<\/div>/
/; # why is tth using such ugly markup for \newline ?
@@ -567,7 +601,7 @@ sub tth {
sub html_to_dom {
my ($text) = @_;
$text = '
'.$text.'';
- my $textref = html_to_xml::html_to_xml(\$text);
+ my $textref = Apache::html_to_xml::html_to_xml(\$text);
utf8::upgrade($$textref); # otherwise the XML parser fails when the HTML parser turns into a character
my $dom_doc = XML::LibXML->load_xml(string => $textref);
my $root = $dom_doc->documentElement;
@@ -1808,12 +1842,31 @@ sub remove_useless_notsolved {
}
}
+# Use
for multi-line comments without elements.
+sub fix_comments {
+ my ($root) = @_;
+ my $doc = $root->ownerDocument;
+ my @comments = $root->getElementsByTagName('comment');
+ foreach my $comment (@comments) {
+ my $first = $comment->firstChild;
+ if (defined $first) {
+ if ($first->nodeType == XML_TEXT_NODE && $first->nodeValue =~ /\n/ &&
+ !defined $first->nextSibling) {
+ my $pre = $doc->createElement('pre');
+ $comment->removeChild($first);
+ $comment->appendChild($pre);
+ $pre->appendChild($first);
+ }
+ }
+ }
+}
+
# adds a paragraph inside if needed and calls fix_paragraph for all paragraphs (including new ones)
sub fix_paragraphs_inside {
my ($node, $all_block) = @_;
# blocks in which paragrahs will be added:
- my @blocks_with_p = ('loncapa','library','problem','part','problemtype','window','block','while','postanswerdate','preduedate','solved','notsolved','languageblock','translated','lang','instructorcomment','togglebox','standalone','form');
- my @fix_p_if_br_or_p = (@responses,'foil','item','text','label','hintgroup','hintpart','hint','web','windowlink','div','li','dd','td','th','blockquote');
+ my @blocks_with_p = ('loncapa','library','problem','part','problemtype','window','block','while','postanswerdate','preduedate','languageblock','instructorcomment','togglebox','standalone','body','form');
+ my @fix_p_if_br_or_p = (@responses,'foil','item','text','label','hintgroup','hintpart','hint','web','windowlink','div','li','dd','td','th','blockquote','solved','notsolved');
if ((string_in_array(\@blocks_with_p, $node->nodeName) && paragraph_needed($node)) ||
(string_in_array(\@fix_p_if_br_or_p, $node->nodeName) && paragraph_inside($node))) {
# if non-empty, add paragraphs where needed between all br and remove br
@@ -1833,6 +1886,18 @@ 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) {
@@ -2009,6 +2074,20 @@ 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);
@@ -2246,6 +2325,33 @@ 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 {
my ($root) = @_;
@@ -2394,8 +2500,29 @@ sub pretty {
my $type = $node->nodeType;
if ($type == XML_ELEMENT_NODE) {
my $name = $node->nodeName;
- if ((string_in_array($all_block, $name) || string_in_array(\@inline_like_block, $name)) &&
- !string_in_array(\@preserve_elements, $name)) {
+ if (string_in_array(\@preserve_elements, $name)) {
+ # remove newlines at the beginning and the end of preserve elements
+ if (defined $node->firstChild && ($node->firstChild->nodeType == XML_TEXT_NODE ||
+ $node->firstChild->nodeType == XML_CDATA_SECTION_NODE)) {
+ my $text = $node->firstChild->nodeValue;
+ $text =~ s/^\n+//;
+ if ($text eq '') {
+ $node->removeChild($node->firstChild);
+ } else {
+ $node->firstChild->setData($text);
+ }
+ }
+ if (defined $node->lastChild && ($node->lastChild->nodeType == XML_TEXT_NODE ||
+ $node->lastChild->nodeType == XML_CDATA_SECTION_NODE)) {
+ my $text = $node->lastChild->nodeValue;
+ $text =~ s/\n+$//;
+ if ($text eq '') {
+ $node->removeChild($node->lastChild);
+ } else {
+ $node->lastChild->setData($text);
+ }
+ }
+ } elsif (string_in_array($all_block, $name) || string_in_array(\@inline_like_block, $name)) {
# make sure there is a newline at the beginning and at the end if there is anything inside
if (defined $node->firstChild && !string_in_array(\@no_newline_inside, $name)) {
my $first = $node->firstChild;
@@ -2478,26 +2605,6 @@ sub pretty {
if ($text eq '') {
$node->removeChild($node->lastChild);
} else {
- $node->lastChild->setData($text);
- }
- }
- } elsif (string_in_array(\@preserve_elements, $name)) {
- # collapse newlines at the beginning and the end of scripts
- if (defined $node->firstChild && $node->firstChild->nodeType == XML_TEXT_NODE) {
- my $text = $node->firstChild->nodeValue;
- $text =~ s/^\n( *\n)+/\n/;
- if ($text eq '') {
- $node->removeChild($node->firstChild);
- } else {
- $node->firstChild->setData($text);
- }
- }
- if (defined $node->lastChild && $node->lastChild->nodeType == XML_TEXT_NODE) {
- my $text = $node->lastChild->nodeValue;
- $text =~ s/\n( *\n)+$/\n/;
- if ($text eq '') {
- $node->removeChild($node->lastChild);
- } else {
$node->lastChild->setData($text);
}
}