shein_review_2026-06-10

shein_review_en.js

// test page: https://kr.shein.com/Short-Sleeve-T-Shirt-Versatile-Short-Sleeve-Tee-Women-s-Short-Sleeve-Top-Slim-Fit-Solid-Color-Undershirt-Casual-Black-Summer-review-142262475.html?mall_code=1&goods_spu=z250419597558&local_site_query_flag=&sku=sz25041959755899894&store_code=9108550356&isLowestPriceProductOfBuyBox=NaN&cat_id=1738&sku_code=&isAppointMall=&isTrusted=true&_vts=1781142343247

// ============================================================
// ▶ User configuration (edit as needed)
// ============================================================
// Works from EITHER the product page (...-p-142262475.html) OR the
// dedicated reviews page (...-review-142262475.html).
// On the product page it auto-clicks "View more reviews" (which navigates),
// then resumes automatically on the reviews page via sessionStorage.
const MAX_PAGE_COUNT = 10;                          // Number of scroll-load batches (1 = current view only)
const WAIT_AFTER_LOAD = 5000;                       // Wait after page load / new batch (ms)
const WAIT_BEFORE_NAV = 1000;                       // Wait after scrolling, before checking (ms)
const REVIEW_LIST_SELECTOR = '.common-reviews-new__list-item';   // Individual review item selector
const VIEW_ALL_SELECTOR = '.common-reviews__pageViewAll';        // "View more reviews" button (product page)
const REVIEW_ID_ATTR = 'data-comment-id';           // Attribute used to de-duplicate reviews (falls back to text)
const CONTAINER_ID = '__review_collector';          // Collector container ID

// ============================================================
// ▶ Internal logic (no edits needed)
// ============================================================
(async () => {
  const sleep = (ms) => new Promise((r) => setTimeout(r, ms));

  // Build a stable de-dup key for a review element.
  // On the product page the id lives on the list item (REVIEW_ID_ATTR);
  // on the reviews page the item has no id, so we also probe descendants
  // for [data-comment-id]/[data-review-id], and finally fall back to text.
  const keyFor = (el) => {
    let id = el.getAttribute(REVIEW_ID_ATTR);
    if (!id) {
      const inner = el.querySelector('[data-comment-id], [data-review-id]');
      if (inner) id = inner.getAttribute('data-comment-id') || inner.getAttribute('data-review-id');
    }
    return ((id || el.textContent || '')).trim().slice(0, 120);
  };

  // 1. Create the collector container (reuse if it already exists)
  let container = document.getElementById(CONTAINER_ID);
  if (!container) {
    container = document.createElement('div');
    container.id = CONTAINER_ID;
    Object.assign(container.style, {
      position: 'fixed',
      right: '200px',
      bottom: '200px',
      zIndex: '999999',
      background: 'rgba(0, 0, 0, 0.85)',
      color: '#0f0',
      fontSize: '18px',
      fontWeight: 'bold',
      fontFamily: 'monospace',
      padding: '16px 24px',
      borderRadius: '12px',
      border: '2px solid #0f0',
      boxShadow: '0 4px 20px rgba(0,255,0,.3)',
      cursor: 'default',
      userSelect: 'none',
      overflow: 'hidden',
    });
    document.body.appendChild(container);
  }

  // Collection store (hidden area) — a display:none wrapper inside the container
  let store = container.querySelector('#__review_store');
  if (!store) {
    store = document.createElement('div');
    store.id = '__review_store';
    store.style.display = 'none';
    container.appendChild(store);
  }

  // Count display label
  let label = container.querySelector('#__review_label');
  if (!label) {
    label = document.createElement('div');
    label.id = '__review_label';
    container.appendChild(label);
  }

  const updateLabel = () => {
    label.textContent = `Reviews collected: ${store.children.length}`;
  };

  // Restore any previously saved progress (after navigating from the product page)
  const saved = sessionStorage.getItem('__review_store_html');
  if (saved && store.children.length === 0) {
    store.innerHTML = saved;
    updateLabel();
  }

  // De-dup set, rebuilt from whatever is already in the store
  const seen = new Set();
  [...store.children].forEach((el) => {
    const key = keyFor(el);
    if (key) seen.add(key);
  });

  // 2. Collect reviews currently in the DOM (de-duplicated)
  const collectCurrentPage = () => {
    const reviews = document.querySelectorAll(
      `${REVIEW_LIST_SELECTOR}:not(#${CONTAINER_ID} *)`
    );
    let added = 0;
    reviews.forEach((el) => {
      const key = keyFor(el);
      if (!key || seen.has(key)) return;
      seen.add(key);
      store.appendChild(el.cloneNode(true));
      added++;
    });
    console.log(`[Scraper] Collected ${added} new reviews (total: ${store.children.length})`);
    updateLabel();
  };

  // 3a. If we're on the PRODUCT page, jump to the reviews page first.
  //     Clicking "View more reviews" triggers a full navigation, so we save
  //     progress (like the Amazon script did) and let the script re-run on load.
  const onReviewsPage = /-review-\d+\.html/.test(location.pathname);
  if (!onReviewsPage) {
    collectCurrentPage(); // grab the 3 preview reviews shown on the product page
    const viewAll = document.querySelector(VIEW_ALL_SELECTOR);
    if (viewAll) {
      console.log('[Scraper] On product page — opening full reviews page…');
      sessionStorage.setItem('__review_store_html', store.innerHTML);
      viewAll.click();
      return; // page navigates; re-run this script on the reviews page to continue
    }
    console.log('[Scraper] No "View more reviews" button found — staying on this page.');
  }

  // 3b. Reviews page: load the next batch via infinite scroll.
  //     Returns true if more reviews appeared, false at end of list.
  const goNextPage = async () => {
    const before = document.querySelectorAll(REVIEW_LIST_SELECTOR).length;
    window.scrollTo({ top: document.documentElement.scrollHeight, behavior: 'smooth' });
    await sleep(WAIT_BEFORE_NAV);
    window.scrollTo(0, document.documentElement.scrollHeight);
    await sleep(WAIT_AFTER_LOAD);
    return document.querySelectorAll(REVIEW_LIST_SELECTOR).length > before;
  };

  // 4. Main loop
  console.log(`[Scraper] Starting collection — up to ${MAX_PAGE_COUNT} batches`);
  for (let page = 1; page <= MAX_PAGE_COUNT; page++) {
    collectCurrentPage();
    console.log(`[Scraper] Batch ${page}/${MAX_PAGE_COUNT} collected`);
    if (page === MAX_PAGE_COUNT) break;
    const moved = await goNextPage();
    if (!moved) {
      console.log('[Scraper] No more reviews loaded — collection finished');
      collectCurrentPage();
      break;
    }
  }

  console.log(`[Scraper] Collection complete! Total reviews: ${store.children.length}`);
  updateLabel();
  sessionStorage.removeItem('__review_store_html');

  // ▶ To retrieve the collected HTML later, run in the console:
  //   document.getElementById('__review_store').innerHTML
})();