contents
        $this->keepWhitespace++;
      }
    }

    if ($this->html[0] == '<') {
      $token = substr($this->html, 0, 9);
      if (substr($token, 0, 2) == 'html, '>');
        $this->setNode('pi', $pos + 1);
        return true;
      }
      if (substr($token, 0, 4) == '');
        if ($pos === false) {
          # could not find a closing -->, use next gt instead
          # this is firefox' behaviour
          $pos = strpos($this->html, '>') + 1;
        } else {
          $pos += 3;
        }
        $this->setNode('comment', $pos);

        $skipWhitespace = true;
        return true;
      }
      if ($token == 'setNode('doctype', strpos($this->html, '>')+1);

        $skipWhitespace = true;
        return true;
      }
      if ($token == 'html = substr($this->html, 9);

        $this->setNode('text', strpos($this->html, ']]>')+3);

        # remove trailing ]]> and trim
        $this->node = substr($this->node, 0, -3);
        $this->handleWhitespaces();

        $skipWhitespace = true;
        return true;
      }
      if ($this->parseTag()) {
        # seems to be a tag
        # handle whitespaces
        if ($this->isBlockElement) {
          $skipWhitespace = true;
        } else {
          $skipWhitespace = false;
        }
        return true;
      }
    }
    if ($this->keepWhitespace) {
      $skipWhitespace = false;
    }
    # when we get here it seems to be a text node
    $pos = strpos($this->html, '<');
    if ($pos === false) {
      $pos = strlen($this->html);
    }
    $this->setNode('text', $pos);
    $this->handleWhitespaces();
    if ($skipWhitespace && $this->node == ' ') {
      return $this->nextNode();
    }
    $skipWhitespace = false;
    return true;
  }
  /**
   * parse tag, set tag name and attributes, see if it's a closing tag and so forth...
   *
   * @param void
   * @return bool
   */
  function parseTag() {
    static $a_ord, $z_ord, $special_ords;
    if (!isset($a_ord)) {
      $a_ord = ord('a');
      $z_ord = ord('z');
      $special_ords = array(
        ord(':'), // for xml:lang
        ord('-'), // for http-equiv
      );
    }

    $tagName = '';

    $pos = 1;
    $isStartTag = $this->html[$pos] != '/';
    if (!$isStartTag) {
      $pos++;
    }
    # get tagName
    while (isset($this->html[$pos])) {
      $pos_ord = ord(strtolower($this->html[$pos]));
      if (($pos_ord >= $a_ord && $pos_ord <= $z_ord) || (!empty($tagName) && is_numeric($this->html[$pos]))) {
        $tagName .= $this->html[$pos];
        $pos++;
      } else {
        $pos--;
        break;
      }
    }

    $tagName = strtolower($tagName);
    if (empty($tagName) || !isset($this->blockElements[$tagName])) {
      # something went wrong => invalid tag
      $this->invalidTag();
      return false;
    }
    if ($this->noTagsInCode && end($this->openTags) == 'code' && !($tagName == 'code' && !$isStartTag)) {
      # we supress all HTML tags inside code tags
      $this->invalidTag();
      return false;
    }

    # get tag attributes
    /** TODO: in html 4 attributes do not need to be quoted **/
    $isEmptyTag = false;
    $attributes = array();
    $currAttrib = '';
    while (isset($this->html[$pos+1])) {
      $pos++;
      # close tag
      if ($this->html[$pos] == '>' || $this->html[$pos].$this->html[$pos+1] == '/>') {
        if ($this->html[$pos] == '/') {
          $isEmptyTag = true;
          $pos++;
        }
        break;
      }

      $pos_ord = ord(strtolower($this->html[$pos]));
      if ( ($pos_ord >= $a_ord && $pos_ord <= $z_ord) || in_array($pos_ord, $special_ords)) {
        # attribute name
        $currAttrib .= $this->html[$pos];
      } elseif (in_array($this->html[$pos], array(' ', "\t", "\n"))) {
        # drop whitespace
      } elseif (in_array($this->html[$pos].$this->html[$pos+1], array('="', "='"))) {
        # get attribute value
        $pos++;
        $await = $this->html[$pos]; # single or double quote
        $pos++;
        $value = '';
        while (isset($this->html[$pos]) && $this->html[$pos] != $await) {
          $value .= $this->html[$pos];
          $pos++;
        }
        $attributes[$currAttrib] = $value;
        $currAttrib = '';
      } else {
        $this->invalidTag();
        return false;
      }
    }
    if ($this->html[$pos] != '>') {
      $this->invalidTag();
      return false;
    }

    if (!empty($currAttrib)) {
      # html 4 allows something like  instead of 
      $attributes[$currAttrib] = $currAttrib;
    }
    if (!$isStartTag) {
      if (!empty($attributes) || $tagName != end($this->openTags)) {
        # end tags must not contain any attributes
        # or maybe we did not expect a different tag to be closed
        $this->invalidTag();
        return false;
      }
      array_pop($this->openTags);
      if (in_array($tagName, $this->preformattedTags)) {
        $this->keepWhitespace--;
      }
    }
    $pos++;
    $this->node = substr($this->html, 0, $pos);
    $this->html = substr($this->html, $pos);
    $this->tagName = $tagName;
    $this->tagAttributes = $attributes;
    $this->isStartTag = $isStartTag;
    $this->isEmptyTag = $isEmptyTag || in_array($tagName, $this->emptyTags);
    if ($this->isEmptyTag) {
      # might be not well formed
      $this->node = preg_replace('# */? *>$#', ' />', $this->node);
    }
    $this->nodeType = 'tag';
    $this->isBlockElement = $this->blockElements[$tagName];
    return true;
  }
  /**
   * handle invalid tags
   *
   * @param void
   * @return void
   */
  function invalidTag() {
    $this->html = substr_replace($this->html, '<', 0, 1);
  }
  /**
   * update all vars and make $this->html shorter
   *
   * @param string $type see description for $this->nodeType
   * @param int $pos to which position shall we cut?
   * @return void
   */
  function setNode($type, $pos) {
    if ($this->nodeType == 'tag') {
      # set tag specific vars to null
      # $type == tag should not be called here
      # see this::parseTag() for more
      $this->tagName = null;
      $this->tagAttributes = null;
      $this->isStartTag = null;
      $this->isEmptyTag = null;
      $this->isBlockElement = null;

    }
    $this->nodeType = $type;
    $this->node = substr($this->html, 0, $pos);
    $this->html = substr($this->html, $pos);
  }
  /**
   * check if $this->html begins with $str
   *
   * @param string $str
   * @return bool
   */
  function match($str) {
    return substr($this->html, 0, strlen($str)) == $str;
  }
  /**
   * truncate whitespaces
   *
   * @param void
   * @return void
   */
  function handleWhitespaces() {
    if ($this->keepWhitespace) {
      # 
 or  before...
      return;
    }
    # truncate multiple whitespaces to a single one
    $this->node = preg_replace('#\s+#s', ' ', $this->node);
  }
  /**
   * normalize self::node
   *
   * @param void
   * @return void
   */
  function normalizeNode() {
    $this->node = '<';
    if (!$this->isStartTag) {
      $this->node .= '/'.$this->tagName.'>';
      return;
    }
    $this->node .= $this->tagName;
    foreach ($this->tagAttributes as $name => $value) {
      $this->node .= ' '.$name.'="'.str_replace('"', '"', $value).'"';
    }
    if ($this->isEmptyTag) {
      $this->node .= ' /';
    }
    $this->node .= '>';
  }
}

/**
 * indent a HTML string properly
 *
 * @param string $html
 * @param string $indent optional
 * @return string
 */
function indentHTML($html, $indent = "  ", $noTagsInCode = false) {
  $parser = new parseHTML;
  $parser->noTagsInCode = $noTagsInCode;
  $parser->html = $html;
  $html = '';
  $last = true; # last tag was block elem
  $indent_a = array();
  while($parser->nextNode()) {
    if ($parser->nodeType == 'tag') {
      $parser->normalizeNode();
    }
    if ($parser->nodeType == 'tag' && $parser->isBlockElement) {
      $isPreOrCode = in_array($parser->tagName, array('code', 'pre'));
      if (!$parser->keepWhitespace && !$last && !$isPreOrCode) {
        $html = rtrim($html)."\n";
      }
      if ($parser->isStartTag) {
        $html .= implode($indent_a);
        if (!$parser->isEmptyTag) {
          array_push($indent_a, $indent);
        }
      } else {
        array_pop($indent_a);
        if (!$isPreOrCode) {
          $html .= implode($indent_a);
        }
      }
      $html .= $parser->node;
      if (!$parser->keepWhitespace && !($isPreOrCode && $parser->isStartTag)) {
        $html .= "\n";
      }
      $last = true;
    } else {
      if ($parser->nodeType == 'tag' && $parser->tagName == 'br') {
        $html .= $parser->node."\n";
        $last = true;
        continue;
      } elseif ($last && !$parser->keepWhitespace) {
        $html .= implode($indent_a);
        $parser->node = ltrim($parser->node);
      }
      $html .= $parser->node;

      if (in_array($parser->nodeType, array('comment', 'pi', 'doctype'))) {
        $html .= "\n";
      } else {
        $last = false;
      }
    }
  }
  return $html;
}
/*
# testcase / example
error_reporting(E_ALL);

$html = '
Simple block on one line:

foo

And nested without indentation:




foo



bar


And with attributes:


    

    


This was broken in 1.0.2b7:



foo

';
#$html = 'asdf';
echo indentHTML($html);
die();
*/