HTML::TreeBuilder, HTML::Element, as_XML()

[AlexTape]

Dear omniscient monks,

i got some html/tei like data and want to parse it to xml format. it is working pretty well for some files.. but not for all.. here is my code:

# pragma
use strict;
use warnings;

# modules
use XML::Simple;
use XML::Tidy;
use Data::Dumper;
use Data::Diver qw( Dive DiveRef DiveError );
use HTML::TreeBuilder;
use XML::Tidy::Tiny;

# little helper
use constant false => 0;
use constant true => 1;

...

        # get instance of treebuilder
        my $root = HTML::TreeBuilder->new();
        # configure treebuilder
        $root->ignore_unknown( false );
        # dump data to the treebuilder
        $root->parse( $fileData );
        # get name for target file
        my $target = $file;
        $target =~ s/$fileExtension$/xml/;
        # open output filehandle
        open( $FH, '>', $target );
        # configure output
        binmode $FH, ":utf8";

        # ERROR HERE
208:    my $data = $root->guts()->as_XML();

        print $FH xml_tidy( $data );
        close $FH;
...

caption has an invalid attribute name 'n' at script.pl line 208

i substite all 'n' in the file.. but got still the same error. for that the 'n' is not the anchor of this error.. i dont know what going on here?!

$root->guts()

is okey.. it is all about the ->as_XML() :-((

kindly, perlig

[Jenda]

IMnsHO, there is a bug in the _valid_name subroutine deep in HTML::Element. There should be

return (0) unless ( $attr =~ /^$START_CHAR$NAME_CHAR*$/ );

not

return (0) unless ( $attr =~ /^$START_CHAR$NAME_CHAR+$/ );

The XML specs say that

Name    ::=    NameStartChar (NameChar)*

[Anonymous Monk]

IMnsHO, there is a bug in the _valid_name subroutine deep in HTML::Element. There should be

I wouldn't go that far , the OP provides no data

[Jenda]

The OP doesn't need to provide data, the code doesn't match the specs linked five lines above the code in question.

[ambrus]

Look in the implementation of XML::Twig for the workarounds it uses when HTML::Tree's as_XML method dies.