public function parseXML($data, $params = array(), $fetchData = false, $silent = false)
{
# this supports <? > and <! ... > ... tags as auto-closure comment tags
if (!isset($params[C_XML_RAW])) {
$params[C_XML_RAW] = false;
}
# Forces XHTML mode if not specified
if (!isset($params[C_XML_AUTOPARSE])) {
$params[C_XML_AUTOPARSE] = $fetchData;
} else {
if ($fetchData) {
$params[C_XML_AUTOPARSE] = true;
}
}
# we must parse the parameters to get ids, classes and links to the $parsedContent
if (!isset($params[C_XML_LAX])) {
$params[C_XML_LAX] = false;
}
# LAX mode will ignore auto-close tags that are not closed, and will automatically fix incorrectly opened/closed tags
if ($params[C_XML_RAW]) {
$fetchData = false;
# pointless with XML
}
if ($fetchData) {
$this->parsedContent = array(C_XHTML_IDS => array(), C_XHTML_CLASSES => array(), C_XHTML_DUPLICATES => array(), C_XHTML_LINKS => array(), C_XHTML_SRCS => array(), C_XHTML_NAMES => array());
}
$autoparse = $params[C_XML_AUTOPARSE];
# will parse the tag parameters to an array
$laxClosure = $params[C_XML_LAX];
# Ignore errors in HTML, such as incorrect open/close tags
$keepcase = isset($params[C_XML_KEEPCASE]);
$isXHTML = false;
# The last tag to be closed had XHTML content (false means only raw text)
$emptyParams = $autoparse ? array() : "";
$pos = 0;
# position in file
$buffer = "";
# what is being processed
$data = str_replace("\r", "\n", $data);
# remove \r, we don't need them (Windows \n\r)
$data = str_replace("\n\n", "\n", $data);
# now remove double \n caused by \r removal (or redundant \n anyway)
$total = strlen($data);
# html size
$intag = false;
# we are inside a < >
$intags = array();
# queue of nested tags
$incode = false;
# we are inside a code (C_XHTML_CODE)
$saida = new ttree();
# what we will generate as output
# nest result in this node (remember the XHTML will be nested on this)
$saida->addbranch(array(0 => $params[C_XML_RAW] ? "xml" : "xhtml", 1 => $emptyParams, 2 => ""));
$branch =& $saida->lastsibling();
$closure = "";
# if inside a literal, which closure was used, " or '
$line = 1;
# current line for debug purposes
while ($pos < $total) {
$car = $data[$pos];
if ($car == "\n") {
$line++;
}
if ($intag) {
if ($car == ">" && $closure == "") {
$intag = false;
// CLOSURE -----------------------------------------------------------
if ($buffer[0] == "/") {
// closes, checking consistency
$tagClose = substr($buffer, 1);
$nextClose = count($intags) > 0 ? $intags[count($intags) - 1] : '';
$thisClose = strtoupper($tagClose);
if ($thisClose == $nextClose) {
// fine, back to previous node
array_pop($intags);
if (!$isXHTML && count($branch->branchs) == 1 && $branch->branchs[0]->data[2] != "" && $branch->branchs[0]->data[0] == "") {
# we had no XHTML inside this tag, so we can compact it!
$branch->data[2] = $branch->branchs[0]->data[2];
$branch->branchs = array();
$branch =& $branch->parent;
} else {
$branch =& $branch->parent;
}
$isXHTML = true;
} else {
if (!$laxClosure) {
// incorrect close tag, and we are running on strict mode
if (!$silent) {
echo "XML:Tag mismatch at {$buffer} expecting {$nextClose} @ line " . $line . " (parent was " . $branch->parent->data[0] . ")";
#echo $data;
}
return false;
} else {
// incorrect close tag, but we are allowed to auto-close it. If this closure is an EXTRA closure, this will cause an error
// we search if this tag exists to be closed, if so, close all of them, if not, ignore this closure
$located = false;
for ($tp = count($intags) - 2; $tp > -1; $tp--) {
if ($thisClose == $intags[$tp]) {
$located = true;
break;
}
}
if ($located) {
$isXHTML = true;
$autoClose = count($intags) - $tp;
# we will close all this tags
for ($c = 0; $c < $autoClose; $c++) {
array_pop($intags);
$branch =& $branch->parent;
}
}
}
}
if (isset($this->xhtmlcode[$thisClose])) {
$incode = false;
}
// Auto-closing tag ----------------------------------------------------------
} else {
if ($buffer[strlen($buffer) - 1] == "/" || $buffer[strlen($buffer) - 1] == "?" && $buffer[0] == "?") {
$buffer = substr($buffer, 0, strlen($buffer) - 1);
# remove auto close
$tag = explode(" ", $buffer);
$tag = $tag[0];
$utag = strtoupper($tag);
$buffer = trim(substr($buffer, strlen($tag) + 1));
$data_do_branch = array(0 => $keepcase ? $tag : $utag, 1 => $buffer, 2 => '');
if ($autoparse) {
$data_do_branch[1] = $this->parseparams($buffer);
if ($fetchData && $utag != "PARAM" && !$incode) {
$this->refreshData($data_do_branch[1]);
}
}
$branch->addbranch($data_do_branch);
// Opening TAG -------------------------------------------------------------
} else {
if ($buffer[0] == "!" || $buffer[0] == "?") {
// stand-alone comment tag, consider it auto-close but don't parse
$tag = explode(" ", $buffer);
$tag = $tag[0];
$utag = strtoupper($tag);
$buffer = substr($buffer, strlen($tag) + 1);
$data_do_branch = array(0 => $tag, 1 => $buffer, 2 => '');
$branch->addbranch($data_do_branch);
} else {
// separates parameters
$tag = explode(" ", $buffer);
$tag = $tag[0];
$utag = strtoupper($tag);
if (!$params[C_XML_RAW] && isset($this->xhtmlacl[$utag])) {
# this should be a auto-close tag!
if ($laxClosure) {
$buffer = trim(substr($buffer, strlen($tag) + 1));
$data_do_branch = array(0 => $keepcase ? $tag : $utag, 1 => $buffer, 2 => '');
if ($autoparse) {
$data_do_branch[1] = $this->parseparams($buffer);
if ($fetchData && $utag != "PARAM" && !$incode) {
$this->refreshData($data_do_branch[1]);
}
} else {
if ($buffer == "") {
$buffer = $emptyParams;
}
}
$branch->addbranch($data_do_branch);
} else {
if (!$silent) {
echo "XML: Auto-close tag not closed at line {$line}: {$tag}";
}
return false;
}
} else {
$buffer = trim(substr($buffer, strlen($tag) + 1));
array_push($intags, $utag);
$data_do_branch = array(0 => $keepcase ? $tag : $utag, 1 => $buffer, 2 => '');
if ($autoparse) {
$data_do_branch[1] = $this->parseparams($buffer);
if ($fetchData && !$incode) {
$this->refreshData($data_do_branch[1]);
}
} else {
if ($buffer == "") {
$buffer = $emptyParams;
}
}
$branch->addbranch($data_do_branch);
$branch =& $branch->lastsibling();
}
}
if (isset($this->xhtmlcode[$utag])) {
$incode = true;
}
$isXHTML = false;
}
}
// ----------------------------------------------------------------------
$buffer = "";
} else {
if ($car != "\n") {
if ($closure == "" && $car == "\t") {
$car = " ";
} else {
if ($closure == "" && ($car == "\"" || $car == "'")) {
$closure = $car;
} else {
if ($closure == $car) {
$closure = "";
}
}
}
$buffer .= $car;
} else {
$buffer .= " ";
// converts \r or \n to " " (\r was hard-converted to \n before)
}
}
} else {
if ($car == "<") {
if ($pos == $total - 1 || $data[$pos + 1] == ">" || preg_match("/^([\n\r\t ='\"<]+)\$/", $data[$pos + 1])) {
// do not count as a tag on these cases
$buffer .= "<";
} else {
if ($buffer != "" && !preg_match("/^([\n\r\t ]+)\$/", $buffer)) {
# ignore empty strings comprised only or these characters (do not add to xml)
$data_do_branch = array(0 => "", 1 => $emptyParams, 2 => preg_replace("/([\t ])+/", " ", $buffer));
$branch->addbranch($data_do_branch);
}
$buffer = "";
if (substr($data, $pos, 4) == "<!--" || substr($data, $pos, 2) == "<?") {
$end = substr($data, $pos, 4) == "<!--" ? strpos($data, "-->", $pos) + 3 : strpos($data, "?>", $pos) + 2;
$buffer .= substr($data, $pos, $end - $pos);
$pos = $end - 1;
if (isset($data[$pos + 1]) && $data[$pos + 1] == "\n") {
$buffer .= "\n";
$pos++;
}
if (!isset($params[C_XML_REMOVECOMMENTS]) || $incode) {
# only add comments if set NOT to remove or we are in code
$data_do_branch = array(0 => "", 1 => $emptyParams, 2 => $buffer);
$branch->addbranch($data_do_branch);
}
$buffer = "";
} else {
$intag = true;
}
$closure = "";
}
} else {
$buffer .= $car;
}
}
$pos++;
}
if ($buffer != "" && !preg_match("/^([\n\r\t ]+)\$/", $buffer)) {
# ignore empty strings comprised only or these characters
$data_do_branch = array(0 => "", 1 => $emptyParams, 2 => preg_replace("/([\t ])+/", " ", $buffer));
$branch->addbranch($data_do_branch);
}
if (!$laxClosure && count($intags) > 0 && !$silent) {
echo "XML: Tag not closed at " . array_pop($intags);
}
# this should never happen on LAX mode
return $saida;
}
