From a0ee772423a67c1fdbad381a62a45d18efed0c57 Mon Sep 17 00:00:00 2001
From: "Edward Z. Yang" <edwardzyang@thewritingpot.com>
Date: Fri, 4 Aug 2006 01:47:48 +0000
Subject: [PATCH] Set up configuration class, implement attr_id_blacklist

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@155 48356398-32a2-884e-a903-53898d9a118a
---
 library/HTMLPurifier.php                      | 10 +++---
 library/HTMLPurifier/Config.php               | 36 +++++++++++++++++++
 library/HTMLPurifier/Definition.php           | 11 ++++--
 library/HTMLPurifier/Strategy.php             |  3 +-
 library/HTMLPurifier/Strategy/Composite.php   |  6 ++--
 .../Strategy/ValidateAttributes.php           | 11 +++++-
 tests/HTMLPurifier/Strategy/CompositeTest.php | 22 ++++++++----
 .../Strategy/ValidateAttributesTest.php       | 10 +++++-
 tests/HTMLPurifier/StrategyAbstractTest.php   |  8 +++--
 9 files changed, 95 insertions(+), 22 deletions(-)
 create mode 100644 library/HTMLPurifier/Config.php

diff --git a/library/HTMLPurifier.php b/library/HTMLPurifier.php
index ee849fa4..741fdd9a 100644
--- a/library/HTMLPurifier.php
+++ b/library/HTMLPurifier.php
@@ -34,12 +34,9 @@ class HTMLPurifier
     
     /**
      * Initializes the purifier.
-     * 
-     * The constructor instantiates all necessary sub-objects to do the job,
-     * because creating some of them (esp. HTMLPurifier_Definition) can be
-     * expensive.
+     * @param $config Configuration for all instances of the purifier
      */
-    function HTMLPurifier() {
+    function HTMLPurifier($config = null) {
         // unimplemented
     }
     
@@ -47,9 +44,10 @@ class HTMLPurifier
      * Purifies HTML.
      * 
      * @param $html String of HTML to purify
+     * @param $config HTMLPurifier_Config object for this specific round
      * @return Purified HTML
      */
-    function purify($html) {
+    function purify($html, $config = null) {
         // unimplemented
     }
     
diff --git a/library/HTMLPurifier/Config.php b/library/HTMLPurifier/Config.php
new file mode 100644
index 00000000..2de62b3d
--- /dev/null
+++ b/library/HTMLPurifier/Config.php
@@ -0,0 +1,36 @@
+<?php
+
+// subclass this to add custom settings
+class HTMLPurifier_Config
+{
+    
+    // which ids do we not allow?
+    var $attr_id_blacklist = array();
+    
+    //////////////////////////////////////////////////////////////////////////
+    // all below properties have not been implemented yet
+    
+    // prefix all ids with this
+    var $attr_id_prefix = '';
+    
+    // if there's a prefix, we may want to transparently rewrite the
+    // URLs we parse too.  However, we can only do it when it's a pure
+    // anchor link, so it's not foolproof
+    var $attr_id_rewrite_urls = false;
+    
+    // determines how the classes array should be construed:
+    // blacklist - allow allow except those in $classes_blacklist
+    // whitelist - only allow those in $classes_whitelist
+    // when one is chosen, the other has no effect
+    var $attr_class_mode = 'blacklist';
+    var $attr_class_blacklist = array();
+    var $attr_class_whitelist = array();
+    
+    function createDefault() {
+        $config = new HTMLPurifier_Config();
+        return $config;
+    }
+    
+}
+
+?>
\ No newline at end of file
diff --git a/library/HTMLPurifier/Definition.php b/library/HTMLPurifier/Definition.php
index 8a382b5f..99f67d1e 100644
--- a/library/HTMLPurifier/Definition.php
+++ b/library/HTMLPurifier/Definition.php
@@ -43,9 +43,16 @@ class HTMLPurifier_Definition
     // used solely by HTMLPurifier_Strategy_RemoveForeignElements
     var $info_tag_transform = array();
     
-    function instance() {
+    // WARNING! Prototype is not passed by reference, so in order to get
+    // a copy of the real one, you'll have to destroy your copy and
+    // use instance() to get it.
+    // Usually, however, modifying the returned definition (reference) should be
+    // sufficient
+    function &instance($prototype = null) {
         static $instance = null;
-        if (!$instance) {
+        if ($prototype) {
+            $instance = $prototype;
+        } elseif (!$instance) {
             $instance = new HTMLPurifier_Definition();
             $instance->setup();
         }
diff --git a/library/HTMLPurifier/Strategy.php b/library/HTMLPurifier/Strategy.php
index 257e9364..14dd7a18 100644
--- a/library/HTMLPurifier/Strategy.php
+++ b/library/HTMLPurifier/Strategy.php
@@ -15,9 +15,10 @@ class HTMLPurifier_Strategy
      * Executes the strategy on the tokens.
      * 
      * @param $tokens Array of HTMLPurifier_Token objects to be operated on.
+     * @param $config Configuration options
      * @returns Processed array of token objects.
      */
-    function execute($tokens) {
+    function execute($tokens, $config = null) {
         trigger_error('Cannot call abstract function', E_USER_ERROR);
     }
     
diff --git a/library/HTMLPurifier/Strategy/Composite.php b/library/HTMLPurifier/Strategy/Composite.php
index 9ddaaac6..178edae9 100644
--- a/library/HTMLPurifier/Strategy/Composite.php
+++ b/library/HTMLPurifier/Strategy/Composite.php
@@ -1,6 +1,7 @@
 <?php
 
 require_once 'HTMLPurifier/Strategy.php';
+require_once 'HTMLPurifier/Config.php';
 
 class HTMLPurifier_Strategy_Composite
 {
@@ -11,9 +12,10 @@ class HTMLPurifier_Strategy_Composite
         trigger_error('Attempt to instantiate abstract object', E_USER_ERROR);
     }
     
-    function execute($tokens) {
+    function execute($tokens, $config = null) {
+        if (!$config) $config = HTMLPurifier_Config::createDefault();
         foreach ($this->strategies as $strategy) {
-            $tokens = $strategy->execute($tokens);
+            $tokens = $strategy->execute($tokens, $config);
         }
         return $tokens;
     }
diff --git a/library/HTMLPurifier/Strategy/ValidateAttributes.php b/library/HTMLPurifier/Strategy/ValidateAttributes.php
index 417ec3cd..09897e4e 100644
--- a/library/HTMLPurifier/Strategy/ValidateAttributes.php
+++ b/library/HTMLPurifier/Strategy/ValidateAttributes.php
@@ -13,9 +13,18 @@ class HTMLPurifier_Strategy_ValidateAttributes extends HTMLPurifier_Strategy
         $this->definition = HTMLPurifier_Definition::instance();
     }
     
-    function execute($tokens) {
+    function execute($tokens, $config = null) {
+        
+        // load default configuration object if none passed
+        if (!$config) $config = HTMLPurifier_Config::createDefault();
+        
+        // setup ID accumulator and load it with blacklisted IDs
         $accumulator = new HTMLPurifier_IDAccumulator();
+        $accumulator->load($config->attr_id_blacklist);
+        
+        // DEFINITION CALL
         $d_defs = $this->definition->info_global_attr;
+        
         foreach ($tokens as $key => $token) {
             if ($token->type !== 'start' && $token->type !== 'end') continue;
             
diff --git a/tests/HTMLPurifier/Strategy/CompositeTest.php b/tests/HTMLPurifier/Strategy/CompositeTest.php
index 769971f7..b557b201 100644
--- a/tests/HTMLPurifier/Strategy/CompositeTest.php
+++ b/tests/HTMLPurifier/Strategy/CompositeTest.php
@@ -2,6 +2,7 @@
 
 require_once('HTMLPurifier/Strategy.php');
 require_once('HTMLPurifier/Strategy/Composite.php');
+require_once('HTMLPurifier/Config.php');
 
 class HTMLPurifier_Strategy_Composite_Test
     extends HTMLPurifier_Strategy_Composite
@@ -20,6 +21,7 @@ class HTMLPurifier_Strategy_CompositeTest extends UnitTestCase
     function test() {
         
         generate_mock_once('HTMLPurifier_Strategy');
+        generate_mock_once('HTMLPurifier_Config');
         
         // setup a bunch of mock strategies to inject into our composite test
         
@@ -39,18 +41,24 @@ class HTMLPurifier_Strategy_CompositeTest extends UnitTestCase
         $input_3 = 'Processed by 1 and 2';
         $input_4 = 'Processed by 1, 2 and 3'; // expected output
         
-        $mock_1->expectOnce('execute', array($input_1));
-        $mock_1->setReturnValue('execute', $input_2, array($input_1));
+        $config = new HTMLPurifier_ConfigMock();
         
-        $mock_2->expectOnce('execute', array($input_2));
-        $mock_2->setReturnValue('execute', $input_3, array($input_2));
+        $params_1 = array($input_1, $config);
+        $params_2 = array($input_2, $config);
+        $params_3 = array($input_3, $config);
         
-        $mock_3->expectOnce('execute', array($input_3));
-        $mock_3->setReturnValue('execute', $input_4, array($input_3));
+        $mock_1->expectOnce('execute', $params_1);
+        $mock_1->setReturnValue('execute', $input_2, $params_1);
+        
+        $mock_2->expectOnce('execute', $params_2);
+        $mock_2->setReturnValue('execute', $input_3, $params_2);
+        
+        $mock_3->expectOnce('execute', $params_3);
+        $mock_3->setReturnValue('execute', $input_4, $params_3);
         
         // perform test
         
-        $output = $composite->execute($input_1);
+        $output = $composite->execute($input_1, $config);
         $this->assertIdentical($input_4, $output);
         
         // tally the calls
diff --git a/tests/HTMLPurifier/Strategy/ValidateAttributesTest.php b/tests/HTMLPurifier/Strategy/ValidateAttributesTest.php
index 8a7cc919..54f92c47 100644
--- a/tests/HTMLPurifier/Strategy/ValidateAttributesTest.php
+++ b/tests/HTMLPurifier/Strategy/ValidateAttributesTest.php
@@ -1,5 +1,6 @@
 <?php
 
+require_once('HTMLPurifier/Config.php');
 require_once('HTMLPurifier/StrategyAbstractTest.php');
 require_once('HTMLPurifier/Strategy/ValidateAttributes.php');
 
@@ -13,6 +14,7 @@ class HTMLPurifier_Strategy_ValidateAttributesTest extends
         
         $inputs = array();
         $expect = array();
+        $config = array();
         
         $inputs[0] = '';
         $expect[0] = '';
@@ -38,7 +40,13 @@ class HTMLPurifier_Strategy_ValidateAttributesTest extends
         $inputs[6] = '<div id=" valid ">Trim whitespace.</div>';
         $expect[6] = '<div id="valid">Trim whitespace.</div>';
         
-        $this->assertStrategyWorks($strategy, $inputs, $expect);
+        // test configuration id blacklist
+        $inputs[7] = '<div id="invalid">Invalid</div>';
+        $expect[7] = '<div>Invalid</div>';
+        $config[7] = HTMLPurifier_Config::createDefault();
+        $config[7]->attr_id_blacklist = array('invalid');
+        
+        $this->assertStrategyWorks($strategy, $inputs, $expect, $config);
         
     }
     
diff --git a/tests/HTMLPurifier/StrategyAbstractTest.php b/tests/HTMLPurifier/StrategyAbstractTest.php
index 97ba1bc6..e6e9df92 100644
--- a/tests/HTMLPurifier/StrategyAbstractTest.php
+++ b/tests/HTMLPurifier/StrategyAbstractTest.php
@@ -25,10 +25,14 @@ class HTMLPurifier_StrategyAbstractTest extends UnitTestCase
         $this->gen = new HTMLPurifier_Generator();
     }
     
-    function assertStrategyWorks($strategy, $inputs, $expect) {
+    function assertStrategyWorks($strategy, $inputs, $expect, $config = array()) {
         foreach ($inputs as $i => $input) {
             $tokens = $this->lex->tokenizeHTML($input);
-            $result_tokens = $strategy->execute($tokens);
+            if (isset($config[$i])) {
+                $result_tokens = $strategy->execute($tokens, $config[$i]);
+            } else {
+                $result_tokens = $strategy->execute($tokens);
+            }
             $result = $this->gen->generateFromTokens($result_tokens);
             $this->assertEqual($expect[$i], $result, "Test $i: %s");
             paintIf($result, $result != $expect[$i]);