git.m6w6.name Git - m6w6/ext-http/blob - lib/FeedAggregator.php

   1 <?php
   2
   3 /**
   4  * Simple Feed Aggregator
   5  * $Id$
   6  *
   7  * @copyright   Michael Wallner, <mike@iworks.at>
   8  * @license     BSD, revised
   9  * @package     pecl/http
  10  * @version     $Revision$
  11  */
  12 class FeedAggregator
  13 {
  14     /**
  15      * Cache directory
  16      *
  17      * @var string
  18      */
  19     public $directory;
  20
  21     /**
  22      * Feeds
  23      *
  24      * @var array
  25      */
  26     protected $feeds = array();
  27
  28     /**
  29      * Constructor
  30      *
  31      * @param string $directory
  32      */
  33     public function __construct($directory = 'feeds')
  34     {
  35         $this->setDirectory($directory);
  36     }
  37
  38     /**
  39      * Set cache directory
  40      *
  41      * @param string $directory
  42      */
  43     public function setDirectory($directory)
  44     {
  45         $this->directory = $directory;
  46         foreach (glob($this->directory .'/*.xml') as $feed) {
  47             $this->feeds[basename($feed, '.xml')] = filemtime($feed);
  48         }
  49     }
  50
  51     /**
  52      * Strips all special chars
  53      *
  54      * @param string $url
  55      * @return string
  56      */
  57     public function url2name($url)
  58     {
  59         return preg_replace('/[^\w\.-]+/', '_', $url);
  60     }
  61
  62     /**
  63      * Checks if $url is a known feed
  64      *
  65      * @param string $url
  66      * @return bool
  67      */
  68     public function hasFeed($url)
  69     {
  70         return isset($this->feeds[$this->url2name($url)]);
  71     }
  72
  73     /**
  74      * Add an URL as feed
  75      *
  76      * @param string $url
  77      * @return void
  78      * @throws Exception
  79      */
  80     public function addFeed($url)
  81     {
  82         $r = $this->setupRequest($url);
  83         $r->send();
  84         $this->handleResponse($r);
  85     }
  86
  87     /**
  88      * Add several URLs as feeds
  89      *
  90      * @param array $urls
  91      * @return void
  92      * @throws Exception
  93      */
  94     public function addFeeds(array $urls)
  95     {
  96         $pool = new HttpRequestPool;
  97         foreach ($urls as $url) {
  98             $pool->attach($r = $this->setupRequest($url));
  99         }
 100         $pool->send();
 101
 102         foreach ($pool as $request) {
 103             $this->handleResponse($request);
 104         }
 105     }
 106
 107     /**
 108      * Load a feed (from cache)
 109      *
 110      * @param string $url
 111      * @return string
 112      * @throws Exception
 113      */
 114     public function getFeed($url)
 115     {
 116         $this->addFeed($url);
 117         return $this->loadFeed($this->url2name($url));
 118     }
 119
 120     /**
 121      * Load several feeds (from cache)
 122      *
 123      * @param array $urls
 124      * @return array
 125      * @throws Exception
 126      */
 127     public function getFeeds(array $urls)
 128     {
 129         $feeds = array();
 130         $this->addFeeds($urls);
 131         foreach ($urls as $url) {
 132             $feeds[] = $this->loadFeed($this->url2name($url));
 133         }
 134         return $feeds;
 135     }
 136
 137     protected function saveFeed($file, $contents)
 138     {
 139         if (file_put_contents($this->directory .'/'. $file .'.xml', $contents)) {
 140             $this->feeds[$file] = time();
 141         } else {
 142             throw new Exception("Could not save feed contents to $file.xml");
 143         }
 144     }
 145
 146     protected function loadFeed($file)
 147     {
 148         if (isset($this->feeds[$file])) {
 149             if ($data = file_get_contents($this->directory .'/'. $file .'.xml')) {
 150                 return $data;
 151             } else {
 152                 throw new Exception("Could not load feed contents from $file.xml");
 153             }
 154         } else {
 155             throw new Exception("Unknown feed/file $file.xml");
 156         }
 157     }
 158
 159     protected function setupRequest($url, $escape = true)
 160     {
 161         $r = new HttpRequest($url);
 162         $r->setOptions(array('redirect' => true));
 163
 164         $file = $escape ? $this->url2name($url) : $url;
 165
 166         if (isset($this->feeds[$file])) {
 167             $r->setOptions(array('lastmodified' => $this->feeds[$file]));
 168         }
 169
 170         return $r;
 171     }
 172
 173     protected function handleResponse(HttpRequest $r)
 174     {
 175         if ($r->getResponseCode() != 304) {
 176             if ($r->getResponseCode() != 200) {
 177                 throw new Exception("Unexpected response code ". $r->getResponseCode());
 178             }
 179             if (!strlen($body = $r->getResponseBody())) {
 180                 throw new Exception("Received empty feed from ". $r->getUrl());
 181             }
 182             $this->saveFeed($this->url2name($r->getUrl()), $body);
 183         }
 184     }
 185 }
 186
 187 ?>