import { z } from 'zod';
import { BaseConnector, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import type { NormalizedRecord, NormalizedSale } from '@rareindex/shared';
import { dateDMY, makeSale, md, money, splitMarkdownItems, vehicleAttributes } from '../_carlib/index.js';

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

export const ItemSchema = z.object({
  slug: z.string(),
  url: z.string(),
  title: z.string(),
  priceText: z.string(),
  dateText: z.string().nullable(),
  country: z.string().nullable(),
  town: z.string().nullable(),
  image: z.string().nullable(),
});
export const PagePayloadSchema = z.object({ kind: z.literal('sold_page'), page: z.number(), total: z.number().nullable(), items: z.array(ItemSchema) });
export type PagePayload = z.infer<typeof PagePayloadSchema>;

const CURRENCY_LINE = /^(?:£|€|A\$|US\$|CA\$|NZ\$|AED|CHF|kr|\$)\s?[0-9][0-9,.]*/;

export function parseSoldPage(markdown: string, page: number): PagePayload {
  const total = markdown.match(/Showing\s+([\d,]+)\s+lots/i)?.[1];
  const chunks = splitMarkdownItems(markdown, /^- \[!\[/m);
  const items: z.infer<typeof ItemSchema>[] = [];
  for (const c of chunks) {
    const href = c.match(/\((https:\/\/collectingcars\.com\/for-sale\/[a-z0-9-]+)\)/)?.[1];
    if (!href) continue;
    const slug = href.split('/').pop()!;
    const title = c.match(/\*\*([^*]+)\*\*/)?.[1]?.trim() ?? c.match(/^- \[!\[([^\]]+)\]/)?.[1] ?? null;
    if (!title) continue;
    const lines = c.split('\n').map((l) => l.replace(/\\+$/, '').trim()).filter(Boolean);
    const priceText = lines.find((l) => CURRENCY_LINE.test(l)) ?? null;
    if (!priceText) continue; // no price → not sold
    const dateText = lines.find((l) => /^\d{2}\/\d{2}\/\d{4}$/.test(l)) ?? null;
    const flag = c.match(/!\[([^\]]+)\]\(https:\/\/flagcdn\.com\/[a-z]{2}\.svg\)([^\]\n]*)/);
    items.push({ slug, url: href, title: md.clean(title), priceText, dateText, country: flag?.[1]?.trim() ?? null, town: flag?.[2]?.trim() || null, image: md.image(c) });
  }
  return { kind: 'sold_page', page, total: total ? Number(total.replace(/,/g, '')) : null, items };
}

const MEMORABILIA = /\b(watch|helmet|poster|sign|number plate|registration|memorabilia|model|artwork|painting|print|petrol pump|engine|wheel set|wheels|seat|suit|jacket|book|literature|steering wheel|trophy)\b/i;

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

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const pages = Number(this.meta.config.pagesPerRun ?? 10);
    const backfill = ctx.options.mode === 'backfill';
    const start = backfill ? Number(ctx.options.cursor?.nextPage ?? 1) : 1;
    const newest = !backfill && typeof ctx.options.cursor?.newestSold === 'string' ? new Date(ctx.options.cursor.newestSold as string) : null;
    let count = 0;
    let maxSold: Date | null = newest;
    for (let page = start; page < start + pages; page++) {
      if (ctx.signal?.aborted || this.reached(ctx, count)) break;
      const url = `${BASE}/sold${page > 1 ? `?page=${page}` : ''}`;
      await this.throttle();
      const res = await ctx.fetch(url, {
        expect: ['title', 'price', 'date', 'status'],
        parse: (r) => {
          const f = r.markdown ? parseSoldPage(r.markdown, page).items[0] : null;
          return f ? { title: f.title, price: money(f.priceText)?.amount ?? null, date: f.dateText, status: 'sold' } : null;
        },
      });
      if (!res.success || !res.markdown) {
        ctx.anomaly('page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
        break;
      }
      const payload = parseSoldPage(res.markdown, page);
      if (payload.items.length === 0) {
        ctx.anomaly('empty_page', url);
        break;
      }
      count++;
      yield { url, externalId: `sold:${page}:${payload.items[0]!.slug}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
      const dates = payload.items.map((i) => dateDMY(i.dateText)).filter((d): d is Date => Boolean(d));
      for (const d of dates) if (!maxSold || d > maxSold) maxSold = d;
      const oldest = dates.length ? new Date(Math.min(...dates.map((d) => d.getTime()))) : null;
      if (backfill) await ctx.setCursor({ nextPage: page + 1, updatedAt: new Date().toISOString() });
      else if (newest && oldest && oldest < newest) break;
    }
    if (!backfill && maxSold) await ctx.setCursor({ newestSold: maxSold.toISOString(), updatedAt: new Date().toISOString() });
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PagePayloadSchema.parse(raw.payload);
    const out: NormalizedSale[] = [];
    for (const it of p.items) {
      const m = money(it.priceText, 'GBP');
      const saleDate = dateDMY(it.dateText);
      if (!m || !saleDate) continue;
      const memorabilia = MEMORABILIA.test(it.title) && !/^\d{4}\s/.test(it.title);
      const attributes = vehicleAttributes(it.title, { country: it.country, identifiers: { collectingcars_slug: it.slug }, metadata: { town: it.town }, ...(memorabilia ? { categorySlug: 'automotive_memorabilia' as const } : {}) });
      out.push(makeSale({ meta: this.meta, sourceUrl: it.url, externalId: it.slug, rawTitle: it.title, attributes, price: m.amount, currency: m.currency, saleDate, buyerPremiumIncluded: false, auctionHouse: 'Collecting Cars', imageUrls: it.image ? [it.image.replace(/\?.*$/, '')] : [], location: [it.town, it.country].filter(Boolean).join(', ') || null, observedAt: raw.fetchedAt, parserVersion: PARSER_VERSION }));
    }
    return out;
  }
}

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