import { z } from 'zod';
import { BaseConnector, html as H, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import { AssetAttributesSchema, NormalizedListingSchema, type NormalizedRecord } from '@rareindex/shared';
import { conditionFromParens, moneyNum, normalizeConditionWord, readSeedCursor, yearOrDecade } from '../_g10-lib/index.js';

/**
 * Peyton Street Pens (Santa Cruz, CA) — vintage & modern fountain-pen dealer on BigCommerce. Category
 * pages (/<category>/?page=N, 24 cards) are server-rendered: each <li class="product"> card carries the
 * product URL, title, SKU, brand, price ($, USD) and CDN image. Dealer asking prices → listings.
 */
const BASE = 'https://www.peytonstreetpens.com';
const PARSER_VERSION = '1.0.0';
const PAGE_SIZE = 24;

export const SeedSchema = z.object({ path: z.string(), brand: z.string().nullable().optional() });
export type Seed = z.infer<typeof SeedSchema>;

export const CardSchema = z.object({ productId: z.string().nullable(), sku: z.string().nullable(), url: z.string(), title: z.string(), brand: z.string().nullable(), price: z.number().nullable(), rrp: z.number().nullable(), image: z.string().nullable(), soldOut: z.boolean().default(false) });
export type Card = z.infer<typeof CardSchema>;

export const PagePayloadSchema = z.object({ kind: z.literal('listing_page'), url: z.string(), seed: SeedSchema, page: z.number().int(), totalPages: z.number().int().nullable(), cards: z.array(CardSchema), snapshot: z.string().optional() });
export type PagePayload = z.infer<typeof PagePayloadSchema>;

const ConfigSchema = z.object({
  seeds: z.array(SeedSchema).min(1),
  pagesPerSeed: z.number().int().min(1).default(2),
  seedsPerRun: z.number().int().min(1).default(4),
  /** titles matching this are accessories/consumables, not pens */
  exclude: z.string().default('\\b(ink|inks|inkwell|ink well|converter|refills?|leads|notebook|paper|pen case|pen wrap|pouch|box only|parts|sac|o-ring|gift certificate|book|poster|t-shirt|stand|display)\\b'),
});

/** BigCommerce "brand" on this shop is often the product type (PEN SET, BALLPOINT…) rather than a maker. */
const GENERIC_BRANDS = /^(pen set|pen sets|ballpoint|rollerball|fountain pen|mechanical pencil|pencil|pens?|desk set|dip pen|other|misc|unknown|various|n\/a|none|vintage|new|used)$/i;
const PEN_TYPE = /\b(fountain pen|ballpoint pen|ballpoint|rollerball|roller ball|mechanical pencil|pencil|pen & pencil set|pen and pencil set|pen set|desk set|dip pen|felt tip|fineliner|pen\b)/i;

export function parseCategoryHtml(htmlText: string): { cards: Card[]; totalPages: number | null } {
  const $ = H.load(htmlText);
  const cards: Card[] = [];
  $('li.product').each((_, el) => {
    const e = $(el);
    const a = e.find('.card-title a').first();
    const url = a.attr('href');
    const title = H.text(a);
    if (!url || !title) return;
    const priceText = H.text(e.find('.price--withoutTax').filter((_, p) => !$(p).hasClass('price--rrp')).first());
    const rrpText = H.text(e.find('.price--rrp').first());
    const img = e.find('img.card-image').first();
    const image = img.attr('data-src') ?? (img.attr('src')?.includes('/products/') ? img.attr('src')! : null);
    cards.push({
      productId: e.find('[data-product-id]').first().attr('data-product-id') ?? null,
      sku: H.text(e.find('.sku-value').first()),
      url: url.startsWith('http') ? url : `${BASE}${url}`,
      title,
      brand: H.text(e.find('.card-text.brand a').first()),
      price: moneyNum(priceText),
      rrp: moneyNum(rrpText),
      image: image ? image.replace(/\/stencil\/\d+x\d+\//, '/stencil/1280x1280/') : null,
      soldOut: /\b(sold out|sold)\b/i.test(e.find('.card-figcaption, .sale-flag, .card-badge, .card-text').text()) || /\bSOLD\b/.test(title),
    });
  });
  const pages = $('a[href*="?page="]').map((_, a) => Number(($(a).attr('href') ?? '').match(/[?&]page=(\d+)/)?.[1] ?? 0)).get().filter((n) => n > 0);
  return { cards, totalPages: pages.length ? Math.max(...pages) : cards.length ? 1 : null };
}

/** First `n` product cards + pagination anchors (fixture snapshot). */
export function trimCategoryHtml(htmlText: string, n = 3): string {
  const $ = H.load(htmlText);
  const cards = $('li.product').toArray().slice(0, n).map((el) => $.html(el));
  const pages = [...new Set($('a[href*="?page="]').map((_, a) => $(a).attr('href') ?? '').get())].map((h) => `<a href="${h}">p</a>`).join('');
  return `<!doctype html><html><body><ul class="productGrid">${cards.join('\n')}</ul><nav class="pagination">${pages}</nav></body></html>`;
}

export function pageUrl(seedPath: string, page: number): string {
  const p = seedPath.endsWith('/') ? seedPath : `${seedPath}/`;
  return `${BASE}${p}${page > 1 ? `?page=${page}` : ''}`;
}

/** "Sheaffer Craftsman Fountain Pen & Pencil Set (1950s) - Burgundy w/GT, …" → brand/model/type. */
export function parsePenTitle(title: string, cardBrand: string | null): { brand: string | null; model: string | null; penType: string | null; variant: string | null } {
  const head = title.split(/\s+-\s+|\s*\(/)[0]?.trim() ?? title;
  const typeM = head.match(PEN_TYPE);
  const penType = typeM ? typeM[0].replace(/\s+/g, ' ').toLowerCase() : null;
  const beforeType = typeM ? head.slice(0, typeM.index).trim() : head;
  let brand = cardBrand && !GENERIC_BRANDS.test(cardBrand) ? cardBrand.replace(/\s+/g, ' ').trim() : null;
  if (brand && /^[A-Z0-9 &'.-]+$/.test(brand) && brand.length > 3) brand = brand.split(' ').map((w) => (w.length > 2 ? w[0] + w.slice(1).toLowerCase() : w)).join(' ');
  const words = beforeType.split(/\s+/).filter(Boolean);
  if (!brand && words.length) brand = /^(montblanc|mont|parker|sheaffer|waterman|wahl|eversharp|pelikan|pilot|namiki|sailor|platinum|nakaya|aurora|omas|visconti|montegrappa|conklin|esterbrook|conway|lamy|kaweco|cross|dunhill|cartier|tiffany|delta|stipula|ranga|lotus|leonardo|scribo|faber-castell|graf|caran|twsbi|danitrio|s\.t\.|st\.)$/i.test(words[0]!) ? (words[0] === 'Conway' || words[0] === 'Wahl' || /^(mont|caran|graf|s\.t\.|st\.)$/i.test(words[0]!) ? words.slice(0, 2).join(' ') : words[0]!) : words[0]!;
  let model: string | null = null;
  if (brand) {
    const rest = beforeType.replace(new RegExp(`^${brand.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}\\s*`, 'i'), '').trim();
    model = rest && rest.length <= 40 ? rest : null;
  }
  const variant = title.match(/\s+-\s+([^()]+?)(?:\s*\(|$)/)?.[1]?.trim() ?? null;
  return { brand, model: model || null, penType, variant: variant && variant.length <= 80 ? variant : null };
}

export class PeytonStreetPensConnector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  protected override minIntervalMs = 3000;
  private readonly cfg: z.infer<typeof ConfigSchema>;

  constructor(meta: ConnectorMeta) {
    super(meta);
    this.cfg = ConfigSchema.parse(meta.config);
  }

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const seeds = ctx.options.seeds?.length ? ctx.options.seeds.map((s) => SeedSchema.parse({ path: s.startsWith('/') ? s : new URL(s).pathname })) : this.cfg.seeds;
    const backfill = ctx.options.mode === 'backfill';
    const maxPages = backfill ? this.policy.backfillMaxPages : this.cfg.pagesPerSeed;
    const start = readSeedCursor(ctx.options.cursor, seeds.length);
    const seedsThisRun = backfill ? seeds.length : Math.min(seeds.length, this.cfg.seedsPerRun);
    let count = 0;
    let items = 0;
    for (let k = 0; k < seedsThisRun; k++) {
      const seedIndex = (start.seedIndex + k) % seeds.length;
      const seed = seeds[seedIndex]!;
      let page = k === 0 ? start.page : 1;
      for (; page <= maxPages; page++) {
        if (ctx.signal?.aborted || this.reached(ctx, count)) return;
        const url = pageUrl(seed.path, page);
        await this.throttle(url);
        const res = await ctx.fetch(url, {
          engines: ['api'],
          responseType: 'text',
          failOnHttpError: false,
          expect: ['title', 'price', 'currency'],
          parse: (r) => {
            const p = r.html ? parseCategoryHtml(r.html) : null;
            const priced = p?.cards.find((c) => c.price);
            return p?.cards.length ? { title: p.cards[0]!.title, price: priced?.price ?? null, currency: priced ? 'USD' : null } : null;
          },
          minQuality: 0.3,
        });
        if (res.httpStatus === 404) {
          ctx.anomaly('selector_missing', `${url}: category not found (404) — update config.seeds`);
          break;
        }
        const parsed = res.success && res.html ? parseCategoryHtml(res.html) : null;
        if (!parsed) {
          ctx.anomaly('page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
          break;
        }
        if (!parsed.cards.length) {
          if (page === 1) ctx.anomaly('parse_failure_page', `${url}: no product cards`);
          break;
        }
        count++;
        items += parsed.cards.length;
        const payload: PagePayload = { kind: 'listing_page', url, seed, page, totalPages: parsed.totalPages, cards: parsed.cards };
        yield { url, externalId: `${seed.path}:p${page}`, kind: 'listing', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
        await ctx.setCursor({ seedIndex, page: page + 1, at: new Date().toISOString() });
        await ctx.progress({ page, totalPages: parsed.totalPages, itemsProcessed: items });
        if (parsed.cards.length < PAGE_SIZE || (parsed.totalPages !== null && page >= parsed.totalPages)) break;
      }
      const nextSeed = (seedIndex + 1) % seeds.length;
      await ctx.setCursor({ seedIndex: nextSeed, page: 1, at: new Date().toISOString(), ...(backfill && nextSeed === 0 ? { done: true } : {}) });
    }
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PagePayloadSchema.parse(raw.payload);
    const exclude = new RegExp(this.cfg.exclude, 'i');
    const out: NormalizedRecord[] = [];
    for (const c of p.cards) {
      if (!c.price) continue;
      const isPen = PEN_TYPE.test(c.title) || /\bpen\b/i.test(c.title);
      const isLighter = /\blighter\b/i.test(c.title);
      if (!isLighter && (!isPen || (exclude.test(c.title) && !/\b(fountain pen|ballpoint|rollerball|pencil set|pen set)\b/i.test(c.title)))) continue;
      const t = parsePenTitle(c.title, c.brand && !GENERIC_BRANDS.test(c.brand.trim()) ? c.brand : (p.seed.brand ?? null));
      const condRaw = conditionFromParens(c.title);
      const { year, decade } = yearOrDecade(c.title);
      const nib = c.title.match(/\b(extra[- ]fine|fine|medium|broad|stub|italic|flex(?:ible)?|oblique)\b[^,()-]*\bnib\b/i)?.[0] ?? null;
      const attributes = AssetAttributesSchema.parse({
        categorySlug: isLighter ? 'lighters' : 'pens',
        brand: t.brand,
        model: t.model,
        name: t.brand && t.model ? `${t.brand} ${t.model}${t.penType ? ` ${t.penType}` : ''}` : c.title.split(/\s+-\s+|\s*\(/)[0]!.trim(),
        variant: t.variant,
        year,
        identifiers: { ...(c.sku ? { peyton_sku: c.sku } : {}), ...(c.productId ? { peyton_product_id: c.productId } : {}) },
        metadata: { pen_type: t.penType, decade, nib, list_price: c.rrp, seed: p.seed.path, restored: /\brestored\b/i.test(c.title), new_old_stock: /\b(new old stock|nos)\b/i.test(c.title) },
      });
      out.push(
        NormalizedListingSchema.parse({
          kind: 'listing',
          connectorId: this.meta.id,
          sourceId: this.meta.sourceId,
          sourceUrl: c.url,
          externalId: c.sku ?? c.productId ?? c.url,
          rawTitle: c.title,
          imageUrls: c.image ? [c.image] : [],
          attributes,
          grade: { grader: null, grade: null, qualifier: null, certificationNumber: null },
          condition: { condition: normalizeConditionWord(condRaw), conditionRaw: condRaw, completeness: /\b(new in box|in box|with box)\b/i.test(c.title) ? 'box_papers' : null },
          observedAt: raw.fetchedAt,
          confidence: 0.78,
          parserVersion: PARSER_VERSION,
          listingType: 'fixed_price',
          price: c.price,
          currency: 'USD',
          seller: 'Peyton Street Pens',
          location: 'Santa Cruz, California, United States',
          availability: c.soldOut ? 'sold' : 'available',
        }),
      );
    }
    return out;
  }
}

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