# -----------------------------------------------------------------------------
# Food-Ka — Agrégateur de produits d'épicerie (province de Québec)
# Auteur : Simon-Pierre Boucher — contact@spboucher.ai
# connectors/_metro.py : socle commun Metro / Super C (même plateforme)
#   Tuiles produit rendues côté serveur (data-product-code) mais site derrière
#   un anti-bot agressif -> Scrapfly (ASP). Pagination : « <allée>-page-N ».
# -----------------------------------------------------------------------------
from __future__ import annotations

import re

from bs4 import BeautifulSoup

from ..schema import Product, parse_price
from .base import BaseConnector

_PER_KG_RE = re.compile(r"(\d+(?:,\d+)?)\s*\$\s*/\s*kg")

# (slug d'allée, catégorie canonique)
AISLES = [
    ("fruits-et-legumes", "Fruits et légumes"),
    ("viandes-et-volailles", "Viandes et volailles"),
    ("poissons-et-fruits-de-mer", "Poissons et fruits de mer"),
    ("produits-laitiers-et-oeufs", "Produits laitiers et œufs"),
    ("pains-et-patisseries", "Boulangerie"),
    ("produits-surgeles", "Surgelés"),
    ("garde-manger", "Garde-manger"),
    ("boissons", "Boissons"),
    ("collations", "Collations et confiseries"),
]


class MetroPlatformConnector(BaseConnector):
    """Base Metro/Super C — sous-classes : définir source_id et domain."""

    domain: str = ""              # ex. "www.metro.ca"
    aisle_prefix: str = "/epicerie-en-ligne/allees"   # superc : "/allees"
    pages_per_aisle: int = 2      # ~45 produits/page

    def _parse_tiles(self, html: str, category: str) -> list[Product]:
        soup = BeautifulSoup(html, "html.parser")
        products: list[Product] = []
        for tile in soup.select("[data-product-code]"):
            code = tile.get("data-product-code")
            name = tile.get("data-product-name") or ""
            if not code or not name:
                continue
            link_el = tile.select_one("a.product-details-link[href]")
            link = link_el["href"] if link_el else ""
            img_el = tile.select_one(".pt__visual img[src]")
            image = (img_el["src"] if img_el else "") or ""
            brand_el = tile.select_one(".head__brand")
            size_el = tile.select_one(".head__unit-details")
            price_el = tile.select_one(".price-update")
            before_el = tile.select_one(".pricing__before-price")
            unit_el = tile.select_one(".pricing__secondary-price")
            price_label = price_el.get_text(" ", strip=True) if price_el else ""
            unit_label = unit_el.get_text(" ", strip=True) if unit_el else ""
            before_txt = before_el.get_text(" ", strip=True) if before_el else ""
            price = None
            regular = parse_price(before_txt) if before_txt else None
            # Produit au poids : le prix affiché est une estimation à l'unité
            # mais le « prix régulier » est au kg — on compare tout au kg.
            if "kg" in before_txt:
                m_cur = _PER_KG_RE.search(unit_label)
                m_reg = _PER_KG_RE.search(before_txt)
                if m_cur:
                    price = float(m_cur.group(1).replace(",", "."))
                    price_label = f"{m_cur.group(1)} $ / kg"
                    regular = (float(m_reg.group(1).replace(",", "."))
                               if m_reg else None)
                else:   # pas de prix au kg comparable : ne rien inventer
                    regular = None
            on_sale = bool(tile.select_one(".pi-price-promo, .promo-price")) or regular is not None
            products.append(Product(
                source=self.source_id,
                external_id=str(code),
                url=(f"https://{self.domain}{link}" if link.startswith("/") else link),
                name=name,
                brand=brand_el.get_text(strip=True) if brand_el else "",
                category=category,
                category_raw=tile.get("data-product-category") or "",
                size_label=size_el.get_text(" ", strip=True) if size_el else "",
                price=price,
                price_label=price_label,
                regular_price=regular,
                on_sale=on_sale,
                unit_price_label=unit_label,
                images=[image] if image.startswith("http") else [],
            ))
        return products

    def fetch(self) -> list[Product]:
        products: list[Product] = []
        seen: set[str] = set()
        for slug, category in AISLES:
            for page in range(1, self.pages_per_aisle + 1):
                suffix = slug if page == 1 else f"{slug}-page-{page}"
                url = f"https://{self.domain}{self.aisle_prefix}/{suffix}"
                try:
                    html = self.get_scrapfly(url)
                except Exception:   # une allée qui casse ne bloque pas le reste
                    break
                batch = self._parse_tiles(html, category)
                new = 0
                for prod in batch:
                    if prod.uid not in seen:
                        seen.add(prod.uid)
                        products.append(prod)
                        new += 1
                if new == 0:   # page vide ou répétée : allée épuisée
                    break
        return products