import { z } from 'zod';
import { BaseConnector, html as H, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import { normalizeCondition } from '@rareindex/taxonomy';
import { CurrencySchema, extractYear, type AssetAttributes, type NormalizedListing, type NormalizedRecord } from '@rareindex/shared';

/**
 * Chrono24 connector — watch listings (asking prices) from public model pages.
 * The page's schema.org ItemList of Offers is the extraction contract; one raw record per page.
 */

const BASE = 'https://www.chrono24.com';
const PARSER_VERSION = '1.0.0';

export const OfferSchema = z.object({ name: z.string(), price: z.number(), currency: z.string(), url: z.string(), image: z.string().nullable() });
export const PagePayloadSchema = z.object({
  kind: z.literal('model_page'),
  url: z.string(),
  brand: z.string(),
  model: z.string(),
  breadcrumb: z.array(z.string()),
  offers: z.array(OfferSchema),
  totalListings: z.number().nullable(),
});
export type PagePayload = z.infer<typeof PagePayloadSchema>;

const BRAND_CATEGORY: Record<string, string> = { rolex: 'rolex', patekphilippe: 'patek_philippe', 'patek-philippe': 'patek_philippe', audemarspiguet: 'audemars_piguet', 'audemars-piguet': 'audemars_piguet', omega: 'omega' };
const BRAND_NAME: Record<string, string> = { rolex: 'Rolex', patekphilippe: 'Patek Philippe', audemarspiguet: 'Audemars Piguet', omega: 'Omega', cartier: 'Cartier', tudor: 'Tudor', vacheronconstantin: 'Vacheron Constantin', breitling: 'Breitling', iwc: 'IWC', jaegerlecoultre: 'Jaeger-LeCoultre', alangesoehne: 'A. Lange & Söhne', richardmille: 'Richard Mille', grandseiko: 'Grand Seiko', tagheuer: 'TAG Heuer', panerai: 'Panerai', hublot: 'Hublot', zenith: 'Zenith', breguet: 'Breguet', blancpain: 'Blancpain' };

/** Watch reference heuristics: 116500LN, 126610LV, 5711/1A-010, 15400ST.OO.1220ST.01, 311.30.42.30.01.005, RM 011 */
const REF_RE = /\b(\d{4,6}[A-Z]{0,3}(?:\/\d[A-Z0-9]*)?(?:-\d{3})?|\d{5}[A-Z]{2}\.[A-Z]{2}\.\d{4}[A-Z]{2}\.\d{2}|\d{3}\.\d{2}\.\d{2}\.\d{2}\.\d{2}\.\d{3}|RM\s?\d{2,3}(?:-\d{2})?)\b/;

export function parseModelPage(htmlText: string, url: string): PagePayload {
  const ld = H.jsonLd(htmlText);
  const crumbs: string[] = [];
  const offers: z.infer<typeof OfferSchema>[] = [];
  for (const block of ld) {
    const type = block['@type'];
    if (type === 'BreadcrumbList') {
      for (const it of (block.itemListElement as Array<{ item?: { name?: string } }>) ?? []) if (it.item?.name) crumbs.push(it.item.name);
    }
    // Listings are published as an AggregateOffer whose `offers` array holds one Offer per watch.
    const offerList = type === 'AggregateOffer' ? ((block.offers as Array<Record<string, unknown>>) ?? []) : type === 'ItemList' ? (((block.itemListElement as Array<Record<string, unknown>>) ?? []).map((el) => (el.item as Record<string, unknown> | undefined) ?? el)) : [];
    if (offerList.length) {
      const defaultCurrency = String(block.priceCurrency ?? '');
      for (const off of offerList) {
        const t = off?.['@type'];
        if (t !== 'Offer' && t !== 'Product') continue;
        const priceRaw = off.price ?? (off.offers as { price?: unknown } | undefined)?.price;
        const price = Number(priceRaw);
        const currency = String(off.priceCurrency ?? (off.offers as { priceCurrency?: unknown } | undefined)?.priceCurrency ?? defaultCurrency);
        const href = String(off.url ?? '');
        if (!Number.isFinite(price) || price <= 0 || !href) continue;
        const img = off.image;
        const image = Array.isArray(img) ? ((img[0] as { contentUrl?: string; url?: string } | string) ?? null) : (img as string | null | undefined) ?? null;
        offers.push({ name: String(off.name ?? '').replace(/\s+/g, ' ').trim(), price, currency, url: href, image: typeof image === 'string' ? image : (image?.contentUrl ?? image?.url ?? null) });
      }
    }
  }
  const pathBrand = url.match(/chrono24\.com\/([a-z-]+)\//)?.[1] ?? '';
  const brand = BRAND_NAME[pathBrand] ?? crumbs[1]?.replace(/\s+watches?$/i, '') ?? pathBrand;
  const model = (crumbs[crumbs.length - 1] ?? '').replace(/\s+watches?$/i, '');
  const total = htmlText.match(/([\d,.]+)\s+(?:listings|watches)\b/)?.[1] ?? null;
  return { kind: 'model_page', url, brand, model, breadcrumb: crumbs, offers, totalListings: total ? Number(total.replace(/[,.]/g, '')) : null };
}

export class Chrono24Connector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  protected override minIntervalMs = 2000;

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const seeds = ctx.options.seeds?.length ? ctx.options.seeds : ((this.meta.config.seeds as string[] | undefined) ?? []);
    const pages = Number(this.meta.config.pagesPerSeed ?? 1);
    const pageSize = Number(this.meta.config.pageSize ?? 120);
    let count = 0;
    for (const seed of seeds) {
      if (ctx.signal?.aborted) return;
      for (let page = 1; page <= pages; page++) {
        if (this.reached(ctx, count)) return;
        const path = page === 1 ? seed : seed.replace(/--mod(\d+)\.htm$/, `--mod$1-${page}.htm`);
        const url = `${path.startsWith('http') ? path : BASE + path}?pageSize=${pageSize}&showpage=${page}`;
        await this.throttle();
        const res = await ctx.fetch(url, {
          renderJs: false,
          country: 'us',
          responseType: 'text',
          expect: ['title', 'price', 'currency', 'images'],
          parse: (r) => {
            if (!r.html) return null;
            const p = parseModelPage(r.html, url);
            const first = p.offers[0];
            return { title: p.model || null, price: first?.price ?? null, currency: first?.currency ?? null, images: p.offers.filter((o) => o.image).map((o) => o.image) };
          },
        });
        if (!res.success || !res.html) {
          ctx.anomaly('page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
          break;
        }
        const payload = parseModelPage(res.html, url.split('?')[0]!);
        if (payload.offers.length === 0) {
          ctx.anomaly('empty_page', url);
          break;
        }
        count++;
        yield { url: url.split('?')[0]!, externalId: `${seed}#${page}`, kind: 'listing', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
      }
    }
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PagePayloadSchema.parse(raw.payload);
    const pathBrand = p.url.match(/chrono24\.com\/([a-z-]+)\//)?.[1] ?? '';
    const categorySlug = BRAND_CATEGORY[pathBrand] ?? 'other_watches';
    const out: NormalizedListing[] = [];
    const seen = new Set<string>();
    for (const o of p.offers) {
      const id = o.url.match(/--id(\d+)\.htm/)?.[1] ?? o.url;
      if (seen.has(id)) continue;
      seen.add(id);
      const cur = CurrencySchema.safeParse(o.currency);
      if (!cur.success) continue;
      // First reference-looking token that is not a plain year (e.g. "UNWORN 2024 126500LN" → 126500LN).
      const ref = [...o.name.matchAll(new RegExp(REF_RE.source, 'g'))].map((m) => m[1]!.replace(/\s+/g, ' ')).find((r) => !/^(18|19|20)\d{2}$/.test(r)) ?? null;
      const year = extractYear(o.name);
      const lower = o.name.toLowerCase();
      const conditionRaw = /unworn|brand new|new\b/.test(lower) ? 'Unworn' : /\bmint\b|like new|excellent/.test(lower) ? 'Excellent' : /pre-owned|used/.test(lower) ? 'Pre-owned' : null;
      const completeness = /full set|box (?:and|&|\/) papers|box\/papers|complete set/.test(lower) ? 'full_set' : /papers/.test(lower) ? 'papers_only' : /\bbox\b/.test(lower) ? 'box_only' : null;
      const attributes: AssetAttributes = {
        categorySlug,
        subcategorySlug: null,
        franchise: null,
        brand: p.brand,
        series: null,
        set: null,
        setCode: null,
        name: `${p.brand} ${p.model}`.trim(),
        model: p.model,
        reference: ref,
        number: null,
        year,
        edition: null,
        variant: null,
        language: null,
        region: null,
        country: null,
        material: /platinum/.test(lower) ? 'platinum' : /yellow gold|rose gold|everose|white gold|18k|gold/.test(lower) ? 'gold' : /two[- ]tone|rolesor/.test(lower) ? 'two-tone' : /steel|stainless/.test(lower) ? 'steel' : /titanium/.test(lower) ? 'titanium' : /ceramic/.test(lower) ? 'ceramic' : null,
        size: o.name.match(/\b(\d{2}(?:\.\d)?)\s?mm\b/i)?.[1] ? `${o.name.match(/\b(\d{2}(?:\.\d)?)\s?mm\b/i)![1]}mm` : null,
        color: null,
        rarity: null,
        productionQuantity: null,
        originalMsrp: null,
        originalMsrpCurrency: null,
        identifiers: { ...(ref ? { reference: ref } : {}), chrono24_id: id },
        metadata: { model_page: p.url },
      };
      out.push({
        kind: 'listing',
        connectorId: this.meta.id,
        sourceId: this.meta.sourceId,
        sourceUrl: o.url,
        externalId: id,
        rawTitle: o.name,
        description: null,
        imageUrls: o.image ? [o.image] : [],
        attributes,
        grade: { grader: null, grade: null, qualifier: null, certificationNumber: null },
        condition: { condition: normalizeCondition(categorySlug, conditionRaw), conditionRaw, completeness },
        observedAt: raw.fetchedAt,
        confidence: 0.8,
        parserVersion: PARSER_VERSION,
        listingType: 'fixed_price',
        price: o.price,
        currency: cur.data,
        seller: null,
        sellerReputation: null,
        location: null,
        shippingCost: null,
        quantity: 1,
        listedAt: null,
        endsAt: null,
        availability: 'available',
        bidCount: null,
      });
    }
    return out;
  }
}

export default (meta: ConnectorMeta) => new Chrono24Connector(meta);