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  * Adds a trailing slash at the end of URL if necessary.
  56  *
  57  * @param string $url URL to check/edit.
  58  *
  59  * @return string $url URL with a end trailing slash.
  60  */
  61 function add_trailing_slash($url)
  62 {
  63     return $url . (!endsWith($url, '/') ? '/' : '');
  64 }
  65
  66 /**
  67  * URL representation and cleanup utilities
  68  *
  69  * Form
  70  *   scheme://[username:password@]host[:port][/path][?query][#fragment]
  71  *
  72  * Examples
  73  *   http://username:password@hostname:9090/path?arg1=value1&arg2=value2#anchor
  74  *   https://host.name.tld
  75  *   https://h2.g2/faq/?vendor=hitchhiker&item=guide&dest=galaxy#answer
  76  *
  77  * @see http://www.faqs.org/rfcs/rfc3986.html
  78  */
  79 class Url
  80 {
  81     private static $annoyingQueryParams = array(
  82         // Facebook
  83         'action_object_map=',
  84         'action_ref_map=',
  85         'action_type_map=',
  86         'fb_',
  87         'fb=',
  88
  89         // Scoop.it
  90         '__scoop',
  91
  92         // Google Analytics & FeedProxy
  93         'utm_',
  94
  95         // ATInternet
  96         'xtor='
  97     );
  98
  99     private static $annoyingFragments = array(
 100         // ATInternet
 101         'xtor=RSS-',
 102
 103         // Misc.
 104         'tk.rss_all'
 105     );
 106
 107     /*
 108      * URL parts represented as an array
 109      *
 110      * @see http://php.net/parse_url
 111      */
 112     protected $parts;
 113
 114     /**
 115      * Parses a string containing a URL
 116      *
 117      * @param string $url a string containing a URL
 118      */
 119     public function __construct($url)
 120     {
 121         $this->parts = parse_url(trim($url));
 122
 123         if (!empty($url) && empty($this->parts['scheme'])) {
 124             $this->parts['scheme'] = 'http';
 125         }
 126     }
 127
 128     /**
 129      * Returns a string representation of this URL
 130      */
 131     public function toString()
 132     {
 133         return unparse_url($this->parts);
 134     }
 135
 136     /**
 137      * Removes undesired query parameters
 138      */
 139     protected function cleanupQuery()
 140     {
 141         if (! isset($this->parts['query'])) {
 142             return;
 143         }
 144
 145         $queryParams = explode('&', $this->parts['query']);
 146
 147         foreach (self::$annoyingQueryParams as $annoying) {
 148             foreach ($queryParams as $param) {
 149                 if (startsWith($param, $annoying)) {
 150                     $queryParams = array_diff($queryParams, array($param));
 151                     continue;
 152                 }
 153             }
 154         }
 155
 156         if (count($queryParams) == 0) {
 157             unset($this->parts['query']);
 158             return;
 159         }
 160
 161         $this->parts['query'] = implode('&', $queryParams);
 162     }
 163
 164     /**
 165      * Removes undesired fragments
 166      */
 167     protected function cleanupFragment()
 168     {
 169         if (! isset($this->parts['fragment'])) {
 170             return;
 171         }
 172
 173         foreach (self::$annoyingFragments as $annoying) {
 174             if (startsWith($this->parts['fragment'], $annoying)) {
 175                 unset($this->parts['fragment']);
 176                 break;
 177             }
 178         }
 179     }
 180
 181     /**
 182      * Removes undesired query parameters and fragments
 183      *
 184      * @return string the string representation of this URL after cleanup
 185      */
 186     public function cleanup()
 187     {
 188         $this->cleanupQuery();
 189         $this->cleanupFragment();
 190         return $this->toString();
 191     }
 192
 193     /**
 194      * Get URL scheme.
 195      *
 196      * @return string the URL scheme or false if none is provided.
 197      */
 198     public function getScheme() {
 199         if (!isset($this->parts['scheme'])) {
 200             return false;
 201         }
 202         return $this->parts['scheme'];
 203     }
 204
 205     /**
 206      * Test if the Url is an HTTP one.
 207      *
 208      * @return true is HTTP, false otherwise.
 209      */
 210     public function isHttp() {
 211         return strpos(strtolower($this->parts['scheme']), 'http') !== false;
 212     }
 213 }