azon_seeker/scripts/test-amazon-search-parser.ts

import assert from 'node:assert/strict';
import { JSDOM } from 'jsdom';
import { collectAmazonSearchPageData } from '../src/page-worker/web-injectors/amazon-search-parser';

const html = `
  <main id="search">
    <section data-component-type="sponsored-brand">
      <span aria-label="Sponsored">Sponsored</span>
      <a href="/stores/example"><span class="brand-logo">Example brand</span></a>
      <h2>Example brand headline</h2>
      <article data-asin="B000000001">
        <h3>Brand product one</h3>
        <a href="/dp/B000000001"><img src="https://example.com/brand-1.jpg" /></a>
      </article>
      <article data-asin="B000000002">
        <h3>Brand product two</h3>
        <a href="/dp/B000000002"><img src="https://example.com/brand-2.jpg" /></a>
      </article>
    </section>

    <div class="s-main-slot">
      <div class="ordinary-result-grid">
        <article data-component-type="s-search-result" data-asin="B000000003">
        <span>Amazon's Choice</span>
        <h2>Organic product</h2>
        <a href="/dp/B000000003"><img class="s-image" src="https://example.com/organic.jpg" /></a>
        <span class="a-price"><span class="a-offscreen">$29.99</span></span>
        </article>

        <article data-component-type="s-search-result" data-asin="B000000004">
        <a aria-label="View Sponsored information or leave ad feedback"></a>
        <h2>Sponsored product</h2>
        <a href="/sspa/click?url=%2Fdp%2FB000000004"><img class="s-image" src="https://example.com/sp.jpg" /></a>
        <span class="a-price"><span class="a-offscreen">$33.93</span></span>
        <span class="a-price a-text-price" data-a-strike="true"><span class="a-offscreen">$39.99</span></span>
        <span class="a-price a-text-price apex-priceperunit-value"><span class="a-offscreen">$0.28</span></span>
        </article>

        <article data-component-type="s-search-result" data-asin="B000000007">
        <h2>Second organic product</h2>
        <a href="/dp/B000000007"><img class="s-image" src="https://example.com/organic-2.jpg" /></a>
        </article>
      </div>

      <section class="custom-sponsored-module">
        <span>Sponsored</span>
        <article data-asin="B000000005">
          <h3>Unknown sponsored product</h3>
          <a href="/dp/B000000005"><img src="https://example.com/unknown.jpg" /></a>
        </article>
      </section>

      <section class="bottom-sponsored-video-module">
        <video></video>
        <span>Sponsored</span>
        <a href="/stores/video-brand"><span class="brand-logo">Video brand</span></a>
        <h2>Video brand headline</h2>
        <div class="puis-card-container">
          <div class="puisg-row">
            <article data-asin="B000000008">
              <h3>Sponsored Brands video product</h3>
              <a href="/dp/B000000008"><img src="https://example.com/brand-video.jpg" /></a>
            </article>
          </div>
        </div>
      </section>

      <section style="display: none">
        <span>Sponsored</span>
        <article data-asin="B000000009">
          <h3 aria-label="Sponsored Ad - hidden preloaded product">Hidden preloaded product</h3>
          <a href="/sspa/click?url=%2Fdp%2FB000000009">
            <img src="https://example.com/hidden.jpg" />
          </a>
        </article>
      </section>

      <section aria-hidden="true">
        <span>Sponsored</span>
        <article data-asin="B000000010">
          <h3>Hidden accessibility product</h3>
          <a href="/dp/B000000010"><img src="https://example.com/hidden-aria.jpg" /></a>
        </article>
      </section>
    </div>

    <aside id="rightResultsATF" data-component-type="sponsored-display">
      <span>Sponsored</span>
      <article data-asin="B000000006">
        <h3>Display product</h3>
        <a href="/dp/B000000006"><img src="https://example.com/display.jpg" /></a>
      </article>
    </aside>
  </main>
`;

const dom = new JSDOM(html, { url: 'https://www.amazon.com/s?k=mirror' });
Object.assign(globalThis, {
  document: dom.window.document,
  Node: dom.window.Node,
  Element: dom.window.Element,
  HTMLElement: dom.window.HTMLElement,
  HTMLAnchorElement: dom.window.HTMLAnchorElement,
  HTMLImageElement: dom.window.HTMLImageElement,
});

async function main() {
  const items = await collectAmazonSearchPageData({ pattern: 'pattern-2' });
  assert.equal(items.length, 8);

  const byAsin = new Map(items.map((item) => [item.asin, item]));
  assert.equal(byAsin.get('B000000003')?.adType, 'organic');
  assert.equal(byAsin.get('B000000004')?.adType, 'sponsoredProducts');
  assert.equal(byAsin.get('B000000004')?.price, '$33.93');
  assert.equal(byAsin.get('B000000004')?.strikethroughPrice, '$39.99');
  assert.equal(byAsin.get('B000000004')?.link, 'https://www.amazon.com/dp/B000000004');
  assert.equal(byAsin.get('B000000004')?.placement, 'middleOfSearch');
  assert.equal(byAsin.get('B000000001')?.adType, 'sponsoredBrands');
  assert.equal(byAsin.get('B000000002')?.adType, 'sponsoredBrands');
  assert.equal(byAsin.get('B000000001')?.placement, 'topOfSearch');
  assert.equal(byAsin.get('B000000001')?.placementId, byAsin.get('B000000002')?.placementId);
  assert.equal(byAsin.get('B000000001')?.adProductIndex, 1);
  assert.equal(byAsin.get('B000000002')?.adProductIndex, 2);
  assert.equal(byAsin.get('B000000005')?.adType, 'sponsoredProducts');
  assert.equal(byAsin.get('B000000005')?.placement, 'bottomOfSearch');
  assert.equal(byAsin.get('B000000006')?.adType, 'sponsoredDisplay');
  assert.equal(byAsin.get('B000000006')?.placement, 'sidebar');
  assert.equal(byAsin.get('B000000008')?.adType, 'sponsoredBrands');
  assert.equal(byAsin.get('B000000008')?.adFormat, 'video');
  assert.equal(byAsin.get('B000000008')?.placement, 'bottomOfSearch');
  assert.equal(items.filter((item) => item.asin === 'B000000008').length, 1);
  assert.equal(byAsin.has('B000000009'), false);
  assert.equal(byAsin.has('B000000010'), false);
  assert.deepEqual(
    items.map((item) => item.pageRank),
    [1, 2, 3, 4, 5, 6, 7, 8],
  );

  const hiddenDisclosureDom = new JSDOM(
    `
      <main id="search">
        <div class="s-main-slot">
          <article data-component-type="s-search-result" data-asin="B000000011">
            <h2>Visible organic product</h2>
            <span
              aria-label="Sponsored Ad - hidden disclosure"
              style="position: absolute; width: 1px; height: 1px; overflow: hidden; clip: rect(0, 0, 0, 0)"
            >Sponsored</span>
            <a href="/sspa/click?url=%2Fdp%2FB000000011">
              <img class="s-image" src="https://example.com/visible-product.jpg" />
            </a>
          </article>
        </div>
      </main>
    `,
    { url: 'https://www.amazon.com/s?k=mirror' },
  );
  Object.assign(globalThis, {
    document: hiddenDisclosureDom.window.document,
    Node: hiddenDisclosureDom.window.Node,
    Element: hiddenDisclosureDom.window.Element,
    HTMLElement: hiddenDisclosureDom.window.HTMLElement,
    HTMLAnchorElement: hiddenDisclosureDom.window.HTMLAnchorElement,
    HTMLImageElement: hiddenDisclosureDom.window.HTMLImageElement,
  });
  const hiddenDisclosureItems = await collectAmazonSearchPageData({ pattern: 'pattern-2' });
  assert.equal(hiddenDisclosureItems.length, 1);
  assert.equal(hiddenDisclosureItems[0].adType, 'organic');

  const groupedAdsDom = new JSDOM(
    `
      <main id="search">
        <div class="s-main-slot">
          <section class="shared-sponsored-products">
            <span aria-label="Sponsored">Sponsored</span>
            <a href="/stores/reolink" style="display: none">Hidden brand destination</a>
            <article data-asin="B000000012">
              <h3>Grouped sponsored product one</h3>
              <a href="/dp/B000000012"><img src="https://example.com/grouped-1.jpg" /></a>
            </article>
            <article data-asin="B000000013">
              <h3>Grouped sponsored product two</h3>
              <a href="/dp/B000000013"><img src="https://example.com/grouped-2.jpg" /></a>
            </article>
          </section>
        </div>
        <aside class="left-sponsored-display" data-component-type="sponsored-display">
          <span aria-label="Sponsored">Sponsored</span>
          <article>
            <h3>Left rail display product</h3>
            <a href="/dp/B000000014"><img src="https://example.com/display-left.jpg" /></a>
          </article>
        </aside>
      </main>
    `,
    { url: 'https://www.amazon.com/s?k=mirror&page=8' },
  );
  Object.assign(globalThis, {
    document: groupedAdsDom.window.document,
    Node: groupedAdsDom.window.Node,
    Element: groupedAdsDom.window.Element,
    HTMLElement: groupedAdsDom.window.HTMLElement,
    HTMLAnchorElement: groupedAdsDom.window.HTMLAnchorElement,
    HTMLImageElement: groupedAdsDom.window.HTMLImageElement,
  });
  const groupedAds = await collectAmazonSearchPageData({ pattern: 'pattern-2' });
  const groupedAdsByAsin = new Map(groupedAds.map((item) => [item.asin, item]));
  assert.equal(groupedAdsByAsin.get('B000000012')?.adType, 'sponsoredProducts');
  assert.equal(groupedAdsByAsin.get('B000000013')?.adType, 'sponsoredProducts');
  assert.equal(
    groupedAdsByAsin.get('B000000012')?.placementId,
    groupedAdsByAsin.get('B000000013')?.placementId,
  );
  assert.equal(groupedAdsByAsin.get('B000000012')?.adProductIndex, 1);
  assert.equal(groupedAdsByAsin.get('B000000013')?.adProductIndex, 2);
  assert.equal(groupedAdsByAsin.get('B000000014')?.adType, 'sponsoredDisplay');
  assert.equal(groupedAdsByAsin.get('B000000014')?.placement, 'sidebar');

  console.log('Amazon search parser fixtures passed.');
}

main().catch((error) => {
  console.error(error);
  process.exitCode = 1;
});