Installation
PyPasser is organized into a modular design. The core package is lightweight with minimal dependencies (httpx[socks] and cryptography), allowing you to install only what your application requires.
Core Installation
Install the base package from PyPI:
The base package supports:
- reCAPTCHA V3 (Pure HTTP anchor URL mode)
- ArvanCloud Interstitial (Pure HTTP JSFuck challenge clearance)
Capability Extras
PyPasser defines optional dependency groups for specialized capabilities:
# Browser automation (Playwright)
pip install "PyPasser[browsers]"
# Stealth browser engines (Patchright + Camoufox) — Recommended for Turnstile & V2
pip install "PyPasser[stealth]"
# Speech-to-text recognition (Vosk + ffmpeg + pydub)
pip install "PyPasser[speech]"
# Computer vision & template matching (Pillow + NumPy)
pip install "PyPasser[vision]"
# All capabilities bundled
pip install "PyPasser[all]"
Extras Requirement Matrix
Consult this table to determine which extras your target solver requires:
| Target Module | Required Extra | What It Installs |
|---|---|---|
pypasser.captcha.recaptcha.v3 (HTTP) |
None (Core) | httpx[socks], cryptography |
pypasser.captcha.recaptcha.v3 (Page mode) |
[browsers] or [stealth] |
playwright or patchright |
pypasser.captcha.recaptcha.v2 |
[browsers] + [speech] |
playwright, vosk, imageio-ffmpeg, pydub |
pypasser.captcha.geetest.v4 |
[vision] + [speech] |
pillow, numpy, vosk, imageio-ffmpeg |
pypasser.captcha.lemin |
[vision] |
pillow, numpy |
pypasser.captcha.turnstile |
[stealth] (recommended) or [browsers] |
patchright or playwright |
pypasser.captcha.arcaptcha |
[vision] + [speech] |
pillow, numpy, vosk, imageio-ffmpeg |
pypasser.antibot.cloudflare |
[stealth] (recommended) or [browsers] |
patchright or playwright |
pypasser.antibot.arvancloud |
None (Core) | httpx[socks] |
Browser Driver Setup
If you use any browser-based solver (recaptcha.v2, turnstile, antibot.cloudflare), ensure your browser binaries are installed:
Offline Speech Models (Vosk)
Solvers utilizing audio recognition (recaptcha.v2, arcaptcha, geetest.v4 voice) download their required Vosk language models automatically on the first run into ~/.cache/vosk/ or the local system cache.
- English model:
vosk-model-small-en-us-0.15(~40 MB) - Persian model:
vosk-model-fa-0.5(~1.4 GB, only for ARCaptcha voice)
Subsequent runs run completely offline from the local cache.