@jackwener/opencli 1.6.1 → 1.6.3

package/src/clis/amazon/bestsellers.ts

@@ -1,180 +0,0 @@
-import { CommandExecutionError } from '../../errors.js';
-import { cli, Strategy } from '../../registry.js';
-import type { IPage } from '../../types.js';
-import {
-  buildProvenance,
-  cleanText,
-  extractAsin,
-  extractReviewCountFromCardText,
-  firstMeaningfulLine,
-  normalizeProductUrl,
-  parsePriceText,
-  parseRatingValue,
-  parseReviewCount,
-  resolveBestsellersUrl,
-  uniqueNonEmpty,
-  assertUsableState,
-  gotoAndReadState,
-} from './shared.js';
-
-interface BestsellersPagePayload {
-  href?: string;
-  title?: string;
-  list_title?: string;
-  cards?: Array<{
-    rank_text?: string | null;
-    asin?: string | null;
-    title?: string | null;
-    href?: string | null;
-    price_text?: string | null;
-    rating_text?: string | null;
-    review_count_text?: string | null;
-    card_text?: string | null;
-  }>;
-  page_links?: string[];
-}
-
-function normalizeBestsellerCandidate(
-  candidate: NonNullable<BestsellersPagePayload['cards']>[number],
-  rank: number,
-  listTitle: string | null,
-  sourceUrl: string,
-): Record<string, unknown> {
-  const productUrl = normalizeProductUrl(candidate.href);
-  const asin = extractAsin(candidate.asin ?? '') ?? extractAsin(productUrl ?? '') ?? null;
-  const title = cleanText(candidate.title) || firstMeaningfulLine(candidate.card_text);
-  const price = parsePriceText(cleanText(candidate.price_text) || candidate.card_text);
-  const ratingText = cleanText(candidate.rating_text) || null;
-  const reviewCountText = cleanText(candidate.review_count_text)
-    || extractReviewCountFromCardText(candidate.card_text)
-    || null;
-  const provenance = buildProvenance(sourceUrl);
-
-  return {
-    rank,
-    asin,
-    title: title || null,
-    product_url: productUrl,
-    list_title: listTitle,
-    ...provenance,
-    price_text: price.price_text,
-    price_value: price.price_value,
-    currency: price.currency,
-    rating_text: ratingText,
-    rating_value: parseRatingValue(ratingText),
-    review_count_text: reviewCountText,
-    review_count: parseReviewCount(reviewCountText),
-  };
-}
-
-async function readBestsellersPage(page: IPage, url: string): Promise<BestsellersPagePayload> {
-  const state = await gotoAndReadState(page, url, 2500, 'bestsellers');
-  assertUsableState(state, 'bestsellers');
-
-  return await page.evaluate(`
-    (() => ({
-      href: window.location.href,
-      title: document.title || '',
-      list_title:
-        document.querySelector('#zg_banner_text')?.textContent
-        || document.querySelector('h1')?.textContent
-        || '',
-      cards: Array.from(document.querySelectorAll('.p13n-sc-uncoverable-faceout'))
-        .map((card) => ({
-          rank_text:
-            card.querySelector('.zg-bdg-text')?.textContent
-            || card.querySelector('[class*="rank"]')?.textContent
-            || '',
-          asin: card.id || '',
-          title:
-            card.querySelector('[class*="line-clamp"]')?.textContent
-            || card.querySelector('img')?.getAttribute('alt')
-            || '',
-          href: card.querySelector('a[href*="/dp/"]')?.href || '',
-          price_text: card.querySelector('.a-price .a-offscreen')?.textContent || '',
-          rating_text: card.querySelector('[aria-label*="out of 5 stars"]')?.getAttribute('aria-label') || '',
-          review_count_text:
-            card.querySelector('a[href*="#customerReviews"]')?.textContent
-            || card.querySelector('.a-size-small')?.textContent
-            || '',
-          card_text: card.innerText || '',
-        })),
-      page_links: Array.from(document.querySelectorAll('li.a-normal a, li.a-selected a'))
-        .map((anchor) => anchor.href || '')
-        .filter((href) => /\\/zgbs\\//.test(href) && /(?:[?&]pg=|ref=zg_bs_pg_)/.test(href)),
-    }))()
-  `) as BestsellersPagePayload;
-}
-
-cli({
-  site: 'amazon',
-  name: 'bestsellers',
-  description: 'Amazon Best Sellers pages for category candidate discovery',
-  domain: 'amazon.com',
-  strategy: Strategy.COOKIE,
-  navigateBefore: false,
-  args: [
-    {
-      name: 'input',
-      positional: true,
-      help: 'Best sellers URL or /zgbs path. Omit to use the root Best Sellers page.',
-    },
-    {
-      name: 'limit',
-      type: 'int',
-      default: 100,
-      help: 'Maximum number of ranked items to return (default 100)',
-    },
-  ],
-  columns: ['rank', 'asin', 'title', 'price_text', 'rating_value', 'review_count'],
-  func: async (page, kwargs) => {
-    const limit = Math.max(1, Number(kwargs.limit) || 100);
-    const initialUrl = resolveBestsellersUrl(typeof kwargs.input === 'string' ? kwargs.input : undefined);
-
-    const queue = [initialUrl];
-    const visited = new Set<string>();
-    const seenAsins = new Set<string>();
-    const results: Record<string, unknown>[] = [];
-    let listTitle: string | null = null;
-
-    while (queue.length > 0 && results.length < limit) {
-      const nextUrl = queue.shift()!;
-      if (visited.has(nextUrl)) continue;
-      visited.add(nextUrl);
-
-      const payload = await readBestsellersPage(page, nextUrl);
-      const sourceUrl = cleanText(payload.href) || nextUrl;
-      listTitle = cleanText(payload.list_title) || cleanText(payload.title) || listTitle;
-      const cards = payload.cards ?? [];
-
-      for (const card of cards) {
-        const normalized = normalizeBestsellerCandidate(card, results.length + 1, listTitle, sourceUrl);
-        const asin = cleanText(String(normalized.asin ?? ''));
-        if (!asin || seenAsins.has(asin)) continue;
-        seenAsins.add(asin);
-        results.push(normalized);
-        if (results.length >= limit) break;
-      }
-
-      const pageLinks = uniqueNonEmpty(payload.page_links ?? []);
-      for (const href of pageLinks) {
-        if (!visited.has(href) && !queue.includes(href)) {
-          queue.push(href);
-        }
-      }
-    }
-
-    if (results.length === 0) {
-      throw new CommandExecutionError(
-        'amazon bestsellers did not expose any ranked items',
-        'Open the same best sellers page in Chrome, verify it is a real Amazon ranking page, and retry.',
-      );
-    }
-
-    return results.slice(0, limit);
-  },
-});
-
-export const __test__ = {
-  normalizeBestsellerCandidate,
-};

package/src/clis/amazon/discussion.test.ts

@@ -1,38 +0,0 @@
-import { describe, expect, it } from 'vitest';
-import { __test__ } from './discussion.js';
-
-describe('amazon discussion normalization', () => {
-  it('normalizes review summary and sample reviews', () => {
-    const result = __test__.normalizeDiscussionPayload({
-      href: 'https://www.amazon.com/product-reviews/B0FJS72893',
-      average_rating_text: '3.9 out of 5',
-      total_review_count_text: '27 global ratings',
-      qa_links: [],
-      review_samples: [
-        {
-          title: '5.0 out of 5 stars Great value and quality',
-          rating_text: '5.0 out of 5 stars',
-          author: 'GTreader2',
-          date_text: 'Reviewed in the United States on February 21, 2026',
-          body: 'Small but mighty.',
-          verified: true,
-        },
-      ],
-    });
-
-    expect(result.asin).toBe('B0FJS72893');
-    expect(result.average_rating_value).toBe(3.9);
-    expect(result.total_review_count).toBe(27);
-    expect(result.review_samples).toEqual([
-      {
-        title: 'Great value and quality',
-        rating_text: '5.0 out of 5 stars',
-        rating_value: 5,
-        author: 'GTreader2',
-        date_text: 'Reviewed in the United States on February 21, 2026',
-        body: 'Small but mighty.',
-        verified_purchase: true,
-      },
-    ]);
-  });
-});

package/src/clis/amazon/discussion.ts

@@ -1,131 +0,0 @@
-import { CommandExecutionError } from '../../errors.js';
-import { cli, Strategy } from '../../registry.js';
-import type { IPage } from '../../types.js';
-import {
-  buildDiscussionUrl,
-  buildProvenance,
-  cleanText,
-  extractAsin,
-  normalizeProductUrl,
-  parseRatingValue,
-  parseReviewCount,
-  trimRatingPrefix,
-  uniqueNonEmpty,
-  assertUsableState,
-  gotoAndReadState,
-} from './shared.js';
-
-interface DiscussionPayload {
-  href?: string;
-  title?: string;
-  average_rating_text?: string | null;
-  total_review_count_text?: string | null;
-  qa_links?: string[];
-  review_samples?: Array<{
-    title?: string | null;
-    rating_text?: string | null;
-    author?: string | null;
-    date_text?: string | null;
-    body?: string | null;
-    verified?: boolean;
-  }>;
-}
-
-function normalizeDiscussionPayload(payload: DiscussionPayload): Record<string, unknown> {
-  const sourceUrl = cleanText(payload.href) || buildDiscussionUrl(payload.href ?? '');
-  const asin = extractAsin(payload.href ?? '') ?? null;
-  const averageRatingText = cleanText(payload.average_rating_text) || null;
-  const totalReviewCountText = cleanText(payload.total_review_count_text) || null;
-  const provenance = buildProvenance(sourceUrl);
-
-  return {
-    asin,
-    product_url: asin ? normalizeProductUrl(asin) : null,
-    discussion_url: sourceUrl,
-    ...provenance,
-    average_rating_text: averageRatingText,
-    average_rating_value: parseRatingValue(averageRatingText),
-    total_review_count_text: totalReviewCountText,
-    total_review_count: parseReviewCount(totalReviewCountText),
-    qa_urls: uniqueNonEmpty(payload.qa_links ?? []),
-    review_samples: (payload.review_samples ?? []).map((sample) => ({
-      title: trimRatingPrefix(sample.title) || null,
-      rating_text: cleanText(sample.rating_text) || null,
-      rating_value: parseRatingValue(sample.rating_text),
-      author: cleanText(sample.author) || null,
-      date_text: cleanText(sample.date_text) || null,
-      body: cleanText(sample.body) || null,
-      verified_purchase: sample.verified === true,
-    })),
-  };
-}
-
-async function readDiscussionPayload(page: IPage, input: string, limit: number): Promise<DiscussionPayload> {
-  const url = buildDiscussionUrl(input);
-  const state = await gotoAndReadState(page, url, 2500, 'discussion');
-  assertUsableState(state, 'discussion');
-
-  return await page.evaluate(`
-    (() => ({
-      href: window.location.href,
-      title: document.title || '',
-      average_rating_text: document.querySelector('[data-hook="rating-out-of-text"]')?.textContent || '',
-      total_review_count_text: document.querySelector('[data-hook="total-review-count"]')?.textContent || '',
-      qa_links: Array.from(document.querySelectorAll('a[href*="ask/questions"]')).map((anchor) => anchor.href || ''),
-      review_samples: Array.from(document.querySelectorAll('[data-hook="review"]')).slice(0, ${limit}).map((card) => ({
-        title: card.querySelector('[data-hook="review-title"]')?.textContent || '',
-        rating_text:
-          card.querySelector('[data-hook="review-star-rating"]')?.textContent
-          || card.querySelector('[data-hook="cmps-review-star-rating"]')?.textContent
-          || '',
-        author: card.querySelector('.a-profile-name')?.textContent || '',
-        date_text: card.querySelector('[data-hook="review-date"]')?.textContent || '',
-        body: card.querySelector('[data-hook="review-body"]')?.textContent || '',
-        verified: !!card.querySelector('[data-hook="avp-badge"]'),
-      })),
-    }))()
-  `) as DiscussionPayload;
-}
-
-cli({
-  site: 'amazon',
-  name: 'discussion',
-  description: 'Amazon review summary and sample customer discussion from product review pages',
-  domain: 'amazon.com',
-  strategy: Strategy.COOKIE,
-  navigateBefore: false,
-  args: [
-    {
-      name: 'input',
-      required: true,
-      positional: true,
-      help: 'ASIN or product URL, for example B0FJS72893',
-    },
-    {
-      name: 'limit',
-      type: 'int',
-      default: 10,
-      help: 'Maximum number of review samples to return (default 10)',
-    },
-  ],
-  columns: ['asin', 'average_rating_value', 'total_review_count'],
-  func: async (page, kwargs) => {
-    const input = String(kwargs.input ?? '');
-    const limit = Math.max(1, Number(kwargs.limit) || 10);
-    const payload = await readDiscussionPayload(page, input, limit);
-    const normalized = normalizeDiscussionPayload(payload);
-
-    if (!normalized.average_rating_text && !normalized.total_review_count_text) {
-      throw new CommandExecutionError(
-        'amazon discussion page did not expose review summary',
-        'The review page may have changed or hit a robot check. Open the review page in Chrome and retry.',
-      );
-    }
-
-    return [normalized];
-  },
-});
-
-export const __test__ = {
-  normalizeDiscussionPayload,
-};

package/src/clis/amazon/offer.test.ts

@@ -1,35 +0,0 @@
-import { describe, expect, it } from 'vitest';
-import { __test__ } from './offer.js';
-
-describe('amazon offer normalization', () => {
-  it('extracts sold-by and fulfillment facts from product offer text', () => {
-    const result = __test__.normalizeOfferPayload({
-      href: 'https://www.amazon.com/dp/B0FJS72893',
-      price_text: '$15.99',
-      merchant_info: '',
-      sold_by: 'KUATUDIRECT',
-      ships_from_text: 'Ships from Amazon',
-      offer_link: null,
-      review_url: 'https://www.amazon.com/dp/B0FJS72893#customerReviews',
-      qa_url: null,
-    });
-
-    expect(result.asin).toBe('B0FJS72893');
-    expect(result.sold_by).toBe('KUATUDIRECT');
-    expect(result.ships_from).toBe('Amazon');
-    expect(result.is_amazon_sold).toBe(false);
-    expect(result.is_amazon_fulfilled).toBe(true);
-  });
-
-  it('parses merchant info fallback text', () => {
-    expect(__test__.extractSoldBy('Sold by Example Seller and Fulfilled by Amazon.')).toBe('Example Seller');
-    expect(__test__.extractShipsFrom('Ships from Amazon')).toBe('Amazon');
-  });
-
-  it('detects delivery-location blocking in the buy box text', () => {
-    expect(__test__.isDeliveryLocationBlocked(
-      'This item cannot be shipped to your selected delivery location. Similar items shipping to Hong Kong',
-    )).toBe(true);
-    expect(__test__.isDeliveryLocationBlocked('Ships from Amazon')).toBe(false);
-  });
-});

package/src/clis/amazon/offer.ts

@@ -1,185 +0,0 @@
-import { CommandExecutionError } from '../../errors.js';
-import { cli, Strategy } from '../../registry.js';
-import type { IPage } from '../../types.js';
-import {
-  buildProductUrl,
-  buildProvenance,
-  cleanText,
-  extractAsin,
-  isAmazonEntity,
-  normalizeProductUrl,
-  PRIMARY_PRICE_SELECTORS,
-  parsePriceText,
-  assertUsableState,
-  gotoAndReadState,
-} from './shared.js';
-
-interface OfferPayload {
-  href?: string;
-  title?: string;
-  price_text?: string | null;
-  merchant_info?: string | null;
-  sold_by?: string | null;
-  ships_from_text?: string | null;
-  offer_link?: string | null;
-  review_url?: string | null;
-  qa_url?: string | null;
-  buybox_text?: string | null;
-}
-
-const OFFER_FACT_SELECTOR = [
-  '#sellerProfileTriggerId',
-  '#shipsFromSoldByInsideBuyBox_feature_div',
-  '#fulfillerInfoFeature_feature_div',
-  '#merchantInfoFeature_feature_div',
-  '#tabular-buybox-container',
-  '#merchant-info',
-].join(', ');
-
-function collapseAdjacentWords(text: string): string {
-  const parts = cleanText(text).split(' ').filter(Boolean);
-  const deduped: string[] = [];
-  for (const part of parts) {
-    if (deduped[deduped.length - 1] === part) continue;
-    deduped.push(part);
-  }
-  return deduped.join(' ');
-}
-
-function extractShipsFrom(text: string): string | null {
-  const normalized = cleanText(text);
-  const match = normalized.match(/Ships from\s+(.+?)(?=Sold by|and Fulfilled by|$)/i);
-  return match ? collapseAdjacentWords(match[1].replace(/Ships from/ig, '')) : null;
-}
-
-function extractSoldBy(text: string): string | null {
-  const normalized = cleanText(text);
-  const match = normalized.match(/Sold by\s+(.+?)(?=and Fulfilled by|Ships from|$)/i);
-  return match ? collapseAdjacentWords(match[1]) : null;
-}
-
-function isDeliveryLocationBlocked(text: string | null | undefined): boolean {
-  const normalized = cleanText(text).toLowerCase();
-  return normalized.includes('cannot be shipped to your selected delivery location')
-    || normalized.includes('similar items shipping to')
-    || normalized.includes('deliver to hong kong');
-}
-
-function normalizeOfferPayload(payload: OfferPayload): Record<string, unknown> {
-  const asin = extractAsin(payload.href ?? '') ?? null;
-  const sourceUrl = cleanText(payload.href) || buildProductUrl(payload.href ?? '');
-  const price = parsePriceText(payload.price_text);
-  const merchantInfo = cleanText(payload.merchant_info) || null;
-  const soldBy = cleanText(payload.sold_by)
-    || extractSoldBy(payload.ships_from_text ?? '')
-    || extractSoldBy(merchantInfo ?? '')
-    || null;
-  const shipsFrom = extractShipsFrom(payload.ships_from_text ?? '')
-    || extractShipsFrom(merchantInfo ?? '')
-    || cleanText(payload.ships_from_text)
-    || null;
-  const provenance = buildProvenance(sourceUrl);
-
-  return {
-    asin,
-    product_url: normalizeProductUrl(payload.href),
-    ...provenance,
-    price_text: price.price_text,
-    price_value: price.price_value,
-    currency: price.currency,
-    merchant_info_text: merchantInfo,
-    sold_by: soldBy,
-    ships_from: shipsFrom,
-    offer_listing_url: cleanText(payload.offer_link) || null,
-    review_url: cleanText(payload.review_url) || null,
-    qa_url: cleanText(payload.qa_url) || null,
-    is_amazon_sold: isAmazonEntity(soldBy),
-    is_amazon_fulfilled: isAmazonEntity(shipsFrom) || /fulfilled by amazon/i.test(merchantInfo ?? ''),
-  };
-}
-
-async function readOfferPayload(page: IPage, input: string): Promise<OfferPayload> {
-  const url = buildProductUrl(input);
-  const state = await gotoAndReadState(page, url, 2500, 'offer');
-  assertUsableState(state, 'offer');
-
-  // Reconnecting to an existing Amazon target can surface the product page
-  // before the buy-box / merchant blocks are reattached to the DOM.
-  await page.wait({ selector: OFFER_FACT_SELECTOR, timeout: 6 }).catch(() => {});
-
-  return await page.evaluate(`
-    (() => ({
-      href: window.location.href,
-      title: document.title || '',
-      price_text: (() => {
-        const selectors = ${JSON.stringify(PRIMARY_PRICE_SELECTORS)};
-        for (const selector of selectors) {
-          const text = document.querySelector(selector)?.textContent || '';
-          if (text.trim()) return text;
-        }
-        return '';
-      })(),
-      merchant_info: document.querySelector('#merchant-info')?.textContent || '',
-      sold_by: document.querySelector('#sellerProfileTriggerId')?.textContent || '',
-      ships_from_text:
-        document.querySelector('#shipsFromSoldByInsideBuyBox_feature_div')?.textContent
-        || document.querySelector('#fulfillerInfoFeature_feature_div')?.textContent
-        || document.querySelector('#merchantInfoFeature_feature_div')?.textContent
-        || document.querySelector('#tabular-buybox-container')?.textContent
-        || '',
-      offer_link: document.querySelector('a[href*="/gp/offer-listing/"]')?.href || '',
-      review_url: document.querySelector('a[href*="#customerReviews"]')?.href || '',
-      qa_url: document.querySelector('a[href*="ask/questions"]')?.href || '',
-      buybox_text:
-        document.querySelector('#desktop_qualifiedBuyBox')?.textContent
-        || document.querySelector('#buybox')?.textContent
-        || '',
-    }))()
-  `) as OfferPayload;
-}
-
-cli({
-  site: 'amazon',
-  name: 'offer',
-  description: 'Amazon seller, buy box, and fulfillment facts from the product page',
-  domain: 'amazon.com',
-  strategy: Strategy.COOKIE,
-  navigateBefore: false,
-  args: [
-    {
-      name: 'input',
-      required: true,
-      positional: true,
-      help: 'ASIN or product URL, for example B0FJS72893',
-    },
-  ],
-  columns: ['asin', 'price_text', 'sold_by', 'ships_from', 'is_amazon_sold', 'is_amazon_fulfilled'],
-  func: async (page, kwargs) => {
-    const input = String(kwargs.input ?? '');
-    const payload = await readOfferPayload(page, input);
-    const normalized = normalizeOfferPayload(payload);
-
-    if (!normalized.sold_by && !normalized.ships_from && !normalized.merchant_info_text) {
-      if (isDeliveryLocationBlocked(payload.buybox_text)) {
-        throw new CommandExecutionError(
-          'amazon offer buy box is blocked by the current delivery location',
-          'The shared Chrome profile is not set to the target US delivery address. Switch Amazon delivery location to the requested US destination, reopen the product page, and retry.',
-        );
-      }
-
-      throw new CommandExecutionError(
-        'amazon offer surface did not expose seller or fulfillment facts',
-        'The product page may have changed. Open the product page in Chrome, make sure the buy box is visible, and retry.',
-      );
-    }
-
-    return [normalized];
-  },
-});
-
-export const __test__ = {
-  extractShipsFrom,
-  extractSoldBy,
-  isDeliveryLocationBlocked,
-  normalizeOfferPayload,
-};

package/src/clis/amazon/product.test.ts

@@ -1,26 +0,0 @@
-import { describe, expect, it } from 'vitest';
-import { __test__ } from './product.js';
-
-describe('amazon product normalization', () => {
-  it('normalizes product facts from the product page', () => {
-    const result = __test__.normalizeProductPayload({
-      href: 'https://www.amazon.com/dp/B0FJS72893',
-      title: 'Amazon.com: KVTUKIAIT Desktop Shelf Organizer',
-      product_title: 'White Desktop Shelf Organizer for Top of Desk',
-      byline: 'Visit the KVTUKIAIT Store',
-      price_text: '$15.99',
-      rating_text: '3.9 out of 5 stars',
-      review_count_text: '27 ratings',
-      review_url: 'https://www.amazon.com/dp/B0FJS72893#customerReviews',
-      qa_url: null,
-      bullets: ['SPACE-SAVING DESK SHELF ORGANIZER', 'SMALL AND STYLISH AESTHETIC DECOR'],
-      breadcrumbs: ['Office Products', 'Desktop & Off-Surface Shelves'],
-    });
-
-    expect(result.asin).toBe('B0FJS72893');
-    expect(result.price_value).toBe(15.99);
-    expect(result.rating_value).toBe(3.9);
-    expect(result.review_count).toBe(27);
-    expect(result.breadcrumbs).toEqual(['Office Products', 'Desktop & Off-Surface Shelves']);
-  });
-});