ieee-download

name	ieee-download
description	Downloads PDF from IEEE Xplore articles. Requires institutional or subscriber access. Use when the user wants to download a paper PDF by article number.
argument-hint	[article number(s) or URL]
disable-model-invocation	true

IEEE Xplore PDF Download

Download PDF files from IEEE Xplore articles to the user's local disk.

Prerequisites

• The user must have access to the article (institutional subscription, open access, or personal subscription).
• If the article is behind a paywall and the user has no access, the download will fail.

PDF URL Patterns (Discovered via Testing)

IEEE Xplore has a two-layer PDF serving architecture:

Layer	URL	Purpose
Wrapper	{BASE_URL}/stamp/stamp.jsp?tp=&arnumber={ARNUMBER}	HTML page with iframe
Actual PDF	{BASE_URL}/stampPDF/getPDF.jsp?tp=&arnumber={ARNUMBER}&ref=	Direct PDF binary

Key discovery: The getPDF.jsp URL pattern is predictable — you can skip stamp.jsp entirely and navigate directly to the PDF. This cuts tool calls from 6 to 2 per paper.

Single Article Download (Optimized: 2 tool calls)

Step 1: Pre-check access, then navigate directly to PDF

First, check if the user has access by visiting the document page:

async () => {
  // Quick access check on the current document page
  const bodyText = document.body.innerText;
  const hasAccess = bodyText.includes('Access provided by');
  const accessLine = bodyText.match(/Access provided by[^\n]*/)?.[0] || '';
  return { hasAccess, accessLine };
}

If hasAccess is false, tell the user: "当前未登录或无权限访问此文章，请先在浏览器中完成机构登录。"

If access is confirmed, navigate directly to the getPDF URL (skip stamp.jsp):

navigate_page({
  url: "{BASE_URL}/stampPDF/getPDF.jsp?tp=&arnumber={ARNUMBER}&ref=",
  initScript: "Object.defineProperty(navigator, 'webdriver', {get: () => undefined})"
})

Step 2: Trigger download

(arnumber, title) => {
  if (document.contentType === 'application/pdf') {
    // Construct a readable filename
    const safeName = (title || '').replace(/[^\w\s-]/g, '').replace(/\s+/g, '_').substring(0, 60);
    const filename = arnumber + (safeName ? '-' + safeName : '') + '.pdf';
    const a = document.createElement('a');
    a.href = window.location.href;
    a.download = filename;
    document.body.appendChild(a);
    a.click();
    document.body.removeChild(a);
    return { downloaded: true, filename };
  }
  return { downloaded: false, contentType: document.contentType, error: 'Not a PDF. Access may be denied.' };
}

Single Article Download (Fallback: 4 tool calls)

Use this only if the direct getPDF URL doesn't work (e.g., unusual access controls).

Step 1: Navigate to stamp.jsp wrapper

navigate_page({
  url: "{BASE_URL}/stamp/stamp.jsp?tp=&arnumber={ARNUMBER}",
  initScript: "Object.defineProperty(navigator, 'webdriver', {get: () => undefined})"
})

Step 2: Extract actual PDF URL from iframe and redirect

async () => {
  await new Promise(r => setTimeout(r, 4000));

  // Find the PDF iframe
  const iframe = document.querySelector('iframe[src*="getPDF"]');
  if (iframe) {
    // Redirect to the PDF URL directly (saves one navigate_page call)
    window.location.href = iframe.src;
    return { status: 'redirecting', pdfUrl: iframe.src };
  }

  // Check if PDF loaded directly
  if (document.contentType === 'application/pdf') {
    return { status: 'direct', pdfUrl: window.location.href };
  }

  // Access denied — page redirected back to document page
  if (window.location.href.includes('/document/')) {
    return { status: 'no-access', message: 'Redirected to document page. User likely has no access to this PDF.' };
  }

  return { : , : ...(, ), : . };
}

Step 3: Wait for PDF to load and trigger download

async (arnumber) => {
  await new Promise(r => setTimeout(r, 3000));
  if (document.contentType === 'application/pdf') {
    const a = document.createElement('a');
    a.href = window.location.href;
    a.download = arnumber + '.pdf';
    document.body.appendChild(a);
    a.click();
    document.body.removeChild(a);
    return { downloaded: true, filename: arnumber + '.pdf' };
  }
  return { downloaded: false, contentType: document.contentType };
}

Batch Download (Sequential)

For downloading multiple papers efficiently, loop through a list of article numbers:

For each arnumber in list:
  1. navigate_page → {BASE_URL}/stampPDF/getPDF.jsp?tp=&arnumber={ARNUMBER}&ref=
  2. evaluate_script → check contentType, trigger download
  3. If contentType !== 'application/pdf' → log as failed, continue to next

Rate limiting: Wait at least 3 seconds between downloads to avoid triggering bot detection. The await new Promise(r => setTimeout(r, 3000)) in the evaluate_script provides this naturally.

Access pre-check: Before starting a batch download, verify the user's institutional access is active by checking any document page for "Access provided by". This avoids wasting time on papers the user can't download.

Batch Download from Search Results (UI-based)

On the search results page, there is a "Download PDFs" button. To use it:

1. Select articles using checkboxes
2. Click the "Download PDFs" button

(arnumbers) => {
  const items = document.querySelectorAll('.List-results-items .result-item');
  let selected = 0;
  items.forEach(item => {
    const titleLink = item.querySelector('h3 a[href*="/document/"]');
    const docNum = titleLink?.href?.match(/\/document\/(\d+)/)?.[1] || '';
    if (arnumbers.includes(docNum)) {
      const checkbox = item.querySelector('input[type="checkbox"]');
      if (checkbox && !checkbox.checked) {
        checkbox.click();
        selected++;
      }
    }
  });

  const downloadBtn = [...document.querySelectorAll('button, .Menu-item')].find(b => b.textContent.trim().includes('Download PDF'));
  if (downloadBtn) {
    downloadBtn.click();
    return { success: true, selected, message: 'Download initiated for selected articles.' };
  }

  return { error: 'Download button not found.' };
}

Access Denied Detection

When stamp.jsp or getPDF.jsp redirects back to the document page (URL contains /document/ instead of /stampPDF/), it means the user has no access to the PDF. Common causes:

• Not logged in (check for "Institutional Sign In" on page)
• Institution doesn't subscribe to this content (e.g., older IEEE standards)
• Session expired

Action: Tell the user to check their login status and institutional access.

Notes

• This skill is set to disable-model-invocation: true — it must be explicitly invoked with /ieee-download.
• Optimized path (direct getPDF.jsp): 2 tool calls per paper.
• Fallback path (stamp.jsp → iframe → getPDF): 3-4 tool calls per paper.
• The window.location.href = iframe.src redirect trick inside evaluate_script saves one navigate_page call.
• Open access articles can be downloaded without authentication.
• For subscription articles, the user must be authenticated (institutional access or personal subscription).
• After solving any captcha once, subsequent downloads in the same session work without interruption.
• If download fails, suggest the user check their access status or try logging in.