# -----------------------------------------------------------------------------
# Lou-Ka — Agrégateur de logements à louer (province de Québec)
# Auteur : Simon-Pierre Boucher — contact@spboucher.ai
# connectors/excell.py : Les immeubles EXCELL (immeublesexcell.com) — agence
#   immobilière d'Abitibi-Ouest (La Sarre, Macamic, Taschereau…) avec une
#   section « Logements » à louer.
#   Techno : ColdFusion servi côté serveur (details.cfm?LogementID=N) — simple
#   parse HTML des cartes `ficheMaison` + page détail pour les attributs
#   (type, meublé, animaux, chauffage, électroménagers, étage).
#   Granularité : LOGEMENT (une carte = un logement, LogementID stable).
# -----------------------------------------------------------------------------
from __future__ import annotations

import hashlib
import html as _html
import re

import requests

from .base import BaseConnector
from ..schema import Listing

BASE = "https://www.immeublesexcell.com"
LIST_URL = f"{BASE}/fr/logements/"

_CARD_SPLIT_RE = re.compile(r'<div class="ficheMaison">')
_ID_RE = re.compile(r'LogementID_(\d+)')
_PRICE_RE = re.compile(r'<h4>([^<]+)</h4>')
_ADDR_RE = re.compile(r'<p>\s*([^<]+?)<br')
_IMG_RE = re.compile(r'background:url\(([^)]+)\)')
# attributs « <span class="valeurAttribut">X</span>…<span class="valeurAttributDetails">Y</span> »
_ATTR_RE = re.compile(
    r'<span class="valeurAttribut">([^<]*)</span>\s*<br\s*/?>\s*'
    r'(?:<span class="valeurAttributDetails">)?([^<]*)', re.S)
_POSTAL_RE = re.compile(r'\b[A-Za-z]\d[A-Za-z]\s?\d[A-Za-z]\d\b')
_DETAIL_IMG_RE = re.compile(r'(?:src|href)="([^"]*images/logements[^"]*)"')

# labels détail -> clés structurées
_DETAIL_KEYS = {"type", "mobilié", "mobilier", "chauffage", "animaux",
                "étage", "frigidaire", "laveuse", "sécheuse", "extérieur",
                "poêle", "lave-vaisselle"}


def _abs(url: str) -> str:
    url = url.strip().strip("'\"")
    if url.startswith("http"):
        return url
    return BASE + "/" + url.lstrip("./").lstrip("/")


class ExcellConnector(BaseConnector):
    """Les immeubles EXCELL — logements à louer en Abitibi-Ouest."""

    source_id = "excell"
    request_delay = 0.8

    def _parse_detail(self, ext_id: str) -> dict:
        """Attributs de la page détail (dict JSON-sérialisable, pour cache)."""
        url = f"{LIST_URL}details.cfm?LogementID={ext_id}"
        h = re.sub(r"\s+", " ", self.get(url).text)
        attrs: dict[str, str] = {}
        for k, v in _ATTR_RE.findall(h):
            k, v = _html.unescape(k).strip(), _html.unescape(v).strip()
            if k and v and k.lower() in _DETAIL_KEYS | {"location"}:
                attrs[k] = v
        images = [_abs(u) for u in _DETAIL_IMG_RE.findall(h)]
        return {"attrs": attrs, "images": images}

    def fetch(self) -> list[Listing]:
        page = self.get(LIST_URL).text
        out: list[Listing] = []
        seen: set[str] = set()
        for block in _CARD_SPLIT_RE.split(page)[1:]:
            b = re.sub(r"\s+", " ", block)
            m = _ID_RE.search(b)
            if not m:
                continue
            ext_id = m.group(1)
            if ext_id in seen:
                continue
            seen.add(ext_id)
            url = f"{LIST_URL}details.cfm?LogementID={ext_id}"

            price_m = _PRICE_RE.search(b)
            price_label = _html.unescape(price_m.group(1)).strip() if price_m else ""
            addr_m = _ADDR_RE.search(b)
            raw_addr = _html.unescape(addr_m.group(1)).strip() if addr_m else ""
            raw_addr = _POSTAL_RE.sub("", raw_addr).strip(" ,")
            parts = [p.strip() for p in raw_addr.split(",") if p.strip()]
            city = parts[-1] if len(parts) > 1 else ""
            address = ", ".join(parts[:-1]) if len(parts) > 1 else raw_addr

            # attributs de la carte : pièces (5 1/2), chambres, sdb, dispo
            unit_type, bedrooms, bathrooms, avail = "", None, None, ""
            for val, lbl in _ATTR_RE.findall(b):
                val = _html.unescape(val).strip()
                lbl = _html.unescape(lbl).strip()
                low = lbl.lower()
                if "pièce" in low and val:
                    unit_type = val
                elif "chambre" in low and val:
                    try:
                        bedrooms = float(val)
                    except ValueError:
                        pass
                elif "bain" in low and val:
                    try:
                        bathrooms = float(val)
                    except ValueError:
                        pass
                elif not val and lbl:        # icône calendrier : début location
                    avail = lbl

            imgs: list[str] = []
            img_m = _IMG_RE.search(b)
            if img_m and "no-image" not in img_m.group(1):
                imgs.append(_abs(img_m.group(1)))

            # page détail (cache : clé = hash de la carte)
            key = hashlib.sha256(b.encode()).hexdigest()[:16]
            try:
                det = self.detail(ext_id, key,
                                  lambda e=ext_id: self._parse_detail(e))
            except requests.RequestException:
                det = {}
            attrs = det.get("attrs") or {}
            for u in det.get("images") or []:
                if u not in imgs:
                    imgs.append(u)

            # filtre résidentiel : le site ne liste que des logements ;
            # on écarte tout de même les locaux commerciaux éventuels
            typ = (attrs.get("Type") or "").lower()
            if typ and any(w in typ for w in ("commercial", "bureau", "local",
                                              "entrepôt", "garage")):
                continue

            furnished = None
            mob = (attrs.get("Mobilié") or attrs.get("Mobilier") or "").lower()
            if mob:
                furnished = "meublé" in mob and "non" not in mob
            pets = None
            ani = (attrs.get("Animaux") or "").lower()
            if ani:
                pets = "oui" if ani.startswith("oui") else \
                       "non" if ani.startswith("non") else "conditions"

            if "chambre" in typ:                # chambre en maison de chambres
                unit_type = "Chambre"

            amenities = [f"{k} : {v}" for k, v in attrs.items()
                         if k.lower() in _DETAIL_KEYS]
            avail = attrs.get("Location") or avail

            out.append(Listing(
                source=self.source_id,
                external_id=ext_id,             # LogementID ColdFusion (stable)
                url=url,
                title=f"{address}, {city}".strip(" ,"),
                address=address,
                city=city,
                unit_type=unit_type,
                bedrooms=bedrooms,
                bathrooms=bathrooms,
                price_label=price_label,        # ex. « 975 $ / mois »
                availability=f"Location : {avail}" if avail else "",
                furnished=furnished,
                pets=pets,
                amenities=amenities,
                details={"region": "Abitibi-Témiscamingue"},
                images=imgs[:30],
            ))
        return out