function wrsqz_extractTextFromMathML($formula, $encoded=true){
//Algorythm: We scan the mathML tag by tag.
//If a tag is one of the allowed (math, mrow) we save it at the stack
//and continue with the next.
//If the tag is not allowed (mfenced, mfrac,...) we skip all mathML until its
//closure (</mfenced>, </mfrac>)
//If the tag is <mtext> we rearange the formula
//If a tag is a closure of allowed tag, we pop it from the stack.
//rapid return if nothing to do.
if(strpos($formula,'mtext')===false) return $formula;
//initializations
$opentag = $encoded ? '«' : '<';
$closetag = $encoded ? '»' : '>';
//tags where an <mtext> can live inside.
$allowedtags = array('math', 'mrow');
$pattern = $opentag.'([^'.$opentag.$closetag.']*)'.$closetag; //regexp that matches a single tag label
mb_ereg_search_init($formula, $pattern);
$stack = array(); //stack of opened tags
$omittedcontent=false; //there is math content before the current point?
$lasttag=null; //last tag of the stack
$length = strlen($formula);
$beginformula = strpos($formula, $opentag); //position of the first character of the last formula (in bytes).
$pos=array(0,0);
//CAUTION: If you change this function, be very carefull with multibyte
// and non-multibyte functions.
while(($pos[0]+$pos[1])<$length){
$pos = mb_ereg_search_pos($pattern);
if($pos[0]+$pos[1] < $length){
//this will be always true but the last iteration
mb_ereg_search_setpos($pos[0]+$pos[1]);
}
$tag = substr($formula, $pos[0],$pos[1]);
$trimmedTag = mb_substr($tag,1,-1);
//skip autoclosed tags
if(mb_substr($trimmedTag,-1) == '/'){
continue;
}
//discard attributes
if(($spacepos = mb_strpos($trimmedTag,' '))!==false){
$trimmedTag=mb_substr($trimmedTag,0,$spacepos);
}
if(in_array($trimmedTag,$allowedtags)){
//allowed tag
$stack[]=array($trimmedTag,$tag);
$lasttag = $trimmedTag;
}else if($trimmedTag == '/'.$lasttag){
//close allowed tag
array_pop($stack);
$lasttag = end($stack);
$lasttag = $lasttag[0];
//discard empty formulas
if(empty($stack) && !$omittedcontent){
$formula1 = substr($formula, 0, $beginformula);
if($pos[0]+$pos[1]<$length){
//this isn't the end.
$formula2 = substr($formula, $pos[0]+$pos[1]);
$formula = $formula1 . $formula2;
$length = strlen($formula);
mb_ereg_search_init($formula, $pattern);
mb_ereg_search_setpos($beginformula);
}else{
//this is the last iteration. $length and mb_ereg_search
//string and position will be wrong, but it doesn't matter.
$formula = $formula1;
}
}
}else if($trimmedTag == 'mtext'){
$pos2 = mb_ereg_search_pos($opentag.'/mtext'.$closetag);
$text = substr($formula, $pos[0]+$pos[1], $pos2[0]-($pos[0]+$pos[1]));
//Decode some chars in text
if($encoded) $text=wrsqz_mathmlDecode($text);
$text = str_replace('·','·',$text);
$text = str_replace(''',''',$text);
$formula1 = substr($formula, 0, $pos[0]); //until <mtext>
$formula2 = substr($formula, $pos2[0]+$pos2[1]); //from </mtext>
if($omittedcontent){
//we have a non-empty formula before the text so we must close it
//compute the tail (close tags) of the formula before the text
//and the head (open tags) of the formula after the text.
$copystack = $stack; //copy stack
$tail1 = '';
$head2 = '';
while($stacktag = array_pop($copystack)){
$tail1.= $opentag.'/'.$stacktag[0].$closetag;
$head2 = $stacktag[1] . $head2;
}
$formula1 = $formula1 . $tail1;
$formula2 = $head2 . $formula2;
//update $formula
$formula = $formula1 . $text . $formula2;
$beginformula = $pos[0]+strlen($tail1)+strlen($text);
$position = $beginformula+strlen($head2);
}else{
//we have an empty formula before the text so we must skip it.
$head = substr($formula1, 0, $beginformula); //all before the empty formula
$formula1 = substr($formula1, $beginformula);
$formula = $head . $text . $formula1 . $formula2;
$beginformula += strlen($text);
$position = $beginformula +strlen($formula1);
}
//update parameters with the new formula.
$length = strlen($formula);
$omittedcontent = false;
mb_ereg_search_init($formula, $pattern);
mb_ereg_search_setpos($position);
}else{
//not allowed tag: go to its closure and remember that we omitted content
$pos = mb_ereg_search_pos($opentag.'/'.$trimmedTag.$closetag);
if($pos === false){
return $formula; //this is an error in XML (unclosed tag);
}
$omittedcontent=true;
mb_ereg_search_setpos($pos[0]+$pos[1]);
}
}
return $formula;
}
function wrsqz_import_from_xml($questionType, $dbType, &$data, &$question, &$format, &$extra) {
if (isset($data['@']['type']) && $data['@']['type'] == $questionType . 'wiris') {
$qfxml = new qformat_xml();
if ($questionType == 'essay') {
$question = $qfxml->import_essay($data);
}else if ($questionType == 'match') {
$question = $qfxml->import_matching($data);
}else if ($questionType == 'multichoice') {
$question = $qfxml->import_multichoice($data);
}else if ($questionType == 'truefalse') {
$question = $qfxml->import_truefalse($data);
}else if ($questionType == 'shortanswer') {
$question = $qfxml->import_shortanswer($data);
}else if ($questionType == 'multianswer') {
$questiontext = $data['#']['wirisquestiontext'][0]['#']['text'];
$container = new stdClass;
$container->questiontext = $qfxml->import_text($questiontext);
$question = wrsqz_qtype_multianswer_extract_question($container);
$question->course = $qfxml->course;
$question->generalfeedback = $qfxml->getpath($data, array('#','generalfeedback',0,'#','text',0,'#'), '', true );
$question->name = $qfxml->import_text( $data['#']['name'][0]['#']['text'] );
}
unset($qfxml);
$question->qtype = $questionType . 'wiris';
$program = wrsqz_mathmlDecode($format->getpath($data, array('#', 'wirisquestion', 0, '#'), 0));
$imported = wrsqz_importCASSession($program);
if($imported===false) return false;
$question->hiddenCASValue = $imported[0];
$translation = $imported[1];
if ($questionType == 'shortanswer') {
mb_parse_str($format->getpath($data, array('#', 'wiriseditor', 0, '#'), 0), $eqoptionArray);
$question->wirisEditor = (isset($eqoptionArray['editor']) && $eqoptionArray['editor'] == 'true') ? 'editor=true' : '';
$question->multipleAnswers = (isset($eqoptionArray['multipleAnswers']) && $eqoptionArray['multipleAnswers'] == 'true') ? 'multipleAnswers=true' : '';
$question->wirisCASForComputations = (isset($eqoptionArray['wirisCASForComputations']) && $eqoptionArray['wirisCASForComputations'] == 'true') ? 'wirisCASForComputations=true' : '';
if(isset($eqoptionArray['testFunctionName'])){
foreach ($eqoptionArray['testFunctionName'] as $index => $functionName){
if(!empty($functionName)){
$question->testFunctionName[$index]=$functionName;
}
}
}
}else if ($questionType == 'multichoice') {
$question->gradeOverride = explode(';', $format->getpath($data, array('#', 'wirisoverrideanswer', 0, '#'), 0));
}else if ($questionType == 'truefalse') {
$question->wirisAnswer = $format->getpath($data, array('#', 'wirisoverrideanswer', 0, '#'), 0);
}else if ($questionType == 'multianswer'){
//mb_parse_str($format->getpath($data, array('#', 'wiriseditor', 0, '#'), 0), $eqoptionArray);
//$question->wirisEditor = (isset($eqoptionArray['editor']) && $eqoptionArray['editor'] == 'true') ? 'editor=true' : '';
}
$options = $format->getpath($data, array('#', 'wirisoptions',0,'#'), array());
$question->wirisCASForComputations = $format->getpath($options, array('wirisCASForComputations',0,'#'),0);
//TODO: translate cas if needed.
$question->hiddenInitialCASValue = wrsqz_mathmlDecode($format->getpath($options, array('hiddenInitialCASValue',0,'#'),0));
if(!empty($translation)){
wrsqz_replaceVarReferencesInFlatQuestion($questionType, $question, $translation);
}
return $question;
}
return false;
}
