PlaywrightCapture 1.24.4__tar.gz → 1.24.5__tar.gz

{playwrightcapture-1.24.4 → playwrightcapture-1.24.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PlaywrightCapture
-Version: 1.24.4
+Version: 1.24.5
 Summary: A simple library to capture websites using playwright
 Home-page: https://github.com/Lookyloo/PlaywrightCapture
 License: BSD-3-Clause

{playwrightcapture-1.24.4 → playwrightcapture-1.24.5}/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:
@@ -682,9 +700,6 @@ class Capture():
                         except Exception:
                             raise e
                 else:
-                    if not self._exception_is_network_error(initial_error):
-                        # TODO: Do something?
-                        self.logger.warning(f'Unexpected error: {initial_error}')
                     raise initial_error
             else:
                 await page.bring_to_front()
@@ -869,7 +884,11 @@ class Capture():
                             'Navigation failed because page was closed!',
                             'Protocol error (Page.bringToFront): Not attached to an active page',
                             'Peer failed to perform TLS handshake: The TLS connection was non-properly terminated.',
-                            'Load cannot follow more than 20 redirections']:
+                            '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

{playwrightcapture-1.24.4 → playwrightcapture-1.24.5}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "PlaywrightCapture"
-version = "1.24.4"
+version = "1.24.5"
 description = "A simple library to capture websites using playwright"
 authors = ["Raphaël Vinot <raphael.vinot@circl.lu>"]
 license = "BSD-3-Clause"
@@ -44,7 +44,7 @@ pytest = "^8.1.1"
 mypy = "^1.9.0"
 types-dateparser = "^1.1.4.20240331"
 types-requests = "^2.31.0.20240406"
-types-pytz = "^2024.1.0.20240203"
+types-pytz = "^2024.1.0.20240417"
 
 
 [build-system]