<?php
/**
 * This set of functions queries either the Yahoo! term extraction API or the Tagyu API and stuffs the
 * results into a DB. For further refining, see step 5 in the blog post linked to below (part 2). 
 *
 * Requires Keith Devens' PHP XML  library, located at http://keithdevens.com/software/phpxml,
 * and a Yahoo! Application key (see here: http://developer.yahoo.net/search/index.html)
 * Read the related blog post here: 
 * http://www.fiftyfoureleven.com/weblog/web-development/programming-and-scripts/apis/retrotag-with-tagyu-yahoo-terms-apis-1
 * http://www.fiftyfoureleven.com/weblog/web-development/programming-and-scripts/apis/retrotag-with-tagyu-yahoo-terms-apis-2
 * Man that's a long couple of urls.
 *
 * 
 *
 * @author: Mike Papageorge, http://www.fiftyfoureleven.com/contact
 * @lastUpDate: 01/12/2005 
 * @license: http://creativecommons.org/licenses/by/2.5/
 * @location: http://www.fiftyfoureleven.com/site/code/retrotag.txt
 */
	error_reporting(E_ALL);

// Variables etc...

	$mode 		= 'Yahoo!'; // Switch between either Yahoo! or Tagyu
	$yahooAppId	= 'yourYahooAppId';
	$TagUser 	= 'YourTagyuUserName';
	$TagPass 	= 'YourTagyuPassword';
	$db			= 'yourDb';
	$user 		= 'userName';
	$pass 		= 'passWord';
	$host 		= 'localhost';
	$table 		= 'post2tag';
	$query 		= "select id, content from weblog"; // The query for your content
	// Keep in mind that with Tagyu you only get 1000 queries per day, so if you have >1000 posts
	// to tag, you need to use 'limit' in your query. See comment #1 on the 2nd weblog link above for more info.
	
	
	$contentRow	= 'content'; 	// The row that your content is held in. For example in 
						// "select id, content from weblog", 'content' would be $contentRow.
	include('xml.php'); // Include Keith Devens XML library: http://keithdevens.com/software/phpxml
	
	// Table for the data:
		/*
			CREATE TABLE `post2tag` (
			`id` int(11) NOT NULL auto_increment,
			`tagvalue` varchar(255) NOT NULL default '',
			`post_id` smallint(6) NOT NULL default '0',
			`tag_id` smallint(6) NOT NULL default '0',
			PRIMARY KEY  (`id`)
			);
		*/

	//
	//
	//	Here's all the action:
	//
	//
		// Connect to our DB
		if(!$link = mysql_connect($host, $user, $pass)) {die('no connection');}
		if(!mysql_select_db($db)) {die('no db');}
		
		// Get our result:
		if(!$result = mysql_query($query)) {die('query error');}
		// Start the while loop:
		while($row = mysql_fetch_array($result)) {
			
			// Here's a switch for choosing which API we're using:
			if($mode == 'Yahoo!') {
				$url = "http://api.search.yahoo.com/ContentAnalysisService/V1/termExtraction"; // Yahoo!
				$querystring = "appid=$yahooAppId&context=$row[$contentRow]"; // Yahoo!
				echo $querystring;
			} elseif($mode == 'Tagyu') {
				$url = 'http://tagyu.com/api/suggest/'; // Tagyu
				$querystring = "suggest=$row[$contentRow]"; // Tagyu
			} else die('Mode is set incorrectly');
	
			// Initialize a curl handle:
			$ch = curl_init();   
			// Now we start to set some of the options:
				// Set request method to POST 
				curl_setopt($ch, CURLOPT_POST, 1); 
				// Pass the url that we're posting to:
				curl_setopt($ch, CURLOPT_URL, $url); 
				// Tell cURL to return the data into a variable:
				curl_setopt($ch, CURLOPT_RETURNTRANSFER,1); 
				// Set a timeout value for connecting to the server, in seconds:
				curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 15); 
				// Set a timeout value for the entire process, in seconds: ****!!!!
				curl_setopt($ch, CURLOPT_TIMEOUT, 15); 
				// To Pass a username and password:
				($mode == 'Tagyu') ? curl_setopt($ch, CURLOPT_USERPWD, "$username:$password"):'';
				// Pass the query string to the handle:
				curl_setopt($ch, CURLOPT_POSTFIELDS, $querystring); 
			// Execute the request!
			$chresult = curl_exec($ch); // run the whole process
			// Close our handle:
			curl_close($ch);
			
			// Have a look at the XML: uncomment the following two lines - this 'exits' here, so it won't affect your DB.
			// header("Content-type: application/xml");
			// echo $chresult;exit;
				
			
			// Sleep 1 second to be polite to the API server.
			sleep(1);
			
			// Pass the result to our XML library and unserialize the data:
			$data = XML_unserialize($chresult);
			
			// Have a look at $data: uncomment the following line - this 'exits' here, so it won't affect your DB.
			// echo '<pre>';print_r($data);echo '</pre>';exit;

			// Here we put the result data into $r so we only need one foreach below:
			if($mode == 'Yahoo!') {
				$r =  $data['ResultSet']['Result']; // Yahoo! data
			} else {
				$r =  $data['suggestions']['tag']; // Tagyu data
			}
	
			foreach($r as $value) {
				$tagquery 	= "insert into $table set tagvalue='$value', post_id = $row[id]";
				$tagresult 	= mysql_query($tagquery);
			}
		}
?>