# ==============================================================================
# Author: Simon-Pierre Boucher <contact@spboucher.ai>
# File:   creaka/connectors/podcastindex.py
# Desc:   Connecteur ENRICHISSEMENT « podcastindex » — API Podcast Index
#         (OFFICIELLE, gratuite ; clé + secret requis, inscription gratuite :
#         https://api.podcastindex.org). Palier 2 (§9, §10). Sans clés :
#         passage sauté PROPREMENT (SkipSource, comme twitch).
# ==============================================================================
"""Enrichissement des balados via l'API Podcast Index.

``/podcasts/byfeedurl`` retourne pour un flux : episodeCount,
lastUpdateTime, itunesId, image/artwork, catégories — reach/fraîcheur/avatar
des balados. Authentification officielle : en-têtes X-Auth-Key + X-Auth-Date
+ Authorization = sha1(key + secret + date).

Complément (avec clés) du connecteur sans clé `balados-rss` : mêmes champs,
mais en UNE requête légère par balado au lieu de télécharger le flux complet.
"""
from __future__ import annotations

import hashlib
import os
import time
from datetime import datetime, timedelta, timezone

from ..normalize import map_niche
from ..schema import Creator, now_iso
from .base import BaseConnector, SkipSource
from .balados_itunes import DORMANT_AFTER_DAYS, _GENRE_NICHE

API_URL = "https://api.podcastindex.org/api/1.0/podcasts/byfeedurl"


class PodcastIndexConnector(BaseConnector):
    source_id = "podcastindex"
    kind = "enrichment"
    request_delay = 0.4
    max_feeds = 200               # cap par passage
    revisit_days = 7              # re-visite hebdomadaire par balado

    def __init__(self) -> None:
        super().__init__()
        self.errors = 0

    def _auth_headers(self, key: str, secret: str) -> dict:
        ts = str(int(time.time()))
        digest = hashlib.sha1((key + secret + ts).encode("utf-8")).hexdigest()
        return {"X-Auth-Key": key, "X-Auth-Date": ts, "Authorization": digest}

    def _needs_visit(self, metrics: dict) -> bool:
        raw = metrics.get("pi_checked")
        if not raw:
            return True
        try:
            checked = datetime.fromisoformat(raw.replace("Z", "+00:00"))
        except ValueError:
            return True
        return (datetime.now(timezone.utc) - checked
                > timedelta(days=self.revisit_days))

    def enrich(self, creators: list[Creator]) -> list[Creator]:
        key = os.environ.get("PODCASTINDEX_API_KEY", "")
        secret = os.environ.get("PODCASTINDEX_API_SECRET", "")
        if not key or not secret:
            # pas de clés → passage sauté PROPREMENT (§15) ; le connecteur
            # sans clé balados-rss couvre les mêmes champs en attendant
            raise SkipSource("clés PODCASTINDEX_API_KEY/PODCASTINDEX_API_SECRET "
                             "manquantes (.env — inscription gratuite : "
                             "https://api.podcastindex.org)")
        enriched: list[Creator] = []
        fetched = 0
        now = datetime.now(timezone.utc)
        for cr in creators:
            if fetched >= self.max_feeds:
                break
            acc = next((a for a in cr.platforms
                        if a.platform == "podcast" and a.metrics.get("feed_url")),
                       None)
            if acc is None or not self._needs_visit(acc.metrics):
                continue
            try:
                resp = self.get(API_URL,
                                params={"url": acc.metrics["feed_url"]},
                                headers=self._auth_headers(key, secret))
                feed = resp.json().get("feed") or {}
            except Exception:
                self.errors += 1
                continue
            fetched += 1
            if not feed or not isinstance(feed, dict):
                continue
            acc.metrics["pi_checked"] = now_iso()
            acc.last_checked = now_iso()
            if feed.get("episodeCount"):
                acc.metrics["episodes"] = feed["episodeCount"]
            if feed.get("itunesId"):
                acc.metrics["itunes_id"] = feed["itunesId"]
            last_ts = feed.get("lastUpdateTime") or feed.get("newestItemPubdate")
            if last_ts:
                last = datetime.fromtimestamp(int(last_ts), tz=timezone.utc)
                acc.metrics["last_episode"] = last.strftime("%Y-%m-%dT%H:%M:%SZ")
                if now - last > timedelta(days=DORMANT_AFTER_DAYS):
                    acc.metrics["dormant"] = True
                    if all(a.platform == "podcast" for a in cr.platforms):
                        cr.status = "inactive"     # §13, réversible au retour
                else:
                    acc.metrics.pop("dormant", None)
            image = (feed.get("image") or feed.get("artwork") or "").strip()
            if image.startswith("http") and not cr.avatar_url:
                cr.avatar_url = image
            cats = feed.get("categories") or {}
            niches = {map_niche(_GENRE_NICHE.get(c, c))
                      for c in cats.values() if isinstance(c, str)}
            niches.discard("autre")
            if niches:
                cr.niches = sorted(set(cr.niches or []) | niches)
            enriched.append(cr)
        return enriched