Reputation: 662
I tried to count all products grouped by their types, using this SQL query:
select pt.id,
pt.name,
count(p.id) as n
from prd.product_types as pt,
prd.products as p
where pt.id = p.type
group by pt.id,
pt.name
order by pt.name
These two tables (prd.product_types and prd.products) are mapped by these two DBIC classes:
package ki::Schema::Result::ProductTypes;
use strict;
use warnings;
use utf8;
use Moose;
use MooseX::NonMoose;
use MooseX::MarkAsMethods autoclean => 1;
extends 'DBIx::Class::Core';
__PACKAGE__->load_components("InflateColumn::DateTime", "TimeStamp", "EncodedColumn");
__PACKAGE__->table("prd.product_types");
__PACKAGE__->add_columns(
"id",
{
data_type => "uuid",
default_value => \"uuid_generate_v4()",
is_nullable => 0,
size => 16,
},
"name",
{ data_type => "varchar", is_nullable => 0, size => 128 },
);
__PACKAGE__->set_primary_key("id");
__PACKAGE__->add_unique_constraint("uk_product_types", ["name"]);
__PACKAGE__->has_many(
"products",
"ki::Schema::Result::Products",
{ "foreign.type" => "self.id" },
{ cascade_copy => 0, cascade_delete => 0 },
);
__PACKAGE__->meta->make_immutable;
1;
and
package ki::Schema::Result::Products;
use strict;
use warnings;
use utf8;
use Moose;
use MooseX::NonMoose;
use MooseX::MarkAsMethods autoclean => 1;
extends 'DBIx::Class::Core';
__PACKAGE__->load_components("InflateColumn::DateTime", "TimeStamp", "EncodedColumn");
__PACKAGE__->table("prd.products");
__PACKAGE__->add_columns(
"id",
{
data_type => "uuid",
default_value => \"uuid_generate_v4()",
is_nullable => 0,
size => 16,
},
"type",
{ data_type => "uuid", is_foreign_key => 1, is_nullable => 0, size => 16 },
"name",
{ data_type => "varchar", is_nullable => 0, size => 128 },
);
__PACKAGE__->set_primary_key("id");
__PACKAGE__->add_unique_constraint("uk_products", ["name"]);
__PACKAGE__->belongs_to(
"type",
"ki::Schema::Result::ProductTypes",
{ id => "type" },
{ is_deferrable => 0, on_delete => "NO ACTION", on_update => "NO ACTION" },
);
__PACKAGE__->meta->make_immutable;
1;
When I set the join it in my Catalyst controller, I get the "No such relationship":
DBIx::Class::ResultSource::_resolve_join(): No such relationship prd.products on ProductTypes
The result set is this:
$c->stash(product_types_rs => $c->model('DB_T::ProductTypes'));
and the join looks like this:
$c->stash(categories => [$c->stash->{product_types_rs}->search({}, {
join => [qw/ prd.products /],
select => [ { count => 'prd.products.id' } ],
as => [qw/ n /],
group_by => [qw/ id /],
order_by => 'name ASC'
})]);
The query above is a one to one reproduction of the query artist-cd from the DBIC documentation. But there is something what I can not understand and it is wrong.
Back with some news:
the only way what worked up to now is this:
$c->stash(categories => [$c->stash->{product_types_rs}->search({},
{
'+select' => [{ COUNT => 'products.id', -as => 'n'}],
join => ['products'],
group_by => 'me.id',
order_by => { -desc => ' COUNT( products.id )' },
})])
and its SQL looks like this:
SELECT me.id, me.name, COUNT( products.id ) AS n
FROM prd.product_types me LEFT JOIN prd.products products
ON products.type = me.id
GROUP BY me.id
ORDER BY COUNT( products.id ) DESC
I tested the generated SQL query and I get the wanted result. There is a remaining issue. In my view, I do use this
[% FOREACH category IN categories -%]
<li>([% category.n %])</li>
[% END -%]
loop to iterate the result set. But the value of n is not displayed even if the Dumper->Dump() displays me the correct datas:
...
'_column_data' => {
'name' => "XTorckAlsa",
'n' => 1,
'id' => '88b94b12-4169-4964-9022-3eebbc5c05c5'
},
...
'_column_data' => {
'name' => "Pledicts11",
'n' => 5,
'id' => 'ebda7223-99e3-48d6-b662-c7d17d124787'
},
I am still missing something.
Upvotes: 1
Views: 113
Reputation: 2204
You named the relationship products
so that's what you need to use in the join
parameter. A table name is never used in any DBIC parameter, only in the ResultSource definition.
You also never have more than a single dot in one of the parameters.
It's recommended to use columns
instead of select
and as
because it's less prone to user errors.
order_by
supports non-literal SQL syntax as well which is
[{ -asc => 'columnname' }]
Upvotes: 1