[LON-CAPA-cvs] cvs: loncom /homework/cleanxml post_xml.pm
damieng
damieng at source.lon-capa.org
Thu Jan 14 11:28:06 EST 2016
damieng Thu Jan 14 16:28:06 2016 EDT
Modified files:
/loncom/homework/cleanxml post_xml.pm
Log:
fixed calls to html_to_xml and tth
Index: loncom/homework/cleanxml/post_xml.pm
diff -u loncom/homework/cleanxml/post_xml.pm:1.6 loncom/homework/cleanxml/post_xml.pm:1.7
--- loncom/homework/cleanxml/post_xml.pm:1.6 Fri Jan 8 20:32:42 2016
+++ loncom/homework/cleanxml/post_xml.pm Thu Jan 14 16:28:06 2016
@@ -1,7 +1,7 @@
# The LearningOnline Network
# Third step to clean a file.
#
-# $Id: post_xml.pm,v 1.6 2016/01/08 20:32:42 damieng Exp $
+# $Id: post_xml.pm,v 1.7 2016/01/14 16:28:06 damieng Exp $
#
# Copyright Michigan State University Board of Trustees
#
@@ -41,6 +41,8 @@
use XML::LibXML;
use HTML::TokeParser; # used to parse sty files
use Tie::IxHash; # for ordered hashes
+use tth;
+use Apache::html_to_xml;
no warnings 'recursion'; # yes, fix_paragraph is using heavy recursion, I know
@@ -552,11 +554,11 @@
# Returns the HTML equivalent of LaTeX input, using tth
sub tth {
my ($text) = @_;
- my ($fh, $tmp_path) = tempfile();
- binmode($fh, ':utf8');
- print $fh $text;
- close $fh;
- my $output = `tth -r -w2 -u -y0 < $tmp_path 2>/dev/null`;
+ my $output = &tth::tth($text);
+ my $errorstring = &tth::ttherror();
+ if ($errorstring) {
+ die $errorstring;
+ }
# hopefully the temp file will not be removed before this point (otherwise we should use unlink_on_destroy 0)
$output =~ s/^\s*|\s*$//;
$output =~ s/<div class="p"><!----><\/div>/<br\/>/; # why is tth using such ugly markup for \newline ?
@@ -567,7 +569,7 @@
sub html_to_dom {
my ($text) = @_;
$text = '<root>'.$text.'</root>';
- 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;
More information about the LON-CAPA-cvs
mailing list