--- loncom/homework/cleanxml/xml_to_loncapa.pm	2015/12/23 20:33:10	1.3
+++ loncom/homework/cleanxml/xml_to_loncapa.pm	2016/01/20 00:41:31	1.7
@@ -1,7 +1,7 @@
 # The LearningOnline Network
 # convert_file takes a well-formed XML file content and converts it to LON-CAPA syntax.
 #
-# $Id: xml_to_loncapa.pm,v 1.3 2015/12/23 20:33:10 damieng Exp $
+# $Id: xml_to_loncapa.pm,v 1.7 2016/01/20 00:41:31 damieng Exp $
 #
 # Copyright Michigan State University Board of Trustees
 #
@@ -49,13 +49,21 @@ my @simple_data = ('polygon', 'rectangle
 
 my @inline_responses = ('stringresponse','optionresponse','numericalresponse','formularesponse','mathresponse','organicresponse','reactionresponse','customresponse','externalresponse');
 
+# see http://www.w3.org/TR/html-polyglot/#empty-elements
+# and http://tiffanybbrown.com/2011/03/23/html5-does-not-allow-self-closing-tags/
+# HTML elements that do not have an empty content, and must never use a self-closing tag:
+my @non_empty_html = ('title','style','script','noscript','body','section','header','footer','article','aside','nav','h1','h2','h3','h4','h5','h6','div','p','li','dt','dd','caption','td','th','span','a','em','strong','b','i','sup','sub','pre','code','kbd','samp','cite','q','tt','ins','del','var','small','big','address','blockquote','bdo','ruby','rb','rp','rt','rtc','figure','figcaption','object','applet','video','audio','canvas','label','option','textarea','fieldset','legend','button','iframe');
+
 
 # Converts a file and return the modified contents
 sub convert_file {
   my ($contents) = @_;
 
   my $dom_doc = XML::LibXML->load_xml(string => $contents);
-  add_outtext($dom_doc);
+  my $root = $dom_doc->documentElement();
+  if (defined $root && $root->nodeName ne 'html') {
+    add_outtext($dom_doc);
+  }
   return node_to_string($dom_doc);
 }
 
@@ -73,7 +81,7 @@ sub node_to_string {
     if (defined $parent->parentNode) {
       $grandparent_name = $parent->parentNode->nodeName;
     }
-    my @no_escape = ('m', 'script', 'display', 'parse', 'answer');
+    my @no_escape = ('m', 'script', 'style', 'display', 'parse', 'answer');
     if (string_in_array(\@no_escape, $parent_name) &&
         ($parent_name ne 'answer' ||
         (defined $grandparent_name &&
@@ -95,7 +103,7 @@ sub node_to_string {
       $s .= escape_attribute($attribute->nodeValue);
       $s .= '"';
     }
-    if ($node->hasChildNodes() || $tag eq 'script') {
+    if ($node->hasChildNodes() || string_in_array(\@non_empty_html, $tag)) {
       $s .= '>';
       foreach my $child ($node->childNodes) {
         $s .= node_to_string($child);
@@ -247,6 +255,21 @@ sub add_endouttext {
   } else {
     $parent->appendChild($endouttext);
   }
+  # replace spaces afterwards by a \n + indentation
+  my $next = $endouttext->nextSibling;
+  if (defined $next && $next->nodeType == XML_TEXT_NODE) {
+    my $v = $next->nodeValue;
+    if ($v =~ /^ /) {
+      $v =~ s/^ +//;
+      if ($parent->firstChild->nodeType == XML_TEXT_NODE &&
+          $parent->firstChild->nodeValue =~ /^\n +$/) {
+        $v = $parent->firstChild->nodeValue.$v;
+      } else {
+        $v = "\n".$v;
+      }
+      $next->setData($v);
+    }
+  }
 }
 
 # Convert paragraph children when one contains an inline response into content + <br>