pythonlib: Add more leak warnings 0.2.8

Warns the user for when the passed parameters are likely to cause a leak.

.gitignore

@@ -1,20 +1,32 @@
+# Local builds
 /camoufox-*
 /firefox-*
 /mozilla-unified
-/extra-docs
-/.vscode
-_old/
 dist/
 bin/
-venv/
-/bundle/fonts/extra
 launch
 launch.exe
+
+# Internal testing
+/extra-docs
+/tests
+/.vscode
+/bundle/fonts/extra
+pythonlib/*.png
+scripts/*.png
+test*
+
+# Old data
+_old/
 *.old
-__pycache__/
-*.pyc
+
+# Logs
 wget-log
 *.kate-swp
 *.log
-test.py
+
+# Python interface
+venv/
+__pycache__/
+*.pyc
 *.mmdb

pythonlib/camoufox/exceptions.py

@@ -116,11 +116,3 @@ class InvalidOS(ValueError):
     """
 
     ...
-
-
-class DetectionWarning(RuntimeWarning):
-    """
-    Raised when a the user has a setting enabled that can cause detection.
-    """
-
-    ...

pythonlib/camoufox/fingerprints.py

@@ -1,14 +1,13 @@
-import os.path
 import re
 from dataclasses import asdict
 from typing import Any, Dict, Optional
 
 from browserforge.fingerprints import Fingerprint, FingerprintGenerator
-from yaml import CLoader, load
+
+from camoufox.pkgman import load_yaml
 
 # Load the browserforge.yaml file
-with open(os.path.join(os.path.dirname(__file__), 'browserforge.yml'), 'r') as f:
-    BROWSERFORGE_DATA = load(f, Loader=CLoader)
+BROWSERFORGE_DATA = load_yaml('browserforge.yml')
 
 FP_GENERATOR = FingerprintGenerator(browser='firefox', os=('linux', 'macos', 'windows'))
 

pythonlib/camoufox/pkgman.py

@@ -16,6 +16,7 @@ import requests
 from platformdirs import user_cache_dir
 from tqdm import tqdm
 from typing_extensions import TypeAlias
+from yaml import CLoader, load
 
 from .exceptions import UnsupportedArchitecture, UnsupportedOS
 
@@ -337,3 +338,11 @@ def unzip(
     with ZipFile(zip_file) as zf:
         for member in tqdm(zf.infolist(), desc=desc):
             zf.extract(member, extract_path)
+
+
+def load_yaml(file: str) -> dict:
+    """
+    Loads a local YAML file and returns it as a dictionary.
+    """
+    with open(Path(__file__).parent / file, 'r') as f:
+        return load(f, Loader=CLoader)

pythonlib/camoufox/utils.py

@@ -1,6 +1,5 @@
 import os
 import sys
-import warnings
 from os import environ
 from pprint import pprint
 from random import randrange
@@ -20,7 +19,6 @@ from .addons import (
     threaded_try_load_addons,
 )
 from .exceptions import (
-    DetectionWarning,
     InvalidOS,
     InvalidPropertyType,
     NonFirefoxFingerprint,
@@ -30,6 +28,7 @@ from .fingerprints import from_browserforge, generate_fingerprint
 from .ip import Proxy, public_ip, valid_ipv4, valid_ipv6
 from .locale import geoip_allowed, get_geolocation, normalize_locale
 from .pkgman import OS_NAME, get_path, installed_verstr
+from .warnings import LeakWarning
 from .xpi_dl import add_default_addons
 
 LAUNCH_FILE = {
@@ -197,12 +196,7 @@ def check_custom_fingerprint(fingerprint: Fingerprint) -> None:
             'If this is intentional, pass `i_know_what_im_doing=True`.'
         )
 
-    warnings.warn(
-        'Passing your own fingerprint is not recommended. '
-        'BrowserForge fingerprints are automatically generated within Camoufox '
-        'based on the provided `os` and `screen` constraints.',
-        category=DetectionWarning,
-    )
+    LeakWarning.warn('custom_fingerprint', False)
 
 
 def check_valid_os(os: ListOrString) -> None:
@@ -249,6 +243,45 @@ def set_into(target: Dict[str, Any], key: str, value: Any) -> None:
         target[key] = value
 
 
+def is_domain_set(
+    config: Dict[str, Any],
+    *properties: str,
+) -> bool:
+    """
+    Checks if a domain is set in the config.
+    """
+    for prop in properties:
+        # If the . prefix exists, check if the domain is a prefix of any key in the config
+        if prop.endswith('.'):
+            if any(key.startswith(prop) for key in config):
+                return True
+        # Otherwise, check if the domain is a direct key in the config
+        else:
+            if prop in config:
+                return True
+    return False
+
+
+def warn_manual_config(config: Dict[str, Any]) -> None:
+    """
+    Warns the user if they are manually setting properties that Camoufox already sets internally.
+    """
+    # Manual locale setting
+    if is_domain_set(
+        config, 'navigator.language', 'navigator.languages', 'headers.Accept-Language'
+    ):
+        LeakWarning.warn('locale', False)
+    # Manual User-Agent setting
+    if is_domain_set(config, 'headers.User-Agent'):
+        LeakWarning.warn('header-ua', False)
+    # Manual navigator setting
+    if is_domain_set(config, 'navigator.'):
+        LeakWarning.warn('navigator', False)
+    # Manual screen/window setting
+    if is_domain_set(config, 'screen.', 'window.', 'document.body.'):
+        LeakWarning.warn('viewport', False)
+
+
 def get_launch_options(
     *,
     config: Optional[Dict[str, Any]] = None,
@@ -282,24 +315,27 @@ def get_launch_options(
     if config is None:
         config = {}
 
+    # Set default values for optional arguments
     if addons is None:
         addons = []
     if args is None:
         args = []
     if firefox_user_prefs is None:
         firefox_user_prefs = {}
+    if i_know_what_im_doing is None:
+        i_know_what_im_doing = False
 
     # Warn the user if headless is being used
     # https://github.com/daijro/camoufox/issues/26
     if headless:
-        warnings.warn(
-            'It is currently not recommended to use headless mode in Camoufox. '
-            'Some WAFs are able to detect headless browsers. The issue is currently being investigated.',
-            category=DetectionWarning,
-        )
+        LeakWarning.warn('headless', i_know_what_im_doing)
     elif headless is None:
         headless = False
 
+    # Warn the user for manual config settings
+    if not i_know_what_im_doing:
+        warn_manual_config(config)
+
     # Assert the target OS is valid
     if os:
         check_valid_os(os)
@@ -366,6 +402,14 @@ def get_launch_options(
         geolocation = get_geolocation(geoip)
         config.update(geolocation.as_config())
 
+    # Raise a warning when a proxy is being used without spoofing geolocation
+    elif (
+        proxy
+        and 'localhost' not in proxy.get('server', '')
+        and not is_domain_set('geolocation', config)
+    ):
+        LeakWarning.warn('proxy_without_geoip')
+
     # Set locale
     if locale:
         parsed_locale = normalize_locale(locale)
@@ -391,10 +435,13 @@ def get_launch_options(
     if block_webrtc:
         firefox_user_prefs['media.peerconnection.enabled'] = False
     if allow_webgl:
+        LeakWarning.warn('allow_webgl', i_know_what_im_doing)
         firefox_user_prefs['webgl.disabled'] = False
 
-    # Launch
+    # Load the addons
     threaded_try_load_addons(get_debug_port(args), addons)
+
+    # Prepare environment variables to pass to Camoufox
     env_vars = {
         **get_env_vars(config, target_os),
         **(cast(Dict[str, Union[str, float, bool]], environ) if env is None else env),

pythonlib/camoufox/warnings.py

@@ -0,0 +1,44 @@
+import inspect
+import warnings
+from pathlib import Path
+from typing import Optional
+
+from camoufox.pkgman import load_yaml
+
+WARNINGS_DATA = load_yaml('warnings.yml')
+
+
+class LeakWarning(RuntimeWarning):
+    """
+    Raised when a the user has a setting enabled that can cause detection.
+    """
+
+    @staticmethod
+    def warn(warning_key: str, i_know_what_im_doing: Optional[bool] = None) -> None:
+        """
+        Warns the user if a passed parameter can cause leaks.
+        """
+        warning = WARNINGS_DATA[warning_key]
+        if i_know_what_im_doing:
+            return
+        if i_know_what_im_doing is not None:
+            warning += '\nIf this is intentional, pass `i_know_what_im_doing=True`.'
+
+        # Get caller information
+        current_module = Path(__file__).parent
+        frame = inspect.currentframe()
+        while frame:
+            if not Path(frame.f_code.co_filename).is_relative_to(current_module):
+                break
+            frame = frame.f_back
+
+        if frame:
+            warnings.warn_explicit(
+                warning,
+                category=LeakWarning,
+                filename=frame.f_code.co_filename,
+                lineno=frame.f_lineno,
+            )
+            return
+
+        warnings.warn(warning, category=LeakWarning)

pythonlib/camoufox/warnings.yml

@@ -0,0 +1,32 @@
+headless: >-
+    Headless mode in Camoufox is not recommended at this time.
+    Some WAFs are able to detect headless browsers. The issue is currently being investigated.
+
+navigator: >-
+    Manually setting navigator properties is not recommended.
+    Device information is automatically generated within Camoufox
+    based on the provided `os`.
+
+locale: >-
+    Use the `locale` parameter in Camoufox instead of setting the config manually.
+
+header-ua: >-
+    Do not set the header.User-Agent manually. Camoufox will generate a User-Agent for you.
+
+viewport: >-
+    Manually setting screen & window properties is not recommended.
+    Screen dimensions are randomly generated within Camoufox
+    based on the provided screen constraints. See here:
+    https://github.com/daijro/camoufox/tree/main/pythonlib#browserforge-integration.
+
+custom_fingerprint: >-
+    Passing your own fingerprint is not recommended.
+    BrowserForge fingerprints are automatically generated within Camoufox
+    based on the provided `os` and `screen` constraints.
+
+proxy_without_geoip: >-
+    When using a proxy, it is heavily recommended that you pass `geoip=True`.
+
+allow_webgl: >-
+    Enabling WebGL can lead to Canvas fingerprinting and detection.
+    Camoufox will automatically spoof your vendor and renderer, but it cannot spoof your WebGL fingerprint.

pythonlib/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "camoufox"
-version = "0.2.7"
+version = "0.2.8"
 description = "Wrapper around Playwright to help launch Camoufox"
 authors = ["daijro <daijro.dev@gmail.com>"]
 license = "MIT"