"""Parser for Dash Storage (dashstorage.com/storage/)."""

from __future__ import annotations

import re

from bs4 import BeautifulSoup

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


class Facility001590Parser(BaseParser):
    """Extract storage units from Dash Storage.

    The site lists unit sizes (e.g. 5'x10', 10'x20') in <h2> tags within
    .fl-rich-text divs under a "Styles Available" section. Each size is
    paired with a description of what fits. No prices are published.
    """

    platform = "custom_facility_001590"

    # Matches sizes like 5'x10', 10'x20', 12'x40', 14'x40'
    _SIZE_RE = re.compile(
        r"(\d+)\s*['\u2032]?\s*[xX\u00d7]\s*(\d+)\s*['\u2032]?"
    )

    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()

        seen: set[str] = set()

        # Unit sizes appear in <h2> tags within .fl-rich-text divs inside
        # nested column groups under the "Styles Available" heading.
        for h2 in soup.find_all("h2"):
            text = h2.get_text(strip=True)
            m = self._SIZE_RE.search(text)
            if not m:
                continue

            width = int(m.group(1))
            length = int(m.group(2))

            # Skip implausible sizes (e.g. image dimensions)
            if width < 3 or length < 3:
                continue

            size_str = f"{width}'x{length}'"
            if size_str in seen:
                continue
            seen.add(size_str)

            # Look for description in the sibling column
            description = ""
            col_group = h2.find_parent(class_="fl-col-group-nested")
            if col_group:
                desc_div = col_group.find("p")
                if desc_div:
                    description = desc_div.get_text(separator=" ", strip=True)

            unit = UnitResult()
            unit.size = size_str
            w, ln, sq = self.normalize_size(size_str)
            if w is not None:
                unit.metadata = {"width": w, "length": ln, "sqft": sq}
            unit.price = None
            unit.description = description[:200] if description else size_str
            result.units.append(unit)

        if not result.units:
            result.warnings.append("No units found on page")
        else:
            result.warnings.append("No pricing published on this site")

        return result