mirror of
https://github.com/feder-cr/invisible_playwright.git
synced 2026-06-07 08:35:12 +02:00
test: fingerprint surface + consistency e2e tests
96 e2e tests reproducing the canonical anti-bot / fingerprinting
libraries' checks against a local InvisiblePlaywright session
on about:blank (no network).
Surface (28 tests):
- BotD: webdriver, app_version, UA tokens, function_bind, productSub,
process, eval.length, languages, plugins, mimeTypes, distinctive
window/document globals, html attributes, window size, webgl_debug
- sannysoft: chrome consistency, permissions.query, iframe chrome,
iframe languages
- FpJS: canvas 2D, audio offline, color-gamut, color-depth
- PIN-locked: screen.width/height, hw.concurrency, audio.sampleRate,
audio.maxChannelCount
- fpscanner: UA<->platform, no userAgentData on Firefox
Consistency (68 tests):
- Math determinism: 17 transcendentals + Math.pow
- Worker scope vs main: 5 navigator props via Blob worker
- Iframe scope vs window: 4 props + screen
- UA self-consistency: UA<->platform, UA<->oscpu, UA<->appVersion
- Native function self-toString: 8 native APIs
- AudioContext / WebGL determinism
- Locale<->Intl: DateTimeFormat / NumberFormat / Collator
- Descriptor shape lies: 16 navigator props (each must be a getter,
not Object.defineProperty(value=...))
- performance.timeOrigin + .now() monotonic
- Window dimension invariants
- Firefox UA invariants (vendor='', appName='Netscape',
appVersion short form)
All marked @pytest.mark.e2e so they're excluded from the default
suite that the pre-push hook runs. Invoke explicitly:
pytest -m e2e -v
Or against a local build:
INVPW_BINARY_PATH=/path/to/firefox.exe pytest -m e2e -v
Sources: github.com/fingerprintjs/BotD, abrahamjuliot/creepjs,
fingerprintjs/fingerprintjs, antoinevastel/fpscanner,
niespodd/browser-fingerprinting, bot.sannysoft.com.
Verified: 95 passed, 1 skipped (Chromium-only invariant), 0 failed
against firefox-5 local build.
This commit is contained in:
parent
22b1171518
commit
cf59e98fa9
2 changed files with 977 additions and 0 deletions
547
tests/test_fingerprint_consistency.py
Normal file
547
tests/test_fingerprint_consistency.py
Normal file
|
|
@ -0,0 +1,547 @@
|
|||
"""Fingerprint consistency / lie-detection tests.
|
||||
|
||||
Complementary to test_fingerprint_surface.py: those tests ask "do you
|
||||
look like a real browser?" — these ask "are your fingerprint surfaces
|
||||
INTERNALLY CONSISTENT?"
|
||||
|
||||
Anti-bot systems catch spoofers not by checking each signal in
|
||||
isolation but by cross-checking related signals. If you spoof UA to
|
||||
"Windows" but leave navigator.platform as "Linux x86_64", or you spoof
|
||||
WebGL renderer in the main thread but not in a Web Worker, the
|
||||
inconsistency proves the spoof is fake.
|
||||
|
||||
Sources studied (all FOSS, MIT-licensed):
|
||||
- creepjs/src/lies/index.ts — the canonical lie detector
|
||||
- creepjs/src/worker/index.ts — main-vs-worker scope cross-check
|
||||
- creepjs/src/math/index.ts — Math.x(p) deterministic equality
|
||||
- creepjs/src/navigator/index.ts — UA/platform/oscpu invariants
|
||||
- niespodd/browser-fingerprinting README — worker hwConcurrency,
|
||||
plugin chain, perf.timeOrigin
|
||||
|
||||
Everything runs against `about:blank` with NO network and NO proxy.
|
||||
|
||||
Run only this file:
|
||||
pytest tests/test_fingerprint_consistency.py -m e2e -v
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import sys
|
||||
|
||||
import pytest
|
||||
|
||||
from invisible_playwright import InvisiblePlaywright
|
||||
from invisible_playwright.constants import BINARY_ENTRY_REL
|
||||
|
||||
|
||||
PIN = {
|
||||
"screen.width": 1920,
|
||||
"screen.height": 1080,
|
||||
"screen.avail_width": 1920,
|
||||
"screen.avail_height": 1040,
|
||||
"screen.dpr": 1.0,
|
||||
"hardware.concurrency": 8,
|
||||
"audio.sample_rate": 48000,
|
||||
"audio.max_channel_count": 2,
|
||||
}
|
||||
|
||||
|
||||
@pytest.fixture(scope="session")
|
||||
def firefox_binary():
|
||||
"""See test_fingerprint_surface.firefox_binary for the lookup chain."""
|
||||
import os
|
||||
env_path = os.environ.get("INVPW_BINARY_PATH")
|
||||
if env_path:
|
||||
from pathlib import Path
|
||||
if Path(env_path).exists():
|
||||
return env_path
|
||||
pytest.skip(f"INVPW_BINARY_PATH={env_path!r} does not exist")
|
||||
if sys.platform not in BINARY_ENTRY_REL:
|
||||
pytest.skip(f"unsupported platform: {sys.platform}")
|
||||
from invisible_playwright.download import cache_dir_for_version
|
||||
entry = cache_dir_for_version() / BINARY_ENTRY_REL[sys.platform]
|
||||
if not entry.exists():
|
||||
pytest.skip(
|
||||
"patched Firefox not cached; run "
|
||||
"`python -m invisible_playwright fetch` first, or set "
|
||||
"INVPW_BINARY_PATH to a local build"
|
||||
)
|
||||
return str(entry)
|
||||
|
||||
|
||||
@pytest.fixture(scope="module")
|
||||
def page(firefox_binary):
|
||||
with InvisiblePlaywright(
|
||||
seed=42,
|
||||
pin=PIN,
|
||||
binary_path=firefox_binary,
|
||||
headless=True,
|
||||
) as browser:
|
||||
ctx = browser.new_context()
|
||||
p = ctx.new_page()
|
||||
p.goto("about:blank", timeout=30_000)
|
||||
yield p
|
||||
|
||||
|
||||
def _ev(page, expr):
|
||||
return page.evaluate(expr)
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 1. Math determinism — same input MUST yield same output
|
||||
# Source: creepjs/src/math/index.ts
|
||||
# A wrapper that adds noise to Math.* (canvas-spoofing prefs) exposes
|
||||
# itself here: two consecutive calls with the same input must be
|
||||
# byte-identical.
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
@pytest.mark.parametrize("fn,arg", [
|
||||
("cos", "1e308"),
|
||||
("acos", "0.5"),
|
||||
("asin", "0.5"),
|
||||
("atan", "Math.PI"),
|
||||
("atanh", "0.5"),
|
||||
("cbrt", "Math.PI"),
|
||||
("cosh", "Math.PI"),
|
||||
("exp", "Math.PI"),
|
||||
("expm1", "Math.PI"),
|
||||
("log", "Math.PI"),
|
||||
("log1p", "Math.PI"),
|
||||
("log10", "Math.PI"),
|
||||
("sin", "Math.PI"),
|
||||
("sinh", "Math.PI"),
|
||||
("sqrt", "Math.PI"),
|
||||
("tan", "Math.PI"),
|
||||
("tanh", "Math.PI"),
|
||||
])
|
||||
def test_math_determinism(page, fn, arg):
|
||||
"""Math.<fn>(<arg>) must return the same value across 100 calls."""
|
||||
first, last, all_equal = _ev(page, f"""() => {{
|
||||
const r = [];
|
||||
for (let i = 0; i < 100; i++) r.push(Math.{fn}({arg}));
|
||||
return [r[0], r[99], r.every(x => Object.is(x, r[0]))];
|
||||
}}""")
|
||||
assert all_equal, (
|
||||
f"Math.{fn}({arg}) drifts across calls: first={first}, last={last}"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_math_pow_two_arg_determinism(page):
|
||||
ok = _ev(page, """() => {
|
||||
const a = Math.pow(Math.PI, 2);
|
||||
for (let i = 0; i < 50; i++) {
|
||||
if (!Object.is(Math.pow(Math.PI, 2), a)) return false;
|
||||
}
|
||||
return true;
|
||||
}""")
|
||||
assert ok
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 2. Worker scope vs main thread — navigator properties MUST agree
|
||||
# Source: creepjs/src/worker/index.ts
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
def _worker_navigator_dict(page, props):
|
||||
expr = """async (props) => {
|
||||
const code = `
|
||||
self.onmessage = (e) => {
|
||||
const out = {};
|
||||
for (const p of e.data) {
|
||||
try { out[p] = self.navigator[p]; }
|
||||
catch (err) { out[p] = '<error: ' + err.message + '>'; }
|
||||
}
|
||||
if (out.languages && Array.isArray(out.languages)) {
|
||||
out.languages = [...out.languages];
|
||||
}
|
||||
self.postMessage(out);
|
||||
};
|
||||
`;
|
||||
const blob = new Blob([code], { type: 'application/javascript' });
|
||||
const url = URL.createObjectURL(blob);
|
||||
const worker = new Worker(url);
|
||||
try {
|
||||
const result = await new Promise((resolve, reject) => {
|
||||
worker.onmessage = (e) => resolve(e.data);
|
||||
worker.onerror = (e) => reject(new Error(e.message));
|
||||
worker.postMessage(props);
|
||||
setTimeout(() => reject(new Error('worker timeout')), 5000);
|
||||
});
|
||||
return result;
|
||||
} finally {
|
||||
worker.terminate();
|
||||
URL.revokeObjectURL(url);
|
||||
}
|
||||
}"""
|
||||
return page.evaluate(expr, list(props))
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_worker_userAgent_matches_main(page):
|
||||
main = _ev(page, "navigator.userAgent")
|
||||
worker = _worker_navigator_dict(page, ("userAgent",))
|
||||
assert worker["userAgent"] == main, (
|
||||
f"UA drift main vs worker:\n main: {main!r}\n worker: {worker['userAgent']!r}"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_worker_hardwareConcurrency_matches_main(page):
|
||||
main = _ev(page, "navigator.hardwareConcurrency")
|
||||
worker = _worker_navigator_dict(page, ("hardwareConcurrency",))
|
||||
assert worker["hardwareConcurrency"] == main
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_worker_language_matches_main(page):
|
||||
main = _ev(page, "navigator.language")
|
||||
worker = _worker_navigator_dict(page, ("language",))
|
||||
assert worker["language"] == main
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_worker_languages_matches_main(page):
|
||||
main = _ev(page, "[...navigator.languages]")
|
||||
worker = _worker_navigator_dict(page, ("languages",))
|
||||
assert list(worker["languages"]) == list(main)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_worker_platform_matches_main(page):
|
||||
main = _ev(page, "navigator.platform")
|
||||
worker = _worker_navigator_dict(page, ("platform",))
|
||||
assert worker["platform"] == main
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 3. Iframe scope vs window scope
|
||||
# Source: creepjs/src/lies/index.ts (getBehemothIframe pattern)
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
def _iframe_navigator_dict(page, props):
|
||||
expr = """(props) => {
|
||||
const iframe = document.createElement('iframe');
|
||||
iframe.style.display = 'none';
|
||||
document.body.appendChild(iframe);
|
||||
const out = {};
|
||||
for (const p of props) {
|
||||
try { out[p] = iframe.contentWindow.navigator[p]; }
|
||||
catch (e) { out[p] = '<error: ' + e.message + '>'; }
|
||||
}
|
||||
if (Array.isArray(out.languages)) out.languages = [...out.languages];
|
||||
document.body.removeChild(iframe);
|
||||
return out;
|
||||
}"""
|
||||
return page.evaluate(expr, list(props))
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_iframe_userAgent_matches_window(page):
|
||||
main = _ev(page, "navigator.userAgent")
|
||||
iframe = _iframe_navigator_dict(page, ("userAgent",))
|
||||
assert iframe["userAgent"] == main
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_iframe_language_matches_window(page):
|
||||
main = _ev(page, "navigator.language")
|
||||
iframe = _iframe_navigator_dict(page, ("language",))
|
||||
assert iframe["language"] == main
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_iframe_hardwareConcurrency_matches_window(page):
|
||||
main = _ev(page, "navigator.hardwareConcurrency")
|
||||
iframe = _iframe_navigator_dict(page, ("hardwareConcurrency",))
|
||||
assert iframe["hardwareConcurrency"] == main
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_iframe_screen_matches_window(page):
|
||||
main = _ev(page, "[screen.width, screen.height]")
|
||||
iframe = _ev(page, """() => {
|
||||
const f = document.createElement('iframe');
|
||||
f.style.display = 'none';
|
||||
document.body.appendChild(f);
|
||||
const v = [f.contentWindow.screen.width, f.contentWindow.screen.height];
|
||||
document.body.removeChild(f);
|
||||
return v;
|
||||
}""")
|
||||
assert iframe == main
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 4. UA self-consistency (creepjs/src/navigator/index.ts)
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_navigator_platform_matches_userAgent_OS(page):
|
||||
ua = _ev(page, "navigator.userAgent")
|
||||
platform = _ev(page, "navigator.platform")
|
||||
if "Windows" in ua:
|
||||
assert "Win" in platform
|
||||
elif "Mac" in ua:
|
||||
assert "Mac" in platform
|
||||
elif "Linux" in ua or "X11" in ua:
|
||||
assert "Linux" in platform or "X11" in platform
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_navigator_oscpu_matches_userAgent(page):
|
||||
"""Firefox-only: navigator.oscpu must correlate with UA OS."""
|
||||
ua = _ev(page, "navigator.userAgent")
|
||||
oscpu = _ev(page, "navigator.oscpu || ''")
|
||||
if not oscpu:
|
||||
pytest.skip("navigator.oscpu not exposed")
|
||||
if "Windows" in ua:
|
||||
assert "Windows" in oscpu
|
||||
elif "Linux" in ua:
|
||||
assert "Linux" in oscpu
|
||||
elif "Mac" in ua:
|
||||
assert "Mac" in oscpu
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_userAgent_contains_appVersion_chromium_only(page):
|
||||
"""Chromium invariant: UA contains appVersion. Firefox uses a short
|
||||
appVersion form so the check is gated on `'chrome' in window`."""
|
||||
if not _ev(page, "'chrome' in window"):
|
||||
pytest.skip("Chromium-only invariant")
|
||||
ua = _ev(page, "navigator.userAgent")
|
||||
av = _ev(page, "navigator.appVersion")
|
||||
assert av in ua
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 5. Native function self-toString (creepjs/src/lies/index.ts hasKnownToString)
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
def _is_native_toString(text, fn_name):
|
||||
"""Mirror of CreepJS hasKnownToString — accept the engine-specific
|
||||
native patterns (single-line on V8, multi-line on SpiderMonkey)."""
|
||||
import re as _re
|
||||
name = _re.escape(fn_name)
|
||||
patterns = [
|
||||
rf"^function {name}\(\) \{{ \[native code\] \}}$",
|
||||
rf"^function get {name}\(\) \{{ \[native code\] \}}$",
|
||||
rf"^function {name}\(\) \{{[\s\S]*\[native code\][\s\S]*\}}$",
|
||||
rf"^function get {name}\(\) \{{[\s\S]*\[native code\][\s\S]*\}}$",
|
||||
]
|
||||
return any(_re.match(p, text) for p in patterns)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
@pytest.mark.parametrize("native_fn,name", [
|
||||
("Function.prototype.toString", "toString"),
|
||||
("Function.prototype.bind", "bind"),
|
||||
("Function.prototype.call", "call"),
|
||||
("Function.prototype.apply", "apply"),
|
||||
("Object.getOwnPropertyDescriptor", "getOwnPropertyDescriptor"),
|
||||
("Object.defineProperty", "defineProperty"),
|
||||
("Array.prototype.slice", "slice"),
|
||||
("JSON.stringify", "stringify"),
|
||||
])
|
||||
def test_native_function_self_toString_matches(page, native_fn, name):
|
||||
"""Each native function's `.toString()` must match its engine's
|
||||
native pattern. A Proxy wrapper or function-rewrite leaks here."""
|
||||
text = _ev(page, f"{native_fn}.toString()")
|
||||
assert _is_native_toString(text, name), (
|
||||
f"{native_fn}.toString() not native-shape: {text!r}"
|
||||
)
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 6. AudioContext / WebGL determinism
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_audio_offline_context_deterministic(page):
|
||||
"""OfflineAudioContext: same graph → byte-identical output."""
|
||||
ok = _ev(page, """async () => {
|
||||
async function render() {
|
||||
const ctx = new (window.OfflineAudioContext ||
|
||||
window.webkitOfflineAudioContext)(1, 5000, 44100);
|
||||
const osc = ctx.createOscillator();
|
||||
osc.connect(ctx.destination);
|
||||
osc.start(0);
|
||||
const buf = await ctx.startRendering();
|
||||
return Array.from(buf.getChannelData(0).slice(0, 50));
|
||||
}
|
||||
const a = await render();
|
||||
const b = await render();
|
||||
return JSON.stringify(a) === JSON.stringify(b);
|
||||
}""")
|
||||
assert ok
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_webgl_getParameter_deterministic(page):
|
||||
"""WebGL parameters must not drift across reads."""
|
||||
ok = _ev(page, """() => {
|
||||
const c = document.createElement('canvas');
|
||||
const gl = c.getContext('webgl');
|
||||
if (!gl) return false;
|
||||
const params = [gl.MAX_TEXTURE_SIZE, gl.MAX_VIEWPORT_DIMS,
|
||||
gl.MAX_RENDERBUFFER_SIZE, gl.MAX_VERTEX_ATTRIBS];
|
||||
const ref = JSON.stringify(params.map(p => gl.getParameter(p)));
|
||||
for (let i = 0; i < 50; i++) {
|
||||
if (JSON.stringify(params.map(p => gl.getParameter(p))) !== ref) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
return true;
|
||||
}""")
|
||||
assert ok
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 7. Locale ↔ Intl cross-consistency
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_navigator_language_matches_Intl_locale(page):
|
||||
"""navigator.language base must agree with Intl.DateTimeFormat locale."""
|
||||
nav = _ev(page, "navigator.language").split("-")[0]
|
||||
intl = _ev(page,
|
||||
"Intl.DateTimeFormat().resolvedOptions().locale").split("-")[0]
|
||||
assert nav == intl, (
|
||||
f"navigator.language base={nav!r} vs Intl={intl!r}"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_navigator_language_matches_Intl_NumberFormat(page):
|
||||
nav = _ev(page, "navigator.language").split("-")[0]
|
||||
num = _ev(page,
|
||||
"Intl.NumberFormat().resolvedOptions().locale").split("-")[0]
|
||||
assert nav == num
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_navigator_language_matches_Intl_Collator(page):
|
||||
nav = _ev(page, "navigator.language").split("-")[0]
|
||||
col = _ev(page,
|
||||
"(new Intl.Collator()).resolvedOptions().locale").split("-")[0]
|
||||
assert nav == col
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 8. Property descriptor shape lies
|
||||
# Spoofers using Object.defineProperty(navigator, prop, {value: ...})
|
||||
# leave a 'value' field on the descriptor — real native props use a getter.
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
_DESCRIPTOR_NATIVE_PROPS = [
|
||||
"userAgent", "platform", "hardwareConcurrency", "language", "languages",
|
||||
"vendor", "appVersion", "appName", "appCodeName", "doNotTrack",
|
||||
"cookieEnabled", "onLine", "product", "productSub", "buildID", "oscpu",
|
||||
]
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
@pytest.mark.parametrize("prop", _DESCRIPTOR_NATIVE_PROPS)
|
||||
def test_navigator_property_descriptor_is_getter_not_value(page, prop):
|
||||
"""Each spoofable navigator.* property must be defined via a native
|
||||
getter — NOT Object.defineProperty(..., {value: x}). The value-field
|
||||
descriptor is the lazy spoof leak CreepJS catches."""
|
||||
has_lie = _ev(page, f"""() => {{
|
||||
let proto = navigator;
|
||||
let descriptor = null;
|
||||
while (proto && !descriptor) {{
|
||||
descriptor = Object.getOwnPropertyDescriptor(proto, {prop!r});
|
||||
proto = Object.getPrototypeOf(proto);
|
||||
}}
|
||||
if (!descriptor) return null;
|
||||
return 'value' in descriptor;
|
||||
}}""")
|
||||
if has_lie is None:
|
||||
pytest.skip(f"navigator.{prop} not exposed")
|
||||
assert has_lie is False, (
|
||||
f"navigator.{prop} descriptor exposes 'value' field — lazy spoof"
|
||||
)
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 9. performance.timeOrigin + monotonic
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_performance_timeOrigin_stable(page):
|
||||
assert _ev(page,
|
||||
"performance.timeOrigin === performance.timeOrigin")
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_performance_now_monotonic(page):
|
||||
ok = _ev(page, """() => {
|
||||
let prev = performance.now();
|
||||
for (let i = 0; i < 100; i++) {
|
||||
const cur = performance.now();
|
||||
if (cur < prev) return false;
|
||||
prev = cur;
|
||||
}
|
||||
return true;
|
||||
}""")
|
||||
assert ok
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 10. Window dimension invariants
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_window_inner_not_larger_than_outer(page):
|
||||
inner, outer = _ev(page, "[window.innerWidth, window.outerWidth]")
|
||||
assert inner <= outer
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_screen_avail_not_larger_than_screen(page):
|
||||
aw, w = _ev(page, "[screen.availWidth, screen.width]")
|
||||
ah, h = _ev(page, "[screen.availHeight, screen.height]")
|
||||
assert aw <= w and ah <= h
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# 11. Firefox UA invariants
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_firefox_UA_implies_empty_vendor(page):
|
||||
"""Firefox: navigator.vendor === ''"""
|
||||
if "Firefox" not in _ev(page, "navigator.userAgent"):
|
||||
pytest.skip("Firefox-only invariant")
|
||||
if "Chrome" in _ev(page, "navigator.userAgent"):
|
||||
pytest.skip("Chrome+Firefox UA — likely synthetic")
|
||||
assert _ev(page, "navigator.vendor") == ""
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_firefox_appVersion_short_form(page):
|
||||
"""Real Firefox's appVersion is '5.0 (Windows)' form, not the full UA."""
|
||||
if "Firefox" not in _ev(page, "navigator.userAgent"):
|
||||
pytest.skip("Firefox-only invariant")
|
||||
av = _ev(page, "navigator.appVersion")
|
||||
ua = _ev(page, "navigator.userAgent")
|
||||
assert av.startswith("5.0 (")
|
||||
assert len(av) < len(ua)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_firefox_UA_implies_appName_Netscape(page):
|
||||
"""navigator.appName === 'Netscape' (historical invariant)."""
|
||||
if "Firefox" not in _ev(page, "navigator.userAgent"):
|
||||
pytest.skip("Firefox-only invariant")
|
||||
assert _ev(page, "navigator.appName") == "Netscape"
|
||||
430
tests/test_fingerprint_surface.py
Normal file
430
tests/test_fingerprint_surface.py
Normal file
|
|
@ -0,0 +1,430 @@
|
|||
"""Fingerprint surface tests — replicate the checks performed by the canonical
|
||||
anti-bot detection libraries against an OFFLINE browser session.
|
||||
|
||||
Each test asserts the SAME thing the upstream detector would flag. A pass
|
||||
here means our patched build appears human to that detector; a fail
|
||||
means a real stealth hole that anti-bot kits would exploit in production.
|
||||
|
||||
Detector libraries studied (all FOSS, MIT-licensed):
|
||||
- github.com/fingerprintjs/BotD — 19 detectors, the most
|
||||
widely deployed client-side
|
||||
bot detector
|
||||
- github.com/abrahamjuliot/creepjs — headless / stealth / lies
|
||||
modules
|
||||
- github.com/fingerprintjs/fingerprintjs — canvas / audio / color /
|
||||
touch consistency
|
||||
- github.com/antoinevastel/fpscanner — UA / platform / oscpu
|
||||
cross-checks
|
||||
- bot.sannysoft.com — classic Puppeteer harness
|
||||
|
||||
Everything runs against `about:blank` with NO network and NO proxy. The
|
||||
suite is intended to be part of the release-gate: pre-push hook runs
|
||||
`pytest -m e2e` and these tests must be green on every release.
|
||||
|
||||
Run only this file:
|
||||
pytest tests/test_fingerprint_surface.py -m e2e -v
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import re
|
||||
import sys
|
||||
|
||||
import pytest
|
||||
|
||||
from invisible_playwright import InvisiblePlaywright
|
||||
from invisible_playwright.constants import BINARY_ENTRY_REL
|
||||
|
||||
|
||||
# ────────────────────────────────────────────────────────────────────
|
||||
# Inline PIN — a coherent mid-range Windows desktop. Not user-config:
|
||||
# these specific values are what the surface tests assert against.
|
||||
# Keep PIN small (only fields that JS exposes) and stable across runs.
|
||||
# ────────────────────────────────────────────────────────────────────
|
||||
|
||||
PIN = {
|
||||
"screen.width": 1920,
|
||||
"screen.height": 1080,
|
||||
"screen.avail_width": 1920,
|
||||
"screen.avail_height": 1040,
|
||||
"screen.dpr": 1.0,
|
||||
"hardware.concurrency": 8,
|
||||
"audio.sample_rate": 48000,
|
||||
"audio.max_channel_count": 2,
|
||||
}
|
||||
|
||||
|
||||
@pytest.fixture(scope="session")
|
||||
def firefox_binary():
|
||||
"""Locate the patched Firefox binary. Three lookup paths:
|
||||
1. ``INVPW_BINARY_PATH`` env var (for dev iteration against a local build)
|
||||
2. Cached binary under ``cache_dir_for_version()`` (post-fetch)
|
||||
3. Skip cleanly (no implicit network download)."""
|
||||
import os
|
||||
env_path = os.environ.get("INVPW_BINARY_PATH")
|
||||
if env_path:
|
||||
from pathlib import Path
|
||||
if Path(env_path).exists():
|
||||
return env_path
|
||||
pytest.skip(f"INVPW_BINARY_PATH={env_path!r} does not exist")
|
||||
if sys.platform not in BINARY_ENTRY_REL:
|
||||
pytest.skip(f"unsupported platform: {sys.platform}")
|
||||
from invisible_playwright.download import cache_dir_for_version
|
||||
entry = cache_dir_for_version() / BINARY_ENTRY_REL[sys.platform]
|
||||
if not entry.exists():
|
||||
pytest.skip(
|
||||
"patched Firefox not cached; run "
|
||||
"`python -m invisible_playwright fetch` first, or set "
|
||||
"INVPW_BINARY_PATH to a local build"
|
||||
)
|
||||
return str(entry)
|
||||
|
||||
|
||||
@pytest.fixture(scope="module")
|
||||
def page(firefox_binary):
|
||||
"""One headless browser shared across the whole module.
|
||||
~20s startup paid once, then every test runs in ~50ms."""
|
||||
with InvisiblePlaywright(
|
||||
seed=42,
|
||||
pin=PIN,
|
||||
binary_path=firefox_binary,
|
||||
headless=True,
|
||||
) as browser:
|
||||
ctx = browser.new_context()
|
||||
p = ctx.new_page()
|
||||
p.goto("about:blank", timeout=30_000)
|
||||
yield p
|
||||
|
||||
|
||||
def _ev(page, expr):
|
||||
return page.evaluate(expr)
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# BotD detectors (github.com/fingerprintjs/BotD/tree/main/src/detectors)
|
||||
# Each detector becomes one pytest. The failure name maps to the BotKind
|
||||
# constant BotD would emit on the wire.
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_webdriver_property_is_falsey(page):
|
||||
"""BotD: navigator.webdriver === true → HeadlessChrome verdict."""
|
||||
assert not _ev(page, "navigator.webdriver"), (
|
||||
"navigator.webdriver is truthy — instant HeadlessChrome verdict"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_app_version_no_headless_token(page):
|
||||
"""BotD detectAppVersion: /headless|electron|slimerjs/i in appVersion."""
|
||||
av = _ev(page, "navigator.appVersion")
|
||||
for token in ("headless", "electron", "slimerjs"):
|
||||
assert not re.search(token, av, re.I), (
|
||||
f"navigator.appVersion contains {token!r}: {av!r}"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_user_agent_no_headless_or_selenium_token(page):
|
||||
"""BotD: /headless|selenium|phantom/i in UA."""
|
||||
ua = _ev(page, "navigator.userAgent")
|
||||
for token in ("headless", "selenium", "phantom"):
|
||||
assert not re.search(token, ua, re.I), (
|
||||
f"navigator.userAgent contains {token!r}: {ua!r}"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_function_bind_is_function(page):
|
||||
"""BotD detectFunctionBind: missing Function.prototype.bind = PhantomJS."""
|
||||
assert _ev(page, "typeof Function.prototype.bind === 'function'")
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_product_sub_is_gecko_value(page):
|
||||
"""BotD detectProductSub: Firefox must return '20100101'; '20030107'
|
||||
on a Firefox UA = Chrome-stub leaked under spoof."""
|
||||
assert _ev(page, "navigator.productSub") == "20100101", (
|
||||
"navigator.productSub must be '20100101' on Firefox"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_no_process_global(page):
|
||||
"""BotD detectProcess: window.process indicates Electron."""
|
||||
assert not _ev(page,
|
||||
"typeof window.process !== 'undefined' && "
|
||||
"window.process.type === 'renderer'"
|
||||
)
|
||||
assert not _ev(page,
|
||||
"typeof window.process !== 'undefined' && "
|
||||
"window.process.versions != null && "
|
||||
"typeof window.process.versions.electron !== 'undefined'"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_eval_length_matches_engine(page):
|
||||
"""BotD detectEvalLengthInconsistency: `eval.toString().length` must be
|
||||
37 on Gecko (33 on Chromium, 39 on IE). Mismatch = engine spoof."""
|
||||
assert _ev(page, "eval.toString().length") == 37
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_languages_array_non_empty(page):
|
||||
"""BotD detectLanguagesLengthInconsistency: empty navigator.languages
|
||||
is the classic HeadlessChrome tell."""
|
||||
assert _ev(page, "navigator.languages.length") > 0
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_plugins_instance_of_PluginArray(page):
|
||||
"""BotD detectPluginsArray: navigator.plugins must be a real PluginArray."""
|
||||
assert _ev(page, "navigator.plugins instanceof PluginArray")
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_mime_types_consistent_prototype_chain(page):
|
||||
"""BotD areMimeTypesConsistent: navigator.mimeTypes and each entry
|
||||
must have proper prototype chain. Spoofers using plain arrays fail."""
|
||||
consistent = _ev(page, """() => {
|
||||
if (typeof navigator.mimeTypes === 'undefined' ||
|
||||
typeof MimeTypeArray === 'undefined') return false;
|
||||
let ok = Object.getPrototypeOf(navigator.mimeTypes) === MimeTypeArray.prototype;
|
||||
for (let i = 0; i < navigator.mimeTypes.length; i++) {
|
||||
ok = ok && Object.getPrototypeOf(navigator.mimeTypes[i]) === MimeType.prototype;
|
||||
}
|
||||
return ok;
|
||||
}""")
|
||||
assert consistent, "navigator.mimeTypes prototype chain inconsistent"
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_no_distinctive_window_props(page):
|
||||
"""BotD checkDistinctiveProperties: scan window for automation globals."""
|
||||
DISTINCTIVE = [
|
||||
"awesomium", "RunPerfTest", "CefSharp", "fmget_targets", "geb",
|
||||
"__nightmare", "nightmare", "__phantomas", "callPhantom", "_phantom",
|
||||
"wdioElectron", "__webdriverFunc", "_WEBDRIVER_ELEM_CACHE",
|
||||
"ChromeDriverw", "domAutomation", "domAutomationController",
|
||||
]
|
||||
leaks = [n for n in DISTINCTIVE
|
||||
if _ev(page, f"typeof window[{n!r}] !== 'undefined'")]
|
||||
assert not leaks, f"Distinctive bot globals leaked: {leaks}"
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_no_distinctive_document_props(page):
|
||||
"""BotD: document-side automation globals (webdriver/selenium/cdc)."""
|
||||
DOC_LEAKS = [
|
||||
"__webdriver_evaluate", "__selenium_evaluate",
|
||||
"__webdriver_script_function", "__webdriver_script_func",
|
||||
"__webdriver_script_fn", "__fxdriver_evaluate",
|
||||
"__driver_unwrapped", "__webdriver_unwrapped",
|
||||
"__driver_evaluate", "__selenium_unwrapped",
|
||||
"__fxdriver_unwrapped",
|
||||
"$cdc_asdjflasutopfhvcZLmcf", "$cdc_asdjflasutopfhvcZLmcfl_",
|
||||
"$chrome_asyncScriptInfo", "__$webdriverAsyncExecutor",
|
||||
]
|
||||
leaks = [n for n in DOC_LEAKS
|
||||
if _ev(page, f"typeof document[{n!r}] !== 'undefined'")]
|
||||
assert not leaks, f"document carries automation property names: {leaks}"
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_document_html_attributes_clean(page):
|
||||
"""BotD detectDocumentAttributes: html element attrs contain 'selenium'
|
||||
/ 'webdriver' / 'driver' → Selenium verdict."""
|
||||
attrs = _ev(page,
|
||||
"Array.from(document.documentElement.attributes).map(a => a.name + '=' + a.value)")
|
||||
bad = [a for a in attrs if any(t in a.lower()
|
||||
for t in ("selenium", "webdriver", "driver"))]
|
||||
assert not bad, f"HTML attributes contain bot tokens: {bad}"
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_window_size_nonzero(page):
|
||||
"""BotD detectWindowSize: headless without window manager → 0x0."""
|
||||
ow = _ev(page, "window.outerWidth")
|
||||
oh = _ev(page, "window.outerHeight")
|
||||
assert ow > 0 and oh > 0, (
|
||||
f"outerWidth/Height = {ow}/{oh} — headless without window manager"
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_botd_webgl_debug_renderer_info_available(page):
|
||||
"""BotD detectWebGL: WEBGL_debug_renderer_info extension must exist."""
|
||||
has_ext = _ev(page, """() => {
|
||||
const c = document.createElement('canvas');
|
||||
const gl = c.getContext('webgl') || c.getContext('experimental-webgl');
|
||||
return !!gl && !!gl.getExtension('WEBGL_debug_renderer_info');
|
||||
}""")
|
||||
assert has_ext
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# sannysoft.com — classic Puppeteer detection harness
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_sannysoft_chrome_object_consistency(page):
|
||||
"""Firefox UA + window.chrome present = bot-framework leak."""
|
||||
if "Firefox" in _ev(page, "navigator.userAgent"):
|
||||
assert not _ev(page, "typeof window.chrome !== 'undefined'")
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_sannysoft_permissions_query_works(page):
|
||||
"""navigator.permissions.query() must return a proper PermissionStatus."""
|
||||
ok = _ev(page, """async () => {
|
||||
if (!navigator.permissions || !navigator.permissions.query) return false;
|
||||
try {
|
||||
const r = await navigator.permissions.query({name: 'notifications'});
|
||||
return r && typeof r.state === 'string';
|
||||
} catch (e) { return false; }
|
||||
}""")
|
||||
assert ok
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_sannysoft_iframe_chrome_not_leaked(page):
|
||||
"""iframe.contentWindow.chrome must not leak on Firefox UA."""
|
||||
if "Firefox" not in _ev(page, "navigator.userAgent"):
|
||||
pytest.skip("Firefox-only invariant")
|
||||
leaks = _ev(page, """() => {
|
||||
const iframe = document.createElement('iframe');
|
||||
iframe.style.display = 'none';
|
||||
document.body.appendChild(iframe);
|
||||
const is = typeof iframe.contentWindow.chrome !== 'undefined';
|
||||
document.body.removeChild(iframe);
|
||||
return is;
|
||||
}""")
|
||||
assert not leaks
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_sannysoft_iframe_languages_not_empty(page):
|
||||
"""Iframe-scope navigator.languages must have ≥1 entry."""
|
||||
n = _ev(page, """() => {
|
||||
const f = document.createElement('iframe');
|
||||
f.style.display = 'none';
|
||||
document.body.appendChild(f);
|
||||
const len = f.contentWindow.navigator.languages.length;
|
||||
document.body.removeChild(f);
|
||||
return len;
|
||||
}""")
|
||||
assert n > 0
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# FingerprintJS — fingerprint surface coherence
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_fpjs_canvas_2d_context_returns_valid(page):
|
||||
ok = _ev(page, """() => {
|
||||
const c = document.createElement('canvas');
|
||||
c.width = 100; c.height = 100;
|
||||
const ctx = c.getContext('2d');
|
||||
if (!ctx) return false;
|
||||
ctx.fillText('test', 10, 10);
|
||||
const data = c.toDataURL();
|
||||
return data.length > 100 && data.startsWith('data:image/png;base64');
|
||||
}""")
|
||||
assert ok
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_fpjs_audio_context_works(page):
|
||||
ok = _ev(page, """async () => {
|
||||
try {
|
||||
const ctx = new (window.OfflineAudioContext ||
|
||||
window.webkitOfflineAudioContext)(1, 5000, 44100);
|
||||
const osc = ctx.createOscillator();
|
||||
osc.connect(ctx.destination);
|
||||
osc.start(0);
|
||||
const buf = await ctx.startRendering();
|
||||
return buf && buf.length > 0;
|
||||
} catch (e) { return false; }
|
||||
}""")
|
||||
assert ok
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_fpjs_color_gamut_query_works(page):
|
||||
"""matchMedia('(color-gamut: ...)') must match at least srgb."""
|
||||
ok = _ev(page, """matchMedia('(color-gamut: srgb)').matches ||
|
||||
matchMedia('(color-gamut: p3)').matches ||
|
||||
matchMedia('(color-gamut: rec2020)').matches""")
|
||||
assert ok
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_fpjs_screen_color_depth_realistic(page):
|
||||
"""Atypical color depths are headless-distinctive."""
|
||||
cd = _ev(page, "screen.colorDepth")
|
||||
assert cd in (24, 30, 32)
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# PIN-locked surfaces (the values declared in PIN above)
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_pin_screen_width_lands_in_screen_object(page):
|
||||
assert _ev(page, "screen.width") == PIN["screen.width"]
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_pin_screen_height_lands_in_screen_object(page):
|
||||
assert _ev(page, "screen.height") == PIN["screen.height"]
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_pin_hardware_concurrency_lands_in_navigator(page):
|
||||
assert (_ev(page, "navigator.hardwareConcurrency")
|
||||
== PIN["hardware.concurrency"])
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_pin_audio_sample_rate_lands_in_AudioContext(page):
|
||||
assert _ev(page,
|
||||
"(new (window.AudioContext||window.webkitAudioContext)()).sampleRate"
|
||||
) == PIN["audio.sample_rate"]
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_pin_audio_max_channels_lands_in_destination(page):
|
||||
assert _ev(page,
|
||||
"(new (window.AudioContext||window.webkitAudioContext)())"
|
||||
".destination.maxChannelCount"
|
||||
) == PIN["audio.max_channel_count"]
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# fpscanner-style cross-checks
|
||||
# ===========================================================================
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_fpscanner_ua_vs_platform_consistent(page):
|
||||
"""UA OS substring must agree with navigator.platform OS substring."""
|
||||
ua = _ev(page, "navigator.userAgent")
|
||||
platform = _ev(page, "navigator.platform")
|
||||
if "Windows" in ua:
|
||||
assert "Win" in platform, f"UA Win but platform={platform!r}"
|
||||
elif "Mac" in ua:
|
||||
assert "Mac" in platform
|
||||
elif "Linux" in ua:
|
||||
assert "Linux" in platform or "X11" in platform
|
||||
|
||||
|
||||
@pytest.mark.e2e
|
||||
def test_fpscanner_no_userAgentData_on_firefox(page):
|
||||
"""navigator.userAgentData is Chromium-only. Presence on Firefox UA = bot."""
|
||||
if "Firefox" in _ev(page, "navigator.userAgent"):
|
||||
assert not _ev(page, "'userAgentData' in navigator")
|
||||
Loading…
Add table
Add a link
Reference in a new issue