<?php
# search2feed
# @author: bronco@warriordudimanche.net / warriordudimanche.net
# @version 0.1
# @license free and opensource
# @use: search2feed.php?q=your+query+here
function Random_referer(){
return array_rand(array(
'http://oudanstoncul.com.free.fr/',
'http://googlearretedenousfliquer.fr/',
'http://stopspyingme.fr/',
'http://spyyourassfuckinggoogle.fr/',
'http://dontfuckinglookatme.fr/',
'http://matemonculgoogle.fr/',
'http://auxarmescitoyens.fr/',
'http://jetlametsavecdugravier.con/',
'http://lesdeuxpiedsdanstagueule.fr/',
'http://moncoudedanstabouche.con/',
'http://monpieddanston.uk/',
'http://bienfaitpourvosgueul.es/',
'http://pandanstesdents.fr/',
'http://tupuessouslesbras.fr/',
'http://mangetescrottesdenez.fr/',
'http://jtepourristesstats.fr/',
'http://ontecompissevigoureusement.com/',
'http://lepoingleveetlemajeuraussi.com/',
));
}
function return_UTF8($var){
$utf8=str_replace("\xef\xbb\xbf\xA0", '',utf8_encode($var));// SANS BOM !!!!
if (strpos($var,"\xA0")){return $utf8;}
if (strpos($utf8,'Ã')){return $var;}else {return $utf8;}
}
function file_curl_contents($url){
$ch = curl_init();
curl_setopt($ch, CURLOPT_HTTPHEADER, array('Accept-Charset: UTF-8'));
curl_setopt($ch, CURLOPT_HEADER, 0);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, FALSE);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
curl_setopt($ch, CURLOPT_URL, $url);
if (!ini_get("safe_mode") && !ini_get('open_basedir') ) {curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);}
curl_setopt($ch, CURLOPT_MAXREDIRS, 10);
curl_setopt($ch, CURLOPT_REFERER, random_referer());// notez le referer "custom"
$data = curl_exec($ch);
$response_headers = curl_getinfo($ch);
// Google seems to be sending ISO encoded page + htmlentities, why??
if($response_headers['content_type'] == 'text/html; charset=ISO-8859-1') $data = html_entity_decode(iconv('ISO-8859-1', 'UTF-8//TRANSLIT', $data));
# $data = curl_exec($ch);
curl_close($ch);
return return_utf8($data);
}
function parse_query($query,$title='Search2feed',$nb, $link='http://warriordudimanche.net',$lang='fr',$safesearch='&safe=off'){
$REGEX_WEB='#(?<=<h3 class="r"><a href="/url\?q=)([^&]+).*?>(.*?)</a>.*?(?<=<span class="st">)(.*?)(?=</span>)#';
$REGEX_PAGES='#&start=([0-9]+)|&start=([0-9]+)#';
$lang=strip_tags($lang);
$URL='https://www.google.com/search?hl='.$lang.$safesearch.'&id=hp&q=';
$page=file_curl_contents($URL.str_replace(' ','+',urlencode($query)).'&num='.$nb);
if (!$page){return false;}
preg_match_all($REGEX_WEB, $page, $r);
preg_match_all($REGEX_PAGES,$page,$p);
$p=count($p[2]);
$content=array();
foreach ($r[1] as $key=>$val){
$content[$key]['description']=strip_tags($r[3][$key]);
$content[$key]['link']=strip_tags($r[1][$key]);
$content[$key]['title']=strip_tags($r[2][$key]);
$content[$key]['guid']=strip_tags($r[1][$key]).'#1';
$content[$key]['pubDate']=@date('r');
}
$rss=array(
'infos'=>array(
'type'=>'rss',
'description'=>strip_tags($query),
'title'=>$title,
'link'=>$link,
),
'items'=>$content
);
return $rss;
}
function array2feed($array=null){
if (!$array){return false;}
if (empty($array['infos']['type'])){$array['infos']['type']='rss';}else{$array['infos']['type']=strtolower($array['infos']['type']);}
if (empty($array['infos']['description'])){$array['infos']['description']='';}
$r="\n";$t="\t";
$tpl=array('rss'=>array(),'atom'=>array());
$tpl['rss']['header']='<?xml version="1.0" encoding="utf-8" ?>'.$r.'<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/">'.$r.$t.'<channel>'.$r;
$tpl['atom']['header']='<feed xmlns="http://www.w3.org/2005/Atom">'.$r;
$tpl['rss']['footer']=$t.'</channel></rss>'.$r;
$tpl['atom']['footer']='</feed>'.$r;
$tpl['rss']['content-type']='Content-Type: application/rss+xml';
$tpl['atom']['content-type']='Content-Type: application/atom+xml;charset=utf-8';
header($tpl[$array['infos']['type']]['content-type']);
$feed=$tpl[$array['infos']['type']]['header'];
//create the feed's info content
foreach($array['infos'] as $key=>$value){
if ($array['infos']['type']=='atom'){ // ATOM
if ($key=='link'){$feed.=$t.$t.'<link href="'.$value.'" rel="self" type="application/atom+xml"/>'.$r;}
elseif ($key=='author'){$feed.=$t.$t.'<author><name>'.$value.'</name></author>'.$r;}
elseif ($key=='licence'){$feed.=$t.$t.'<'.$key.' href="'.$value.'" rel="license"/>'.$r;} // in atom feed, licence is the link to the licence type
elseif ($key!='version'&&$key!='type'){$feed.=$t.$t.'<'.$key.'>'.$value.'</'.$key.'>'.$r;}
}else{ // RSS
if ($key!='version'&&$key!='type'){$feed.=$t.$t.'<'.$key.'>'.$value.'</'.$key.'>'.$r;}
}
}
//then the items content
foreach ($array['items'] as $item){
if ($array['infos']['type']=='atom'){ $feed.=$t.$t.$t.'<entry>'.$r;}else{$feed.=$t.$t.$t.'<item>'.$r;}
foreach($item as $key=>$value){
if ($array['infos']['type']=='atom'){ // ATOM
if ($key=='link'){$feed.=$t.$t.$t.$t.'<link href="'.$value.'" rel="alternate" type="text/html"/>'.$r;}
elseif ($key=='content'){$feed.=$t.$t.$t.$t.'<content type="text">'.htmlspecialchars($value).'</content>'.$r;}
else{$feed.=$t.$t.$t.$t.'<'.$key.'>'.$value.'</'.$key.'>'.$r;}
}else{ // RSS
if ($key=='date'||$key=='pubDate'||$key=='title'||$key=='link'){$feed.=$t.$t.$t.$t.'<'.$key.'>'.htmlspecialchars($value).'</'.$key.'>'.$r;}
elseif($key=='guid'){ $feed.=$t.$t.$t.$t.'<guid isPermaLink="false">'.$value.'</guid>'.$r;}
else{$feed.=$t.$t.$t.$t.'<'.$key.'><![CDATA['.$value.']]></'.$key.'>'.$r;}
}
}
if ($array['infos']['type']=='atom'){ $feed.=$t.$t.$t.'</entry>'.$r;}else{$feed.=$t.$t.$t.'</item>'.$r;}
}
$feed.=$tpl[$array['infos']['type']]['footer'];
return $feed;
}
// récupération des parametres get
if (isset($_GET['q'])){$query=strip_tags($_GET['q']);}else{exit('pas de query');}
if (isset($_GET['title'])){$title=strip_tags($_GET['title']);}else{$title='GoogolSearch for '.$query;}
if (isset($_GET['nb'])){$nb=strip_tags($_GET['nb']);}else{$nb=20;}
echo array2feed(parse_query($query,$title,$nb));
?>