Return to validate_xml.pl CVS log

Up to [LON-CAPA] / modules / damieng / clean_xml

Annotation of modules/damieng/clean_xml/validate_xml.pl, revision 1.1

1.1     ! damieng     1: #!/usr/bin/perl
        !             2: 
        !             3: # Validates a file or directory against loncapa.xsd with libxml2
        !             4: 
        !             5: use strict;
        !             6: use utf8;
        !             7: use warnings;
        !             8: 
        !             9: use File::Basename;
        !            10: use Try::Tiny;
        !            11: use XML::LibXML;
        !            12: 
        !            13: 
        !            14: binmode(STDOUT, ':encoding(UTF-8)');
        !            15: binmode(STDERR, ':encoding(UTF-8)');
        !            16: 
        !            17: if (scalar(@ARGV) != 1) {
        !            18:   print STDERR "Usage: perl validate_xml.pl file|directory\n";
        !            19:   exit(1);
        !            20: }
        !            21: 
        !            22: # find the command-line argument encoding
        !            23: use I18N::Langinfo qw(langinfo CODESET);
        !            24: my $codeset = langinfo(CODESET);
        !            25: use Encode qw(decode);
        !            26: @ARGV = map { decode $codeset, $_ } @ARGV;
        !            27: 
        !            28: my $pathname = "$ARGV[0]";
        !            29: 
        !            30: my $script_dir = dirname(__FILE__);
        !            31: my $xmlschema = XML::LibXML::Schema->new(location => $script_dir.'/loncapa.xsd');
        !            32: 
        !            33: if (-d "$pathname") {
        !            34:   validate_dir($pathname);
        !            35: } elsif (-f $pathname) {
        !            36:   validate_file($pathname);
        !            37: }
        !            38: 
        !            39: 
        !            40: # Validates a directory recursively, selecting only .(problem|exam|survey|html|library).xml files.
        !            41: sub validate_dir {
        !            42:   my ($dirpath) = @_;
        !            43:   
        !            44:   opendir (my $dh, $dirpath) or die $!;
        !            45:   while (my $entry = readdir($dh)) {
        !            46:     next if ($entry =~ m/^\./); # ignore entries starting with a period
        !            47:     my $pathname = $dirpath.'/'.$entry;
        !            48:     if (-d $pathname) {
        !            49:       validate_dir($pathname);
        !            50:     } elsif (-f $pathname) {
        !            51:       if ($pathname =~ /\.(problem|exam|survey|html?|library)\.xml$/) {
        !            52:         validate_file($pathname);
        !            53:       }
        !            54:     }
        !            55:   }
        !            56:   closedir($dh);
        !            57: }
        !            58: 
        !            59: # Validates a file against loncapa.xsd with libxml2
        !            60: sub validate_file {
        !            61:   my ($pathname) = @_;
        !            62:   
        !            63:   my $doc = XML::LibXML->load_xml(location => $pathname);
        !            64:   try {
        !            65:     $xmlschema->validate($doc);
        !            66:     print "$pathname is valid\n";
        !            67:   } catch {
        !            68:     $_ =~ s/%20/ /g;
        !            69:     print "$_\n";
        !            70:   }
        !            71: }