/**
* Will return an DOM object tree from the well formed XML.
*
* @param string $strXMLText
* @return CDataXMLDocument
*/
function __parse(&$strXMLText)
{
static $search = array(">", "<", "'", """, "&");
static $replace = array(">", "<", "'", '"', "&");
$oXMLDocument = new CDataXMLDocument();
// strip comments
$strXMLText =& CDataXML::__stripComments($strXMLText);
// stip the !doctype
// The DOCTYPE declaration can consists of an internal DTD in square brackets
$cnt = 0;
$strXMLText = preg_replace("%<\\!DOCTYPE[^\\[>]*\\[.*?\\]>%is", "", $strXMLText, -1, $cnt);
if ($cnt == 0) {
$strXMLText = preg_replace("%<\\!DOCTYPE[^>]*>%is", "", $strXMLText);
}
// get document version and encoding from header
preg_match_all("#<\\?(.*?)\\?>#i", $strXMLText, $arXMLHeader_tmp);
foreach ($arXMLHeader_tmp[0] as $strXMLHeader_tmp) {
preg_match_all("/([a-zA-Z:]+=\".*?\")/i", $strXMLHeader_tmp, $arXMLParam_tmp);
foreach ($arXMLParam_tmp[0] as $strXMLParam_tmp) {
if ($strXMLParam_tmp != '') {
$arXMLAttribute_tmp = explode("=\"", $strXMLParam_tmp);
if ($arXMLAttribute_tmp[0] == "version") {
$oXMLDocument->version = substr($arXMLAttribute_tmp[1], 0, strlen($arXMLAttribute_tmp[1]) - 1);
} elseif ($arXMLAttribute_tmp[0] == "encoding") {
$oXMLDocument->encoding = substr($arXMLAttribute_tmp[1], 0, strlen($arXMLAttribute_tmp[1]) - 1);
}
}
}
}
// strip header
$strXMLText =& preg_replace("#<\\?.*?\\?>#", "", $strXMLText);
$oXMLDocument->root =& $oXMLDocument->children;
/** @var CDataXMLNode $currentNode */
$currentNode =& $oXMLDocument;
$tok = strtok($strXMLText, "<");
$arTag = explode(">", $tok);
if (count($arTag) < 2) {
//There was whitespace before <, so make another try
$tok = strtok("<");
$arTag = explode(">", $tok);
if (count($arTag) < 2) {
//It's a broken XML
return false;
}
}
while ($tok !== false) {
$tagName = $arTag[0];
$tagContent = $arTag[1];
// find tag name with attributes
// check if it's an endtag </tagname>
if ($tagName[0] == "/") {
$tagName = substr($tagName, 1);
// strip out namespace; nameSpace:Name
if ($this->delete_ns) {
$colonPos = strpos($tagName, ":");
if ($colonPos > 0) {
$tagName = substr($tagName, $colonPos + 1);
}
}
if ($currentNode->name != $tagName) {
// Error parsing XML, unmatched tags $tagName
return false;
}
$currentNode = $currentNode->_parent;
// convert special chars
if (!$this->TrimWhiteSpace || trim($tagContent) != "") {
$currentNode->content = str_replace($search, $replace, $tagContent);
}
} elseif (strncmp($tagName, "![CDATA[", 8) === 0) {
//because cdata may contain > and < chars
//it is special processing needed
$cdata = "";
for ($i = 0, $c = count($arTag); $i < $c; $i++) {
$cdata .= $arTag[$i] . ">";
if (substr($cdata, -3) == "]]>") {
$tagContent = $arTag[$i + 1];
break;
}
}
if (substr($cdata, -3) != "]]>") {
$cdata = substr($cdata, 0, -1) . "<";
do {
$tok = strtok(">");
//unfortunatly strtok eats > followed by >
$cdata .= $tok . ">";
//util end of string or end of cdata found
} while ($tok !== false && substr($tok, -2) != "]]");
//$tagName = substr($tagName, 0, -1);
}
$cdataSection = substr($cdata, 8, -3);
// new CDATA node
$subNode = new CDataXMLNode();
$subNode->name = "cdata-section";
$subNode->content = $cdataSection;
$currentNode->children[] = $subNode;
$currentNode->content .= $subNode->content;
// convert special chars
if (!$this->TrimWhiteSpace || trim($tagContent) != "") {
$currentNode->content = str_replace($search, $replace, $tagContent);
}
} else {
// normal start tag
$firstSpaceEnd = strpos($tagName, " ");
$firstNewlineEnd = strpos($tagName, "\n");
if ($firstNewlineEnd != false) {
if ($firstSpaceEnd != false) {
$tagNameEnd = min($firstSpaceEnd, $firstNewlineEnd);
} else {
$tagNameEnd = $firstNewlineEnd;
}
} else {
if ($firstSpaceEnd != false) {
$tagNameEnd = $firstSpaceEnd;
} else {
$tagNameEnd = 0;
}
}
if ($tagNameEnd > 0) {
$justName = substr($tagName, 0, $tagNameEnd);
} else {
$justName = $tagName;
}
// strip out namespace; nameSpace:Name
if ($this->delete_ns) {
$colonPos = strpos($justName, ":");
if ($colonPos > 0) {
$justName = substr($justName, $colonPos + 1);
}
}
// remove trailing / from the name if exists
$justName = rtrim($justName, "/");
$subNode = new CDataXMLNode();
$subNode->_parent = $currentNode;
$subNode->name = $justName;
// find attributes
if ($tagNameEnd > 0) {
$attributePart = substr($tagName, $tagNameEnd);
// attributes
unset($attr);
$attr = CDataXML::__parseAttributes($attributePart);
if ($attr != false) {
$subNode->attributes = $attr;
}
}
// convert special chars
if (!$this->TrimWhiteSpace || trim($tagContent) != "") {
$subNode->content = str_replace($search, $replace, $tagContent);
}
$currentNode->children[] = $subNode;
if (substr($tagName, -1) != "/") {
$currentNode = $subNode;
}
}
//Next iteration
$tok = strtok("<");
$arTag = explode(">", $tok);
//There was whitespace before < just after CDATA section, so make another try
if (count($arTag) < 2 && strncmp($tagName, "![CDATA[", 8) === 0) {
$currentNode->content .= $arTag[0];
// convert special chars
if (!$this->TrimWhiteSpace || trim($tagContent) != "") {
$currentNode->content = str_replace($search, $replace, $tagContent);
}
$tok = strtok("<");
$arTag = explode(">", $tok);
}
}
return $oXMLDocument;
}
