PlaywrightCapture 1.40.8__tar.gz → 1.41.0__tar.gz

{playwrightcapture-1.40.8 → playwrightcapture-1.41.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PlaywrightCapture
-Version: 1.40.8
+Version: 1.41.0
 Summary: A simple library to capture websites using playwright
 License-Expression: BSD-3-Clause
 License-File: LICENSE
@@ -19,7 +19,7 @@ Classifier: Topic :: Internet
 Classifier: Topic :: Security
 Provides-Extra: recaptcha
 Requires-Dist: SpeechRecognition (>=3.17.0) ; extra == "recaptcha"
-Requires-Dist: aiohttp-socks (>=0.11.0)
+Requires-Dist: aiohttp-socks (>=0.12.0)
 Requires-Dist: aiohttp[speedups] (>=3.14.3)
 Requires-Dist: async-timeout (>=5.0.1) ; python_version < "3.11"
 Requires-Dist: beautifulsoup4[charset-normalizer,lxml] (>=4.15.0)
@@ -32,7 +32,7 @@ Requires-Dist: playwright-stealth (>=2.0.3)
 Requires-Dist: pure-magic-rs (>=0.4.3)
 Requires-Dist: pydub-ng (>=0.2.0) ; extra == "recaptcha"
 Requires-Dist: pyfaup-rs (>=0.4.6,<0.5.0)
-Requires-Dist: python-socks (>=2.7.1,<3.0.0)
+Requires-Dist: python-socks (>=3.0.0,<4.0.0)
 Requires-Dist: rfc3161-client (>=1.0.4,<2.0.0)
 Requires-Dist: typing-extensions (>=4.12.2,<5.0.0) ; python_version < "3.12"
 Requires-Dist: w3lib (>=2.4.1)

{playwrightcapture-1.40.8 → playwrightcapture-1.41.0}/playwrightcapture/capture.py

@@ -31,7 +31,7 @@ from bs4 import BeautifulSoup
 from charset_normalizer import from_bytes
 from lookyloo_models import Cookie, CaptureSettings
 from playwright._impl._errors import TargetClosedError
-from playwright.async_api import async_playwright, Frame, Error, Page, Download, Request, Route
+from playwright.async_api import async_playwright, Frame, Error, Page, Download, Request, Route, ConsoleMessage
 from playwright.async_api import TimeoutError as PlaywrightTimeoutError
 from playwright_stealth import Stealth, ALL_EVASIONS_DISABLED_KWARGS  # type: ignore[attr-defined]
 from pure_magic_rs import MagicDb
@@ -101,6 +101,9 @@ class CaptureResponse(TypedDict, total=False):
     # dict[<entry name>] = '<base64 encoded timestamp response>'
     trusted_timestamps: dict[str, str] | None
 
+    # The entries in the debug console gathered during the captue
+    console_messages: list[dict[str, str | int | float]] | None
+
     # One day, playwright will support getting the favicon from the capture itself
     # favicon: Optional[bytes]
     # in the meantime, we need a workaround: https://github.com/Lookyloo/PlaywrightCapture/issues/45
@@ -226,6 +229,9 @@ class Capture():
         # Trusted Timestamp Settings are provided by LacusCore
         self.tt_settings = tt_settings
 
+        # Gather console messages
+        self.console_messages: list[dict[str, str | int | float]] = []
+
         # Prepare the env to use for playwright
         if env:
             self._env.update(env)
@@ -391,6 +397,19 @@ class Capture():
                 self.logger.info(f'Unable to force download: {e}')
                 await route.continue_()
 
+        async def handle_console_msg(msg: ConsoleMessage) -> None:
+            to_add = {'timestamp': msg.timestamp, 'type': msg.type, 'text': msg.text}
+
+            if msg.location.get('url') is not None:
+                to_add['url'] = msg.location['url']
+            if msg.location.get('line') is not None:
+                # can be 0
+                to_add['line'] = msg.location['line']
+            if msg.location.get('column') is not None:
+                # can be 0
+                to_add['column'] = msg.location['column']
+            self.console_messages.append(to_add)
+
         np_retry = 3
         while np_retry > 0:
             try:
@@ -435,6 +454,7 @@ class Capture():
         page.on("requestfinished", store_request)
         page.on("dialog", lambda dialog: dialog.accept())
         page.on("download", handle_download)
+        page.on("console", handle_console_msg)
         return page
 
     def __prepare_proxy_playwright(self, proxy: str) -> ProxySettings:
@@ -1304,6 +1324,8 @@ class Capture():
                 await self._get_trusted_timestamps(to_return)
             except Exception as e:
                 self.logger.warning(f'Unable to get trusted timestamps: {e}')
+        if self.console_messages:
+            to_return['console_messages'] = self.console_messages
 
     def __check_local_url(self, url: str) -> tuple[bool, str]:
         try:
@@ -2029,6 +2051,7 @@ class Capture():
             'Error receiving data: Connection reset by peer',
             'Internal SOCKSv5 proxy server error.',
             'Host unreachable through SOCKSv5 server.',
+            'Connection to SOCKSv5 proxy server lost',
             'Operation was cancelled',
             'The URL can’t be shown',
             'Frame was detached',

{playwrightcapture-1.40.8 → playwrightcapture-1.41.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "PlaywrightCapture"
-version = "1.40.8"
+version = "1.41.0"
 description = "A simple library to capture websites using playwright"
 authors = [
     {name="Raphaël Vinot", email= "raphael.vinot@circl.lu"}
@@ -18,10 +18,10 @@ dependencies = [
     "playwright-stealth (>=2.0.3)",
     "async-timeout (>=5.0.1) ; python_version < \"3.11\"",
     "aiohttp[speedups] (>=3.14.3)",
-    "aiohttp-socks (>=0.11.0)",
+    "aiohttp-socks (>=0.12.0)",
     "typing-extensions (>=4.12.2,<5.0.0) ; python_version < \"3.12\"",
     "dnspython (>=2.7.0,<3.0.0)",
-    "python-socks (>=2.7.1,<3.0.0)",
+    "python-socks (>=3.0.0,<4.0.0)",
     "rfc3161-client (>=1.0.4,<2.0.0)",
     "orjson (>=3.11.4,<4.0.0)",
     "pure-magic-rs (>=0.4.3)",
@@ -52,8 +52,8 @@ recaptcha = [
 [tool.poetry.group.dev.dependencies]
 types-beautifulsoup4 = "^4.12.0.20250516"
 pytest = "^9.1.1"
-mypy = "^2.3.0"
-types-dateparser = "^1.4.1.20260806"
+mypy = "^2.3.1"
+types-dateparser = "^1.4.2.20260813"
 types-pytz = "^2026.3.1.20260727"