- слоты браузеров: asyncio.Semaphore, размер из BROWSER_SLOTS, при нехватке — 503 + Retry-After; устранена утечка счётчика при отказе (#12, #49) - session_id принимается только как UUID, путь профиля проверяется (#5) - секреты клиента в логах маскируются, полный дамп — только на DEBUG (#6) - ответ: status всегда "ok"/"err", поле code (no_slots / request_failed / challenge_failed / action_failed), response_model (#7) - непройденный антибот → challenge_failed, действия не выполняются (#8) - таймауты провайдеров в мс вынесены в константы; detect переживает навигацию посреди проверки; пауза между неудачными solve вместо холостого цикла (#9) - обязательные поля действий и формат прокси проверяются схемой → 422 до запуска браузера (#13) - прокси необязателен (#15) - user_agent сохраняется в профиле сессии и реально передаётся в браузер (#16) - duration_ms у действий и у запроса (#18), опция screenshot (#19) - общий конфиг логирования (LOG_LEVEL), print → логгер, CloudflareProvider, удалён cached_persistent, GET /health (#22) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
141 lines
3.8 KiB
Python
141 lines
3.8 KiB
Python
import asyncio
|
|
import time
|
|
|
|
from patchright.async_api import Page
|
|
|
|
from antibot.orchestrator import pass_challenges
|
|
from captcha import load_captcha
|
|
from api.schemas import ActionRequest
|
|
|
|
_actions: dict = {}
|
|
|
|
|
|
class ChallengeFailed(Exception):
|
|
pass
|
|
|
|
|
|
def register(fun) -> None:
|
|
_actions[fun.__name__] = fun
|
|
return fun
|
|
|
|
|
|
async def _query(page: Page, selector: str):
|
|
element = await page.query_selector(selector)
|
|
if element is None:
|
|
raise LookupError(f"selector not found: {selector}")
|
|
return element
|
|
|
|
|
|
@register
|
|
async def fill(page: Page, act: ActionRequest, **kwargs) -> str | None:
|
|
await page.fill(act.selector, act.value)
|
|
|
|
|
|
@register
|
|
async def captcha(page: Page, act: ActionRequest, captcha_key: str, **kwargs) -> str | None:
|
|
captcha_solver = load_captcha(captcha_key, act.value)
|
|
await captcha_solver.solve(act.selector, page)
|
|
|
|
|
|
@register
|
|
async def delay(page: Page, act: ActionRequest, **kwargs) -> str | None:
|
|
await asyncio.sleep(int(act.value))
|
|
|
|
|
|
@register
|
|
async def mouse_move(page: Page, act: ActionRequest, **kwargs) -> str | None:
|
|
element = await _query(page, act.selector)
|
|
await element.hover(scroll="auto")
|
|
|
|
|
|
@register
|
|
async def click(page: Page, act: ActionRequest, **kwargs) -> str | None:
|
|
element = page.locator(act.selector)
|
|
await element.click()
|
|
|
|
|
|
@register
|
|
async def submit(page: Page, act: ActionRequest, **kwargs) -> str | None:
|
|
element = page.locator(act.selector)
|
|
|
|
async with page.expect_event(
|
|
"framenavigated",
|
|
predicate=lambda frame: frame == page.main_frame,
|
|
):
|
|
await element.click()
|
|
|
|
|
|
@register
|
|
async def inner_html(page: Page, act: ActionRequest, **kwargs) -> str | None:
|
|
element = await _query(page, act.selector)
|
|
return await element.inner_html()
|
|
|
|
|
|
@register
|
|
async def wait_for(page: Page, act: ActionRequest, **kwargs) -> str | None:
|
|
await page.wait_for_selector(act.selector, timeout=act.timeout * 1000)
|
|
|
|
|
|
@register
|
|
async def pass_challenge(
|
|
page: Page, act: ActionRequest, *,
|
|
antibot_name, success_locator=None, captcha_key=None, detect_locator=None,
|
|
**kwargs
|
|
) -> str | None:
|
|
result = await pass_challenges(
|
|
antibot_name, page,
|
|
captcha=captcha_key,
|
|
success_locator=act.selector or success_locator,
|
|
detect_locator=detect_locator
|
|
)
|
|
if not result.cleared:
|
|
raise Exception(f"challenge not passed: {result.reason}")
|
|
return result.reason
|
|
|
|
|
|
async def execute(
|
|
page: Page, actions: list[ActionRequest],
|
|
captcha_key: str | None = None, antibot_name: str | None = None,
|
|
success_locator: str | None = None,
|
|
detect_locator: str | None = None) -> list[dict]:
|
|
"""Выполнение очереди автоматизации.
|
|
|
|
Если антибот не пройден — бросает ChallengeFailed, действия не выполняются: на странице-заглушке
|
|
они бы упали с невнятной ошибкой селектора."""
|
|
actions_result = []
|
|
ctx = {
|
|
"captcha_key": captcha_key,
|
|
"antibot_name": antibot_name,
|
|
"success_locator": success_locator,
|
|
"detect_locator": detect_locator,
|
|
}
|
|
outcome = await pass_challenges(
|
|
antibot_name, page,
|
|
captcha=captcha_key,
|
|
success_locator=success_locator,
|
|
detect_locator=detect_locator,
|
|
)
|
|
if not outcome.cleared:
|
|
stage = outcome.last_stage.value if outcome.last_stage else None
|
|
raise ChallengeFailed(f"{antibot_name} not passed: {outcome.reason} (stage: {stage})")
|
|
|
|
for act in actions:
|
|
new_line = {
|
|
"type": act.type,
|
|
}
|
|
started = time.monotonic()
|
|
try:
|
|
result = await _actions[act.type](page, act, **ctx)
|
|
except Exception as e:
|
|
new_line["ok"] = False
|
|
new_line["result"] = str(e) or type(e).__name__
|
|
else:
|
|
new_line["ok"] = True
|
|
if result:
|
|
new_line["result"] = result
|
|
new_line["duration_ms"] = int((time.monotonic() - started) * 1000)
|
|
actions_result.append(new_line)
|
|
if not new_line["ok"]:
|
|
break
|
|
return actions_result
|