@jackwener/opencli 1.6.1 → 1.6.2

package/src/clis/1688/item.ts

@@ -0,0 +1,282 @@
+import { CommandExecutionError } from '../../errors.js';
+import { cli, Strategy } from '../../registry.js';
+import type { IPage } from '../../types.js';
+import { isRecord } from '../../utils.js';
+import {
+  assertAuthenticatedState,
+  buildDetailUrl,
+  buildProvenance,
+  canonicalizeSellerUrl,
+  cleanMultilineText,
+  cleanText,
+  extractLocation,
+  extractMemberId,
+  extractOfferId,
+  extractShopId,
+  gotoAndReadState,
+  normalizePriceTiers,
+  parseMoqText,
+  parsePriceText,
+  toNumber,
+  uniqueNonEmpty,
+} from './shared.js';
+
+interface BuyerProtectionModel {
+  serviceName?: string;
+  shortBuyerDesc?: string;
+  packageBuyerDesc?: string;
+  textDesc?: string;
+  agreeDeliveryHours?: number;
+}
+
+interface ItemBrowserPayload {
+  href?: string;
+  title?: string;
+  bodyText?: string;
+  offerTitle?: string;
+  offerId?: string | number;
+  seller?: {
+    companyName?: string;
+    memberId?: string;
+    winportUrl?: string;
+    sellerWinportUrlMap?: Record<string, string>;
+  };
+  trade?: {
+    beginAmount?: string | number;
+    priceDisplay?: string;
+    unit?: string;
+    saleCount?: string | number;
+    offerIDatacenterSellInfo?: Record<string, unknown>;
+    offerPriceModel?: {
+      currentPrices?: Array<{ beginAmount?: string | number; price?: string | number }>;
+    };
+  };
+  gallery?: {
+    mainImage?: string[];
+    offerImgList?: string[];
+    wlImageInfos?: Array<{ fullPathImageURI?: string }>;
+  };
+  shipping?: {
+    deliveryLimitText?: string;
+    logisticsText?: string;
+    protectionInfos?: BuyerProtectionModel[];
+    buyerProtectionModel?: BuyerProtectionModel[];
+  };
+  services?: BuyerProtectionModel[];
+}
+
+interface VisibleAttribute {
+  key: string;
+  value: string;
+}
+
+function normalizeItemPayload(payload: ItemBrowserPayload): Record<string, unknown> {
+  const href = cleanText(payload.href);
+  const bodyText = cleanMultilineText(payload.bodyText);
+  const sellerName = cleanText(payload.seller?.companyName);
+  const sellerUrlRaw = cleanText(
+    payload.seller?.winportUrl
+      ?? payload.seller?.sellerWinportUrlMap?.defaultUrl
+      ?? payload.seller?.sellerWinportUrlMap?.indexUrl,
+  );
+  const sellerUrl = canonicalizeSellerUrl(sellerUrlRaw);
+  const offerId = cleanText(String(payload.offerId ?? '')) || extractOfferId(href) || null;
+  const memberId = cleanText(payload.seller?.memberId) || extractMemberId(sellerUrlRaw || href) || null;
+  const shopId = extractShopId(sellerUrl ?? href);
+  const unit = cleanText(payload.trade?.unit);
+  const priceDisplay = cleanText(payload.trade?.priceDisplay);
+  const priceRange = parsePriceText(priceDisplay ? `¥${priceDisplay}` : bodyText);
+  const moqText = extractMoqText(bodyText, payload.trade?.beginAmount, unit);
+  const moq = parseMoqText(moqText);
+  const services = uniqueServices(payload);
+  const serviceBadges = uniqueNonEmpty(services.map((service) => cleanText(service.serviceName)));
+  const attributes = normalizeVisibleAttributes(payload.trade?.offerIDatacenterSellInfo);
+  const priceTiers = normalizePriceTiers(payload.trade?.offerPriceModel?.currentPrices ?? [], unit || null);
+  const images = uniqueNonEmpty([
+    ...(payload.gallery?.mainImage ?? []),
+    ...(payload.gallery?.offerImgList ?? []),
+    ...((payload.gallery?.wlImageInfos ?? []).map((item) => item.fullPathImageURI ?? '')),
+  ]);
+
+  const detailUrl = offerId ? buildDetailUrl(offerId) : href;
+  const provenance = buildProvenance(href || detailUrl);
+
+  return {
+    offer_id: offerId,
+    member_id: memberId,
+    shop_id: shopId,
+    title: cleanText(payload.offerTitle) || stripAlibabaSuffix(payload.title) || firstNonEmptyLine(bodyText) || null,
+    item_url: detailUrl,
+    main_images: images,
+    price_text: priceRange.price_text || null,
+    price_tiers: priceTiers,
+    currency: priceRange.currency,
+    moq_text: moq.moq_text || null,
+    moq_value: moq.moq_value,
+    seller_name: sellerName || null,
+    seller_url: sellerUrl,
+    shop_name: sellerName || null,
+    origin_place: extractLocation(bodyText),
+    delivery_days_text: extractDeliveryDaysText(bodyText, services, payload.shipping),
+    customization_text: extractKeywordLine(bodyText, ['来样定制', '来图定制', '支持定制', '可定制', '定制']),
+    private_label_text: extractKeywordLine(bodyText, ['贴牌', '贴标', '定制logo', '打logo', 'OEM', 'ODM']),
+    visible_attributes: attributes,
+    sales_text: extractSalesText(bodyText),
+    service_badges: serviceBadges,
+    stock_quantity: extractStockQuantity(bodyText),
+    ...provenance,
+  };
+}
+
+function normalizeVisibleAttributes(raw: unknown): VisibleAttribute[] {
+  if (!isRecord(raw)) return [];
+  return Object.entries(raw)
+    .filter(([key, value]) => key !== 'sellPointModel' && cleanText(key) && cleanText(String(value)))
+    .map(([key, value]) => ({ key: cleanText(key), value: cleanText(String(value)) }));
+}
+
+function uniqueServices(payload: ItemBrowserPayload): BuyerProtectionModel[] {
+  const combined = [
+    ...(Array.isArray(payload.services) ? payload.services : []),
+    ...(Array.isArray(payload.shipping?.protectionInfos) ? payload.shipping.protectionInfos : []),
+    ...(Array.isArray(payload.shipping?.buyerProtectionModel) ? payload.shipping.buyerProtectionModel : []),
+  ];
+
+  const seen = new Set<string>();
+  const result: BuyerProtectionModel[] = [];
+  for (const service of combined) {
+    const key = cleanText(service.serviceName);
+    if (!key || seen.has(key)) continue;
+    seen.add(key);
+    result.push(service);
+  }
+  return result;
+}
+
+function stripAlibabaSuffix(title: string | undefined): string {
+  return cleanText(title).replace(/\s*-\s*阿里巴巴$/, '').trim();
+}
+
+function firstNonEmptyLine(text: string): string {
+  return text.split('\n').map((line) => cleanText(line)).find(Boolean) ?? '';
+}
+
+function extractMoqText(bodyText: string, beginAmount: string | number | undefined, unit: string): string {
+  const lineMatch = bodyText.match(/\d+(?:\.\d+)?\s*(件|个|套|箱|包|双|台|把|只)\s*起批/);
+  if (lineMatch) return lineMatch[0];
+
+  const moqValue = toNumber(beginAmount);
+  if (moqValue !== null) {
+    return `${moqValue}${unit || ''}起批`;
+  }
+
+  return '';
+}
+
+function extractDeliveryDaysText(
+  bodyText: string,
+  services: BuyerProtectionModel[],
+  shipping: ItemBrowserPayload['shipping'],
+): string | null {
+  const shippingText = cleanText(shipping?.deliveryLimitText) || cleanText(shipping?.logisticsText);
+  if (shippingText) return shippingText;
+
+  const textMatch = bodyText.match(/\d+\s*(?:小时|天)(?:内)?发货/);
+  if (textMatch) return textMatch[0];
+
+  const hourMatch = services.find((service) => typeof service.agreeDeliveryHours === 'number');
+  if (hourMatch && typeof hourMatch.agreeDeliveryHours === 'number') {
+    return `${hourMatch.agreeDeliveryHours}小时内发货`;
+  }
+
+  return null;
+}
+
+function extractKeywordLine(bodyText: string, keywords: string[]): string | null {
+  const lines = bodyText.split('\n').map((line) => cleanText(line)).filter(Boolean);
+  for (const line of lines) {
+    if (keywords.some((keyword) => line.includes(keyword))) {
+      return line;
+    }
+  }
+  return null;
+}
+
+function extractSalesText(bodyText: string): string | null {
+  const match = bodyText.match(/(?:全网销量|已售)\s*\d+(?:\.\d+)?\+?\s*[件套个单]?/);
+  return match ? cleanText(match[0]) : null;
+}
+
+function extractStockQuantity(bodyText: string): number | null {
+  const match = bodyText.match(/库存\s*(\d+)/);
+  return match ? Number.parseInt(match[1], 10) : null;
+}
+
+async function readItemPayload(page: IPage, itemUrl: string): Promise<ItemBrowserPayload> {
+  const state = await gotoAndReadState(page, itemUrl, 2500, 'item');
+  assertAuthenticatedState(state, 'item');
+
+  const payload = await page.evaluate(`
+    (() => {
+      const root = window.context ?? {};
+      const model = root.result?.global?.globalData?.model ?? null;
+      const toJson = (value) => JSON.parse(JSON.stringify(value ?? null));
+      return {
+        href: window.location.href,
+        title: document.title || '',
+        bodyText: document.body ? document.body.innerText || '' : '',
+        offerTitle: model?.offerTitleModel?.subject ?? '',
+        offerId: model?.tradeModel?.offerId ?? '',
+        seller: toJson(model?.sellerModel),
+        trade: toJson(model?.tradeModel),
+        gallery: toJson(root.result?.data?.gallery?.fields ?? null),
+        shipping: toJson(root.result?.data?.shippingServices?.fields ?? null),
+        services: toJson(root.result?.data?.shippingServices?.fields?.protectionInfos ?? []),
+      };
+    })()
+  `) as ItemBrowserPayload;
+
+  const resolvedOfferId = cleanText(String(payload.offerId ?? '')) || extractOfferId(cleanText(payload.href));
+  if (!resolvedOfferId) {
+    throw new CommandExecutionError(
+      '1688 item page did not expose product context',
+      '当前 tab 非商品详情上下文，请切到 detail.1688.com 商品页并重试',
+    );
+  }
+
+  return payload;
+}
+
+cli({
+  site: '1688',
+  name: 'item',
+  description: '1688 商品详情（公开商品字段、价格阶梯、卖家基础信息）',
+  domain: 'www.1688.com',
+  strategy: Strategy.COOKIE,
+  navigateBefore: false,
+  args: [
+    {
+      name: 'input',
+      required: true,
+      positional: true,
+      help: '1688 商品 URL 或 offer ID（如 887904326744）',
+    },
+  ],
+  columns: ['offer_id', 'title', 'price_text', 'moq_text', 'seller_name', 'origin_place'],
+  func: async (page, kwargs) => {
+    const itemUrl = buildDetailUrl(String(kwargs.input ?? ''));
+    const payload = await readItemPayload(page, itemUrl);
+    return [normalizeItemPayload(payload)];
+  },
+});
+
+export const __test__ = {
+  normalizeItemPayload,
+  normalizeVisibleAttributes,
+  stripAlibabaSuffix,
+  extractMoqText,
+  extractDeliveryDaysText,
+  extractKeywordLine,
+  extractSalesText,
+  extractStockQuantity,
+};

package/src/clis/1688/search.test.ts

@@ -0,0 +1,81 @@
+import { describe, expect, it } from 'vitest';
+import { __test__ } from './search.js';
+
+describe('1688 search normalization', () => {
+  it('normalizes search candidates into structured result rows', () => {
+    const result = __test__.normalizeSearchCandidate({
+      item_url: 'https://detail.1688.com/offer/887904326744.html',
+      title: '宿舍置物架桌面加高架',
+      container_text: '宿舍置物架桌面加高架 ¥56.00 2套起批 山东青岛 已售300+套',
+      price_text: '¥ 56 .00',
+      sales_text: '300+套',
+      moq_text: '2套起批',
+      tag_items: ['退货包运费', '回头率52%'],
+      hover_items: ['验厂报告'],
+      seller_name: '青岛沁澜衣品服装有限公司',
+      seller_url: 'https://yinuoweierfushi.1688.com/page/index.html?spm=a123',
+    }, 'https://s.1688.com/selloffer/offer_search.htm?charset=utf8&keywords=置物架');
+
+    expect(result.rank).toBe(0);
+    expect(result.offer_id).toBe('887904326744');
+    expect(result.shop_id).toBe('yinuoweierfushi');
+    expect(result.item_url).toBe('https://detail.1688.com/offer/887904326744.html');
+    expect(result.seller_url).toBe('https://yinuoweierfushi.1688.com');
+    expect(result.price_text).toBe('¥56.00');
+    expect(result.price_min).toBe(56);
+    expect(result.price_max).toBe(56);
+    expect(result.moq_value).toBe(2);
+    expect(result.location).toBe('山东青岛');
+    expect(result.sales_text).toBe('300+套');
+    expect(result.badges).toEqual(expect.arrayContaining(['退货包运费', '验厂报告']));
+    expect(result.return_rate_text).toBe('回头率52%');
+  });
+
+  it('does not use hover_price_text as MOQ source', () => {
+    const result = __test__.normalizeSearchCandidate({
+      item_url: 'https://detail.1688.com/offer/887904326744.html',
+      title: 'test',
+      container_text: 'test ¥56.00',
+      price_text: '¥ 56 .00',
+      hover_price_text: '¥56.00 3件起批',
+      moq_text: null,
+    }, 'https://s.1688.com/selloffer/offer_search.htm?charset=utf8&keywords=test');
+    // hover_price_text should not be used for MOQ extraction
+    expect(result.moq_text).toBeNull();
+    expect(result.moq_value).toBeNull();
+  });
+
+  it('extracts offer id from mobile detail search links', () => {
+    const result = __test__.normalizeSearchCandidate({
+      item_url: 'http://detail.m.1688.com/page/index.html?offerId=910933345396&sortType=&pageId=',
+      title: '',
+      container_text: '桌面书桌办公室工位收纳展示新中式博古架多层茶具厨房摆放置物架 ¥24.3 已售20+件',
+      price_text: '¥ 14 .28',
+      sales_text: '1500+件',
+      moq_text: '≥2个',
+      seller_name: '泰商国际贸易（宁阳）有限公司',
+      seller_url: 'http://tsgjmy.1688.com/',
+    }, 'https://s.1688.com/selloffer/offer_search.htm?charset=utf8&keywords=桌面置物架');
+
+    expect(result.offer_id).toBe('910933345396');
+    expect(result.shop_id).toBe('tsgjmy');
+    expect(result.item_url).toBe('https://detail.1688.com/offer/910933345396.html');
+    expect(result.title).toContain('桌面书桌办公室工位收纳展示');
+    expect(result.price_text).toBe('¥14.28');
+    expect(result.sales_text).toBe('1500+件');
+    expect(result.moq_text).toBe('≥2个');
+    expect(result.moq_value).toBe(2);
+  });
+
+  it('prefers offer id and falls back to item url for dedupe key', () => {
+    expect(__test__.buildDedupeKey({
+      offer_id: '123456',
+      item_url: 'https://detail.1688.com/offer/123456.html',
+    })).toBe('offer:123456');
+    expect(__test__.buildDedupeKey({
+      offer_id: null,
+      item_url: 'https://detail.1688.com/offer/123456.html',
+    })).toBe('url:https://detail.1688.com/offer/123456.html');
+    expect(__test__.buildDedupeKey({ offer_id: null, item_url: null })).toBeNull();
+  });
+});