From e5191b3ada7a9a23d0039972b88ea99a447da092 Mon Sep 17 00:00:00 2001
From: "Edward Z. Yang" <edwardzyang@thewritingpot.com>
Date: Sat, 23 Jun 2007 20:52:57 +0000
Subject: [PATCH] [2.0.1] Scrap auto_close in favor of ChildDef->elements
 heuristic.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1213 48356398-32a2-884e-a903-53898d9a118a
---
 NEWS                                             |  2 ++
 library/HTMLPurifier/ChildDef.php                |  5 +++++
 library/HTMLPurifier/ChildDef/Chameleon.php      |  1 +
 library/HTMLPurifier/ChildDef/Custom.php         |  6 ++++++
 library/HTMLPurifier/ChildDef/Required.php       |  2 +-
 library/HTMLPurifier/ChildDef/Table.php          |  2 ++
 library/HTMLPurifier/ElementDef.php              |  8 --------
 library/HTMLPurifier/HTMLModule/List.php         |  3 +--
 library/HTMLPurifier/HTMLModule/Tables.php       |  5 +----
 library/HTMLPurifier/HTMLModule/Text.php         |  8 +-------
 library/HTMLPurifier/HTMLModule/Tidy.php         |  2 +-
 library/HTMLPurifier/Injector/AutoParagraph.php  |  2 +-
 library/HTMLPurifier/Strategy/MakeWellFormed.php |  4 ++--
 tests/HTMLPurifier/ChildDef/CustomTest.php       | 10 ++++++++++
 tests/HTMLPurifier/ElementDefTest.php            | 12 ------------
 tests/HTMLPurifier/Strategy/CoreTest.php         |  4 ++--
 16 files changed, 36 insertions(+), 40 deletions(-)

diff --git a/NEWS b/NEWS
index f4cc74b1..7587a6ee 100644
--- a/NEWS
+++ b/NEWS
@@ -10,6 +10,8 @@ NEWS ( CHANGELOG and HISTORY )                                     HTMLPurifier
 ==========================
 
 2.0.1, unknown release date
+! Tag auto-closing now based on a ChildDef heuristic rather than a
+  manually set auto_close array; some behavior may change
 - Clean up special case code for <script> tags
 - Reorder includes for DefinitionCache decorators, fixes a possible
   missing class error
diff --git a/library/HTMLPurifier/ChildDef.php b/library/HTMLPurifier/ChildDef.php
index bed43cac..4ac5fead 100644
--- a/library/HTMLPurifier/ChildDef.php
+++ b/library/HTMLPurifier/ChildDef.php
@@ -36,6 +36,11 @@ class HTMLPurifier_ChildDef
      */
     var $allow_empty;
     
+    /**
+     * Lookup array of all elements that this definition could possibly allow
+     */
+    var $elements = array();
+    
     /**
      * Validates nodes according to definition and returns modification.
      * 
diff --git a/library/HTMLPurifier/ChildDef/Chameleon.php b/library/HTMLPurifier/ChildDef/Chameleon.php
index afe0299f..81dc1d43 100644
--- a/library/HTMLPurifier/ChildDef/Chameleon.php
+++ b/library/HTMLPurifier/ChildDef/Chameleon.php
@@ -35,6 +35,7 @@ class HTMLPurifier_ChildDef_Chameleon extends HTMLPurifier_ChildDef
     function HTMLPurifier_ChildDef_Chameleon($inline, $block) {
         $this->inline = new HTMLPurifier_ChildDef_Optional($inline);
         $this->block  = new HTMLPurifier_ChildDef_Optional($block);
+        $this->elements = $this->inline->elements;
     }
     
     function validateChildren($tokens_of_children, $config, &$context) {
diff --git a/library/HTMLPurifier/ChildDef/Custom.php b/library/HTMLPurifier/ChildDef/Custom.php
index 9f90b04b..bdf26d1c 100644
--- a/library/HTMLPurifier/ChildDef/Custom.php
+++ b/library/HTMLPurifier/ChildDef/Custom.php
@@ -44,6 +44,12 @@ class HTMLPurifier_ChildDef_Custom extends HTMLPurifier_ChildDef
         // COMPLICATED! AND MIGHT BE BUGGY! I HAVE NO CLUE WHAT I'M
         // DOING! Seriously: if there's problems, please report them.
         
+        // collect all elements into the $elements array
+        preg_match_all("/$el/", $reg, $matches);
+        foreach ($matches[0] as $match) {
+            $this->elements[$match] = true;
+        }
+        
         // setup all elements as parentheticals with leading commas
         $reg = preg_replace("/$el/", '(,\\0)', $reg);
         
diff --git a/library/HTMLPurifier/ChildDef/Required.php b/library/HTMLPurifier/ChildDef/Required.php
index 4d253398..c03c0ca2 100644
--- a/library/HTMLPurifier/ChildDef/Required.php
+++ b/library/HTMLPurifier/ChildDef/Required.php
@@ -25,7 +25,7 @@ class HTMLPurifier_ChildDef_Required extends HTMLPurifier_ChildDef
             $elements = array_flip($elements);
             foreach ($elements as $i => $x) {
                 $elements[$i] = true;
-                if (empty($i)) unset($elements[$i]);
+                if (empty($i)) unset($elements[$i]); // remove blank
             }
         }
         $this->elements = $elements;
diff --git a/library/HTMLPurifier/ChildDef/Table.php b/library/HTMLPurifier/ChildDef/Table.php
index 3534cdd0..09260f0c 100644
--- a/library/HTMLPurifier/ChildDef/Table.php
+++ b/library/HTMLPurifier/ChildDef/Table.php
@@ -9,6 +9,8 @@ class HTMLPurifier_ChildDef_Table extends HTMLPurifier_ChildDef
 {
     var $allow_empty = false;
     var $type = 'table';
+    var $elements = array('tr' => true, 'tbody' => true, 'thead' => true,
+        'tfoot' => true, 'caption' => true, 'colgroup' => true, 'col' => true);
     function HTMLPurifier_ChildDef_Table() {}
     function validateChildren($tokens_of_children, $config, &$context) {
         if (empty($tokens_of_children)) return false;
diff --git a/library/HTMLPurifier/ElementDef.php b/library/HTMLPurifier/ElementDef.php
index 4c90cce8..c99802a4 100644
--- a/library/HTMLPurifier/ElementDef.php
+++ b/library/HTMLPurifier/ElementDef.php
@@ -71,13 +71,6 @@ class HTMLPurifier_ElementDef
     
     
     
-    /**
-     * Lookup table of tags that close this tag. Used during parsing
-     * to make sure we don't attempt to nest unclosed tags.
-     * @public
-     */
-    var $auto_close = array();
-    
     /**
      * Does the element have a content model (#PCDATA | Inline)*? This
      * is important for chameleon ins and del processing in 
@@ -151,7 +144,6 @@ class HTMLPurifier_ElementDef
         }
         $this->_mergeAssocArray($this->attr_transform_pre, $def->attr_transform_pre);
         $this->_mergeAssocArray($this->attr_transform_post, $def->attr_transform_post);
-        $this->_mergeAssocArray($this->auto_close, $def->auto_close);
         $this->_mergeAssocArray($this->excludes, $def->excludes);
         
         if(!empty($def->content_model)) {
diff --git a/library/HTMLPurifier/HTMLModule/List.php b/library/HTMLPurifier/HTMLModule/List.php
index 894ac5aa..43a8f8ee 100644
--- a/library/HTMLPurifier/HTMLModule/List.php
+++ b/library/HTMLPurifier/HTMLModule/List.php
@@ -26,8 +26,7 @@ class HTMLPurifier_HTMLModule_List extends HTMLPurifier_HTMLModule
         $this->addElement('ul', true, 'List', 'Required: li', 'Common');
         $this->addElement('dl', true, 'List', 'Required: dt | dd', 'Common');
         
-        $li =& $this->addElement('li', true, false, 'Flow', 'Common');
-        $li->auto_close = array('li' => true);
+        $this->addElement('li', true, false, 'Flow', 'Common');
         
         $this->addElement('dd', true, false, 'Flow', 'Common');
         $this->addElement('dt', true, false, 'Inline', 'Common');
diff --git a/library/HTMLPurifier/HTMLModule/Tables.php b/library/HTMLPurifier/HTMLModule/Tables.php
index 043c8101..1e92db28 100644
--- a/library/HTMLPurifier/HTMLModule/Tables.php
+++ b/library/HTMLPurifier/HTMLModule/Tables.php
@@ -56,10 +56,7 @@ class HTMLPurifier_HTMLModule_Tables extends HTMLPurifier_HTMLModule
             $cell_align
         );
         $this->addElement('col',      true, false, 'Empty',         'Common', $cell_col);
-        $colgroup =& $this->addElement('colgroup', true, false, 'Optional: col', 'Common', $cell_col);
-        $colgroup->auto_close = $this->makeLookup(
-            'thead', 'tbody', 'tfoot', 'tr'
-        );
+        $this->addElement('colgroup', true, false, 'Optional: col', 'Common', $cell_col);
         
         $this->addElement('tbody', true, false, 'Required: tr', 'Common', $cell_align);
         $this->addElement('thead', true, false, 'Required: tr', 'Common', $cell_align);
diff --git a/library/HTMLPurifier/HTMLModule/Text.php b/library/HTMLPurifier/HTMLModule/Text.php
index c8abe1d7..3d9dde7e 100644
--- a/library/HTMLPurifier/HTMLModule/Text.php
+++ b/library/HTMLPurifier/HTMLModule/Text.php
@@ -55,13 +55,7 @@ class HTMLPurifier_HTMLModule_Text extends HTMLPurifier_HTMLModule
         $this->addElement('h6', true, 'Heading', 'Inline', 'Common');
         
         // Block Structural -----------------------------------------------
-        $p =& $this->addElement('p', true, 'Block', 'Inline', 'Common');
-        // this seems really ad hoc: implementing some general
-        // heuristics would probably be better
-        $p->auto_close = $this->makeLookup(
-            'address', 'blockquote', 'dd', 'dir', 'div', 'dl', 'dt',
-            'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'hr', 'ol', 'p', 'pre',
-            'table', 'ul' );
+        $this->addElement('p', true, 'Block', 'Inline', 'Common');
         $this->addElement('div', true, 'Block', 'Flow', 'Common');
         
     }
diff --git a/library/HTMLPurifier/HTMLModule/Tidy.php b/library/HTMLPurifier/HTMLModule/Tidy.php
index b81bf3ad..88d463bb 100644
--- a/library/HTMLPurifier/HTMLModule/Tidy.php
+++ b/library/HTMLPurifier/HTMLModule/Tidy.php
@@ -238,4 +238,4 @@ class HTMLPurifier_HTMLModule_Tidy extends HTMLPurifier_HTMLModule
     
 }
 
-?>
\ No newline at end of file
+?>
diff --git a/library/HTMLPurifier/Injector/AutoParagraph.php b/library/HTMLPurifier/Injector/AutoParagraph.php
index 5b72b3ce..8a60d59e 100644
--- a/library/HTMLPurifier/Injector/AutoParagraph.php
+++ b/library/HTMLPurifier/Injector/AutoParagraph.php
@@ -116,7 +116,7 @@ class HTMLPurifier_Injector_AutoParagraph extends HTMLPurifier_Injector
      */
     function _isInline($token, $config) {
         $definition = $config->getHTMLDefinition();
-        return !isset($definition->info['p']->auto_close[$token->name]);
+        return isset($definition->info['p']->child->elements[$token->name]);
     }
     
 }
diff --git a/library/HTMLPurifier/Strategy/MakeWellFormed.php b/library/HTMLPurifier/Strategy/MakeWellFormed.php
index e8e605df..2529f9b7 100644
--- a/library/HTMLPurifier/Strategy/MakeWellFormed.php
+++ b/library/HTMLPurifier/Strategy/MakeWellFormed.php
@@ -105,8 +105,8 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
                     // this can be replaced with a more general algorithm:
                     // if the token is not allowed by the parent, auto-close
                     // the parent
-                    if (isset($parent_info->auto_close[$token->name])) {
-                        // close th e parent, then append the token
+                    if (!isset($parent_info->child->elements[$token->name])) {
+                        // close the parent, then append the token
                         $result[] = new HTMLPurifier_Token_End($parent->name);
                         $result[] = $token;
                         $current_nesting[] = $token;
diff --git a/tests/HTMLPurifier/ChildDef/CustomTest.php b/tests/HTMLPurifier/ChildDef/CustomTest.php
index e4f00a62..e8253a84 100644
--- a/tests/HTMLPurifier/ChildDef/CustomTest.php
+++ b/tests/HTMLPurifier/ChildDef/CustomTest.php
@@ -10,6 +10,9 @@ class HTMLPurifier_ChildDef_CustomTest extends HTMLPurifier_ChildDefHarness
         
         $this->obj = new HTMLPurifier_ChildDef_Custom('(a,b?,c*,d+,(a,b)*)');
         
+        $this->assertEqual($this->obj->elements, array('a' => true,
+            'b' => true, 'c' => true, 'd' => true));
+        
         $this->assertResult('', false);
         $this->assertResult('<a /><a />', false);
         
@@ -21,6 +24,8 @@ class HTMLPurifier_ChildDef_CustomTest extends HTMLPurifier_ChildDefHarness
     
     function testNesting() {
         $this->obj = new HTMLPurifier_ChildDef_Custom('(a,b,(c|d))+');
+        $this->assertEqual($this->obj->elements, array('a' => true,
+            'b' => true, 'c' => true, 'd' => true));
         $this->assertResult('', false);
         $this->assertResult('<a /><b /><c /><a /><b /><d />');
         $this->assertResult('<a /><b /><c /><d />', false);
@@ -28,6 +33,8 @@ class HTMLPurifier_ChildDef_CustomTest extends HTMLPurifier_ChildDefHarness
     
     function testNestedEitherOr() {
         $this->obj = new HTMLPurifier_ChildDef_Custom('b,(a|(c|d))+');
+        $this->assertEqual($this->obj->elements, array('a' => true,
+            'b' => true, 'c' => true, 'd' => true));
         $this->assertResult('', false);
         $this->assertResult('<b /><a /><c /><d />');
         $this->assertResult('<b /><d /><a /><a />');
@@ -37,6 +44,7 @@ class HTMLPurifier_ChildDef_CustomTest extends HTMLPurifier_ChildDefHarness
     
     function testNestedQuantifier() {
         $this->obj = new HTMLPurifier_ChildDef_Custom('(b,c+)*');
+        $this->assertEqual($this->obj->elements, array('b' => true, 'c' => true));
         $this->assertResult('');
         $this->assertResult('<b /><c />');
         $this->assertResult('<b /><c /><c /><c />');
@@ -47,6 +55,7 @@ class HTMLPurifier_ChildDef_CustomTest extends HTMLPurifier_ChildDefHarness
     function testEitherOr() {
         
         $this->obj = new HTMLPurifier_ChildDef_Custom('a|b');
+        $this->assertEqual($this->obj->elements, array('a' => true, 'b' => true));
         $this->assertResult('', false);
         $this->assertResult('<a />');
         $this->assertResult('<b />');
@@ -57,6 +66,7 @@ class HTMLPurifier_ChildDef_CustomTest extends HTMLPurifier_ChildDefHarness
     function testCommafication() {
         
         $this->obj = new HTMLPurifier_ChildDef_Custom('a,b');
+        $this->assertEqual($this->obj->elements, array('a' => true, 'b' => true));
         $this->assertResult('<a /><b />');
         $this->assertResult('<ab />', false);
         
diff --git a/tests/HTMLPurifier/ElementDefTest.php b/tests/HTMLPurifier/ElementDefTest.php
index 23ef6e17..527b4acc 100644
--- a/tests/HTMLPurifier/ElementDefTest.php
+++ b/tests/HTMLPurifier/ElementDefTest.php
@@ -33,10 +33,6 @@ class HTMLPurifier_ElementDefTest extends UnitTestCase
         $def1->child = $overloaded_old;
         $def1->content_model = 'old';
         $def1->content_model_type = $overloaded_old;
-        $def1->auto_close = array(
-            'old' => true,
-            'removed-old' => true
-        );
         $def1->descendants_are_inline = false;
         $def1->excludes = array(
             'old' => true,
@@ -60,10 +56,6 @@ class HTMLPurifier_ElementDefTest extends UnitTestCase
         $def2->child = $new;
         $def2->content_model = 'new';
         $def2->content_model_type = $overloaded_new;
-        $def2->auto_close = array(
-            'new' => true,
-            'removed-old' => false
-        );
         $def2->descendants_are_inline = true;
         $def2->excludes = array(
             'new' => true,
@@ -90,10 +82,6 @@ class HTMLPurifier_ElementDefTest extends UnitTestCase
         $this->assertIdentical($def1->child, $new);
         $this->assertIdentical($def1->content_model, 'old | new');
         $this->assertIdentical($def1->content_model_type, $overloaded_new);
-        $this->assertIdentical($def1->auto_close, array(
-            'old' => true,
-            'new' => true
-        ));
         $this->assertIdentical($def1->descendants_are_inline, true);
         $this->assertIdentical($def1->excludes, array(
             'old' => true,
diff --git a/tests/HTMLPurifier/Strategy/CoreTest.php b/tests/HTMLPurifier/Strategy/CoreTest.php
index f2be4f50..30d8869b 100644
--- a/tests/HTMLPurifier/Strategy/CoreTest.php
+++ b/tests/HTMLPurifier/Strategy/CoreTest.php
@@ -20,7 +20,7 @@ class HTMLPurifier_Strategy_CoreTest extends HTMLPurifier_StrategyHarness
         );
         $this->assertResult(
             '<b><div>Fix nesting.</div></b>',
-            '<b>Fix nesting.</b>'
+            '<b></b><div>Fix nesting.</div>'
         );
         $this->assertResult(
             '<asdf>Foreign element removal.</asdf>',
@@ -28,7 +28,7 @@ class HTMLPurifier_Strategy_CoreTest extends HTMLPurifier_StrategyHarness
         );
         $this->assertResult(
             '<foo><b><div>All three.</div></b>',
-            '<b>All three.</b>'
+            '<b></b><div>All three.</div>'
         );
         
     }