lacuscore 1.10.4__py3-none-any.whl → 1.10.5__py3-none-any.whl

lacuscore/helpers.py

@@ -6,7 +6,7 @@ import json
 
 from enum import IntEnum, unique
 from logging import LoggerAdapter
-from typing import MutableMapping, Any, TypedDict, Mapping
+from typing import MutableMapping, Any, TypedDict, Mapping, Literal
 
 from defang import refang  # type: ignore[import-untyped]
 from pydantic import BaseModel, field_validator, model_validator, ValidationError
@@ -87,7 +87,7 @@ class CaptureSettings(BaseModel):
     url: str | None = None
     document_name: str | None = None
     document: str | None = None
-    browser: str | None = None
+    browser: Literal['chromium', 'firefox', 'webkit'] | None = None
     device_name: str | None = None
     user_agent: str | None = None
     proxy: str | dict[str, str] | None = None
@@ -111,6 +111,24 @@ class CaptureSettings(BaseModel):
     depth: int = 0
     rendered_hostname_only: bool = True  # Note: only used if depth is > 0
 
+    @model_validator(mode="before")
+    @classmethod
+    def empty_str_to_none(cls, data: Any) -> dict[str, Any] | Any:
+        if isinstance(data, dict):
+            # Make sure all the strings are stripped, and None if empty.
+            to_return: dict[str, Any] = {}
+            for k, v in data.items():
+                if isinstance(v, str):
+                    if v_stripped := v.strip():
+                        if v_stripped[0] in ['{', '[']:
+                            to_return[k] = from_json(v_stripped)
+                        else:
+                            to_return[k] = v_stripped
+                else:
+                    to_return[k] = v
+            return to_return
+        return data
+
     @model_validator(mode='after')
     def check_capture_element(self) -> CaptureSettings:
         if self.document_name and not self.document:
@@ -126,142 +144,142 @@ class CaptureSettings(BaseModel):
 
     @field_validator('url', mode='after')
     @classmethod
-    def load_url(cls, v: str | None) -> str | None:
-        if isinstance(v, str):
-            url = v.strip()
-            url = refang(url)  # In case we get a defanged url at this stage.
-            if (not url.lower().startswith('data:')
-                    and not url.lower().startswith('http:')
-                    and not url.lower().startswith('https:')
-                    and not url.lower().startswith('file:')):
-                url = f'http://{url}'
-            return url
-        return v
+    def load_url(cls, url: str | None) -> str | None:
+        if isinstance(url, str):
+            _url = refang(url)  # In case we get a defanged url at this stage.
+            if (not _url.lower().startswith('data:')
+                    and not _url.lower().startswith('http:')
+                    and not _url.lower().startswith('https:')
+                    and not _url.lower().startswith('file:')):
+                _url = f'http://{_url}'
+            return _url
+        return url
 
     @field_validator('document_name', mode='after')
     @classmethod
-    def load_document_name(cls, v: str | None) -> str | None:
-        if isinstance(v, str):
-            name = v.strip()
-            if '.' not in name:
+    def load_document_name(cls, document_name: str | None) -> str | None:
+        if isinstance(document_name, str):
+            if '.' not in document_name:
                 # The browser will simply display the file as text if there is no extension.
                 # Just add HTML as a fallback, as it will be the most comon one.
-                name = f'{name}.html'
-            return name
-        return v
+                document_name = f'{document_name}.html'
+            return document_name
+        return None
+
+    @field_validator('browser', mode='before')
+    @classmethod
+    def load_browser(cls, browser: Any) -> str | None:
+        if isinstance(browser, str) and browser in ['chromium', 'firefox', 'webkit']:
+            return browser
+        # There are old captures where the browser is not a playwright browser name, so we ignore it.
+        return None
 
     @field_validator('proxy', mode='before')
     @classmethod
-    def load_proxy_json(cls, v: Any) -> str | dict[str, str] | None:
-        if not v:
+    def load_proxy_json(cls, proxy: Any) -> str | dict[str, str] | None:
+        if not proxy:
             return None
-        if isinstance(v, str):
-            if v.startswith('{'):
-                return from_json(v)
+        if isinstance(proxy, str):
             # Just the proxy
-            return v
-        elif isinstance(v, dict):
-            return v
+            return proxy
+        elif isinstance(proxy, dict):
+            return proxy
         return None
 
     @field_validator('cookies', mode='before')
     @classmethod
-    def load_cookies_json(cls, v: Any) -> list[dict[str, Any]] | None:
-        if not v:
+    def load_cookies_json(cls, cookies: Any) -> list[dict[str, Any]] | None:
+        if not cookies:
             return None
-        if isinstance(v, str):
-            if v.startswith('['):
-                return from_json(v)
+        if isinstance(cookies, str):
             # Cookies are invalid, ignoring.
-        elif isinstance(v, list):
-            return v
+            pass
+        elif isinstance(cookies, list):
+            return cookies
         return None
 
     @field_validator('headers', mode='before')
     @classmethod
-    def load_headers_json(cls, v: Any) -> dict[str, str] | None:
-        if not v:
+    def load_headers_json(cls, headers: Any) -> dict[str, str] | None:
+        if not headers:
             return None
-        if isinstance(v, str):
-            if v[0] == '{':
-                return from_json(v)
-            else:
-                # make it a dict
-                new_headers = {}
-                for header_line in v.splitlines():
-                    if header_line and ':' in header_line:
-                        splitted = header_line.split(':', 1)
-                        if splitted and len(splitted) == 2:
-                            header, h_value = splitted
-                            if header.strip() and h_value.strip():
-                                new_headers[header.strip()] = h_value.strip()
-                return new_headers
-        elif isinstance(v, dict):
-            return v
+        if isinstance(headers, str):
+            # make it a dict
+            new_headers = {}
+            for header_line in headers.splitlines():
+                if header_line and ':' in header_line:
+                    splitted = header_line.split(':', 1)
+                    if splitted and len(splitted) == 2:
+                        header, h_value = splitted
+                        if header.strip() and h_value.strip():
+                            new_headers[header.strip()] = h_value.strip()
+            return new_headers
+        elif isinstance(headers, dict):
+            return headers
         return None
 
     @field_validator('http_credentials', mode='before')
     @classmethod
-    def load_http_creds_json(cls, v: Any) -> dict[str, str] | None:
-        if not v:
+    def load_http_creds_json(cls, http_credentials: Any) -> dict[str, str] | None:
+        if not http_credentials:
             return None
-        if isinstance(v, str):
-            if v.startswith('{'):
-                return from_json(v)
-        elif isinstance(v, dict):
-            return v
+        if isinstance(http_credentials, str):
+            # ignore
+            return None
+        elif isinstance(http_credentials, dict):
+            return http_credentials
         return None
 
     @field_validator('http_credentials', mode='after')
     @classmethod
-    def check_http_creds(cls, v: dict[str, str] | None) -> dict[str, str] | None:
-        if not v:
-            return v
-        if 'username' in v and 'password' in v:
-            return v
-        raise CaptureSettingsError(f'HTTP credentials must have a username and a password: {v}')
+    def check_http_creds(cls, http_credentials: dict[str, str] | None) -> dict[str, str] | None:
+        if not http_credentials:
+            return None
+        if 'username' in http_credentials and 'password' in http_credentials:
+            return http_credentials
+        raise CaptureSettingsError(f'HTTP credentials must have a username and a password: {http_credentials}')
 
     @field_validator('geolocation', mode='before')
     @classmethod
-    def load_geolocation_json(cls, v: Any) -> dict[str, float] | None:
-        if not v:
+    def load_geolocation_json(cls, geolocation: Any) -> dict[str, float] | None:
+        if not geolocation:
+            return None
+        if isinstance(geolocation, str):
+            # ignore
             return None
-        if isinstance(v, str):
-            if v.startswith('{'):
-                return from_json(v)
-        elif isinstance(v, dict):
-            return v
+        elif isinstance(geolocation, dict):
+            return geolocation
         return None
 
     @field_validator('geolocation', mode='after')
     @classmethod
-    def check_geolocation(cls, v: dict[str, float] | None) -> dict[str, float] | None:
-        if not v:
-            return v
-        if 'latitude' in v and 'longitude' in v:
-            return v
-        raise CaptureSettingsError(f'A geolocation must have a latitude and a longitude: {v}')
+    def check_geolocation(cls, geolocation: dict[str, float] | None) -> dict[str, float] | None:
+        if not geolocation:
+            return None
+        if 'latitude' in geolocation and 'longitude' in geolocation:
+            return geolocation
+        raise CaptureSettingsError(f'A geolocation must have a latitude and a longitude: {geolocation}')
 
     @field_validator('viewport', mode='before')
     @classmethod
-    def load_viewport_json(cls, v: Any) -> dict[str, int] | None:
-        if not v:
+    def load_viewport_json(cls, viewport: Any) -> dict[str, int] | None:
+        if not viewport:
+            return None
+        if isinstance(viewport, str):
+            # ignore
             return None
-        if isinstance(v, str):
-            if v.startswith('{'):
-                return from_json(v)
-        elif isinstance(v, dict):
-            return v
+        elif isinstance(viewport, dict):
+            return viewport
         return None
 
     @field_validator('viewport', mode='after')
     @classmethod
-    def check_viewport(cls, v: dict[str, int] | None) -> dict[str, int] | None:
-        if not v:
-            return v
-        if 'width' in v and 'height' in v:
-            return v
-        raise CaptureSettingsError(f'A viewport must have a width and a height: {v}')
+    def check_viewport(cls, viewport: dict[str, int] | None) -> dict[str, int] | None:
+        if not viewport:
+            return None
+        if 'width' in viewport and 'height' in viewport:
+            return viewport
+        raise CaptureSettingsError(f'A viewport must have a width and a height: {viewport}')
 
     def redis_dump(self) -> Mapping[str | bytes, bytes | float | int | str]:
         mapping_capture: dict[str | bytes, bytes | float | int | str] = {}

lacuscore/lacuscore.py

@@ -433,8 +433,8 @@ class LacusCore():
 
             # Set default as chromium
             browser_engine: BROWSER = "chromium"
-            if to_capture.browser in ['chromium', 'firefox', 'webkit']:
-                browser_engine = to_capture.browser  # type: ignore[assignment]
+            if to_capture.browser:
+                browser_engine = to_capture.browser
             elif to_capture.user_agent:
                 parsed_string = user_agent_parser.ParseUserAgent(to_capture.user_agent)
                 browser_family = parsed_string['family'].lower()

lacuscore/task_logger.py

@@ -6,7 +6,6 @@ from typing import Any, Coroutine, Optional, TypeVar, Tuple
 
 import asyncio
 import functools
-import logging
 
 from .helpers import LacusCoreLogAdapter
 
@@ -43,7 +42,7 @@ def create_task(
 def _handle_task_result(
     task: asyncio.Task[Any],
     *,
-    logger: logging.Logger,
+    logger: 'LacusCoreLogAdapter',
     message: str,
     message_args: Tuple[Any, ...] = (),
 ) -> None:

{lacuscore-1.10.4.dist-info → lacuscore-1.10.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: lacuscore
-Version: 1.10.4
+Version: 1.10.5
 Summary: Core of Lacus, usable as a module
 Home-page: https://github.com/ail-project/LacusCore
 License: BSD-3-Clause
@@ -29,7 +29,7 @@ Requires-Dist: async-timeout (>=4.0.3,<5.0.0) ; python_version < "3.11"
 Requires-Dist: defang (>=0.5.3,<0.6.0)
 Requires-Dist: dnspython (>=2.6.1,<3.0.0)
 Requires-Dist: eval-type-backport (>=0.2.0,<0.3.0) ; python_version < "3.10"
-Requires-Dist: playwrightcapture[recaptcha] (>=1.25.6,<2.0.0)
+Requires-Dist: playwrightcapture[recaptcha] (>=1.25.7,<2.0.0)
 Requires-Dist: pydantic (>=2.8.2,<3.0.0)
 Requires-Dist: redis[hiredis] (>=5.0.7,<6.0.0)
 Requires-Dist: requests (>=2.32.3,<3.0.0)

lacuscore-1.10.5.dist-info/RECORD

@@ -0,0 +1,10 @@
+lacuscore/__init__.py,sha256=aLBshQPT9IBDKn5qWrX9A_exqtLFPyLsQiPWdfpAFjA,537
+lacuscore/helpers.py,sha256=cGapH4apNJIkKCF0vzumKCJ_ngAsqoq4IjA9iZWbkfU,10856
+lacuscore/lacus_monitoring.py,sha256=UOfE_1-_rhVeKJXQ_m9XxYkr7VwyQnA6iK-x_tcXJfo,2775
+lacuscore/lacuscore.py,sha256=eQMJBf0ZYkCkACu3OWCdidpNEYn0wZ1piIVHP1kv_kU,39372
+lacuscore/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+lacuscore/task_logger.py,sha256=kWBThqfv_alu1YA3jEqP4GsqXIVKLbzyI7w14aJ2g9I,1908
+lacuscore-1.10.5.dist-info/LICENSE,sha256=4C4hLYrIkUD96Ggk-y_Go1Qf7PBZrEm9PSeTGe2nd4s,1516
+lacuscore-1.10.5.dist-info/METADATA,sha256=5SVAbyOXH-4Qp-0KU2pbjWppMHHrGWhtGIrM31VQ0hk,2747
+lacuscore-1.10.5.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+lacuscore-1.10.5.dist-info/RECORD,,

lacuscore-1.10.4.dist-info/RECORD

@@ -1,10 +0,0 @@
-lacuscore/__init__.py,sha256=aLBshQPT9IBDKn5qWrX9A_exqtLFPyLsQiPWdfpAFjA,537
-lacuscore/helpers.py,sha256=GAn47fHikhVciuKtFK0-EbxHbBUNiImtUH6EJ_Vdgu4,9594
-lacuscore/lacus_monitoring.py,sha256=UOfE_1-_rhVeKJXQ_m9XxYkr7VwyQnA6iK-x_tcXJfo,2775
-lacuscore/lacuscore.py,sha256=rFOqUJ0cY58iFK0EwDRSqlSydLjEzUZZiWG5xHCNN_M,39437
-lacuscore/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-lacuscore/task_logger.py,sha256=8WbdJdKnGeFCxt9gtCNLI9vAQQZbsy2I5PRQpHP7XFU,1916
-lacuscore-1.10.4.dist-info/LICENSE,sha256=4C4hLYrIkUD96Ggk-y_Go1Qf7PBZrEm9PSeTGe2nd4s,1516
-lacuscore-1.10.4.dist-info/METADATA,sha256=N0kfRd8E1FjpauTYBnbUsWtQbSejNT8ae9gaNX9Zp0c,2747
-lacuscore-1.10.4.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-lacuscore-1.10.4.dist-info/RECORD,,