<?php
$dir = "pages/";
//Trends Parsing
$trends_url = 'http://www.google.com/trends/hottrends/atom/hourly';
$trends = file_get_contents($trends_url);
$trends = html_entity_decode(htmlentities($trends, ENT_COMPAT
, 'UTF-8'));
preg_match_all("/<a href=\"[^>]*>(.*?)<\/a>/is", $trends, $keywords, PREG_PATTERN_ORDER
);
shuffle($keywords[1]);
$kw = array();
$c = rand (5, 6); // Выборка количества трендов для генерации
$k = array_keys($keywords[1]);
$v = array_values($keywords[1]);
while($c>0) {
$q=array_rand($k);
$kw[$k[$q]]=$v[$q];
unset($k[$q]);
$c--;
$cur_kw = $kw[$q]; //текущий тренд
echo "<h1>".$cur_kw."</h1>";
$count = rand (4, 9);// Количество страниц для парсинга
//Google Serps Parsing
$google_request = 'http://www.google.com/ie?hl=en&num='.$count.'&start=0&lr=&q='.urlencode(trim($kw[$q]));
//$google_request = 'http://www.google.com/search?hl=en&q=keyword&btnG=Google+Search&meta=lr%3Dlang_en';
$google_pattern = '/<a title="(.*)" href=(.*)>(.*)<\/a>/isU';
$google_result = file_get_contents($google_request);
$google_result = str_replace(">", "-", $google_result);
$google_result = str_replace("<b>", "", $google_result);
$google_result = str_replace("</b>", "", $google_result);
$google_result = str_replace("<b>", "", $google_result);
$google_result = str_replace("</b>", "", $google_result);
$google_result = str_replace("’", "'", $google_result);
$google_result = str_replace(" ...", ".", $google_result);
$google_result = str_replace("...", ".", $google_result);
$google_result = str_replace(" | ", ", ", $google_result);
$google_result = str_replace("'", "'", $google_result);
preg_match_all ($google_pattern, $google_result, $google_data);
//MSN Serps Parsing
$msn_url = 'http://search.msn.com/results.aspx?format=rss&count='.$count.'&q='.urlencode(trim($kw[$q]));
$msn_request = file_get_contents($msn_url);
$msn_request = html_entity_decode(htmlentities($msn_request, ENT_COMPAT
, 'UTF-8'));
$msn_request = str_replace(" ...", ".", $msn_request);
$msn_request = str_replace("Live Search: ", "", $msn_request);
preg_match_all("/<title[^>]*>(.*?)<\/title>/is", $msn_request, $msn_title, PREG_PATTERN_ORDER
);
preg_match_all("/<description[^>]*>(.*?)<\/description>/is", $msn_request, $msn_desc, PREG_PATTERN_ORDER
);
//Blogsearch Serps parsing
$blogs_url = 'http://blogsearch.google.com/blogsearch_feeds?hl=en&ie=utf-8&num='.$count.'&output=atom&q='.urlencode($kw[$q]);
$blogs_request = file_get_contents($blogs_url);
$blogs_request = html_entity_decode(htmlentities($blogs_request, ENT_COMPAT
, 'UTF-8'));
$blogs_request=str_replace("<b>", "", $blogs_request);
$blogs_request=str_replace("</b>", "", $blogs_request);
$blogs_request=str_replace("... ", "", $blogs_request);
$blogs_request=str_replace(" ...", "", $blogs_request);
$blogs_request=str_replace("'", "", $blogs_request);
$blogs_request=str_replace("--", "", $blogs_request);
$blogs_request=str_replace("Google Blog Search: ", "", $blogs_request);
preg_match_all("/<title[^>]*>(.*?)<\/title>/is", $blogs_request, $blogs_title, PREG_PATTERN_ORDER
);
preg_match_all("/<content[^>]*>(.*?)<\/content>/is", $blogs_request, $blogs_desc, PREG_PATTERN_ORDER
);
for ($i = 0; $i<count
($google_data[1]); $i++){
$googlelinktitle = $google_data[0][$i]; //Активная ссылка с заголовком в анкоре из серпа google
$googlesnippet = $google_data[1][$i]; //Текст google сниппета
$googlelinkurl = $google_data[2][$i]; //Активная ссылка с url сайта в анкоре из серпа google
$googletitle = $google_data[3][$i]; //Заголовок сайта из сниппета google
$msnlinktitle = $msn_title[1][$i]; //Заголовок сайта из сниппета msn
$msnsnippet = $msn_desc[1][$i]; //Текст msn сниппета
$blogslinktitle = $blogs_title[1][$i]; //Заголовок сайта из сниппета blogsearch
$blogssnippet = $blogs_desc[1][$i]; //Текст blogsearch сниппета
$cur_kw = $kw[$q]; //текущий тренд
//Content
$pagetext = $googlesnippet."<br>".$blogslinktitle." ".$msn_snippet."<br>".$blogssnippet."<br><hr>";
$pagelink = "<a target='_blank' href='".$googlelinkurl."'>".$googletitle."</a> <br>";
echo $pagetext.$pagelink;
}
}
?>