import { z } from 'zod';
import { BaseConnector, html as H, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import type { NormalizedRecord } from '@rareindex/shared';
import { lotAttributes, makeSale, money, vehicleAttributes } from '../../firecrawl/_carlib/index.js';

/**
 * Historics Auctioneers — results list → per-sale "Past lots" grid (96 lots per page) with "Sold £X".
 * One raw record per lots page; one sale per lot with a price.
 */
const BASE = 'https://www.historics.co.uk';
const PARSER_VERSION = '1.0.0';

export const AuctionSchema = z.object({ au: z.string(), slug: z.string(), title: z.string(), saleNumber: z.string().nullable(), endedOn: z.string().nullable(), lotCount: z.number().nullable() });
export type Auction = z.infer<typeof AuctionSchema>;
export const LotSchema = z.object({ lotId: z.string(), url: z.string(), lotNo: z.string().nullable(), title: z.string(), subtitle: z.string().nullable(), soldText: z.string().nullable(), priceGbp: z.number().nullable(), image: z.string().nullable() });
export type Lot = z.infer<typeof LotSchema>;
export const PagePayloadSchema = z.object({ kind: z.literal('lots_page'), url: z.string(), auction: AuctionSchema, page: z.number(), lots: z.array(LotSchema) });

const MONTHS: Record<string, number> = { jan: 0, feb: 1, mar: 2, apr: 3, may: 4, jun: 5, jul: 6, aug: 7, sep: 8, oct: 9, nov: 10, dec: 11 };
/** "18th Jul, 2026 9:30" | "6th Aug, 2026 19:30" → UTC midnight. */
export function parseUkDate(s: string | null | undefined): Date | null {
  const m = s?.match(/(\d{1,2})(?:st|nd|rd|th)?\s+([A-Za-z]{3})[a-z]*,?\s+(\d{4})/);
  if (!m) return null;
  const mo = MONTHS[m[2]!.toLowerCase()];
  return mo === undefined ? null : new Date(Date.UTC(Number(m[3]), mo, Number(m[1])));
}

/** Parse /auction-results: calendar items with title, "Date:"/"Ends:" and "Sale number:". */
export function parseResultsList(htmlText: string): Auction[] {
  const $ = H.load(htmlText);
  const out: Auction[] = [];
  const seen = new Set<string>();
  $('.auction-calendar-item').each((_, el) => {
    const href = $(el).find('a[href*="/auction/details/"]').first().attr('href') ?? '';
    const m = href.match(/\/auction\/details\/([^/?]+)\?au=(\d+)/);
    if (!m || seen.has(m[2]!)) return;
    const text = $(el).text().replace(/\s+/g, ' ').trim();
    const title = H.text($(el).find('.auction-calendar-text h2, .auction-calendar-text h3, .auction-calendar-text h4').first()) ?? text.split(/\s(?:Date|Ends):/)[0]!.trim();
    const dateTxt = text.match(/(?:Date|Ends):\s*([^S]+?)\s+Sale number/i)?.[1] ?? text.match(/(?:Date|Ends):\s*(\d{1,2}(?:st|nd|rd|th)?\s+[A-Za-z]{3,9},?\s+\d{4})/i)?.[1] ?? null;
    const saleNumber = text.match(/Sale number:\s*([A-Z]{1,2}O?\d{2,4})/i)?.[1] ?? null;
    const lots = text.match(/Lots:\s*(\d+)/i)?.[1];
    seen.add(m[2]!);
    out.push({ au: m[2]!, slug: m[1]!, title, saleNumber, endedOn: parseUkDate(dateTxt)?.toISOString() ?? null, lotCount: lots ? Number(lots) : null });
  });
  return out;
}

/** Parse a lots page: cards with lot number, title, "Sold £X". */
export function parseLotsPage(htmlText: string): Lot[] {
  const $ = H.load(htmlText);
  const out: Lot[] = [];
  $('.auction-lot').each((_, el) => {
    const a = $(el).find('.auction-lot-title a').first();
    const href = a.attr('href') ?? '';
    const lotId = href.match(/[?&]lot=(\d+)/)?.[1];
    if (!lotId) return;
    const titleEl = a.find('.lot-title').clone();
    const subtitle = H.text(titleEl.find('.sub-title'));
    titleEl.find('.sub-title').remove();
    const full = H.text(titleEl) ?? '';
    const lotNo = full.match(/^Lot\s+([A-Z]?\d+[A-Z]?)\s*-\s*/i)?.[1] ?? null;
    const title = full.replace(/^Lot\s+[A-Z]?\d+[A-Z]?\s*-\s*/i, '').trim();
    if (!title) return;
    const soldText = H.text($(el).find('strong').filter((_, s) => /^Sold/i.test($(s).text().trim())).first());
    const priceTxt = soldText?.match(/Sold\s*£\s*([\d,]+(?:\.\d+)?)/i)?.[1] ?? null;
    const img = $(el).find('.auction-lot-image img').attr('src') ?? null;
    out.push({ lotId, url: `${BASE}${href.split('&so=')[0]!.replace(/&amp;/g, '&')}`, lotNo, title, subtitle, soldText, priceGbp: priceTxt ? Number(priceTxt.replace(/,/g, '')) : null, image: img ? img.replace(/\?v=.*$/, '') : null });
  });
  return out;
}

export function categoryFor(auction: Auction, title: string): { categorySlug: string; vehicle: boolean } {
  const sale = (auction.saleNumber ?? '').toUpperCase();
  if (/registration|number plate|cherished/i.test(title) || /^[A-Z]{1,2}\d*\s*[A-Z]{0,3}$/.test(title.replace(/\s+/g, ' ').trim()) && sale.startsWith('W')) return { categorySlug: 'license_plates', vehicle: false };
  if (sale.startsWith('W') && !/^(19|20)\d{2}\b/.test(title)) return { categorySlug: 'automotive_memorabilia', vehicle: false };
  return { categorySlug: 'automobiles', vehicle: true };
}

export class HistoricsConnector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  protected override minIntervalMs = 10_000; // robots.txt crawl-delay: 10

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const perRun = Number(this.meta.config.auctionsPerRun ?? 1);
    const pagesPerRun = Number(this.meta.config.pagesPerRun ?? 3);
    const progress = { ...((ctx.options.cursor?.progress as Record<string, number> | undefined) ?? {}) };
    const complete = new Set<string>((ctx.options.cursor?.complete as string[] | undefined) ?? []);
    await this.throttle();
    const list = await ctx.fetch(`${BASE}/auction-results`, { engines: ['api'], responseType: 'text', expect: ['title', 'date'], parse: (r) => (r.html ? { title: parseResultsList(r.html)[0]?.title, date: parseResultsList(r.html)[0]?.endedOn } : null) });
    if (!list.success || !list.html) {
      ctx.anomaly('page_fetch_failed', `results list: ${list.error ?? list.httpStatus}`);
      return;
    }
    const auctions = parseResultsList(list.html).filter((a) => a.endedOn && new Date(a.endedOn).getTime() < Date.now());
    const todo = auctions.filter((a) => !complete.has(a.au)).slice(0, perRun);
    let pagesFetched = 0;
    let count = 0;
    for (const auction of todo) {
      let page = (progress[auction.au] ?? 0) + 1;
      while (true) {
        if (ctx.signal?.aborted || this.reached(ctx, count) || pagesFetched >= pagesPerRun) return void (await ctx.setCursor({ progress, complete: [...complete] }));
        const url = `${BASE}/auction/details/${auction.slug}?au=${auction.au}&pp=96&pn=${page}`;
        await this.throttle();
        const res = await ctx.fetch(url, {
          engines: ['api'],
          responseType: 'text',
          expect: ['title', 'price', 'currency', 'status'],
          parse: (r) => {
            const lots = r.html ? parseLotsPage(r.html) : [];
            const sold = lots.find((l) => l.priceGbp);
            return lots.length ? { title: lots[0]!.title, price: sold?.priceGbp ?? null, currency: sold ? 'GBP' : null, status: sold ? 'sold' : null } : null;
          },
        });
        pagesFetched++;
        const lots = res.success && res.html ? parseLotsPage(res.html) : [];
        if (!lots.length) {
          if (!res.success) ctx.anomaly('page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
          complete.add(auction.au);
          break;
        }
        count++;
        yield { url, externalId: `auction:${auction.au}:page:${page}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload: { kind: 'lots_page' as const, url, auction, page, lots }, fetchedAt: res.fetchedAt };
        progress[auction.au] = page;
        if (lots.length < 96) {
          complete.add(auction.au);
          break;
        }
        page++;
      }
    }
    await ctx.setCursor({ progress, complete: [...complete] });
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PagePayloadSchema.parse(raw.payload);
    if (!p.auction.endedOn) return [];
    const saleDate = new Date(p.auction.endedOn);
    const out: NormalizedRecord[] = [];
    for (const lot of p.lots) {
      if (!lot.priceGbp || lot.priceGbp <= 0) continue;
      const m = money(`£${lot.priceGbp}`, 'GBP');
      if (!m) continue;
      const cat = categoryFor(p.auction, lot.title);
      const meta = { auction: p.auction.title, sale_number: p.auction.saleNumber, subtitle: lot.subtitle, sold_text: lot.soldText };
      const attributes = cat.vehicle ? vehicleAttributes(lot.title, { country: 'GB', identifiers: { historics_lot: lot.lotId }, metadata: meta }) : lotAttributes({ categorySlug: cat.categorySlug, name: lot.title, country: 'GB', identifiers: { historics_lot: lot.lotId }, metadata: meta });
      out.push(
        makeSale({
          meta: this.meta,
          sourceUrl: lot.url,
          externalId: lot.lotId,
          rawTitle: lot.title,
          attributes,
          price: m.amount,
          currency: 'GBP',
          saleDate,
          buyerPremiumIncluded: false,
          auctionHouse: 'Historics Auctioneers',
          lotNumber: lot.lotNo,
          imageUrls: lot.image ? [lot.image] : [],
          description: lot.subtitle,
          location: 'United Kingdom',
          observedAt: raw.fetchedAt,
          parserVersion: PARSER_VERSION,
        }),
      );
    }
    return out;
  }
}

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