# -----------------------------------------------------------------------------
# Lou-Ka — Agrégateur de logements à louer (province de Québec)
# Auteur : Simon-Pierre Boucher — contact@spboucher.ai
# connectors/cent4.py : connecteur Terrasse Cent4 (terrassecent4.com —
#   condos locatifs neufs 3½/4½ au 1700, chemin Saint-Jean, La Prairie,
#   promoteur Vestric). WordPress (thème Voyou) rendu serveur : la page
#   /les-unites liste les unités disponibles par bâtiment (Terrasse 2/3)
#   dans des <li class="vy_units_listing_item"> — numéro, étage, typologie,
#   pi², loyer, mois de disponibilité + lien « fiche de l'unité ».
#   La fiche (via self.detail, cache BD) ajoute inclusions et options.
#   Granularité = UNITÉ. external_id = slug de la fiche (terrasse-2-unite-202).
# -----------------------------------------------------------------------------
from __future__ import annotations

import html as _html
import re

from ..schema import Listing
from .base import BaseConnector

BASE = "https://terrassecent4.com"
LIST_URL = f"{BASE}/les-unites"
ADDRESS = "1700, chemin Saint-Jean, La Prairie"
CITY = "La Prairie"

BUILDING_RE = re.compile(
    r"<div class=\"vy_units_list_item\" data-filter-building='([^']+)'")
ITEM_RE = re.compile(
    r'<li class="vy_units_listing_item"[\s\S]*?</li>')
FIELD_RE = re.compile(
    r'vy_units_listing_(unit|floor|type|square|price|availability)[\s\S]*?'
    r'<span>([\s\S]*?)</span>')
FICHE_RE = re.compile(r'href="(https://terrassecent4\.com/units/[^"]+)"')
AVAILABLE_RE = re.compile(r'data-filter-available="1"')
INCL_RE = re.compile(r"Inclusions([\s\S]*?)(?:Options|Imprimez)", re.I)
OPT_RE = re.compile(r"Options([\s\S]*?)Imprimez", re.I)
SCRIPT_RE = re.compile(r"<(script|style)[\s\S]*?</\1>", re.I)
TAG_RE = re.compile(r"<[^>]+>")
IMG_RE = re.compile(          # Bedrock : médias sous /app/uploads/ (data-src)
    r'https://terrassecent4\.com/app/uploads/[^"\s)]+'
    r"\.(?:jpe?g|png|webp)", re.I)


def _flat(html: str) -> str:
    t = _html.unescape(SCRIPT_RE.sub(" ", html))
    return re.sub(r"[\s  ]+", " ", TAG_RE.sub(" | ", t))


def _clean(txt: str) -> str:
    return re.sub(r"\s+", " ", _html.unescape(TAG_RE.sub(" ", txt))).strip()


class Cent4Connector(BaseConnector):
    source_id = "cent4"
    request_delay = 0.8

    def fetch(self) -> list[Listing]:
        try:
            html = self.get(LIST_URL).text
        except Exception:
            return []

        listings: list[Listing] = []
        seen: set[str] = set()
        # les items sont regroupés par bloc bâtiment ; on suit le bâtiment
        # courant en balayant le HTML dans l'ordre
        pos_buildings = [(m.start(), m.group(1))
                         for m in BUILDING_RE.finditer(html)]
        for m in ITEM_RE.finditer(html):
            chunk = m.group(0)
            if not AVAILABLE_RE.search(chunk):
                continue
            building = ""
            for pos, b in pos_buildings:
                if pos < m.start():
                    building = b.replace("-", " ").title()
            fields = {k: _clean(v) for k, v in FIELD_RE.findall(chunk)}
            unit = fields.get("unit", "")
            fm = FICHE_RE.search(chunk)
            url = fm.group(1) if fm else LIST_URL
            ext_id = (url.rsplit("/", 1)[-1] if fm
                      else f"{building.lower().replace(' ', '-')}-{unit}")
            if not unit or ext_id in seen:
                continue
            seen.add(ext_id)

            price = None
            pm = re.search(r"(\d[\d\s  ]{2,8})\$", fields.get("price", ""))
            if pm:
                price = float(re.sub(r"[\s  ]", "", pm.group(1)))
            area = None
            am = re.search(r"(\d{3,5})\s*pi", fields.get("square", ""))
            if am:
                area = float(am.group(1))
            availability = re.sub(r"^Disponibilité\s*\|?\s*", "",
                                  fields.get("availability", "")).strip()

            detail = self.detail(ext_id, f"{price}|{availability}",
                                 lambda u=url: self._detail(u))

            listings.append(Listing(
                source=self.source_id,
                external_id=ext_id,
                url=url,
                title=f"Terrasse Cent4 ({building}) — unité {unit}",
                address=ADDRESS,
                city=CITY,
                unit_type=fields.get("type", ""),
                price=price,
                price_label=f"{int(price)} $ par mois" if price else "",
                availability=(f"Disponibilité : {availability}"
                              if availability else ""),
                area_sqft=area,
                description=detail.get("description", ""),
                amenities=detail.get("amenities", []),
                details={"building": building,
                         "floor": fields.get("floor", "")},
                images=detail.get("images", []),
            ))
        return listings

    def _detail(self, url: str) -> dict:
        """Fiche d'unité : inclusions, options, photos."""
        try:
            html = self.get(url).text
        except Exception:
            return {}
        flat = _flat(html)
        amenities: list[str] = []
        for rx in (INCL_RE, OPT_RE):
            fm = rx.search(flat)
            if fm:
                for part in fm.group(1).split("|"):
                    part = part.strip(" .")
                    if 3 <= len(part) <= 120 and part not in amenities:
                        amenities.append(part)
        images = [u for u in dict.fromkeys(IMG_RE.findall(html))
                  if not re.search(r"logo|icon|favicon|hummingbird|-\d+x\d+\.",
                                   u, re.I)][:10]
        desc = ("Condo locatif neuf à la Terrasse Cent4, complexe de "
                f"Vestric à La Prairie ({ADDRESS}).")
        return {"description": desc, "amenities": amenities, "images": images}