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  * Converts an URL with an IDN host to a ASCII one.
  67  *
  68  * @param string $url Input URL.
  69  *
  70  * @return string converted URL.
  71  */
  72 function url_with_idn_to_ascii($url)
  73 {
  74     $parts = parse_url($url);
  75     $parts['host'] = idn_to_ascii($parts['host']);
  76
  77     $httpUrl = new \http\Url($parts);
  78     return $httpUrl->toString();
  79 }
  80 /**
  81  * URL representation and cleanup utilities
  82  *
  83  * Form
  84  *   scheme://[username:password@]host[:port][/path][?query][#fragment]
  85  *
  86  * Examples
  87  *   http://username:password@hostname:9090/path?arg1=value1&arg2=value2#anchor
  88  *   https://host.name.tld
  89  *   https://h2.g2/faq/?vendor=hitchhiker&item=guide&dest=galaxy#answer
  90  *
  91  * @see http://www.faqs.org/rfcs/rfc3986.html
  92  */
  93 class Url
  94 {
  95     private static $annoyingQueryParams = array(
  96         // Facebook
  97         'action_object_map=',
  98         'action_ref_map=',
  99         'action_type_map=',
 100         'fb_',
 101         'fb=',
 102
 103         // Scoop.it
 104         '__scoop',
 105
 106         // Google Analytics & FeedProxy
 107         'utm_',
 108
 109         // ATInternet
 110         'xtor='
 111     );
 112
 113     private static $annoyingFragments = array(
 114         // ATInternet
 115         'xtor=RSS-',
 116
 117         // Misc.
 118         'tk.rss_all'
 119     );
 120
 121     /*
 122      * URL parts represented as an array
 123      *
 124      * @see http://php.net/parse_url
 125      */
 126     protected $parts;
 127
 128     /**
 129      * Parses a string containing a URL
 130      *
 131      * @param string $url a string containing a URL
 132      */
 133     public function __construct($url)
 134     {
 135         $url = self::cleanupUnparsedUrl(trim($url));
 136         $this->parts = parse_url($url);
 137
 138         if (!empty($url) && empty($this->parts['scheme'])) {
 139             $this->parts['scheme'] = 'http';
 140         }
 141     }
 142
 143     /**
 144      * Clean up URL before it's parsed.
 145      * ie. handle urlencode, url prefixes, etc.
 146      *
 147      * @param string $url URL to clean.
 148      *
 149      * @return string cleaned URL.
 150      */
 151     protected static function cleanupUnparsedUrl($url)
 152     {
 153         return self::removeFirefoxAboutReader($url);
 154     }
 155
 156     /**
 157      * Remove Firefox Reader prefix if it's present.
 158      *
 159      * @param string $input url
 160      *
 161      * @return string cleaned url
 162      */
 163     protected static function removeFirefoxAboutReader($input)
 164     {
 165         $firefoxPrefix = 'about://reader?url=';
 166         if (startsWith($input, $firefoxPrefix)) {
 167             return urldecode(ltrim($input, $firefoxPrefix));
 168         }
 169         return $input;
 170     }
 171
 172     /**
 173      * Returns a string representation of this URL
 174      */
 175     public function toString()
 176     {
 177         return unparse_url($this->parts);
 178     }
 179
 180     /**
 181      * Removes undesired query parameters
 182      */
 183     protected function cleanupQuery()
 184     {
 185         if (! isset($this->parts['query'])) {
 186             return;
 187         }
 188
 189         $queryParams = explode('&', $this->parts['query']);
 190
 191         foreach (self::$annoyingQueryParams as $annoying) {
 192             foreach ($queryParams as $param) {
 193                 if (startsWith($param, $annoying)) {
 194                     $queryParams = array_diff($queryParams, array($param));
 195                     continue;
 196                 }
 197             }
 198         }
 199
 200         if (count($queryParams) == 0) {
 201             unset($this->parts['query']);
 202             return;
 203         }
 204
 205         $this->parts['query'] = implode('&', $queryParams);
 206     }
 207
 208     /**
 209      * Removes undesired fragments
 210      */
 211     protected function cleanupFragment()
 212     {
 213         if (! isset($this->parts['fragment'])) {
 214             return;
 215         }
 216
 217         foreach (self::$annoyingFragments as $annoying) {
 218             if (startsWith($this->parts['fragment'], $annoying)) {
 219                 unset($this->parts['fragment']);
 220                 break;
 221             }
 222         }
 223     }
 224
 225     /**
 226      * Removes undesired query parameters and fragments
 227      *
 228      * @return string the string representation of this URL after cleanup
 229      */
 230     public function cleanup()
 231     {
 232         $this->cleanupQuery();
 233         $this->cleanupFragment();
 234         return $this->toString();
 235     }
 236
 237     /**
 238      * Converts an URL with an International Domain Name host to a ASCII one.
 239      * This requires PHP-intl. If it's not available, just returns this->cleanup().
 240      *
 241      * @return string converted cleaned up URL.
 242      */
 243     public function indToAscii()
 244     {
 245         $out = $this->cleanup();
 246         if (! function_exists('idn_to_ascii') || ! isset($this->parts['host'])) {
 247             return $out;
 248         }
 249         $asciiHost = idn_to_ascii($this->parts['host']);
 250         return str_replace($this->parts['host'], $asciiHost, $out);
 251     }
 252
 253     /**
 254      * Get URL scheme.
 255      *
 256      * @return string the URL scheme or false if none is provided.
 257      */
 258     public function getScheme() {
 259         if (!isset($this->parts['scheme'])) {
 260             return false;
 261         }
 262         return $this->parts['scheme'];
 263     }
 264
 265     /**
 266      * Get URL host.
 267      *
 268      * @return string the URL host or false if none is provided.
 269      */
 270     public function getHost() {
 271         if (empty($this->parts['host'])) {
 272             return false;
 273         }
 274         return $this->parts['host'];
 275     }
 276
 277     /**
 278      * Test if the Url is an HTTP one.
 279      *
 280      * @return true is HTTP, false otherwise.
 281      */
 282     public function isHttp() {
 283         return strpos(strtolower($this->parts['scheme']), 'http') !== false;
 284     }
 285 }