=head1 NAME DateTime::Format::Builder::Tutorial - Quick class on using Builder =head1 CREATING A CLASS As most people who are writing modules know, you start a package with a package declaration and some indication of module version: package DateTime::Format::ICal; our $VERSION = '0.04'; After that, you call Builder with some options. There are only a few (detailed later). Right now, we're only interested in I<parsers>. use DateTime::Format::Builder ( parsers => { ... } ); The I<parsers> option takes a reference to a hash of method names and specifications: parsers => { parse_datetime => ... , parse_datetime_with_timezone => ... , ... } Builder will create methods in your class, each method being a parser that follows the given specifications. It is B<strongly> recommended that one method is called I<parse_datetime>, be it a Builder created method or one of your own. In addition to creating any of the parser methods it also creates a C<new()> method that can instantiate (or clone) objects of this class. This behaviour can be modified with the I<constructor> option, but we don't need to know that yet. Each value corresponding to a method name in the parsers list is either a single specification, or a list of specifications. We'll start with the simple case. parse_briefdate => { params => [ qw( year month day ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)$/, }, This will result in a method named I<parse_briefdate> which will take strings in the form C<20040716> and return DateTime objects representing that date. A user of the class might write: use DateTime::Format::ICal; my $date = "19790716"; my $dt = DateTime::Format::ICal->parse_briefdate( $date ); print "My birth month is ", $dt->month_name, "\n"; The C<regex> is applied to the input string, and if it matches, then C<$1>, C<$2>, ... are mapped to the I<params> given and handed to C<< DateTime->new() >>. Essentially: my $rv = DateTime->new( year => $1, month => $2, day => $3 ); There are more complicated things one can do within a single specification, but we'll cover those later. Often, you'll want a method to be able to take one string, and run it against multiple parser specifications. It would be very irritating if the user had to work out what format the datetime string was in and then which method was most appropriate. So, Builder lets you specify multiple specifications: parse_datetime => [ { params => [ qw( year month day hour minute second ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)(\d\d)$/, }, { params => [ qw( year month day hour minute ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)$/, }, { params => [ qw( year month day hour ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)$/, }, { params => [ qw( year month day ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)$/, }, ], It's an arrayref of specifications. A parser will be created that will try each of these specifications sequentially, in the order you specified. There's a flaw with this though. In this example, we're building a parser for ICal datetimes. One can place a timezone id at the start of an ICal datetime. You might extract such an id with the following code: if ( $date =~ s/^TZID=([^:]+):// ) { $time_zone = $1; } # Z at end means UTC elsif ( $date =~ s/Z$// ) { $time_zone = 'UTC'; } else { $time_zone = 'floating'; } C<$date> would end up without the id, and C<$time_zone> would contain something appropriate to give to DateTime's I<set_time_zone> method, or I<time_zone> argument. But how to get this scrap of code into your parser? You might be tempted to call the parser something else and build a small wrapper. There's no need though because an option is provided for preprocesing dates: parse_datetime => [ [ preprocess => \&_parse_tz ], # Only changed line! { params => [ qw( year month day hour minute second ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)(\d\d)$/, }, { params => [ qw( year month day hour minute ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)$/, }, { params => [ qw( year month day hour ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)$/, }, { params => [ qw( year month day ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)$/, }, ], It will necessitate I<_parse_tz> to be written, and that routine looks like this: sub _parse_tz { my %args = @_; my ($date, $p) = @args{qw( input parsed )}; if ( $date =~ s/^TZID=([^:]+):// ) { $p->{time_zone} = $1; } # Z at end means UTC elsif ( $date =~ s/Z$// ) { $p->{time_zone} = 'UTC'; } else { $p->{time_zone} = 'floating'; } return $date; } On input it is given a hash containing two items: the input date and a hashref that will be used in the parsing. The return value from the routine is what the parser specifications will run against, and anything in the I<parsed> hash (C<$p> in the example) will be put in the call to C<< DateTime->new(...) >>. So, we now have a happily working ICal parser. It parses the assorted formats, and can also handle timezones. Is there anything else it needs to do? No. But we can make it work more efficiently. At present, the specifications are tested sequentially. However, each one applies to strings of particular lengths. Thus we could be efficient and have the parser only test the given strings against a parser that handles that string length. Again, Builder makes it easy: parse_datetime => [ [ preprocess => \&_parse_tz ], { length => 15, # We handle strings of exactly 15 chars params => [ qw( year month day hour minute second ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)(\d\d)$/, }, { length => 13, # exactly 13 chars... params => [ qw( year month day hour minute ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)$/, }, { length => 11, # 11.. params => [ qw( year month day hour ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)$/, }, { length => 8, # yes. params => [ qw( year month day ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)$/, }, ], Now the created parser will create a parser that only runs specifications against appropriate strings. So our complete code looks like: package DateTime::Format::ICal; use strict; our $VERSION = '0.04'; use DateTime::Format::Builder ( parsers => { parse_datetime => [ [ preprocess => \&_parse_tz ], { length => 15, params => [ qw( year month day hour minute second ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)(\d\d)$/, }, { length => 13, params => [ qw( year month day hour minute ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)(\d\d)$/, }, { length => 11, params => [ qw( year month day hour ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)T(\d\d)$/, }, { length => 8, params => [ qw( year month day ) ], regex => qr/^(\d\d\d\d)(\d\d)(\d\d)$/, }, ], }, ); sub _parse_tz { my %args = @_; my ($date, $p) = @args{qw( input parsed )}; if ( $date =~ s/^TZID=([^:]+):// ) { $p->{time_zone} = $1; } # Z at end means UTC elsif ( $date =~ s/Z$// ) { $p->{time_zone} = 'UTC'; } else { $p->{time_zone} = 'floating'; } return $date; } 1; And that's an ICal parser. The actual L<DateTime::Format::ICal> module also includes formatting methods and parsing for durations, but Builder doesn't support those yet. A drop in replacement (at the time of writing the replacement) can be found in the F<examples> directory of the Builder distribution, along with similar variants of other common modules. =head1 SUPPORT Any errors you see in this document, please log them with CPAN RT system via the web or email: http://perl.dellah.org/rt/dtbuilder bug-datetime-format-builder@rt.cpan.org This makes it much easier for me to track things and thus means your problem is less likely to be neglected. =head1 LICENSE AND COPYRIGHT Copyright E<copy> Iain Truskett, 2003. All rights reserved. You can redistribute this document and/or modify it under the same terms as Perl itself. The full text of the licenses can be found in the F<Artistic> and F<COPYING> files included with this document. =head1 AUTHOR Iain Truskett <spoon@cpan.org> =head1 SEE ALSO C<datetime@perl.org> mailing list. http://datetime.perl.org/ L<perl>, L<DateTime>, L<DateTime::Format::Builder> =cut