout=max_wait)
         except PlaywrightTimeoutError:
@@ -1103,8 +1130,11 @@ class Capture():
                 'net::ERR_CONNECTION_CLOSED',
                 'net::ERR_CONNECTION_REFUSED',
                 'net::ERR_CONNECTION_RESET',
+                'net::ERR_CONNECTION_TIMED_OUT',
                 'net::ERR_EMPTY_RESPONSE',
+                'net::ERR_HTTP_RESPONSE_CODE_FAILURE',
                 'net::ERR_HTTP2_PROTOCOL_ERROR',
+                'net::ERR_INVALID_RESPONSE',
                 'net::ERR_NAME_NOT_RESOLVED',
                 'net::ERR_SOCKS_CONNECTION_FAILED',
                 'net::ERR_SSL_UNRECOGNIZED_NAME_ALERT',
@@ -1251,8 +1281,14 @@ class Capture():
                         # unable to identify the mimetype
                         self.logger.debug(f'Unable to identify the mimetype for favicon from {u}')
                     else:
-                        if mimetype.startswith('image'):
+                        if not mimetype:
+                            # empty, ignore
+                            pass
+                        elif mimetype.startswith('image'):
                             to_return.add(favicon)
+                        elif mimetype.startswith('text'):
+                            # Just ignore, it's probably a 404 page
+                            pass
                         else:
                             self.logger.warning(f'Unexpected mimetype for favicon from {u}: {mimetype}')
                 self.logger.debug(f'Done with favicon from {u}.')

{playwrightcapture-1.24.3.dist-info → playwrightcapture-1.24.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PlaywrightCapture
-Version: 1.24.3
+Version: 1.24.5
 Summary: A simple library to capture websites using playwright
 Home-page: https://github.com/Lookyloo/PlaywrightCapture
 License: BSD-3-Clause
@@ -29,7 +29,7 @@ Requires-Dist: puremagic (>=1.21,<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: requests[socks] (>=2.31.0,<3.0.0) ; extra == "recaptcha"
-Requires-Dist: setuptools (>=69.2.0,<70.0.0)
+Requires-Dist: setuptools (>=69.5.1,<70.0.0)
 Requires-Dist: tzdata (>=2024.1,<2025.0)
 Requires-Dist: w3lib (>=2.1.2,<3.0.0)
 Project-URL: Repository, https://github.com/Lookyloo/PlaywrightCapture

{playwrightcapture-1.24.3.dist-info → playwrightcapture-1.24.5.dist-info}/RECORD

@@ -1,9 +1,9 @@
 playwrightcapture/__init__.py,sha256=F90Y8wYS13tDjgsfjuFrCfmzQfdnH44G-ovuilJfLEE,511
-playwrightcapture/capture.py,sha256=KuIaHzPtZqZKYNDO8WfPx22D6Ms6xKezZmgg4I4yCNs,61485
+playwrightcapture/capture.py,sha256=-30zUyPeS8VPt-8MZhkYmFiJEvq9LwAZFp0ChhWeRsY,63307
 playwrightcapture/exceptions.py,sha256=LhGJQCGHzEu7Sx2Dfl28OFeDg1OmrwufFjAWXlxQnEA,366
 playwrightcapture/helpers.py,sha256=SXQLEuxMs8-bcWykMiUVosHzzxBKuS-QC0gBV3OmKmo,1764
 playwrightcapture/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-playwrightcapture-1.24.3.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
-playwrightcapture-1.24.3.dist-info/METADATA,sha256=yuyoEKN1Qa5Dcsx0YkWGST9sPFSP7RlClu9FjMWz53A,3077
-playwrightcapture-1.24.3.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-playwrightcapture-1.24.3.dist-info/RECORD,,
+playwrightcapture-1.24.5.dist-info/LICENSE,sha256=uwFc39fTLacBUG-XTuxX6IQKTKhg4z14gWOLt3ex4Ho,1775
+playwrightcapture-1.24.5.dist-info/METADATA,sha256=fT8WNXOeQ5hQVkN7cQhd29L1S3CwShSvd5jpq53orYE,3077
+playwrightcapture-1.24.5.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+playwrightcapture-1.24.5.dist-info/RECORD,,