# ==============================================================================
# Author: Simon-Pierre Boucher <contact@spboucher.ai>
# File:   src/main.py (ka-onlyfans)
# Desc:   Profils OnlyFans publics (page de profil hors connexion) — extraction
#         best-effort des compteurs embarqués : posts, photos, vidéos, J'aime
#         (favoritedCount), prix d'abonnement, badge, bio, avatar/bannière.
#         EXPÉRIMENTAL : OnlyFans change souvent son rendu ; found=False n'est
#         pas une erreur du pipeline.
# ==============================================================================
from __future__ import annotations

import asyncio
import html as htmllib
import json
import re

from apify import Actor

from .net import Fetcher

PROFILE_URL = "https://onlyfans.com/{u}"

_NUM_RES = {
    "posts_count": re.compile(r'"postsCount"\s*:\s*(\d+)'),
    "photos_count": re.compile(r'"photosCount"\s*:\s*(\d+)'),
    "videos_count": re.compile(r'"videosCount"\s*:\s*(\d+)'),
    "likes": re.compile(r'"favoritedCount"\s*:\s*(\d+)'),
    "streams_count": re.compile(r'"finishedStreamsCount"\s*:\s*(\d+)'),
    "audios_count": re.compile(r'"audiosCount"\s*:\s*(\d+)'),
    "medias_count": re.compile(r'"mediasCount"\s*:\s*(\d+)'),
}
_JOIN_RE = re.compile(r'"joinDate"\s*:\s*"((?:[^"\\]|\\.)*)"')
_PERFORMER_RE = re.compile(r'"isPerformer"\s*:\s*(true|false)')
_PRICE_RE = re.compile(r'"subscribePrice"\s*:\s*([\d.]+)')
_STR_RES = {
    "full_name": re.compile(r'"name"\s*:\s*"((?:[^"\\]|\\.)*)"'),
    "biography": re.compile(r'"rawAbout"\s*:\s*"((?:[^"\\]|\\.)*)"'),
    "location": re.compile(r'"location"\s*:\s*"((?:[^"\\]|\\.)*)"'),
    "website": re.compile(r'"website"\s*:\s*"((?:[^"\\]|\\.)*)"'),
    "avatar": re.compile(r'"avatar"\s*:\s*"((?:[^"\\]|\\.)*)"'),
    "banner": re.compile(r'"header"\s*:\s*"((?:[^"\\]|\\.)*)"'),
}
_VERIFIED_RE = re.compile(r'"isVerified"\s*:\s*(true|false)')
_OG_TITLE_RE = re.compile(r'<meta property="og:title" content="([^"]*)"')


def _dec(raw: str) -> str:
    try:
        return htmllib.unescape(json.loads(f'"{raw}"'))
    except Exception:
        return raw


async def main() -> None:
    async with Actor:
        inp = await Actor.get_input() or {}
        usernames = [u.strip().lstrip("@").lower()
                     for u in (inp.get("usernames") or []) if u and u.strip()]
        proxy = await Actor.create_proxy_configuration(
            actor_proxy_input=inp.get("proxyConfiguration"))
        fetcher = Fetcher(proxy, delay=float(inp.get("requestDelay") or 2))
        sem = asyncio.Semaphore(int(inp.get("concurrency") or 2))

        async def one(u: str) -> None:
            async with sem:
                miss = {"kind": "profile", "platform": "onlyfans",
                        "username": u, "found": False}
                resp = None
                for attempt in range(3):  # mur Cloudflare → nouvelle IP
                    try:
                        resp = await fetcher.get(PROFILE_URL.format(u=u),
                                                 session_id=f"{u}o{attempt}")
                    except Exception as exc:
                        await Actor.push_data({**miss,
                                               "error": str(exc)[:200]})
                        return
                    if resp.status_code == 404:
                        await Actor.push_data({**miss, "error": "http_404"})
                        return
                    if resp.status_code == 200 and \
                            "postsCount" in (resp.text or ""):
                        break
                text = resp.text or ""
                if resp.status_code != 200 or "postsCount" not in text:
                    await Actor.push_data(
                        {**miss, "error": f"wall_{resp.status_code}"})
                    return
                rec: dict = {"kind": "profile", "platform": "onlyfans",
                             "found": True, "username": u}
                for key, rx in _NUM_RES.items():
                    m = rx.search(text)
                    rec[key] = int(m.group(1)) if m else None
                for key, rx in _STR_RES.items():
                    m = rx.search(text)
                    rec[key] = _dec(m.group(1)) if m else None
                if rec.get("biography"):
                    rec["biography"] = re.sub(r"<[^>]+>", " ",
                                              rec["biography"])[:1000].strip()
                m = _PRICE_RE.search(text)
                rec["subscribe_price_usd"] = float(m.group(1)) if m else None
                rec["is_free"] = (rec["subscribe_price_usd"] == 0.0
                                  if rec["subscribe_price_usd"] is not None
                                  else None)
                m = _VERIFIED_RE.search(text)
                rec["is_verified"] = (m.group(1) == "true") if m else None
                m = _JOIN_RE.search(text)
                rec["join_date"] = _dec(m.group(1)) if m else None
                m = _PERFORMER_RE.search(text)
                rec["is_performer"] = (m.group(1) == "true") if m else None
                if not rec.get("full_name"):
                    m = _OG_TITLE_RE.search(text)
                    if m:
                        rec["full_name"] = htmllib.unescape(
                            m.group(1)).split(" OnlyFans")[0].strip()
                # « followers » au sens crea-ka : le compteur public = J'aime
                rec["followers"] = None
                await Actor.push_data(rec)

        await asyncio.gather(*[one(u) for u in usernames])
        Actor.log.info(f"terminé : {len(usernames)} profils")