import type { ConnectorMeta, CrawlContext } from '@rareindex/connectors';
import type { ExtractionResult } from '@rareindex/shared';
import { parseEuDate, parseEuMoney } from '../_g8-auctions-eu-apac-lib/index.js';
import { SaleResultsConnector, absolute, chunksBetween, pick, textOf, type HouseConfig, type ParsedLot, type ParsedSalePage, type SaleRef } from '../_g8-auctions-eu-apac-lib/sale-results.js';

/**
 * Millon (Paris, Bruxelles, Nice…) — Drupal site, fully server-rendered. Past sales: /catalogue/ventes-passees?page=N
 * (0-based, 36 cards/page, ~1 740 sales); sale page /catalogue/vente<id>-<slug>?page=N (0-based, 50 lots/page) with
 * cards "Lot 12 · title · sub-title · description · Adjugé à | 160 €". The sale year is not printed on the sale
 * page ("lun 06 Juil - lun 07 Sep"); each lot page states "Vente <title> du 6 juillet 2026", so the connector reads
 * the first lot page of a sale once to get the date. "Adjugé à" is not labelled hammer/with-fees → premium basis null.
 */
const BASE = 'https://www.millon.com';
const PAGE = 50;

export function parseMillonIndex(htmlText: string): SaleRef[] {
  const out: SaleRef[] = [];
  const seen = new Set<string>();
  for (const chunk of chunksBetween(htmlText, /<article class="card[^"]*">/)) {
    const href = chunk.match(/href="(\/catalogue\/vente(\d+)[^"/?#]*)"/);
    if (!href || seen.has(href[2]!)) continue;
    seen.add(href[2]!);
    const title = pick(chunk, /<h5 class="card-title"><a[^>]*>([\s\S]*?)<\/a>/) ?? pick(chunk, /title="([^"]+)"/) ?? href[2]!;
    const dayText = pick(chunk, /day-of-sale">([\s\S]*?)<\/span>/);
    const location = pick(chunk, /selling-location">([\s\S]*?)<\/span>/);
    out.push({ id: href[2]!, title, url: `${BASE}${href[1]}`, date: null, location: location || null, extra: { day_text: dayText } });
  }
  return out;
}

/** Total pages from the Drupal pagerer widget (`max="49"` → 49 pages, 0-based query). */
export function parsePagerMax(htmlText: string): number | null {
  const m = htmlText.match(/class="pagerer-page"[^>]*max="(\d+)"/);
  return m ? Number(m[1]) : null;
}

/** Lot page → sale date ("Vente <a>…</a> du 6 juillet 2026"). */
export function parseLotPageDate(htmlText: string): string | null {
  const m = htmlText.match(/sale--info">[\s\S]*?<\/a>\s*(?:du|le|des)\s+([^<]+?)\s*<\/p>/);
  const d = m ? parseEuDate(m[1]!) : null;
  return d ? d.toISOString() : null;
}

export function parseMillonSalePage(htmlText: string, sale: SaleRef, page: number): ParsedSalePage | null {
  if (!/class="card[^"]*"/.test(htmlText) || !/<span class="lot">/.test(htmlText)) return null;
  const lots: ParsedLot[] = [];
  for (const chunk of chunksBetween(htmlText, /<article class="card[^"]*">/)) {
    const lotNo = pick(chunk, /<span class="lot">\s*Lot\s*([^<]*)<\/span>/);
    const href = chunk.match(/href="(\/catalogue\/vente\d+[^"]*\/lot\d+[^"]*)"/)?.[1] ?? null;
    const title = pick(chunk, /<h5 class="card-title"><a[^>]*>([\s\S]*?)<\/a>/);
    if (!lotNo || !title || !href) continue;
    const sub = pick(chunk, /<p class="sub-title">([\s\S]*?)<\/p>/);
    const desc = pick(chunk, /<div class="description">([\s\S]*?)<\/div>/);
    const est = chunk.match(/<span class="label">\s*Estimation[^<]*<\/span>\s*([^<]+)</);
    const adj = chunk.match(/<span class="label">\s*Adjugé[^<]*<\/span>\s*([^<]+)</);
    const price = adj ? parseEuMoney(textOf(adj[1]), 'EUR') : null;
    const estM = est ? textOf(est[1]).match(/([\d\s.,]+)\s*€?\s*-\s*([\d\s.,]+)\s*€/) : null;
    const image = chunk.match(/srcset="(\/sites\/millon\/files\/[^"\s]+)/)?.[1] ?? null;
    lots.push({
      lotNo: lotNo.replace(/\s+/g, ''),
      title,
      subtitle: sub || null,
      description: desc || null,
      url: absolute(BASE, href)!,
      image: image ? absolute(BASE, image) : null,
      price: price?.amount ?? null,
      currency: 'EUR',
      premiumIncluded: null,
      estimateLow: estM ? parseEuMoney(`${estM[1]} €`, 'EUR')?.amount ?? null : null,
      estimateHigh: estM ? parseEuMoney(`${estM[2]} €`, 'EUR')?.amount ?? null : null,
      date: null,
      sold: price !== null,
      extra: { adjuge_text: adj ? textOf(adj[1]) : null },
    });
  }
  const max = parsePagerMax(htmlText);
  const title = pick(htmlText, /<h1 class="title-large">([\s\S]*?)<\/h1>/);
  const type = pick(htmlText, /<p class="type">([\s\S]*?)<\/p>/);
  const location = pick(htmlText, /<span class="sale-status">([\s\S]*?)<\/span>/) ?? pick(htmlText, /selling-location">([\s\S]*?)<\/span>/);
  return { lots, hasMore: max !== null ? page < max : lots.length >= PAGE, totalLots: null, sale: { title: title ?? undefined, location: location || undefined, extra: { sale_type: type ?? null, day_text: pick(htmlText, /day-of-sale">([\s\S]*?)<\/span>/) } } };
}

export class MillonConnector extends SaleResultsConnector {
  readonly version = '1.0.0';
  readonly house: HouseConfig = { houseName: 'Millon', defaultCurrency: 'EUR', location: 'Paris, France', idKey: 'millon_lot', premiumIncluded: null, fallbackSlug: 'antiques', minIntervalMs: 2500, maxPagesPerSale: 30 };
  protected override minIntervalMs = 2500;

  async listSales(ctx: CrawlContext): Promise<SaleRef[]> {
    const pages = ctx.options.mode === 'backfill' ? Number(this.meta.config.backfillIndexPages ?? 60) : Number(this.meta.config.indexPages ?? 1);
    const all: SaleRef[] = [];
    for (let p = 0; p < pages; p++) {
      const url = `${BASE}/catalogue/ventes-passees${p > 0 ? `?page=${p}` : ''}`;
      await this.throttle(url);
      const res = await ctx.fetch(url, { engines: ['api'], responseType: 'text', minQuality: 0, timeoutMs: 60_000 });
      if (!res.success || !res.html) {
        ctx.anomaly('past_list_failed', `${url}: ${res.error ?? res.httpStatus}`);
        break;
      }
      const sales = parseMillonIndex(res.html);
      if (sales.length === 0) break;
      // keep index order (most recent first) as a pseudo-date so incremental runs pick newest sales first
      for (const s of sales) all.push({ ...s, extra: { ...s.extra, index_rank: all.length } });
      const max = parsePagerMax(res.html);
      if (max !== null && p + 1 >= max) break;
    }
    return all;
  }

  salePageUrl(sale: SaleRef, page: number): string {
    return page > 1 ? `${sale.url}?page=${page - 1}` : sale.url;
  }

  parseSalePage(res: ExtractionResult, sale: SaleRef, page: number): ParsedSalePage | null {
    return res.html ? parseMillonSalePage(res.html, sale, page) : null;
  }

  /** Page 1 also fetches the first lot page to learn the sale date (year is absent from the sale page). */
  protected override async fetchSalePage(ctx: CrawlContext, sale: SaleRef, page: number): Promise<{ url: string; res: ExtractionResult; parsed: ParsedSalePage | null }> {
    const r = await super.fetchSalePage(ctx, sale, page);
    if (r.parsed && page === 1 && !sale.date && r.parsed.lots[0]) {
      const lotUrl = r.parsed.lots[0].url;
      await this.throttle(lotUrl);
      const lp = await ctx.fetch(lotUrl, { engines: ['api'], responseType: 'text', minQuality: 0, timeoutMs: 60_000 });
      const date = lp.success && lp.html ? parseLotPageDate(lp.html) : null;
      if (!date) ctx.anomaly('selector_missing', `${sale.id}: sale date not found on ${lotUrl}`);
      r.parsed.sale = { ...(r.parsed.sale ?? {}), date: date ?? undefined };
    }
    return r;
  }
}

export default function createConnector(meta: ConnectorMeta) {
  return new MillonConnector(meta);
}