--- loncom/homework/cleanxml/post_xml.pm	2016/01/21 22:09:38	1.9
+++ loncom/homework/cleanxml/post_xml.pm	2016/11/10 21:53:56	1.11
@@ -1,7 +1,7 @@
 # The LearningOnline Network
 # Third step to clean a file.
 #
-# $Id: post_xml.pm,v 1.9 2016/01/21 22:09:38 damieng Exp $
+# $Id: post_xml.pm,v 1.11 2016/11/10 21:53:56 damieng Exp $
 #
 # Copyright Michigan State University Board of Trustees
 #
@@ -136,6 +136,8 @@ sub post_xml {
   
   fix_empty_lc_elements($root);
   
+  reduce_empty_p($root);
+  
   lowercase_attribute_values($root);
   
   replace_numericalresponse_unit_attribute($root);
@@ -1856,6 +1858,18 @@ sub fix_paragraphs_inside {
           push(@new_children, $doc->createElement('p'));
         }
         $p = undef;
+        # ignore the next node if it is a br (the paragraph default margin will take as much space)
+        # (ignoring whitespace)
+        while (defined $next && $next->nodeType == XML_TEXT_NODE && $next->nodeValue =~ /^[ \t\f\n\r]*$/) {
+          my $next2 = $next->nextSibling;
+          $node->removeChild($next);
+          $next = $next2;
+        }
+        if (defined $next && $next->nodeType == XML_ELEMENT_NODE && $next->nodeName eq 'br') {
+          my $next2 = $next->nextSibling;
+          $node->removeChild($next);
+          $next = $next2;
+        }
       } elsif ($child->nodeType == XML_ELEMENT_NODE && string_in_array(\@inline_like_block, $child->nodeName)) {
         # inline_like_block: use the paragraph if there is one, otherwise do not create one
         if (defined $p) {
@@ -2032,6 +2046,20 @@ sub fix_paragraph {
               if (!defined $left || !$left_needs_p) {
                 $replacement->appendChild($middle);
               }
+              # ignore the next node if it is a br (the paragraph default margin will take as much space)
+              my $first_right;
+              if (defined $right) {
+                $first_right = $right->firstChild;
+                # ignore non-nbsp whitespace
+                while (defined $first_right && $first_right->nodeType == XML_TEXT_NODE &&
+                    $first_right->nodeValue =~ /^[ \t\f\n\r]*$/) {
+                  $first_right = $first_right->nextSibling;
+                }
+              }
+              if (defined $first_right && $first_right->nodeType == XML_ELEMENT_NODE &&
+                  $first_right->nodeName eq 'br') {
+                $right->removeChild($first_right);
+              }
             } else {
               fix_paragraphs_inside($n, $all_block);
               $replacement->appendChild($n);
@@ -2268,6 +2296,33 @@ sub fix_empty_lc_elements {
     }
   }
 }
+
+# remove consecutive empty paragraphs (they will not show anyway)
+sub reduce_empty_p {
+  my ($node) = @_;
+  my $next;
+  for (my $child=$node->firstChild; defined $child; $child=$next) {
+    $next = $child->nextSibling;
+    while (defined $next && $next->nodeType == XML_TEXT_NODE && $next->nodeValue =~ /^[ \t\f\n\r]*$/) {
+      $next = $next->nextSibling;
+    }
+    if ($child->nodeType == XML_ELEMENT_NODE && $child->nodeName eq 'p' && defined $next &&
+        $next->nodeType == XML_ELEMENT_NODE && $next->nodeName eq 'p') {
+      my $first = $child->firstChild;
+      if (!defined $first || (!defined $first->nextSibling &&
+          $first->nodeType == XML_TEXT_NODE && $first->nodeValue =~ /^[ \t\f\n\r]*$/)) {
+        $first = $next->firstChild;
+        if (!defined $first || (!defined $first->nextSibling &&
+            $first->nodeType == XML_TEXT_NODE && $first->nodeValue =~ /^[ \t\f\n\r]*$/)) {
+          $node->removeChild($child);
+        }
+      }
+    }
+    if ($child->nodeType == XML_ELEMENT_NODE) {
+      reduce_empty_p($child);
+    }
+  }
+}
 
 # turn some attribute values into lowercase when they should be
 sub lowercase_attribute_values {