lib/htmlpurifier/HTMLPurifier/URIParser.php

   1 <?php
   2
   3 require_once 'HTMLPurifier/URI.php';
   4
   5 /**
   6  * Parses a URI into the components and fragment identifier as specified
   7  * by RFC 2396.
   8  * @todo Replace regexps with a native PHP parser
   9  */
  10 class HTMLPurifier_URIParser
  11 {
  12
  13     /**
  14      * Parses a URI
  15      * @param $uri string URI to parse
  16      * @return HTMLPurifier_URI representation of URI
  17      */
  18     function parse($uri) {
  19         $r_URI = '!'.
  20             '(([^:/?#<>\'"]+):)?'. // 2. Scheme
  21             '(//([^/?#<>\'"]*))?'. // 4. Authority
  22             '([^?#<>\'"]*)'.       // 5. Path
  23             '(\?([^#<>\'"]*))?'.   // 7. Query
  24             '(#([^<>\'"]*))?'.     // 8. Fragment
  25             '!';
  26
  27         $matches = array();
  28         $result = preg_match($r_URI, $uri, $matches);
  29
  30         if (!$result) return false; // *really* invalid URI
  31
  32         // seperate out parts
  33         $scheme     = !empty($matches[1]) ? $matches[2] : null;
  34         $authority  = !empty($matches[3]) ? $matches[4] : null;
  35         $path       = $matches[5]; // always present, can be empty
  36         $query      = !empty($matches[6]) ? $matches[7] : null;
  37         $fragment   = !empty($matches[8]) ? $matches[9] : null;
  38
  39         // further parse authority
  40         if ($authority !== null) {
  41             // ridiculously inefficient: it's a stacked regex!
  42             $HEXDIG = '[A-Fa-f0-9]';
  43             $unreserved = 'A-Za-z0-9-._~'; // make sure you wrap with []
  44             $sub_delims = '!$&\'()'; // needs []
  45             $pct_encoded = "%$HEXDIG$HEXDIG";
  46             $r_userinfo = "(?:[$unreserved$sub_delims:]|$pct_encoded)*";
  47             $r_authority = "/^(($r_userinfo)@)?(\[[^\]]+\]|[^:]*)(:(\d*))?/";
  48             $matches = array();
  49             preg_match($r_authority, $authority, $matches);
  50             $userinfo   = !empty($matches[1]) ? $matches[2] : null;
  51             $host       = !empty($matches[3]) ? $matches[3] : '';
  52             $port       = !empty($matches[4]) ? (int) $matches[5] : null;
  53         } else {
  54             $port = $host = $userinfo = null;
  55         }
  56
  57         return new HTMLPurifier_URI(
  58             $scheme, $userinfo, $host, $port, $path, $query, $fragment);
  59     }
  60
  61 }
  62