[2.1.0] Implement MakeAbsolute URI filter

- Move some directives with complex dependencies to URIDefinition
- Fix a missing extends
- Add hierarchical information to URI schemes
- Fix bug in URIHarness.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1346 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/AttrDef/URI.php

@@ -7,54 +7,59 @@ require_once 'HTMLPurifier/URISchemeRegistry.php';
 require_once 'HTMLPurifier/AttrDef/URI/Host.php';
 require_once 'HTMLPurifier/PercentEncoder.php';
 
-HTMLPurifier_ConfigSchema::define(
-    'URI', 'DefaultScheme', 'http', 'string',
-    'Defines through what scheme the output will be served, in order to '.
-    'select the proper object validator when no scheme information is present.'
-);
+// special case filtering directives 
 
 HTMLPurifier_ConfigSchema::define(
-    'URI', 'Host', null, 'string/null',
-    'Defines the domain name of the server, so we can determine whether or '.
-    'an absolute URI is from your website or not.  Not strictly necessary, '.
-    'as users should be using relative URIs to reference resources on your '.
-    'website.  It will, however, let you use absolute URIs to link to '.
-    'subdomains of the domain you post here: i.e. example.com will allow '.
-    'sub.example.com.  However, higher up domains will still be excluded: '.
-    'if you set %URI.Host to sub.example.com, example.com will be blocked. '.
-    'This directive has been available since 1.2.0.'
-);
+    'URI', 'Munge', null, 'string/null', '
+<p>
+    Munges all browsable (usually http, https and ftp)
+    absolute URI\'s into another URI, usually a URI redirection service.
+    This directive accepts a URI, formatted with a <code>%s</code> where 
+    the url-encoded original URI should be inserted (sample: 
+    <code>http://www.google.com/url?q=%s</code>).
+</p>
+<p>
+    Uses for this directive:
+</p>
+<ul>
+    <li>
+        Prevent PageRank leaks, while being fairly transparent 
+        to users (you may also want to add some client side JavaScript to 
+        override the text in the statusbar). <strong>Notice</strong>:
+        Many security experts believe that this form of protection does not deter spam-bots. 
+    </li>
+    <li>
+        Redirect users to a splash page telling them they are leaving your
+        website. While this is poor usability practice, it is often mandated
+        in corporate environments.
+    </li>
+</ul>
+<p>
+    This directive has been available since 1.3.0.
+</p>
+');
+
+// disabling directives
 
 HTMLPurifier_ConfigSchema::define(
-    'URI', 'DisableResources', false, 'bool',
-    'Disables embedding resources, essentially meaning no pictures. You can '.
-    'still link to them though. See %URI.DisableExternalResources for why '.
-    'this might be a good idea. This directive has been available since 1.3.0.'
-);
-
-HTMLPurifier_ConfigSchema::define(
-    'URI', 'Munge', null, 'string/null',
-    'Munges all browsable (usually http, https and ftp) URI\'s into some URL '.
-    'redirection service. Pass this directive a URI, with %s inserted where '.
-    'the url-encoded original URI should be inserted (sample: '.
-    '<code>http://www.google.com/url?q=%s</code>). '.
-    'This prevents PageRank leaks, while being as transparent as possible '.
-    'to users (you may also want to add some client side JavaScript to '.
-    'override the text in the statusbar). Warning: many security experts '.
-    'believe that this form of protection does not deter spam-bots. '.
-    'You can also use this directive to redirect users to a splash page '.
-    'telling them they are leaving your website. '.
-    'This directive has been available since 1.3.0.'
-);
-
-HTMLPurifier_ConfigSchema::define(
-    'URI', 'Disable', false, 'bool',
-    'Disables all URIs in all forms. Not sure why you\'d want to do that '.
-    '(after all, the Internet\'s founded on the notion of a hyperlink). '.
-    'This directive has been available since 1.3.0.'
-);
+    'URI', 'Disable', false, 'bool', '
+<p>
+    Disables all URIs in all forms. Not sure why you\'d want to do that 
+    (after all, the Internet\'s founded on the notion of a hyperlink). 
+    This directive has been available since 1.3.0.
+</p>
+');
 HTMLPurifier_ConfigSchema::defineAlias('Attr', 'DisableURI', 'URI', 'Disable');
 
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'DisableResources', false, 'bool', '
+<p>
+    Disables embedding resources, essentially meaning no pictures. You can 
+    still link to them though. See %URI.DisableExternalResources for why 
+    this might be a good idea. This directive has been available since 1.3.0.
+</p>
+');
+
 /**
  * Validates a URI as defined by RFC 3986.
  * @note Scheme-specific mechanics deferred to HTMLPurifier_URIScheme
@@ -118,7 +123,7 @@ class HTMLPurifier_AttrDef_URI extends HTMLPurifier_AttrDef
         
         // munge scheme off if necessary (this must be last)
         if (!is_null($uri->scheme) && is_null($uri->host)) {
-            if ($config->get('URI', 'DefaultScheme') == $uri->scheme) {
+            if ($uri_def->defaultScheme == $uri->scheme) {
                 $uri->scheme = null;
             }
         }

library/HTMLPurifier/URI.php

@@ -37,11 +37,12 @@ class HTMLPurifier_URI
             if (!$scheme_obj) return false; // invalid scheme, clean it out
         } else {
             // no scheme: retrieve the default one
-            $scheme_obj = $registry->getScheme($config->get('URI', 'DefaultScheme'), $config, $context);
+            $def = $config->getDefinition('URI');
+            $scheme_obj = $registry->getScheme($def->defaultScheme, $config, $context);
             if (!$scheme_obj) {
                 // something funky happened to the default scheme object
                 trigger_error(
-                    'Default scheme object "' . $config->get('URI', 'DefaultScheme') . '" was not readable',
+                    'Default scheme object "' . $def->defaultScheme . '" was not readable',
                     E_USER_WARNING
                 );
                 return false;
@@ -107,5 +108,12 @@ class HTMLPurifier_URI
         return $result;
     }
     
+    /**
+     * Returns a copy of the URI object
+     */
+    function copy() {
+        return unserialize(serialize($this));
+    }
+    
 }
 

library/HTMLPurifier/URIDefinition.php

@@ -2,10 +2,12 @@
 
 require_once 'HTMLPurifier/Definition.php';
 require_once 'HTMLPurifier/URIFilter.php';
+require_once 'HTMLPurifier/URIParser.php';
 
 require_once 'HTMLPurifier/URIFilter/DisableExternal.php';
 require_once 'HTMLPurifier/URIFilter/DisableExternalResources.php';
 require_once 'HTMLPurifier/URIFilter/HostBlacklist.php';
+require_once 'HTMLPurifier/URIFilter/MakeAbsolute.php';
 
 HTMLPurifier_ConfigSchema::define(
     'URI', 'DefinitionID', null, 'string/null', '
@@ -25,6 +27,48 @@ HTMLPurifier_ConfigSchema::define(
 </p>
 ');
 
+// informative URI directives
+
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'DefaultScheme', 'http', 'string', '
+<p>
+    Defines through what scheme the output will be served, in order to 
+    select the proper object validator when no scheme information is present.
+</p>
+');
+
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'Host', null, 'string/null', '
+<p>
+    Defines the domain name of the server, so we can determine whether or 
+    an absolute URI is from your website or not.  Not strictly necessary, 
+    as users should be using relative URIs to reference resources on your 
+    website.  It will, however, let you use absolute URIs to link to 
+    subdomains of the domain you post here: i.e. example.com will allow 
+    sub.example.com.  However, higher up domains will still be excluded: 
+    if you set %URI.Host to sub.example.com, example.com will be blocked. 
+    <strong>Note:</strong> This directive overrides %URI.Base because
+    a given page may be on a sub-domain, but you wish HTML Purifier to be
+    more relaxed and allow some of the parent domains too.
+    This directive has been available since 1.2.0.
+</p>
+');
+
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'Base', null, 'string/null', '
+<p>
+    The base URI is the URI of the document this purified HTML will be
+    inserted into.  This information is important if HTML Purifier needs
+    to calculate absolute URIs from relative URIs, such as when %URI.MakeAbsolute
+    is on.  You may use a non-absolute URI for this value, but behavior
+    may vary (%URI.MakeAbsolute deals nicely with both absolute and 
+    relative paths, but forwards-compatibility is not guaranteed).
+    <strong>Warning:</strong> If set, the scheme on this URI
+    overrides the one specified by %URI.DefaultScheme. This directive has
+    been available since 2.1.0.
+</p>
+');
+
 class HTMLPurifier_URIDefinition extends HTMLPurifier_Definition
 {
     
@@ -32,10 +76,26 @@ class HTMLPurifier_URIDefinition extends HTMLPurifier_Definition
     var $filters = array();
     var $registeredFilters = array();
     
+    /**
+     * HTMLPurifier_URI object of the base specified at %URI.Base
+     */
+    var $base;
+    
+    /**
+     * String host to consider "home" base
+     */
+    var $host;
+    
+    /**
+     * Name of default scheme based on %URI.DefaultScheme and %URI.Base
+     */
+    var $defaultScheme;
+    
     function HTMLPurifier_URIDefinition() {
         $this->registerFilter(new HTMLPurifier_URIFilter_DisableExternal());
         $this->registerFilter(new HTMLPurifier_URIFilter_DisableExternalResources());
         $this->registerFilter(new HTMLPurifier_URIFilter_HostBlacklist());
+        $this->registerFilter(new HTMLPurifier_URIFilter_MakeAbsolute());
     }
     
     function registerFilter($filter) {
@@ -43,6 +103,11 @@ class HTMLPurifier_URIDefinition extends HTMLPurifier_Definition
     }
     
     function doSetup($config) {
+        $this->setupFilters($config);
+        $this->setupMemberVariables($config);
+    }
+    
+    function setupFilters($config) {
         foreach ($this->registeredFilters as $name => $filter) {
             $conf = $config->get('URI', $name);
             if ($conf !== false && $conf !== null) {
@@ -53,6 +118,18 @@ class HTMLPurifier_URIDefinition extends HTMLPurifier_Definition
         unset($this->registeredFilters);
     }
     
+    function setupMemberVariables($config) {
+        $this->host = $config->get('URI', 'Host');
+        $base_uri = $config->get('URI', 'Base');
+        if (!is_null($base_uri)) {
+            $parser = new HTMLPurifier_URIParser();
+            $this->base = $parser->parse($base_uri);
+            $this->defaultScheme = $this->base->scheme;
+            if (is_null($this->host)) $this->host = $this->base->host;
+        }
+        if (is_null($this->defaultScheme)) $this->defaultScheme = $config->get('URI', 'DefaultScheme');
+    }
+    
     function filter(&$uri, $config, &$context) {
         foreach ($this->filters as $name => $x) {
             $result = $this->filters[$name]->filter($uri, $config, $context);

library/HTMLPurifier/URIFilter/HostBlacklist.php

@@ -10,7 +10,7 @@ HTMLPurifier_ConfigSchema::define(
     'This directive has been available since 1.3.0.'
 );
 
-class HTMLPurifier_URIFilter_HostBlacklist
+class HTMLPurifier_URIFilter_HostBlacklist extends HTMLPurifier_URIFilter
 {
     var $name = 'HostBlacklist';
     var $blacklist = array();

library/HTMLPurifier/URIFilter/MakeAbsolute.php

@@ -0,0 +1,115 @@
+<?php
+
+// does not support network paths
+
+require_once 'HTMLPurifier/URIFilter.php';
+
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'MakeAbsolute', false, 'bool', '
+<p>
+    Converts all URIs into absolute forms. This is useful when the HTML
+    being filtered assumes a specific base path, but will actually be
+    viewed in a different context (and setting an alternate base URI is
+    not possible). %URI.Base must be set for this directive to work.
+    This directive has been available since 2.1.0.
+</p>
+');
+
+class HTMLPurifier_URIFilter_MakeAbsolute extends HTMLPurifier_URIFilter
+{
+    var $name = 'MakeAbsolute';
+    var $base;
+    var $basePathStack = array();
+    function prepare($config) {
+        $def = $config->getDefinition('URI');
+        $this->base = $def->base;
+        if (is_null($this->base)) {
+            trigger_error('URI.MakeAbsolute is being ignored due to lack of value for URI.Base configuration', E_USER_ERROR);
+            return;
+        }
+        $this->base->fragment = null; // fragment is invalid for base URI
+        $stack = explode('/', $this->base->path);
+        array_pop($stack); // discard last segment
+        $stack = $this->_collapseStack($stack); // do pre-parsing
+        $this->basePathStack = $stack;
+    }
+    function filter(&$uri, $config, &$context) {
+        if (is_null($this->base)) return true; // abort early
+        if (
+            $uri->path === '' && is_null($uri->scheme) &&
+            is_null($uri->host) && is_null($uri->query) && is_null($uri->fragment)
+        ) {
+            // reference to current document
+            $uri = $this->base->copy();
+            return true;
+        }
+        if (!is_null($uri->scheme)) {
+            // absolute URI already: don't change
+            if (!is_null($uri->host)) return true;
+            $scheme_obj = $uri->getSchemeObj($config, $context);
+            if (!$scheme_obj->hierarchical) {
+                // non-hierarchal URI with explicit scheme, don't change
+                return true;
+            }
+            // special case: had a scheme but always is hierarchical and had no authority
+        }
+        if (!is_null($uri->host)) {
+            // network path, don't bother
+            return true;
+        }
+        if ($uri->path === '') {
+            $uri->path = $this->base->path;
+        }elseif ($uri->path[0] !== '/') {
+            // relative path, needs more complicated processing
+            $stack = explode('/', $uri->path);
+            $new_stack = array_merge($this->basePathStack, $stack);
+            $new_stack = $this->_collapseStack($new_stack);
+            $uri->path = implode('/', $new_stack);
+        }
+        // re-combine
+        $uri->scheme = $this->base->scheme;
+        if (is_null($uri->userinfo)) $uri->userinfo = $this->base->userinfo;
+        if (is_null($uri->host))     $uri->host     = $this->base->host;
+        if (is_null($uri->port))     $uri->port     = $this->base->port;
+        return true;
+    }
+    
+    /**
+     * Resolve dots and double-dots in a path stack
+     * @private
+     */
+    function _collapseStack($stack) {
+        $result = array();
+        for ($i = 0; isset($stack[$i]); $i++) {
+            $is_folder = false;
+            // absorb an internally duplicated slash
+            if ($stack[$i] == '' && $i && isset($stack[$i+1])) continue;
+            if ($stack[$i] == '..') {
+                if (!empty($result)) {
+                    $segment = array_pop($result);
+                    if ($segment === '' && empty($result)) {
+                        // error case: attempted to back out too far:
+                        // restore the leading slash
+                        $result[] = '';
+                    } elseif ($segment === '..') {
+                        $result[] = '..'; // cannot remove .. with ..
+                    }
+                } else {
+                    // relative path, preserve the double-dots
+                    $result[] = '..';
+                }
+                $is_folder = true;
+                continue;
+            }
+            if ($stack[$i] == '.') {
+                // silently absorb
+                $is_folder = true;
+                continue;
+            }
+            $result[] = $stack[$i];
+        }
+        if ($is_folder) $result[] = '';
+        return $result;
+    }
+}
+

library/HTMLPurifier/URIScheme.php

@@ -19,6 +19,12 @@ class HTMLPurifier_URIScheme
      */
     var $browsable = false;
     
+    /**
+     * Whether or not the URI always uses <hier_part>, resolves edge cases
+     * with making relative URIs absolute
+     */
+    var $hierarchical = false;
+    
     /**
      * Validates the components of a URI
      * @note This implementation should be called by children if they define

library/HTMLPurifier/URIScheme/ftp.php

@@ -9,6 +9,7 @@ class HTMLPurifier_URIScheme_ftp extends HTMLPurifier_URIScheme {
     
     var $default_port = 21;
     var $browsable = true; // usually
+    var $hierarchical = true;
     
     function validate(&$uri, $config, &$context) {
         parent::validate($uri, $config, $context);

library/HTMLPurifier/URIScheme/http.php

@@ -9,6 +9,7 @@ class HTMLPurifier_URIScheme_http extends HTMLPurifier_URIScheme {
     
     var $default_port = 80;
     var $browsable = true;
+    var $hierarchical = true;
     
     function validate(&$uri, $config, &$context) {
         parent::validate($uri, $config, $context);