"""Parser for Morningstar Storage facility 084967 (CANDEE platform)."""

from __future__ import annotations

import json
import re

from bs4 import BeautifulSoup

from src.parsers.base import BaseParser, ParseResult, UnitResult


class Facility084967Parser(BaseParser):
    """Extract storage units from Morningstar Storage (CANDEE reservation system).

    The CANDEE platform renders unit data in div.unitMasterData elements with
    data-price (monthly rate) and a nested div.unitName containing the size
    string (e.g. "5x5 Heated & Cooled Ground").  data-size holds the sq ft.
    """

    platform = "custom_facility_084967"

    _DIM_RE = re.compile(r"(\d+\.?\d*)\s*[xX×]\s*(\d+\.?\d*)")

    def parse(self, html: str, url: str = "") -> ParseResult:
        soup = BeautifulSoup(html, "lxml")
        result = ParseResult(platform=self.platform, parser_name=self.__class__.__name__)

        seen: set[tuple[str, str]] = set()

        for unit_el in soup.find_all(class_="unitMasterData"):
            price_raw = unit_el.get("data-price", "").strip()
            price = self.normalize_price(price_raw)
            if price is None:
                continue

            # unitName contains the display name e.g. "5x5 Heated & Cooled Ground"
            name_el = unit_el.find(class_="unitName")
            if not name_el:
                continue
            name_text = name_el.get_text(strip=True)

            # Extract dimension from the name text
            dim_m = self._DIM_RE.search(name_text)
            if dim_m:
                size_text = dim_m.group(0)
            else:
                # Fall back to sq ft if no WxL found
                sqft_raw = unit_el.get("data-size", "")
                size_text = f"{sqft_raw} sqft" if sqft_raw else name_text

            key = (size_text, price_raw)
            if key in seen:
                continue
            seen.add(key)

            # Features from data-features JSON array
            features_raw = unit_el.get("data-features", "")
            desc: str | None = None
            if features_raw:
                try:
                    features_list = json.loads(features_raw)
                    # Deduplicate while preserving order
                    seen_feats: list[str] = []
                    for f in features_list:
                        if f not in seen_feats:
                            seen_feats.append(f)
                    desc = ", ".join(seen_feats)
                except (json.JSONDecodeError, TypeError):
                    pass

            unit = UnitResult(
                size=size_text,
                price=price,
                description=desc,
            )
            w, ln, sq = self.normalize_size(size_text)
            if w is not None:
                unit.metadata = {"width": w, "length": ln, "sqft": sq}
            elif unit_el.get("data-size"):
                try:
                    unit.metadata = {"sqft": float(unit_el["data-size"])}
                except ValueError:
                    pass
            result.units.append(unit)

        if not result.units:
            result.warnings.append("No units found in CANDEE unitMasterData elements")

        return result