application/http/UrlUtils.php

   1 <?php
   2
   3 /**
   4  * Converts an array-represented URL to a string
   5  *
   6  * Source: http://php.net/manual/en/function.parse-url.php#106731
   7  *
   8  * @see http://php.net/manual/en/function.parse-url.php
   9  *
  10  * @param array $parsedUrl an array-represented URL
  11  *
  12  * @return string the string representation of the URL
  13  */
  14 function unparse_url($parsedUrl)
  15 {
  16     $scheme   = isset($parsedUrl['scheme']) ? $parsedUrl['scheme'] . '://' : '';
  17     $host     = isset($parsedUrl['host']) ? $parsedUrl['host'] : '';
  18     $port     = isset($parsedUrl['port']) ? ':' . $parsedUrl['port'] : '';
  19     $user     = isset($parsedUrl['user']) ? $parsedUrl['user'] : '';
  20     $pass     = isset($parsedUrl['pass']) ? ':' . $parsedUrl['pass']  : '';
  21     $pass     = ($user || $pass) ? "$pass@" : '';
  22     $path     = isset($parsedUrl['path']) ? $parsedUrl['path'] : '';
  23     $query    = isset($parsedUrl['query']) ? '?' . $parsedUrl['query'] : '';
  24     $fragment = isset($parsedUrl['fragment']) ? '#' . $parsedUrl['fragment'] : '';
  25
  26     return "$scheme$user$pass$host$port$path$query$fragment";
  27 }
  28
  29 /**
  30  * Removes undesired query parameters and fragments
  31  *
  32  * @param string url UrlUtils to be cleaned
  33  *
  34  * @return string the string representation of this URL after cleanup
  35  */
  36 function cleanup_url($url)
  37 {
  38     $obj_url = new \Shaarli\Http\Url($url);
  39     return $obj_url->cleanup();
  40 }
  41
  42 /**
  43  * Get URL scheme.
  44  *
  45  * @param string url UrlUtils for which the scheme is requested
  46  *
  47  * @return mixed the URL scheme or false if none is provided.
  48  */
  49 function get_url_scheme($url)
  50 {
  51     $obj_url = new \Shaarli\Http\Url($url);
  52     return $obj_url->getScheme();
  53 }
  54
  55 /**
  56  * Adds a trailing slash at the end of URL if necessary.
  57  *
  58  * @param string $url URL to check/edit.
  59  *
  60  * @return string $url URL with a end trailing slash.
  61  */
  62 function add_trailing_slash($url)
  63 {
  64     return $url . (!endsWith($url, '/') ? '/' : '');
  65 }
  66
  67 /**
  68  * Replace not whitelisted protocols by 'http://' from given URL.
  69  *
  70  * @param string $url       URL to clean
  71  * @param array  $protocols List of allowed protocols (aside from http(s)).
  72  *
  73  * @return string URL with allowed protocol
  74  */
  75 function whitelist_protocols($url, $protocols)
  76 {
  77     if (startsWith($url, '?') || startsWith($url, '/') || startsWith($url, '#')) {
  78         return $url;
  79     }
  80     $protocols = array_merge(['http', 'https'], $protocols);
  81     $protocol = preg_match('#^(\w+):/?/?#', $url, $match);
  82     // Protocol not allowed: we remove it and replace it with http
  83     if ($protocol === 1 && ! in_array($match[1], $protocols)) {
  84         $url = str_replace($match[0], 'http://', $url);
  85     } elseif ($protocol !== 1) {
  86         $url = 'http://' . $url;
  87     }
  88     return $url;
  89 }