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 { hintFromLabel, slugFromTitle } from '../_auction-lib/categories.js';
import { amount, decodeEntities, houseCategory, isBundleTitle, lotAttributes, makeSale, parseUsDate, safeYear, saleGrade } from '../_g7-auctions-na-lib/index.js';

const SITE = 'https://www.doyle.com';
const HOUSE = 'Doyle';
const PARSER_VERSION = '1.0.0';
const PAGE_SIZE = 96;

export const AuctionSchema = z.object({ auId: z.string(), code: z.string().nullable(), name: z.string(), dateText: z.string().nullable(), url: z.string() });
export const LotSchema = z.object({
  lotId: z.string(),
  lotNumber: z.string().nullable(),
  title: z.string(),
  url: z.string(),
  image: z.string().nullable(),
  soldText: z.string().nullable(),
  soldPrice: z.number().nullable(),
  estimateText: z.string().nullable(),
  estimateLow: z.number().nullable(),
  estimateHigh: z.number().nullable(),
  categoryId: z.string().nullable(),
});
export const PayloadSchema = z.object({ kind: z.literal('doyle_results_page'), auction: AuctionSchema, page: z.number(), hasNext: z.boolean(), lots: z.array(LotSchema) });
export type Payload = z.infer<typeof PayloadSchema>;
export type DoyleAuction = z.infer<typeof AuctionSchema>;

/** /past-auctions/ → calendar items (newest first, as displayed). */
export function parsePastAuctions(htmlText: string): DoyleAuction[] {
  const $ = H.load(htmlText);
  const out: DoyleAuction[] = [];
  const seen = new Set<string>();
  $('.auction-calendar-item').each((_, el) => {
    const item = $(el);
    const href = item.find('a[href*="/auction/"][href*="au="]').first().attr('href');
    const auId = href?.match(/[?&]au=(\d+)/)?.[1];
    if (!href || !auId || seen.has(auId)) return;
    seen.add(auId);
    const name = H.text(item.find('h3, H3').first()) ?? '';
    // Live sales print "Date: …"; online-only sales print "Ends: …" (the close date = sale date).
    const dateText = H.text(item.find('strong').filter((_, s) => /^\s*(Date|Ends?):/i.test($(s).text())).first())?.replace(/^(Date|Ends?):\s*/i, '') ?? null;
    const code = href.match(/\/auction\/([0-9a-z]+)-/i)?.[1]?.toUpperCase() ?? null;
    out.push({ auId, code, name: decodeEntities(name), dateText, url: `${SITE}${href.split('#')[0]}` });
  });
  return out;
}

/** "Mon d, yyyy hh:mm EST" header on an auction page (fallback when the calendar date is missing, e.g. seeds). */
export function parseAuctionHeaderDate(htmlText: string): string | null {
  const text = htmlText.replace(/<script[\s\S]*?<\/script>|<style[\s\S]*?<\/style>/g, ' ').replace(/<[^>]+>/g, ' ').replace(/\s+/g, ' ');
  return text.match(/\b((?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]* \d{1,2}, 20\d{2}(?: \d{1,2}:\d{2} [A-Z]{2,4})?)\b/)?.[1] ?? null;
}

function lotUrl(href: string): string {
  const u = new URL(href.replace(/&amp;/g, '&'), SITE);
  const lot = u.searchParams.get('lot');
  const au = u.searchParams.get('au');
  const out = new URL(u.pathname, SITE);
  if (lot) out.searchParams.set('lot', lot);
  if (au) out.searchParams.set('au', au);
  return out.toString();
}

/** One results page (/auction/search/?au=…&pp=96&pn=N&g=1) → lot cards. */
export function parseResultsPage(htmlText: string, auction: DoyleAuction, page: number): Payload {
  const $ = H.load(htmlText);
  const lots: Payload['lots'] = [];
  const seen = new Set<string>();
  $('div.auction-lot').each((_, el) => {
    const card = $(el);
    const link = card.find('p.auction-lot-title a').first();
    const href = link.attr('href') ?? card.find('a[href*="/auction/lot/"]').first().attr('href');
    const lotId = href?.match(/[?&]lot=(\d+)/)?.[1];
    if (!href || !lotId || seen.has(lotId)) return;
    seen.add(lotId);
    const span = card.find('span.lot-title').first();
    const parts = (span.html() ?? '').split(/<br\s*\/?>/i);
    const lotNumber = H.text(H.load(parts[0] ?? '')('body'))?.replace(/^Lot\s*/i, '').trim() || null;
    const title = decodeEntities(H.text(H.load(parts.slice(1).join(' '))('body')) ?? '') || decodeEntities(H.text(span) ?? '');
    if (!title) return;
    const categoryId = (span.attr('class') ?? '').match(/\bcat-(\d+)/)?.[1] ?? null;
    let soldText: string | null = null;
    let estimateText: string | null = null;
    card.find('strong').each((_, s) => {
      const t = H.text($(s)) ?? '';
      if (/^Sold for/i.test(t)) soldText = t;
      else if (/^Estimate/i.test(t)) estimateText = t;
    });
    const est = (estimateText ?? '').match(/\$([\d,]+)\s*-\s*\$([\d,]+)/);
    const img = card.find('img[src*="/stock/"]').first().attr('src') ?? null;
    lots.push({
      lotId,
      lotNumber,
      title,
      url: lotUrl(href),
      image: img,
      soldText,
      soldPrice: soldText ? amount((soldText as string).replace(/^Sold for/i, '')) : null,
      estimateText,
      estimateLow: est ? amount(est[1]) : null,
      estimateHigh: est ? amount(est[2]) : null,
      categoryId,
    });
  });
  const hasNext = new RegExp(`[?&](?:amp;)?pn=${page + 1}(?:&|'|"|$)`).test(htmlText);
  return { kind: 'doyle_results_page', auction, page, hasNext, lots };
}

/** Doyle sale name + lot title → taxonomy slug (null when nothing confident). */
export function doyleCategory(saleName: string, title: string): string | null {
  const s = saleName.toLowerCase();
  // Natural history lots appear inside book/decorative sales; decide them before department fallbacks
  // (and before the handbag keyword "clutch" can misfire on "clutch of eggs").
  if (/\b(meteorite|pallasite|chondrite)\b/i.test(title)) return 'meteorites';
  if (/\b(fossil|ammonite|trilobite|dinosaur|sauropod|megalodon|mammoth|mosasaur|petrified|fossilized|coprolite)\b/i.test(title)) return 'fossils';
  if (/\b(mineral specimen|geode|quartz cluster|amethyst|tourmaline|fluorite|azurite|malachite|crystal cluster|agate slice)\b/i.test(title)) return 'minerals';
  if (/couture|handbag|fashion|luxury accessor/.test(s)) return slugFromTitle(title, 'fashion') ?? 'fashion_streetwear';
  if (/book|autograph|map|manuscript|bibliophil|librar|print(ed)? & manuscript/.test(s)) return slugFromTitle(title, 'books') ?? 'books';
  if (/photograph/.test(s)) return slugFromTitle(title, 'photographs') ?? 'photography';
  if (/jewel|gem/.test(s) && !/watch/.test(s)) return slugFromTitle(title, 'jewelry') ?? 'jewelry';
  if (/watch/.test(s)) return slugFromTitle(title, /\b(watch|wristwatch|chronograph|pocket watch)\b/i.test(title) ? 'watches' : 'jewelry') ?? 'jewelry';
  if (/coin|bank ?note|stamp|currency|numismat/.test(s)) return slugFromTitle(title, 'coins') ?? 'coins';
  // Silver-only sales; mixed sales ("… Furniture, Old Master Paintings, Silver") fall through to the title sweep below.
  if (/silver|vertu/.test(s) && !/furniture|painting|decorative|works of art/.test(s)) return slugFromTitle(title, 'silver') ?? 'silver';
  // Mixed sales ("English & Continental Furniture, Old Master Paintings, Silver"): let the title decide, default antiques.
  if (/furniture|decorative|works of art|at home|estate|collects/.test(s) && /painting|silver|art\b/.test(s)) {
    if (/\b(oil on|acrylic|watercolou?r|gouache|lithograph|etching|engraving|screenprint|woodcut|drawing|pastel|bronze|sculpture|mixed media)\b/i.test(title)) return slugFromTitle(title, 'art') ?? 'art';
    return slugFromTitle(title, 'furniture') ?? 'antiques';
  }
  if (/contemporary|post-war|modern art/.test(s)) return slugFromTitle(title, 'contemporary') ?? 'contemporary_art';
  if (/painting|prints|drawing|american art|european art|impressionist|old master|fine art|sculpture|works on paper|artist/.test(s)) return slugFromTitle(title, 'art') ?? 'art';
  if (/design|mid-century|20th century decorative/.test(s)) return slugFromTitle(title, 'design') ?? 'design_furniture';
  if (/asian|chinese|japanese|russian|furniture|decorative|english|continental|at home|estate|collects|interior|american story|americana|works of art|antique|rug|carpet/.test(s)) return slugFromTitle(title, 'furniture') ?? 'antiques';
  const generic = houseCategory(saleName, title, null);
  if (generic) return generic;
  return hintFromLabel(saleName) === 'unknown' ? null : 'antiques';
}

/**
 * Doyle (New York) — auction results. Public server-rendered pages over plain HTTPS: /past-auctions/ (calendar of
 * closed sales with date) and /auction/search/?au=<id>&pp=96&pn=N&g=1 (result cards "Sold for $X", estimate).
 * robots.txt asks for crawl-delay 10, honoured. See meta.json accessNotes.
 */
export class DoyleConnector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  protected override minIntervalMs = 10_000; // robots.txt crawl-delay: 10

  private async html(ctx: CrawlContext, url: string): Promise<{ html: string | null; status: number | null; fetchedAt: Date }> {
    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('page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
      return { html: null, status: res.httpStatus, fetchedAt: res.fetchedAt };
    }
    return { html: res.html, status: res.httpStatus, fetchedAt: res.fetchedAt };
  }

  private async listAuctions(ctx: CrawlContext): Promise<DoyleAuction[]> {
    if (ctx.options.seeds?.length) {
      const out: DoyleAuction[] = [];
      for (const s of ctx.options.seeds) {
        const auId = s.match(/[?&]au=(\d+)/)?.[1];
        if (!auId) continue;
        const code = s.match(/\/auction\/([0-9a-z]+)-/i)?.[1]?.toUpperCase() ?? null;
        out.push({ auId, code, name: '', dateText: null, url: s.startsWith('http') ? s : `${SITE}${s}` });
      }
      return out;
    }
    const r = await this.html(ctx, `${SITE}/past-auctions/`);
    if (!r.html) return [];
    const list = parsePastAuctions(r.html);
    if (!list.length) ctx.anomaly('selector_missing', 'past-auctions: no .auction-calendar-item found');
    return list;
  }

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const auctionsPerRun = ctx.options.mode === 'probe' ? 1 : Number(this.meta.config.auctionsPerRun ?? 2);
    const maxPages = ctx.options.mode === 'probe' ? 1 : Number(this.meta.config.pagesPerAuction ?? 15);
    const done = new Set<string>(Array.isArray(ctx.options.cursor?.doneAuctions) ? (ctx.options.cursor!.doneAuctions as string[]) : []);
    const all = await this.listAuctions(ctx);
    // Incremental: newest first (list order). Backfill: oldest first so progress walks the archive forward.
    const ordered = ctx.options.mode === 'backfill' ? [...all].reverse() : all;
    const pending = ordered.filter((a) => !done.has(a.auId));
    let processed = 0;
    let count = 0;
    let items = 0;
    for (const auction of pending) {
      if (ctx.signal?.aborted || processed >= auctionsPerRun || this.reached(ctx, count)) break;
      if (!auction.dateText) {
        // Seeds carry no calendar date: read the auction page header once.
        const page = await this.html(ctx, auction.url);
        if (page.html) {
          auction.dateText = parseAuctionHeaderDate(page.html);
          if (!auction.name) auction.name = decodeEntities(H.text(H.load(page.html)('h1').first()) ?? H.text(H.load(page.html)('title')) ?? '');
        }
        if (!auction.dateText) ctx.anomaly('missing_auction_date', auction.url);
      }
      let complete = true;
      for (let page = 1; page <= maxPages; page++) {
        if (ctx.signal?.aborted || this.reached(ctx, count)) {
          complete = false;
          break;
        }
        const url = `${SITE}/auction/search/?au=${auction.auId}&pp=${PAGE_SIZE}${page > 1 ? `&pn=${page}` : ''}&g=1`;
        const r = await this.html(ctx, url);
        if (!r.html) {
          complete = false;
          break;
        }
        const payload = parseResultsPage(r.html, auction, page);
        if (payload.lots.length === 0) {
          if (page === 1) ctx.anomaly('parse_failure_page', `${url}: no lot cards`);
          break;
        }
        count++;
        items += payload.lots.length;
        yield { url, externalId: `auction:${auction.auId}:page:${page}`, kind: 'sale', engine: 'api', httpStatus: r.status, payload, fetchedAt: r.fetchedAt };
        if (!payload.hasNext) break;
        if (page === maxPages) complete = false;
      }
      processed++;
      if (complete) done.add(auction.auId);
      const doneList = [...done].slice(-500);
      await ctx.setCursor({ doneAuctions: doneList, updatedAt: new Date().toISOString() });
      await ctx.progress({ page: all.filter((a) => done.has(a.auId)).length, totalPages: all.length, itemsProcessed: items, cursor: { doneAuctions: doneList } });
    }
    if (ctx.options.mode === 'backfill' && all.length && all.every((a) => done.has(a.auId))) await ctx.setCursor({ doneAuctions: [...done].slice(-500), done: true, updatedAt: new Date().toISOString() });
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PayloadSchema.parse(raw.payload);
    const saleDate = parseUsDate(p.auction.dateText);
    if (!saleDate || saleDate.getTime() > Date.now() + 86_400_000) return [];
    const out: NormalizedRecord[] = [];
    for (const l of p.lots) {
      if (!l.soldPrice) continue; // unsold / withdrawn / passed: no realised price
      const slug = doyleCategory(p.auction.name, l.title);
      if (!slug) continue;
      const g = saleGrade(l.title);
      const attributes = lotAttributes({
        categorySlug: slug,
        name: l.title,
        year: safeYear(l.title),
        identifiers: { doyle_lot_id: l.lotId },
        metadata: { auction_id: p.auction.auId, auction_code: p.auction.code, auction_name: p.auction.name, estimate_low: l.estimateLow, estimate_high: l.estimateHigh, doyle_category_id: l.categoryId },
      });
      out.push(
        makeSale({
          meta: this.meta,
          sourceUrl: l.url,
          externalId: l.lotId,
          rawTitle: l.title,
          attributes,
          price: l.soldPrice,
          currency: 'USD',
          saleDate,
          // Doyle lot pages print "Includes Buyer's Premium" under "Sold for".
          buyerPremiumIncluded: true,
          auctionHouse: HOUSE,
          lotNumber: l.lotNumber,
          imageUrls: l.image ? [l.image] : [],
          location: 'US',
          observedAt: raw.fetchedAt,
          parserVersion: PARSER_VERSION,
          grader: g.grader,
          grade: g.grade,
          isBundle: isBundleTitle(l.title),
          confidence: 0.85,
        }),
      );
    }
    return out;
  }
}

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