beastie scraper

scraper.js

// each h2 is a section
var $h2s = $("h2");
console.log($h2s.length)
var alphabet = ['A','B','C','D','E','F','G','H','I','J','K'];
function Quiz() {
    var self = this;
    self.concepts = [];
    self.print = function() {
        var txtPrint = "";
        // for each concept
        for (var i = 0; i < self.concepts.length; i++) {
            var curConcept = self.concepts[i];
            // print concept
            txtPrint += curConcept.name +"\n"
            // for each question in concept
            for (var j = 0; j < curConcept.questions.length;j++) {
                var curQuestion = curConcept.questions[j];
                txtPrint += curQuestion.number+"\n";
                txtPrint += curQuestion.text + "\n"
            }
        }
        return txtPrint;
    }
}
var quiz = new Quiz();
function Concept(name) {
    var self = this;
    self.name = name;
    self.questions = [];
}
function Question(number, text) {
    var self = this;
    self.number = number;
    self.text = text;
}
for (var i = 0; i < $h2s.length; i++) {
    var $h2 = $($h2s[i]);
    var $itemsInSection = $h2.nextUntil("h2");
    quiz.concepts.push(new Concept($h2.text().replace("Concept: ","").trim()));
    console.log($h2.text())
    for (var j = 0;j < $itemsInSection.length;j++) {
        var curQuestion = new Question();
        var $li = $($itemsInSection[j]).find("li").first();
        var questionText = $li.text().replace(
            $li.children().text()
                .replace("strong","")
                .replace("weak","")
                .replace('nn','')
                .trim()
                .replace('N		P','')
                .replace('B			N		P	A			N		P','')
                .trim()
            ,'')
                .replace(/\t/g, '')
                .replace(/\r\n/g,'')
                .replace(/\n/g,'')
                .trim();
        // if not t or f
        var $questionNumber = $li.parent();
        curQuestion.number = parseInt($questionNumber.attr("start"));
        console.log($questionNumber.attr("start"))
        var $answers = $li.children().children("li");
        //console.log($answers);
        var totalAnswer = questionText+"\n";
        for (var k = 0; k < $answers.length; k++) {
            var $answer = $($answers[k]);
            var letter = alphabet[k];
            var txtAnswer = letter + ". "+$answer.text().trim()
            .replace("\n","")
            .replace(/\t/g, '')+"\n";
            //console.log(txtAnswer)
            totalAnswer += txtAnswer;
        }
        console.log(totalAnswer)
        curQuestion.text = totalAnswer;
        quiz.concepts[i].questions.push(curQuestion);
    }
}
console.log(quiz.concepts);