getTexts.js

var webPage = require('webpage');
var HOME = 'http://my.domain.com/';
var stepIndex = 0;

var HOME_CONTENT = 'Where ara the info in home';
var POST_MENU = 'Go to menu';
var TOP_MENU = 'Top Menu';
var CONTENT_POST = 'Where are the info in pages';

var PAGES = {};
var POST = [];

function posibleError (msg, line, source) {
	console.log('>', msg);
}

function posibleAlert (msg) {
	console.log('<', msg);
}

/**
 * [getPostOrPages description]
 * @param  {String} name Selector
 * @return {Object}      Post/Pages
 */
function getPostOrPages (name) {
	function clean (node) {
		var newNode = [];
		for (var i = 0; i < node.length; i++) {
			if(node[i] && !node[i].body && node[i].tagName ){
				if(node[i].tagName == 'DIV'){
					var z = clean($(node[i]
							.innerHTML
							.replace(/\t|\n|\s{2,}/gim, '')
							.replace(/(\<br\>)/gim, '</p><p>')));

					for (var w = 0; w < z.length; w++) {
						newNode.push(z[w]);
					}
				} else if(/P|H[1-6]|UL|I|TABLE|LI|IMG/i.test(node[i].tagName)){
					newNode.push(node[i]);
				}
			}
		}

		return newNode;
	}

	var post = { text : [] };
	var child = clean($(name).children());

	for (var i = 0; i < child.length; i++) {
		if(child[i] && child[i].tagName && !child[i].body && child[i].innerHTML.length){
			if( child[i].tagName != 'BR' && child[i].tagName != 'H1' ){
				post.text.push(child[i].outerHTML);	
			} else if (  child[i].tagName == 'H1' ) {
				post.title = child[i].innerText;
			} else if (  child[i].tagName == 'I' && !post.autor ) {
				post.autor = child[i].innerText;
			}
		}
	}

	return post;
}

/**
 * [getHome description]
 * @param  {String} name Selector
 * @return {Array}      Element
 */
function getHome(name) {
	var data = [];
	var post = {};
	var cild = $(name).children();
	for (var i = 0; i < cild.length; i++) {
		if(cild[i].tagName === 'H2'){
			if(post.title){
				data.push(post);
			}
			post = { title : cild[i].innerText, text : [] };
		} else {
			if(post.text && cild[i].tagName != 'BR'	){
				if(cild[i].tagName != 'I'){
					post.text.push(cild[i]
							.outerHTML
							.replace(/\t|\n|(\s){2,}|(\<br\>){2,}/gim, ''));
				} else {
					post.autor = cild[i].innerText;
				}
			}
		}
	}
	if(post.title){
		data.push(post);
	}

	return data;
}

/**
 * [getPostMenu description]
 * @param  {String} name Selector
 * @return {Array}       Array URL
 */
function getPostMenu (name) {
	var urs = [];
	var items = $(name);
	for (var i = 0; i < items.length; i++) {
		urs.push(items[i].href);
	}
	return urs;
}

/**
 * [getMenu description]
 * @param  {Number} i   Number in array
 * @param  {String} top Selector
 * @return {String}     Go to
 */
function getMenu (i, top) {
	return $(top)[i].href;
}

/**
 * [startPage description]
 * @param  {String}  url    Go to
 * @param  {Boolean} isPost [description]
 */
function startPage(url, isPost) {
	var page = webPage.create();

	page.onConsoleMessage = posibleError;
	page.onAlert = posibleAlert;
	
	stepIndex++;

	if(url){
		page.open(url, function(status){
			if (status === 'success') {
				console.log('Start :',  url);

				page.injectJs('jquery.min.js');

				var newUrl;
				
				if(!phantom.state){
					if(isPost || url.indexOf('index.html') < 0){
						PAGES[ url ] = page.evaluate(getPostOrPages, CONTENT_POST);
					}

					if(isPost){
						PAGES[ url ].post = true;
						newUrl = POST[stepIndex];
					} else {
						if(url.indexOf('index.html') >= 0){
							POST = page.evaluate(getPostMenu, POST_MENU);
							PAGES[ url ] = page.evaluate(getHome, HOME_CONTENT);
							PAGES[ url ].list = true;
						} else {
							PAGES[ url ].page = true;
						}

						newUrl = page.evaluate(getMenu, stepIndex, TOP_MENU));
					}

					if(!newUrl){
						stepIndex = 0;
						newUrl = POST[stepIndex];
						isPost = true;
					}

					startPage(newUrl, isPost);
				} else {
					phantom.state();
				}
			}
		});
	} else {
		console.log(JSON.stringify(PAGES, null, '\t'));
	}
}

startPage(HOME + 'index.html');