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 { dateMDY, makeSale, md, money, splitMarkdownItems, vehicleAttributes } from '../_carlib/index.js';

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

export const ItemSchema = z.object({
  id: z.string(),
  url: z.string(),
  title: z.string(),
  statusText: z.string(),
  ended: z.string().nullable(),
  subtitle: z.string().nullable(),
  image: z.string().nullable(),
});
export const PagePayloadSchema = z.object({ kind: z.literal('past_page'), page: z.number(), items: z.array(ItemSchema) });
export type PagePayload = z.infer<typeof PagePayloadSchema>;

/** Parse the Firecrawl markdown of /past-auctions/ into compact items. */
export function parsePastPage(markdown: string, page: number): PagePayload {
  const chunks = splitMarkdownItems(markdown, /^- \[!\[/m);
  const items: z.infer<typeof ItemSchema>[] = [];
  for (const c of chunks) {
    const link = md.link(c, /https:\/\/carsandbids\.com\/auctions\/([A-Za-z0-9]+)\/[a-z0-9-]+/);
    if (!link) continue;
    const id = link.href.match(/\/auctions\/([A-Za-z0-9]+)\//)?.[1];
    if (!id) continue;
    const title = c.match(/\]\(https:\/\/carsandbids\.com\/auctions\/[A-Za-z0-9]+\/[a-z0-9-]+\s+"([^"]+)"\)/)?.[1] ?? link.text;
    const statusText = c.match(/-\s*((?:Sold for|Bid to)\s+[^\]\n]+)\]/)?.[1]?.trim() ?? '';
    const ended = c.match(/Ended\s+(\d{1,2}\/\d{1,2}\/\d{2,4})/)?.[1] ?? null;
    const lines = c.split('\n').map((l) => l.trim()).filter(Boolean);
    const titleLineIdx = lines.findIndex((l) => l.startsWith(`[${title}](`) || l.startsWith(`[${title.replace(/"/g, '')}]`));
    const subtitle = titleLineIdx >= 0 ? lines.slice(titleLineIdx + 1).find((l) => !/^Ended|^Featured|^Watch|^!\[|^- \[/.test(l) && !l.startsWith('[')) ?? null : null;
    if (!statusText) continue;
    items.push({ id, url: link.href, title: md.clean(title), statusText, ended, subtitle: subtitle ? md.clean(subtitle) : null, image: md.image(c) });
  }
  return { kind: 'past_page', page, items };
}

export class CarsAndBidsConnector 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?.newestEnded === 'string' ? new Date(ctx.options.cursor.newestEnded as string) : null;
    let count = 0;
    let maxEnded: Date | null = newest;
    for (let page = start; page < start + pages; page++) {
      if (ctx.signal?.aborted || this.reached(ctx, count)) break;
      const url = `${BASE}/past-auctions/${page > 1 ? `?page=${page}` : ''}`;
      await this.throttle();
      const res = await ctx.fetch(url, {
        expect: ['title', 'price', 'date', 'status'],
        parse: (r) => {
          const p = r.markdown ? parsePastPage(r.markdown, page) : null;
          const f = p?.items[0];
          return f ? { title: f.title, price: money(f.statusText, 'USD')?.amount ?? null, date: f.ended, status: f.statusText } : null;
        },
      });
      if (!res.success || !res.markdown) {
        ctx.anomaly('page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
        break;
      }
      const payload = parsePastPage(res.markdown, page);
      if (payload.items.length === 0) {
        ctx.anomaly('empty_page', url);
        break;
      }
      count++;
      yield { url, externalId: `past:${page}:${payload.items[0]!.id}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
      const dates = payload.items.map((i) => dateMDY(i.ended)).filter((d): d is Date => Boolean(d));
      for (const d of dates) if (!maxEnded || d > maxEnded) maxEnded = 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 && maxEnded) await ctx.setCursor({ newestEnded: maxEnded.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) {
      if (!/^Sold for/i.test(it.statusText)) continue;
      const m = money(it.statusText, 'USD');
      const saleDate = dateMDY(it.ended);
      if (!m || !saleDate) continue;
      const attributes = vehicleAttributes(it.title, { identifiers: { carsandbids_id: it.id }, metadata: { highlights: it.subtitle } });
      out.push(makeSale({ meta: this.meta, sourceUrl: it.url, externalId: it.id, rawTitle: it.title, attributes, price: m.amount, currency: m.currency, saleDate, buyerPremiumIncluded: false, auctionHouse: 'Cars & Bids', imageUrls: it.image ? [it.image] : [], description: it.subtitle, observedAt: raw.fetchedAt, parserVersion: PARSER_VERSION }));
    }
    return out;
  }
}

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