Google Scholar Info Extractor

google_scholar_info.js

function getCitationInfo() {
  t = document.querySelector("table#gsc_rsb_st")
  cite_hist = document.querySelector("div.gsc_md_hist_b")
  overall = Array.from(t.querySelectorAll("tr")).map((tr, i) => {
    selector = i == 0 ? "th" : "td";
    row = Array.from(tr.querySelectorAll(selector))
    return row.map(td => td.innerText)
  })
  temporal = {
    years: Array.from(cite_hist.querySelectorAll("span.gsc_g_t")).map(s => +s.innerText),
    citations: Array.from(cite_hist.querySelectorAll("a.gsc_g_a")).map(a => +a.text)
  }
  return {
    overall,
    temporal
  }
}

function getPaperInfo(p) {

  link = p.querySelector("td.gsc_a_t a")
  url = link.getAttribute("data-href")
  title = link.text
  citations = +p.querySelector("td.gsc_a_c a").text
  year = +p.querySelector("td.gsc_a_y").innerText
  authors = p.querySelector("td.gsc_a_t div.gs_gray").innerText.split(", ")
  venue = p.querySelector("td.gsc_a_t div.gs_gray:last-child").innerText
  return {
    url,
    title,
    citations,
    year,
    authors,
    venue
  };
}

function getAllInfo() {
  papers_e = Array.from(document.querySelectorAll("#gsc_a_tw tr.gsc_a_tr"));
  papers = papers_e.map(getPaperInfo);
  img = document.querySelector("img#gsc_prf_pup-img").src
  name = document.querySelector("div#gsc_prf_in").innerText
  affiliation_a = document.querySelector("div.gsc_prf_il a")
  affiliation_url = affiliation_a.getAttribute("href")
  affiliation_name = affiliation_a.text
  affiliation = {
      url: affiliation_url,
      name: affiliation_name
    },
    citations = getCitationInfo()
  return {
    name,
    img,
    affiliation,
    citations,
    papers
  };
}

getAllInfo();