Image Downloader & Preview

How it works

A scanner is injected on demand — never on page load — and collects every image the page references, not just the ones already painted. It reads currentSrc, walks srcset for the largest candidate, checks the common lazy-loading attributes, and sweeps CSS background images.

function bestFromSrcset(srcset) {
  let best = null, bestScore = -1;
  for (const part of srcset.split(',')) {
    const [url, descriptor = ''] = part.trim().split(/\s+/);
    // "1024w" → 1024, "2x" → 2000, so widths and densities never mix badly
    const score = descriptor.endsWith('w') ? parseFloat(descriptor)
                : descriptor.endsWith('x') ? parseFloat(descriptor) * 1000
                : 1;
    if (score > bestScore) { bestScore = score; best = url; }
  }
  return best;
}

Two download paths, for one specific reason

A single image goes through the downloads API. Chrome fetches it with the page's own credentials, so there is no CORS involved and no host permission needed:

chrome.downloads.download({ url: img.src, filename, saveAs: false });

A ZIP is different — the bytes have to be in the extension's hands to compress them. That means fetch, which means host permissions, which are requested for exactly the origins involved at the moment you click, and never up front:

const origins = [...new Set(images.map((i) => new URL(i.src).origin + '/*'))];
if (!(await chrome.permissions.contains({ origins }))) {
  await chrome.permissions.request({ origins });
}

const blob = await (await fetch(img.src, { cache: 'force-cache' })).blob();
zip.file(uniqueName(filename, used), blob);

Fetches run sequentially rather than all at once: two hundred parallel requests to one host gets throttled, and the progress bar stops meaning anything.

Naming

Modern CDN URLs frequently have no file extension — /photo/1234?w=800 is an image. Filtering on a .jpg suffix, which is the obvious approach, silently discards most images on most sites. Names are derived from the response's content type instead, and de-duplicated so several files called photo.jpg do not collapse into one ZIP entry.

const EXT_BY_MIME = {
  'image/jpeg': 'jpg', 'image/png': 'png', 'image/webp': 'webp',
  'image/avif': 'avif', 'image/gif': 'gif', 'image/svg+xml': 'svg',
};

In the popup

Filter by URL or alt text, hide anything below a minimum dimension, sort by size, and a "load all" button that scrolls the page to trigger lazy loading before rescanning.