// ==UserScript==
// @name Scholar's Toolkit
// @namespace greatest.deepsurf.us
// @version 3.0
// @description Checks for free PDFs from Google Scholar, Sci-Hub, LibGen, Anna's Archive, Sci-net, Semantic Scholar, Unpaywall, Openrxiv (medRxiv/bioRxiv), and ArXiv. When hovering a DOI, it also displays journal name, ISSN, publisher, metrics (SJR, H-Index, JIF), and integrity status (PubPeer, Retraction Database, Beall's Predatory List).
// @author Bui Quoc Dung (with community contributions)
// @match *://*/*
// @grant GM_xmlhttpRequest
// @connect *
// @license AGPL-3.0-or-later
// ==/UserScript==
(function () {
'use strict';
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%; margin-top: 6px; }
.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,
});
});
}
async function httpGet(url) {
return await httpRequest({ method: 'GET', url });
}
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(/\[(PDF|Maybe|COM|RET|PRE)\]/g, '<b>$&</b>');
cell.className = isNo ? 'status-no' : 'status-yes';
cell.appendChild(link);
}
async function fetchCrossref(doi) {
try {
const r = await httpGet(`https://api.crossref.org/works/${doi}`);
const js = JSON.parse(r.responseText).message;
return {
journal: (js["container-title"]?.[0] || "").replace(/&/g, '&'),
publisher: (js.publisher || "").replace(/&/g, '&'),
issn: js.ISSN?.[0] || ""
};
} catch(e) { return {}; }
}
function querySJRByISSN(issn, cb){
const SJR_SEARCH_URL = 'https://www.scimagojr.com/journalsearch.php?q=';
const SJR_BASE_URL = 'https://www.scimagojr.com/';
if(!issn) return cb(null);
GM_xmlhttpRequest({
method:'GET', url:SJR_SEARCH_URL+encodeURIComponent(issn),
onload:res=>{
const doc=new DOMParser().parseFromString(res.responseText,"text/html");
const link=doc.querySelector('.search_results a'); if(!link) return cb(null);
const url=SJR_BASE_URL+link.getAttribute('href');
GM_xmlhttpRequest({
method:'GET',url,
onload:r2=>{
const d=new DOMParser().parseFromString(r2.responseText,"text/html");
const ps=d.querySelectorAll('p.hindexnumber'); if(ps.length<2) return cb(null);
const sjr=ps[0].childNodes[0]?.textContent.trim();
const quart=ps[0].querySelector('span')?.textContent.trim();
const h=ps[1].textContent.trim();
let text = sjr ? `SJR: ${sjr}` : "SJR: N/A";
if(quart) text += ` (${quart})`;
if(h) text += ` | H-index: ${h}`;
cb({text,link:url});
},
onerror:()=>cb(null)
});
},
onerror:()=>cb(null)
});
}
function queryJIFByISSN(issn,cb){
const WOS_JOURNAL_URL = 'https://wos-journal.info/?jsearch=';
if(!issn) return cb(null);
GM_xmlhttpRequest({
method:'GET', url:WOS_JOURNAL_URL+encodeURIComponent(issn),
onload:res=>{
const doc=new DOMParser().parseFromString(res.responseText,"text/html");
const t=doc.querySelectorAll('.title.col-4.col-md-3');
const c=doc.querySelectorAll('.content.col-8.col-md-9');
if(!t.length||t.length!==c.length) return cb(null);
let j=null;
for(let i=0;i<t.length;i++){
if(t[i].textContent.trim()==='Journal Impact Factor (JIF):'){
j=c[i].textContent.trim(); break;
}
}
cb(j && !isNaN(j) ? {value:j,link:WOS_JOURNAL_URL+issn} : null);
},
onerror:()=>cb(null)
});
}
async function checkBeallsList(journalName, publisherName, cell) {
const mainUrl = 'https://beallslist.net/';
if (!journalName && !publisherName) {
return updateLink(cell, '[No] Beall', mainUrl, true);
}
const getList = async (url, cacheKey, selector = ".entry-content ul li") => {
const cached = sessionStorage.getItem(cacheKey);
if (cached) return new Set(JSON.parse(cached));
const response = await httpRequest({ method: 'GET', url });
const doc = new DOMParser().parseFromString(response.responseText, "text/html");
const items = doc.querySelectorAll(selector);
const itemSet = new Set();
items.forEach(item => {
const text = item.textContent.trim().toLowerCase();
if (text) itemSet.add(text);
});
sessionStorage.setItem(cacheKey, JSON.stringify(Array.from(itemSet)));
return itemSet;
};
try {
const [journalSet, publisherSet] = await Promise.all([
getList('https://beallslist.net/standalone-journals/', 'beallsListJournals'),
getList('https://beallslist.net/', 'beallsListPublishers')
]);
const lowerJournal = journalName ? journalName.toLowerCase() : '';
if (lowerJournal && journalSet.has(lowerJournal)) {
updateLink(cell, '[PRE] Beall', mainUrl, false);
return;
}
const lowerPublisher = publisherName ? publisherName.toLowerCase() : '';
if (lowerPublisher) {
for (const predatoryPublisher of publisherSet) {
if (predatoryPublisher.includes(lowerPublisher) || lowerPublisher.includes(predatoryPublisher)) {
updateLink(cell, '[PRE] Beall', mainUrl, false);
return;
}
}
}
updateLink(cell, '[No] Beall', mainUrl, true);
} catch (error) {
console.error("Error checking Beall's List:", error);
updateLink(cell, '[No] Beall', mainUrl, true);
}
}
function checkPubPeer(doi, cell) {
const encodedDoi = encodeURIComponent(doi);
const searchPageUrl = `https://pubpeer.com/search?q=${encodedDoi}`;
const pubpeerUrl = `https://pubpeer.com/search?q=${encodedDoi}`;
const updateFail = () => updateLink(cell, '[No]PubPeer', pubpeerUrl, true);
GM_xmlhttpRequest({
method: 'GET', url: searchPageUrl,
onload: function(response) {
const parser = new DOMParser();
const doc = parser.parseFromString(response.responseText, "text/html");
const metaTag = doc.querySelector('meta[name="csrf-token"]');
if (!metaTag) { return updateFail(); }
const token = metaTag.getAttribute('content');
const apiUrl = `https://pubpeer.com/api/search/?q=${encodedDoi}&token=${token}`;
GM_xmlhttpRequest({
method: 'GET', url: apiUrl, responseType: 'json',
onload: function(apiResponse) {
try {
const data = apiResponse.response;
if (data && data.publications && data.publications.length > 0) {
const commentCount = data.publications[0].comments_total;
const result = commentCount > 0 ? `[COM] PubPeer (${commentCount})` : '[No] PubPeer';
updateLink(cell, result, pubpeerUrl, commentCount === 0);
} else { updateFail(); }
} catch (e) { updateFail(); }
},
onerror: updateFail
});
},
onerror: updateFail
});
}
async function checkRetraction(doi, cell) {
const rwdUrl = 'https://retractiondatabase.org/RetractionSearch.aspx';
try {
const r = await httpGet(`https://api.crossref.org/works/${doi}`);
const js = JSON.parse(r.responseText).message;
const isRetracted = (js['updated-by'] || []).some(u => u.type === 'retraction');
if (isRetracted) {
updateLink(cell, '[RET] RWD', rwdUrl, false);
} else {
updateLink(cell, '[No] RWD', rwdUrl, true);
}
} catch (e) {
updateLink(cell, '[No] RWD', rwdUrl, true);
}
}
async function checkGoogleScholar(doi, cell) {
const GOOGLE_SCHOLAR_URL = 'https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=';
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 SCIHUB_URL = 'https://tesble.com/';
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 LIBGEN_URL = 'https://libgen.bz/';
const LIBGEN_SEARCH_URL = LIBGEN_URL + 'index.php?req=';
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 ANNA_URL = 'https://annas-archive.org';
const ANNA_SCIDB_URL = ANNA_URL + '/scidb/';
const ANNA_CHECK_URL = ANNA_URL + '/search?index=journals&q=';
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 SCINET_URL = 'https://sci-net.xyz/';
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 UNPAYWALL_EMAIL = '[email protected]';
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>`;
}
async function checkArxiv(doi, cell) {
const baseUrl = 'https://arxiv.org';
const searchUrl = `${baseUrl}/search/?query=${encodeURIComponent(doi)}&searchtype=all`;
if (!doi.includes('10.48550/arXiv.'))
return updateLink(cell, '[No] ArXiv', searchUrl, true);
const arxivId = doi.replace(/^10\.48550\/arXiv\./i, '');
const pdfUrl = `${baseUrl}/pdf/${arxivId}.pdf`;
try {
const { status } = await httpRequest({ method: 'HEAD', url: pdfUrl });
updateLink(cell, status === 200 ? '[PDF] ArXiv' : '[No] ArXiv',
status === 200 ? pdfUrl : searchUrl, status !== 200);
} catch {
updateLink(cell, '[No] ArXiv', searchUrl, true);
}
}
function removeCurrentPopup() {
if (currentPopup) { currentPopup.remove(); currentPopup = null; }
}
async function getDoiFromLink(linkElement) {
const DOI_REGEX = /\b(10\.\d{4,}(?:\.\d+)*\/[^\s?#"&]+)/i;
if (linkElement.dataset.doi) return linkElement.dataset.doi;
if (linkElement.dataset.doiFailed) return null;
const url = linkElement.href.toLowerCase();
const keywords = [
'doi','article','journal','abs','content','abstract',
'pubmed','document','fulltext','research','mdpi','springer'
];
if (!keywords.some(k => url.includes(k))) {
linkElement.dataset.doiFailed = 'true';
return null;
}
const cleanDOI = doi => (doi.match(DOI_REGEX)?.[1]?.trim() ?? doi.trim()).replace(/\/(meta|full|abs|pdf)\/?$/i, "").replace(/%e2%81%a9/gi, '').replace(/[.,]+$/, '').replace(/\/[a-zA-Z0-9-]{20,}\/?$/i, '');
let doi =
url.match(DOI_REGEX)?.[1];
if (!doi) {
try {
const res = await httpRequest({ method: 'GET', url: linkElement.href });
doi = res.responseText.match(DOI_REGEX)?.[1];
} catch {}
}
if (doi) {
const final = cleanDOI(doi);
linkElement.dataset.doi = final;
return final;
} 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: ${doi}`;
popup.appendChild(doiLine);
const journalLine = document.createElement('div');
journalLine.className = 'doi-header';
journalLine.textContent = '...';
popup.appendChild(journalLine);
const metricsLine = document.createElement('div');
metricsLine.className = 'doi-header';
metricsLine.innerHTML = `<span>SJR: ...</span> | <span>JIF: ...</span>`;
popup.appendChild(metricsLine);
(async () => {
const { journal, issn, publisher } = await fetchCrossref(doi);
journalLine.textContent = `${journal || 'N/A'} | ISSN: ${issn || 'N/A'}`;
if (cellRefs.Beall) {
checkBeallsList(journal, publisher, cellRefs.Beall);
}
if (!issn) {
metricsLine.innerHTML = `<span>SJR: N/A</span> | <span>JIF: N/A</span>`;
return;
}
querySJRByISSN(issn, (sjrData) => {
const sjrSpan = metricsLine.querySelector('span:first-child');
if (sjrData && sjrData.text) {
sjrSpan.innerHTML = `<a href="${sjrData.link}" target="_blank" rel="noopener noreferrer">${sjrData.text}</a>`;
} else {
sjrSpan.textContent = 'SJR: N/A';
}
});
queryJIFByISSN(issn, (jifData) => {
const jifSpan = metricsLine.querySelector('span:last-child');
if (jifData && jifData.value) {
jifSpan.innerHTML = `<a href="${jifData.link}" target="_blank" rel="noopener noreferrer">JIF: ${jifData.value}</a>`;
} else {
jifSpan.textContent = 'JIF: N/A';
}
});
})();
const table = document.createElement('table');
const rows = [
['RWD', 'PubPeer', 'Beall' ],
['GS', 'Semantic', 'Unpay' ],
['SH', 'LG', 'Anna' ],
['Sci', 'OpenRxiv' , 'Arxiv']
];
const cellRefs = {};
rows.forEach(rowKeys => {
const row = table.insertRow();
rowKeys.forEach(key => {
const cell = row.insertCell();
cell.textContent = '...';
cell.className = 'status-checking';
cellRefs[key] = cell;
});
});
popup.appendChild(table);
const checks = {
PubPeer: checkPubPeer,
RWD: checkRetraction,
GS: checkGoogleScholar,
Semantic: checkSemanticScholar,
Unpay: checkUnpaywall,
SH: checkSciHub,
LG: checkLibgen,
Anna: checkAnna,
Sci: checkSciNet,
OpenRxiv: checkOpenRxiv,
Arxiv: checkArxiv
};
for (const key in checks) {
if(cellRefs[key]) {
checks[key](doi, cellRefs[key]);
}
}
popup.addEventListener('mouseenter', () => clearTimeout(hideTimeout));
popup.addEventListener('mouseleave', () => removeCurrentPopup());
document.body.appendChild(popup);
popup.style.top = `${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 || link.dataset.doiListenerAttached) return;
if (link.closest('.doi-enhancer-popup')) return;
clearTimeout(hideTimeout);
removeCurrentPopup();
link.dataset.doiCheckInProgress = 'true';
try {
const doi = await getDoiFromLink(link);
if (doi) {
link.dataset.doiListenerAttached = 'true';
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');
}
});
})();
