"""Parser for Sound Storage (Wix brochure site, no unit pricing data)."""

from __future__ import annotations

import re

from bs4 import BeautifulSoup

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


class Facility043186Parser(BaseParser):
    """Extract storage units from Sound Storage.

    This is a Wix site with multiple locations (Port Townsend, Friday Harbor).
    The homepage does not publish unit sizes or prices.
    """

    platform = "custom_facility_043186"

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

        for tag in soup.find_all(["script", "style"]):
            tag.decompose()

        text = soup.get_text(separator="\n")

        # Try to extract any size x price patterns
        size_price_re = re.compile(
            r"(\d+)\s*['\u2019\u2032]?\s*[xX×]\s*(\d+)\s*['\u2019\u2032]?"
            r"[^\$]{0,80}"
            r"\$(\d[\d,.]*)",
            re.DOTALL,
        )
        for m in size_price_re.finditer(text):
            w, ln = float(m.group(1)), float(m.group(2))
            if w < 3 or ln < 3:
                continue
            unit = UnitResult()
            unit.size = f"{int(w)}x{int(ln)}"
            unit.price = self.normalize_price(m.group(3))
            unit.metadata = {"width": w, "length": ln, "sqft": w * ln}
            result.units.append(unit)

        # Try size-only extraction
        if not result.units:
            size_re = re.compile(r"(\d+)\s*['\u2019\u2032]?\s*[xX×]\s*(\d+)")
            seen: set[str] = set()
            for m in size_re.finditer(text):
                w, ln = float(m.group(1)), float(m.group(2))
                if w < 3 or ln < 3:
                    continue
                key = f"{int(w)}x{int(ln)}"
                if key in seen:
                    continue
                seen.add(key)
                unit = UnitResult()
                unit.size = key
                unit.metadata = {"width": w, "length": ln, "sqft": w * ln}
                result.units.append(unit)

        if not result.units:
            result.warnings.append("No units found - brochure site without pricing data")

        return result