Data::Dumper::Freezer proper usage - perl

Question

I'm trying to log data structures in an old and big Perl project. In order to do so, I use Data::Dumper, however, some structures are a bit too large and spam the log. So I'm looking for a way to log them in a less verbose manner.
Now Dumper's doc mentions $Data::Dumper::Freezer = <method_name> variable that can be used to fix that. I've tried using that.
Adding a serializer method that returns "shortened" value results in nothing, though the method gets called. Making the serializer method act on $_[0] causes the needed effect, but spoils the original data structure.
I'm confused. What am I doing wrong? How can I fix it?
Here's a refined sample code:
#!/usr/bin/env perl
use strict;
use warnings;
use Data::Dumper;
$Data::Dumper::Indent = 0;
$\="\n";
my $x = Foo->new ( answer => 42, use => "force" );
my $y = { foo => $x };
print "initial plain:\n\t", Dumper( $x );
print "initial compound:\n\t", Dumper( $y );
{
local $Data::Dumper::Freezer = 'freeze_pure';
print "still not abbreviated data:\n\t", Dumper( $y );
};
{
local $Data::Dumper::Freezer = 'freeze_replace';
print "abbreviated data:\n\t", Dumper( $y );
};
print "initial data is still intact:\n\t", Dumper( $x );
print "compound data is corrupted:\n\t", Dumper( $y );
package Foo;
sub new {
my $class = shift;
return bless { #_ }, $class;
};
sub freeze_pure {
my $self = $_[0];
warn "# In freeze_pure";
return bless {
values => join ",", values %$self
}, (ref $self) . "::short";
};
sub freeze_replace {
my $self = $_[0];
warn "# In freeze_replace";
$_[0] = bless {
values => join ",", values %$self
}, (ref $self) . "::short";
return;
};
And output:
initial plain:
$VAR1 = bless( {'use' => 'force','answer' => 42}, 'Foo' );
initial compound:
$VAR1 = {'foo' => bless( {'use' => 'force','answer' => 42}, 'Foo' )};
# In freeze_pure at dumper-freezer.pl line 36.
still not abbreviated data:
$VAR1 = {'foo' => bless( {'use' => 'force','answer' => 42}, 'Foo' )};
# In freeze_replace at dumper-freezer.pl line 42.
abbreviated data:
$VAR1 = {'foo' => bless( {'values' => 'force,42'}, 'Foo::short' )};
initial data is still intact:
$VAR1 = bless( {'use' => 'force','answer' => 42}, 'Foo' );
compound data is corrupted:
$VAR1 = {'foo' => bless( {'values' => 'force,42'}, 'Foo::short' )};

Answer 1

Although the documentation is a bit sparse, the intended use of freezer/toaster is data serialization/de-serialization, not prettification of debugging output.
So, Data::Dumper calls the freezer method, but doesn't use the return value. The idea is probably that if you're going to serialize an object, you won't be messing with it again until you de-serialize it, so there's no problem with changing the object itself.
Here's the relevant section of code from the Data::Dumper source:
# Call the freezer method if it's specified and the object has the
# method. Trap errors and warn() instead of die()ing, like the XS
# implementation.
my $freezer = $s->{freezer};
if ($freezer and UNIVERSAL::can($val, $freezer)) {
eval { $val->$freezer() };
warn "WARNING(Freezer method call failed): $#" if $#;
}
If you just want to reduce the size of the output in your logs, you can remove newlines and indentation by setting $Data::Dumper::Indent to zero:
use Data::Dumper;
use WWW::Mechanize;
$Data::Dumper::Indent = 0;
my $mech = WWW::Mechanize->new;
print Dumper $mech;
Output:
$VAR1 = bless( {'headers' => {},'ssl_opts' => {'verify_hostname' => 1},'forms' => undef,'page_stack' => [],'text' => undef,'requests_redirectable' => ['GET','HEAD','POST'],'timeout' => 180,'onerror' => sub { "DUMMY" },'current_form' => undef,'links' => undef,'max_redirect' => 7,'quiet' => 0,'images' => undef,'noproxy' => 0,'stack_depth' => 8675309,'show_progress' => undef,'protocols_forbidden' => undef,'no_proxy' => [],'handlers' => {'request_prepare' => bless( [{'owner' => 'LWP::UserAgent::cookie_jar','callback' => sub { "DUMMY" },'line' => '/home/foo/perl5/perlbrew/perls/perl-5.16.3/lib/site_perl/5.16.3/LWP/UserAgent.pm:705'}], 'HTTP::Config' ),'response_header' => bless( [{'owner' => 'LWP::UserAgent::parse_head','callback' => sub { "DUMMY" },'m_media_type' => 'html','line' => '/home/foo/perl5/perlbrew/perls/perl-5.16.3/lib/site_perl/5.16.3/LWP/UserAgent.pm:684'}], 'HTTP::Config' ),'response_done' => bless( [{'owner' => 'LWP::UserAgent::cookie_jar','callback' => sub { "DUMMY" },'line' => '/home/foo/perl5/perlbrew/perls/perl-5.16.3/lib/site_perl/5.16.3/LWP/UserAgent.pm:708'}], 'HTTP::Config' )},'onwarn' => sub { "DUMMY" },'protocols_allowed' => undef,'use_eval' => 1,'local_address' => undef,'autocheck' => 1,'title' => undef,'def_headers' => bless( {'user-agent' => 'WWW-Mechanize/1.75'}, 'HTTP::Headers' ),'cookie_jar' => bless( {'COOKIES' => {}}, 'HTTP::Cookies' ),'proxy' => {},'max_size' => undef}, 'WWW::Mechanize' );
This is still a lot of output, but it's certainly more compact than:
$VAR1 = bless( {
'headers' => {},
'ssl_opts' => {
'verify_hostname' => 1
},
'forms' => undef,
'page_stack' => [],
'text' => undef,
'requests_redirectable' => [
'GET',
'HEAD',
'POST'
],
'timeout' => 180,
'onerror' => sub { "DUMMY" },
'current_form' => undef,
'links' => undef,
'max_redirect' => 7,
'quiet' => 0,
'images' => undef,
'noproxy' => 0,
'stack_depth' => 8675309,
'show_progress' => undef,
'protocols_forbidden' => undef,
'no_proxy' => [],
'handlers' => {
'request_prepare' => bless( [
{
'owner' => 'LWP::UserAgent::cookie_jar',
'callback' => sub { "DUMMY" },
'line' => '/home/foo/perl5/perlbrew/perls/perl-5.16.3/lib/site_perl/5.16.3/LWP/UserAgent.pm:705'
}
], 'HTTP::Config' ),
'response_header' => bless( [
{
'owner' => 'LWP::UserAgent::parse_head',
'callback' => sub { "DUMMY" },
'm_media_type' => 'html',
'line' => '/home/foo/perl5/perlbrew/perls/perl-5.16.3/lib/site_perl/5.16.3/LWP/UserAgent.pm:684'
}
], 'HTTP::Config' ),
'response_done' => bless( [
{
'owner' => 'LWP::UserAgent::cookie_jar',
'callback' => sub { "DUMMY" },
'line' => '/home/foo/perl5/perlbrew/perls/perl-5.16.3/lib/site_perl/5.16.3/LWP/UserAgent.pm:708'
}
], 'HTTP::Config' )
},
'onwarn' => sub { "DUMMY" },
'protocols_allowed' => undef,
'use_eval' => 1,
'local_address' => undef,
'autocheck' => 1,
'title' => undef,
'def_headers' => bless( {
'user-agent' => 'WWW-Mechanize/1.75'
}, 'HTTP::Headers' ),
'cookie_jar' => bless( {
'COOKIES' => {}
}, 'HTTP::Cookies' ),
'proxy' => {},
'max_size' => undef
}, 'WWW::Mechanize' );
Alternatively, you could try Data::Dump, which allows you to filter the output using Data::Dump::Filtered. I prefer Data::Dump to Data::Dumper anyway because I think it has more sensible defaults (e.g. outputting escape sequences for whitespace other than spaces).
I haven't used the filtering feature yet, but brian d foy wrote a nice article about it with several examples.