feature!: добавлен механизм обхода антибот защиты
This commit is contained in:
1 parent
f5871710ab
commit
1d2fdeae70
17 files changed
+735
-16
No files matched your search
Whitespace-only changes.
@@ -0,0 +1,81 @@
|
||||
"""
|
||||
абстракции: ChallengeInfo, Detector, Solver
|
||||
|
||||
Пользователь передает нам:
|
||||
|
||||
- селектор со страницы (достаточно общий чтобы покрыть весь сайт, но не страницу антибот защиты)
|
||||
- ключ сервиса распознавания капчи
|
||||
"""
|
||||
from copy import deepcopy
|
||||
from dataclasses import dataclass
|
||||
from enum import Enum
|
||||
from typing import Type
|
||||
|
||||
from playwright.async_api import Page
|
||||
|
||||
|
||||
class Stage(str, Enum):
|
||||
INTERSTITIAL = "interstitial"
|
||||
CHECKBOX = "checkbox"
|
||||
INTERACTIVE = "interactive"
|
||||
|
||||
|
||||
@dataclass
|
||||
class ChallengeInfo:
|
||||
stage: Stage
|
||||
provider: str # "cloudflare_turnstile", "datadome", "hcaptcha", "recaptcha_v2", ...
|
||||
success: bool = False
|
||||
success_locator: str | None = None # локатор на странице за защитой
|
||||
site_key: str | None = None # для стадии C, чтобы передать в 2captcha
|
||||
meta: dict | None = None # что-то специфичное для провайдера
|
||||
|
||||
|
||||
class Solver:
|
||||
"""Алгоритм прохождения. Один солвер = один антибот провайдер"""
|
||||
|
||||
def __init__(self, **kwargs):
|
||||
self._opt = kwargs
|
||||
|
||||
async def solve(self, page: Page, info: ChallengeInfo):
|
||||
"""Возвращает True, если стадия пройдена (или больше не требуется действий)."""
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class Detector:
|
||||
"""Один детектор = одна сигнатура одного провайдера."""
|
||||
|
||||
async def detect(self, page: Page, info: ChallengeInfo):
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class AntiBotProvider:
|
||||
provider: str
|
||||
detector: Detector
|
||||
solver: Solver
|
||||
_solver_cls: Type[Solver]
|
||||
_detector_cls: Type[Detector]
|
||||
_success_locator: str | None = None
|
||||
|
||||
def __init__(self):
|
||||
self.detector = self._detector_cls()
|
||||
|
||||
async def detect(self, page: Page) -> ChallengeInfo:
|
||||
info = ChallengeInfo(
|
||||
stage = Stage.INTERSTITIAL,
|
||||
provider = self.provider,
|
||||
success_locator = self._success_locator,
|
||||
)
|
||||
# обновление info
|
||||
await self.detector.detect(page, info)
|
||||
return info
|
||||
|
||||
async def solve(self, page: Page, info: ChallengeInfo):
|
||||
await self.solver.solve(page, info)
|
||||
|
||||
def configure(self, success_locator=None, **kwargs):
|
||||
self.solver = self._solver_cls(**kwargs)
|
||||
self._success_locator = success_locator
|
||||
|
||||
def clear(self):
|
||||
del self.solver
|
||||
del self.detector
|
||||
@@ -0,0 +1,6 @@
|
||||
class ProviderNotDetected(Exception):
|
||||
pass
|
||||
|
||||
|
||||
class IframeLoadError(Exception):
|
||||
pass
|
||||
@@ -0,0 +1,62 @@
|
||||
"""
|
||||
человекоподобные движения мыши (безье+ускорение), тайминги кликов
|
||||
|
||||
Camoufox — это форк Firefox, заточенный под антидетект-автоматизацию (спуфинг fingerprint'а, WebGL, canvas,
|
||||
шрифтов и т.д.), с Python-обёрткой поверх Playwright-протокола (Juggler).
|
||||
|
||||
У него есть параметр humanize при запуске:
|
||||
|
||||
```
|
||||
from camoufox.sync_api import Camoufox
|
||||
|
||||
with Camoufox(humanize=True) as browser:
|
||||
...
|
||||
```
|
||||
|
||||
Или с указанием максимального времени на движение:
|
||||
|
||||
```
|
||||
Camoufox(humanize=1.5) # курсор доберётся до цели не более чем за 1.5 сек
|
||||
```
|
||||
|
||||
Когда humanize включён, при вызовах вроде page.click() / locator.click() библиотека:
|
||||
|
||||
- строит траекторию курсора от текущей позиции до цели не по прямой, а по кривой (используется подход,
|
||||
похожий на ghost-cursor — Безье-путь с шумом),
|
||||
- разбивает движение на промежуточные точки с нелинейной скоростью (ускорение/замедление к цели, как у реальной руки),
|
||||
- добавляет небольшую случайную задержку перед самим кликом.
|
||||
|
||||
Это реализовано на уровне их собственного форка драйвера, а не через голый Playwright API — обычный Playwright (
|
||||
page.click()) телепортирует курсор мгновенно и такого не делает.
|
||||
"""
|
||||
import random
|
||||
import string
|
||||
|
||||
from playwright.async_api import Page
|
||||
|
||||
|
||||
async def typing(page: Page, locator: str, text: str):
|
||||
element = page.locator(locator)
|
||||
|
||||
await element.scroll_into_view_if_needed()
|
||||
assert await element.is_visible()
|
||||
assert await element.is_enabled()
|
||||
assert await element.is_editable()
|
||||
await element.focus()
|
||||
await element.click()
|
||||
|
||||
for char in text:
|
||||
await page.keyboard.press(char)
|
||||
|
||||
if char in string.ascii_lowercase:
|
||||
delays = (50, 100)
|
||||
elif char in string.ascii_uppercase:
|
||||
delays = (100, 150)
|
||||
elif char in string.digits:
|
||||
delays = (100, 150)
|
||||
elif char in string.punctuation:
|
||||
delays = (100, 200)
|
||||
else:
|
||||
raise SystemError("character not supported: {}".format(char))
|
||||
|
||||
await page.wait_for_timeout(random.uniform(*delays))
|
||||
Whitespace-only changes.
@@ -0,0 +1,20 @@
|
||||
from playwright.async_api import Page
|
||||
|
||||
from ..base import Detector, ChallengeInfo
|
||||
from ..base import Stage
|
||||
from ..registry import register_detector
|
||||
|
||||
|
||||
@register_detector
|
||||
class CloudflareDetector(Detector):
|
||||
provider = 'cloudflare_turnstile'
|
||||
stage = Stage.INTERSTITIAL
|
||||
|
||||
async def detect(self, page: Page) -> ChallengeInfo:
|
||||
result = ChallengeInfo(
|
||||
stage=self.stage,
|
||||
provider=self.provider,
|
||||
)
|
||||
# TODO: добавить функцию определения что страница с проверкой показана
|
||||
raise NotImplementedError
|
||||
return result
|
||||
@@ -0,0 +1,81 @@
|
||||
import logging
|
||||
|
||||
from playwright.async_api import Page
|
||||
|
||||
from ..base import Detector, ChallengeInfo, Solver, AntiBotProvider
|
||||
from ..base import Stage
|
||||
from ..registry import register
|
||||
from ..exceptions import IframeLoadError
|
||||
|
||||
log = logging.getLogger(__name__)
|
||||
log.setLevel(logging.DEBUG)
|
||||
log.addHandler(logging.StreamHandler())
|
||||
|
||||
|
||||
class ReCaptchaSolver(Solver):
|
||||
|
||||
def __init__(self, captcha_solver) -> None:
|
||||
super().__init__(captcha_solver=captcha_solver)
|
||||
self.captcha_solver = self._opt["captcha_solver"]
|
||||
|
||||
async def solve(self, page: Page, info: ChallengeInfo):
|
||||
if info.stage == Stage.INTERSTITIAL:
|
||||
await page.wait_for_timeout(15)
|
||||
elif info.stage == Stage.CHECKBOX:
|
||||
checkbox_selector = "#recaptcha-anchor"
|
||||
element = page.locator(checkbox_selector)
|
||||
async with page.expect_response("**/recaptcha/enterprise/reload**", timeout=30_000):
|
||||
await element.click()
|
||||
else:
|
||||
element = page.locator("body > div:nth-child(2) > div:nth-child(1)")
|
||||
if await element.is_visible():
|
||||
await element.click()
|
||||
|
||||
self.captcha_solver.recaptcha(page)
|
||||
# TODO: execute js
|
||||
|
||||
class ReCaptchaDetector(Detector):
|
||||
|
||||
async def detect(self, page: Page, info: ChallengeInfo):
|
||||
info.meta = {}
|
||||
checkbox_selector = "#recaptcha-anchor"
|
||||
interactive_marker_selector = "body > div:nth-child(2) > div:nth-child(4) > iframe:nth-child(1)"
|
||||
|
||||
# попытка определить что мы вообще на странице блокировки
|
||||
marker_selector = "#captcha-form"
|
||||
if not await page.query_selector(marker_selector):
|
||||
log.debug("Page is not a challenge form")
|
||||
if info.success_locator:
|
||||
info.success = bool(await page.query_selector(info.success_locator))
|
||||
info.meta["reason"] = [
|
||||
"Success locator not present and marker too.",
|
||||
"Marker not present but success locator is it!"
|
||||
][info.success]
|
||||
else:
|
||||
info.success = True
|
||||
# noinspection unresolved-references
|
||||
info.meta["reason"] = "Page is not a challenge form"
|
||||
return
|
||||
|
||||
# ожидаем когда будет загружен чекбокс
|
||||
if not await page.wait_for_selector(checkbox_selector, timeout=30_000):
|
||||
log.debug("Checkbox not present")
|
||||
# noinspection unresolved-references
|
||||
info.meta["reason"] = "Checkbox not present"
|
||||
raise IframeLoadError("Checkbox not present")
|
||||
else:
|
||||
# нужно сделать клик
|
||||
info.stage = Stage.CHECKBOX
|
||||
|
||||
if await page.wait_for_selector(interactive_marker_selector, timeout=15_000):
|
||||
# нужно решение
|
||||
info.stage = Stage.INTERACTIVE
|
||||
|
||||
return
|
||||
|
||||
|
||||
@register
|
||||
class ReCaptchaProvider(AntiBotProvider):
|
||||
provider = 'recaptcha'
|
||||
_solver_cls = ReCaptchaSolver
|
||||
_detector_cls = ReCaptchaDetector
|
||||
@@ -0,0 +1,57 @@
|
||||
"""
|
||||
цикл: detect -> solve -> verify -> retry
|
||||
"""
|
||||
import logging
|
||||
import time
|
||||
from dataclasses import dataclass
|
||||
|
||||
from captcha import load_captcha
|
||||
|
||||
from .base import Stage
|
||||
from .registry import select_one
|
||||
from .exceptions import ProviderNotDetected
|
||||
|
||||
log = logging.getLogger(__name__)
|
||||
log.setLevel(logging.DEBUG)
|
||||
log.addHandler(logging.StreamHandler())
|
||||
|
||||
|
||||
@dataclass
|
||||
class ChallengeOutcome:
|
||||
cleared: bool
|
||||
last_stage: Stage | None = None # на чём застряли, если cleared=False
|
||||
last_provider: str | None = None
|
||||
reason: str | None = None # "no captcha_key" / "timeout" / ...
|
||||
|
||||
|
||||
async def pass_challenges(
|
||||
antibot_name, page, *, captcha=None, success_locator: str | None = None, timeout_ms=120_000) -> ChallengeOutcome:
|
||||
# установка дедлайна
|
||||
deadline = time.monotonic() + timeout_ms / 1000
|
||||
|
||||
# 1. определить кто блокирует
|
||||
try:
|
||||
if captcha:
|
||||
captcha_solver = load_captcha(captcha, antibot_name)
|
||||
else:
|
||||
captcha_solver = None
|
||||
provider = await select_one(
|
||||
page, antibot_name, captcha_solver=captcha_solver, success_locator=success_locator)
|
||||
except ProviderNotDetected:
|
||||
return ChallengeOutcome(False, None, None, "Provider not detected")
|
||||
|
||||
info = None
|
||||
while time.monotonic() < deadline:
|
||||
info = await provider.detect(page) # noqa
|
||||
# 3. проверяем что решение успешно
|
||||
if info.success: # noqa
|
||||
return ChallengeOutcome(True, reason=(info.meta or {}).get("reason"))
|
||||
# 2. передаем страницу в класс и ждем решения
|
||||
await provider.solve(page, info) # noqa
|
||||
else:
|
||||
return ChallengeOutcome(
|
||||
False,
|
||||
info.stage if info else None,
|
||||
info.provider if info else None,
|
||||
"timeout"
|
||||
)
|
||||
@@ -0,0 +1,32 @@
|
||||
"""
|
||||
реестр детекторов, порядок проверки
|
||||
"""
|
||||
from playwright.async_api import Page
|
||||
|
||||
from .base import AntiBotProvider
|
||||
from .exceptions import ProviderNotDetected
|
||||
|
||||
_detectors: list[AntiBotProvider] = []
|
||||
|
||||
|
||||
def register(provider_cls) -> AntiBotProvider:
|
||||
_detectors.append(provider_cls())
|
||||
return provider_cls
|
||||
|
||||
|
||||
async def select_any(page: Page, **kwargs) -> AntiBotProvider:
|
||||
for provider in _detectors:
|
||||
if await provider.detect(page):
|
||||
provider.configure(**kwargs)
|
||||
return provider
|
||||
else:
|
||||
raise ProviderNotDetected()
|
||||
|
||||
|
||||
async def select_one(page: Page, provider_name: str, **kwargs) -> AntiBotProvider:
|
||||
for provider in _detectors:
|
||||
if provider.provider == provider_name:
|
||||
provider.configure(**kwargs)
|
||||
return provider
|
||||
else:
|
||||
raise ProviderNotDetected()
|
||||
Whitespace-only changes.
@@ -0,0 +1,20 @@
|
||||
"""
|
||||
стадия B: универсальный click-solver + сигнатуры провайдеров
|
||||
"""
|
||||
|
||||
from ..registry import detect_any
|
||||
from ..base import Stage
|
||||
|
||||
|
||||
async def solve_checkbox(page, info, max_attempts=3):
|
||||
for attempt in range(max_attempts):
|
||||
frame = page.locator(page, info.frame_locator)
|
||||
|
||||
|
||||
|
||||
# после клика — либо чисто, либо эскалация в картинки (стадия C)
|
||||
if await detect_any(page, Stage.CHECKBOX) is None:
|
||||
return True
|
||||
if await detect_any(page, Stage.INTERACTIVE):
|
||||
return False # передаём выше, пусть оркестратор переключит стадию
|
||||
return False
|
||||
@@ -0,0 +1,4 @@
|
||||
# стадия C: мост к src/captcha (hcaptcha/recaptcha/...)
|
||||
async def solve_interactive(page, info, captcha):
|
||||
await captcha.solve(info.frame_locator or info.meta["container_selector"], page)
|
||||
return True
|
||||
@@ -0,0 +1,15 @@
|
||||
"""
|
||||
стадия A: универсальный wait-solver + сигнатуры провайдеров
|
||||
"""
|
||||
from ..base import Stage
|
||||
from ..registry import detect_any
|
||||
|
||||
|
||||
async def solve_interstitial(page, info, timeout_ms=15000, poll_ms=500):
|
||||
elapsed = 0
|
||||
while elapsed < timeout_ms:
|
||||
if await detect_any(page, Stage.INTERSTITIAL) is None:
|
||||
return True
|
||||
await page.wait_for_timeout(poll_ms)
|
||||
elapsed += poll_ms
|
||||
return False
|
||||
@@ -0,0 +1,60 @@
|
||||
"""
|
||||
|
||||
dl.formRow:nth-child(3) > dd:nth-child(2) > div:nth-child(1)
|
||||
<div data-xf-init="h-captcha" data-sitekey="83433494-1c9f-446b-85c3-e27b8b7375e1" data-invisible="">
|
||||
|
||||
iframe[data-hcaptcha-widget-id="0dm47zf0gn8t"]
|
||||
|
||||
textarea[id="g-recaptcha-response-0dm47zf0gn8t"]
|
||||
textarea[id="h-captcha-response-0dm47zf0gn8t"]
|
||||
|
||||
|
||||
"""
|
||||
import logging
|
||||
|
||||
from .base import Captcha, Page
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
logger.setLevel(logging.DEBUG)
|
||||
logger.addHandler(logging.StreamHandler())
|
||||
|
||||
|
||||
class SolverReCaptcha(Captcha):
|
||||
ctype = "recaptcha"
|
||||
|
||||
async def solve(self, selector, page: Page):
|
||||
div_container = page.locator(selector)
|
||||
await div_container.wait_for()
|
||||
|
||||
if not div_container:
|
||||
return None
|
||||
|
||||
site_key = await div_container.get_attribute("data-sitekey")
|
||||
is_invisible = bool(await div_container.get_attribute("data-invisible"))
|
||||
|
||||
iframe = div_container.locator("iframe:nth-child(1)")
|
||||
await iframe.wait_for()
|
||||
textarea_id_postfix = await iframe.get_attribute("data-hcaptcha-widget-id")
|
||||
|
||||
textarea_g = await page.query_selector(f"#g-recaptcha-response-{textarea_id_postfix}")
|
||||
textarea_h = await page.query_selector(f"#h-captcha-response-{textarea_id_postfix}")
|
||||
|
||||
code = await self.recaptcha(page.url, site_key, is_invisible)
|
||||
await textarea_g.evaluate((
|
||||
"(el, value) => {"
|
||||
"el.value = value;"
|
||||
"el.dispatchEvent(new Event('input', {bubbles: true}));"
|
||||
"}"
|
||||
), code)
|
||||
await textarea_h.evaluate((
|
||||
"(el, value) => {"
|
||||
"el.value = value;"
|
||||
"el.dispatchEvent(new Event('input', {bubbles: true}));"
|
||||
"}"
|
||||
), code)
|
||||
return None
|
||||
|
||||
async def recaptcha(self, url, site_key, is_invisible):
|
||||
resp = self._solver.recaptcha(site_key, url, invisible=is_invisible)
|
||||
logger.debug(f"recaptcha response: {resp}")
|
||||
return resp["code"]
|
||||
@@ -1,3 +1,5 @@
|
||||
from typing import Any
|
||||
|
||||
from pydantic import BaseModel
|
||||
|
||||
|
||||
@@ -14,8 +16,6 @@ class ActionRequest(BaseModel):
|
||||
|
||||
class SolveRequest(BaseModel):
|
||||
"""
|
||||
# TODO: добавить locale и time_zone
|
||||
|
||||
Объект запроса для автоматизации на сайте.
|
||||
"""
|
||||
url: str
|
||||
@@ -23,11 +23,13 @@ class SolveRequest(BaseModel):
|
||||
timeout: int | None = None
|
||||
screen: str = "1280x720"
|
||||
user_agent: str | None = None
|
||||
cookies: list[dict[str, Any]] | None = None
|
||||
# List[{name: str, value: str, domain: str, path: str, expires: float, httpOnly: bool, secure: bool, sameSite: Union["Lax", "None", "Strict"], partitionKey: Union[str, None]}]
|
||||
|
||||
actions: list[ActionRequest]
|
||||
|
||||
html: bool = False
|
||||
|
||||
captcha: str | None = None
|
||||
captcha_selector: str | None = None
|
||||
captcha_success_selector: str | None = None
|
||||
captcha_type: str | None = None
|
||||
+34
-13
@@ -1,5 +1,6 @@
|
||||
import json
|
||||
import logging
|
||||
import traceback
|
||||
import uuid
|
||||
from pathlib import Path
|
||||
from typing import Annotated
|
||||
@@ -16,6 +17,7 @@ from playwright.async_api import ProxySettings
|
||||
from ua_parser import user_agent_parser
|
||||
|
||||
import captcha
|
||||
from antibot.orchestrator import pass_challenges
|
||||
from engine import actions
|
||||
from engine.schemas import SolveRequest
|
||||
|
||||
@@ -109,16 +111,16 @@ async def solve(
|
||||
config = load_or_create_identity(user_data_dir, user_agent=payload.user_agent)
|
||||
ctx_opts = {**config["context_options"], "config": config["config"]}
|
||||
|
||||
if payload.captcha:
|
||||
# "dl.formRow:nth-child(3) > dd:nth-child(2) > div:nth-child(1)"
|
||||
# hcaptcha
|
||||
# rucaptcha api_key
|
||||
captcha_solver = captcha.load_captcha(
|
||||
payload.captcha,
|
||||
payload.captcha_type
|
||||
)
|
||||
else:
|
||||
captcha_solver = None
|
||||
# if payload.captcha:
|
||||
# # "dl.formRow:nth-child(3) > dd:nth-child(2) > div:nth-child(1)"
|
||||
# # hcaptcha
|
||||
# # rucaptcha api_key
|
||||
# captcha_solver = captcha.load_captcha(
|
||||
# payload.captcha,
|
||||
# payload.captcha_type
|
||||
# )
|
||||
# else:
|
||||
# captcha_solver = None
|
||||
|
||||
async with AsyncCamoufox(
|
||||
persistent_context=True,
|
||||
@@ -127,16 +129,35 @@ async def solve(
|
||||
window=scr_size,
|
||||
i_know_what_im_doing=True,
|
||||
geoip=True,
|
||||
humanize=True,
|
||||
**ctx_opts,
|
||||
) as camoufo:
|
||||
await camoufo.add_init_script(config["init_script"])
|
||||
page = camoufo.pages[0] if camoufo.pages else await camoufo.new_page()
|
||||
if payload.cookies:
|
||||
await camoufo.add_cookies(payload.cookies)
|
||||
page.set_default_timeout(30_000)
|
||||
page.set_default_navigation_timeout(60_000)
|
||||
|
||||
await page.goto(url=payload.url)
|
||||
# if captcha_solver:
|
||||
# assert payload.captcha_selector
|
||||
# await captcha_solver.solve(payload.captcha_selector, page)
|
||||
|
||||
try:
|
||||
await pass_challenges(
|
||||
payload.captcha_type, page,
|
||||
captcha=payload.captcha, success_locator=payload.captcha_success_selector
|
||||
)
|
||||
except Exception:
|
||||
result = {
|
||||
"status": "err",
|
||||
"user_agent": ctx_opts["user_agent"],
|
||||
"cookies": None,
|
||||
"actions": [],
|
||||
"error": traceback.format_exc(),
|
||||
}
|
||||
return result
|
||||
|
||||
actions_result = await actions.execute(page, payload.actions, payload.captcha)
|
||||
# test: is captcha present?
|
||||
status = not "error" in actions_result[-1]
|
||||
error = ["Something error", None][status]
|
||||
cookies = await camoufo.cookies()
|
||||
|
||||
Reference in new issue
Block a user