"""
Lemmy community discovery.

Searches major Lemmy instances for communities matching a keyword and
returns candidates for user review. Nothing is stored until the user
imports via PUT /subs/{community@instance}.
"""
from __future__ import annotations

import logging
from typing import Any

import httpx

from app.services.lemmy.client import KNOWN_INSTANCES, _DEFAULT_USER_AGENT

logger = logging.getLogger(__name__)

# Instances to fan search across (ordered by size)
_SEARCH_INSTANCES = ["lemmy.world", "lemmy.ml", "beehaw.org", "sh.itjust.works", "reddthat.com"]

_TIMEOUT = 10.0


def search_lemmy(
    keyword: str,
    limit: int = 15,
    known_subs: set[str] | None = None,
) -> list[dict[str, Any]]:
    """
    Search multiple Lemmy instances for communities matching keyword.

    Deduplicates by actor_id (canonical AP URL) so federated communities
    don't appear twice. Returns candidates sorted by subscriber count.
    """
    seen_actor_ids: set[str] = set()
    results: list[dict] = []

    for instance in _SEARCH_INSTANCES:
        try:
            r = httpx.get(
                f"https://{instance}/api/v3/search",
                params={
                    "q": keyword,
                    "type_": "Communities",
                    "limit": 20,
                    "sort": "TopAll",
                },
                headers={"User-Agent": _DEFAULT_USER_AGENT},
                timeout=_TIMEOUT,
                follow_redirects=True,
            )
            if r.status_code != 200:
                logger.debug("Lemmy search on %s returned %d", instance, r.status_code)
                continue

            communities = r.json().get("communities", [])

            for cv in communities:
                comm = cv.get("community", {})
                counts = cv.get("counts", {})

                actor_id: str = comm.get("actor_id", "")
                if not actor_id or actor_id in seen_actor_ids:
                    continue
                seen_actor_ids.add(actor_id)

                # Derive home instance from actor_id (e.g. https://lemmy.world/c/selfhosted)
                try:
                    home_instance = actor_id.split("/")[2]
                except IndexError:
                    home_instance = instance

                if comm.get("nsfw"):
                    continue

                name = comm.get("name", "")
                sub_key = f"{name}@{home_instance}"
                subscribers = counts.get("subscribers", 0)
                description = (comm.get("description") or "").strip()

                results.append({
                    "sub": sub_key,
                    "title": comm.get("title") or name,
                    "subscribers": subscribers,
                    "description": description[:280],
                    "promo_allowed": None,  # Lemmy has no standardized promo rules
                    "flair_required": False,
                    "available_flairs": [],
                    "rule_warning": False,
                    "notes": None,
                    "already_tracked": (sub_key.lower() in known_subs) if known_subs else False,
                    "platform": "lemmy",
                })

        except Exception:
            logger.warning("Error searching Lemmy instance %s", instance, exc_info=True)

    results.sort(key=lambda x: x["subscribers"], reverse=True)
    return results[:limit]