Scrapper for the website societe.com

societe.com.js

var nodeio = require('node.io');
var main_url = 'http://www.societe.com';
var options = {
    max: 20,
    timeout: 20,
    jsdom: true
};

exports.job = new nodeio.Job(options, {
    input: [{
        type: 'initial',
        url: main_url + '/cgi-bin/mainsrch?champ=' + encodeURIComponent(process.argv[3])
    }],
    run: function(input) {
        var self = this;
        var output = [];

        if(input.type === 'initial') {
            this.getHtml(input.url, function(err, $) {
                //Handle any request / parsing errors
                if(err) this.exit(err);
                try {

                    $('table.GlobalTable a.Blue12').each(function(index, el) {
                        var url_ = main_url + el.href;

                        self.add({
                            type: 'subrequest',
                            url: main_url + el.href,
                            name: $(el).text()
                        });
                        self.skip();
                    });
                } catch(err) {
                    console.log(err);
                }
            });
        } else {
            //console.log(self.assigned_input.name);
            this.getHtml(input.url, function(err, $) {

                if(err) this.exit(err);

                try {

                    var siret = "",
                        ca_soc = "",
                        siege_soc = "",
                        matriculation = "",
                        categorie = "",
                        juridique = "",
                        ca_ = "",
                        ceo_ = [];

                    var foundin = $('td:contains("Forme juridique"):last');
                    foundin.parent().parent().find('td').each(function(index, el) {
                        if($(el).text().indexOf("social") != -1) {
                            $(el).parent("tr").children().each(function(index, el) {
                                if(index == 1) siege_soc = $(el).text().trim();
                            });
                        }
                        if($(el).text().indexOf("juridique") != -1) {
                            $(el).parent("tr").children().each(function(index, el) {
                                if(index == 1) juridique = $(el).text().trim();
                            });
                        }
                        if($(el).text().indexOf("gorie") != -1) { //for categorie
                            $(el).parent("tr").children().each(function(index, el) {
                                if(index == 1) categorie = $(el).text().trim();
                            });
                        }
                    });

                    var foundin = $('td:contains("Immatriculation"):last');
                    foundin.parent().parent().find('td').each(function(index, el) {
                        if($(el).text().indexOf("SIRET") != -1) {
                            $(el).parent("tr").children().each(function(index, el) {
                                if(index == 1) {
                                    siret = $(el).text().trim();
                                }
                            });
                        }
                        if($(el).text().indexOf("Capital social") != -1) {
                            $(el).parent("tr").children().each(function(index, el) {
                                ca_soc = $(el).text().trim();
                            });
                        }
                        if($(el).text().indexOf("matriculation") != -1) { //for categorie
                            $(el).parent("tr").children().each(function(index, el) {
                                if(index == 1) matriculation = $(el).text().trim();
                            });
                        }
                    });

                    var foundin = $('td > div > b:contains("Dirigeants"):last');

                    foundin.closest("td").children('table:first').find('td').each(function(index, el) {
                        if(($(el).text().indexOf("ant") != -1) || //Gérant
                        ($(el).text().indexOf("ateur") != -1) || //Administrateur
                        ($(el).text().indexOf("DG") != -1) || //DG
                        ($(el).text().indexOf("dent") != -1)) { // Président
                            $(el).parent("tr").children().each(function(index, el) {
                                if(index == 1) {
                                    ceo_.push($(el).text().trim());
                                }
                            });
                        }
                    });

                    output = "[" + $("h1").text() + "] " + this.last.url + " | " + siret + " | " + ca_soc + " | " + categorie + " | " + ceo_.join(" ") + " | " + siege_soc + " | " + juridique;
                    //this.emit(output);
                    console.log(output)
                } catch(err) {
                    console.log(err);
                }
            });
        }
    }
});