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, parseCardGrade, psaCategorySlug, titleCase, toInt, yearOf } from '../_g2-grading-lib/text.js';

/**
 * PSA certificate verification — https://www.psacard.com/cert/<cert> (public; no login).
 * The page is a Next.js app-router page behind a Cloudflare managed challenge: plain HTTP gets a
 * 403 "Just a moment…" page, Firecrawl (and Scrapfly as fallback) render the same public page a
 * visitor sees. PSA APR and the population report proper require a Collectors login and are never
 * fetched; the per-grade "PSA Population / Pop Higher" figures shown on the cert page itself are used.
 */
const PARSER_VERSION = '1.0.0';
const PATTERNS = [/^https?:\/\/(?:www\.)?psacard\.com\/(?:[a-z]{2}-[a-z]{2}\/)?cert\/(\d{5,12})(?:\/[a-z]+)?\/?(?:[?#].*)?$/i];

/** Canonical public verify URL (matches data/taxonomy/graders.json verifyUrl). */
export function certUrl(cert: string): string {
  return `https://www.psacard.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 PsaPage {
  title: string;
  fields: Record<string, string>;
  population: number | null;
  popHigher: number | null;
  populationUrl: string | null;
  estimate: number | null;
  estimateText: string | null;
  specId: string | null;
  images: string[];
}

/** Pure parser over the trimmed `<main>` fragment. Exported for tests. */
export function parsePsaPage(snapshot: string): PsaPage | null {
  const $ = H.load(snapshot);
  const title = clean($('h1').first().text());
  const fields: Record<string, string> = {};
  $('dl dt').each((_, dt) => {
    const label = clean($(dt).text());
    const dd = $(dt).next('dd');
    if (label && dd.length) fields[label] = clean(dd.text());
  });
  if (!fields['Cert Number'] && !title) return null;
  const stat = (label: string): { text: string; href: string | null } | null => {
    const p = $('p')
      .toArray()
      .find((e) => clean($(e).text()).toLowerCase() === label.toLowerCase());
    if (!p) return null;
    const next = $(p).next();
    return { text: clean(next.text()), href: next.attr('href') ?? null };
  };
  const pop = stat('PSA Population');
  const higher = stat('PSA Pop Higher');
  const est = stat('PSA Estimate');
  const specHref = $('a[href*="/spec/psa/"]').first().attr('href') ?? null;
  const specId = specHref?.match(/\/spec\/psa\/(\d+)/)?.[1] ?? null;
  const images = [...new Set($('img[alt^="Cert image"]').toArray().map((e) => $(e).attr('src') ?? '').filter((u) => /^https?:\/\//.test(u)))];
  return { title, fields, population: toInt(pop?.text), popHigher: toInt(higher?.text), populationUrl: pop?.href ?? null, estimate: money(est?.text), estimateText: est?.text ?? null, specId, images };
}

export class PsaCertConnector extends CertLookupConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  readonly grader = 'psa';
  readonly idKey = 'psa_cert';
  readonly format = 'html' as const;
  override readonly urlPatterns = PATTERNS;
  protected override waitForMs = 4000;

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

  classify(doc: string): PageStatus {
    const $ = H.load(doc);
    const hasCert = $('dl dt')
      .toArray()
      .some((dt) => /^cert number$/i.test(clean($(dt).text())));
    if (hasCert && $('h1').length) return 'found';
    const text = clean($('main').text() || $('body').text());
    if (/(could not|couldn.t|cannot|unable to) (find|locate)|not (a )?valid|no (results|record)|does not exist|not found|invalid cert|private set registry/i.test(text)) return 'not_found';
    return 'unknown';
  }

  trim(doc: string): string {
    const $ = H.load(doc);
    $('script, style, noscript, iframe, svg, table, [role="dialog"], header, footer, nav').remove();
    const main = $('main').first();
    return main.length ? $.html(main) : $.html($('body'));
  }

  parse(snapshot: string, info: { cert: string; url: string }): ParsedCert | null {
    const page = parsePsaPage(snapshot);
    if (!page) return null;
    const f = page.fields;
    const cert = f['Cert Number'] ?? info.cert;
    const g = parseCardGrade(f['Item Grade']);
    const brand = f['Brand/Title'] ?? null;
    const subject = f['Subject'] ?? null;
    const cat = psaCategorySlug(f['Category'], brand, subject);
    const text = `${brand ?? ''} ${subject ?? ''} ${page.title}`;
    const isPokemon = cat.slug === 'pokemon';
    const identifiers: Record<string, string> = {};
    if (page.specId) identifiers.psa_spec_id = page.specId;
    const metadata: Record<string, unknown> = {
      psa_category: f['Category'] ?? null,
      label_type: f['Label Type'] ?? null,
      reverse_cert_barcode: f['Reverse Cert/Barcode'] ?? null,
      psa_estimate_usd: page.estimate,
      psa_estimate_text: page.estimateText,
      population_url: page.populationUrl,
      fields: f,
    };
    return {
      title: page.title || [f['Year'], brand, subject].filter(Boolean).join(' '),
      attributes: {
        categorySlug: cat.slug,
        franchise: isPokemon ? 'Pokémon' : null,
        brand: isPokemon ? 'The Pokémon Company' : brand ? titleCase(brand) : null,
        set: brand ? titleCase(brand) : null,
        name: subject ? titleCase(subject) : page.title,
        number: f['Card Number'] ?? null,
        year: yearOf(f['Year']),
        variant: f['Variety/Pedigree'] ?? f['Variety'] ?? null,
        language: /japanese/i.test(text) ? 'Japanese' : isPokemon ? 'English' : null,
        identifiers,
        metadata,
      },
      grade: g.grade,
      qualifier: g.qualifier,
      gradeLabel: g.label || null,
      images: page.images,
      population: page.population !== null ? { gradeKey: g.grade ?? g.label, atGrade: page.population, higher: page.popHigher, url: page.populationUrl } : null,
      confidence: cat.confident ? 0.92 : 0.8,
    };
  }
}

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