// ==UserScript==
// @name Paper links to free PDFs
// @namespace greatest.deepsurf.us
// @version 1.9
// @description Checks for free PDFs from Google Scholar, Sci-Hub, LibGen, Anna's Archive, Sci-net, Semantic Scholar, Unpaywall, and Openrxiv (medRxiv/bioRxiv) when hovering over a DOI link.
// @author Bui Quoc Dung
// @match *://*/*
// @grant GM.xmlHttpRequest
// @connect *
// @license AGPL-3.0-or-later
// ==/UserScript==
(function () {
'use strict';
const SCIHUB_URL = 'https://tesble.com/';
const LIBGEN_URL = 'https://libgen.bz/';
const LIBGEN_SEARCH_URL = LIBGEN_URL + 'index.php?req=';
const ANNA_URL = 'https://annas-archive.org';
const ANNA_SCIDB_URL = ANNA_URL + '/scidb/';
const ANNA_CHECK_URL = ANNA_URL + '/search?index=journals&q=';
const SCINET_URL = 'https://sci-net.xyz/';
const GOOGLE_SCHOLAR_URL = 'https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=';
const UNPAYWALL_EMAIL = '[email protected]';
const DOI_REGEX = /\b(10\.\d{4,}(?:\.\d+)*\/(?:(?!["&'<>])\S)+)\b/i;
const styles = `
.doi-enhancer-popup {
position: absolute; z-index: 9999; background-color: white;
border: 1px solid #ccc; border-radius: 6px; padding: 6px;
box-shadow: 0 4px 8px rgba(0,0,0,0.2); font-family: sans-serif;
font-size: 13px; max-width: 600px;
}
.doi-enhancer-popup .doi-header {
margin-bottom: 6px; color: #333; word-break: break-word;
padding-left: 7px;
}
.doi-enhancer-popup table { border-collapse: collapse; width: 100%; }
.doi-enhancer-popup td {
padding: 4px 6px; text-align: center;
border-right: 1px solid #eee; white-space: nowrap;
}
.doi-enhancer-popup td:last-child { border-right: none; }
.doi-enhancer-popup a { color: #007bff; text-decoration: none; }
.doi-enhancer-popup a:hover { text-decoration: underline; }
.doi-enhancer-popup .status-no a { color: #888; }
.doi-enhancer-popup .status-checking { color: #999; }
`;
const styleEl = document.createElement('style');
styleEl.textContent = styles;
document.head.appendChild(styleEl);
let currentPopup = null;
let hideTimeout = null;
function httpRequest(details) {
return new Promise((resolve, reject) => {
GM.xmlHttpRequest({
...details,
timeout: 15000,
onload: resolve,
onerror: reject,
ontimeout: reject,
});
});
}
function updateLink(cell, text, href, isNo = false) {
cell.innerHTML = '';
const link = document.createElement('a');
link.href = href;
link.target = '_blank';
link.rel = 'noopener noreferrer';
link.innerHTML = text.replace('[PDF]', '<b>[PDF]</b>').replace('[Maybe]', '<b>[Maybe]</b>');
cell.className = isNo ? 'status-no' : 'status-yes';
cell.appendChild(link);
}
async function checkGoogleScholar(doi, cell) {
const url = GOOGLE_SCHOLAR_URL + encodeURIComponent(doi);
try {
const res = await httpRequest({ method: 'GET', url });
const doc = new DOMParser().parseFromString(res.responseText, 'text/html');
const gsLink = doc.querySelector('.gs_or_ggsm a');
if (gsLink) updateLink(cell, '[PDF] Google', gsLink.href);
else updateLink(cell, '[No] Google', url, true);
} catch {
updateLink(cell, '[No] Google', url, true);
}
}
async function checkSciHub(doi, cell) {
const url = SCIHUB_URL + doi;
try {
const res = await httpRequest({ method: 'GET', url });
const hasPDF = /iframe|embed/.test(res.responseText);
updateLink(cell, hasPDF ? '[PDF] Sci-Hub' : '[No] Sci-Hub', url, !hasPDF);
} catch {
updateLink(cell, '[No] Sci-Hub', url, true);
}
}
async function checkLibgen(doi, cell) {
const url = LIBGEN_SEARCH_URL + encodeURIComponent(doi);
try {
const res = await httpRequest({ method: 'GET', url });
const doc = new DOMParser().parseFromString(res.responseText, 'text/html');
const linkEl = doc.querySelector('.table.table-striped a[href^="edition.php?id="]');
if (linkEl) {
const detailUrl = LIBGEN_URL + linkEl.getAttribute('href');
const detailRes = await httpRequest({ method: 'GET', url: detailUrl });
const detailDoc = new DOMParser().parseFromString(detailRes.responseText, 'text/html');
const hasPDF = !!detailDoc.querySelector('table');
updateLink(cell, hasPDF ? '[PDF] LibGen' : '[No] LibGen', url, !hasPDF);
} else updateLink(cell, '[No] LibGen', url, true);
} catch {
updateLink(cell, '[No] LibGen', url, true);
}
}
async function checkAnna(doi, cell, retry = 0) {
const checkUrl = ANNA_CHECK_URL + encodeURIComponent(doi);
const directUrl = ANNA_SCIDB_URL + doi;
try {
const res = await httpRequest({ method: 'GET', url: checkUrl });
const text = res.responseText;
if (text.includes("Rate limited") && retry < 10) {
setTimeout(() => checkAnna(doi, cell, retry + 1), 5000);
return;
}
const doc = new DOMParser().parseFromString(text, 'text/html');
const found = doc.querySelector('.mt-4.uppercase.text-xs.text-gray-500') ||
[...doc.querySelectorAll('div.text-gray-500')].some(div => div.textContent.includes(doi));
if (found) {
const res2 = await httpRequest({ method: 'GET', url: directUrl });
const doc2 = new DOMParser().parseFromString(res2.responseText, 'text/html');
const hasPDF = doc2.querySelector('.pdfViewer, #viewerContainer, iframe[src*="viewer.html?file="]');
updateLink(cell, hasPDF ? '[PDF] Anna' : '[Maybe] Anna', directUrl);
} else updateLink(cell, '[No] Anna', checkUrl, true);
} catch {
updateLink(cell, '[No] Anna', checkUrl, true);
}
}
async function checkSciNet(doi, cell) {
const url = SCINET_URL + doi;
try {
const res = await httpRequest({ method: 'GET', url });
const hasPDF = /iframe|pdf|embed/.test(res.responseText);
updateLink(cell, hasPDF ? '[PDF] Sci-net' : '[No] Sci-net', url, !hasPDF);
} catch {
updateLink(cell, '[No] Sci-net', url, true);
}
}
async function checkSemanticScholar(doi, cell) {
const apiUrl = `https://api.semanticscholar.org/graph/v1/paper/${encodeURIComponent(doi)}?fields=title,openAccessPdf`;
try {
const res = await httpRequest({ method: 'GET', url: apiUrl });
const data = JSON.parse(res.responseText);
const pdfUrl = data?.openAccessPdf?.url;
if (pdfUrl) updateLink(cell, '[PDF] Semantic', pdfUrl);
else updateLink(cell, '[No] Semantic', `https://www.semanticscholar.org/search?q=${encodeURIComponent(doi)}`, true);
} catch {
updateLink(cell, '[No] Semantic', `https://www.semanticscholar.org/search?q=${encodeURIComponent(doi)}`, true);
}
}
async function checkUnpaywall(doi, cell) {
const url = `https://api.unpaywall.org/v2/${encodeURIComponent(doi)}?email=${encodeURIComponent(UNPAYWALL_EMAIL)}`;
try {
const res = await httpRequest({ method: 'GET', url });
const data = JSON.parse(res.responseText);
if (data?.is_oa && data.best_oa_location?.url)
updateLink(cell, '[PDF] Unpaywall', data.best_oa_location.url);
else updateLink(cell, '[No] Unpaywall', `https://unpaywall.org/`, true);
} catch {
updateLink(cell, '[No] Unpaywall', `https://unpaywall.org/`, true);
}
}
async function checkOpenRxiv(doi, cell) {
const doiUrl = `https://doi.org/${doi}`;
try {
const res = await httpRequest({ method: 'GET', url: doiUrl, redirect: 'follow' });
if (res.finalUrl) {
const final = res.finalUrl;
if (final.includes('biorxiv.org')) {
const pdfUrl = final.replace(/\/$/, '') + '.full.pdf';
updateLink(cell, '[PDF] BioRxiv', pdfUrl);
return;
}
if (final.includes('medrxiv.org')) {
const pdfUrl = final.replace(/\/$/, '') + '.full.pdf';
updateLink(cell, '[PDF] MedRxiv', pdfUrl);
return;
}
}
} catch (e) {
}
updateLink(cell, '[No]', `https://doi.org/${doi}`, true);
cell.innerHTML += `
<a href="https://www.biorxiv.org/content/${doi}" target="_blank">BioRxiv</a>
<a href="https://www.medrxiv.org/content/${doi}" target="_blank">MedRxiv</a>`;
}
function removeCurrentPopup() {
if (currentPopup) { currentPopup.remove(); currentPopup = null; }
}
async function getDoiFromLink(linkElement) {
if (linkElement.dataset.doi) return linkElement.dataset.doi;
if (linkElement.dataset.doiFailed) return null;
const url = linkElement.href;
const keywords = ['doi', 'article', 'journal', 'abs', 'content', 'abstract', 'pubmed', 'document', 'fulltext', 'research', 'mdpi', 'springer'];
if (!keywords.some(k => url.toLowerCase().includes(k))) {
linkElement.dataset.doiFailed = 'true';
return null;
}
const cleanDOI = doi => {
const match = doi.match(/^10\.\d{4,}(?:\.\d+)*\/[^\s\/?#]+/);
return match ? match[0] : doi.trim();
};
let doi = url.match(/\/doi\/(?:abs|full|pdf)?\/(10\.\d{4,}(?:\.\d+)*\/[^?#]+)/i)?.[1] ||
url.match(/10\.\d{4,}(?:\.\d+)*\/[^\s"'<>{}[\]|\\^`]+/)?.[0];
if (!doi) {
try {
const res = await httpRequest({ method: 'GET', url });
const html = res.responseText;
doi = html.match(/10\.\d{4,}(?:\.\d+)*\/[^\s"'<>{}[\]()]+/)?.[0];
} catch {}
}
if (doi) {
linkElement.dataset.doi = cleanDOI(doi);
return linkElement.dataset.doi;
} else {
linkElement.dataset.doiFailed = 'true';
return null;
}
}
function showPopup(linkElement, doi, mouseX, mouseY) {
clearTimeout(hideTimeout);
removeCurrentPopup();
const popup = document.createElement('div');
popup.className = 'doi-enhancer-popup';
currentPopup = popup;
const doiLine = document.createElement('div');
doiLine.className = 'doi-header';
doiLine.textContent = `DOI: ${doi}`;
popup.appendChild(doiLine);
const table = document.createElement('table');
const row1 = table.insertRow();
const cellGS = row1.insertCell(); cellGS.textContent = '...'; cellGS.className = 'status-checking';
const cellSemantic = row1.insertCell(); cellSemantic.textContent = '...'; cellSemantic.className = 'status-checking';
const cellUnpay = row1.insertCell(); cellUnpay.textContent = '...'; cellUnpay.className = 'status-checking';
const row2 = table.insertRow();
const cellSH = row2.insertCell(); cellSH.textContent = '...'; cellSH.className = 'status-checking';
const cellLG = row2.insertCell(); cellLG.textContent = '...'; cellLG.className = 'status-checking';
const cellAnna = row2.insertCell(); cellAnna.textContent = '...'; cellAnna.className = 'status-checking';
const row3 = table.insertRow();
const cellSci = row3.insertCell(); cellSci.textContent = '...'; cellSci.className = 'status-checking';
const cellOpenRxiv = row3.insertCell(); cellOpenRxiv.textContent = '...'; cellOpenRxiv.className = 'status-checking';
const cellDummy = row3.insertCell(); cellDummy.textContent = ''; cellDummy.className = 'status-no';
popup.appendChild(table);
checkGoogleScholar(doi, cellGS);
checkSemanticScholar(doi, cellSemantic);
checkUnpaywall(doi, cellUnpay);
checkSciHub(doi, cellSH);
checkLibgen(doi, cellLG);
checkAnna(doi, cellAnna);
checkSciNet(doi, cellSci);
checkOpenRxiv(doi, cellOpenRxiv);
popup.addEventListener('mouseenter', () => clearTimeout(hideTimeout));
popup.addEventListener('mouseleave', () => removeCurrentPopup());
document.body.appendChild(popup);
popup.style.top = `${mouseY + 10}px`;
popup.style.left = `${mouseX + 10}px`;
}
document.addEventListener('mouseover', async (event) => {
const link = event.target.closest('a');
if (!link || !link.href || link.dataset.doiCheckInProgress) return;
if (link.closest('.doi-enhancer-popup')) return;
clearTimeout(hideTimeout);
removeCurrentPopup();
link.dataset.doiCheckInProgress = 'true';
try {
const doi = await getDoiFromLink(link);
if (doi) {
link.addEventListener('mouseenter', (e) => showPopup(link, doi, e.pageX, e.pageY));
link.addEventListener('mouseleave', () => hideTimeout = setTimeout(removeCurrentPopup, 150));
showPopup(link, doi, event.pageX, event.pageY);
}
} finally {
link.removeAttribute('data-doiCheckInProgress');
}
});
})();
