azon_seeker/src/page-worker/web-injectors/amazon-search-pagination.ts

/**
 * Resolves the page being collected. Amazon's selected pagination element can
 * point at the next page while the current results are still rendered, so the
 * document URL is the authoritative source whenever it carries a page number.
 */
export async function collectAmazonCurrentPage(): Promise<number> {
  const currentUrl = new URL(window.location.href);
  const parsePage = (value: string | null | undefined) => {
    const page = Number.parseInt(value || '', 10);
    return Number.isInteger(page) && page > 0 ? page : null;
  };

  const urlPage = parsePage(currentUrl.searchParams.get('page'));
  if (urlPage) return urlPage;

  const refPage = parsePage(
    /(?:^|_)pg_(\d+)(?:_|$)/.exec(currentUrl.searchParams.get('ref') || '')?.[1],
  );
  if (refPage) return refPage;

  const selected = document.querySelector<HTMLElement>('.s-pagination-item.s-pagination-selected');
  return parsePage(selected?.textContent?.trim()) || 1;
}

/**
 * Reads the next search-page URL without navigating from the injected script.
 * Navigation must happen only after browser.scripting.executeScript has returned,
 * otherwise Amazon can move the frame into BFCache and destroy the script context.
 */
export async function collectAmazonNextPageUrl() {
  const nextButton = document.querySelector<HTMLAnchorElement>('.s-pagination-next');
  if (!nextButton || nextButton.classList.contains('s-pagination-disabled')) return null;

  const href = nextButton.getAttribute('href');
  return href ? new URL(href, window.location.href).toString() : null;
}