Switch branches/tags
Find file
Fetching contributors…
Cannot retrieve contributors at this time
509 lines (364 sloc) 17.4 KB
Class::Base - useful base class for deriving other modules
package My::Funky::Module;
use base qw( Class::Base );
# custom initialiser method
sub init {
my ($self, $config) = @_;
# copy various params into $self
$self->params($config, qw( FOO BAR BAZ ))
|| return undef;
# to indicate a failure
return $self->error('bad constructor!')
if $something_bad;
# or to indicate general happiness and well-being
return $self;
package main;
# new() constructor folds args into hash and calls init()
my $object = My::Funky::Module->new( foo => 'bar', ... )
|| die My::Funky::Module->error();
# error() class/object method to get/set errors
$object->error('something has gone wrong');
print $object->error();
# debugging() method (de-)activates the debug() method
# debug() prints to STDERR if debugging enabled
$object->debug('The ', $animal, ' sat on the ', $place);
This module implements a simple base class from which other modules can
be derived, thereby inheriting a number of useful methods such as
"new()", "init()", "params()", "clone()", "error()" and "debug()".
For a number of years, I found myself re-writing this module for
practically every Perl project of any significant size. Or rather, I
would copy the module from the last project and perform a global search
and replace to change the names. Each time it got a little more polished
and eventually, I decided to Do The Right Thing and release it as a
module in it's own right.
It doesn't pretend to be an all-encompassing solution for every kind of
object creation problem you might encounter. In fact, it only supports
blessed hash references that are created using the popular, but by no
means universal convention of calling "new()" with a list or reference
to a hash array of named parameters. Constructor failure is indicated by
returning undef and setting the "$ERROR" package variable in the
module's class to contain a relevant message (which you can also fetch
by calling "error()" as a class method).
my $object = My::Module->new(
file => 'myfile.html',
msg => 'Hello World'
) || die $My::Module::ERROR;
my $object = My::Module->new({
file => 'myfile.html',
msg => 'Hello World',
}) || die My::Module->error();
The "new()" method handles the conversion of a list of arguments into a
hash array and calls the "init()" method to perform any initialisation.
In many cases, it is therefore sufficient to define a module like so:
package My::Module;
use Class::Base;
use base qw( Class::Base );
sub init {
my ($self, $config) = @_;
# copy some config items into $self
$self->params($config, qw( FOO BAR )) || return undef;
return $self;
# other application-specific methods
Then you can go right ahead and use it like this:
use My::Module;
my $object = My::Module->new( FOO => 'the foo value',
BAR => 'the bar value' )
|| die $My::Module::ERROR;
Despite its limitations, Class::Base can be a surprisingly useful module
to have lying around for those times where you just want to create a
regular object based on a blessed hash reference and don't want to worry
too much about duplicating the same old code to bless a hash, define
configuration values, provide an error reporting mechanism, and so on.
Simply derive your module from "Class::Base" and leave it to worry about
most of the detail. And don't forget, you can always redefine your own
"new()", "error()", or other method, if you don't like the way the
Class::Base version works.
Subclassing Class::Base
This module is what object-oriented afficionados would describe as an
"abstract base class". That means that it's not designed to be used as a
stand-alone module, rather as something from which you derive your own
modules. Like this:
package My::Funky::Module
use base qw( Class::Base );
You can then use it like this:
use My::Funky::Module;
my $module = My::Funky::Module->new();
Construction and Initialisation Methods
If you want to apply any per-object initialisation, then simply write an
"init()" method. This gets called by the "new()" method which passes a
reference to a hash reference of configuration options.
sub init {
my ($self, $config) = @_;
return $self;
When you create new objects using the "new()" method you can either pass
a hash reference or list of named arguments. The "new()" method does the
right thing to fold named arguments into a hash reference for passing to
the "init()" method. Thus, the following are equivalent:
# hash reference
my $module = My::Funky::Module->new({
foo => 'bar',
wiz => 'waz',
# list of named arguments (no enclosing '{' ... '}')
my $module = My::Funky::Module->new(
foo => 'bar',
wiz => 'waz'
Within the "init()" method, you can either handle the configuration
sub init {
my ($self, $config) = @_;
$self->{ file } = $config->{ file }
|| return $self->error('no file specified');
return $self;
or you can call the "params()" method to do it for you:
sub init {
my ($self, $config) = @_;
$self->params($config, 'file')
|| return $self->error('no file specified');
return $self;
Error Handling
The "init()" method should return $self to indicate success or undef to
indicate a failure. You can use the "error()" method to report an error
within the "init()" method. The "error()" method returns undef, so you
can use it like this:
sub init {
my ($self, $config) = @_;
# let's make 'foobar' a mandatory argument
$self->{ foobar } = $config->{ foobar }
|| return $self->error("no foobar argument");
return $self;
When you create objects of this class via "new()", you should now check
the return value. If undef is returned then the error message can be
retrieved by calling "error()" as a class method.
my $module = My::Funky::Module->new()
|| die My::Funky::Module->error();
Alternately, you can inspect the "$ERROR" package variable which will
contain the same error message.
my $module = My::Funky::Module->new()
|| die $My::Funky::Module::ERROR;
Of course, being a conscientious Perl programmer, you will want to be
sure that the "$ERROR" package variable is correctly defined.
package My::Funky::Module
use base qw( Class::Base );
our $ERROR;
You can also call "error()" as an object method. If you pass an argument
then it will be used to set the internal error message for the object
and return undef. Typically this is used within the module methods to
report errors.
sub another_method {
my $self = shift;
# set the object error
return $self->error('something bad happened');
If you don't pass an argument then the "error()" method returns the
current error value. Typically this is called from outside the object to
determine its status. For example:
my $object = My::Funky::Module->new()
|| die My::Funky::Module->error();
|| die $object->error();
Debugging Methods
The module implements two methods to assist in writing debugging code:
debug() and debugging(). Debugging can be enabled on a per-object or
per-class basis, or as a combination of the two.
When creating an object, you can set the "DEBUG" flag (or lower case
"debug" if you prefer) to enable or disable debugging for that one
my $object = My::Funky::Module->new( debug => 1 )
|| die My::Funky::Module->error();
my $object = My::Funky::Module->new( DEBUG => 1 )
|| die My::Funky::Module->error();
If you don't explicitly specify a debugging flag then it assumes the
value of the "$DEBUG" package variable in your derived class or 0 if
that isn't defined.
You can also switch debugging on or off via the "debugging()" method.
$object->debugging(0); # debug off
$object->debugging(1); # debug on
The "debug()" method examines the internal debugging flag (the "_DEBUG"
member within the "$self" hash) and if it finds it set to any true value
then it prints to STDERR all the arguments passed to it. The output is
prefixed by a tag containing the class name of the object in square
brackets (but see the "id()" method below for details on how to change
that value).
For example, calling the method as:
$object->debug('foo', 'bar');
prints the following output to STDERR:
[My::Funky::Module] foobar
When called as class methods, "debug()" and "debugging()" instead use
the "$DEBUG" package variable in the derived class as a flag to control
debugging. This variable also defines the default "DEBUG" flag for any
objects subsequently created via the new() method.
package My::Funky::Module
use base qw( Class::Base );
our $ERROR;
our $DEBUG = 0 unless defined $DEBUG;
# some time later, in a module far, far away
package main;
# debugging off (by default)
my $object1 = My::Funky::Module->new();
# turn debugging on for My::Funky::Module objects
$My::Funky::Module::DEBUG = 1;
# alternate syntax
# debugging on (implicitly from $DEBUG package var)
my $object2 = My::Funky::Module->new();
# debugging off (explicit override)
my $object3 = My::Funky::Module->new(debug => 0);
If you call "debugging()" without any arguments then it returns the
value of the internal object flag or the package variable accordingly.
print "debugging is turned ", $object->debugging() ? 'on' : 'off';
Class constructor method which expects a reference to a hash array of
parameters or a list of "name => value" pairs which are automagically
folded into a hash reference. The method blesses a hash reference and
then calls the "init()" method, passing the reference to the hash array
of configuration parameters.
Returns a reference to an object on success or undef on error. In the
latter case, the "error()" method can be called as a class method, or
the "$ERROR" package variable (in the derived class' package) can be
inspected to return an appropriate error message.
my $object = My::Class->new( foo => 'bar' ) # params list
|| die $My::Class::$ERROR; # package var
my $object = My::Class->new({ foo => 'bar' }) # params hashref
|| die My::Class->error; # class method
Object initialiser method which is called by the "new()" method, passing
a reference to a hash array of configuration parameters. The method may
be derived in a subclass to perform any initialisation required. It
should return "$self" on success, or "undef" on error, via a call to the
"error()" method.
package My::Module;
use base qw( Class::Base );
sub init {
my ($self, $config) = @_;
# let's make 'foobar' a mandatory argument
$self->{ foobar } = $config->{ foobar }
|| return $self->error("no foobar argument");
return $self;
params($config, @keys)
The "params()" method accept a reference to a hash array as the first
argument containing configuration values such as those passed to the
"init()" method. The second argument can be a reference to a list of
parameter names or a reference to a hash array mapping parameter names
to default values. If the second argument is not a reference then all
the remaining arguments are taken as parameter names. Thus the method
can be called as follows:
sub init {
my ($self, $config) = @_;
# either...
$self->params($config, qw( foo bar ));
# or...
$self->params($config, [ qw( foo bar ) ]);
# or...
$self->params($config, { foo => 'default foo value',
bar => 'default bar value' } );
return $self;
The method looks for values in $config corresponding to the keys
specified and copies them, if defined, into $self.
Keys can be specified in UPPER CASE and the method will look for either
upper or lower case equivalents in the "$config" hash. Thus you can call
"params()" from "init()" like so:
sub init {
my ($self, $config) = @_;
$self->params($config, qw( FOO BAR ))
return $self;
but use either case for parameters passed to "new()":
my $object = My::Module->new( FOO => 'the foo value',
BAR => 'the bar value' )
|| die My::Module->error();
my $object = My::Module->new( foo => 'the foo value',
bar => 'the bar value' )
|| die My::Module->error();
Note however that the internal key within "$self" used to store the
value will be in the case provided in the call to "params()" (upper case
in this example). The method doesn't look for upper case equivalents
when they are specified in lower case.
When called in list context, the method returns a list of all the values
corresponding to the list of keys, some of which may be undefined
(allowing you to determine which values were successfully set if you
need to). When called in scalar context it returns a reference to the
same list.
The "clone()" method performs a simple shallow copy of the object hash
and creates a new object blessed into the same class. You may want to
provide your own "clone()" method to perform a more complex cloning
my $clone = $object->clone();
error($msg, ...)
General purpose method for getting and setting error messages. When
called as a class method, it returns the value of the "$ERROR" package
variable (in the derived class' package) if called without any
arguments, or sets the same variable when called with one or more
arguments. Multiple arguments are concatenated together.
# set error
My::Module->error('set the error string');
My::Module->error('set ', 'the ', 'error string');
# get error
print My::Module->error();
print $My::Module::ERROR;
When called as an object method, it operates on the "_ERROR" member of
the object, returning it when called without any arguments, or setting
it when called with arguments.
# set error
$object->error('set the error string');
# get error
print $object->error();
The method returns "undef" when called with arguments. This allows it to
be used within object methods as shown:
sub my_method {
my $self = shift;
# set error and return undef in one
return $self->error('bad, bad, error')
if $something_bad;
debug($msg, $msg, ...)
Prints all arguments to STDERR if the internal "_DEBUG" flag (when
called as an object method) or "$DEBUG" package variable (when called as
a class method) is set to a true value. Otherwise does nothing. The
output is prefixed by a string of the form "[Class::Name]" where the
name of the class is that returned by the "id()" method.
Used to get (no arguments) or set ($flag defined) the value of the
internal "_DEBUG" flag (when called as an object method) or "$DEBUG"
package variable (when called as a class method).
The "debug()" method calls this method to return an identifier for the
object for printing in the debugging message. By default it returns the
class name of the object (i.e. "ref $self"), but you can of course
subclass the method to return some other value. When called with an
argument it uses that value to set its internal "_ID" field which will
be returned by subsequent calls to "id()".
Andy Wardley <>
This is version 0.03 of Class::Base.
This module began life as the Template::Base module distributed as part
of the Template Toolkit.
Thanks to Brian Moseley and Matt Sergeant for suggesting various
enhancments, some of which went into version 0.02.
Copyright (C) 1996-2002 Andy Wardley. All Rights Reserved.
This module is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.