network-stats-mashup/includes/class.cricket_spider.php

<?php
class CricketSpider {
  protected $html_file,
            $urls = array();
  public    $trunks = array();

  public static $cache_dir = "cache";

  public function __construct($html_file){
    $this->html_file = $html_file; 
    $this->create_trunks($this->urls);
  }

  private function create_trunks(){
    $urls = CricketSpider::get_urls_from_html_file($this->html_file);
    foreach($urls[3] as $index => $url){
      //echo 'url:' . $urls[3][$index] . "\n";
      //echo 'alt:' . $urls[2][$index] . "\n";
      $this->trunks[] = new Trunk($urls[3][$index], $urls[2][$index]);
    }
  }

  /* checks if a given file is less than 1 day old */
  public static function probe_cache($file){
    $today = time();
    $start_of_today = mktime(0, 0, 0, date("n", $today), date("j", $today), date("Y", $today));

    if(@filemtime($file) >= strtotime("-1 day", $start_of_today)){
      return true;
    }
    return false;
  }

  /* returns a unique code the url either a urlencoded string or an md5 hash if no meaningful identifier was found in the url */
  public static function code_from_url($url){
    $name_string = array();
    if(preg_match('/target=%2F(.*);/', $url, $name_string)){
      return $name_string[1];
    } else {
      return md5($url);
    }
  }

  /* returns ana array of urls found in a local file */
  public static function get_urls_from_html_file($file){
    $page_urls = array();
    $urls_cache = CricketSpider::$cache_dir . '/' . md5($file);
    /* get a list of Cricket urls from an html file */
    if(CricketSpider::probe_cache($urls_cache)){
      /* load cache */
      $page_urls = unserialize(file_get_contents($urls_cache));
    } else {
      /* makes a list of urls present in the image map html */
      preg_match_all('/(alt="(.+)")?\s?href="(.*)"/', file_get_contents('./'.$file), $page_urls);
      /* create cache */
      file_put_contents($urls_cache, serialize($page_urls));
    }
    return $page_urls;
  }
}
?>
1	<?php
2	class CricketSpider {
3	protected $html_file,
4	$urls = array();
5	public $trunks = array();
6
7	public static $cache_dir = "cache";
8
9	public function __construct($html_file){
10	$this->html_file = $html_file;
11	$this->create_trunks($this->urls);
12	}
13
14	private function create_trunks(){
15	$urls = CricketSpider::get_urls_from_html_file($this->html_file);
16	foreach($urls[3] as $index => $url){
17	//echo 'url:' . $urls[3][$index] . "\n";
18	//echo 'alt:' . $urls[2][$index] . "\n";
19	$this->trunks[] = new Trunk($urls[3][$index], $urls[2][$index]);
20	}
21	}
22
23	/* checks if a given file is less than 1 day old */
24	public static function probe_cache($file){
25	$today = time();
26	$start_of_today = mktime(0, 0, 0, date("n", $today), date("j", $today), date("Y", $today));
27
28	if(@filemtime($file) >= strtotime("-1 day", $start_of_today)){
29	return true;
30	}
31	return false;
32	}
33
34	/* returns a unique code the url either a urlencoded string or an md5 hash if no meaningful identifier was found in the url */
35	public static function code_from_url($url){
36	$name_string = array();
37	if(preg_match('/target=%2F(.*);/', $url, $name_string)){
38	return $name_string[1];
39	} else {
40	return md5($url);
41	}
42	}
43
44	/* returns ana array of urls found in a local file */
45	public static function get_urls_from_html_file($file){
46	$page_urls = array();
47	$urls_cache = CricketSpider::$cache_dir . '/' . md5($file);
48	/* get a list of Cricket urls from an html file */
49	if(CricketSpider::probe_cache($urls_cache)){
50	/* load cache */
51	$page_urls = unserialize(file_get_contents($urls_cache));
52	} else {
53	/* makes a list of urls present in the image map html */
54	preg_match_all('/(alt="(.+)")?\s?href="(.*)"/', file_get_contents('./'.$file), $page_urls);
55	/* create cache */
56	file_put_contents($urls_cache, serialize($page_urls));
57	}
58	return $page_urls;
59	}
60	}
61	?>
grid.support@nikhef.nl	ViewVC Help
Powered by ViewVC 1.1.28