"""TikTok page signal detection.

Single entry point: PageSignals.from_page(...)
"""

from __future__ import annotations

import re
from typing import ClassVar

from pydantic import BaseModel


class PageSignals(BaseModel):
    """Normalised browser observations the LLM can act on."""

    block_signal: str = (
        "none"  # none | login_redirect | captcha | soft_block | empty_response
    )
    page_load_quality: str = "full"  # full | partial | empty
    redirect_detected: bool = False
    slow_response: bool = False
    response_time_ms: int = 0
    items_load_ratio: float | None = None
    zero_metrics: bool = False

    # ------------------------------------------------------------------ constants
    _SLOW_MS: ClassVar[int] = 15_000
    _EMPTY_THRESHOLD: ClassVar[int] = 3_000
    _LOGIN_URL_TOKENS: ClassVar[tuple] = ("/login", "loginpage", "login?")
    _CAPTCHA_TOKENS: ClassVar[tuple] = (
        "captcha",
        "verifyaccount",
        "verify-account",
        "robot-check",
    )

    # ------------------------------------------------------------------ factory
    @classmethod
    def from_page(
        cls,
        *,
        html: str,
        requested_url: str,
        final_url: str,
        response_time_ms: int,
        items_loaded: int = 0,
        count_text: str | None = None,
        metrics: dict | None = None,
    ) -> PageSignals:
        return cls(
            block_signal=cls._block_signal(html, final_url),
            page_load_quality=cls._load_quality(items_loaded),
            redirect_detected=cls._is_redirect(requested_url, final_url),
            slow_response=response_time_ms > cls._SLOW_MS,
            response_time_ms=response_time_ms,
            items_load_ratio=cls._load_ratio(items_loaded, count_text),
            zero_metrics=cls._zero_metrics(metrics or {}),
        )

    # ------------------------------------------------------------------ private helpers
    @classmethod
    def _block_signal(cls, html: str, final_url: str) -> str:
        lower_url = final_url.lower()
        if any(t in lower_url for t in cls._LOGIN_URL_TOKENS):
            return "login_redirect"
        lower_html = html.lower()
        if any(t in lower_html for t in cls._CAPTCHA_TOKENS):
            return "captcha"
        if len(html) < cls._EMPTY_THRESHOLD:
            return "empty_response"
        if "tiktok.com" not in lower_url:
            return "soft_block"
        return "none"

    @classmethod
    def _load_quality(cls, items_loaded: int, threshold: int = 5) -> str:
        if items_loaded == 0:
            return "empty"
        if items_loaded < threshold:
            return "partial"
        return "full"

    @classmethod
    def _is_redirect(cls, requested: str, final: str) -> bool:
        def norm(u: str) -> str:
            return u.rstrip("/").split("?")[0].lower()

        return norm(requested) != norm(final)

    @classmethod
    def _load_ratio(cls, loaded: int, count_text: str | None) -> float | None:
        total = cls._parse_count(count_text)
        if not total:
            return None
        return round(loaded / total, 4)

    @staticmethod
    def _parse_count(text: str | None) -> int | None:
        if not text:
            return None
        m = re.search(r"(\d[\d,.]*)\s*([KkMmBb]?)", text)
        if not m:
            return None
        num = float(m.group(1).replace(",", ""))
        suffix = m.group(2).upper()
        return int(
            num * {"K": 1_000, "M": 1_000_000, "B": 1_000_000_000}.get(suffix, 1)
        )

    @staticmethod
    def _zero_metrics(metrics: dict) -> bool:
        return all(not metrics.get(k) for k in ("views", "likes", "comments"))