Extract question answers from GRE Magoosh Review.

magoosh-gre-question-scraper.js

function scrapeQuestionsFromMagooshReviewPage() {
  // I think this only works on Firefox, since they add the tbody element.
  const table = $('tbody tr')
  const questions = [];

  for(let node of table) {
    const cleanText = t => t.replace(/\n/g,'');
    const parsedNode = {
      correct: node.querySelector('.correct-sprite') != null,
      title: node.querySelectorAll('td')[1].textContent,
      section: node.querySelectorAll('td')[2].textContent,
      subject: node.querySelectorAll('td')[3].textContent,
      difficulty: node.querySelectorAll('td')[4].textContent,
      userPace: node.querySelectorAll('td')[5].textContent,
      globalPace: node.querySelectorAll('td')[6].textContent,
      latestDate: node.querySelectorAll('td')[7].textContent,
    }
    const processedNode = {
      ...parsedNode,
      difficulty: cleanText(parsedNode.difficulty),
      userPace: cleanText(parsedNode.userPace),
      globalPace: cleanText(parsedNode.globalPace),
    };
    questions.push(processedNode);
  }
  return questions;
}

function groupBy(data, groupingKey) {
  return data.reduce(
    (accumulated, currentSample) => {
      (accumulated[currentSample[groupingKey]] = accumulated[currentSample[groupingKey]] || []).push(currentSample);
      return accumulated;
    }, {}
  );
}

function performanceSummary(questions) {
  const groups = groupBy(questions, 'subject');
  const subjects = Object.keys(groups);
  const resultsBySubject = subjects.reduce((accum, curr) => {
    const currentSubject = groups[curr];
    const currentSubjectResults = currentSubject.map(s => s.correct);
    const currentSubjectScore = currentSubjectResults.reduce((accum, curr) => accum + curr);
    const scoreFractionString = `${+currentSubjectScore}/${currentSubjectResults.length}`;
    const scorePercentage = Math.round(+currentSubjectScore/currentSubjectResults.length * 10000) / 100;
    accum[curr] = `${scoreFractionString} (${scorePercentage}%)`;
    return accum;
  }, {});
  return resultsBySubject;
}

const questions = scrapeQuestionsFromMagooshReviewPage();
const scoreSummary = performanceSummary(questions);

console.table(scoreSummary);