maintenance/PH5P.patch

   1 --- C:\Users\Edward\Webs\htmlpurifier\maintenance\PH5P.php      2008-07-07 09:12:12.000000000 -0400
   2 +++ C:\Users\Edward\Webs\htmlpurifier\maintenance/PH5P.new.php  2008-12-06 02:29:34.988800000 -0500
   3 @@ -65,7 +65,7 @@
   4
   5      public function __construct($data) {
   6          $data = str_replace("\r\n", "\n", $data);
   7 -        $date = str_replace("\r", null, $data);
   8 +        $data = str_replace("\r", null, $data);
   9
  10          $this->data = $data;
  11          $this->char = -1;
  12 @@ -211,7 +211,10 @@
  13          // If nothing is returned, emit a U+0026 AMPERSAND character token.
  14          // Otherwise, emit the character token that was returned.
  15          $char = (!$entity) ? '&' : $entity;
  16 -        $this->emitToken($char);
  17 +        $this->emitToken(array(
  18 +            'type' => self::CHARACTR,
  19 +            'data' => $char
  20 +        ));
  21
  22          // Finally, switch to the data state.
  23          $this->state = 'data';
  24 @@ -708,7 +711,7 @@
  25          } elseif($char === '&') {
  26              /* U+0026 AMPERSAND (&)
  27              Switch to the entity in attribute value state. */
  28 -            $this->entityInAttributeValueState('non');
  29 +            $this->entityInAttributeValueState();
  30
  31          } elseif($char === '>') {
  32              /* U+003E GREATER-THAN SIGN (>)
  33 @@ -738,7 +741,8 @@
  34              ? '&'
  35              : $entity;
  36
  37 -        $this->emitToken($char);
  38 +        $last = count($this->token['attr']) - 1;
  39 +        $this->token['attr'][$last]['value'] .= $char;
  40      }
  41
  42      private function bogusCommentState() {
  43 @@ -1066,6 +1070,11 @@
  44                      $this->char++;
  45
  46                      if(in_array($id, $this->entities)) {
  47 +                        if ($e_name[$c-1] !== ';') {
  48 +                            if ($c < $len && $e_name[$c] == ';') {
  49 +                                $this->char++; // consume extra semicolon
  50 +                            }
  51 +                        }
  52                          $entity = $id;
  53                          break;
  54                      }
  55 @@ -2084,7 +2093,7 @@
  56                      /* Reconstruct the active formatting elements, if any. */
  57                      $this->reconstructActiveFormattingElements();
  58
  59 -                    $this->insertElement($token);
  60 +                    $this->insertElement($token, true, true);
  61                  break;
  62              }
  63              break;
  64 @@ -3465,7 +3474,18 @@
  65          }
  66      }
  67
  68 -    private function insertElement($token, $append = true) {
  69 +    private function insertElement($token, $append = true, $check = false) {
  70 +        // Proprietary workaround for libxml2's limitations with tag names
  71 +        if ($check) {
  72 +            // Slightly modified HTML5 tag-name modification,
  73 +            // removing anything that's not an ASCII letter, digit, or hyphen
  74 +            $token['name'] = preg_replace('/[^a-z0-9-]/i', '', $token['name']);
  75 +            // Remove leading hyphens and numbers
  76 +            $token['name'] = ltrim($token['name'], '-0..9');
  77 +            // In theory, this should ever be needed, but just in case
  78 +            if ($token['name'] === '') $token['name'] = 'span'; // arbitrary generic choice
  79 +        }
  80 +
  81          $el = $this->dom->createElement($token['name']);
  82
  83          foreach($token['attr'] as $attr) {
  84 @@ -3659,7 +3679,7 @@
  85          }
  86      }
  87
  88 -    private function generateImpliedEndTags(array $exclude = array()) {
  89 +    private function generateImpliedEndTags($exclude = array()) {
  90          /* When the steps below require the UA to generate implied end tags,
  91          then, if the current node is a dd element, a dt element, an li element,
  92          a p element, a td element, a th  element, or a tr element, the UA must
  93 @@ -3673,7 +3693,8 @@
  94          }
  95      }
  96
  97 -    private function getElementCategory($name) {
  98 +    private function getElementCategory($node) {
  99 +        $name = $node->tagName;
 100          if(in_array($name, $this->special))
 101              return self::SPECIAL;
 102