utils.py

import asyncio
import random
import logging

logger = logging.getLogger(__name__)

CAPTCHA_SIGNALS = [
    "captcha",
    "капча",
    "robot",
    "i am not a robot",
    "я не робот",
    "blocked",
    "access denied",
    "403 forbidden",
    "cloudflare",
    "ddos-guard",
    "please verify",
    "проверьте, что вы",
]


def random_delay(min_sec: float, max_sec: float):
    import time
    time.sleep(random.uniform(min_sec, max_sec))


async def async_random_delay(min_sec: float, max_sec: float):
    await asyncio.sleep(random.uniform(min_sec, max_sec))


def is_captcha_page(content: str) -> bool:
    lower = content.lower()
    return any(signal in lower for signal in CAPTCHA_SIGNALS)


async def retry_async(
    coro_fn,
    max_attempts: int = 3,
    backoff_factor: float = 2.0,
    label: str = "",
):
    """
    Call coro_fn() up to max_attempts times, with exponential backoff.
    Raises the last exception if all attempts fail.
    """
    last_exc = None
    for attempt in range(1, max_attempts + 1):
        try:
            return await coro_fn()
        except Exception as exc:
            last_exc = exc
            wait = backoff_factor ** (attempt - 1) * 2
            logger.warning(
                f"[retry {attempt}/{max_attempts}] {label}: "
                f"{type(exc).__name__}: {exc} — ждём {wait:.1f}с"
            )
            await asyncio.sleep(wait)
    raise last_exc


def find_pain_points(texts: list[str], keywords: list[str]) -> tuple[bool, str]:
    """
    Scan review texts for pain-point keywords.

    Returns (found: bool, quote: str) where quote is the best matching
    sentence (up to 300 chars) prefixed with the triggering keyword.

    Matching is case-insensitive and checks whole-word presence to reduce
    false positives (e.g. "записаться" should not fire on "записи книги").
    """
    import re
    kw_lower = [k.lower() for k in keywords]

    best_quote = ""
    for text in texts:
        text_lower = text.lower()
        for kw in kw_lower:
            if kw in text_lower:
                # split into sentences on ., !, ?, \n
                sentences = re.split(r"[.!?\n]+", text)
                for s in sentences:
                    if kw in s.lower():
                        quote = s.strip()
                        if len(quote) > len(best_quote):
                            best_quote = quote[:300]
                if best_quote:
                    return True, f'[{kw}] "{best_quote}"'

    return False, ""