"""XML sitemaps and sitemap indexes."""
from __future__ import annotations

import re
from dataclasses import dataclass
from xml.etree import ElementTree as ET

_NS = re.compile(r"\{[^}]+\}")


@dataclass
class SitemapEntry:
    loc: str
    lastmod: str | None = None
    is_index: bool = False


def parse_sitemap(content: bytes | str) -> list[SitemapEntry]:
    if isinstance(content, str):
        content = content.encode()
    try:
        root = ET.fromstring(content)
    except ET.ParseError:
        return []
    tag = _NS.sub("", root.tag).lower()
    is_index = tag == "sitemapindex"
    out: list[SitemapEntry] = []
    for node in root:
        loc = None
        lastmod = None
        for child in node:
            name = _NS.sub("", child.tag).lower()
            if name == "loc" and child.text:
                loc = child.text.strip()
            elif name == "lastmod" and child.text:
                lastmod = child.text.strip()
        if loc:
            out.append(SitemapEntry(loc=loc, lastmod=lastmod, is_index=is_index))
    return out


__all__ = ["SitemapEntry", "parse_sitemap"]