application/Url.php

   1 <?php
   2 /**
   3  * Converts an array-represented URL to a string
   4  *
   5  * Source: http://php.net/manual/en/function.parse-url.php#106731
   6  *
   7  * @see http://php.net/manual/en/function.parse-url.php
   8  *
   9  * @param array $parsedUrl an array-represented URL
  10  *
  11  * @return string the string representation of the URL
  12  */
  13 function unparse_url($parsedUrl)
  14 {
  15     $scheme   = isset($parsedUrl['scheme']) ? $parsedUrl['scheme'].'://' : '';
  16     $host     = isset($parsedUrl['host']) ? $parsedUrl['host'] : '';
  17     $port     = isset($parsedUrl['port']) ? ':'.$parsedUrl['port'] : '';
  18     $user     = isset($parsedUrl['user']) ? $parsedUrl['user'] : '';
  19     $pass     = isset($parsedUrl['pass']) ? ':'.$parsedUrl['pass']  : '';
  20     $pass     = ($user || $pass) ? "$pass@" : '';
  21     $path     = isset($parsedUrl['path']) ? $parsedUrl['path'] : '';
  22     $query    = isset($parsedUrl['query']) ? '?'.$parsedUrl['query'] : '';
  23     $fragment = isset($parsedUrl['fragment']) ? '#'.$parsedUrl['fragment'] : '';
  24
  25     return "$scheme$user$pass$host$port$path$query$fragment";
  26 }
  27
  28 /**
  29  * Removes undesired query parameters and fragments
  30  *
  31  * @param string url Url to be cleaned
  32  *
  33  * @return string the string representation of this URL after cleanup
  34  */
  35 function cleanup_url($url)
  36 {
  37   $obj_url = new Url($url);
  38   return $obj_url->cleanup();
  39 }
  40
  41 /**
  42  * Get URL scheme.
  43  *
  44  * @param string url Url for which the scheme is requested
  45  *
  46  * @return mixed the URL scheme or false if none is provided.
  47  */
  48 function get_url_scheme($url)
  49 {
  50   $obj_url = new Url($url);
  51   return $obj_url->getScheme();
  52 }
  53
  54 /**
  55  * URL representation and cleanup utilities
  56  *
  57  * Form
  58  *   scheme://[username:password@]host[:port][/path][?query][#fragment]
  59  *
  60  * Examples
  61  *   http://username:password@hostname:9090/path?arg1=value1&arg2=value2#anchor
  62  *   https://host.name.tld
  63  *   https://h2.g2/faq/?vendor=hitchhiker&item=guide&dest=galaxy#answer
  64  *
  65  * @see http://www.faqs.org/rfcs/rfc3986.html
  66  */
  67 class Url
  68 {
  69     private static $annoyingQueryParams = array(
  70         // Facebook
  71         'action_object_map=',
  72         'action_ref_map=',
  73         'action_type_map=',
  74         'fb_',
  75         'fb=',
  76
  77         // Scoop.it
  78         '__scoop',
  79
  80         // Google Analytics & FeedProxy
  81         'utm_',
  82
  83         // ATInternet
  84         'xtor='
  85     );
  86
  87     private static $annoyingFragments = array(
  88         // ATInternet
  89         'xtor=RSS-',
  90
  91         // Misc.
  92         'tk.rss_all'
  93     );
  94
  95     /*
  96      * URL parts represented as an array
  97      *
  98      * @see http://php.net/parse_url
  99      */
 100     protected $parts;
 101
 102     /**
 103      * Parses a string containing a URL
 104      *
 105      * @param string $url a string containing a URL
 106      */
 107     public function __construct($url)
 108     {
 109         $this->parts = parse_url($url);
 110
 111         if (!empty($url) && empty($this->parts['scheme'])) {
 112             $this->parts['scheme'] = 'http';
 113         }
 114     }
 115
 116     /**
 117      * Returns a string representation of this URL
 118      */
 119     public function toString()
 120     {
 121         return unparse_url($this->parts);
 122     }
 123
 124     /**
 125      * Removes undesired query parameters
 126      */
 127     protected function cleanupQuery()
 128     {
 129         if (! isset($this->parts['query'])) {
 130             return;
 131         }
 132
 133         $queryParams = explode('&', $this->parts['query']);
 134
 135         foreach (self::$annoyingQueryParams as $annoying) {
 136             foreach ($queryParams as $param) {
 137                 if (startsWith($param, $annoying)) {
 138                     $queryParams = array_diff($queryParams, array($param));
 139                     continue;
 140                 }
 141             }
 142         }
 143
 144         if (count($queryParams) == 0) {
 145             unset($this->parts['query']);
 146             return;
 147         }
 148
 149         $this->parts['query'] = implode('&', $queryParams);
 150     }
 151
 152     /**
 153      * Removes undesired fragments
 154      */
 155     protected function cleanupFragment()
 156     {
 157         if (! isset($this->parts['fragment'])) {
 158             return;
 159         }
 160
 161         foreach (self::$annoyingFragments as $annoying) {
 162             if (startsWith($this->parts['fragment'], $annoying)) {
 163                 unset($this->parts['fragment']);
 164                 break;
 165             }
 166         }
 167     }
 168
 169     /**
 170      * Removes undesired query parameters and fragments
 171      *
 172      * @return string the string representation of this URL after cleanup
 173      */
 174     public function cleanup()
 175     {
 176         $this->cleanupQuery();
 177         $this->cleanupFragment();
 178         return $this->toString();
 179     }
 180
 181     /**
 182      * Get URL scheme.
 183      *
 184      * @return string the URL scheme or false if none is provided.
 185      */
 186     public function getScheme() {
 187         if (!isset($this->parts['scheme'])) {
 188             return false;
 189         }
 190         return $this->parts['scheme'];
 191     }
 192 }