import { html as H, type ConnectorMeta } from '@rareindex/connectors';
import { CertLookupConnector, type PageStatus, type ParsedCert } from '../_g2-grading-lib/cert-base.js';
import { clean, money, parseCoinGrade, toInt, yearOf } from '../_g2-grading-lib/text.js';

/**
 * PCGS certificate verification — https://www.pcgs.com/cert/<cert> (public; coins and banknotes).
 * Svelte-rendered page behind a Cloudflare managed challenge → Firecrawl render (Scrapfly fallback).
 * The page carries the coin's PCGS# (deterministic id shared with connectors/firecrawl/pcgs-priceguide),
 * grade, price-guide value and the PCGS population at that grade + higher.
 */
const PARSER_VERSION = '1.0.0';
const PATTERNS = [/^https?:\/\/(?:www\.)?pcgs\.com\/cert\/(\d{6,10})\/?(?:[?#].*)?$/i, /^https?:\/\/(?:www\.)?pcgs\.com\/cert\/verify\?(?:.*&)?certno=(\d{6,10})/i];

export function certUrl(cert: string): string {
  return `https://www.pcgs.com/cert/${clean(cert).replace(/\D/g, '')}`;
}

export function certFromUrl(url: string): string | null {
  for (const re of PATTERNS) {
    const m = url.match(re);
    if (m) return m[1]!;
  }
  return null;
}

export interface PcgsPage {
  title: string;
  cert: string | null;
  rows: Record<string, string>;
  links: Record<string, string>;
  images: string[];
}

/** Pure parser over the trimmed fragment. Exported for tests. */
export function parsePcgsPage(snapshot: string): PcgsPage | null {
  const $ = H.load(snapshot);
  const title = clean($('.text-display5').first().text());
  const cert = clean($('.text-display5').first().parent().text()).match(/#(\d{6,10})/)?.[1] ?? clean($.root().text()).match(/#(\d{6,10})/)?.[1] ?? null;
  const rows: Record<string, string> = {};
  const links: Record<string, string> = {};
  $('table tr').each((_, tr) => {
    const tds = $(tr).find('td');
    if (tds.length < 2) return;
    const label = clean($(tds[0]).text());
    if (!label) return;
    rows[label] = clean($(tds[1]).text());
    const href = $(tds[1]).find('a[href]').first().attr('href');
    if (href) links[label] = href;
  });
  if (!title || !rows['Grade']) return null;
  const images = [...new Set($('img[alt*="enlarge" i], img[src*="/pcgs/cert/"]').toArray().map((e) => $(e).attr('src') ?? '').filter((u) => /^https?:\/\//.test(u)))];
  return { title, cert, rows, links, images };
}

export class PcgsCertConnector extends CertLookupConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  readonly grader = 'pcgs';
  readonly idKey = 'pcgs_cert';
  readonly format = 'html' as const;
  override readonly urlPatterns = PATTERNS;

  certUrl(cert: string): string {
    return certUrl(cert);
  }
  certFromUrl(url: string): string | null {
    return certFromUrl(url);
  }

  classify(doc: string): PageStatus {
    const $ = H.load(doc);
    const hasGradeRow = $('table tr td')
      .toArray()
      .some((td) => /^grade$/i.test(clean($(td).text())));
    if ($('.text-display5').length && hasGradeRow) return 'found';
    const text = clean($('body').text() || $.root().text());
    if (/not found|no record|invalid cert|does not exist|could not be found|unable to (find|locate)|no results|not in our|we were unable|isn.t in the pcgs/i.test(text)) return 'not_found';
    return 'unknown';
  }

  trim(doc: string): string {
    const $ = H.load(doc);
    $('script, style, noscript, iframe, svg, header, footer, nav, form').remove();
    let node = $('.text-display5').first();
    if (!node.length) {
      const body = $('body');
      return `<div class="g2-trimmed">${clean(body.text()).slice(0, 4000)}</div>`;
    }
    for (let i = 0; i < 10 && node.length; i++) {
      if (node.find('table').length) break;
      node = node.parent();
    }
    return $.html(node.length ? node : $('body'));
  }

  parse(snapshot: string, info: { cert: string; url: string }): ParsedCert | null {
    const page = parsePcgsPage(snapshot);
    if (!page) return null;
    const r = page.rows;
    const cert = page.cert ?? info.cert;
    const g = parseCoinGrade(r['Grade']);
    const pcgsNo = clean(r['PCGS #'] ?? '').replace(/\D/g, '') || null;
    const region = r['Region'] ?? null;
    const isNote = /note|banknote|currency/i.test(`${page.title} ${r['Denomination'] ?? ''} ${r['Holder Type'] ?? ''}`) && !/coin/i.test(r['Holder Type'] ?? '');
    const identifiers: Record<string, string> = {};
    if (pcgsNo) identifiers.pcgs_number = pcgsNo;
    const population = toInt(r['Population']);
    return {
      title: `${page.title} PCGS ${r['Grade']}`.trim(),
      attributes: {
        categorySlug: isNote ? 'banknotes' : 'coins',
        name: page.title,
        year: yearOf(r['Date, Mintmark'] ?? page.title),
        country: /united states/i.test(region ?? '') ? 'US' : null,
        region,
        variant: r['Variety'] ?? r['Designation'] ?? null,
        productionQuantity: toInt(r['Mintage']),
        identifiers,
        metadata: {
          pcgs_cert: cert,
          date_mintmark: r['Date, Mintmark'] ?? null,
          denomination: r['Denomination'] ?? null,
          price_guide_value_usd: money(r['Price Guide Value']),
          holder_type: r['Holder Type'] ?? null,
          security: r['Security'] ?? null,
          coinfacts_url: page.links['PCGS #'] ?? null,
          population_url: page.links['Population'] ?? null,
          fields: r,
        },
      },
      grade: g.grade,
      qualifier: g.qualifier,
      gradeLabel: g.label || null,
      images: page.images,
      population: population !== null ? { gradeKey: g.grade ?? g.label, atGrade: population, higher: toInt(r['Pop Higher']), url: page.links['Population'] ?? null } : null,
      confidence: pcgsNo ? 0.95 : 0.85,
    };
  }
}

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