<?php if (! defined('BASEPATH')) exit('No direct script access allowed');
include_once("simple_html_dom.php");
class Crawl
{
    
    var $html_content = '';
    var $arr_att_clean = array();
    
    function Crawl(){
        // nothing to do
    }
    
	function robot(){
		return $html = new simple_html_dom();
    }
	
    function getHTML($link)
	{
        if($this->html_content==''){
            $html = file_get_html($link);
            $this->html_content = $html;    
        }else{
            $html = $this->html_content;
        }
        //trit($html);   
		return $html ; //type object
    }
    
    
	function getNodes($html, $rule, $num=-1)
	{
		$arr = array(); 
		if($num==-1)
			foreach($html->find($rule) as $e)
				array_push($arr, $e->innertext);
		else
			foreach($html->find($rule,$num) as $e)
				array_push($arr, $e->innertext);
        return $arr;
    }
	
	function getLink($html, $rule, $num=-1)
	{
		$arr = array();
		if($num==-1)
			foreach($html->find($rule) as $e)
				array_push($arr, $e->href);
		else
			foreach($html->find($rule,$num) as $e)
				array_push($arr, $e->href);
        return $arr;
    }
	
    function getContent($link, $att_content)
	{
        if($this->html_content==''){
            $html = file_get_html($link);
            $this->html_content = $html;    
        }else{
            $html = $this->html_content;
        }
                
        foreach($html->find($att_content) as $e){
            $content_html = $e->innertext;
        }
        $html = str_get_html($content_html);
       
        foreach($this->arr_att_clean as $att_clean){
            foreach($html->find($att_clean) as $e){
                $e->outertext = '';
            }
        }
        
        $ret = $html->save();
        return $ret;
    }
    
    function removeLink($content)
	{
        $html = str_get_html($content);
        // link content
        foreach($html->find('a') as $e){
            $e->outertext = $e->innertext;
        }
        $ret = $html->save();
        return $ret;
    }
    
    function removeLastElement($content, $element)
	{
        $html = str_get_html($content);
        // link content
        $html->find($element, -1)->outertext = '';
        $ret = $html->save();
        return $ret;
    }
    
    function removeFirstElement($content, $element)
	{
        $html = str_get_html($content);        
        $html->find($element, 0)->outertext = '';
        $ret = $html->save();
        return $ret;
    }

}