=head1 NAME
-DBIx::Class::ResultClass::HashRefInflator
+DBIx::Class::ResultClass::HashRefInflator - Get raw hashrefs from a resultset
=head1 SYNOPSIS
- my $rs = $schema->resultset('CD');
+ use DBIx::Class::ResultClass::HashRefInflator;
+ my $rs = $schema->resultset('CD');
$rs->result_class('DBIx::Class::ResultClass::HashRefInflator');
+ while (my $hashref = $rs->next) {
+ ...
+ }
+
+ OR as an attribute:
+
+ my $rs = $schema->resultset('CD')->search({}, {
+ result_class => 'DBIx::Class::ResultClass::HashRefInflator',
+ });
+ while (my $hashref = $rs->next) {
+ ...
+ }
=head1 DESCRIPTION
-DBIx::Class is not built for speed: it's built for convenience and
-ease of use. But sometimes you just need to get the data, and skip the
-fancy objects. That is what this class provides.
+DBIx::Class is faster than older ORMs like Class::DBI but it still isn't
+designed primarily for speed. Sometimes you need to quickly retrieve the data
+from a massive resultset, while skipping the creation of fancy result objects.
+Specifying this class as a C<result_class> for a resultset will change C<< $rs->next >>
+to return a plain data hash-ref (or a list of such hash-refs if C<< $rs->all >> is used).
-There are two ways of using this class.
+There are two ways of applying this class to a resultset:
=over
=back
+=cut
+
+##############
+# NOTE
+#
+# Generally people use this to gain as much speed as possible. If a new &mk_hash is
+# implemented, it should be benchmarked using the maint/benchmark_hashrefinflator.pl
+# script (in addition to passing all tests of course :)
+
+# This coderef is a simple recursive function
+# Arguments: ($me, $prefetch, $is_root) from inflate_result() below
+my $mk_hash;
+$mk_hash = sub {
+
+ my $hash = {
+ # the main hash could be an undef if we are processing a skipped-over join
+ $_[0] ? %{$_[0]} : (),
+
+ # the second arg is a hash of arrays for each prefetched relation
+ map {
+ ref $_[1]->{$_}[0] eq 'ARRAY' # multi rel or not?
+ ? ( $_ => [ map
+ { $mk_hash->(@$_) || () }
+ @{$_[1]->{$_}}
+ ] )
+ : ( $_ => $mk_hash->( @{$_[1]->{$_}} ) )
+
+ } ( $_[1] ? ( keys %{$_[1]} ) : () )
+ };
+
+ # if there is at least one defined column *OR* we are at the root of
+ # the resultset - consider the result real (and not an emtpy has_many
+ # rel containing one empty hashref)
+ # an empty arrayref is an empty multi-sub-prefetch - don't consider
+ # those either
+ return $hash if $_[2];
+
+ for (values %$hash) {
+ return $hash if (
+ defined $_
+ and
+ (ref $_ ne 'ARRAY' or scalar @$_)
+ );
+ }
+
+ return undef;
+};
+
=head1 METHODS
=head2 inflate_result
-Inflates the result and prefetched data into a hash-ref using L<mk_hash>.
+Inflates the result and prefetched data into a hash-ref (invoked by L<DBIx::Class::ResultSet>)
=cut
+##################################################################################
+# inflate_result is invoked as:
+# HRI->inflate_result ($resultsource_instance, $main_data_hashref, $prefetch_data_hashref)
sub inflate_result {
- my ($self, $source, $me, $prefetch) = @_;
-
- return mk_hash($me, $prefetch);
+ return $mk_hash->($_[2], $_[3], 'is_root');
}
-=head2 mk_hash
-
-This does all the work of inflating the (pre)fetched data.
-=cut
+=head1 CAVEATS
-sub mk_hash {
- my ($me, $rest) = @_;
-
- # $me is the hashref of cols/data from the immediate resultsource
- # $rest is a deep hashref of all the data from the prefetched
- # related sources.
-
- # to avoid emtpy has_many rels contain one empty hashref
- return undef if (not keys %$me);
-
- my $def;
-
- foreach (values %$me) {
- if (defined $_) {
- $def = 1;
- last;
- }
- }
- return undef unless $def;
-
- return { %$me,
- map {
- ( $_ =>
- ref($rest->{$_}[0]) eq 'ARRAY'
- ? [ grep defined, map mk_hash(@$_), @{$rest->{$_}} ]
- : mk_hash( @{$rest->{$_}} )
- )
- } keys %$rest
- };
-}
+=over
-=head1 CAVEAT
+=item *
This will not work for relationships that have been prefetched. Consider the
following:
my $cds = $artist->cds;
$cds->result_class('DBIx::Class::ResultClass::HashRefInflator');
- my $first = $cds->first;
+ my $first = $cds->first;
-C<$first> will B<not> be a hashref, it will be a normal CD row since
+C<$first> will B<not> be a hashref, it will be a normal CD row since
HashRefInflator only affects resultsets at inflation time, and prefetch causes
relations to be inflated when the master C<$artist> row is inflated.
+=item *
+
+Column value inflation, e.g., using modules like
+L<DBIx::Class::InflateColumn::DateTime>, is not performed.
+The returned hash contains the raw database values.
+
+=back
+
=cut
1;