<?php
/*notes:
Arrays needed for data pulls from Google:
keys, content_host, search tokens
Arrays for URL storage to get strings
IAM_URL, MYHOBBY_URL, MY_FAVORITE_URL, MYJOB_URL, IN_MY_LIFE_URL, MY_PET_URL, MY_POLITICAL_VIEW_URL, EMBARRASING_MOMENT_URL, HAPPIEST_MOMENT_URL, TURNON_URL, TURNOFF_URL, SEEKING_URL,
Arrays for token variants
*/
//set the state of the script;
$debug = 1;
//build necessary arrays;
$search_tokens = file('incs/token_searches.txt'); //REQUIRED*///*Search tokens are the sentence opening words like "I am a, My favorite" etc
$keys = file('incs/keys.txt'); //REQUIRED*//Keys are relevant by nature allows for extensibility but also gives higher quality information. i.e. sports or food or cooking etc.
$content_host = file('incs/content_hosts.txt');//REQUIRED*//Content host is the target blogs we are getting this content from.
//***********************END REQUIRED*****************************//
//Iterate data for url pulls
foreach($content_host as $host){
foreach($search_tokens as $sent_start){
foreach($keys as $keyword){
$host.$sent_start.$keyword = file_get_contents('http://www.google.com/search?hl=en&q=inurl%3A'.trim($host).'+'.trim($sent_start).'+-intitle%3A-'.trim($sent_start).'+%22'.trim($keyword).'%22&num=100&btnG=Google+Search');
preg_match_all('|<div class=g><h2 class=r><a href="(.+?)" class=l>|',$host.$sent_start.$keyword,$sent_start_matches);
}
}
}
?>