benizi has asked for the wisdom of the Perl Monks concerning the following question:

I'm sure I'm missing some interaction between the various options, but I was wondering if someone (mirod?) could tell me how to accomplish the following. I have an XML document that I want to process with XML::Twig. I want the output document to retain the formatting characteristics of the input document. I also want to use twig_roots, since the file will not fit into memory, and is record-based. (i.e. the processing of each record is self-contained.). I used twig_print_outside_roots, because I want to specify a filehandle for the default prints/flushes. (Is there something more appropriate for that purpose?) The problem is that the root (wrapper) element's start tag is being output twice.

Example input:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE foo SYSTEM "/path">
<foo version="blah">
<record>stuff</record>
<record>stuff 2</record>
</foo>

[download]

Desired output:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE foo SYSTEM "/path">
<foo version="blah">
<record>altered stuff</record>
<record>altered stuff 2</record>
</foo>

[download]

My attempt:

#!/usr/bin/perl
use strict;
use warnings;
use XML::Twig;
open my $outfh, '>', "out.xml" or die ">out.xml:$!";
my $p = XML::Twig->new(
    twig_print_outside_roots => $outfh,
    twig_roots => {
        record => sub {
            $_->set_text("altered ".$_->text);
            shift->flush
        }
    },
    empty_tags => 'html',
    keep_encoding => 1,
    keep_spaces => 1,
);
$p->parsefile("in.xml");
print "DONE\n";

[download]

Actual output:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE foo SYSTEM "/path">
<foo version="blah">
<foo version="blah"><record>altered stuff</record> # the extra <foo ve
+rsion="blah"> is the problem.
<record>altered stuff 2</record>
</foo>

[download]

Replies are listed 'Best First'.
Re: XML::Twig outputting root element start tag twice
by Tanktalus (Canon) on Apr 18, 2006 at 18:34 UTC

    First, you're missing the end-flush. Before you're really "DONE", you need to add $p->flush(). That gets the extra </foo> tag you're missing in your output. Not that you want it, but once you fix the other problem, you'll want it back.

    Second, it's the twig_print_outside_roots flag that's doing it. Remove that. Instead, change your flush calls (including the new one) to have the param "$outfh". Now you'll flush to that file.

    That leaves me with: 

    #!/usr/bin/perl
use strict;
use warnings;
use XML::Twig;
open my $outfh, '>', "out.xml" or die ">out.xml:$!";

my $p = XML::Twig->new(
                       #twig_print_outside_roots => $outfh,
                       twig_roots => {
                           record => sub {
                               $_->set_text("altered ".$_->text);
                               shift->flush($outfh),
                           }
                       },
                       empty_tags => 'html',
                       keep_encoding => 1,
                       keep_spaces => 1,
                      );
$p->parsefile("in.xml");
$p->flush($outfh);
print "DONE\n";

    [download]
    As to why, ... I'm not sure.

    Hope that helps,

    Update: Ok, I see you really want the twig_print_outside_roots feature. It doesn't seem to do what you want it to, though. I am curious, though, as to why the formatting matters - this is XML, after all...

[reply]
[d/l]
[select]

      Explicitly adding the $outfh is part of what I was avoiding, as it's not in the scope of the actual handlers in the real-life example. (XML::Twig has so much DWIMmery, I assumed specifying an output filehandle would be something pretty trivial.)

      As to the formatting, it's because, while I'm using XML::Twig, other people in the project aren't (yet!), and the line-based -ness of the format is easier for them to handle. (Plus, I simply prefer the aesthetics of it.)

[reply]

        Why is the output filehandle not in scope? If it is available when you create the twig, you should be able to use it in the handlers (you can use a closure to pass it to the handlers). You could also use select to send all output to the filehandle, even though I would consider not so good for the maintenability of the code.

        Finally, if you are using the latest version of XML::Twig, you don't need the final flush, it's done automagically.

[reply]

Back to Seekers of Perl Wisdom