Personal sample of using Node.js for processing web page contents using request and cheerio

scraper.js

var request = require('request');
var cheerio = require('cheerio');

var urls = [
    'https://developer.android.com/reference/com/google/android/gms/location/Geofence.html',
    'http://developer.android.com/reference/com/google/android/gms/maps/model/LatLng.html'
];

urls.forEach(function (elem) {
    request({
        method: 'GET',
        url: elem
    }, function(error, response, body) {
        if (!error && response.statusCode === 200)
        {
            $ = cheerio.load(body);

            var classObj = {};

            // Process header
            var header = $('#jd-header');
            classObj.name = header.find("[itemprop='name']").text().trim();
            classObj.classModifiers = header.contents().first().text().trim().replace(/\n/g, '').replace(/\s+/g, ' ');

            // Robustly process looking for the optional properties of the class
            var properties = header.contents().each(function (i, elem) {
                var property = $(elem);

                if (property.text().match(/extends/))
                {
                    classObj.extends = property.text().trim();
                }
                else if (property.text().match(/implements/))
                {
                    classObj.implements = property.text().trim().replace(/\n/g, '').replace(/\s+/g, ' ');
                }
            });

            // Nested Classes
            classObj.nestedClasses = [];
            $('#nestedclasses').children('tr.api').each(function (i, elem) {
                var nestedClass = {};

                nestedClass.classModifiers = $(elem).find('.jd-typecol').text().trim();
                nestedClass.name = $(elem).find('.jd-linkcol').text().trim();

                classObj.nestedClasses.push(nestedClass);
            });

            // Constants
            classObj.constants = [];
            $('#constants').children('tr.api').each(function (i, elem) {
                var constant = {};

                constant.typeName = $(elem).find('.jd-typecol').text().trim();
                constant.name = $(elem).find('.jd-linkcol').text().trim();
                constant.desc = $(elem).find('.jd-descrcol').text().trim();

                classObj.constants.push(constant);
            });

            // Public Fields
            classObj.fields = [];
            $('#lfields').children('tr.api').each(function (i, elem) {
                var field = {};

                field.typeName = $(elem).find('.jd-typecol').text().trim().replace(/\n/g, '').replace(/\s+/g, ' ');
                field.name = $(elem).find('.jd-linkcol').text().trim();
                field.desc = $(elem).find('.jd-descrcol').text().trim();

                classObj.fields.push(field);
            });

            // Constructors
            classObj.constructors = [];
            $('#pubctors').find('.jd-linkcol').each(function (i, elem) {
                var constructor = {};

                constructor.name = $(elem).find('a').text().trim();
                constructor.desc = $(elem).find('.jd-descrdiv').text().trim();
                classObj.constructors.push(constructor);
            });

            classObj.methods = [];
            // Methods
            $('#pubmethods').children('tr.api').each(function (i, elem) {
                var method = {};

                method.returnTypeName = $(elem).find('.jd-typecol').text().trim().replace(/\n/g, '').replace(/\s+/g, ' ');;
                method.name = $(elem).find('.jd-linkcol .sympad a').text().trim();
                method.desc = $(elem).find('.jd-linkcol .jd-descrdiv').text().trim();
                method.params = $(elem).find('.jd-linkcol nobr').first().contents().last().text().trim();

                classObj.methods.push(method);
            });

            console.log(JSON.stringify(classObj, null, 4));
        }
    });
}, this);