node-bot

Node.js implementation of Extract Widget bot using https://github.com/Ayms/node-dom/, https://github.com/Ayms/node-gadgets/ and https://github.com/Ayms/node-googleSearch/

Purpose:

Real-time extraction of web pages information (html, text, etc) based on given criterias.

It can be used as a server or an API, then parameters are passed in the URL, or directly as an independant node.js module.

The difference with node-gadgets is that for performances reasons it does not return the full gadgets, only the relevant information (shopbot example : seeking for "nike lebron 9" will return real-time the price of the shoes on nike store's web site)

Install :

npm install node-bot

or

git clone http://github.com/Ayms/node-bot.git
cd node-bot
npm link .

Complementary modules : node-ewa

 Note : node-ewa is not a public module for now, so you can only use node-bot's server/API mode. 

Use :

getelements.js :

As a module :

	var getElements = require('node-bot').getElements;
	
	var $E=encodeURIComponent;
	
	var response={
		end:function(gadgets) {
			console.log(gadgets);
			//output format, see below
		}
	};
	
	with node-googleSearch
	
	var params='search='+$E('nike shoes')'+'&name='+$E(nike_shoes)+'&regexp='+$E('\\$|€')+'&nbmax=20';
	
	without node-googleSearch
	
	var params='url=http://store.nike.com/us/en_us/%3Fl%3Dshop,pdp,ctr-inline/cid-1/pid-417121/pgid-437002&search='+$E('nike shoes')'+'&name='+$E(nike_shoes)+'&regexp='+$E('\\$|€')+'&nbmax=20';

	getElements(params,response);

As a server/API :

	var http = require('http'),  
	URL = require('url'),
	getElements = require('node-bot').getElements;

	var handleRequest = function (request, response) {
	  
		var qs = URL.parse(request.url);
		  
		if (qs.pathname == '/getelements'){
			getElements(qs.query,response);
		};
	};

	http.createServer(handleRequest).listen(myport);

To call it directly :

with node-googleSearch

http://myserver:myport/getelements?name=nike_shoes&search='nikestore nike lebron9'&regexp=$|€&nbmax=20

without node-googleSearch

http://myserver:myport/getelements?url=http://store.nike.com/us/en_us/%3Fl%3Dshop,pdp,ctr-inline/cid-1/pid-417121/pgid-437002&name=nike_shoes&search='nikestore nike lebron9'&regexp=$|€&nbmax=20

Example with encoded parameters to retrieve the price of "lebron9" shoes on nike store :

with node-googleSearch

http://213.246.53.127:1341/getelements?name=nike_shoes&search=nike%20lebron%209&regexp=%5C%24%7C%C3%A2%E2%80%9A%C2%AC

without node-googleSearch

http://213.246.53.127:1341/getelements?url=http%3A%2F%2Fstore.nike.com%2Fus%2Fen_us%2F%253Fl%253Dshop%2Cpdp%2Cctr-inline%2Fcid-1%2Fpid-417121%2Fpgid-437002&name=nike_shoes&search=nike%20lebron%209&regexp=%5C%24%7C%C3%A2%E2%80%9A%C2%AC&nbmax=20

To call it from a script :

	var xscript=document.createElement('SCRIPT');
	xscript.type="text/javascript";
	var params='name=nike_shoes'+'&search='+$E(nike shoes nikestore)+'&regexp='+$E('\\$|€')+'&nbmax=20'; //add url parameter if you already know it (do not use node-googleSearch)
	xscript.src='http://myserver:myport/getelements?'+params;
	document.head.appendChild(xscript);

	xscript.onload or onreadystatechange --> do what you have to do with the output

Output format (see more details below) : nike_shoes=(Array containing the gadgets) (where 'nike_shoes' does correspond to the parameter 'name')

	So to use it you can do :
	
	xscript.onload=function() {
		var res=eval('nike_shoes'+this.shoe_number); //for example if the parameter name depends on some var in your code
		if (res.length>0) {
			...
		}
		
		or simply
		
		if (nike_shoes.length>0) {
			...
		}
	}

See example here : http://www.extractwidget.com/nodejs/test/babyliss.html (see API code at the end of html file)

Note : if your regexp does contain "\" and if you pass it through a js var (Example above : $E('\\$|€')) make sure to double it.

Note2 : make sure that the encoding of your files/browsers is utf-8

Parameters :

url : the url of the site where you want to extract gadgets from, if absent the url is retrieved with node-googleSearch using the value of search string (example : "nikestore nike shoes" will return the first url returned by Google Search that matches this string).

name : the name that will become the name of the global var containing the output in its 'gadgets' property (example : nike_shoes.gadgets).

regexp : while building the DOM, node-dom will use that regular expression to detect the objects that you are looking for (example : regexp=$|€ --> you are looking for gadgets in the page that are related to a price in $ or €).

search : indicates that once the gadgets have been selected with the regexp, you can filter these gadgets based on the value of search field (example : "nikestore nike shoes" url can contain other products than shoes, node-bot will return only the results matching "nike shoes").

nbmax : important parameter for performances, the value does specify a limit for the weight of searched gadgets so node-bot does not spend a lot of time processing gadgets that are not relevant. The default value is 100, recommended value is 20.

Output :

The output is an Array of :

[gadget html,width,height,gadget name,reserved,base,price,html of regexp object]

No json format here for now for historical reasons and backward compatibility with existing projects (TODO later).

The first three parameters in the output are not filled by node-bot.

See https://github.com/Ayms/node-gadgets/ documentation for more details.

Tunnelling with node-Tor :

TODO

Tests and API :

jCore server (http://www.jcore.fr) : http://213.246.53.127:1341/getelements?params

Example with node-googleSearch :

http://213.246.53.127:1341/getelements?name=nike_shoes&search=nike%20lebron%209&regexp=%5C%24%7C%C3%A2%E2%80%9A%C2%AC&nbmax=20

Example without node-googleSearch :

http://213.246.53.127:1341/getelements?url=http%3A%2F%2Fstore.nike.com%2Fus%2Fen_us%2F%253Fl%253Dshop%2Cpdp%2Cctr-inline%2Fcid-1%2Fpid-417121%2Fpgid-437002&name=nike_shoes&search=nike%20lebron%209&regexp=%5C%24%7C%C3%A2%E2%80%9A%C2%AC&nbmax=20

You can use the API on jCore server : http://213.246.53.127:1341 (if by any unforeseen reasons the server is down, please advise).

Links above might not return a correct result due to changes on nike store web site, then you can try :

Webble project : http://www.webble.it/mindex5.php (quick test : click on OK, then on first link that appears)

Example of API code and use, retrieve real time the price of babyliss homelight product on different merchant sites :

http://www.extractwidget.com/nodejs/test/babyliss.html (click on "acheter maintenant" then wait for prices to be displayed in green)

See tests.txt in ./test