import { z } from 'zod';
import { adapters, BaseConnector, html as H, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import type { NormalizedRecord, NormalizedSale } from '@rareindex/shared';
import { makeSale } from '../../firecrawl/_carlib/index.js';
import { clean, coinYear, isNumisBundle, numisAttributes, numisCategory, parseAuctionDate, parseCoinGrade, realized } from '../../firecrawl/_g5-numismatics-lib/index.js';

/**
 * Fritz Rudolf Künker (Osnabrück) — Europe's largest coin auction house. Every past lot keeps a static
 * Shopware product page (≈ 725 000 URLs in the product sitemaps) with auction name, lot number, sale
 * date, estimate and "Hammer price" in EUR, plus a properties table (Nominal/Year, Mint, Condition with
 * PCGS/NGC slab data, references). Plain HTTPS; the JS lot listings (/widgets/) are robots-disallowed and
 * never used — discovery goes through the sitemaps.
 */

const SITE = 'https://www.kuenker.de';
const PARSER_VERSION = '1.0.0';

export const PayloadSchema = z.object({
  kind: z.literal('lot_page'),
  url: z.string(),
  productId: z.string(),
  title: z.string(),
  section1: z.string().nullable(),
  section2: z.string().nullable(),
  auctionName: z.string().nullable(),
  lotNumber: z.string().nullable(),
  dateText: z.string().nullable(),
  statusText: z.string().nullable(),
  estimateText: z.string().nullable(),
  hammerText: z.string().nullable(),
  properties: z.record(z.string(), z.string()),
  description: z.string().nullable(),
  images: z.array(z.string()),
});
export type Payload = z.infer<typeof PayloadSchema>;

export function parseLotPage(htmlText: string, url: string): Payload | null {
  const $ = H.load(htmlText);
  const productId = url.match(/\/(\d+)\/?(?:[?#].*)?$/)?.[1] ?? null;
  const h1 = $('h1.product-detail-name').first();
  const section1 = H.text(h1.find('.section-1').first());
  const section2 = H.text(h1.find('.section-2').first());
  const ld = H.jsonLd(htmlText, 'Product')[0] as { name?: string; sku?: string; image?: string[] | string; description?: string } | undefined;
  // the JSON-LD name carries the full issuer line ("SCHWEDEN KÖNIGREICH Oskar II. …"); the mobile h1 may drop the country
  const title = (ld?.name ? clean(ld.name) : '') || clean(`${section1 ?? ''} ${section2 ?? ''}`);
  if (!productId || !title) return null;
  const info = $('.auction-lot-info').first();
  const auctionName = H.text(info.find('b').first());
  const lotNumber = (H.text(info.find('span').first()) ?? '').match(/(\d+[A-Za-z]?)/)?.[1] ?? null;
  const dateBox = $('.date-status-container').first();
  const dateText = H.text(dateBox.find('span').first());
  const statusText = clean(dateBox.clone().children('span').remove().end().text()) || null;
  const estimateText = ($('.estimated-price').first().text().match(/:\s*([^\n]+)/)?.[1] ?? '').trim() || null;
  const hammerBlock = $('.bid-status-finished').first();
  const hammerText = /hammer price|zuschlag/i.test(hammerBlock.text()) ? H.text(hammerBlock.find('.price').first()) : null;
  const properties: Record<string, string> = {};
  $('table.product-detail-properties-table tr.properties-row').each((_, tr) => {
    const k = H.text($(tr).find('th').first());
    const v = H.text($(tr).find('td').first());
    if (k && v) properties[k] = v;
  });
  const images: string[] = [];
  const ldImages = Array.isArray(ld?.image) ? ld!.image : ld?.image ? [ld.image] : [];
  for (const i of ldImages) if (typeof i === 'string') images.push(i.replace(/\\\//g, '/'));
  const og = $('meta[property="og:image"]').attr('content');
  if (og && !images.includes(og)) images.push(og);
  const description = ld?.description ? clean(ld.description.replace(/&lt;/g, '<').replace(/&gt;/g, '>').replace(/&quot;/g, '"')) : null;
  return { kind: 'lot_page', url, productId, title, section1, section2, auctionName, lotNumber, dateText, statusText, estimateText, hammerText, properties, description, images: images.slice(0, 3) };
}

/** "Fast Stempelglanz / In US-Plastikholder der PCGS mit der Bewertung MS 64 PL (812624.64/38930174)." → slab ids. */
export function slabIds(condition: string | null | undefined): { pcgsNumber: string | null; cert: string | null } {
  const m = condition?.match(/\((\d{4,7})\.(\d{2})\/(\d{7,10})\)/);
  if (m) return { pcgsNumber: m[1]!, cert: m[3]! };
  const c = condition?.match(/\b(?:Zertifikat|cert(?:ificate)?|No\.?|Nr\.?)\s*#?\s*(\d{7,10})\b/i);
  return { pcgsNumber: null, cert: c?.[1] ?? null };
}

interface Cursor {
  sitemapIndex?: number;
  urlIndex?: number;
  done?: boolean;
  updatedAt?: string;
}

export class KuenkerConnector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  protected override minIntervalMs = 1500;
  override readonly urlPatterns = [/kuenker\.de\/[^/]+\/\d+\/?$/i];

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const lotsPerRun = Number(this.meta.config.lotsPerRun ?? 100);
    const cursor = { ...((ctx.options.cursor ?? {}) as Cursor) };
    const seeds = ctx.options.seeds?.filter((s) => /kuenker\.de\//.test(s)) ?? [];
    let urls: string[] = seeds;
    if (!urls.length) {
      await this.throttle();
      const index = await ctx.fetch(`${SITE}/sitemap.xml`, { engines: ['api'], responseType: 'text', minQuality: 0, force: true });
      const children = index.success && index.html ? adapters.parseSitemapIndex(index.html).filter((c) => /sitemap-products-/.test(c.loc)) : [];
      if (!children.length) {
        ctx.anomaly('sitemap_fetch_failed', `sitemap.xml: ${index.error ?? index.httpStatus ?? 'no product sitemaps'}`);
        return;
      }
      children.sort((a, b) => Number(a.loc.match(/products-(\d+)/)?.[1] ?? 0) - Number(b.loc.match(/products-(\d+)/)?.[1] ?? 0));
      const si = Math.min(cursor.sitemapIndex ?? 0, children.length - 1);
      await this.throttle();
      const sm = await ctx.fetch(children[si]!.loc, { engines: ['api'], responseType: 'text', minQuality: 0, force: true });
      const entries = sm.success && sm.html ? adapters.parseUrlset(sm.html) : [];
      if (!entries.length) {
        ctx.anomaly('sitemap_fetch_failed', `${children[si]!.loc}: ${sm.error ?? sm.httpStatus}`);
        return;
      }
      const start = cursor.urlIndex ?? 0;
      urls = entries.slice(start, start + lotsPerRun).map((e) => e.loc);
      const nextIndex = start + urls.length;
      const exhausted = nextIndex >= entries.length;
      cursor.sitemapIndex = exhausted ? (si + 1 >= children.length ? 0 : si + 1) : si;
      cursor.urlIndex = exhausted ? 0 : nextIndex;
      cursor.done = ctx.options.mode === 'backfill' && exhausted && si + 1 >= children.length;
      if (ctx.options.mode === 'backfill') await ctx.progress({ page: si * 25_000 + nextIndex, totalPages: children.length * 25_000, itemsProcessed: 0 });
    }
    let count = 0;
    for (const url of urls) {
      if (ctx.signal?.aborted || this.reached(ctx, count)) break;
      if (ctx.options.mode !== 'backfill' && !(await ctx.shouldFetch(url))) continue;
      await this.throttle();
      const res = await ctx.fetch(url, {
        engines: ['api', 'firecrawl'],
        responseType: 'text',
        expect: ['title', 'price', 'date'],
        parse: (r) => {
          const p = r.html ? parseLotPage(r.html, url) : null;
          return p ? { title: p.title, price: p.hammerText ?? p.estimateText, date: p.dateText } : null;
        },
      });
      const payload = res.success && res.html ? parseLotPage(res.html, url) : null;
      if (!payload) {
        ctx.anomaly(res.success ? 'parse_failure_page' : 'page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
        continue;
      }
      count++;
      yield { url, externalId: payload.productId, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
    }
    await ctx.setCursor({ ...cursor, updatedAt: new Date().toISOString() });
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PayloadSchema.parse(raw.payload);
    const price = realized(p.hammerText, 'EUR');
    const saleDate = parseAuctionDate(p.dateText);
    if (!price || !saleDate) return []; // unsold, withdrawn or still running → no sale
    const condition = p.properties.Condition ?? p.properties.Erhaltung ?? null;
    const text = `${p.title} ${condition ?? ''}`;
    const g = parseCoinGrade(text);
    const slab = slabIds(condition);
    const categorySlug = numisCategory(p.title, `${p.auctionName ?? ''} ${p.section1 ?? ''}`, /medaille|medal|orden/i.test(p.title) && !/münze|coin|taler|dukat|mark|pfennig|kreuzer/i.test(p.title) ? 'medals' : 'coins');
    const identifiers: Record<string, string> = { kuenker_product_id: p.productId };
    if (slab.pcgsNumber) identifiers.pcgs_number = slab.pcgsNumber;
    if (slab.cert && g.grader) identifiers[`${g.grader}_cert`] = slab.cert;
    const estimate = realized(p.estimateText, 'EUR');
    const attributes = numisAttributes({
      categorySlug,
      title: p.title,
      section: p.section1,
      materialHint: p.properties.Weight ?? p.properties.Gewicht ?? null,
      year: coinYear(p.properties['Nominal/Year'] ?? p.properties['Nominal/Jahr'] ?? '').year,
      identifiers,
      metadata: {
        auction: p.auctionName,
        status: p.statusText,
        estimate: estimate?.amount ?? null,
        estimate_currency: estimate?.currency ?? null,
        hammer_price: price.amount,
        buyer_premium: "excluded — page label 'Hammer price' (Zuschlag); Künker's premium is stated in the auction terms",
        nominal_year: p.properties['Nominal/Year'] ?? p.properties['Nominal/Jahr'] ?? null,
        mint: p.properties.Mint ?? p.properties.Prägestätte ?? null,
        rarity: p.properties.Rarity ?? p.properties.Seltenheit ?? null,
        weight: p.properties.Weight ?? p.properties.Gewicht ?? null,
        references: p.properties.Quotes ?? p.properties.Zitate ?? null,
        condition_text: condition,
      },
    });
    if (p.properties.Mint && !attributes.variant) attributes.variant = p.properties.Mint.replace(/\.$/, '');
    const sale = makeSale({
      meta: this.meta,
      sourceUrl: p.url,
      externalId: p.productId,
      rawTitle: p.title,
      description: p.description,
      attributes,
      price: price.amount,
      currency: price.currency,
      saleDate,
      buyerPremiumIncluded: false,
      auctionHouse: 'Fritz Rudolf Künker',
      lotNumber: p.lotNumber,
      imageUrls: p.images,
      observedAt: raw.fetchedAt,
      parserVersion: PARSER_VERSION,
      confidence: g.grader ? 0.9 : 0.82,
      isBundle: isNumisBundle(p.title) || /\bLot\s+von\b|\bLots?\b.*\bStück\b/i.test(p.title),
      conditionRaw: condition?.split('/')[0]?.trim() ?? g.conditionRaw,
      location: 'DE',
    });
    sale.grade = { grader: g.grader, grade: g.grade, qualifier: g.qualifier, certificationNumber: slab.cert ?? g.certificationNumber };
    return [sale];
  }
}

export default (meta: ConnectorMeta) => new KuenkerConnector(meta);