# Xing-Scraper ## Description Scrapes the contact list of a contact (visible page) to CouchDB Works on the "Kontakte" Tab of a contact (if contacts are deiplayed by policy) This version scrapes only the 10 contacts visible, the next version will skip to the next page until last page has been reached ## Prerequisites: 1. This Javascript code has …

gistfile1.js

// ==UserScript==
// @name       Xing-Scraper
// @namespace  http://use.i.E.your.homepage/
// @version    0.95
// @description   scrapes the contact list of a contact (visible page) to CouchDB
// @match      https://www.xing.com/app/profile*
// @copyright  2012+, You
// ==/UserScript==

// Works on the "Kontakte" Tab of a contact (if contacts are deiplayed by policy)
// This version scrapes only the 10 contacts visible, ...
// ...the next version will skip to the next page until last page has been reached
// Prerequisites:
// 1. This Javascript code has to be injected into the loaded Xing-page
// ... either via Greasemonkey/Usersripts or similar
// ....or by using a bookmarklet (not tried yet, but better for occasional use)
// 2. Scraping produces a Javascript-object. To save this object this script uses CouchDB
// ... so you will need a CouchDB database and know the URL of it
// ... I use Iriscouch, free (for small use) online service https://www.iriscouch.com
var couchdburl="https://XXXXXXXX.iriscouch.com:6984/xing"; //YOUR URL goes here!!!
// ... you have to create a databse with the name "xing"
// 3. Prevent the Browser to block AJAX-requests to non-Xing-URLs
// ... (FYI "same origin policy") http://en.wikipedia.org/wiki/Same_origin_policy
// ... my way: start Chrome with parameter "--disable-web-security"
// ... eg. on OSX enter terminal and launch "open /Applications/Google\ Chrome.app/ --args --disable-web-security"
// ... (Discussion: tried "jsonp" instead but jQuery changed PUT and POST requests to GET - did not work at all!!) 


// This is just a place to print some feedback to the page. "alert()" does not work in xing.com
$("div#profile").append('<span id="ooo">&nbsp;</span>');
//$("div.wrap_content_top").append('<span id="ooo">&nbsp;</span>');
$("span#ooo").css("background-color","yellow");


// Little utility function: "containsEl" does array contain a value already
Array.prototype.containsel = function(v) {
    for(var i = 0; i < this.length; i++) {
        if(this[i] === v) return true;
    }
    return false;
};

// source contact information
var sourceid = $("ul.subnav > li").eq(1).find("a").attr("href").replace(/^.*\/profile\/(.*)$/,"$1");
  //... more contact data like company, name, function, fon, email etc.
  //... and PUT this information as well

// if present scrape one page of linked contacts (10)
$("table#profile-contacts").find("tr").each(function(i){
  var el = $("td", this).find("a.user-photo");
	var hovercardurl=el.attr("data-hover-card");
	var photourl=el.find("img").attr("src");
	el = $("td", this).eq(1).find("a");
	var name = el.eq(0).text();
	var link = el.eq(0).attr("href");
	id=link.replace(/^.*\/profile\/(.*?)\/.*$/,"$1"); 
	var firma = el.eq(-1).text();

	var contact={};
	jQuery.ajax({type:"GET", async:false, url:couchdburl+"/"+id,dataType:"json",success:function(data,status){contact = data;} });
	if (contact["_id"] !== undefined) {
		// add source
		if (contact["source"] !== undefined) {
			if (!contact["source"].containsel(sourceid)) contact["source"] = contact["source"].concat(sourceid); //??? geht das add element to array?
			//else do nothing ... maybe prevent empty update to be sent to the databse
		} else 
			contact["source"] = [sourceid];
	} else {
		// create contact
        contact = {"_id":id, "name":name, "company":firma, "source":[sourceid], "url":"https://xing.com"+link, "photourl":"https://xing.com"+photourl, "hovercardurl":hovercardurl};
	}
	//$("span#ooo").append("<br>Scrape "+id);
	jQuery.ajax({
        type:"POST", 		// ... seems to work with POST instead of PUT too (make sure _id and for update _rev is in the data object!
        url:couchdburl,    // ... can use it with _id in the data instead path+id, 
        contentType:"application/json; charset=UTF-8", 
        data: JSON.stringify(contact),
        error:function(xhr,errmsg,err){ $("span#ooo").append(" Error:"+errmsg+JSON.stringify(err)); },
        //complete:function(xhr,status){ $("span#ooo").append(" Status:"+status); },
        success:function(data,status){ $("span#ooo").append(" OK:"+JSON.stringify(data)); }
	 }); 
	
});