res/simple_html_dom.php

<?php
/**
 * Website: http://sourceforge.net/projects/simplehtmldom/
 * Additional projects that may be used: http://sourceforge.net/projects/debugobject/
 * Acknowledge: Jose Solorzano (https://sourceforge.net/projects/php-html/)
 * Contributions by:
 *	 Yousuke Kumakura (Attribute filters)
 *	 Vadim Voituk (Negative indexes supports of "find" method)
 *	 Antcs (Constructor with automatically load contents either text or file/url)
 *
 * all affected sections have comments starting with "PaperG"
 *
 * Paperg - Added case insensitive testing of the value of the selector.
 * Paperg - Added tag_start for the starting index of tags - NOTE: This works but not accurately.
 *  This tag_start gets counted AFTER \r\n have been crushed out, and after the remove_noice calls so it will not reflect the REAL position of the tag in the source,
 *  it will almost always be smaller by some amount.
 *  We use this to determine how far into the file the tag in question is.  This "percentage will never be accurate as the $dom->size is the "real" number of bytes the dom was created from.
 *  but for most purposes, it's a really good estimation.
 * Paperg - Added the forceTagsClosed to the dom constructor.  Forcing tags closed is great for malformed html, but it CAN lead to parsing errors.
 * Allow the user to tell us how much they trust the html.
 * Paperg add the text and plaintext to the selectors for the find syntax.  plaintext implies text in the innertext of a node.  text implies that the tag is a text node.
 * This allows for us to find tags based on the text they contain.
 * Create find_ancestor_tag to see if a tag is - at any level - inside of another specific tag.
 * Paperg: added parse_charset so that we know about the character set of the source document.
 *  NOTE:  If the user's system has a routine called get_last_retrieve_url_contents_content_type availalbe, we will assume it's returning the content-type header from the
 *  last transfer or curl_exec, and we will parse that and use it in preference to any other method of charset detection.
 *
 * Found infinite loop in the case of broken html in restore_noise.  Rewrote to protect from that.
 * PaperG (John Schlick) Added get_display_size for "IMG" tags.
 *
 * Licensed under The MIT License
 * Redistributions of files must retain the above copyright notice.
 *
 * @author S.C. Chen <me578022@gmail.com>
 * @author John Schlick
 * @author Rus Carroll
 * @version Rev. 1.7 (214)
 * @package PlaceLocalInclude
 * @subpackage ums_simple_html_dom
 */

/**
 * All of the Defines for the classes below.
 * @author S.C. Chen <me578022@gmail.com>
 */
define('HDOM_TYPE_ELEMENT', 1);
define('HDOM_TYPE_COMMENT', 2);
define('HDOM_TYPE_TEXT',	3);
define('HDOM_TYPE_ENDTAG',  4);
define('HDOM_TYPE_ROOT',	5);
define('HDOM_TYPE_UNKNOWN', 6);
define('HDOM_QUOTE_DOUBLE', 0);
define('HDOM_QUOTE_SINGLE', 1);
define('HDOM_QUOTE_NO',	 3);
define('HDOM_INFO_BEGIN',   0);
define('HDOM_INFO_END',	 1);
define('HDOM_INFO_QUOTE',   2);
define('HDOM_INFO_SPACE',   3);
define('HDOM_INFO_TEXT',	4);
define('HDOM_INFO_INNER',   5);
define('HDOM_INFO_OUTER',   6);
define('HDOM_INFO_ENDSPACE',7);
define('DEFAULT_TARGET_CHARSET', 'UTF-8');
define('DEFAULT_BR_TEXT', "\r\n");
define('DEFAULT_SPAN_TEXT', " ");
define('MAX_FILE_SIZE', 10485760);

/** Contents between curly braces "{" and "}" are interpreted as text */
define('HDOM_SMARTY_AS_TEXT', 1);

// helper functions
// -----------------------------------------------------------------------------
// get html dom from file
// $maxlen is defined in the code as PHP_STREAM_COPY_ALL which is defined as -1.
function ums_file_get_html($url, $use_include_path = false, $context=null, $offset = 0, $maxLen=-1, $lowercase = true, $forceTagsClosed=true, $target_charset = DEFAULT_TARGET_CHARSET, $stripRN=true, $defaultBRText=DEFAULT_BR_TEXT, $defaultSpanText=DEFAULT_SPAN_TEXT)
{
	// Ensure maximum length is greater than zero
	if($maxLen <= 0) { $maxLen = MAX_FILE_SIZE; }

	// We DO force the tags to be terminated.
	$dom = new ums_simple_html_dom(null, $lowercase, $forceTagsClosed, $target_charset, $stripRN, $defaultBRText, $defaultSpanText);
	// For sourceforge users: uncomment the next line and comment the retrieve_url_contents line 2 lines down if it is not already done.
	global $wp_filesystem;
    if ( ! is_a( $wp_filesystem, 'WP_Filesystem_Base') ){
        include_once(ABSPATH . 'wp-admin/includes/file.php');$creds = request_filesystem_credentials( site_url() );
        wp_filesystem($creds);
    }
    $contents = $wp_filesystem->get_contents($url);
	// Paperg - use our own mechanism for getting the contents as we want to control the timeout.
	
	if (empty($contents) || strlen($contents) > $maxLen)
	{
		return false;
	}
	// The second parameter can force the selectors to all be lowercase.
	$dom->load($contents, $lowercase, $stripRN);
	return $dom;
}

// get html dom from string
function ums_str_get_html($str, $lowercase=true, $forceTagsClosed=true, $target_charset = DEFAULT_TARGET_CHARSET, $stripRN=true, $defaultBRText=DEFAULT_BR_TEXT, $defaultSpanText=DEFAULT_SPAN_TEXT)
{
	$dom = new ums_simple_html_dom(null, $lowercase, $forceTagsClosed, $target_charset, $stripRN, $defaultBRText, $defaultSpanText);
	if (empty($str) || strlen($str) > MAX_FILE_SIZE)
	{
		$dom->clear();
		return false;
	}
	$dom->load($str, $lowercase, $stripRN);
	return $dom;
}

// dump html dom tree
function ums_dump_html_tree($node, $show_attr=true, $deep=0)
{
	$node->dump($node);
}


/**
 * simple html dom node
 * PaperG - added ability for "find" routine to lowercase the value of the selector.
 * PaperG - added $tag_start to track the start position of the tag in the total byte index
 *
 * @package PlaceLocalInclude
 */
class ums_simple_html_dom_node
{
	/**
	 * Node type
	 *
	 * Default is {@see HDOM_TYPE_TEXT}
	 *
	 * @var int
	 */
	public $nodetype = HDOM_TYPE_TEXT;

	/**
	 * Tag name
	 *
	 * Default is 'text'
	 *
	 * @var string
	 */
	public $tag = 'text';

	/**
	 * List of attributes
	 *
	 * @var array
	 */
	public $attr = array();

	/**
	 * List of child node objects
	 *
	 * @var array
	 */
	public $children = array();
	public $nodes = array();

	/**
	 * The parent node object
	 *
	 * @var object|null
	 */
	public $parent = null;

	// The "info" array - see HDOM_INFO_... for what each element contains.
	public $_ = array();

	/**
	 * Start position of the tag in the document
	 *
	 * @var int
	 */
	public $tag_start = 0;

	/**
	 * The DOM object
	 *
	 * @var object|null
	 */
	private $dom = null;

	/**
	 * Construct new node object
	 *
	 * Adds itself to the list of DOM Nodes {@see ums_simple_html_dom::$nodes}
	 */
	function __construct($dom)
	{
		$this->dom = $dom;
		$dom->nodes[] = $this;
	}

	function __destruct()
	{
		$this->clear();
	}

	function __toString()
	{
		return $this->outertext();
	}

	// clean up memory due to php5 circular references memory leak...
	function clear()
	{
		$this->dom = null;
		$this->nodes = null;
		$this->parent = null;
		$this->children = null;
	}

	// dump node's tree
	function dump($show_attr=true, $deep=0)
	{
		$lead = str_repeat('	', $deep);

		echo $lead.$this->tag;
		if ($show_attr && count($this->attr)>0)
		{
			echo '(';
			foreach ($this->attr as $k=>$v)
				echo "[$k]=>\"".$this->$k.'", ';
			echo ')';
		}
		echo "\n";

		if ($this->nodes)
		{
			foreach ($this->nodes as $c)
			{
				$c->dump($show_attr, $deep+1);
			}
		}
	}


	// Debugging function to dump a single dom node with a bunch of information about it.
	function dump_node($echo=true)
	{

		$string = $this->tag;
		if (count($this->attr)>0)
		{
			$string .= '(';
			foreach ($this->attr as $k=>$v)
			{
				$string .= "[$k]=>\"".$this->$k.'", ';
			}
			$string .= ')';
		}
		if (count($this->_)>0)
		{
			$string .= ' $_ (';
			foreach ($this->_ as $k=>$v)
			{
				if (is_array($v))
				{
					$string .= "[$k]=>(";
					foreach ($v as $k2=>$v2)
					{
						$string .= "[$k2]=>\"".$v2.'", ';
					}
					$string .= ")";
				} else {
					$string .= "[$k]=>\"".$v.'", ';
				}
			}
			$string .= ")";
		}

		if (isset($this->text))
		{
			$string .= " text: (" . $this->text . ")";
		}

		$string .= " HDOM_INNER_INFO: '";
		if (isset($node->_[HDOM_INFO_INNER]))
		{
			$string .= $node->_[HDOM_INFO_INNER] . "'";
		}
		else
		{
			$string .= ' NULL ';
		}

		$string .= " children: " . count($this->children);
		$string .= " nodes: " . count($this->nodes);
		$string .= " tag_start: " . $this->tag_start;
		$string .= "\n";

		if ($echo)
		{
			echo $string;
			return;
		}
		else
		{
			return $string;
		}
	}

	/**
	 * Return or set parent node
	 *
	 * @param object|null $parent (optional) The parent node, `null` to return
	 * the current parent node.
	 * @return object|null The parent node
	 */
	function parent($parent=null)
	{
		// I am SURE that this doesn't work properly.
		// It fails to unset the current node from it's current parents nodes or children list first.
		if ($parent !== null)
		{
			$this->parent = $parent;
			$this->parent->nodes[] = $this;
			$this->parent->children[] = $this;
		}

		return $this->parent;
	}

	/**
	 * @return bool True if the node has at least one child node
	 */
	function has_child()
	{
		return !empty($this->children);
	}

	/**
	 * Get child node at specified index
	 *
	 * @param int $idx The index of the child node to return, `-1` to return all
	 * child nodes.
	 * @return object|array|null The child node at the specified index, all child
	 * nodes or null if the index is invalid.
	 */
	function children($idx=-1)
	{
		if ($idx===-1)
		{
			return $this->children;
		}
		if (isset($this->children[$idx]))
		{
			return $this->children[$idx];
		}
		return null;
	}

	/**
	 * Get first child node
	 *
	 * @return object|null The first child node or null if the current node has
	 * no child nodes.
	 *
	 * @todo Use `empty()` instead of `count()` to improve performance on large
	 * arrays.
	 */
	function first_child()
	{
		if (count($this->children)>0)
		{
			return $this->children[0];
		}
		return null;
	}

	/**
	 * Get last child node
	 *
	 * @return object|null The last child node or null if the current node has
	 * no child nodes.
	 *
	 * @todo Use `end()` to slightly improve performance on large arrays.
	 */
	function last_child()
	{
		if (($count=count($this->children))>0)
		{
			return $this->children[$count-1];
		}
		return null;
	}

	/**
	 * Get next sibling node
	 *
	 * @return object|null The sibling node or null if the current node has no
	 * sibling nodes.
	 */
	function next_sibling()
	{
		if ($this->parent===null)
		{
			return null;
		}

		$idx = 0;
		$count = count($this->parent->children);
		while ($idx<$count && $this!==$this->parent->children[$idx])
		{
			++$idx;
		}
		if (++$idx>=$count)
		{
			return null;
		}
		return $this->parent->children[$idx];
	}

	/**
	 * Get previous sibling node
	 *
	 * @return object|null The sibling node or null if the current node has no
	 * sibling nodes.
	 */
	function prev_sibling()
	{
		if ($this->parent===null) return null;
		$idx = 0;
		$count = count($this->parent->children);
		while ($idx<$count && $this!==$this->parent->children[$idx])
			++$idx;
		if (--$idx<0) return null;
		return $this->parent->children[$idx];
	}

	/**
	 * Traverse ancestors to the first matching tag.
	 *
	 * @param string $tag Tag to find
	 * @return object|null First matching node in the DOM tree or null if no
	 * match was found.
	 *
	 * @todo Null is returned implicitly by calling ->parent on the root node.
	 * This behaviour could change at any time, rendering this function invalid.
	 */
	function find_ancestor_tag($tag)
	{
		global $debug_object;
		if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

		// Start by including ourselves in the comparison.
		$returnDom = $this;

		while (!is_null($returnDom))
		{
			if (is_object($debug_object)) { $debug_object->debug_log(2, "Current tag is: " . $returnDom->tag); }

			if ($returnDom->tag == $tag)
			{
				break;
			}
			$returnDom = $returnDom->parent;
		}
		return $returnDom;
	}

	/**
	 * Get node's inner text (everything inside the opening and closing tags)
	 *
	 * @return string
	 */
	function innertext()
	{
		if (isset($this->_[HDOM_INFO_INNER])) return $this->_[HDOM_INFO_INNER];
		if (isset($this->_[HDOM_INFO_TEXT])) return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);

		$ret = '';
		foreach ($this->nodes as $n)
			$ret .= $n->outertext();
		return $ret;
	}

	/**
	 * Get node's outer text (everything including the opening and closing tags)
	 *
	 * @return string
	 */
	function outertext()
	{
		global $debug_object;
		if (is_object($debug_object))
		{
			$text = '';
			if ($this->tag == 'text')
			{
				if (!empty($this->text))
				{
					$text = " with text: " . $this->text;
				}
			}
			$debug_object->debug_log(1, 'Innertext of tag: ' . $this->tag . $text);
		}

		if ($this->tag==='root') return $this->innertext();

		// trigger callback
		if ($this->dom && $this->dom->callback!==null)
		{
			call_user_func_array($this->dom->callback, array($this));
		}

		if (isset($this->_[HDOM_INFO_OUTER])) return $this->_[HDOM_INFO_OUTER];
		if (isset($this->_[HDOM_INFO_TEXT])) return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);

		// render begin tag
		if ($this->dom && $this->dom->nodes[$this->_[HDOM_INFO_BEGIN]])
		{
			$ret = $this->dom->nodes[$this->_[HDOM_INFO_BEGIN]]->makeup();
		} else {
			$ret = "";
		}

		// render inner text
		if (isset($this->_[HDOM_INFO_INNER]))
		{
			// If it's a br tag...  don't return the HDOM_INNER_INFO that we may or may not have added.
			if ($this->tag != "br")
			{
				$ret .= $this->_[HDOM_INFO_INNER];
			}
		} else {
			if ($this->nodes)
			{
				foreach ($this->nodes as $n)
				{
					$ret .= $this->convert_text($n->outertext());
				}
			}
		}

		// render end tag
		if (isset($this->_[HDOM_INFO_END]) && $this->_[HDOM_INFO_END]!=0)
			$ret .= '</'.$this->tag.'>';
		return $ret;
	}

	/**
	 * Get node's plain text (everything excluding all tags)
	 *
	 * @return string
	 */
	function text()
	{
		if (isset($this->_[HDOM_INFO_INNER])) return $this->_[HDOM_INFO_INNER];
		switch ($this->nodetype)
		{
			case HDOM_TYPE_TEXT: return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
			case HDOM_TYPE_COMMENT: return '';
			case HDOM_TYPE_UNKNOWN: return '';
		}
		if (strcasecmp($this->tag, 'script')===0) return '';
		if (strcasecmp($this->tag, 'style')===0) return '';

		$ret = '';
		// In rare cases, (always node type 1 or HDOM_TYPE_ELEMENT - observed for some span tags, and some p tags) $this->nodes is set to NULL.
		// NOTE: This indicates that there is a problem where it's set to NULL without a clear happening.
		// WHY is this happening?
		if (!is_null($this->nodes))
		{
			foreach ($this->nodes as $n)
			{
				// Start paragraph after a blank line
				if ($n->tag == 'p')
				{
					$ret .= "\n\n";
				}

				$ret .= $this->convert_text($n->text());

				// If this node is a span... add a space at the end of it so multiple spans don't run into each other.  This is plaintext after all.
				if ($n->tag == "span")
				{
					$ret .= $this->dom->default_span_text;
				}
			}
		}
		return trim($ret);
	}

	/**
	 * Get node's xml text (inner text as a CDATA section)
	 *
	 * @return string
	 */
	function xmltext()
	{
		$ret = $this->innertext();
		$ret = str_ireplace('<![CDATA[', '', $ret);
		$ret = str_replace(']]>', '', $ret);
		return $ret;
	}

	// build node's text with tag
	function makeup()
	{
		// text, comment, unknown
		if (isset($this->_[HDOM_INFO_TEXT])) return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);

		$ret = '<'.$this->tag;
		$i = -1;

		foreach ($this->attr as $key=>$val)
		{
			++$i;

			// skip removed attribute
			if ($val===null || $val===false)
				continue;

			$ret .= $this->_[HDOM_INFO_SPACE][$i][0];
			//no value attr: nowrap, checked selected...
			if ($val===true)
				$ret .= $key;
			else {
				switch ($this->_[HDOM_INFO_QUOTE][$i])
				{
					case HDOM_QUOTE_DOUBLE: $quote = '"'; break;
					case HDOM_QUOTE_SINGLE: $quote = '\''; break;
					default: $quote = '';
				}
				$ret .= $key.$this->_[HDOM_INFO_SPACE][$i][1].'='.$this->_[HDOM_INFO_SPACE][$i][2].$quote.$val.$quote;
			}
		}
		$ret = $this->dom->restore_noise($ret);
		return $ret . $this->_[HDOM_INFO_ENDSPACE] . '>';
	}

	// find elements by css selector
	//PaperG - added ability for find to lowercase the value of the selector.
	function find($selector, $idx=null, $lowercase=false)
	{
		$selectors = $this->parse_selector($selector);
		if (($count=count($selectors))===0) return array();
		$found_keys = array();

		// find each selector
		for ($c=0; $c<$count; ++$c)
		{
			// The change on the below line was documented on the sourceforge code tracker id 2788009
			// used to be: if (($levle=count($selectors[0]))===0) return array();
			if (($levle=count($selectors[$c]))===0) return array();
			if (!isset($this->_[HDOM_INFO_BEGIN])) return array();

			$head = array($this->_[HDOM_INFO_BEGIN]=>1);

			// handle descendant selectors, no recursive!
			for ($l=0; $l<$levle; ++$l)
			{
				$ret = array();
				foreach ($head as $k=>$v)
				{
					$n = ($k===-1) ? $this->dom->root : $this->dom->nodes[$k];
					//PaperG - Pass this optional parameter on to the seek function.
					$n->seek($selectors[$c][$l], $ret, $lowercase);
				}
				$head = $ret;
			}

			foreach ($head as $k=>$v)
			{
				if (!isset($found_keys[$k]))
				{
					$found_keys[$k] = 1;
				}
			}
		}

		// sort keys
		ksort($found_keys);

		$found = array();
		foreach ($found_keys as $k=>$v)
			$found[] = $this->dom->nodes[$k];

		// return nth-element or array
		if (is_null($idx)) return $found;
		else if ($idx<0) $idx = count($found) + $idx;
		return (isset($found[$idx])) ? $found[$idx] : null;
	}

	// seek for given conditions
	// PaperG - added parameter to allow for case insensitive testing of the value of a selector.
	protected function seek($selector, &$ret, $lowercase=false)
	{
		global $debug_object;
		if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

		list($tag, $key, $val, $exp, $no_key) = $selector;

		// xpath index
		if ($tag && $key && is_numeric($key))
		{
			$count = 0;
			foreach ($this->children as $c)
			{
				if ($tag==='*' || $tag===$c->tag) {
					if (++$count==$key) {
						$ret[$c->_[HDOM_INFO_BEGIN]] = 1;
						return;
					}
				}
			}
			return;
		}

		$end = (!empty($this->_[HDOM_INFO_END])) ? $this->_[HDOM_INFO_END] : 0;
		if ($end==0) {
			$parent = $this->parent;
			while (!isset($parent->_[HDOM_INFO_END]) && $parent!==null) {
				$end -= 1;
				$parent = $parent->parent;
			}
			$end += $parent->_[HDOM_INFO_END];
		}

		for ($i=$this->_[HDOM_INFO_BEGIN]+1; $i<$end; ++$i) {
			$node = $this->dom->nodes[$i];

			$pass = true;

			if ($tag==='*' && !$key) {
				if (in_array($node, $this->children, true))
					$ret[$i] = 1;
				continue;
			}

			// compare tag
			if ($tag && $tag!=$node->tag && $tag!=='*') {$pass=false;}
			// compare key
			if ($pass && $key) {
				if ($no_key) {
					if (isset($node->attr[$key])) $pass=false;
				} else {
					if (($key != "plaintext") && !isset($node->attr[$key])) $pass=false;
				}
			}
			// compare value
			if ($pass && $key && $val  && $val!=='*') {
				// If they have told us that this is a "plaintext" search then we want the plaintext of the node - right?
				if ($key == "plaintext") {
					// $node->plaintext actually returns $node->text();
					$nodeKeyValue = $node->text();
				} else {
					// this is a normal search, we want the value of that attribute of the tag.
					$nodeKeyValue = $node->attr[$key];
				}
				if (is_object($debug_object)) {$debug_object->debug_log(2, "testing node: " . $node->tag . " for attribute: " . $key . $exp . $val . " where nodes value is: " . $nodeKeyValue);}

				//PaperG - If lowercase is set, do a case insensitive test of the value of the selector.
				if ($lowercase) {
					$check = $this->match($exp, strtolower($val), strtolower($nodeKeyValue));
				} else {
					$check = $this->match($exp, $val, $nodeKeyValue);
				}
				if (is_object($debug_object)) {$debug_object->debug_log(2, "after match: " . ($check ? "true" : "false"));}

				// handle multiple class
				if (!$check && strcasecmp($key, 'class')===0) {
					foreach (explode(' ',$node->attr[$key]) as $k) {
						// Without this, there were cases where leading, trailing, or double spaces lead to our comparing blanks - bad form.
						if (!empty($k)) {
							if ($lowercase) {
								$check = $this->match($exp, strtolower($val), strtolower($k));
							} else {
								$check = $this->match($exp, $val, $k);
							}
							if ($check) break;
						}
					}
				}
				if (!$check) $pass = false;
			}
			if ($pass) $ret[$i] = 1;
			unset($node);
		}
		// It's passed by reference so this is actually what this function returns.
		if (is_object($debug_object)) {$debug_object->debug_log(1, "EXIT - ret: ", $ret);}
	}

	protected function match($exp, $pattern, $value) {
		global $debug_object;
		if (is_object($debug_object)) {$debug_object->debug_log_entry(1);}

		switch ($exp) {
			case '=':
				return ($value===$pattern);
			case '!=':
				return ($value!==$pattern);
			case '^=':
				return preg_match("/^".preg_quote($pattern,'/')."/", $value);
			case '$=':
				return preg_match("/".preg_quote($pattern,'/')."$/", $value);
			case '*=':
				if ($pattern[0]=='/') {
					return preg_match($pattern, $value);
				}
				return preg_match("/".$pattern."/i", $value);
		}
		return false;
	}

	protected function parse_selector($selector_string) {
		global $debug_object;
		if (is_object($debug_object)) {$debug_object->debug_log_entry(1);}

		// pattern of CSS selectors, modified from mootools
		// Paperg: Add the colon to the attrbute, so that it properly finds <tag attr:ibute="something" > like google does.
		// Note: if you try to look at this attribute, yo MUST use getAttribute since $dom->x:y will fail the php syntax check.
// Notice the \[ starting the attbute?  and the @? following?  This implies that an attribute can begin with an @ sign that is not captured.
// This implies that an html attribute specifier may start with an @ sign that is NOT captured by the expression.
// farther study is required to determine of this should be documented or removed.
//		$pattern = "/([\w-:\*]*)(?:\#([\w-]+)|\.([\w-]+))?(?:\[@?(!?[\w-]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is";
		$pattern = "/([\w:\*-]*)(?:\#([\w-]+)|\.([\w-]+))?(?:\[@?(!?[\w:-]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is";
		preg_match_all($pattern, trim($selector_string).' ', $matches, PREG_SET_ORDER);
		if (is_object($debug_object)) {$debug_object->debug_log(2, "Matches Array: ", $matches);}

		$selectors = array();
		$result = array();
		//print_r($matches);

		foreach ($matches as $m) {
			$m[0] = trim($m[0]);
			if ($m[0]==='' || $m[0]==='/' || $m[0]==='//') continue;
			// for browser generated xpath
			if ($m[1]==='tbody') continue;

			list($tag, $key, $val, $exp, $no_key) = array($m[1], null, null, '=', false);
			if (!empty($m[2])) {$key='id'; $val=$m[2];}
			if (!empty($m[3])) {$key='class'; $val=$m[3];}
			if (!empty($m[4])) {$key=$m[4];}
			if (!empty($m[5])) {$exp=$m[5];}
			if (!empty($m[6])) {$val=$m[6];}
			if($tag === null)
			{
				$tag = '';
			}
			if($key === null)
			{
				$key = '';
			}
			// convert to lowercase
			if ($this->dom->lowercase) {$tag=strtolower($tag); $key=strtolower($key);}
			//elements that do NOT have the specified attribute
			if (isset($key[0]) && $key[0]==='!') {$key=substr($key, 1); $no_key=true;}

			$result[] = array($tag, $key, $val, $exp, $no_key);
			if (trim($m[7])===',') {
				$selectors[] = $result;
				$result = array();
			}
		}
		if (count($result)>0)
			$selectors[] = $result;
		return $selectors;
	}

	function __get($name)
	{
		if (isset($this->attr[$name]))
		{
			return $this->convert_text($this->attr[$name]);
		}
		switch ($name)
		{
			case 'outertext': return $this->outertext();
			case 'innertext': return $this->innertext();
			case 'plaintext': return $this->text();
			case 'xmltext': return $this->xmltext();
			default: return array_key_exists($name, $this->attr);
		}
	}

	function __set($name, $value)
	{
		global $debug_object;
		if (is_object($debug_object)) {$debug_object->debug_log_entry(1);}

		switch ($name)
		{
			case 'outertext': return $this->_[HDOM_INFO_OUTER] = $value;
			case 'innertext':
				if (isset($this->_[HDOM_INFO_TEXT])) return $this->_[HDOM_INFO_TEXT] = $value;
				return $this->_[HDOM_INFO_INNER] = $value;
		}
		if (!isset($this->attr[$name]))
		{
			$this->_[HDOM_INFO_SPACE][] = array(' ', '', '');
			$this->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_DOUBLE;
		}
		$this->attr[$name] = $value;
	}

	function __isset($name)
	{
		switch ($name)
		{
			case 'outertext': return true;
			case 'innertext': return true;
			case 'plaintext': return true;
		}
		//no value attr: nowrap, checked selected...
		return (array_key_exists($name, $this->attr)) ? true : isset($this->attr[$name]);
	}

	function __unset($name) {
		if (isset($this->attr[$name]))
			unset($this->attr[$name]);
	}

	// PaperG - Function to convert the text from one character set to another if the two sets are not the same.
	function convert_text($text)
	{
		global $debug_object;
		if (is_object($debug_object)) {$debug_object->debug_log_entry(1);}

		$converted_text = $text;

		$sourceCharset = "";
		$targetCharset = "";

		if ($this->dom)
		{
			$sourceCharset = strtoupper($this->dom->_charset);
			$targetCharset = strtoupper($this->dom->_target_charset);
		}
		if (is_object($debug_object)) {$debug_object->debug_log(3, "source charset: " . $sourceCharset . " target charaset: " . $targetCharset);}

		if (!empty($sourceCharset) && !empty($targetCharset) && (strcasecmp($sourceCharset, $targetCharset) != 0))
		{
			// Check if the reported encoding could have been incorrect and the text is actually already UTF-8
			if ((strcasecmp($targetCharset, 'UTF-8') == 0) && ($this->is_utf8($text)))
			{
				$converted_text = $text;
			}
			else
			{
				$converted_text = iconv($sourceCharset, $targetCharset, $text);
			}
		}

		// Lets make sure that we don't have that silly BOM issue with any of the utf-8 text we output.
		if ($targetCharset == 'UTF-8')
		{
			if (substr($converted_text, 0, 3) == "\xef\xbb\xbf")
			{
				$converted_text = substr($converted_text, 3);
			}
			if (substr($converted_text, -3) == "\xef\xbb\xbf")
			{
				$converted_text = substr($converted_text, 0, -3);
			}
		}

		return $converted_text;
	}

	/**
	* Returns true if $string is valid UTF-8 and false otherwise.
	*
	* @param mixed $str String to be tested
	* @return boolean
	*/
	static function is_utf8($str)
	{
		$c=0; $b=0;
		$bits=0;
		$len=strlen($str);
		for($i=0; $i<$len; $i++)
		{
			$c=ord($str[$i]);
			if($c > 128)
			{
				if(($c >= 254)) return false;
				elseif($c >= 252) $bits=6;
				elseif($c >= 248) $bits=5;
				elseif($c >= 240) $bits=4;
				elseif($c >= 224) $bits=3;
				elseif($c >= 192) $bits=2;
				else return false;
				if(($i+$bits) > $len) return false;
				while($bits > 1)
				{
					$i++;
					$b=ord($str[$i]);
					if($b < 128 || $b > 191) return false;
					$bits--;
				}
			}
		}
		return true;
	}
	/*
	function is_utf8($string)
	{
		//this is buggy
		return (utf8_encode(utf8_decode($string)) == $string);
	}
	*/

	/**
	 * Function to try a few tricks to determine the displayed size of an img on the page.
	 * NOTE: This will ONLY work on an IMG tag. Returns FALSE on all other tag types.
	 *
	 * @author John Schlick
	 * @version April 19 2012
	 * @return array an array containing the 'height' and 'width' of the image on the page or -1 if we can't figure it out.
	 */
	function get_display_size()
	{
		global $debug_object;

		$width = -1;
		$height = -1;

		if ($this->tag !== 'img')
		{
			return false;
		}

		// See if there is aheight or width attribute in the tag itself.
		if (isset($this->attr['width']))
		{
			$width = $this->attr['width'];
		}

		if (isset($this->attr['height']))
		{
			$height = $this->attr['height'];
		}

		// Now look for an inline style.
		if (isset($this->attr['style']))
		{
			// Thanks to user gnarf from stackoverflow for this regular expression.
			$attributes = array();
			preg_match_all("/([\w-]+)\s*:\s*([^;]+)\s*;?/", $this->attr['style'], $matches, PREG_SET_ORDER);
			foreach ($matches as $match) {
			  $attributes[$match[1]] = $match[2];
			}

			// If there is a width in the style attributes:
			if (isset($attributes['width']) && $width == -1)
			{
				// check that the last two characters are px (pixels)
				if (strtolower(substr($attributes['width'], -2)) == 'px')
				{
					$proposed_width = substr($attributes['width'], 0, -2);
					// Now make sure that it's an integer and not something stupid.
					if (filter_var($proposed_width, FILTER_VALIDATE_INT))
					{
						$width = $proposed_width;
					}
				}
			}

			// If there is a width in the style attributes:
			if (isset($attributes['height']) && $height == -1)
			{
				// check that the last two characters are px (pixels)
				if (strtolower(substr($attributes['height'], -2)) == 'px')
				{
					$proposed_height = substr($attributes['height'], 0, -2);
					// Now make sure that it's an integer and not something stupid.
					if (filter_var($proposed_height, FILTER_VALIDATE_INT))
					{
						$height = $proposed_height;
					}
				}
			}

		}

		// Future enhancement:
		// Look in the tag to see if there is a class or id specified that has a height or width attribute to it.

		// Far future enhancement
		// Look at all the parent tags of this image to see if they specify a class or id that has an img selector that specifies a height or width
		// Note that in this case, the class or id will have the img subselector for it to apply to the image.

		// ridiculously far future development
		// If the class or id is specified in a SEPARATE css file thats not on the page, go get it and do what we were just doing for the ones on the page.

		$result = array('height' => $height,
						'width' => $width);
		return $result;
	}

	// camel naming conventions
	function getAllAttributes() {return $this->attr;}
	function getAttribute($name) {return $this->__get($name);}
	function setAttribute($name, $value) {$this->__set($name, $value);}
	function hasAttribute($name) {return $this->__isset($name);}
	function removeAttribute($name) {$this->__set($name, null);}
	function getElementById($id) {return $this->find("#$id", 0);}
	function getElementsById($id, $idx=null) {return $this->find("#$id", $idx);}
	function getElementByTagName($name) {return $this->find($name, 0);}
	function getElementsByTagName($name, $idx=null) {return $this->find($name, $idx);}
	function parentNode() {return $this->parent();}
	function childNodes($idx=-1) {return $this->children($idx);}
	function firstChild() {return $this->first_child();}
	function lastChild() {return $this->last_child();}
	function nextSibling() {return $this->next_sibling();}
	function previousSibling() {return $this->prev_sibling();}
	function hasChildNodes() {return $this->has_child();}
	function nodeName() {return $this->tag;}
	function appendChild($node) {$node->parent($this); return $node;}

}

/**
 * simple html dom parser
 * Paperg - in the find routine: allow us to specify that we want case insensitive testing of the value of the selector.
 * Paperg - change $size from protected to public so we can easily access it
 * Paperg - added ForceTagsClosed in the constructor which tells us whether we trust the html or not.  Default is to NOT trust it.
 *
 * @package PlaceLocalInclude
 */
class ums_simple_html_dom
{
	/**
	 * The root node of the document
	 *
	 * @var object
	 */
	public $root = null;

	/**
	 * List of nodes in the current DOM
	 *
	 * @var array
	 */
	public $nodes = array();

	/**
	 * Callback function to run for each element in the DOM.
	 *
	 * @var callable|null
	 */
	public $callback = null;

	/**
	 * Indicates how tags and attributes are matched
	 *
	 * @var bool When set to **true** tags and attributes will be converted to
	 * lowercase before matching.
	 */
	public $lowercase = false;

	/**
	 * Original document size
	 *
	 * Holds the original document size.
	 *
	 * @var int
	 */
	public $original_size;

	/**
	 * Current document size
	 *
	 * Holds the current document size. The document size is determined by the
	 * string length of ({@see ums_simple_html_dom::$doc}).
	 *
	 * _Note_: Using this variable is more efficient than calling `strlen($doc)`
	 *
	 * @var int
	 * */
	public $size;

	/**
	 * Current position in the document
	 *
	 * @var int
	 */
	protected $pos;

	/**
	 * The document
	 *
	 * @var string
	 */
	protected $doc;

	/**
	 * Current character
	 *
	 * Holds the current character at position {@see ums_simple_html_dom::$pos} in
	 * the document {@see ums_simple_html_dom::$doc}
	 *
	 * _Note_: Using this variable is more efficient than calling `substr($doc, $pos, 1)`
	 *
	 * @var string
	 */
	protected $char;

	protected $cursor;

	/**
	 * Parent node of the next node detected by the parser
	 *
	 * @var object
	 */
	protected $parent;
	protected $noise = array();

	/**
	 * Tokens considered blank in HTML
	 *
	 * @var string
	 */
	protected $token_blank = " \t\r\n";

	/**
	 * Tokens to identify the equal sign for attributes, stopping either at the
	 * closing tag ("/" i.e. "<html />") or the end of an opening tag (">" i.e.
	 * "<html>")
	 *
	 * @var string
	 */
	protected $token_equal = ' =/>';

	/**
	 * Tokens to identify the end of a tag name. A tag name either ends on the
	 * ending slash ("/" i.e. "<html/>") or whitespace ("\s\r\n\t")
	 *
	 * @var string
	 */
	protected $token_slash = " />\r\n\t";

	/**
	 * Tokens to identify the end of an attribute
	 *
	 * @var string
	 */
	protected $token_attr = ' >';

	// Note that this is referenced by a child node, and so it needs to be public for that node to see this information.
	public $_charset = '';
	public $_target_charset = '';

	/**
	 * Innertext for <br> elements
	 *
	 * @var string
	 */
	protected $default_br_text = "";

	/**
	 * Suffix for <span> elements
	 *
	 * @var string
	 */
	public $default_span_text = "";

	/**
	 * Defines a list of self-closing tags (Void elements) according to the HTML
	 * Specification
	 *
	 * _Remarks_:
	 * - Use `isset()` instead of `in_array()` on array elements to boost
	 * performance about 30%
	 * - Sort elements by name for better readability!
	 *
	 * @link https://www.w3.org/TR/html HTML Specification
	 * @link https://www.w3.org/TR/html/syntax.html#void-elements Void elements
	 */
	protected $self_closing_tags = array(
		'area'=>1,
		'base'=>1,
		'br'=>1,
		'col'=>1,
		'embed'=>1,
		'hr'=>1,
		'img'=>1,
		'input'=>1,
		'link'=>1,
		'meta'=>1,
		'param'=>1,
		'source'=>1,
		'track'=>1,
		'wbr'=>1
	);

	/**
	 * Defines a list of tags which - if closed - close all optional closing
	 * elements within if they haven't been closed yet. (So, an element where
	 * neither opening nor closing tag is omissible consistently closes every
	 * optional closing element within)
	 *
	 * _Remarks_:
	 * - Use `isset()` instead of `in_array()` on array elements to boost
	 * performance about 30%
	 * - Sort elements by name for better readability!
	 */
	protected $block_tags = array(
		'body'=>1,
		'div'=>1,
		'form'=>1,
		'root'=>1,
		'span'=>1,
		'table'=>1
	);

	/**
	 * Defines elements whose end tag is omissible.
	 *
	 * * key = Name of an element whose end tag is omissible.
	 * * value = Names of elements whose end tag is omissible, that are closed
	 * by the current element.
	 *
	 * _Remarks_:
	 * - Use `isset()` instead of `in_array()` on array elements to boost
	 * performance about 30%
	 * - Sort elements by name for better readability!
	 *
	 * **Example**
	 *
	 * An `li` element???s end tag may be omitted if the `li` element is immediately
	 * followed by another `li` element. To do that, add following element to the
	 * array:
	 *
	 * ```php
	 * 'li' => array('li'),
	 * ```
	 *
	 * With this, the following two examples are considered equal. Note that the
	 * second example is missing the closing tags on `li` elements.
	 *
	 * ```html
	 * <ul><li>First Item</li><li>Second Item</li></ul>
	 * ```
	 *
	 * <ul><li>First Item</li><li>Second Item</li></ul>
	 *
	 * ```html
	 * <ul><li>First Item<li>Second Item</ul>
	 * ```
	 *
	 * <ul><li>First Item<li>Second Item</ul>
	 *
	 * @var array A two-dimensional array where the key is the name of an
	 * element whose end tag is omissible and the value is an array of elements
	 * whose end tag is omissible, that are closed by the current element.
	 *
	 * @link https://www.w3.org/TR/html/syntax.html#optional-tags Optional tags
	 *
	 * @todo The implementation of optional closing tags doesn't work in all cases
	 * because it only consideres elements who close other optional closing
	 * tags, not taking into account that some (non-blocking) tags should close
	 * these optional closing tags. For example, the end tag for "p" is omissible
	 * and can be closed by an "address" element, whose end tag is NOT omissible.
	 * Currently a "p" element without closing tag stops at the next "p" element
	 * or blocking tag, even if it contains other elements.
	 *
	 * @todo Known sourceforge issue #2977341
	 * B tags that are not closed cause us to return everything to the end of
	 * the document.
	 */
	protected $optional_closing_tags = array(
		'b'=>array('b'=>1), // Not optional, see https://www.w3.org/TR/html/textlevel-semantics.html#the-b-element
		'dd'=>array('dd'=>1, 'dt'=>1),
		'dl'=>array('dd'=>1, 'dt'=>1), // Not optional, see https://www.w3.org/TR/html/grouping-content.html#the-dl-element
		'dt'=>array('dd'=>1, 'dt'=>1),
		'li'=>array('li'=>1),
		'optgroup'=>array('optgroup'=>1, 'option'=>1),
		'option'=>array('optgroup'=>1, 'option'=>1),
		'p'=>array('p'=>1),
		'rp'=>array('rp'=>1, 'rt'=>1),
		'rt'=>array('rp'=>1, 'rt'=>1),
		'td'=>array('td'=>1, 'th'=>1),
		'th'=>array('td'=>1, 'th'=>1),
		'tr'=>array('td'=>1, 'th'=>1, 'tr'=>1),
	);

	function __construct($str=null, $lowercase=true, $forceTagsClosed=true, $target_charset=DEFAULT_TARGET_CHARSET, $stripRN=true, $defaultBRText=DEFAULT_BR_TEXT, $defaultSpanText=DEFAULT_SPAN_TEXT, $options=0)
	{
		if ($str)
		{
			global $wp_filesystem;
            if ( ! is_a( $wp_filesystem, 'WP_Filesystem_Base') ){
                include_once(ABSPATH . 'wp-admin/includes/file.php');$creds = request_filesystem_credentials( site_url() );
               wp_filesystem($creds);
            }
			if (preg_match("/^http:\/\//i",$str) || $wp_filesystem->is_file($str))
			{
				$this->load_file($str);
			}
			else
			{
				$this->load($str, $lowercase, $stripRN, $defaultBRText, $defaultSpanText, $options);
			}
		}
		// Forcing tags to be closed implies that we don't trust the html, but it can lead to parsing errors if we SHOULD trust the html.
		if (!$forceTagsClosed) {
			$this->optional_closing_array=array();
		}
		$this->_target_charset = $target_charset;
	}

	function __destruct()
	{
		$this->clear();
	}

	// load html from string
	function load($str, $lowercase=true, $stripRN=true, $defaultBRText=DEFAULT_BR_TEXT, $defaultSpanText=DEFAULT_SPAN_TEXT, $options=0)
	{
		global $debug_object;

		// prepare
		$this->prepare($str, $lowercase, $defaultBRText, $defaultSpanText);

		// Per sourceforge http://sourceforge.net/tracker/?func=detail&aid=2949097&group_id=218559&atid=1044037
		// Script tags removal now preceeds style tag removal.
		// strip out <script> tags
		$this->remove_noise("'<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>'is");
		$this->remove_noise("'<\s*script\s*>(.*?)<\s*/\s*script\s*>'is");

		// strip out the \r \n's if we are told to.
		if ($stripRN) {
			$this->doc = str_replace("\r", " ", $this->doc);
			$this->doc = str_replace("\n", " ", $this->doc);

			// set the length of content since we have changed it.
			$this->size = strlen($this->doc);
		}

		// strip out cdata
		$this->remove_noise("'<!\[CDATA\[(.*?)\]\]>'is", true);
		// strip out comments
		$this->remove_noise("'<!--(.*?)-->'is");
		// strip out <style> tags
		$this->remove_noise("'<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>'is");
		$this->remove_noise("'<\s*style\s*>(.*?)<\s*/\s*style\s*>'is");
		// strip out preformatted tags
		$this->remove_noise("'<\s*(?:code)[^>]*>(.*?)<\s*/\s*(?:code)\s*>'is");
		// strip out server side scripts
		$this->remove_noise("'(<\?)(.*?)(\?>)'s", true);

		if($options & HDOM_SMARTY_AS_TEXT) { // Strip Smarty scripts
			$this->remove_noise("'(\{\w)(.*?)(\})'s", true);
		}

		// parsing
		$this->parse();
		// end
		$this->root->_[HDOM_INFO_END] = $this->cursor;
		$this->parse_charset();

		// make load function chainable
		return $this;

	}

	// load html from file
	function load_file()
	{
		$args = func_get_args();

		if(count($args) == 0)
        {
            return false;
        }
        global $wp_filesystem;
        if ( ! is_a( $wp_filesystem, 'WP_Filesystem_Base') ){
            include_once(ABSPATH . 'wp-admin/includes/file.php');$creds = request_filesystem_credentials( site_url() );
            wp_filesystem($creds);
        }
        $doc = $wp_filesystem->get_contents($args[0]);
		if($doc !== false) {
			$this->load($doc, true);
		} else {
			return false;
		}
	}

	/**
	 * Set the callback function
	 *
	 * @param callable $function_name Callback function to run for each element
	 * in the DOM.
	 * @return void
	 */
	function set_callback($function_name)
	{
		$this->callback = $function_name;
	}

	/**
	 * Remove callback function
	 *
	 * @return void
	 */
	function remove_callback()
	{
		$this->callback = null;
	}

	// save dom as string
	function save($filepath='')
	{
		$ret = $this->root->innertext();
		if ($filepath!=='') 
        {
            global $wp_filesystem;
            if ( ! is_a( $wp_filesystem, 'WP_Filesystem_Base') ){
                include_once(ABSPATH . 'wp-admin/includes/file.php');$creds = request_filesystem_credentials( site_url() );
                wp_filesystem($creds);
            }
            $wp_filesystem->put_contents($filepath, $ret);
        }
		return $ret;
	}

	// find dom node by css selector
	// Paperg - allow us to specify that we want case insensitive testing of the value of the selector.
	function find($selector, $idx=null, $lowercase=false)
	{
		return $this->root->find($selector, $idx, $lowercase);
	}

	// clean up memory due to php5 circular references memory leak...
	function clear()
	{
		foreach ($this->nodes as $n) {$n->clear(); $n = null;}
		// This add next line is documented in the sourceforge repository. 2977248 as a fix for ongoing memory leaks that occur even with the use of clear.
		if (isset($this->children)) foreach ($this->children as $n) {$n->clear(); $n = null;}
		if (isset($this->parent)) {$this->parent->clear(); unset($this->parent);}
		if (isset($this->root)) {$this->root->clear(); unset($this->root);}
		unset($this->doc);
		unset($this->noise);
	}

	function dump($show_attr=true)
	{
		$this->root->dump($show_attr);
	}

	// prepare HTML data and init everything
	protected function prepare($str, $lowercase=true, $defaultBRText=DEFAULT_BR_TEXT, $defaultSpanText=DEFAULT_SPAN_TEXT)
	{
		$this->clear();

		$this->doc = trim($str);
		$this->size = strlen($this->doc);
		$this->original_size = $this->size; // Save the original size of the html that we got in.  It might be useful to someone.
		$this->pos = 0;
		$this->cursor = 1;
		$this->noise = array();
		$this->nodes = array();
		$this->lowercase = $lowercase;
		$this->default_br_text = $defaultBRText;
		$this->default_span_text = $defaultSpanText;
		$this->root = new ums_simple_html_dom_node($this);
		$this->root->tag = 'root';
		$this->root->_[HDOM_INFO_BEGIN] = -1;
		$this->root->nodetype = HDOM_TYPE_ROOT;
		$this->parent = $this->root;
		if ($this->size>0) $this->char = $this->doc[0];
	}

	/**
	 * Parse HTML content
	 *
	 * @return bool True on success
	 */
	protected function parse()
	{
		while (true) {
			// Read next tag if there is no text between current position and the
			// next opening tag.
			if (($s = $this->copy_until_char('<'))==='')
			{
				if($this->read_tag()) {
					continue;
				} else {
					return true;
				}
			}

			// Add a text node for text between tags
			$node = new ums_simple_html_dom_node($this);
			++$this->cursor;
			$node->_[HDOM_INFO_TEXT] = $s;
			$this->link_nodes($node, false);
		}
	}

	// PAPERG - dkchou - added this to try to identify the character set of the page we have just parsed so we know better how to spit it out later.
	// NOTE:  IF you provide a routine called get_last_retrieve_url_contents_content_type which returns the CURLINFO_CONTENT_TYPE from the last curl_exec
	// (or the content_type header from the last transfer), we will parse THAT, and if a charset is specified, we will use it over any other mechanism.
	protected function parse_charset()
	{
		global $debug_object;

		$charset = null;

		if (function_exists('get_last_retrieve_url_contents_content_type'))
		{
			$contentTypeHeader = get_last_retrieve_url_contents_content_type();
			$success = preg_match('/charset=(.+)/', $contentTypeHeader, $matches);
			if ($success)
			{
				$charset = $matches[1];
				if (is_object($debug_object)) {$debug_object->debug_log(2, 'header content-type found charset of: ' . $charset);}
			}

		}

		if (empty($charset))
		{
			$el = $this->root->find('meta[http-equiv=Content-Type]',0, true);
			if (!empty($el))
			{
				$fullvalue = $el->content;
				if (is_object($debug_object)) {$debug_object->debug_log(2, 'meta content-type tag found' . $fullvalue);}

				if (!empty($fullvalue))
				{
					$success = preg_match('/charset=(.+)/i', $fullvalue, $matches);
					if ($success)
					{
						$charset = $matches[1];
					}
					else
					{
						// If there is a meta tag, and they don't specify the character set, research says that it's typically ISO-8859-1
						if (is_object($debug_object)) {$debug_object->debug_log(2, 'meta content-type tag couldn\'t be parsed. using iso-8859 default.');}
						$charset = 'ISO-8859-1';
					}
				}
			}
		}

		// If we couldn't find a charset above, then lets try to detect one based on the text we got...
		if (empty($charset))
		{
			// Use this in case mb_detect_charset isn't installed/loaded on this machine.
			$charset = false;
			if (function_exists('mb_detect_encoding'))
			{
				// Have php try to detect the encoding from the text given to us.
				$charset = mb_detect_encoding($this->doc . "ascii", $encoding_list = array( "UTF-8", "CP1252" ) );
				if (is_object($debug_object)) {$debug_object->debug_log(2, 'mb_detect found: ' . $charset);}
			}

			// and if this doesn't work...  then we need to just wrongheadedly assume it's UTF-8 so that we can move on - cause this will usually give us most of what we need...
			if ($charset === false)
			{
				if (is_object($debug_object)) {$debug_object->debug_log(2, 'since mb_detect failed - using default of utf-8');}
				$charset = 'UTF-8';
			}
		}

		// Since CP1252 is a superset, if we get one of it's subsets, we want it instead.
		if ((strtolower($charset) == strtolower('ISO-8859-1')) || (strtolower($charset) == strtolower('Latin1')) || (strtolower($charset) == strtolower('Latin-1')))
		{
			if (is_object($debug_object)) {$debug_object->debug_log(2, 'replacing ' . $charset . ' with CP1252 as its a superset');}
			$charset = 'CP1252';
		}

		if (is_object($debug_object)) {$debug_object->debug_log(1, 'EXIT - ' . $charset);}

		return $this->_charset = $charset;
	}

	/**
	 * Parse tag from current document position.
	 *
	 * @return bool True if a tag was found, false otherwise
	 */
	protected function read_tag()
	{
		// Set end position if no further tags found
		if ($this->char!=='<')
		{
			$this->root->_[HDOM_INFO_END] = $this->cursor;
			return false;
		}
		$begin_tag_pos = $this->pos;
		$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next

		// end tag
		if ($this->char==='/')
		{
			$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next

			// Skip whitespace in end tags (i.e. in "</   html>")
			$this->skip($this->token_blank);
			$tag = $this->copy_until_char('>');

			// Skip attributes in end tags
			if (($pos = strpos($tag, ' '))!==false)
				$tag = substr($tag, 0, $pos);

			$parent_lower = strtolower($this->parent->tag);
			$tag_lower = strtolower($tag);

			// The end tag is supposed to close the parent tag. Handle situations
			// when it doesn't
			if ($parent_lower!==$tag_lower)
			{
				// Parent tag does not have to be closed necessarily (optional closing tag)
				// Current tag is a block tag, so it may close an ancestor
				if (isset($this->optional_closing_tags[$parent_lower]) && isset($this->block_tags[$tag_lower]))
				{
					$this->parent->_[HDOM_INFO_END] = 0;
					$org_parent = $this->parent;

					// Traverse ancestors to find a matching opening tag
					// Stop at root node
					while (($this->parent->parent) && strtolower($this->parent->tag)!==$tag_lower)
						$this->parent = $this->parent->parent;

					// If we don't have a match add current tag as text node
					if (strtolower($this->parent->tag)!==$tag_lower) {
						$this->parent = $org_parent; // restore origonal parent
						if ($this->parent->parent) $this->parent = $this->parent->parent;
						$this->parent->_[HDOM_INFO_END] = $this->cursor;
						return $this->as_text_node($tag);
					}
				}
				// Grandparent exists and current tag is a block tag, so our parent doesn't have an end tag
				else if (($this->parent->parent) && isset($this->block_tags[$tag_lower]))
				{
					$this->parent->_[HDOM_INFO_END] = 0; // No end tag
					$org_parent = $this->parent;

					// Traverse ancestors to find a matching opening tag
					// Stop at root node
					while (($this->parent->parent) && strtolower($this->parent->tag)!==$tag_lower)
						$this->parent = $this->parent->parent;

					// If we don't have a match add current tag as text node
					if (strtolower($this->parent->tag)!==$tag_lower)
					{
						$this->parent = $org_parent; // restore origonal parent
						$this->parent->_[HDOM_INFO_END] = $this->cursor;
						return $this->as_text_node($tag);
					}
				}
				// Grandparent exists and current tag closes it
				else if (($this->parent->parent) && strtolower($this->parent->parent->tag)===$tag_lower)
				{
					$this->parent->_[HDOM_INFO_END] = 0;
					$this->parent = $this->parent->parent;
				}
				else // Random tag, add as text node
					return $this->as_text_node($tag);
			}

			// Set end position of parent tag to current cursor position
			$this->parent->_[HDOM_INFO_END] = $this->cursor;
			if ($this->parent->parent) $this->parent = $this->parent->parent;

			$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
			return true;
		}

		// start tag
		$node = new ums_simple_html_dom_node($this);
		$node->_[HDOM_INFO_BEGIN] = $this->cursor;
		++$this->cursor;
		$tag = $this->copy_until($this->token_slash); // Get tag name
		$node->tag_start = $begin_tag_pos;

		// doctype, cdata & comments...
		// <!DOCTYPE html>
		// <![CDATA[ ... ]]>
		// <!-- Comment -->
		if (isset($tag[0]) && $tag[0]==='!') {
			$node->_[HDOM_INFO_TEXT] = '<' . $tag . $this->copy_until_char('>');

			if (isset($tag[2]) && $tag[1]==='-' && $tag[2]==='-') { // Comment ("<!--")
				$node->nodetype = HDOM_TYPE_COMMENT;
				$node->tag = 'comment';
			} else { // Could be doctype or CDATA but we don't care
				$node->nodetype = HDOM_TYPE_UNKNOWN;
				$node->tag = 'unknown';
			}
			if ($this->char==='>') $node->_[HDOM_INFO_TEXT].='>';
			$this->link_nodes($node, true);
			$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
			return true;
		}

		// The start tag cannot contain another start tag, if so add as text
		// i.e. "<<html>"
		if ($pos=strpos($tag, '<')!==false) {
			$tag = '<' . substr($tag, 0, -1);
			$node->_[HDOM_INFO_TEXT] = $tag;
			$this->link_nodes($node, false);
			$this->char = $this->doc[--$this->pos]; // prev
			return true;
		}

		// Handle invalid tag names (i.e. "<html#doc>")
		if (!preg_match("/^\w[\w:-]*$/", $tag)) {
			$node->_[HDOM_INFO_TEXT] = '<' . $tag . $this->copy_until('<>');

			// Next char is the beginning of a new tag, don't touch it.
			if ($this->char==='<') {
				$this->link_nodes($node, false);
				return true;
			}

			// Next char closes current tag, add and be done with it.
			if ($this->char==='>') $node->_[HDOM_INFO_TEXT].='>';
			$this->link_nodes($node, false);
			$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
			return true;
		}

		// begin tag, add new node
		$node->nodetype = HDOM_TYPE_ELEMENT;
		$tag_lower = strtolower($tag);
		$node->tag = ($this->lowercase) ? $tag_lower : $tag;

		// handle optional closing tags
		if (isset($this->optional_closing_tags[$tag_lower]) )
		{
			// Traverse ancestors to close all optional closing tags
			while (isset($this->optional_closing_tags[$tag_lower][strtolower($this->parent->tag)]))
			{
				$this->parent->_[HDOM_INFO_END] = 0;
				$this->parent = $this->parent->parent;
			}
			$node->parent = $this->parent;
		}

		$guard = 0; // prevent infinity loop
		$space = array($this->copy_skip($this->token_blank), '', ''); // [0] Space between tag and first attribute

		// attributes
		do
		{
			// Everything until the first equal sign should be the attribute name
			$name = $this->copy_until($this->token_equal);

			if ($name==='' && $this->char!==null && $space[0]==='')
			{
				break;
			}

			if ($guard===$this->pos) // Escape infinite loop
			{
				$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
				continue;
			}
			$guard = $this->pos;

			// handle endless '<'
			if ($this->pos>=$this->size-1 && $this->char!=='>') { // Out of bounds before the tag ended
				$node->nodetype = HDOM_TYPE_TEXT;
				$node->_[HDOM_INFO_END] = 0;
				$node->_[HDOM_INFO_TEXT] = '<'.$tag . $space[0] . $name;
				$node->tag = 'text';
				$this->link_nodes($node, false);
				return true;
			}

			// handle mismatch '<'
			if ($this->doc[$this->pos-1]=='<') { // Attributes cannot start after opening tag
				$node->nodetype = HDOM_TYPE_TEXT;
				$node->tag = 'text';
				$node->attr = array();
				$node->_[HDOM_INFO_END] = 0;
				$node->_[HDOM_INFO_TEXT] = substr($this->doc, $begin_tag_pos, $this->pos-$begin_tag_pos-1);
				$this->pos -= 2;
				$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
				$this->link_nodes($node, false);
				return true;
			}

			if ($name!=='/' && $name!=='') { // this is a attribute name
				$space[1] = $this->copy_skip($this->token_blank); // [1] Whitespace after attribute name
				$name = $this->restore_noise($name); // might be a noisy name
				if ($this->lowercase) $name = strtolower($name);
				if ($this->char==='=') { // attribute with value
					$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
					$this->parse_attr($node, $name, $space); // get attribute value
				}
				else {
					//no value attr: nowrap, checked selected...
					$node->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_NO;
					$node->attr[$name] = true;
					if ($this->char!='>') $this->char = $this->doc[--$this->pos]; // prev
				}
				$node->_[HDOM_INFO_SPACE][] = $space;
				$space = array($this->copy_skip($this->token_blank), '', ''); // prepare for next attribute
			}
			else // no more attributes
				break;
		} while ($this->char!=='>' && $this->char!=='/'); // go until the tag ended

		$this->link_nodes($node, true);
		$node->_[HDOM_INFO_ENDSPACE] = $space[0];

		// handle empty tags (i.e. "<div/>")
		if ($this->copy_until_char('>')==='/')
		{
			$node->_[HDOM_INFO_ENDSPACE] .= '/';
			$node->_[HDOM_INFO_END] = 0;
		}
		else
		{
			// reset parent
			if (!isset($this->self_closing_tags[strtolower($node->tag)])) $this->parent = $node;
		}
		$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next

		// If it's a BR tag, we need to set it's text to the default text.
		// This way when we see it in plaintext, we can generate formatting that the user wants.
		// since a br tag never has sub nodes, this works well.
		if ($node->tag == "br")
		{
			$node->_[HDOM_INFO_INNER] = $this->default_br_text;
		}

		return true;
	}

	/**
	 * Parse attribute from current document position
	 *
	 * @param object $node Node for the attributes
	 * @param string $name Name of the current attribute
	 * @param array $space Array for spacing information
	 * @return void
	 */
	protected function parse_attr($node, $name, &$space)
	{
		// Per sourceforge: http://sourceforge.net/tracker/?func=detail&aid=3061408&group_id=218559&atid=1044037
		// If the attribute is already defined inside a tag, only pay attention to the first one as opposed to the last one.
		// https://stackoverflow.com/a/26341866
		if (isset($node->attr[$name]))
		{
			return;
		}

		$space[2] = $this->copy_skip($this->token_blank); // [2] Whitespace between "=" and the value
		switch ($this->char) {
			case '"': // value is anything between double quotes
				$node->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_DOUBLE;
				$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
				$node->attr[$name] = $this->restore_noise($this->copy_until_char('"'));
				$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
				break;
			case '\'': // value is anything between single quotes
				$node->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_SINGLE;
				$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
				$node->attr[$name] = $this->restore_noise($this->copy_until_char('\''));
				$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
				break;
			default: // value is anything until the first space or end tag
				$node->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_NO;
				$node->attr[$name] = $this->restore_noise($this->copy_until($this->token_attr));
		}
		// PaperG: Attributes should not have \r or \n in them, that counts as html whitespace.
		$node->attr[$name] = str_replace("\r", "", $node->attr[$name]);
		$node->attr[$name] = str_replace("\n", "", $node->attr[$name]);
		// PaperG: If this is a "class" selector, lets get rid of the preceeding and trailing space since some people leave it in the multi class case.
		if ($name == "class") {
			$node->attr[$name] = trim($node->attr[$name]);
		}
	}

	/**
	 * Link node to parent node
	 *
	 * @param object $node Node to link to parent
	 * @param bool $is_child True if the node is a child of parent
	 * @return void
	 */
	// link node's parent
	protected function link_nodes(&$node, $is_child)
	{
		$node->parent = $this->parent;
		$this->parent->nodes[] = $node;
		if ($is_child)
		{
			$this->parent->children[] = $node;
		}
	}

	/**
	 * Add tag as text node to current node
	 *
	 * @param string $tag Tag name
	 * @return bool True on success
	 */
	protected function as_text_node($tag)
	{
		$node = new ums_simple_html_dom_node($this);
		++$this->cursor;
		$node->_[HDOM_INFO_TEXT] = '</' . $tag . '>';
		$this->link_nodes($node, false);
		$this->char = (++$this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
		return true;
	}

	/**
	 * Seek from the current document position to the first occurrence of a
	 * character not defined by the provided string. Update the current document
	 * position to the new position.
	 *
	 * @param string $chars A string containing every allowed character.
	 * @return void
	 */
	protected function skip($chars)
	{
		$this->pos += strspn($this->doc, $chars, $this->pos);
		$this->char = ($this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
	}

	/**
	 * Copy substring from the current document position to the first occurrence
	 * of a character not defined by the provided string.
	 *
	 * @param string $chars A string containing every allowed character.
	 * @return string Substring from the current document position to the first
	 * occurrence of a character not defined by the provided string.
	 */
	protected function copy_skip($chars)
	{
		$pos = $this->pos;
		$len = strspn($this->doc, $chars, $pos);
		$this->pos += $len;
		$this->char = ($this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
		if ($len===0) return '';
		return substr($this->doc, $pos, $len);
	}

	/**
	 * Copy substring from the current document position to the first occurrence
	 * of any of the provided characters.
	 *
	 * @param string $chars A string containing every character to stop at.
	 * @return string Substring from the current document position to the first
	 * occurrence of any of the provided characters.
	 */
	protected function copy_until($chars)
	{
		$pos = $this->pos;
		$len = strcspn($this->doc, $chars, $pos);
		$this->pos += $len;
		$this->char = ($this->pos<$this->size) ? $this->doc[$this->pos] : null; // next
		return substr($this->doc, $pos, $len);
	}

	/**
	 * Copy substring from the current document position to the first occurrence
	 * of the provided string.
	 *
	 * @param string $char The string to stop at.
	 * @return string Substring from the current document position to the first
	 * occurrence of the provided string.
	 */
	protected function copy_until_char($char)
	{
		if ($this->char===null) return '';

		if (($pos = strpos($this->doc, $char, $this->pos))===false) {
			$ret = substr($this->doc, $this->pos, $this->size-$this->pos);
			$this->char = null;
			$this->pos = $this->size;
			return $ret;
		}

		if ($pos===$this->pos) return '';
		$pos_old = $this->pos;
		$this->char = $this->doc[$pos];
		$this->pos = $pos;
		return substr($this->doc, $pos_old, $pos-$pos_old);
	}

	/**
	 * Remove noise from HTML content
	 *
	 * Noise is stored to {@see ums_simple_html_dom::$noise}
	 *
	 * @param string $pattern The regex pattern used for finding noise
	 * @param bool $remove_tag True to remove the entire match. Default is false
	 * to only remove the captured data.
	 */
	protected function remove_noise($pattern, $remove_tag=false)
	{
		global $debug_object;
		if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

		$count = preg_match_all($pattern, $this->doc, $matches, PREG_SET_ORDER|PREG_OFFSET_CAPTURE);

		for ($i=$count-1; $i>-1; --$i)
		{
			$key = '___noise___'.sprintf('% 5d', count($this->noise)+1000);
			if (is_object($debug_object)) { $debug_object->debug_log(2, 'key is: ' . $key); }
			$idx = ($remove_tag) ? 0 : 1; // 0 = entire match, 1 = submatch
			$this->noise[$key] = $matches[$i][$idx][0];
			$this->doc = substr_replace($this->doc, $key, $matches[$i][$idx][1], strlen($matches[$i][$idx][0]));
		}

		// reset the length of content
		$this->size = strlen($this->doc);
		if ($this->size>0)
		{
			$this->char = $this->doc[0];
		}
	}

	/**
	 * Restore noise to HTML content
	 *
	 * Noise is restored from {@see ums_simple_html_dom::$noise}
	 *
	 * @param string $text A subset of HTML containing noise
	 * @return string The same content with noise restored
	 */
	function restore_noise($text)
	{
		global $debug_object;
		if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

		while (($pos=strpos($text, '___noise___'))!==false)
		{
			// Sometimes there is a broken piece of markup, and we don't GET the pos+11 etc... token which indicates a problem outside of us...
			if (strlen($text) > $pos+15)
			{	// todo: "___noise___1000" (or any number with four or more digits) in the DOM causes an infinite loop which could be utilized by malicious software
				$key = '___noise___'.$text[$pos+11].$text[$pos+12].$text[$pos+13].$text[$pos+14].$text[$pos+15];
				if (is_object($debug_object)) { $debug_object->debug_log(2, 'located key of: ' . $key); }

				if (isset($this->noise[$key]))
				{
					$text = substr($text, 0, $pos).$this->noise[$key].substr($text, $pos+16);
				}
				else
				{
					// do this to prevent an infinite loop.
					$text = substr($text, 0, $pos).'UNDEFINED NOISE FOR KEY: '.$key . substr($text, $pos+16);
				}
			}
			else
			{
				// There is no valid key being given back to us... We must get rid of the ___noise___ or we will have a problem.
				$text = substr($text, 0, $pos).'NO NUMERIC NOISE KEY' . substr($text, $pos+11);
			}
		}
		return $text;
	}

	// Sometimes we NEED one of the noise elements.
	function search_noise($text)
	{
		global $debug_object;
		if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

		foreach($this->noise as $noiseElement)
		{
			if (strpos($noiseElement, $text)!==false)
			{
				return $noiseElement;
			}
		}
	}
	function __toString()
	{
		return $this->root->innertext();
	}

	function __get($name)
	{
		switch ($name)
		{
			case 'outertext':
				return $this->root->innertext();
			case 'innertext':
				return $this->root->innertext();
			case 'plaintext':
				return $this->root->text();
			case 'charset':
				return $this->_charset;
			case 'target_charset':
				return $this->_target_charset;
		}
	}

	// camel naming conventions
	function childNodes($idx=-1) {return $this->root->childNodes($idx);}
	function firstChild() {return $this->root->first_child();}
	function lastChild() {return $this->root->last_child();}
	function createElement($name, $value=null) {return ums_str_get_html("<$name>$value</$name>")->first_child();}
	function createTextNode($value) {return end(ums_str_get_html($value)->nodes);}
	function getElementById($id) {return $this->find("#$id", 0);}
	function getElementsById($id, $idx=null) {return $this->find("#$id", $idx);}
	function getElementByTagName($name) {return $this->find($name, 0);}
	function getElementsByTagName($name, $idx=-1) {return $this->find($name, $idx);}
	function loadFile() {$args = func_get_args();$this->load_file($args);}
}

?>