lib/htmlpurifier/HTMLPurifier/Strategy/MakeWellFormed.php

   1 <?php
   2
   3 require_once 'HTMLPurifier/Strategy.php';
   4 require_once 'HTMLPurifier/HTMLDefinition.php';
   5 require_once 'HTMLPurifier/Generator.php';
   6
   7 /**
   8  * Takes tokens makes them well-formed (balance end tags, etc.)
   9  */
  10 class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
  11 {
  12
  13     function execute($tokens, $config, &$context) {
  14         $definition = $config->getHTMLDefinition();
  15         $generator = new HTMLPurifier_Generator();
  16         $result = array();
  17         $current_nesting = array();
  18         $escape_invalid_tags = $config->get('Core', 'EscapeInvalidTags');
  19         foreach ($tokens as $token) {
  20             if (empty( $token->is_tag )) {
  21                 $result[] = $token;
  22                 continue;
  23             }
  24
  25             // DEFINITION CALL
  26             $info = $definition->info[$token->name]->child;
  27
  28             // test if it claims to be a start tag but is empty
  29             if ($info->type == 'empty' &&
  30                 $token->type == 'start' ) {
  31
  32                 $result[] = new HTMLPurifier_Token_Empty($token->name,
  33                                                          $token->attr);
  34                 continue;
  35             }
  36
  37             // test if it claims to be empty but really is a start tag
  38             if ($info->type != 'empty' &&
  39                 $token->type == 'empty' ) {
  40
  41                 $result[] = new HTMLPurifier_Token_Start($token->name,
  42                                                          $token->attr);
  43                 $result[] = new HTMLPurifier_Token_End($token->name);
  44
  45                 continue;
  46             }
  47
  48             // automatically insert empty tags
  49             if ($token->type == 'empty') {
  50                 $result[] = $token;
  51                 continue;
  52             }
  53
  54             // we give start tags precedence, so automatically accept unless...
  55             // it's one of those special cases
  56             if ($token->type == 'start') {
  57
  58                 // if there's a parent, check for special case
  59                 if (!empty($current_nesting)) {
  60
  61                     $parent = array_pop($current_nesting);
  62                     $parent_name = $parent->name;
  63                     $parent_info = $definition->info[$parent_name];
  64
  65                     if (isset($parent_info->auto_close[$token->name])) {
  66                         $result[] = new HTMLPurifier_Token_End($parent_name);
  67                         $result[] = $token;
  68                         $current_nesting[] = $token;
  69                         continue;
  70                     }
  71
  72                     $current_nesting[] = $parent; // undo the pop
  73                 }
  74
  75                 $result[] = $token;
  76                 $current_nesting[] = $token;
  77                 continue;
  78             }
  79
  80             // sanity check
  81             if ($token->type != 'end') continue;
  82
  83             // okay, we're dealing with a closing tag
  84
  85             // make sure that we have something open
  86             if (empty($current_nesting)) {
  87                 if ($escape_invalid_tags) {
  88                     $result[] = new HTMLPurifier_Token_Text(
  89                         $generator->generateFromToken($token, $config, $context)
  90                     );
  91                 }
  92                 continue;
  93             }
  94
  95             // first, check for the simplest case: everything closes neatly
  96
  97             // current_nesting is modified
  98             $current_parent = array_pop($current_nesting);
  99             if ($current_parent->name == $token->name) {
 100                 $result[] = $token;
 101                 continue;
 102             }
 103
 104             // undo the array_pop
 105             $current_nesting[] = $current_parent;
 106
 107             // okay, so we're trying to close the wrong tag
 108
 109             // scroll back the entire nest, trying to find our tag
 110             // feature could be to specify how far you'd like to go
 111             $size = count($current_nesting);
 112             // -2 because -1 is the last element, but we already checked that
 113             $skipped_tags = false;
 114             for ($i = $size - 2; $i >= 0; $i--) {
 115                 if ($current_nesting[$i]->name == $token->name) {
 116                     // current nesting is modified
 117                     $skipped_tags = array_splice($current_nesting, $i);
 118                     break;
 119                 }
 120             }
 121
 122             // we still didn't find the tag, so translate to text
 123             if ($skipped_tags === false) {
 124                 if ($escape_invalid_tags) {
 125                     $result[] = new HTMLPurifier_Token_Text(
 126                         $generator->generateFromToken($token, $config, $context)
 127                     );
 128                 }
 129                 continue;
 130             }
 131
 132             // okay, we found it, close all the skipped tags
 133             // note that skipped tags contains the element we need closed
 134             $size = count($skipped_tags);
 135             for ($i = $size - 1; $i >= 0; $i--) {
 136                 $result[] = new HTMLPurifier_Token_End($skipped_tags[$i]->name);
 137             }
 138
 139             // done!
 140
 141         }
 142
 143         // we're at the end now, fix all still unclosed tags
 144
 145         if (!empty($current_nesting)) {
 146             $size = count($current_nesting);
 147             for ($i = $size - 1; $i >= 0; $i--) {
 148                 $result[] =
 149                     new HTMLPurifier_Token_End($current_nesting[$i]->name);
 150             }
 151         }
 152
 153         return $result;
 154     }
 155
 156 }
 157
 158 ?>