Implement URIScheme and subclasses except for mailto. Remove fragment from components, as it is scheme independent.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@218 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/AttrDef/URI.php

@@ -49,7 +49,9 @@ class HTMLPurifier_AttrDef_URI extends HTMLPurifier_AttrDef
             $scheme_obj =& $registry->getScheme($scheme, $config);
             if (!$scheme_obj) return ''; // invalid scheme, clean it out
         } else {
-            $scheme_obj =& $registry->getScheme($config->get('URI', 'DefaultScheme'), $config);
+            $scheme_obj =& $registry->getScheme(
+                $config->get('URI', 'DefaultScheme'), $config
+            );
         }
         
         
@@ -99,11 +101,8 @@ class HTMLPurifier_AttrDef_URI extends HTMLPurifier_AttrDef
             
             // userinfo and host are validated within the regexp
             
-            // regenerate authority
-            $authority =
-                ($userinfo === null ? '' : ($userinfo . '@')) .
-                $host .
-                ($port === null ? '' : (':' . $port));
+        } else {
+            $port = $host = $userinfo = null;
         }
         
         
@@ -120,10 +119,21 @@ class HTMLPurifier_AttrDef_URI extends HTMLPurifier_AttrDef
         
         
         // okay, now we defer execution to the subobject for more processing
-        list($authority, $path, $query, $fragment) = 
-        $scheme_obj->validateComponents($authority, $path, $query, $fragment);
+        // note that $fragment is omitted
+        list($userinfo, $host, $port, $path, $query) = 
+            $scheme_obj->validateComponents(
+                $userinfo, $host, $port, $path, $query, $config
+            );
         
         
+        // reconstruct authority
+        $authority = null;
+        if (!is_null($userinfo) || !is_null($host) || !is_null($port)) {
+            $authority = '';
+            if($userinfo !== null) $authority .= $userinfo . '@';
+            $authority .= $host;
+            if($port !== null) $authority .= ':' . $port;
+        }
         
         // reconstruct the result
         $result = '';

library/HTMLPurifier/URIScheme.php

@@ -3,8 +3,13 @@
 class HTMLPurifier_URIScheme
 {
     
-    function validateComponents($authority, $path, $query, $fragment) {
-        return array($authority, $path, $query, $fragment);
+    var $default_port = null;
+    
+    function validateComponents(
+        $userinfo, $host, $port, $path, $query, $config
+    ) {
+        if ($this->default_port == $port) $port = null;
+        return array($userinfo, $host, $port, $path, $query);
     }
     
 }

library/HTMLPurifier/URIScheme/ftp.php

@@ -0,0 +1,21 @@
+<?php
+
+require_once 'HTMLPurifier/URIScheme.php';
+
+class HTMLPurifier_URIScheme_ftp extends HTMLPurifier_URIScheme {
+    
+    var $default_port = 21;
+    
+    function validateComponents(
+        $userinfo, $host, $port, $path, $query, $config
+    ) {
+        list($userinfo, $host, $port, $path, $query) = 
+            parent::validateComponents(
+                $userinfo, $host, $port, $path, $query, $config );
+        // typecode check needed on path
+        return array($userinfo, $host, $port, $path, null);
+    }
+    
+}
+
+?>

library/HTMLPurifier/URIScheme/http.php

@@ -4,8 +4,15 @@ require_once 'HTMLPurifier/URIScheme.php';
 
 class HTMLPurifier_URIScheme_http extends HTMLPurifier_URIScheme {
     
-    function validateComponents($authority, $path, $query, $fragment) {
-        
+    var $default_port = 80;
+    
+    function validateComponents(
+        $userinfo, $host, $port, $path, $query, $config
+    ) {
+        list($userinfo, $host, $port, $path, $query) = 
+            parent::validateComponents(
+                $userinfo, $host, $port, $path, $query, $config );
+        return array(null, $host, $port, $path, $query);
     }
     
 }

library/HTMLPurifier/URIScheme/https.php

@@ -0,0 +1,11 @@
+<?php
+
+require_once 'HTMLPurifier/URIScheme/http.php';
+
+class HTMLPurifier_URIScheme_https extends HTMLPurifier_URIScheme_http {
+    
+    var $default_port = 443;
+    
+}
+
+?>

library/HTMLPurifier/URIScheme/news.php

@@ -0,0 +1,19 @@
+<?php
+
+require_once 'HTMLPurifier/URIScheme.php';
+
+class HTMLPurifier_URIScheme_news extends HTMLPurifier_URIScheme {
+    
+    function validateComponents(
+        $userinfo, $host, $port, $path, $query, $config
+    ) {
+        list($userinfo, $host, $port, $path, $query) = 
+            parent::validateComponents(
+                $userinfo, $host, $port, $path, $query, $config );
+        // typecode check needed on path
+        return array(null, null, null, $path, null);
+    }
+    
+}
+
+?>

library/HTMLPurifier/URIScheme/nntp.php

@@ -0,0 +1,20 @@
+<?php
+
+require_once 'HTMLPurifier/URIScheme.php';
+
+class HTMLPurifier_URIScheme_nntp extends HTMLPurifier_URIScheme {
+    
+    var $default_port = 119;
+    
+    function validateComponents(
+        $userinfo, $host, $port, $path, $query, $config
+    ) {
+        list($userinfo, $host, $port, $path, $query) = 
+            parent::validateComponents(
+                $userinfo, $host, $port, $path, $query, $config );
+        return array(null, $host, $port, $path, null);
+    }
+    
+}
+
+?>