Data::Dumper - stringified perl data structures, suitable for both printing and eval
use Data::Dumper;
# simple procedural interface
print Dumper($foo, $bar);
# extended usage with names
print Data::Dumper->Dump([$foo, $bar], [qw(foo *ary)]);
# configuration variables
{
local $Data::Dump::Purity = 1;
eval Data::Dumper->Dump([$foo, $bar], [qw(foo *ary)]);
}
# OO usage
$d = Data::Dumper->new([$foo, $bar], [qw(foo *ary)]);
...
print $d->Dump;
...
$d->Purity(1)->Terse(1)->Deepcopy(1);
eval $d->Dump;
Given a list of scalars or reference variables, writes out their contents in perl syntax. The references can also be objects. The contents of each variable is output in a single Perl statement. Handles self-referential structures correctly.
The return value can be evaled to get back an identical copy of the
original reference structure.
Any references that are the same as one of those passed in will be named
$VARn (where n is a numeric suffix), and other duplicate references
to substructures within $VARn will be appropriately labeled using arrow
notation. You can specify names for individual values to be dumped if you
use the Dump() method, or you can change the default $VAR prefix to
something else. See $Data::Dumper::Varname and $Data::Dumper::Terse
below.
The default output of self-referential structures can be evaled, but the
nested references to $VARn will be undefined, since a recursive
structure cannot be constructed using one Perl statement. You should set the
Purity flag to 1 to get additional statements that will correctly fill in
these references.
In the extended usage form, the references to be dumped can be given
user-specified names. If a name begins with a *, the output will
describe the dereferenced type of the supplied reference for hashes and
arrays, and coderefs. Output of names will be avoided where possible if
the Terse flag is set.
In many cases, methods that are used to set the internal state of the object will return the object itself, so method calls can be conveniently chained together.
Several styles of output are possible, all controlled by setting
the Indent flag. See Configuration Variables or Methods below
for details.
Data::Dumper object. The first argument is an
anonymous array of values to be dumped. The optional second argument is an
anonymous array of names for the values. The names need not have a leading
$ sign, and must be comprised of alphanumeric characters. You can begin
a name with a * to specify that the dereferenced type must be dumped
instead of the reference itself, for ARRAY and HASH references.
The prefix specified by $Data::Dumper::Varname will be used with a
numeric suffix if the name for a value is undefined.
Data::Dumper will catalog all references encountered while dumping the values. Cross-references (in the form of names of substructures in perl syntax) will be inserted at all possible points, preserving any structural interdependencies in the original set of values. Structure traversal is depth-first, and proceeds in order from the first supplied value to the last.
new), subject to the
configuration options below. In an array context, it returns a list
of strings corresponding to the supplied values.
The second form, for convenience, simply calls the new method on its
arguments before dumping the object immediately.
Reset to explicitly clear the table if needed. Such
references are not dumped; instead, their names are inserted wherever they
are encountered subsequently. This is useful especially for properly
dumping subroutine references.
Expects a anonymous hash of name => value pairs. Same rules apply for names
as in new. If no argument is supplied, will return the ``seen'' list of
name => value pairs, in an array context. Otherwise, returns the object
itself.
$VARn in the
output, where n is a numeric suffix. Will return a list of strings
in an array context.
Several configuration variables can be used to control the kind of output
generated when using the procedural interface. These variables are usually
localized in a block so that other parts of the code are not affected by
the change.
These variables determine the default state of the object created by calling
the new method, but cannot be used to alter the state of the object
thereafter. The equivalent method names should be used instead to query
or set the internal state of the object.
The method forms return the object itself when called with arguments, so that they can be chained together nicely.
evaled to recreate the
supplied reference structures. Setting it to 1 will output additional perl
statements that will correctly recreate nested references. The default is
0.
[\n\t\r], ``unsafe''
characters will be backslashed, and unprintable characters will be output as
quoted octal integers. Since setting this variable imposes a performance
penalty, the default is 0. Dump() will run slower if this flag is set,
since the fast XSUB implementation doesn't support it yet.
$VARn names
will be avoided where possible, but be advised that such output may not
always be parseable by eval.
bless(DATA, CLASS)-METHOD()>. Note that this means that
the method specified will have to perform any modifications required on the
object (like creating new state within it, and/or reblessing it in a
different package) and then return it. The client is responsible for making
sure the method can be called via the object, and that it returns a valid
object. Defaults to an empty string.
bless
builtin operator used to create objects. A function with the specified
name should exist, and should accept the same arguments as the builtin.
Default is bless.
Data::Dumper::Purity is set. (Useful in debugger when we often don't
want to see more than enough). Default is 0, which means there is
no maximum depth.
Run these code snippets to get a quick feel for the behavior of this module. When you are through with these examples, you may want to add or change the various configuration variables described above, to see their behavior. (See the testsuite in the Data::Dumper distribution for more examples.)
use Data::Dumper;
package Foo;
sub new {bless {'a' => 1, 'b' => sub { return "foo" }}, $_[0]};
package Fuz; # a weird REF-REF-SCALAR object
sub new {bless \($_ = \ 'fu\'z'), $_[0]};
package main;
$foo = Foo->new;
$fuz = Fuz->new;
$boo = [ 1, [], "abcd", \*foo,
{1 => 'a', 023 => 'b', 0x45 => 'c'},
\\"p\q\'r", $foo, $fuz];
########
# simple usage
########
$bar = eval(Dumper($boo));
print($@) if $@;
print Dumper($boo), Dumper($bar); # pretty print (no array indices)
$Data::Dumper::Terse = 1; # don't output names where feasible
$Data::Dumper::Indent = 0; # turn off all pretty print
print Dumper($boo), "\n";
$Data::Dumper::Indent = 1; # mild pretty print
print Dumper($boo);
$Data::Dumper::Indent = 3; # pretty print with array indices
print Dumper($boo);
$Data::Dumper::Useqq = 1; # print strings in double quotes
print Dumper($boo);
########
# recursive structures
########
@c = ('c');
$c = \@c;
$b = {};
$a = [1, $b, $c];
$b->{a} = $a;
$b->{b} = $a->[1];
$b->{c} = $a->[2];
print Data::Dumper->Dump([$a,$b,$c], [qw(a b c)]);
$Data::Dumper::Purity = 1; # fill in the holes for eval
print Data::Dumper->Dump([$a, $b], [qw(*a b)]); # print as @a
print Data::Dumper->Dump([$b, $a], [qw(*b a)]); # print as %b
$Data::Dumper::Deepcopy = 1; # avoid cross-refs
print Data::Dumper->Dump([$b, $a], [qw(*b a)]);
$Data::Dumper::Purity = 0; # avoid cross-refs
print Data::Dumper->Dump([$b, $a], [qw(*b a)]);
########
# deep structures
########
$a = "pearl";
$b = [ $a ];
$c = { 'b' => $b };
$d = [ $c ];
$e = { 'd' => $d };
$f = { 'e' => $e };
print Data::Dumper->Dump([$f], [qw(f)]);
$Data::Dumper::Maxdepth = 3; # no deeper than 3 refs down
print Data::Dumper->Dump([$f], [qw(f)]);
########
# object-oriented usage
########
$d = Data::Dumper->new([$a,$b], [qw(a b)]);
$d->Seen({'*c' => $c}); # stash a ref without printing it
$d->Indent(3);
print $d->Dump;
$d->Reset->Purity(0); # empty the seen cache
print join "----\n", $d->Dump;
########
# persistence
########
package Foo;
sub new { bless { state => 'awake' }, shift }
sub Freeze {
my $s = shift;
print STDERR "preparing to sleep\n";
$s->{state} = 'asleep';
return bless $s, 'Foo::ZZZ';
}
package Foo::ZZZ;
sub Thaw {
my $s = shift;
print STDERR "waking up\n";
$s->{state} = 'awake';
return bless $s, 'Foo';
}
package Foo;
use Data::Dumper;
$a = Foo->new;
$b = Data::Dumper->new([$a], ['c']);
$b->Freezer('Freeze');
$b->Toaster('Thaw');
$c = $b->Dump;
print $c;
$d = eval $c;
print Data::Dumper->Dump([$d], ['d']);
########
# symbol substitution (useful for recreating CODE refs)
########
sub foo { print "foo speaking\n" }
*other = \&foo;
$bar = [ \&other ];
$d = Data::Dumper->new([\&other,$bar],['*other','bar']);
$d->Seen({ '*foo' => \&foo });
print $d->Dump;
Due to limitations of Perl subroutine call semantics, you cannot pass an
array or hash. Prepend it with a \ to pass its reference instead. This
will be remedied in time, with the arrival of prototypes in later versions
of Perl. For now, you need to use the extended usage form, and prepend the
name with a * to output it as a hash or array.
Data::Dumper cheats with CODE references. If a code reference is
encountered in the structure being processed, an anonymous subroutine that
contains the string '``DUMMY''' will be inserted in its place, and a warning
will be printed if Purity is set. You can eval the result, but bear
in mind that the anonymous sub that gets created is just a placeholder.
Someday, perl will have a switch to cache-on-demand the string
representation of a compiled piece of code, I hope. If you have prior
knowledge of all the code refs that your data structures are likely
to have, you can use the Seen method to pre-seed the internal reference
table and make the dumped output point to them, instead. See EXAMPLES
above.
The Useqq flag makes Dump() run slower, since the XSUB implementation
does not support it.
SCALAR objects have the weirdest looking bless workaround.
Gurusamy Sarathy gsar@activestate.com
Copyright (c) 1996-98 Gurusamy Sarathy. All rights reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
Version 2.11 (unreleased)
perl(1)