NAME Lexical::Attributes - Proper encapsulation SYNOPSIS use Lexical::Attributes; has $.scalar; has $.key ro; has (@.array, %.hash) is rw; sub method { $self -> another_method; print $.scalar; } DESCRIPTION NOTE: This module has changed significantly between releases 1.3 and 1.4. Code that works with version 1.3 or earlier *will not* work with version 1.4 or later. NOTE: This is experimental software! Certain things will change, specially if they are marked FIXME or mentioned on the TODO list. This module was created out of frustration with Perl's default OO mechanism, which doesn't offer good data encapsulation. I designed the technique of *Inside-Out Objects* several years ago, but I was not really satisfied with it, as it still required a lot of typing. This module uses a source filter to hide the details of the Inside-Out technique from the user. Attributes, the variables that belong to an object, are stored in lexical hashes, instead of piggy-backing on the reference that makes the object. The lexical hashes, one for each attribute, are indexed using the object. However, the details of this technique are hidden behind a source filter. Instead, attributes are declared in a similar way as lexical variables. Except that instead of "my", a Perl6 keyword, "has" is used. Another thing is borrowed from Perl6, and that's the second sigil. Attributes have a dot separating the sigil from the name of attribute. Attributes To declare an attribute, use the Perl6 keyword "has". The simplest way to declare an attribute is: has $.colour; # Gives the object a 'colour' attribute. Now your object has an attribute *colour*. Note the way the attribute is written, in a Perl6 style, it has the sigil (a "$"), a period, and then the attribute name. Attribute names are strings of letters, digits and underscores, and cannot start with a digit. Attribute names are case-sensitive. You can use this attribute in the same way as a normal Perl scalar (except for interpolation). Here's a sub that prints out the colour of the object: sub print_colour { print $.colour; } Array and hash attributes work in a similar way: has @.array; # Gives the object an array attribute. has %.hash; # Gives the object a hash attribute. And you can use them in a similar way as you can with "normal" Perl variables: sub first_element { return $.array [0]; } sub pop_element { return pop @.array; } sub last_index { return $#.array; } sub gimme_key { my $key = shift; return $.hash {$key}; } sub gimme_all_keys { return keys %.hash; } Note however that you *cannot* have a scalar and an array (or a scalar and a hash, or an array and a hash) with the same name. Using both "has $.key;" and "has @.key;" will result in a warning, and the second (and third, fourth, etc) declaration of the attibute will be ignored. If you have several attributes you want to declare, you can use "has" in a similar way as you can "my" and "local". "has" takes a list as argument as well (parenthesis are required): has ($.key1, @.key2, %.key3); Note that the declaration, that is, the "has" keyword followed by an attribute, or a list of attributes, can be followed by an optional *trait* (as discussed below), must be followed by a semi-colon (after optional whitespace). The following will not work: has $.does_not_work = 1; Traits Since inspecting and setting the attributes of an object is a commonly requested action, it's possible to give the attributes *traits* that will achieve this. Traits are given by following the "has" declaration with the keyword "is" and the name of the trait (with the keyword being optional). Examples include: has $.get_set is rw; has $.get ro; has (@.array, %.hash) is priv; # Trait applies to both attributes. The following traits can be given: "pr" or "priv" Using "pr" or "priv" has the same effect as not giving any traits, no accessor for this attribute is generated. "ro" This trait generates an accessor for the attribute, with the same name as the attribute. For scalar attributes, calling the accessor returns the value of the attribute. Any parameters given to the accessor will be ignored. package MyObject; use Lexical::Attributes; has $.colour is ro; sub new {bless \do {my $obj} => shift} sub some_sub { ... # Some code that sets the 'colour' attribute. } 1; # Main program my $obj = MyObject -> new; $obj -> some_sub (...); print $obj -> colour; # Prints the colour. __END__ Accessors for arrays and hashes take optional arguments. If no arguments are given, the accessor will return the array or hash in list context (as a list - just as if you'd use an array or hash in list context). In scalar context, the number of elements of the array or hash are returned. If one or more arguments are given, the corresponding arguments are returned. Some examples: package MyObject; use Lexical::Attributes; has @.colours is ro; has %.fruit is ro; sub new {bless \do {my $obj} => shift} method init { # See below for discussion of 'method'. @.colours = qw /red white blue green yellow/; %.fruit = (cherry => 'red', peach => 'pink', apple => 'green', ); $self; } 1; # Main program my $obj = MyObject -> new -> init; local $, = " "; print $obj -> colours; # red white blue green yellow print $obj -> colours (2); # blue print $obj -> colours (1, 3); # white green print sort $obj -> fruit; # apple cherry green peach pink red print $obj -> fruit ('cherry'); # red print $obj -> fruit ('apple', 'peach') # green pink __END__ "rw" Attributes with the "rw" trait have two accessors generated for them. One accessor, with the same name as the attribute is used to fetch the value - it's identical to the accessor discussed at above, for "ro" attributes. The second accessor is used to store values; its name will be the name of the attribute, prepended by "set_". For scalar values, calling the setting accessor sets the attribute to the first argument. Any other argument are ignored. package MyObject; use Lexical::Attributes; has $.name is rw; sub new {bless \do {my $var} => shift} 1; # Main program my $obj = MyObject -> new; $obj -> set_name ("Abigail"); print $obj -> name; # Prints 'Abigail'. __END__ For aggregates, the situation is a bit more complex. There are four possibilities: No arguments If the settable accessor was called without arguments, the array or hash this accessor is associated with is cleared - that is, set to an empty array or hash. One argument, a reference of the appropriate type If one argument is given, and the argument is a reference of the appropriate type (a reference to an array for array attributes, and a reference to a hash for hash attributes), the array or hash is set to the given argument. Note that the actual reference is stored - no copies are made. One argument, not a reference of the appropriate type In this case, the argument is taken to be an index in the array or hash (so, for array attributes, the argument is cast to an integer if necessary, and to a string for hash attributes), and the corresponding element is deleted, in a similar way "delete" is called on regular arrays and hashes. Note that for arrays, "deleting" something that's in the middle of the array doesn't cause the array to shrink - the element is just undefined. More than one argument Then it's assumed a list of key (or index)/value pairs are given. Values are set to the corresponding keys or indices. Arrays and hashes will grow if needed. package MyObject; use Lexical::Attributes; has @.colours is rw; has %.fruit is rw; sub new {bless \do {my $obj} => shift} 1; # Main program. my $obj = MyObject -> new; local $, = " "; # Set the colours to a specific array. $obj -> set_colours (['red', 'white', 'blue']); print $obj -> colours; # 'red white blue'. print $obj -> colours (1); # 'white'. # Change colour on index 1. $obj -> set_colours (1, 'yellow'); print $obj -> colours; # 'red yellow blue'. # Change/add multiple colours. $obj -> set_colours (1, 'green', 3, 'brown'); print $obj -> colours; # 'red green blue brown'. # Delete colour on index 3. $obj -> set_colours (3); print $obj -> colours; # 'red green blue'. # Clear the array. $obj -> set_colour; print $obj -> colours; # Nothing, array is empty. # Set the fruits to a specific hash. $obj -> set_fruit ({apple => 'green', cherry => 'red', peach => 'pink'}); print $obj -> fruit; # 'apple green peach pink cherry red'. print $obj -> fruit ("apple"); # 'green'. # Change the colour of the apple. $obj -> set_fruit (apple => 'yellow'); print $obj -> fruit; # 'apple yellow peach pink cherry red'. # Change/add multiple fruits. $obj -> set_fruit (apple => 'red', lemon => 'yellow'); print $obj -> fruit; # 'apple red peach pink # cherry red lemon yellow'. # Delete a fruit $obj -> set_fruit ("peach"); print $obj -> fruit; # 'apple red cherry red lemon yellow'. # Delete all fruits. $obj -> set_fruit; print $obj -> fruit; # Nothing, hash is empty. All settable accessors return the object, regardless of the number or types of arguments. This gives the caller the option of chaining modifications: my $obj = Class -> new -> set_age (25) -> set_name ("Jane Doe") -> set_hair_colour ("auburn"); Methods In order for the module to access the attributes, it needs access to the variable holding the current object. It will assume this variable is called $self. This is not likely to be a problem, as it seems to be quite common to name the variable holding the current object $self. To further add the programmer, if a subroutine uses the keyword "method" instead of "sub", it will have a variable called $self, in which the first element of @_ is shifted. Essentially, the line "my $self = shift;" is prepended to the body of the subroutine. Subroutines that do not use the keyword "method" are left as is - these subroutines are typically reserved for class methods, or private subroutines. Examples: # Don't need to declare $self. sub my_method { $.attribute + $self -> other_method; } If you do not use the "method" keyword, you do not put the current object into a variable called $self, and you use one of the lexical attributes, your code is unlikely to work. DESTROY Since the attributes are stored in lexical hashes, attributes do not get garbage collected via a reference counting mechanism when the object goes out of scope. In order to clean up attribute data, action triggered by the call of "DESTROY" is needed. Hence, this module will insert a "DESTROY" subroutine which will take care of cleaning up the attribute data. It will also propagate calling "DESTROY" methods in any inherited classes. If you want to do any other action you'd normally put into "DESTROY", create a method called "DESTRUCT". This method will be called on when the object goes out of scope. The method will be called before attributes values have been cleaned up. There is no need to manually call "DESTRUCT" in inherited classes, as "Lexical::Attributes" will do that for you. In fact, calling "DESTRUCT" in a super class yourself is likely to cause unwanted effects, because that will mean "DESTRUCT" in a superclass is called more than once. Inheritance Inheritance just works. Classes using this technique require *nothing* from their super class implementation, and demand *nothing* from the classes that will inherit them. Super classes can use this technique, or traditional hash based objects, or something else entirely. And it's the same for classes that will inherit our classes. Interpolation Interpolation of scalars and array is possible in "", ``, "//", "m//", "s///", "qq {}" "qr {}", and "qx {}" strings. There's no interpolation in '', "m''", "s'''", "tr//" nor in "qw {}" strings. Overloading Overloading of objects should work in the same way as other types of objects. TODO o Compiling a module is slow. This is probably caused by FILTER_ONLY being slow. o Consider more traits. Methods for pop/push/shift/unshift for arrays, and keys/values/each for hashes would be useful. So are getting/setting keys by index. DEVELOPMENT The current sources of this module are found on github, . AUTHOR Abigail, . COPYRIGHT and LICENSE This program is copyright 2004, 2005, 2009 by Abigail. Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. INSTALLATION To install this module type the following: perl Makefile.PL make make test make install