| Class-DBI-Loader-mysql-Grok documentation | Contained in the Class-DBI-Loader-mysql-Grok distribution. |
Class::DBI::Loader::mysql::Grok - Build Quality Table Relationships Automatically
use Class::DBI::Loader; # optional
use Class::DBI::Loader::mysql::Grok;
my $loader = Class::DBI::Loader->new(
...
namespace => "Music",
relationships => 1,
);
my $class = $loader->find_class('artist'); # $class => Music::Artist
my $artist = $class->retrieve(1);
for my $cd ($artist->cds) {
print $cd->artist->name,"\n";
print $cd->reldate->ymd,"\n"; # a Time::Piece object
}
# etc ...
If you name your tables and columns using some common sense rules, there's no need for you to do any work to have proper db abstraction. The following examples mostly follow the Class::DBI perldoc. To see where they differ (immaterially), see the test script and the accompanying SQL.
The kinds of relationships which are created include:
In the example above, the cd table contains a column which matches the name of another table: artist. In this case, Music::Cd objects have a has_a relationship with Music::Artist. As a result, you can call $cd->artist->name, etc.
Similar to the has_a example above, the fact that the cd table contains a column which matches the name of another table means that Music::Artist objects have a has_many relationship with Music::CD. As a result, you can call $artist->cds->next->title, etc.
When we're working with a mapping table like Music::StyleRef in the Class::DBI perldoc, which maps a many-to-many relationship, the mapping table name must =~ /_ref$/i, and the columns in that table must be named after the tables to which they refer.
The liner_notes table's primary key is named 'cd'. Since that's so, and the table name (liner_notes) !~ /_ref$/i: Music::Cd->might_have(liner_notes_notes => Music::LinerNotes => 'notes');
While not a multi-table relationship, Time::Piece support is included for date, time, datetime, and timestamp types.
None by default, but it does redefine the _relationships routine in Class::DBI::Loader::mysql.
Class::DBI Class::DBI::Loader, Class::DBI::Loader::mysql, Time::Piece
James Tolley, <james@bitperfect.com>
Copyright (C) 2005 by James Tolley
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.6 or, at your option, any later version of Perl 5 you may have available.
| Class-DBI-Loader-mysql-Grok documentation | Contained in the Class-DBI-Loader-mysql-Grok distribution. |
package Class::DBI::Loader::mysql::Grok; our $VERSION = '0.19'; package Class::DBI::Loader::mysql; use strict; use warnings; use Lingua::EN::Inflect qw(PL); use Class::DBI::Loader 0.22; use Class::DBI::Loader::mysql 0.22; use Time::Piece::MySQL 0.05; no warnings 'redefine'; sub _relationships { my $self = shift; print "Entering Grok::_relationships\n" if $self->debug; my @tables = $self->tables; my $dbh = $self->find_class( $tables[0] )->db_Main; # keys are table names, # values are hrefs where keys # are column names and values are pri & type my %tables = (); for my $table ($self->tables) { my $sth = $dbh->prepare(qq[ SELECT * FROM $table WHERE 0=1 ]); $sth->execute; for(my $i = 0; $i < @{$sth->{NAME}}; $i++) { # MySQL-specific $tables{$table}->{$sth->{NAME}->[$i]} = { type => $sth->{mysql_type_name}->[$i], pri => $sth->{mysql_is_pri_key}->[$i], # uni => ($sth->{mysql_is_key}->[$i] && ! $sth->{mysql_is_pri_key}->[$i]), # unique }; } } # for each table, go through the columns to see if they refer to another table # for my $subject_table (@tables) { print "$subject_table\n" if $self->debug; for my $column (sort keys %{$tables{$subject_table}}) { my $subject_class = $self->find_class($subject_table); $subject_class->autoinflate(dates => 'Time::Piece'); my $object_class = $self->find_class($column); if($tables{$subject_table}->{$column}->{type} eq 'time') { print "\t$subject_class->has_a($column, Time::Piece) time\n" if $self->debug; $subject_class->has_a($column, 'Time::Piece', inflate => sub { Time::Piece->strptime(shift(),'%H:%M:%S') }, deflate => sub { shift->strftime('%H:%M:%S') }, ); # this points to another table; don't point to this table # } elsif($tables{$column} and $subject_table ne $column) { print "\t$column matches another table\n" if $self->debug; # referring column is a primary key and this is NOT a _ref table: liner notes # load all of the _other_ columns in $subject_table into $object_class if($tables{$subject_table}->{$column}->{pri} && $subject_table !~ /_ref$/i) { for my $col (keys %{$tables{$subject_table}}) { # this is the primary column, so we can skip it # next if $col eq $column; print "\t\t$object_class->might_have($subject_table.'_'.$col, $subject_class, $col)\n" if $self->debug; $object_class->might_have($subject_table.'_'.$col, $subject_class, $col); # might_have } } else { print "\t\t$subject_class->has_a($column, $object_class)\n" if $self->debug; # this is a non-primary column so there's a has_a relationship here # $subject_class->has_a($column, $object_class); # has_a # as for the has_many in the reverse direction, is this mapping or simple? # if($subject_table =~ /_ref$/i) { # $subject_table is a mapping table my($other_column) = grep { $tables{$_} && $column ne $_ } # not the key, and not this col keys %{$tables{$subject_table}}; # get the column which points to the mapped table my $plural = PL($other_column); print "\t\t$object_class->has_many($plural, [ $subject_class, $other_column ] )\n" if $self->debug; $object_class->has_many($plural, [ $subject_class, $other_column ]); # has_many } else { # simple has_many my $plural = PL($subject_table); print "\t\t$object_class->has_many($plural, $subject_class)\n" if $self->debug; $object_class->has_many($plural, $subject_class); # has_many } } } } } } 1; __END__ # Below is stub documentation for your module. You'd better edit it!