feat: Implement PyDoll engine

[nyx-rattapoom]

Summary

Introduces a new browser engine, PydollEngine, utilizing the pydoll-python library for Chrome browser automation.

Key Changes

• New PydollEngine:

  • Implemented in engines/pydoll/pydoll_engine.py.
  • Integrates pydoll-python's Chrome client and ChromiumOptions.
  • Supports headless/headed modes, custom user agent configuration, and proxy setup (HTTP/HTTPS protocols, including authentication).
  • Inherits from BrowserEngine and implements start(), stop(), navigate(), reload_page(), locator(), get_page_content(), execute_js(), and screenshot().
  • Handles extraction of values from CDP nested responses (via _extract_value_from_cdp).
  • Correctly tracks spawned child processes using psutil.

• Engine Registration:

  • Registered pydoll-chrome_headless and pydoll-chrome configurations in config/engines.py, making them available for selection in the benchmark.

• Dependency Updates:

  • Added pydoll-python>=2.23.0 to requirements.txt.
  • Updated aiofiles version constraints to >=25.1.0.

[KittisakPh]

Vs. Camoufox what is better ?

[nyx-rattapoom]

@KittisakPh
It's quite hard to judge which one is better; it depends on website protection. Some sites will aggressively flag you when you use Firefox. Any undetected CDP might be better, such as nodriver, zendriver, pydoll, and seleniumbase.

As I tried pydoll, it's on par with seleniumbase with a better API interface, less confusing, but CDP patch is easy to detect compared to native patches like camoufox and cloakbrowser.