googlePlayCrawler.js

function pageFunction(context) {
    var PAGES = 100; // each page has 40 reviews
    // called on every page the crawler visits, use it to extract data from it
    var $ = context.jQuery;
    var result = [];
    var extractData = function(page) {
        if( page < PAGES ) {
            var api = "https://play.google.com/store/getreviews?authuser=0";
            $.ajax({  
                url: api,  
                type: 'POST',
                data: "reviewType=0&pageNum=" + page + "&id=com.instagram.android&reviewSortOrder=0&xhr=1",
                dataType: 'text'
            })
            .done(function(data, textStatus, jqXHR) {
                console.log("Reviews data fetched");
                var html = JSON.parse(data.substring(data.indexOf("[[")))[0][2];
                var jQueryObject = $($.parseHTML("<div>" + html + "</div>"));
                $(jQueryObject).find('.single-review').each(function(){
                    var item = {
                        author: $(this).find('.author-name').text().trim(),
                        date: $(this).find('.review-date').text().trim(),
                        rating: $(this).find('.star-rating-non-editable-container').attr('aria-label').trim(),
                        title: $(this).find('.review-title').text().trim(),
                        text: $(this).find('.review-title').parent().contents()
                                  .filter(function() {
                                    return this.nodeType === 3;
                                  }).text().trim()
                    };
                    result.push(item);
                });
                extractData(++page);
            })
            .fail(function(xhr, textStatus, errorThrown) {
                console.log("Reviews API call failed");
                context.finish();
            });
        } else {
            context.finish(result);
        }
    };
    extractData(0);
    context.willFinishLater();
}