rqsession 0.4.7

Browser TLS fingerprint impersonation library powered by Rust (BoringSSL + hyper).

rqsession

中文 | English

A Python HTTP library that impersonates real browser TLS fingerprints, powered by a Rust native extension (BoringSSL + hyper + tokio).

---

What it does

Most anti-bot systems (Cloudflare, Akamai, DataDome, etc.) inspect the TLS ClientHello and HTTP/2 SETTINGS frame to distinguish scrapers from real browsers. Standard requests or httpx produce a recognizable Python fingerprint regardless of what User-Agent you set.

rqsession controls the full fingerprint stack at the Rust level:

Layer	What is controlled
TLS ClientHello	Cipher suite order, supported groups (curves), ALPN, signature algorithms, version range
HTTP/2	SETTINGS frame values (window size, max frame size), connection WINDOW_UPDATE
HTTP headers	Exact header order, browser-specific headers (sec-ch-ua, sec-fetch-*, etc.)

No external proxy process required — it's a compiled .pyd / .so extension, imported directly.

---

Installation

pip install rqsession

Pre-built wheels are available for Windows x86_64 and Linux x86_64 (Python 3.9+).
Other platforms require a local Rust toolchain to build from the source distribution.

---

Quick Start

Synchronous

from rqsession.rust_session import BrowserSession, Chrome120

with BrowserSession(Chrome120) as s:
    resp = s.get("https://httpbin.org/get")
    print(resp.status_code)   # 200
    print(resp.json())

Async

import asyncio
from rqsession.rust_session import AsyncBrowserSession, Chrome120

async def main():
    async with AsyncBrowserSession(Chrome120) as s:
        resp = await s.get("https://httpbin.org/get")
        print(resp.status_code)
        print(resp.json())

asyncio.run(main())

Concurrent async requests

import asyncio
from rqsession.rust_session import AsyncBrowserSession, Chrome120

async def main():
    async with AsyncBrowserSession(Chrome120) as s:
        results = await asyncio.gather(
            s.get("https://httpbin.org/get"),
            s.get("https://httpbin.org/ip"),
            s.get("https://httpbin.org/user-agent"),
        )
    for r in results:
        print(r.status_code)

asyncio.run(main())

---

Built-in Browser Profiles

Import name	Browser	OS
Chrome138	Chrome 138	Windows
Chrome120	Chrome 120	Windows
Chrome119	Chrome 119	Windows
Edge147	Edge 147	Windows
Edge142	Edge 142	Windows
Edge141	Edge 141	Windows
Firefox146	Firefox 146	Windows
Firefox133	Firefox 133	Windows
Tor128	Tor Browser 128	Windows
Safari17	Safari 17	macOS
MacosChrome140	Chrome 140	macOS
AndroidChrome114	Chrome 114	Android
Py37Aiohttp381	Python aiohttp 3.8.1	Windows

from rqsession.rust_session import (
    BrowserSession, AsyncBrowserSession,
    Chrome138, Chrome120, Chrome119,
    Edge147, Edge142, Edge141,
    Firefox146, Firefox133,
    Tor128, Safari17, MacosChrome140,
    AndroidChrome114, Py37Aiohttp381,
)

Each profile carries its own cipher suite order, curves, signature algorithms, HTTP/2 settings, and header order — matching the real browser's wire behavior.

---

Proxy Support

Both sync and async sessions accept a proxy URL at construction time:

# HTTP proxy
s = BrowserSession(Chrome120, proxy="http://127.0.0.1:7890")

# Authenticated proxy
s = BrowserSession(Chrome120, proxy="http://user:pass@host:port")

# Async
async with AsyncBrowserSession(Firefox133, proxy="http://127.0.0.1:7890") as s:
    resp = await s.get("https://example.com")

The proxy tunnel is implemented as an HTTP CONNECT connection at the TCP level, so TLS fingerprinting applies end-to-end through the tunnel.

---

Session API

Constructor parameters

BrowserSession(
    profile,              # built-in constant or BrowserProfile object
    *,
    proxy=None,           # "http://[user:pass@]host:port"
    verify=True,          # set False to skip SSL verification
    ca_bundle=None,       # path to CA bundle; auto-detects certifi when None
)
# AsyncBrowserSession takes the same parameters

Request methods

# GET
resp = s.get(url, headers={...}, params={...})

# POST — form data or JSON
resp = s.post(url, data=b"...", headers={...})
resp = s.post(url, json={"key": "value"})

# Generic
resp = s.request("PUT", url, headers={...}, json={...})

# Async versions — same signature, add `await`
resp = await s.get(url)
resp = await s.post(url, json={...})

# Disable redirect following (default: allow_redirects=True)
resp = s.get(url, allow_redirects=False)   # returns the 3xx directly
resp = await s.get(url, allow_redirects=False)

Response object

Compatible with requests.Response style:

resp.status_code    # int
resp.text           # str (auto-decoded)
resp.content        # bytes
resp.headers        # dict[str, str]
resp.json()         # parsed JSON → dict / list
resp.ok             # True when status < 400
resp.raise_for_status()   # raises on 4xx/5xx
resp.url            # final URL after redirects

Response bodies are automatically decompressed (gzip, deflate, br, zstd).

Cookies

# Persist cookies manually
s.update_cookies({"token": "abc123"})

# Cookies from Set-Cookie response headers are automatically
# stored in the session and sent on subsequent requests.

---

Adding Custom Browser Profiles

Capture a fingerprint from a real browser using tls.peet.ws/api/all, then convert it:

# Save output from tls.peet.ws as chrome136.json, then:
python tools/tls_peet_to_profile.py chrome136.json -n chrome136_windows --install

The --install flag writes the profile directly to rqsession/rust_session/profiles/builtin/.
You can also load profiles at runtime without rebuilding:

from rqsession.rust_session import BrowserSession, load_custom, load_profile_json
from rqsession._rust_core import load_profile

# From profiles/custom/<name>.json
s = BrowserSession(load_custom("my_browser"))

# From a JSON string
s = BrowserSession(load_profile_json('{"name": "...", "tls": {...}, ...}'))

# From an arbitrary file path
s = BrowserSession(load_profile("/path/to/profile.json"))

# List available profiles
from rqsession.rust_session import list_builtin, list_custom
print(list_builtin())   # ['chrome120_windows', 'firefox133_windows', ...]
print(list_custom())    # ['my_browser', ...]

---

Windows asyncio note

Python 3.8+ on Windows uses ProactorEventLoop by default. If you see event-loop errors, switch to SelectorEventLoop before asyncio.run():

import asyncio, sys

if sys.platform == "win32":
    asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())

asyncio.run(main())

---

Legacy layers

Earlier versions of this library included RequestSession and EnhancedRequestSession. These layers are still available for backward compatibility:

from rqsession import RequestSession          # basic requests.Session wrapper
from rqsession import EnhancedRequestSession  # requests.Session wrapper with browser headers

For new projects, use BrowserSession / AsyncBrowserSession instead — they are faster, require no external process, and provide more accurate fingerprinting.

---

License

Apache 2.0 — see LICENSE.