"""RSS / Atom feeds (feedparser) → normalized items."""
from __future__ import annotations

from dataclasses import dataclass, field
from datetime import UTC, datetime
from typing import Any

import feedparser

from aiatlas.sdk.extract.html import clean_text


@dataclass
class FeedItem:
    id: str
    url: str
    title: str
    summary: str | None
    published_at: datetime | None
    updated_at: datetime | None
    authors: list[str] = field(default_factory=list)
    categories: list[str] = field(default_factory=list)
    content_html: str | None = None
    raw: dict[str, Any] = field(default_factory=dict, repr=False)


def _dt(struct: Any) -> datetime | None:
    if not struct:
        return None
    try:
        return datetime(*struct[:6], tzinfo=UTC)
    except Exception:  # noqa: BLE001
        return None


def parse_feed(content: bytes | str) -> tuple[dict[str, Any], list[FeedItem]]:
    parsed = feedparser.parse(content)
    feed_meta = {"title": parsed.feed.get("title"), "link": parsed.feed.get("link"), "updated": parsed.feed.get("updated"),
                 "subtitle": parsed.feed.get("subtitle"), "bozo": bool(parsed.get("bozo"))}
    items: list[FeedItem] = []
    for e in parsed.entries:
        url = e.get("link") or ""
        if not url:
            for link in e.get("links", []):
                if link.get("rel") in (None, "alternate") and link.get("href"):
                    url = link["href"]
                    break
        content_html = None
        if e.get("content"):
            content_html = e["content"][0].get("value")
        summary = e.get("summary") or e.get("description")
        items.append(FeedItem(
            id=e.get("id") or e.get("guid") or url,
            url=url,
            title=clean_text(e.get("title") or "")[:500],
            summary=clean_text(_strip_tags(summary))[:4000] if summary else None,
            published_at=_dt(e.get("published_parsed")) or _dt(e.get("created_parsed")),
            updated_at=_dt(e.get("updated_parsed")),
            authors=[a.get("name") for a in e.get("authors", []) if a.get("name")] or ([e["author"]] if e.get("author") else []),
            categories=[t.get("term") for t in e.get("tags", []) if t.get("term")],
            content_html=content_html,
            raw={k: v for k, v in e.items() if isinstance(v, (str, int, float))},
        ))
    return feed_meta, items


def _strip_tags(s: str) -> str:
    from selectolax.parser import HTMLParser

    try:
        return HTMLParser(s).text(separator=" ")
    except Exception:  # noqa: BLE001
        return s


__all__ = ["FeedItem", "parse_feed"]