import { z } from 'zod';
import { BaseConnector, html as H, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import { parseSourceDate, AssetAttributesSchema, NormalizedSaleSchema, type NormalizedRecord } from '@rareindex/shared';

/**
 * Scotch Whisky Auctions — monthly online whisky auctions (Glasgow) with a public archive of
 * results. One raw record per lot-list page; normalise → one sale per lot with a "Sold for" price.
 */

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

export const AuctionSchema = z.object({ id: z.string(), slug: z.string(), title: z.string(), endedOn: z.string().nullable(), lotCount: z.number().nullable(), ended: z.boolean() });
export type Auction = z.infer<typeof AuctionSchema>;

export const LotSchema = z.object({ itemId: z.string(), url: z.string(), title: z.string(), lotNo: z.string().nullable(), soldText: z.string().nullable(), priceGbp: z.number().nullable(), image: z.string().nullable() });
export const PagePayloadSchema = z.object({ kind: z.literal('lot_page'), url: z.string(), auction: AuctionSchema, page: z.number(), totalPages: z.number().nullable(), lots: z.array(LotSchema) });
export type PagePayload = z.infer<typeof PagePayloadSchema>;

/** Parse /auctions/ : cards "The 181st Auction · Ended July 12, 2026 · There are 4926 lots in this auction". */
export function parseAuctionList(htmlText: string): Auction[] {
  const $ = H.load(htmlText);
  const out: Auction[] = [];
  $('a.auction').each((_, a) => {
    const href = $(a).attr('href') ?? '';
    const m = href.match(/^\/auctions\/(\d+)-([^/]+)\/?$/);
    if (!m) return;
    const title = H.text($(a).find('h4')) ?? '';
    const status = H.text($(a).find('h5')) ?? '';
    const lots = H.text($(a).find('h6'))?.match(/(\d[\d,]*)\s+lots/i)?.[1] ?? null;
    const ended = /^Ended\b/i.test(status);
    const dateTxt = status.replace(/^(Ended|Ends)\s*/i, '');
    const d = parseSourceDate(dateTxt);
    out.push({ id: m[1]!, slug: m[2]!, title, endedOn: d ? d.toISOString() : null, lotCount: lots ? Number(lots.replace(/,/g, '')) : null, ended });
  });
  return out;
}

/** Parse an auction lot-list page (20 lots): title, lot number, "Sold for £100 in July 2026", image. */
export function parseLotPage(htmlText: string, url: string, auction: Auction, page: number): PagePayload {
  const $ = H.load(htmlText);
  const totalPages = Number(htmlText.match(/Page \d+ of (\d+)/)?.[1]) || null;
  const lots: z.infer<typeof LotSchema>[] = [];
  $('a.lot').each((_, a) => {
    const href = $(a).attr('href') ?? '';
    const m = href.match(/\/auctions\/\d+-[^/]+\/(\d+)-[^/]*\/?$/);
    if (!m) return;
    const title = H.text($(a).find('h4'));
    if (!title) return;
    const lotNo = H.text($(a).find('h6'))?.replace(/^Lot no\s*/i, '') ?? null;
    const soldText = H.text($(a).find('p.sold')) ?? null;
    const price = soldText?.match(/Sold for £([\d,]+(?:\.\d+)?)/i)?.[1] ?? null;
    const bg = $(a).find('.aucimg').attr('style') ?? '';
    const image = bg.match(/url\('([^']+)'\)/)?.[1] ?? null;
    lots.push({ itemId: m[1]!, url: BASE + href.replace(/\/?$/, '/'), title, lotNo, soldText, priceGbp: price ? Number(price.replace(/,/g, '')) : null, image });
  });
  return { kind: 'lot_page', url, auction, page, totalPages, lots };
}

const SIZE_RE = /(\d+(?:\.\d+)?)\s?(cl|ml|l|litre|liter)s?\b/i;
const AGE_RE = /(\d{1,2})\s*[- ]?\s*(?:year|yo\b|y\.?o\.?)/i;
const VINTAGE_RE = /\b(19[2-9]\d|20[0-2]\d)\b(?!\s*(?:year|yo))/i;

export interface WhiskyFacts {
  brand: string | null;
  age: number | null;
  vintage: number | null;
  size: string | null;
  categorySlug: 'whisky' | 'rum' | 'cognac';
}

/** Heuristic facts from a lot title ("Macallan 1989 18 Year Old Gran Reserva 70cl"). Unknown → null, never guessed. */
export function whiskyFacts(title: string): WhiskyFacts {
  const t = title.replace(/[‘’]/g, "'");
  const categorySlug: WhiskyFacts['categorySlug'] = /\brum\b|\brhum\b/i.test(t) ? 'rum' : /\bcognac\b|\barmagnac\b/i.test(t) ? 'cognac' : 'whisky';
  const age = t.match(AGE_RE)?.[1] ? Number(t.match(AGE_RE)![1]) : null;
  const vintage = t.match(VINTAGE_RE)?.[1] ? Number(t.match(VINTAGE_RE)![1]) : null;
  const sizeM = t.match(SIZE_RE);
  const size = sizeM ? `${sizeM[1]}${sizeM[2]!.toLowerCase().startsWith('l') ? 'L' : sizeM[2]!.toLowerCase()}` : null;
  // Brand = leading words before the first digit / age / "Year" / "-" ; keep 1–4 words
  const lead = t.split(/\s+(?=\d)|\s+-\s+|\s+\(/)[0] ?? t;
  const words = lead.replace(/^['"]|['"]$/g, '').trim().split(/\s+/).filter(Boolean);
  const brand = words.length ? words.slice(0, Math.min(words.length, 4)).join(' ') : null;
  return { brand: brand && /[a-z]/i.test(brand) ? brand : null, age, vintage, size, categorySlug };
}

export class ScotchWhiskyAuctionsConnector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  override readonly urlPatterns = [/scotchwhiskyauctions\.com\/auctions\/\d+-[^/]+\/\d+-/];
  protected override minIntervalMs = 2000;

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const listUrl = String(this.meta.config.auctionsUrl ?? `${BASE}/auctions/`);
    const auctionsPerRun = Number(this.meta.config.auctionsPerRun ?? 2);
    const pagesPerRun = Number(this.meta.config.pagesPerRun ?? 12);
    const cursor = (ctx.options.cursor ?? {}) as { progress?: Record<string, number>; complete?: string[] };
    const progress: Record<string, number> = { ...(cursor.progress ?? {}) };
    const complete = new Set<string>(cursor.complete ?? []);

    const list = await ctx.fetch(listUrl, { engines: ['api'], responseType: 'text', minQuality: 0 });
    if (!list.success || !list.html) {
      ctx.anomaly('auction_list_failed', list.error ?? String(list.httpStatus));
      return;
    }
    let auctions = parseAuctionList(list.html).filter((a) => a.ended);
    // newest first; incremental = newest auctions not yet complete, backfill = continue with older ones
    auctions.sort((a, b) => Number(b.id) - Number(a.id));
    if (ctx.options.mode !== 'backfill') auctions = auctions.slice(0, 6);
    let count = 0;
    let pagesFetched = 0;
    let auctionsTouched = 0;
    for (const auction of auctions) {
      if (complete.has(auction.id)) continue;
      if (auctionsTouched >= auctionsPerRun) break;
      auctionsTouched++;
      let page = (progress[auction.id] ?? 0) + 1;
      for (;;) {
        if (ctx.signal?.aborted || this.reached(ctx, count) || pagesFetched >= pagesPerRun) return void (await ctx.setCursor({ progress, complete: [...complete] }));
        const url = `${BASE}/auctions/${auction.id}-${auction.slug}/${page > 1 ? `?page=${page}` : ''}`;
        await this.throttle();
        const res = await ctx.fetch(url, {
          engines: ['api'],
          responseType: 'text',
          expect: ['title', 'price', 'currency', 'status'],
          parse: (r) => {
            const p = r.html ? parseLotPage(r.html, url, auction, page) : null;
            const sold = p?.lots.find((l) => l.priceGbp);
            return p && p.lots.length ? { title: p.lots[0]!.title, price: sold?.priceGbp ?? null, currency: sold ? 'GBP' : null, status: sold ? 'sold' : null } : null;
          },
        });
        pagesFetched++;
        const payload = res.success && res.html ? parseLotPage(res.html, url, auction, page) : null;
        if (!payload || payload.lots.length === 0) {
          ctx.anomaly(payload ? 'empty_page' : 'page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
          complete.add(auction.id);
          break;
        }
        count++;
        yield { url, externalId: `auction:${auction.id}:page:${page}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
        progress[auction.id] = page;
        if (payload.totalPages && page >= payload.totalPages) {
          complete.add(auction.id);
          break;
        }
        page++;
      }
      await ctx.setCursor({ progress, complete: [...complete] });
    }
  }

  async lookup(url: string, ctx: CrawlContext): Promise<RawRecordInput[]> {
    const m = url.match(/\/auctions\/(\d+)-([^/]+)\/(\d+)-/);
    if (!m) return [];
    const res = await ctx.fetch(url, { engines: ['api'], responseType: 'text', minQuality: 0 });
    if (!res.success || !res.html) return [];
    const $ = H.load(res.html);
    const title = H.text($('h1').first()) ?? H.text($('h4').first());
    const soldText = res.html.match(/Sold for £[\d,]+(?:\.\d+)?[^<]*/)?.[0] ?? null;
    const lotNo = res.html.match(/Lot no\s*([\d-]+)/i)?.[1] ?? null;
    const ended = res.html.match(/Ended\s+([A-Z][a-z]+ \d{1,2}, \d{4})/)?.[1] ?? null;
    const endedOn = ended ? parseSourceDate(ended)?.toISOString() ?? null : null;
    if (!title) return [];
    const price = soldText?.match(/£([\d,]+(?:\.\d+)?)/)?.[1] ?? null;
    const image = $('meta[property="og:image"]').attr('content') ?? null;
    const auction: Auction = { id: m[1]!, slug: m[2]!, title: '', endedOn, lotCount: null, ended: Boolean(endedOn) };
    const payload: PagePayload = { kind: 'lot_page', url, auction, page: 0, totalPages: null, lots: [{ itemId: m[3]!, url, title, lotNo, soldText, priceGbp: price ? Number(price.replace(/,/g, '')) : null, image }] };
    return [{ url, externalId: `lot:${m[3]}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt }];
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PagePayloadSchema.parse(raw.payload);
    const saleDate = p.auction.endedOn ? new Date(p.auction.endedOn) : null;
    if (!saleDate) return [];
    const out: NormalizedRecord[] = [];
    for (const lot of p.lots) {
      if (!lot.priceGbp || lot.priceGbp <= 0) continue;
      const f = whiskyFacts(lot.title);
      const attributes = AssetAttributesSchema.parse({
        categorySlug: f.categorySlug,
        brand: f.brand,
        name: lot.title,
        year: f.vintage,
        size: f.size,
        country: f.categorySlug === 'whisky' && /scotch|islay|speyside|highland|campbeltown|lowland/i.test(lot.title) ? 'GB' : null,
        identifiers: { swa_item: lot.itemId, ...(lot.lotNo ? { swa_lot: lot.lotNo } : {}) },
        metadata: { age_statement: f.age, auction_id: p.auction.id, auction_title: p.auction.title, sold_text: lot.soldText },
      });
      out.push(
        NormalizedSaleSchema.parse({
          kind: 'sale',
          connectorId: this.meta.id,
          sourceId: this.meta.sourceId,
          sourceUrl: lot.url,
          externalId: lot.itemId,
          rawTitle: lot.title,
          imageUrls: lot.image ? [lot.image] : [],
          attributes,
          grade: { grader: null, grade: null, qualifier: null, certificationNumber: null },
          condition: { condition: null, conditionRaw: null, completeness: null },
          observedAt: raw.fetchedAt,
          confidence: 0.9,
          parserVersion: PARSER_VERSION,
          saleType: 'auction',
          saleDate,
          price: lot.priceGbp,
          currency: 'GBP',
          buyerPremiumIncluded: false,
          quantity: 1,
          isBundle: /\b(x\s?\d|\d+\s?x\s|lot of|set of|\d+\s?bottles)\b/i.test(lot.title),
          location: 'Glasgow, United Kingdom',
          auctionHouse: 'Scotch Whisky Auctions',
          lotNumber: lot.lotNo,
        }),
      );
    }
    return out;
  }
}

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