.
*/ ?>
won't work,
* it needs to have , just a stupid browser thing
* that doesn't make any sense at all. There may very well be
* others feel free to add them here, it wont hurt anything besides
* page load time a very very slight amount. Removing however can!
*/
'a', 'textarea', 'script', 'p', 'style', 'div', 'span'
);
function DocumentParser($xml) {
//$this->count = 0;
$this->xml = $xml;
$this->CreateArray($this->xml);
$this->nodes = $this->HandleStructArray($this->elements);
}
public function GetElementsByTagName($tag) {
$test = $this->count;
$this->searchIndex++;
$this->searchNodes[$this->searchIndex] = array();
$this->TraverseSearch('name', $tag);
return $this->searchNodes[$this->searchIndex];
}
public function GetElementsByID($id) {
$test = $this->count;
$this->searchIndex++;
$this->searchNodes[$this->searchIndex] = array();
$this->TraverseSearch(array('attributes', 'id'), $id);
return $this->searchNodes[$this->searchIndex];
}
public function GetElementsByAttribute($attr, $attrVal) {
$test = $this->count;
$this->searchIndex++;
$this->searchNodes[$this->searchIndex] = array();
$this->TraverseSearch(array('attributes', $attr), $attrVal);
return $this->searchNodes[$this->searchIndex];
}
public function &GetNodes() {
return $this->nodes;
}
public function ToString() {
$this->xmlString = '';
$this->GenerateXML();
$this->xmlString = '' . trim($this->xmlString);
return $this->xmlString;
}
public function Unlink($killNode, $nodeGroup=array()) {
if(empty($nodeGroup)) {
$nodeGroup = &$this->nodes;
}
foreach($nodeGroup as $key=>$val) {
if($val == $killNode) {
if($val->HasChildren()) {
$this->Unlink($killNode, $val);
}
}
}
}
public function GenerateXML($node=array(), $tabs=0) {
$spaces = $tabs * 2;
if(empty($node)) {
$node = &$this->nodes;
}
foreach($node as $nodekey=>$nodeval) {
if($nodeval->GetTag() == 'text') {
$content = trim($nodeval->GetContent());
$content = wordwrap($content, 75 - $spaces);
$content = explode("\n", $content);
foreach($content as $contentkey=>$contentval) {
$this->xmlString .= str_repeat(" ", $spaces) . $contentval . "\n";
}
} else if($nodeval->GetTag() == 'kill') {
//SKIP!
} else if($nodeval->GetTag() == 'placeholder') {
if($nodeval->HasChildren()) {
$this->GenerateXML($nodeval->GetChildren(), $tabs);
}
} else if($nodeval->GetTag() == 'ieblock') {
$this->xmlString .= str_repeat(" ", $spaces) . "\n";
}
else {
$line = str_repeat(" ", $spaces) . "<" . $nodeval->GetTag();
if($nodeval->HasAttributes()) {
$attributes = $nodeval->GetAttributes();
foreach($attributes as $attrkey=>$attrval) {
$line .= " " . $attrkey . "=" . '"' . $attrval . '"';
}
}
if($nodeval->HasChildren()) {
$line .= ">" . "\n";
$this->xmlString .= $line;
$tabs++;
$this->GenerateXML($nodeval->GetChildren(), $tabs);
$tabs--;
$this->xmlString .= str_repeat(" ", $spaces) . "GetTag() . "> \n";
} else if(in_array($nodeval->GetTag(), $this->NeedStrictClose)) {
$this->xmlString .= $line . ">GetTag() . ">" . "\n";
}
else {
$this->xmlString .= $line . " />" . "\n";
}
}
}
}
//This function converts nodes to node objects
private function HandleStructArray($struct, $parentNode='') {
foreach($struct as $shallowArray) {
$index = count($node);
$node[] = new DocumentNode();
$node[$index]->SetTag($shallowArray['name']);
$node[$index]->SetIdentifier(microtime());
if($parentNode) {
$node[$index]->SetParent($parentNode);
}
if($shallowArray['content']) {
$node[$index]->SetContent($shallowArray['content']);
}
if(is_array($shallowArray['attributes'])) {
$node[$index]->SetAttributes($shallowArray['attributes']);
}
if($shallowArray['children'] && !empty($shallowArray['children'])) {
$node[$index]->AppendChildren($this->HandleStructArray($shallowArray['children'], &$node[$index]));
}
}
return $node;
}
private function CreateArray($xml) {
$xml = str_replace('&', '&', $xml);
$parser = xml_parser_create();
xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
xml_parse_into_struct($parser, $xml, $tags);
xml_parser_free($parser);
$elements = array();
$stack = array();
foreach ($tags as $tag) {
$index = count( $elements );
if($tag['type'] == "complete" || $tag['type'] == "open") {
$elements[$index] = array();
$elements[$index]['name'] = $tag['tag'];
$elements[$index]['attributes'] = empty($tag['attributes']) ? "" : $tag['attributes'];
if($tag['type'] == "open") {
$stack[count($stack)] = &$elements;
if($tag['value']) {
$elements[$index]['children'][0]['name'] = "text";
$elements[$index]['children'][0]['content'] = $tag['value'];
}
$elements = &$elements[$index]['children'];
} else if($tag['value']) {
$elements[$index]['children'] = array();
$elements[$index]['children'][0]['name'] = "text";
$elements[$index]['children'][0]['content'] = $tag['value'];
}
}
else if($tag['type'] == "cdata") {
$elements[$index] = array();
$elements[$index]['name'] = "text";
$elements[$index]['content'] = $tag['value'];
}
else if($tag['type'] == "close") {
$elements = &$stack[count($stack) - 1];
unset($stack[count($stack) - 1]);
} else {
die("ERROR: " .$tag['type'] . " NOT RECOGNIZED");
}
}
$this->elements = $elements;
}
private function TraverseSearch($searchKey, $searchValue, $node=array()) {
if(is_array($searchKey)) {
$attributes = true;
}
if(empty($node)) {
$node = &$this->nodes;
}
foreach($node as $nodekey=>$nodeval) {
if(is_object($nodeval)) {
if($nodeval->HasAttributes() && $attributes) {
if($node[$nodekey]->GetAttribute($searchKey[1])) {
$this->searchNodes[$this->searchIndex][] = &$node[$nodekey];
}
}
else if($searchKey == 'name' && $searchValue == $node[$nodekey]->GetTag($searchValue)) {
$this->searchNodes[$this->searchIndex][] = &$node[$nodekey];
}
if($nodeval->HasChildren()) {
$this->TraverseSearch($searchKey, $searchValue, $nodeval->GetChildren());
}
} else {
//Not an object, lets see what happens
}
}
}
}
?>