# -----------------------------------------------------------------------------
# Immo-Ka — Agrégateur de maisons à vendre (province de Québec)
# Auteur : Simon-Pierre Boucher — contact@spboucher.ai
# connectors/kw_prestige.py : Keller Williams Prestige (kwprestige.com)
#   Plateforme macleweb.com. La grille /fr/proprietes/ est servie par l'API
#   interne /ajax/getProps.php (JSON {html, nbPages}) : cartes complètes —
#   prix, adresse, ville (secteur), cac/sdb, mini-description, photo. Les
#   fiches détail sont rendues serveur : galerie complète (data-centris/
#   large_photo), description, tableau Centris, GPS (iframe maps/embed),
#   courtiers inscripteurs.
#   ⚠️ Ne PAS revenir au sitemap.xml : il est figé (lastmod 2025-09) et ses
#   fiches expirées rendent un gabarit vide (prix « 0 », aucune adresse).
#   ⚠️ Le site exige un User-Agent navigateur (curl nu → 403) ; l'UA de
#   BaseConnector passe.
# -----------------------------------------------------------------------------
from __future__ import annotations

import html as _html
import os
import re

from .base import BaseConnector
from . import _detailutil as du
from ..normalize import parse_price
from ..schema import PropertyListing

BASE = "https://www.kwprestige.com"
AJAX = BASE + "/ajax/getProps.php?json=1&default_lang=fr&page={page}&qty=100"
DETAIL_LIMIT = int(os.environ.get("IMMOKA_KW_DETAIL_LIMIT", "400"))

# carte de la grille : <a href='/{slug}-{id}-detail-Fr' class='proprieteitem'>…</a>
_CARD_RE = re.compile(
    r"<a href='(/[^']*?-(\d{6,9})-detail-Fr)' class='proprieteitem'>(.*?)</a>", re.S)
_PRICE_RE = re.compile(r"class='titre2'>(.*?)</p>", re.S)
_ADDR_RE = re.compile(r"class='titresm'>(.*?)</h3>\s*<p>(.*?)</p>", re.S)
_ICON_RE = re.compile(r"</svg>\s*(\d+)\s*</div>")
_MINIDESC_RE = re.compile(r"class='minidescription'>(.*?)</p>", re.S)
_THUMB_RE = re.compile(r"src='(https://www\.kwprestige\.com/data-centris/photo/[^']+)'")

# fiche détail
_LARGE_RE = re.compile(
    r'https://www\.kwprestige\.com/data-centris/large_photo/[^"\'\\ ]+'
    r'\.(?:jpg|jpeg|png|webp)', re.I)
_DPRICE_RE = re.compile(r'<p class="prix">(.*?)</p>', re.S)
_DDESC_RE = re.compile(r'fichecdescription">.*?<p>(.*?)</p>', re.S)
_GPS_RE = re.compile(r'!2d(-?\d{2,3}\.\d+)!3d(-?\d{2}\.\d+)')
_BROKER_RE = re.compile(r'courtierfiche">.*?<h3 class="titresm">(.*?)</h3>', re.S)
_PHONE_RE = re.compile(r'href="tel:([+\d.\-]{7,})"')
_ROOM_RE = re.compile(r'class="nbr_(chambres|sdb|pieces)">\s*<img[^>]*>\s*(\d+)')

# type de propriété depuis le slug d'URL (« {type}-a-vendre-… »)
_TYPE = [
    ("location-d-espace", "Commercial"), ("location-de", "Commercial"),
    ("location", "Commercial"),
    ("maison-etages", "Maison"), ("maison-paliers", "Maison"),
    ("maison-plain-pied", "Maison"), ("maison-mobile", "Maison mobile"),
    ("maison", "Maison"), ("appartement", "Condo"), ("condo", "Condo"),
    ("loft", "Condo"), ("duplex", "Duplex"), ("triplex", "Triplex"),
    ("quadruplex", "Multiplex"), ("quintuplex", "Multiplex"),
    ("propriete-revenu", "Multiplex"), ("terrain", "Terrain"),
    ("chalet", "Chalet"), ("fermette", "Fermette/Agricole"),
    ("ferme", "Fermette/Agricole"),
    ("local-commercial", "Commercial"), ("espace-commercial", "Commercial"),
    ("espace-industriel", "Commercial"), ("espace-bureau", "Commercial"),
    ("commerce", "Commercial"), ("bureau", "Commercial"),
    ("batisse", "Commercial"), ("industriel", "Commercial"),
]


def _txt(s: str) -> str:
    return _html.unescape(re.sub(r"<[^>]+>", " ", s or "")).replace("\xa0", " ").strip()


def _city_sector(label: str) -> tuple[str, str]:
    """« Montréal (Ville-Marie) » -> (Montréal, Ville-Marie)."""
    m = re.match(r"(.+?)\s*\((.+?)\)\s*$", label or "")
    if m:
        return m.group(1).strip(), m.group(2).strip()
    return (label or "").strip(), ""


class KwPrestigeConnector(BaseConnector):
    source_id = "kw_prestige"
    request_delay = 0.5

    def fetch(self) -> list[PropertyListing]:
        by_id: dict[str, PropertyListing] = {}
        page = 1
        while page <= 50:                      # garde-fou
            data = self.get(AJAX.format(page=page)).json()
            for path, cid, block in _CARD_RE.findall(data.get("html") or ""):
                if cid not in by_id:
                    by_id[cid] = self._to_listing(path, cid, block)
            if page >= int(data.get("nbPages") or 1):
                break
            page += 1
        listings = list(by_id.values())
        du.enrich(self, listings, DETAIL_LIMIT, parse_kw_prestige_detail, key="v3")
        return listings

    def _to_listing(self, path: str, cid: str, block: str) -> PropertyListing:
        m_price = _PRICE_RE.search(block)
        price_label = _txt(m_price.group(1)) if m_price else ""
        m_addr = _ADDR_RE.search(block)
        addr = _txt(m_addr.group(1)) if m_addr else ""
        city, sector = _city_sector(_txt(m_addr.group(2)) if m_addr else "")
        nums = _ICON_RE.findall(block)
        m_desc = _MINIDESC_RE.search(block)
        m_img = _THUMB_RE.search(block)
        slug = path.lstrip("/")
        typ_slug = re.split(r"-a-(?:vendre|louer)-", slug)[0]
        ptype = ""
        for key, canon in _TYPE:
            if typ_slug.startswith(key):
                ptype = canon
                break
        low = price_label.lower()
        status = "vendu" if low.startswith(("vendu", "loué", "loue")) else "a-vendre"
        return PropertyListing(
            source=self.source_id,
            external_id=cid,
            url=BASE + path,
            title=addr or _txt(price_label),
            status=status,
            address=addr,
            city=city,
            sector=sector,
            property_type=ptype,
            price=parse_price(price_label),
            price_label=price_label,
            bedrooms=int(nums[0]) if nums else None,
            bathrooms=int(nums[1]) if len(nums) > 1 else None,
            description=_txt(m_desc.group(1)) if m_desc else "",
            images=[m_img.group(1)] if m_img else [],
            mls=cid,
            broker_name="Keller Williams Prestige",
            agency="Keller Williams Prestige",
        )


def parse_kw_prestige_detail(html: str) -> dict:
    """Fiche détail macleweb : galerie complète, prix, description, tableau
    Centris, GPS (iframe Google Maps embed), courtiers, cac/sdb/pièces."""
    out: dict = {}
    seen, imgs = set(), []
    for u in _LARGE_RE.findall(html):
        if u not in seen:
            seen.add(u)
            imgs.append(u)
    if imgs:
        out["images"] = imgs
    m = _DPRICE_RE.search(html)
    if m:
        label = _txt(m.group(1))
        price = parse_price(label)
        if price:                                # « 0 » = gabarit vide/expiré
            out["price"] = price
            out["price_label"] = label
    m = _DDESC_RE.search(html)
    if m:
        d = _txt(m.group(1))
        if len(d) > 40:
            out["description"] = d
    m = _GPS_RE.search(html)
    if m:
        lng, lat = float(m.group(1)), float(m.group(2))
        if 44.5 <= lat <= 63.0 and -80.0 <= lng <= -56.0:
            out["lat"], out["lng"] = lat, lng
    m = _BROKER_RE.search(html)
    if m:
        out["broker_name"] = _txt(m.group(1))
    m = _PHONE_RE.search(html)
    if m:
        out["broker_phone"] = m.group(1)
    for key, n in _ROOM_RE.findall(html):
        if key == "chambres":
            out["bedrooms"] = int(n)
        elif key == "sdb":
            out["bathrooms"] = int(n)
    det = du.centris_details(du.flatten(html))
    if det:
        out.setdefault("details", {}).update(det)
    return out