--- modules/damieng/clean_xml/xml_to_loncapa.pl	2015/04/29 19:26:05	1.1
+++ modules/damieng/clean_xml/xml_to_loncapa.pl	2015/05/16 20:17:13	1.2
@@ -8,6 +8,16 @@ use warnings;
 
 use XML::LibXML;
 
+
+my @loncapa_block = ('parameter','location','answer','foil','image','polygon','rectangle','text','conceptgroup','itemgroup','item','label','data','function','array','unit','answergroup','functionplotresponse','functionplotruleset','functionplotelements','functionplotcustomrule','essayresponse','hintpart','formulahint','numericalhint','reactionhint','organichint','optionhint','radiobuttonhint','stringhint','customhint','mathhint','formulahintcondition','numericalhintcondition','reactionhintcondition','organichintcondition','optionhintcondition','radiobuttonhintcondition','stringhintcondition','customhintcondition','mathhintcondition','imageresponse','foilgroup','datasubmission','textfield','hiddensubmission','radiobuttonresponse','rankresponse','matchresponse','import','style','script','window','block','library','notsolved','part','postanswerdate','preduedate','problem','problemtype','randomlabel','bgimg','labelgroup','randomlist','solved','while','tex','print','web','gnuplot','curve','Task','IntroParagraph','ClosingParagraph','Question','QuestionText','Setup','Instance','InstanceText','Criteria','CriteriaText','GraderNote','languageblock','translated','lang','instructorcomment','dataresponse','togglebox','standalone','comment','drawimage','allow','displayduedate','displaytitle','responseparam','organicstructure','scriptlib','parserlib','drawoptionlist','spline','backgroundplot','plotobject','plotvector','drawvectorsum','functionplotrule','functionplotvectorrule','functionplotvectorsumrule','axis','key','xtics','ytics','title','xlabel','ylabel','hiddenline','dtm','stringresponse','optionresponse','numericalresponse','formularesponse','mathresponse','organicresponse','reactionresponse','customresponse','externalresponse', 'hint', 'hintgroup');
+
+my @loncapa_inline = ('display','m','lm','chem','num','parse','algebra','displayweight','displaystudentphoto'); # not textline
+
+my @html_elements = ('html','meta','head','title','base','link','style','noscript','body','header','footer','aside','h1','h2','h3','h4','h5','h6','li','dd','dt','tbody','tr','caption','thead','tfoot','td','th','span','a','em','strong','b','i','sup','sub','pre','code','kbd','samp','cite','q','tt','ins','del','var','small','big','br','hr','address','blockquote','img','figure','figcaption','object','param','embed','applet','video','source','audio','map','area','canvas','form','label','input','select','optgroup','option','textarea','fieldset','legend','button','iframe','section','div','p','ul','ol','dl','table'); # without script
+
+my @simple_data = ('polygon', 'rectangle', 'vector', 'value', 'answer', 'title', 'xlabel', 'ylabel', 'tic', 'parserlib', 'scriptlib', 'import', 'tex', 'text', 'image', 'display', 'm', 'lm', 'num', 'algebra', 'chem', 'parse', 'title', 'style', 'script', 'ins', 'del', 'label', 'option', 'textarea', 'legend' );
+
+
 binmode(STDOUT, ':encoding(UTF-8)');
 
 if (scalar(@ARGV) != 1) {
@@ -39,6 +49,7 @@ sub convert_file {
   my $dom_doc = XML::LibXML->load_xml(location => $pathname);
   
   open my $out, '>:encoding(UTF-8)', $newpath;
+  add_outtext($dom_doc);
   print $out node_to_string($dom_doc);
   close $out;
 }
@@ -103,6 +114,123 @@ sub escape {
   return $s;
 }
 
+# Adds startouttext and endouttext where useful for the colorfull editor
+sub add_outtext {
+  my ($node) = @_;
+  
+  if ($node->nodeType == XML_DOCUMENT_NODE) {
+    my $root = $node->documentElement();
+    add_outtext($root);
+    return;
+  }
+  if ($node->nodeType != XML_ELEMENT_NODE) {
+    return;
+  }
+  if (string_in_array(\@simple_data, $node->nodeName)) {
+    return;
+  }
+  my $next;
+  my $in_outtext = 0;
+  for (my $child=$node->firstChild; defined $child; $child=$next) {
+    $next = $child->nextSibling;
+    if (!$in_outtext && inside_outtext($child)) {
+      # Add startouttext
+      my $doc = $node->ownerDocument;
+      my $startouttext = $doc->createElement('startouttext');
+      $node->insertBefore($startouttext, $child);
+      $in_outtext = 1;
+    } elsif ($in_outtext && !continue_outtext($child)) {
+      # Add endouttext
+      add_endouttext($node, $child);
+      $in_outtext = 0;
+    }
+    if (!$in_outtext) {
+      add_outtext($child);
+    }
+  }
+  if ($in_outtext) {
+    # Add endouttext
+    add_endouttext($node);
+  }
+}
+
+# Returns 1 if this node should trigger the addition of startouttext before it
+sub inside_outtext {
+  my ($node) = @_;
+  if ($node->nodeType == XML_TEXT_NODE && $node->nodeValue !~ /^\s*$/) {
+    return 1;
+  }
+  if ($node->nodeType == XML_ELEMENT_NODE && string_in_array(\@html_elements, $node->nodeName)) {
+    if (contains_loncapa_block($node)) {
+      return 0;
+    } else {
+      return 1;
+    }
+  }
+  if ($node->nodeType == XML_ELEMENT_NODE && string_in_array(\@loncapa_inline, $node->nodeName)) {
+    return 1;
+  }
+  return 0;
+}
+
+# Returns 1 if the outtext environment can continue with this node
+sub continue_outtext {
+  my ($node) = @_;
+  if (inside_outtext($node)) {
+    return 1;
+  }
+  if ($node->nodeType == XML_TEXT_NODE) {
+    return 1; # continue even if this is just spaces
+  }
+  return 0;
+}
+
+# Returns 1 if the node contains a LON-CAPA block in a descendant.
+sub contains_loncapa_block {
+  my ($node) = @_;
+  foreach my $child ($node->childNodes) {
+    if ($child->nodeType == XML_ELEMENT_NODE) {
+      if (string_in_array(\@loncapa_block, $child->nodeName)) {
+        return 1;
+      }
+      if (contains_loncapa_block($child)) {
+        return 1;
+      }
+    }
+  }
+  return 0;
+}
+
+sub add_endouttext {
+  my ($parent, $before_node) = @_;
+  my $doc = $parent->ownerDocument;
+  my $endouttext = $doc->createElement('endouttext');
+  my $before_before;
+  if (defined $before_node) {
+    $before_before = $before_node->previousSibling;
+  } else {
+    $before_before = $parent->lastChild;
+  }
+  if (defined $before_before && $before_before->nodeType == XML_TEXT_NODE) {
+    # split space at the end of the node before endouttext
+    if ($before_before->nodeValue =~ /^(.*?)(\s+)$/s) {
+      $before_before->setData($1);
+      my $space_node = $doc->createTextNode($2);
+      if (defined $before_node) {
+        $parent->insertBefore($space_node, $before_node);
+      } else {
+        $parent->appendChild($space_node);
+      }
+      $before_node = $space_node;
+    }
+  }
+  if (defined $before_node) {
+    $parent->insertBefore($endouttext, $before_node);
+  } else {
+    $parent->appendChild($endouttext);
+  }
+}
+
 ##
 # Tests if a string is in an array (using eq) (to avoid Smartmatch warnings with $value ~~ @array)
 # @param {Array<string>} array - reference to the array of strings