package Koha::Objects; # Copyright ByWater Solutions 2014 # # This file is part of Koha. # # Koha is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 3 of the License, or # (at your option) any later version. # # Koha is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with Koha; if not, see . use Modern::Perl; use Carp qw( carp ); use List::MoreUtils qw( none ); use Class::Inspector; use Koha::Database; use Koha::Exceptions::Object; =head1 NAME Koha::Objects - Koha Object set base class =head1 SYNOPSIS use Koha::Objects; my $objects = Koha::Objects->search({ borrowernumber => $borrowernumber}); =head1 DESCRIPTION This class must be subclassed. =head1 API =head2 Class Methods =cut =head3 Koha::Objects->new(); my $object = Koha::Objects->new(); =cut sub new { my ($class) = @_; my $self = {}; bless( $self, $class ); } =head3 Koha::Objects->_new_from_dbic(); my $object = Koha::Objects->_new_from_dbic( $resultset ); =cut sub _new_from_dbic { my ( $class, $resultset ) = @_; my $self = { _resultset => $resultset }; bless( $self, $class ); } =head3 Koha::Objects->find(); Similar to DBIx::Class::ResultSet->find this method accepts: \%columns_values | @pk_values, { key => $unique_constraint, %attrs }? Strictly speaking, columns_values should only refer to columns under an unique constraint. It returns undef if no results were found my $object = Koha::Objects->find( { col1 => $val1, col2 => $val2 } ); my $object = Koha::Objects->find( $id ); my $object = Koha::Objects->find( $idpart1, $idpart2, $attrs ); # composite PK =cut sub find { my ( $self, @pars ) = @_; my $object; unless ( !@pars || none { defined($_) } @pars ) { my $result = $self->_resultset()->find(@pars); if ($result) { $object = $self->object_class($result)->_new_from_dbic($result); } } return $object; } =head3 Koha::Objects->find_or_create(); my $object = Koha::Objects->find_or_create( $attrs ); =cut sub find_or_create { my ( $self, $params ) = @_; my $result = $self->_resultset->find_or_create($params); return unless $result; my $object = $self->object_class($result)->_new_from_dbic($result); return $object; } =head3 search # scalar context my $objects = Koha::Objects->search([$params, $attributes]); while (my $object = $objects->next) { do_stuff($object); } This B the I class, and generates a resultset based on the query I<$params> and I<$attributes> that are passed (like in DBIC). =cut sub search { my ( $self, $params, $attributes ) = @_; my $class = ref($self) ? ref($self) : $self; my $rs = $self->_resultset()->search( $params, $attributes ); return $class->_new_from_dbic($rs); } =head3 search_related my $objects = Koha::Objects->search_related( $rel_name, $cond?, \%attrs? ); Searches the specified relationship, optionally specifying a condition and attributes for matching records. =cut sub search_related { my ( $self, $rel_name, @params ) = @_; return if !$rel_name; my $rs = $self->_resultset()->search_related( $rel_name, @params ); return if !$rs; my $object_class = _get_objects_class( $rs->result_class ); eval "require $object_class"; return _new_from_dbic( $object_class, $rs ); } =head3 delete =cut sub delete { my ( $self, $params ) = @_; if ( Class::Inspector->function_exists( $self->object_class, 'delete' ) ) { my $objects_deleted; $self->_resultset->result_source->schema->txn_do( sub { $self->reset; # If we iterated already over the set while ( my $o = $self->next ) { $o->delete($params); $objects_deleted++; } } ); return $objects_deleted; } return $self->_resultset->delete; } =head3 update my $objects = Koha::Objects->new; # or Koha::Objects->search $objects->update( $fields, [ { no_triggers => 0/1 } ] ); This method overloads the DBIC inherited one so if code-level triggers exist (through the use of an overloaded I or I method in the Koha::Object based class) those are called in a loop on the resultset. If B is passed and I, then the DBIC update method is called directly. This feature is important for performance, in cases where no code-level triggers should be triggered. The developer will explicitly ask for this and QA should catch wrong uses as well. =cut sub update { my ( $self, $fields, $options ) = @_; Koha::Exceptions::Object::NotInstantiated->throw( method => 'update', class => $self ) unless ref $self; my $no_triggers = $options->{no_triggers}; if ( !$no_triggers && ( Class::Inspector->function_exists( $self->object_class, 'update' ) or Class::Inspector->function_exists( $self->object_class, 'store' ) ) ) { my $objects_updated; $self->_resultset->result_source->schema->txn_do( sub { while ( my $o = $self->next ) { $o->update($fields); $objects_updated++; } } ); return $objects_updated; } return $self->_resultset->update($fields); } =head3 filter_by_last_update my $filtered_objects = $objects->filter_by_last_update({ from => $from_datetime, to => $to_datetime, days|older_than => $days, min_days => $days, younger_than => $days, exact_days => $days }); You should pass at least one of the parameters: from, to, days|older_than, min_days or younger_than. Make sure that they do not conflict with each other to get meaningful results. Note: from, to and min_days are inclusive! And by nature days|older_than and younger_than are exclusive. The from and to parameters must be DateTime objects. =cut sub filter_by_last_update { my ( $self, $params ) = @_; my $timestamp_column_name = $params->{timestamp_column_name} || 'timestamp'; my $conditions; Koha::Exceptions::MissingParameter->throw("Please pass: days|from|to|older_than|younger_than|exact_days") unless grep { exists $params->{$_} } qw/days from to older_than younger_than min_days exact_days/; foreach my $key (qw(from to)) { if ( exists $params->{$key} and ref $params->{$key} ne 'DateTime' ) { Koha::Exceptions::WrongParameter->throw("'$key' parameter must be a DateTime object"); } } my $dtf = Koha::Database->new->schema->storage->datetime_parser; foreach my $p (qw/days older_than younger_than min_days exact_days/) { next if !exists $params->{$p}; my $days = $params->{$p}; my $operator = { days => '<', older_than => '<', min_days => '<=', 'exact_days' => "=" }->{$p} // '>'; $conditions->{$operator} = \[ 'DATE_SUB(CURDATE(), INTERVAL ? DAY)', $days ]; } if ( exists $params->{from} ) { $conditions->{'>='} = $dtf->format_datetime( $params->{from} ); } if ( exists $params->{to} ) { $conditions->{'<='} = $dtf->format_datetime( $params->{to} ); } return $self->search( { $timestamp_column_name => $conditions } ); } =head3 single my $object = Koha::Objects->search({}, { rows => 1 })->single Returns one and only one object that is part of this set. Returns undef if there are no objects found. This is optimal as it will grab the first returned result without instantiating a cursor. See: http://search.cpan.org/dist/DBIx-Class/lib/DBIx/Class/Manual/Cookbook.pod#Retrieve_one_and_only_one_row_from_a_resultset =cut sub single { my ($self) = @_; my $single = $self->_resultset()->single; return unless $single; return $self->object_class($single)->_new_from_dbic($single); } =head3 Koha::Objects->next(); my $object = Koha::Objects->next(); Returns the next object that is part of this set. Returns undef if there are no more objects to return. =cut sub next { my ($self) = @_; my $result = $self->_resultset()->next(); return unless $result; my $object = $self->object_class($result)->_new_from_dbic($result); return $object; } =head3 Koha::Objects->last; my $object = Koha::Objects->last; Returns the last object that is part of this set. Returns undef if there are no object to return. =cut sub last { my ($self) = @_; my $count = $self->_resultset->count; return unless $count; my ($result) = $self->_resultset->slice( $count - 1, $count - 1 ); my $object = $self->object_class($result)->_new_from_dbic($result); return $object; } =head3 empty my $empty_rs = Koha::Objects->new->empty; Sets the resultset empty. This is handy for consistency on method returns (e.g. if we know in advance we won't have results but want to keep returning an iterator). =cut sub empty { my ($self) = @_; Koha::Exceptions::Object::NotInstantiated->throw( method => 'empty', class => $self ) unless ref $self; $self = $self->search( \'0 = 1' ); $self->_resultset()->set_cache( [] ); return $self; } =head3 Koha::Objects->reset(); Koha::Objects->reset(); resets iteration so the next call to next() will start again with the first object in a set. =cut sub reset { my ($self) = @_; $self->_resultset()->reset(); return $self; } =head3 Koha::Objects->as_list(); Koha::Objects->as_list(); Returns an arrayref of the objects in this set. =cut sub as_list { my ($self) = @_; my @dbic_rows = $self->_resultset()->all(); my @objects = $self->_wrap(@dbic_rows); return wantarray ? @objects : \@objects; } =head3 Koha::Objects->unblessed Returns an unblessed representation of objects. =cut sub unblessed { my ($self) = @_; return [ map { $_->unblessed } $self->as_list ]; } =head3 Koha::Objects->get_column Return all the values of this set for a given column =cut sub get_column { my ( $self, $column_name ) = @_; return $self->_resultset->get_column($column_name)->all; } =head3 Koha::Objects->TO_JSON Returns an unblessed representation of objects, suitable for JSON output. =cut sub TO_JSON { my ($self) = @_; return [ map { $_->TO_JSON } $self->as_list ]; } =head3 Koha::Objects->to_api Returns a representation of the objects, suitable for API output . =cut sub to_api { my ( $self, $params ) = @_; return [ map { $_->to_api($params) } $self->as_list ]; } =head3 attributes_from_api my $attributes = $objects->attributes_from_api( $api_attributes ); Translates attributes from the API to DBIC =cut sub attributes_from_api { my ( $self, $attributes ) = @_; $self->{_singular_object} ||= $self->object_class->new(); return $self->{_singular_object}->attributes_from_api($attributes); } =head3 from_api_mapping my $mapped_attributes_hash = $objects->from_api_mapping; Attributes map from the API to DBIC =cut sub from_api_mapping { my ($self) = @_; $self->{_singular_object} ||= $self->object_class->new(); return $self->{_singular_object}->from_api_mapping; } =head3 prefetch_whitelist my $whitelist = $object->prefetch_whitelist() Returns a hash of prefetchable subs and the type it returns =cut sub prefetch_whitelist { my ($self) = @_; $self->{_singular_object} ||= $self->object_class->new(); $self->{_singular_object}->prefetch_whitelist; } =head3 Koha::Objects->_wrap wraps the DBIC object in a corresponding Koha object =cut sub _wrap { my ( $self, @dbic_rows ) = @_; my @objects = map { $self->object_class($_)->_new_from_dbic($_) } @dbic_rows; return @objects; } =head3 Koha::Objects->_resultset Returns the internal resultset or creates it if undefined =cut sub _resultset { my ($self) = @_; if ( ref($self) ) { $self->{_resultset} ||= Koha::Database->new()->schema()->resultset( $self->_type() ); return $self->{_resultset}; } else { return Koha::Database->new()->schema()->resultset( $self->_type() ); } } sub _get_objects_class { my ($type) = @_; return unless $type; if ( $type->can('koha_objects_class') ) { return $type->koha_objects_class; } $type =~ s|Schema::Result::||; return "${type}s"; } =head3 columns my @columns = Koha::Objects->columns Return the table columns =cut sub columns { my ($class) = @_; return Koha::Database->new->schema->resultset( $class->_type )->result_source->columns; } =head3 AUTOLOAD The autoload method is used call DBIx::Class method on a resultset. Important: If you plan to use one of the DBIx::Class methods you must provide relevant tests in t/db_dependent/Koha/Objects.t Currently count, is_paged, pager, result_class, single and slice are covered. =cut sub AUTOLOAD { my ( $self, @params ) = @_; my @known_methods = qw( count is_paged pager result_class single slice ); my $method = our $AUTOLOAD; $method =~ s/.*:://; unless ( grep { $_ eq $method } @known_methods ) { my $class = ref($self) ? ref($self) : $self; Koha::Exceptions::Object::MethodNotCoveredByTests->throw( error => sprintf( "The method %s->%s is not covered by tests!", $class, $method ), show_trace => 1 ); } my $r = eval { $self->_resultset->$method(@params) }; if ($@) { carp "No method $method found for " . ref($self) . " " . $@; return; } return $r; } =head3 _type The _type method must be set for all child classes. The value returned by it should be the DBIC resultset name. For example, for holds, _type should return 'Reserve'. =cut sub _type { } =head3 _polymorphic_field sub _polymorphic_field { return 'transport'; } The _polymorphic_field method must be set for all child classes that implement their own polymorphic children. It should return the field name that distinguishes the classes. =cut =head3 _polymorphic_map sub _polymorphic_map { return { sftp => 'Koha::File::Transport::SFTP', ftp => 'Koha::File::Transport::FTP', }; } The _polymorphic_map method must be implemented by all child classes that implement their own polymorphic children. It should return a simple hashmap mapping for field value to class name for the polymorphic class using the value from the _polymorphic_field defined above. =cut =head3 object_class This method must be set for all child classes. The value returned by it should be the name of the Koha object class that is returned by this class. For example, for holds, object_class should return 'Koha::Hold'. =cut sub object_class { } sub DESTROY { } =head1 AUTHOR Kyle M Hall =cut 1;