X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FDBIx%2FClass%2FResultSet.pm;h=c944b4567a4a0b9b213be09ce68050d95b4d868e;hb=52864fbd5c6035f8f3961173ad05c2cc58fe9a34;hp=29dea9e153d1e3af6a5a66f50b4420f26e1ac238;hpb=a779441cbba1a366b4df7dd574966dae88e43ecb;p=dbsrgits%2FDBIx-Class.git diff --git a/lib/DBIx/Class/ResultSet.pm b/lib/DBIx/Class/ResultSet.pm index 29dea9e..c944b45 100644 --- a/lib/DBIx/Class/ResultSet.pm +++ b/lib/DBIx/Class/ResultSet.pm @@ -4,7 +4,6 @@ use strict; use warnings; use base qw/DBIx::Class/; use DBIx::Class::Carp; -use DBIx::Class::Exception; use DBIx::Class::ResultSetColumn; use Scalar::Util qw/blessed weaken/; use Try::Tiny; @@ -34,12 +33,12 @@ DBIx::Class::ResultSet - Represents a query used for fetching a set of results. =head1 SYNOPSIS - my $users_rs = $schema->resultset('User'); + my $users_rs = $schema->resultset('User'); while( $user = $users_rs->next) { print $user->username; } - my $registered_users_rs = $schema->resultset('User')->search({ registered => 1 }); + my $registered_users_rs = $schema->resultset('User')->search({ registered => 1 }); my @cds_in_2005 = $schema->resultset('CD')->search({ year => 2005 })->all(); =head1 DESCRIPTION @@ -138,11 +137,15 @@ another. =head3 Resolving conditions and attributes -When a resultset is chained from another resultset, conditions and -attributes with the same keys need resolving. +When a resultset is chained from another resultset (ie: +Csearch(\%extra_cond, \%attrs)>), conditions +and attributes with the same keys need resolving. + +If any of L, L, L are present, they reset the +original selection, and start the selection "clean". -L, L, L, L attributes are merged -into the existing ones from the original resultset. +The L, L, L, L, L attributes +are merged into the existing ones from the original resultset. The L and L attributes, and any search conditions, are merged with an SQL C to the existing condition from the original @@ -191,9 +194,9 @@ See: L, L, L, L, L. =over 4 -=item Arguments: $source, \%$attrs +=item Arguments: L<$source|DBIx::Class::ResultSource>, L<\%attrs?|/ATTRIBUTES> -=item Return Value: $rs +=item Return Value: L<$resultset|/search> =back @@ -202,16 +205,31 @@ L) and an attribute hash (see L below). Does not perform any queries -- these are executed as needed by the other methods. -Generally you won't need to construct a resultset manually. You'll -automatically get one from e.g. a L called in scalar context: +Generally you never construct a resultset manually. Instead you get one +from e.g. a +C<< $schema->L('$source_name') >> +or C<< $another_resultset->L(...) >> (the later called in +scalar context): my $rs = $schema->resultset('CD')->search({ title => '100th Window' }); -IMPORTANT: If called on an object, proxies to new_result instead so +=over + +=item WARNING + +If called on an object, proxies to L instead, so my $cd = $schema->resultset('CD')->new({ title => 'Spoon' }); -will return a CD object, not a ResultSet. +will return a CD object, not a ResultSet, and is equivalent to: + + my $cd = $schema->resultset('CD')->new_result({ title => 'Spoon' }); + +Please also keep in mind that many internals call L directly, +so overloading this method with the idea of intercepting new result object +creation B. See also warning pertaining to L. + +=back =cut @@ -254,9 +272,9 @@ sub new { =over 4 -=item Arguments: $cond, \%attrs? +=item Arguments: L<$cond|DBIx::Class::SQLMaker> | undef, L<\%attrs?|/ATTRIBUTES> -=item Return Value: $resultset (scalar context) || @row_objs (list context) +=item Return Value: $resultset (scalar context) | L<@result_objs|DBIx::Class::Manual::ResultClass> (list context) =back @@ -267,7 +285,8 @@ sub new { # year = 2005 OR year = 2004 In list context, C<< ->all() >> is called implicitly on the resultset, thus -returning a list of row objects instead. To avoid that, use L. +returning a list of L objects instead. +To avoid that, use L. If you need to pass in additional attributes but no additional condition, call it as C. @@ -289,7 +308,7 @@ For more help on using joins with search, see L. Note that L does not process/deflate any of the values passed in the L-compatible search condition structure. This is unlike other -condition-bound methods L, L and L. The user must ensure +condition-bound methods L, L and L. The user must ensure manually that any value passed to this method will stringify to something the RDBMS knows how to deal with. A notable example is the handling of L objects, for more info see: @@ -324,9 +343,9 @@ sub search { =over 4 -=item Arguments: $cond, \%attrs? +=item Arguments: L<$cond|DBIx::Class::SQLMaker>, L<\%attrs?|/ATTRIBUTES> -=item Return Value: $resultset +=item Return Value: L<$resultset|/search> =back @@ -338,20 +357,36 @@ always return a resultset, even in list context. sub search_rs { my $self = shift; - # Special-case handling for (undef, undef). - if ( @_ == 2 && !defined $_[1] && !defined $_[0] ) { - @_ = (); - } + my $rsrc = $self->result_source; + my ($call_cond, $call_attrs); - my $call_attrs = {}; - if (@_ > 1) { - if (ref $_[-1] eq 'HASH') { - # copy for _normalize_selection - $call_attrs = { %{ pop @_ } }; - } - elsif (! defined $_[-1] ) { - pop @_; # search({}, undef) + # Special-case handling for (undef, undef) or (undef) + # Note that (foo => undef) is valid deprecated syntax + @_ = () if not scalar grep { defined $_ } @_; + + # just a cond + if (@_ == 1) { + $call_cond = shift; + } + # fish out attrs in the ($condref, $attr) case + elsif (@_ == 2 and ( ! defined $_[0] or (ref $_[0]) ne '') ) { + ($call_cond, $call_attrs) = @_; + } + elsif (@_ % 2) { + $self->throw_exception('Odd number of arguments to search') + } + # legacy search + elsif (@_) { + carp_unique 'search( %condition ) is deprecated, use search( \%condition ) instead' + unless $rsrc->result_class->isa('DBIx::Class::CDBICompat'); + + for my $i (0 .. $#_) { + next if $i % 2; + $self->throw_exception ('All keys in condition key/value pairs must be plain scalars') + if (! defined $_[$i] or ref $_[$i] ne ''); } + + $call_cond = { @_ }; } # see if we can keep the cache (no $rs changes) @@ -367,8 +402,6 @@ sub search_rs { $cache = $self->get_cache; } - my $rsrc = $self->result_source; - my $old_attrs = { %{$self->{attrs}} }; my $old_having = delete $old_attrs->{having}; my $old_where = delete $old_attrs->{where}; @@ -376,7 +409,10 @@ sub search_rs { my $new_attrs = { %$old_attrs }; # take care of call attrs (only if anything is changing) - if (keys %$call_attrs) { + if ($call_attrs and keys %$call_attrs) { + + # copy for _normalize_selection + $call_attrs = { %$call_attrs }; my @selector_attrs = qw/select as columns cols +select +as +columns include_columns/; @@ -423,28 +459,6 @@ sub search_rs { } - # rip apart the rest of @_, parse a condition - my $call_cond = do { - - if (ref $_[0] eq 'HASH') { - (keys %{$_[0]}) ? $_[0] : undef - } - elsif (@_ == 1) { - $_[0] - } - elsif (@_ % 2) { - $self->throw_exception('Odd number of arguments to search') - } - else { - +{ @_ } - } - - } if @_; - - if( @_ > 1 and ! $rsrc->result_class->isa('DBIx::Class::CDBICompat') ) { - carp_unique 'search( %condition ) is deprecated, use search( \%condition ) instead'; - } - for ($old_where, $call_cond) { if (defined $_) { $new_attrs->{where} = $self->_stack_cond ( @@ -617,11 +631,20 @@ sub _stack_cond { =head2 search_literal +B: C is provided for Class::DBI compatibility and +should only be used in that context. C is a convenience +method. It is equivalent to calling C<< $schema->search(\[]) >>, but if you +want to ensure columns are bound correctly, use L. + +See L and +L for searching techniques that do not +require C. + =over 4 -=item Arguments: $sql_fragment, @bind_values +=item Arguments: $sql_fragment, @standalone_bind_values -=item Return Value: $resultset (scalar context) || @row_objs (list context) +=item Return Value: L<$resultset|/search> (scalar context) | L<@result_objs|DBIx::Class::Manual::ResultClass> (list context) =back @@ -631,21 +654,11 @@ sub _stack_cond { Pass a literal chunk of SQL to be added to the conditional part of the resultset query. -CAVEAT: C is provided for Class::DBI compatibility and should -only be used in that context. C is a convenience method. -It is equivalent to calling $schema->search(\[]), but if you want to ensure -columns are bound correctly, use C. - Example of how to use C instead of C my @cds = $cd_rs->search_literal('cdid = ? AND (artist = ? OR artist = ?)', (2, 1, 2)); my @cds = $cd_rs->search(\[ 'cdid = ? AND (artist = ? OR artist = ?)', [ 'cdid', 2 ], [ 'artist', 1 ], [ 'artist', 2 ] ]); - -See L and -L for searching techniques that do not -require C. - =cut sub search_literal { @@ -654,16 +667,16 @@ sub search_literal { if ( @bind && ref($bind[-1]) eq 'HASH' ) { $attr = pop @bind; } - return $self->search(\[ $sql, map [ __DUMMY__ => $_ ], @bind ], ($attr || () )); + return $self->search(\[ $sql, map [ {} => $_ ], @bind ], ($attr || () )); } =head2 find =over 4 -=item Arguments: \%columns_values | @pk_values, \%attrs? +=item Arguments: \%columns_values | @pk_values, { key => $unique_constraint, L<%attrs|/ATTRIBUTES> }? -=item Return Value: $row_object | undef +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> | undef =back @@ -695,7 +708,7 @@ Note that this fallback behavior may be deprecated in further versions. If you need to search with arbitrary conditions - use L. If the query resulting from this fallback produces more than one row, a warning to the effect is issued, though only the first row is constructed and returned as -C<$row_object>. +C<$result_object>. In addition to C, L recognizes and applies standard L in the same way as L does. @@ -909,7 +922,7 @@ sub _build_unique_cond { and !$ENV{DBIC_NULLABLE_KEY_NOWARN} and - my @undefs = grep { ! defined $final_cond->{$_} } (keys %$final_cond) + my @undefs = sort grep { ! defined $final_cond->{$_} } (keys %$final_cond) ) { carp_unique ( sprintf ( "NULL/undef values supplied for requested unique constraint '%s' (NULL " @@ -927,9 +940,9 @@ sub _build_unique_cond { =over 4 -=item Arguments: $rel, $cond?, \%attrs? +=item Arguments: $rel_name, $cond?, L<\%attrs?|/ATTRIBUTES> -=item Return Value: $new_resultset (scalar context) || @row_objs (list context) +=item Return Value: L<$resultset|/search> (scalar context) | L<@result_objs|DBIx::Class::Manual::ResultClass> (list context) =back @@ -941,7 +954,7 @@ Searches the specified relationship, optionally specifying a condition and attributes for matching records. See L for more information. In list context, C<< ->all() >> is called implicitly on the resultset, thus -returning a list of row objects instead. To avoid that, use L. +returning a list of result objects instead. To avoid that, use L. See also L. @@ -968,7 +981,7 @@ sub search_related_rs { =item Arguments: none -=item Return Value: $cursor +=item Return Value: L<$cursor|DBIx::Class::Cursor> =back @@ -978,22 +991,23 @@ L for more information. =cut sub cursor { - my ($self) = @_; - - my $attrs = $self->_resolved_attrs_copy; + my $self = shift; - return $self->{cursor} - ||= $self->result_source->storage->select($attrs->{from}, $attrs->{select}, - $attrs->{where},$attrs); + return $self->{cursor} ||= do { + my $attrs = { %{$self->_resolved_attrs } }; + $self->result_source->storage->select( + $attrs->{from}, $attrs->{select}, $attrs->{where}, $attrs + ); + }; } =head2 single =over 4 -=item Arguments: $cond? +=item Arguments: L<$cond?|DBIx::Class::SQLMaker> -=item Return Value: $row_object | undef +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> | undef =back @@ -1036,7 +1050,7 @@ sub single { $self->throw_exception('single() only takes search conditions, no attributes. You want ->search( $cond, $attrs )->single()'); } - my $attrs = $self->_resolved_attrs_copy; + my $attrs = { %{$self->_resolved_attrs} }; $self->throw_exception( 'single() can not be used on resultsets prefetching has_many. Use find( \%cond ) or next() instead' @@ -1100,9 +1114,9 @@ sub _collapse_query { =over 4 -=item Arguments: $cond? +=item Arguments: L<$cond?|DBIx::Class::SQLMaker> -=item Return Value: $resultsetcolumn +=item Return Value: L<$resultsetcolumn|DBIx::Class::ResultSetColumn> =back @@ -1122,9 +1136,9 @@ sub get_column { =over 4 -=item Arguments: $cond, \%attrs? +=item Arguments: L<$cond|DBIx::Class::SQLMaker>, L<\%attrs?|/ATTRIBUTES> -=item Return Value: $resultset (scalar context) || @row_objs (list context) +=item Return Value: L<$resultset|/search> (scalar context) | L<@result_objs|DBIx::Class::Manual::ResultClass> (list context) =back @@ -1167,7 +1181,7 @@ sub search_like { =item Arguments: $first, $last -=item Return Value: $resultset (scalar context) || @row_objs (list context) +=item Return Value: L<$resultset|/search> (scalar context) | L<@result_objs|DBIx::Class::Manual::ResultClass> (list context) =back @@ -1196,7 +1210,7 @@ sub slice { =item Arguments: none -=item Return Value: $result | undef +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> | undef =back @@ -1258,23 +1272,28 @@ sub _construct_objects { my $rsrc = $self->result_source; my $attrs = $self->_resolved_attrs; + + if (!$fetch_all and ! $attrs->{order_by} and $attrs->{collapse}) { + # default order for collapsing unless the user asked for something + $attrs->{order_by} = [ map { join '.', $attrs->{alias}, $_} $rsrc->primary_columns ]; + $attrs->{_ordered_for_collapse} = 1; + $attrs->{_order_is_artificial} = 1; + } + my $cursor = $self->cursor; # this will be used as both initial raw-row collector AND as a RV of # _construct_objects. Not regrowing the array twice matters a lot... # a suprising amount actually - my $rows = (delete $self->{stashed_rows}) || []; + my $rows = delete $self->{stashed_rows}; + if ($fetch_all) { # FIXME SUBOPTIMAL - we can do better, cursor->next/all (well diff. methods) should return a ref - $rows = [ @$rows, $cursor->all ]; + $rows = [ ($rows ? @$rows : ()), $cursor->all ]; } - elsif (!$attrs->{collapse}) { - # FIXME SUBOPTIMAL - we can do better, cursor->next/all (well diff. methods) should return a ref - push @$rows, do { my @r = $cursor->next; @r ? \@r : () } - unless @$rows; - } - else { - $attrs->{_ordered_for_collapse} ||= (!$attrs->{order_by}) ? undef : do { + elsif( $attrs->{collapse} ) { + + $attrs->{_ordered_for_collapse} = (!$attrs->{order_by}) ? 0 : do { my $st = $rsrc->schema->storage; my @ord_cols = map { $_->[0] } @@ -1301,64 +1320,106 @@ sub _construct_objects { { $colinfos->{$_}{-colname} => $colinfos->{$_} } @ord_cols })) ? 1 : 0; - }; + } unless defined $attrs->{_ordered_for_collapse}; - if ($attrs->{_ordered_for_collapse}) { - push @$rows, do { my @r = $cursor->next; @r ? \@r : () }; - } - # instead of looping over ->next, use ->all in stealth mode - # FIXME - encapsulation breach, got to be a better way - elsif (! $cursor->{done}) { - push @$rows, $cursor->all; - $cursor->{done} = 1; + if (! $attrs->{_ordered_for_collapse}) { $fetch_all = 1; + + # instead of looping over ->next, use ->all in stealth mode + # *without* calling a ->reset afterwards + # FIXME - encapsulation breach, got to be a better way + if (! $cursor->{_done}) { + $rows = [ ($rows ? @$rows : ()), $cursor->all ]; + $cursor->{_done} = 1; + } } } - return undef unless @$rows; + if (! $fetch_all and ! @{$rows||[]} ) { + # FIXME SUBOPTIMAL - we can do better, cursor->next/all (well diff. methods) should return a ref + if (scalar (my @r = $cursor->next) ) { + $rows = [ \@r ]; + } + } - my $res_class = $self->result_class; - my $inflator = $res_class->can ('inflate_result') - or $self->throw_exception("Inflator $res_class does not provide an inflate_result() method"); + return undef unless @{$rows||[]}; + + my @extra_collapser_args; + if ($attrs->{collapse} and ! $fetch_all ) { + + @extra_collapser_args = ( + # FIXME SUBOPTIMAL - we can do better, cursor->next/all (well diff. methods) should return a ref + sub { my @r = $cursor->next or return; \@r }, # how the collapser gets more rows + ($self->{stashed_rows} = []), # where does it stuff excess + ); + } + + # hotspot - skip the setter + my $res_class = $self->_result_class; + + my $inflator_cref = $self->{_result_inflator}{cref} ||= do { + $res_class->can ('inflate_result') + or $self->throw_exception("Inflator $res_class does not provide an inflate_result() method"); + }; my $infmap = $attrs->{as}; - if (!$attrs->{collapse} and $attrs->{_single_object_inflation}) { - # construct a much simpler array->hash folder for the one-table cases right here + $self->{_result_inflator}{is_hri} = do { ( $inflator_cref == ( + require DBIx::Class::ResultClass::HashRefInflator + && + DBIx::Class::ResultClass::HashRefInflator->can('inflate_result') + ) ) ? 1 : 0 + } unless defined $self->{_result_inflator}{is_hri}; + if ($attrs->{_single_resultclass_inflation}) { + # construct a much simpler array->hash folder for the one-table cases right here + if ($self->{_result_inflator}{is_hri}) { + for my $r (@$rows) { + $r = { map { $infmap->[$_] => $r->[$_] } 0..$#$infmap }; + } + } # FIXME SUBOPTIMAL this is a very very very hot spot # while rather optimal we can *still* do much better, by - # building a smarter [Row|HRI]::inflate_result(), and + # building a smarter Row::inflate_result(), and # switch to feeding it data via a much leaner interface # # crude unscientific benchmarking indicated the shortcut eval is not worth it for # this particular resultset size - if (@$rows < 60) { - my @as_idx = 0..$#$infmap; + elsif (@$rows < 60) { for my $r (@$rows) { - $r = $inflator->($res_class, $rsrc, { map { $infmap->[$_] => $r->[$_] } @as_idx } ); + $r = $inflator_cref->($res_class, $rsrc, { map { $infmap->[$_] => $r->[$_] } (0..$#$infmap) } ); } } else { eval sprintf ( - '$_ = $inflator->($res_class, $rsrc, { %s }) for @$rows', + '$_ = $inflator_cref->($res_class, $rsrc, { %s }) for @$rows', join (', ', map { "\$infmap->[$_] => \$_->[$_]" } 0..$#$infmap ) ); } } - else { - ($self->{_row_parser} ||= eval sprintf 'sub { %s }', $rsrc->_mk_row_parser({ + # Special-case multi-object HRI (we always prune) + elsif ($self->{_result_inflator}{is_hri}) { + ( $self->{_row_parser}{hri} ||= $rsrc->_mk_row_parser({ + eval => 1, inflate_map => $infmap, selection => $attrs->{select}, collapse => $attrs->{collapse}, - }) or die $@)->($rows, $fetch_all ? () : ( - # FIXME SUBOPTIMAL - we can do better, cursor->next/all (well diff. methods) should return a ref - sub { my @r = $cursor->next or return; \@r }, # how the collapser gets more rows - ($self->{stashed_rows} = []), # where does it stuff excess - )); # modify $rows in-place, shrinking/extending as necessary + premultiplied => $attrs->{_main_source_premultiplied}, + hri_style => 1, + }) )->($rows, @extra_collapser_args); + } + # Regular multi-object + else { - $_ = $inflator->($res_class, $rsrc, @$_) for @$rows; + ( $self->{_row_parser}{classic} ||= $rsrc->_mk_row_parser({ + eval => 1, + inflate_map => $infmap, + selection => $attrs->{select}, + collapse => $attrs->{collapse}, + premultiplied => $attrs->{_main_source_premultiplied}, + }) )->($rows, @extra_collapser_args); + $_ = $inflator_cref->($res_class, $rsrc, @$_) for @$rows; } # CDBI compat stuff @@ -1373,9 +1434,9 @@ sub _construct_objects { =over 4 -=item Arguments: $result_source? +=item Arguments: L<$result_source?|DBIx::Class::ResultSource> -=item Return Value: $result_source +=item Return Value: L<$result_source|DBIx::Class::ResultSource> =back @@ -1392,7 +1453,7 @@ is derived. =back -An accessor for the class to use when creating row objects. Defaults to +An accessor for the class to use when creating result objects. Defaults to C<< result_source->result_class >> - which in most cases is the name of the L<"table"|DBIx::Class::Manual::Glossary/"ResultSource"> class. @@ -1406,6 +1467,7 @@ in the original source class will not run. sub result_class { my ($self, $result_class) = @_; if ($result_class) { + unless (ref $result_class) { # don't fire this for an object $self->ensure_class_loaded($result_class); } @@ -1414,6 +1476,8 @@ sub result_class { # permit the user to set result class on one result set only; it only # chains if provided to search() #$self->{attrs}{result_class} = $result_class if ref $self; + + delete $self->{_result_inflator}; } $self->_result_class; } @@ -1422,7 +1486,7 @@ sub result_class { =over 4 -=item Arguments: $cond, \%attrs?? +=item Arguments: L<$cond|DBIx::Class::SQLMaker>, L<\%attrs?|/ATTRIBUTES> =item Return Value: $count @@ -1439,7 +1503,7 @@ sub count { return $self->search(@_)->count if @_ and defined $_[0]; return scalar @{ $self->get_cache } if $self->get_cache; - my $attrs = $self->_resolved_attrs_copy; + my $attrs = { %{ $self->_resolved_attrs } }; # this is a little optimization - it is faster to do the limit # adjustments in software, instead of a subquery @@ -1465,9 +1529,9 @@ sub count { =over 4 -=item Arguments: $cond, \%attrs?? +=item Arguments: L<$cond|DBIx::Class::SQLMaker>, L<\%attrs?|/ATTRIBUTES> -=item Return Value: $count_rs +=item Return Value: L<$count_rs|DBIx::Class::ResultSetColumn> =back @@ -1573,18 +1637,22 @@ sub _count_subq_rs { my ($lquote, $rquote, $sep) = map { quotemeta $_ } ($sql_maker->_quote_chars, $sql_maker->name_sep); - my $sql = $sql_maker->_parse_rs_attrs ({ having => $attrs->{having} }); + my $having_sql = $sql_maker->_parse_rs_attrs ({ having => $attrs->{having} }); + my %seen_having; # search for both a proper quoted qualified string, for a naive unquoted scalarref # and if all fails for an utterly naive quoted scalar-with-function - while ($sql =~ / + while ($having_sql =~ / $rquote $sep $lquote (.+?) $rquote | [\s,] \w+ \. (\w+) [\s,] | [\s,] $lquote (.+?) $rquote [\s,] /gx) { - push @parts, ($1 || $2 || $3); # one of them matched if we got here + my $part = $1 || $2 || $3; # one of them matched if we got here + unless ($seen_having{$part}++) { + push @parts, $part; + } } } @@ -1620,9 +1688,12 @@ sub _bool { =head2 count_literal +B: C is provided for Class::DBI compatibility and +should only be used in that context. See L for further info. + =over 4 -=item Arguments: $sql_fragment, @bind_values +=item Arguments: $sql_fragment, @standalone_bind_values =item Return Value: $count @@ -1641,7 +1712,7 @@ sub count_literal { shift->search_literal(@_)->count; } =item Arguments: none -=item Return Value: @objects +=item Return Value: L<@result_objs|DBIx::Class::Manual::ResultClass> =back @@ -1689,8 +1760,7 @@ another query. sub reset { my ($self) = @_; - delete @{$self}{qw/_attrs stashed_rows stashed_objects/}; - + delete @{$self}{qw/stashed_rows stashed_objects/}; $self->{all_cache_position} = 0; $self->cursor->reset; return $self; @@ -1702,12 +1772,12 @@ sub reset { =item Arguments: none -=item Return Value: $object | undef +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> | undef =back -Resets the resultset and returns an object for the first result (or C -if the resultset is empty). +L the resultset (causing a fresh query to storage) and returns +an object for the first result (or C if the resultset is empty). =cut @@ -1725,159 +1795,146 @@ sub first { sub _rs_update_delete { my ($self, $op, $values) = @_; - my $cond = $self->{cond}; my $rsrc = $self->result_source; my $storage = $rsrc->schema->storage; my $attrs = { %{$self->_resolved_attrs} }; - # "needs" is a strong word here - if the subquery is part of an IN clause - no point of - # even adding the group_by. It will really be used only when composing a poor-man's - # multicolumn-IN equivalent OR set - my $needs_group_by_subq = defined $attrs->{group_by}; - - # simplify the joinmap and maybe decide if a grouping (and thus subquery) is necessary - my $relation_classifications; - if (ref($attrs->{from}) eq 'ARRAY') { - if (@{$attrs->{from}} == 1) { - # not a fucking JOIN at all, quit with the dickery - $relation_classifications = {}; - } else { - $attrs->{from} = $storage->_prune_unused_joins ($attrs->{from}, $attrs->{select}, $cond, $attrs); + my $join_classifications; + my $existing_group_by = delete $attrs->{group_by}; + + # do we need a subquery for any reason? + my $needs_subq = ( + defined $existing_group_by + or + # if {from} is unparseable wrap a subq + ref($attrs->{from}) ne 'ARRAY' + or + # limits call for a subq + $self->_has_resolved_attr(qw/rows offset/) + ); - $relation_classifications = $storage->_resolve_aliastypes_from_select_args ( + # simplify the joinmap, so we can further decide if a subq is necessary + if (!$needs_subq and @{$attrs->{from}} > 1) { + $attrs->{from} = $storage->_prune_unused_joins ($attrs->{from}, $attrs->{select}, $self->{cond}, $attrs); + + # check if there are any joins left after the prune + if ( @{$attrs->{from}} > 1 ) { + $join_classifications = $storage->_resolve_aliastypes_from_select_args ( [ @{$attrs->{from}}[1 .. $#{$attrs->{from}}] ], $attrs->{select}, - $cond, + $self->{cond}, $attrs - ) unless $needs_group_by_subq; # we already know we need a group, no point of resolving them + ); + + # any non-pruneable joins imply subq + $needs_subq = scalar keys %{ $join_classifications->{restricting} || {} }; } } - else { - $needs_group_by_subq ||= 1; # if {from} is unparseable assume the worst - } - - $needs_group_by_subq ||= exists $relation_classifications->{multiplying}; - # if no subquery - life is easy-ish - unless ( - $needs_group_by_subq + # check if the head is composite (by now all joins are thrown out unless $needs_subq) + $needs_subq ||= ( + (ref $attrs->{from}[0]) ne 'HASH' or - keys %$relation_classifications # if any joins at all - need to wrap a subq - or - $self->_has_resolved_attr(qw/rows offset/) # limits call for a subq - ) { + ref $attrs->{from}[0]{ $attrs->{from}[0]{-alias} } + ); + + my ($cond, $guard); + # do we need anything like a subquery? + if (! $needs_subq) { # Most databases do not allow aliasing of tables in UPDATE/DELETE. Thus # a condition containing 'me' or other table prefixes will not work - # at all. What this code tries to do (badly) is to generate a condition - # with the qualifiers removed, by exploiting the quote mechanism of sqla - # - # this is atrocious and should be replaced by normal sqla introspection - # one sunny day - my ($sql, @bind) = do { + # at all. Tell SQLMaker to dequalify idents via a gross hack. + $cond = do { my $sqla = $rsrc->storage->sql_maker; local $sqla->{_dequalify_idents} = 1; - $sqla->_recurse_where($self->{cond}); - } if $self->{cond}; - - return $rsrc->storage->$op( - $rsrc, - $op eq 'update' ? $values : (), - $self->{cond} ? \[$sql, @bind] : (), - ); - } - - # we got this far - means it is time to wrap a subquery - my $idcols = $rsrc->_identifying_column_set || $self->throw_exception( - sprintf( - "Unable to perform complex resultset %s() without an identifying set of columns on source '%s'", - $op, - $rsrc->source_name, - ) - ); - my $existing_group_by = delete $attrs->{group_by}; - - # make a new $rs selecting only the PKs (that's all we really need for the subq) - delete @{$attrs}{qw/collapse select _prefetch_selector_range as/}; - $attrs->{columns} = [ map { "$attrs->{alias}.$_" } @$idcols ]; - $attrs->{group_by} = \ ''; # FIXME - this is an evil hack, it causes the optimiser to kick in and throw away the LEFT joins - my $subrs = (ref $self)->new($rsrc, $attrs); - - if (@$idcols == 1) { - return $storage->$op ( - $rsrc, - $op eq 'update' ? $values : (), - { $idcols->[0] => { -in => $subrs->as_query } }, - ); + \[ $sqla->_recurse_where($self->{cond}) ]; + }; } - elsif ($storage->_use_multicolumn_in) { - # This is hideously ugly, but SQLA does not understand multicol IN expressions - my $sql_maker = $storage->sql_maker; - my ($sql, @bind) = @${$subrs->as_query}; - $sql = sprintf ('(%s) IN %s', # the as_query already comes with a set of parenthesis - join (', ', map { $sql_maker->_quote ($_) } @$idcols), - $sql, + else { + # we got this far - means it is time to wrap a subquery + my $idcols = $rsrc->_identifying_column_set || $self->throw_exception( + sprintf( + "Unable to perform complex resultset %s() without an identifying set of columns on source '%s'", + $op, + $rsrc->source_name, + ) ); - return $storage->$op ( - $rsrc, - $op eq 'update' ? $values : (), - \[$sql, @bind], - ); - } - else { - # if all else fails - get all primary keys and operate over a ORed set - # wrap in a transaction for consistency - # this is where the group_by starts to matter - my $subq_group_by; - if ($needs_group_by_subq) { - $subq_group_by = $attrs->{columns}; - - # make sure if there is a supplied group_by it matches the columns compiled above - # perfectly. Anything else can not be sanely executed on most databases so croak - # right then and there - if ($existing_group_by) { - my @current_group_by = map - { $_ =~ /\./ ? $_ : "$attrs->{alias}.$_" } - @$existing_group_by - ; + # make a new $rs selecting only the PKs (that's all we really need for the subq) + delete $attrs->{$_} for qw/collapse select _prefetch_selector_range as/; + $attrs->{columns} = [ map { "$attrs->{alias}.$_" } @$idcols ]; + $attrs->{group_by} = \ ''; # FIXME - this is an evil hack, it causes the optimiser to kick in and throw away the LEFT joins + my $subrs = (ref $self)->new($rsrc, $attrs); - if ( - join ("\x00", sort @current_group_by) - ne - join ("\x00", sort @$subq_group_by ) - ) { - $self->throw_exception ( - "You have just attempted a $op operation on a resultset which does group_by" - . ' on columns other than the primary keys, while DBIC internally needs to retrieve' - . ' the primary keys in a subselect. All sane RDBMS engines do not support this' - . ' kind of queries. Please retry the operation with a modified group_by or' - . ' without using one at all.' - ); + if (@$idcols == 1) { + $cond = { $idcols->[0] => { -in => $subrs->as_query } }; + } + elsif ($storage->_use_multicolumn_in) { + # no syntax for calling this properly yet + # !!! EXPERIMENTAL API !!! WILL CHANGE !!! + $cond = $storage->sql_maker->_where_op_multicolumn_in ( + $idcols, # how do I convey a list of idents...? can binds reside on lhs? + $subrs->as_query + ), + } + else { + # if all else fails - get all primary keys and operate over a ORed set + # wrap in a transaction for consistency + # this is where the group_by/multiplication starts to matter + if ( + $existing_group_by + or + keys %{ $join_classifications->{multiplying} || {} } + ) { + # make sure if there is a supplied group_by it matches the columns compiled above + # perfectly. Anything else can not be sanely executed on most databases so croak + # right then and there + if ($existing_group_by) { + my @current_group_by = map + { $_ =~ /\./ ? $_ : "$attrs->{alias}.$_" } + @$existing_group_by + ; + + if ( + join ("\x00", sort @current_group_by) + ne + join ("\x00", sort @{$attrs->{columns}} ) + ) { + $self->throw_exception ( + "You have just attempted a $op operation on a resultset which does group_by" + . ' on columns other than the primary keys, while DBIC internally needs to retrieve' + . ' the primary keys in a subselect. All sane RDBMS engines do not support this' + . ' kind of queries. Please retry the operation with a modified group_by or' + . ' without using one at all.' + ); + } } + + $subrs = $subrs->search({}, { group_by => $attrs->{columns} }); } - } - my $guard = $storage->txn_scope_guard; + $guard = $storage->txn_scope_guard; - my @op_condition; - for my $row ($subrs->search({}, { group_by => $subq_group_by })->cursor->all) { - push @op_condition, { map - { $idcols->[$_] => $row->[$_] } - (0 .. $#$idcols) - }; + $cond = []; + for my $row ($subrs->cursor->all) { + push @$cond, { map + { $idcols->[$_] => $row->[$_] } + (0 .. $#$idcols) + }; + } } + } - my $res = $storage->$op ( - $rsrc, - $op eq 'update' ? $values : (), - \@op_condition, - ); + my $res = $storage->$op ( + $rsrc, + $op eq 'update' ? $values : (), + $cond, + ); - $guard->commit; + $guard->commit if $guard; - return $res; - } + return $res; } =head2 update @@ -1886,13 +1943,13 @@ sub _rs_update_delete { =item Arguments: \%values -=item Return Value: $storage_rv +=item Return Value: $underlying_storage_rv =back Sets the specified columns in the resultset to the supplied values in a single query. Note that this will not run any accessor/set_column/update -triggers, nor will it update any row object instances derived from this +triggers, nor will it update any result object instances derived from this resultset (this includes the contents of the L if any). See L if you need to execute any on-update triggers or cascades defined either by you or a @@ -1954,13 +2011,13 @@ sub update_all { =item Arguments: none -=item Return Value: $storage_rv +=item Return Value: $underlying_storage_rv =back Deletes the rows matching this resultset in a single query. Note that this will not run any delete triggers, nor will it alter the -L status of any row object instances +L status of any result object instances derived from this resultset (this includes the contents of the L if any). See L if you need to execute any on-delete triggers or cascades defined either by you or a @@ -2010,28 +2067,55 @@ sub delete_all { =over 4 -=item Arguments: \@data; +=item Arguments: [ \@column_list, \@row_values+ ] | [ \%col_data+ ] + +=item Return Value: L<\@result_objects|DBIx::Class::Manual::ResultClass> (scalar context) | L<@result_objects|DBIx::Class::Manual::ResultClass> (list context) =back -Accepts either an arrayref of hashrefs or alternatively an arrayref of arrayrefs. -For the arrayref of hashrefs style each hashref should be a structure suitable -for submitting to a $resultset->create(...) method. +Accepts either an arrayref of hashrefs or alternatively an arrayref of +arrayrefs. -In void context, C in L is used -to insert the data, as this is a faster method. +=over -Otherwise, each set of data is inserted into the database using -L, and the resulting objects are -accumulated into an array. The array itself, or an array reference -is returned depending on scalar or list context. +=item NOTE -Example: Assuming an Artist Class that has many CDs Classes relating: +The context of this method call has an important effect on what is +submitted to storage. In void context data is fed directly to fastpath +insertion routines provided by the underlying storage (most often +L), bypassing the L and +L calls on the +L class, including any +augmentation of these methods provided by components. For example if you +are using something like L to create primary +keys for you, you will find that your PKs are empty. In this case you +will have to explicitly force scalar or list context in order to create +those values. - my $Artist_rs = $schema->resultset("Artist"); +=back + +In non-void (scalar or list) context, this method is simply a wrapper +for L. Depending on list or scalar context either a list of +L objects or an arrayref +containing these objects is returned. - ## Void Context Example - $Artist_rs->populate([ +When supplying data in "arrayref of arrayrefs" invocation style, the +first element should be a list of column names and each subsequent +element should be a data value in the earlier specified column order. +For example: + + $Arstist_rs->populate([ + [ qw( artistid name ) ], + [ 100, 'A Formally Unknown Singer' ], + [ 101, 'A singer that jumped the shark two albums ago' ], + [ 102, 'An actually cool singer' ], + ]); + +For the arrayref of hashrefs style each hashref should be a structure +suitable for passing to L. Multi-create is also permitted with +this syntax. + + $schema->resultset("Artist")->populate([ { artistid => 4, name => 'Manufactured Crap', cds => [ { title => 'My First CD', year => 2006 }, { title => 'Yet More Tweeny-Pop crap', year => 2007 }, @@ -2045,37 +2129,11 @@ Example: Assuming an Artist Class that has many CDs Classes relating: }, ]); - ## Array Context Example - my ($ArtistOne, $ArtistTwo, $ArtistThree) = $Artist_rs->populate([ - { name => "Artist One"}, - { name => "Artist Two"}, - { name => "Artist Three", cds=> [ - { title => "First CD", year => 2007}, - { title => "Second CD", year => 2008}, - ]} - ]); - - print $ArtistOne->name; ## response is 'Artist One' - print $ArtistThree->cds->count ## reponse is '2' - -For the arrayref of arrayrefs style, the first element should be a list of the -fieldsnames to which the remaining elements are rows being inserted. For -example: - - $Arstist_rs->populate([ - [qw/artistid name/], - [100, 'A Formally Unknown Singer'], - [101, 'A singer that jumped the shark two albums ago'], - [102, 'An actually cool singer'], - ]); - -Please note an important effect on your data when choosing between void and -wantarray context. Since void context goes straight to C in -L this will skip any component that is overriding -C. So if you are using something like L to -create primary keys for you, you will find that your PKs are empty. In this -case you will have to use the wantarray context in order to create those -values. +If you attempt a void-context multi-create as in the example above (each +Artist also has the related list of CDs), and B supply the +necessary autoinc foreign key information, this method will proxy to the +less efficient L, and then throw the Result objects away. In this +case there are obviously no benefits to using this method over L. =cut @@ -2088,10 +2146,7 @@ sub populate { return unless @$data; if(defined wantarray) { - my @created; - foreach my $item (@$data) { - push(@created, $self->create($item)); - } + my @created = map { $self->create($_) } @$data; return wantarray ? @created : \@created; } else { @@ -2146,14 +2201,12 @@ sub populate { ## inherit the data locked in the conditions of the resultset my ($rs_data) = $self->_merge_with_rscond({}); delete @{$rs_data}{@columns}; - my @inherit_cols = keys %$rs_data; - my @inherit_data = values %$rs_data; ## do bulk insert on current row $rsrc->storage->insert_bulk( $rsrc, - [@columns, @inherit_cols], - [ map { [ @$_{@columns}, @inherit_data ] } @$data ], + [@columns, keys %$rs_data], + [ map { [ @$_{@columns}, values %$rs_data ] } @$data ], ); ## do the has_many relationships @@ -2216,11 +2269,11 @@ sub _normalize_populate_args { =item Arguments: none -=item Return Value: $pager +=item Return Value: L<$pager|Data::Page> =back -Return Value a L object for the current resultset. Only makes +Returns a L object for the current resultset. Only makes sense for queries with a C attribute. To get the full count of entries for a paged resultset, call @@ -2263,7 +2316,7 @@ sub pager { =item Arguments: $page_number -=item Return Value: $rs +=item Return Value: L<$resultset|/search> =back @@ -2282,16 +2335,16 @@ sub page { =over 4 -=item Arguments: \%vals +=item Arguments: \%col_data -=item Return Value: $rowobject +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> =back -Creates a new row object in the resultset's result class and returns +Creates a new result object in the resultset's result class and returns it. The row is not inserted into the database at this point, call L to do that. Calling L -will tell you whether the row object has been inserted or not. +will tell you whether the result object has been inserted or not. Passes the hashref of input on to L. @@ -2299,7 +2352,11 @@ Passes the hashref of input on to L. sub new_result { my ($self, $values) = @_; - $self->throw_exception( "new_result needs a hash" ) + + $self->throw_exception( "new_result takes only one argument - a hashref of values" ) + if @_ > 2; + + $self->throw_exception( "new_result expects a hashref" ) unless (ref $values eq 'HASH'); my ($merged_cond, $cols_from_relations) = $self->_merge_with_rscond($values); @@ -2485,7 +2542,7 @@ sub _remove_alias { =item Arguments: none -=item Return Value: \[ $sql, @bind ] +=item Return Value: \[ $sql, L<@bind_values|/DBIC BIND VALUES> ] =back @@ -2498,7 +2555,7 @@ This is generally used as the RHS for a subquery. sub as_query { my $self = shift; - my $attrs = $self->_resolved_attrs_copy; + my $attrs = { %{ $self->_resolved_attrs } }; # For future use: # @@ -2516,9 +2573,9 @@ sub as_query { =over 4 -=item Arguments: \%vals, \%attrs? +=item Arguments: \%col_data, { key => $unique_constraint, L<%attrs|/ATTRIBUTES> }? -=item Return Value: $rowobject +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> =back @@ -2563,9 +2620,9 @@ sub find_or_new { =over 4 -=item Arguments: \%vals +=item Arguments: \%col_data -=item Return Value: a L $object +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> =back @@ -2589,12 +2646,11 @@ This can be applied recursively, and will work correctly for a structure with an arbitrary depth and width, as long as the relationships actually exists and the correct column data has been supplied. - Instead of hashrefs of plain related data (key/value pairs), you may also pass new or inserted objects. New objects (not inserted yet, see -L), will be inserted into their appropriate tables. +L), will be inserted into their appropriate tables. -Effectively a shortcut for C<< ->new_result(\%vals)->insert >>. +Effectively a shortcut for C<< ->new_result(\%col_data)->insert >>. Example of creating a new row. @@ -2632,9 +2688,10 @@ C resultset. Note Hashref. When subclassing ResultSet never attempt to override this method. Since it is a simple shortcut for C<< $self->new_result($attrs)->insert >>, a lot of the internals simply never call it, so your override will be -bypassed more often than not. Override either L -or L depending on how early in the -L process you need to intervene. +bypassed more often than not. Override either L +or L depending on how early in the +L process you need to intervene. See also warning pertaining to +L. =back @@ -2651,9 +2708,9 @@ sub create { =over 4 -=item Arguments: \%vals, \%attrs? +=item Arguments: \%col_data, { key => $unique_constraint, L<%attrs|/ATTRIBUTES> }? -=item Return Value: $rowobject +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> =back @@ -2712,7 +2769,7 @@ database! year => 2005, }); - if( $cd->in_storage ) { + if( !$cd->in_storage ) { # do some stuff $cd->insert; } @@ -2733,16 +2790,16 @@ sub find_or_create { =over 4 -=item Arguments: \%col_values, { key => $unique_constraint }? +=item Arguments: \%col_data, { key => $unique_constraint, L<%attrs|/ATTRIBUTES> }? -=item Return Value: $row_object +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> =back $resultset->update_or_create({ col => $val, ... }); Like L, but if a row is found it is immediately updated via -C<< $found_row->update (\%col_values) >>. +C<< $found_row->update (\%col_data) >>. Takes an optional C attribute to search on a specific unique constraint. @@ -2783,20 +2840,6 @@ L and L instead. Don't forget to call L to save the newly created row to the database! - my $cd = $schema->resultset('CD')->update_or_new( - { - artist => 'Massive Attack', - title => 'Mezzanine', - year => 1998, - }, - { key => 'cd_artist_title' } - ); - - if( $cd->in_storage ) { - # do some stuff - $cd->insert; - } - =cut sub update_or_create { @@ -2817,16 +2860,16 @@ sub update_or_create { =over 4 -=item Arguments: \%col_values, { key => $unique_constraint }? +=item Arguments: \%col_data, { key => $unique_constraint, L<%attrs|/ATTRIBUTES> }? -=item Return Value: $rowobject +=item Return Value: L<$result|DBIx::Class::Manual::ResultClass> =back $resultset->update_or_new({ col => $val, ... }); Like L but if a row is found it is immediately updated via -C<< $found_row->update (\%col_values) >>. +C<< $found_row->update (\%col_data) >>. For example: @@ -2882,7 +2925,7 @@ sub update_or_new { =item Arguments: none -=item Return Value: \@cache_objects | undef +=item Return Value: L<\@result_objs|DBIx::Class::Manual::ResultClass> | undef =back @@ -2901,15 +2944,15 @@ sub get_cache { =over 4 -=item Arguments: \@cache_objects +=item Arguments: L<\@result_objs|DBIx::Class::Manual::ResultClass> -=item Return Value: \@cache_objects +=item Return Value: L<\@result_objs|DBIx::Class::Manual::ResultClass> =back Sets the contents of the cache for the resultset. Expects an arrayref of objects of the same class as those produced by the resultset. Note that -if the cache is set the resultset will return the cached objects rather +if the cache is set, the resultset will return the cached objects rather than re-querying the database even if the cache attr is not set. The contents of the cache can also be populated by using the @@ -2980,9 +3023,9 @@ sub is_ordered { =over 4 -=item Arguments: $relationship_name +=item Arguments: $rel_name -=item Return Value: $resultset +=item Return Value: L<$resultset|/search> =back @@ -2995,7 +3038,6 @@ Returns a related resultset for the supplied relationship name. sub related_resultset { my ($self, $rel) = @_; - $self->{related_resultsets} ||= {}; return $self->{related_resultsets}{$rel} ||= do { my $rsrc = $self->result_source; my $rel_info = $rsrc->relationship_info($rel); @@ -3022,13 +3064,13 @@ sub related_resultset { #XXX - temp fix for result_class bug. There likely is a more elegant fix -groditi delete @{$attrs}{qw(result_class alias)}; - my $new_cache; + my $related_cache; if (my $cache = $self->get_cache) { - if ($cache->[0] && $cache->[0]->related_resultset($rel)->get_cache) { - $new_cache = [ map { @{$_->related_resultset($rel)->get_cache||[]} } - @$cache ]; - } + $related_cache = [ map + { @{$_->related_resultset($rel)->get_cache||[]} } + @$cache + ]; } my $rel_source = $rsrc->related_source($rel); @@ -3051,7 +3093,7 @@ sub related_resultset { where => $attrs->{where}, }); }; - $new->set_cache($new_cache) if $new_cache; + $new->set_cache($related_cache) if $related_cache; $new; }; } @@ -3094,9 +3136,7 @@ source alias of the current result set: =cut sub current_source_alias { - my ($self) = @_; - - return ($self->{attrs} || {})->{alias} || 'me'; + return (shift->{attrs} || {})->{alias} || 'me'; } =head2 as_subselect_rs @@ -3105,7 +3145,7 @@ sub current_source_alias { =item Arguments: none -=item Return Value: $resultset +=item Return Value: L<$resultset|/search> =back @@ -3278,11 +3318,39 @@ sub _chain_relationship { return {%$attrs, from => $from, seen_join => $seen}; } -# too many times we have to do $attrs = { %{$self->_resolved_attrs} } -sub _resolved_attrs_copy { - my $self = shift; - return { %{$self->_resolved_attrs (@_)} }; -} +# FIXME - this needs to go live in Schema with the tree walker... or +# something +my $inflatemap_checker; +$inflatemap_checker = sub { + my ($rsrc, $relpaths) = @_; + + my $rels; + + for (@$relpaths) { + $_ =~ /^ ( [^\.]+ ) \. (.+) $/x + or next; + + push @{$rels->{$1}}, $2; + } + + for my $rel (keys %$rels) { + my $rel_rsrc = try { + $rsrc->related_source ($rel) + } catch { + $rsrc->throw_exception(sprintf( + "Inflation into non-existent relationship '%s' of '%s' requested, " + . "check the inflation specification (columns/as) ending in '...%s.%s'", + $rel, + $rsrc->source_name, + $rel, + ( sort { length($a) <=> length ($b) } @{$rels->{$rel}} )[0], + ))}; + + $inflatemap_checker->($rel_rsrc, $rels->{$rel}); + } + + return; +}; sub _resolved_attrs { my $self = shift; @@ -3307,7 +3375,7 @@ sub _resolved_attrs { if (my $cols = delete $attrs->{columns}) { for my $c (ref $cols eq 'ARRAY' ? @$cols : $cols) { if (ref $c eq 'HASH') { - for my $as (keys %$c) { + for my $as (sort keys %$c) { push @sel, $c->{$as}; push @as, $as; } @@ -3355,6 +3423,14 @@ sub _resolved_attrs { } } + # validate the user-supplied 'as' chain + # folks get too confused by the (logical) exception message, need to + # go to some lengths to clarify the text + # + # FIXME - this needs to go live in Schema with the tree walker... or + # something + $inflatemap_checker->($source, \@as); + $attrs->{select} = \@sel; $attrs->{as} = \@as; @@ -3464,39 +3540,60 @@ sub _resolved_attrs { push @{ $attrs->{as} }, (map { $_->[1] } @prefetch); } - $attrs->{_single_object_inflation} = ! List::Util::first { $_ =~ /\./ } @{$attrs->{as}}; + if ( ! List::Util::first { $_ =~ /\./ } @{$attrs->{as}} ) { + $attrs->{_single_resultclass_inflation} = 1; + $attrs->{collapse} = 0; + } # run through the resulting joinstructure (starting from our current slot) # and unset collapse if proven unnesessary - if ($attrs->{collapse} && ref $attrs->{from} eq 'ARRAY') { + # + # also while we are at it find out if the current root source has + # been premultiplied by previous related_source chaining + # + # this allows to predict whether a root object with all other relation + # data set to NULL is in fact unique + if ($attrs->{collapse}) { - if (@{$attrs->{from}} > 1) { + if (ref $attrs->{from} eq 'ARRAY') { - # find where our table-spec starts and consider only things after us - my @fromlist = @{$attrs->{from}}; - while (@fromlist) { - my $t = shift @fromlist; - $t = $t->[0] if ref $t eq 'ARRAY'; #me vs join from-spec mismatch - last if ($t->{-alias} && $t->{-alias} eq $alias); + if (@{$attrs->{from}} <= 1) { + # no joins - no collapse + $attrs->{collapse} = 0; } + else { + # find where our table-spec starts + my @fromlist = @{$attrs->{from}}; + while (@fromlist) { + my $t = shift @fromlist; + + my $is_multi; + # me vs join from-spec distinction - a ref means non-root + if (ref $t eq 'ARRAY') { + $t = $t->[0]; + $is_multi ||= ! $t->{-is_single}; + } + last if ($t->{-alias} && $t->{-alias} eq $alias); + $attrs->{_main_source_premultiplied} ||= $is_multi; + } - for (@fromlist) { - $attrs->{collapse} = ! $_->[0]{-is_single} - and last; + # no non-singles remaining, nor any premultiplication - nothing to collapse + if ( + ! $attrs->{_main_source_premultiplied} + and + ! List::Util::first { ! $_->[0]{-is_single} } @fromlist + ) { + $attrs->{collapse} = 0; + } } } + else { - # no joins - no collapse - $attrs->{collapse} = 0; + # if we can not analyze the from - err on the side of safety + $attrs->{_main_source_premultiplied} = 1; } } - if (! $attrs->{order_by} and $attrs->{collapse}) { - # default order for collapsing unless the user asked for something - $attrs->{order_by} = [ map { "$alias.$_" } $source->primary_columns ]; - $attrs->{_ordered_for_collapse} = 1; - } - # if both page and offset are specified, produce a combined offset # even though it doesn't make much sense, this is what pre 081xx has # been doing @@ -3621,7 +3718,7 @@ sub _merge_joinpref_attr { $seen_keys->{$import_key} = 1; # don't merge the same key twice } - return $orig; + return @$orig ? $orig : (); } { @@ -3718,7 +3815,7 @@ sub STORABLE_freeze { # A cursor in progress can't be serialized (and would make little sense anyway) # the parser can be regenerated (and can't be serialized) - delete @{$to_serialize}{qw/cursor _row_parser/}; + delete @{$to_serialize}{qw/cursor _row_parser _result_inflator/}; # nor is it sensical to store a not-yet-fired-count pager if ($to_serialize->{pager} and ref $to_serialize->{pager}{total_entries} eq 'CODE') { @@ -3755,6 +3852,10 @@ sub throw_exception { } } +1; + +__END__ + # XXX: FIXME: Attributes docs need clearing up =head1 ATTRIBUTES @@ -3764,6 +3865,10 @@ searching for data. They can be passed to any method which takes an C<\%attrs> argument. See L, L, L, L. +Default attributes can be set on the result class using +L. (Please read +the CAVEATS on that feature before using it!) + These are in no particular order: =head2 order_by @@ -3800,7 +3905,7 @@ syntax as outlined above. =over 4 -=item Value: \@columns +=item Value: \@columns | \%columns | $column =back @@ -3902,14 +4007,6 @@ an explicit list. =back -=head2 +as - -=over 4 - -Indicates additional column names for those added via L. See L. - -=back - =head2 as =over 4 @@ -3952,6 +4049,14 @@ use C instead: You can create your own accessors if required - see L for details. +=head2 +as + +=over 4 + +Indicates additional column names for those added via L. See L. + +=back + =head2 join =over 4 @@ -4015,196 +4120,175 @@ similarly for a third time). For e.g. will return a set of all artists that have both a cd with title 'Down to Earth' and a cd with title 'Popular'. -If you want to fetch related objects from other tables as well, see C +If you want to fetch related objects from other tables as well, see L below. + NOTE: An internal join-chain pruner will discard certain joins while + constructing the actual SQL query, as long as the joins in question do not + affect the retrieved result. This for example includes 1:1 left joins + that are not part of the restriction specification (WHERE/HAVING) nor are + a part of the query selection. + For more help on using joins with search, see L. -=head2 prefetch +=head2 collapse =over 4 -=item Value: ($rel_name | \@rel_names | \%rel_names) +=item Value: (0 | 1) =back -Contains one or more relationships that should be fetched along with -the main query (when they are accessed afterwards the data will -already be available, without extra queries to the database). This is -useful for when you know you will need the related objects, because it -saves at least one query: - - my $rs = $schema->resultset('Tag')->search( - undef, - { - prefetch => { - cd => 'artist' - } - } - ); +When set to a true value, indicates that any rows fetched from joined has_many +relationships are to be aggregated into the corresponding "parent" object. For +example, the resultset: -The initial search results in SQL like the following: - - SELECT tag.*, cd.*, artist.* FROM tag - JOIN cd ON tag.cd = cd.cdid - JOIN artist ON cd.artist = artist.artistid - -L has no need to go back to the database when we access the -C or C relationships, which saves us two SQL statements in this -case. - -Simple prefetches will be joined automatically, so there is no need -for a C attribute in the above search. - -L can be used with the any of the relationship types and -multiple prefetches can be specified together. Below is a more complex -example that prefetches a CD's artist, its liner notes (if present), -the cover image, the tracks on that cd, and the guests on those -tracks. - - # Assuming: - My::Schema::CD->belongs_to( artist => 'My::Schema::Artist' ); - My::Schema::CD->might_have( liner_note => 'My::Schema::LinerNotes' ); - My::Schema::CD->has_one( cover_image => 'My::Schema::Artwork' ); - My::Schema::CD->has_many( tracks => 'My::Schema::Track' ); + my $rs = $schema->resultset('CD')->search({}, { + '+columns' => [ qw/ tracks.title tracks.position / ], + join => 'tracks', + collapse => 1, + }); - My::Schema::Artist->belongs_to( record_label => 'My::Schema::RecordLabel' ); +While executing the following query: - My::Schema::Track->has_many( guests => 'My::Schema::Guest' ); + SELECT me.*, tracks.title, tracks.position + FROM cd me + LEFT JOIN track tracks + ON tracks.cdid = me.cdid +Will return only as many objects as there are rows in the CD source, even +though the result of the query may span many rows. Each of these CD objects +will in turn have multiple "Track" objects hidden behind the has_many +generated accessor C. Without C<< collapse => 1 >>, the return values +of this resultset would be as many CD objects as there are tracks (a "Cartesian +product"), with each CD object containing exactly one of all fetched Track data. - my $rs = $schema->resultset('CD')->search( - undef, - { - prefetch => [ - { artist => 'record_label'}, # belongs_to => belongs_to - 'liner_note', # might_have - 'cover_image', # has_one - { tracks => 'guests' }, # has_many => has_many - ] - } - ); +When a collapse is requested on a non-ordered resultset, an order by some +unique part of the main source (the left-most table) is inserted automatically. +This is done so that the resultset is allowed to be "lazy" - calling +L<< $rs->next|/next >> will fetch only as many rows as it needs to build the next +object with all of its related data. -This will produce SQL like the following: +If an L is already declared, and orders the resultset in a way that +makes collapsing as described above impossible (e.g. C<< ORDER BY +has_many_rel.column >> or C), DBIC will automatically +switch to "eager" mode and slurp the entire resultset before consturcting the +first object returned by L. - SELECT cd.*, artist.*, record_label.*, liner_note.*, cover_image.*, - tracks.*, guests.* - FROM cd me - JOIN artist artist - ON artist.artistid = me.artistid - JOIN record_label record_label - ON record_label.labelid = artist.labelid - LEFT JOIN track tracks - ON tracks.cdid = me.cdid - LEFT JOIN guest guests - ON guests.trackid = track.trackid - LEFT JOIN liner_notes liner_note - ON liner_note.cdid = me.cdid - JOIN cd_artwork cover_image - ON cover_image.cdid = me.cdid - ORDER BY tracks.cd +Setting this attribute on a resultset that does not join any has_many +relations is a no-op. -Now the C, C, C, C, -C, and C of the CD will all be available through the -relationship accessors without the need for additional queries to the -database. +For a more in-depth discussion, see L. -However, there is one caveat to be observed: it can be dangerous to -prefetch more than one L -relationship on a given level. e.g.: +=head2 prefetch - my $rs = $schema->resultset('CD')->search( - undef, - { - prefetch => [ - 'tracks', # has_many - { cd_to_producer => 'producer' }, # has_many => belongs_to (i.e. m2m) - ] - } - ); +=over 4 -In fact, C will emit the following warning: +=item Value: ($rel_name | \@rel_names | \%rel_names) - Prefetching multiple has_many rels tracks and cd_to_producer at top - level will explode the number of row objects retrievable via ->next - or ->all. Use at your own risk. +=back -The collapser currently can't identify duplicate tuples for multiple -L relationships and as a -result the second L -relation could contain redundant objects. +This attribute is a shorthand for specifying a L spec, adding all +columns from the joined related sources as L and setting +L to a true value. For example, the following two queries are +equivalent: -=head3 Using L with L + my $rs = $schema->resultset('Artist')->search({}, { + prefetch => { cds => ['genre', 'tracks' ] }, + }); -L implies a L with the equivalent argument, and is -properly merged with any existing L specification. So the -following: +and - my $rs = $schema->resultset('CD')->search( - {'record_label.name' => 'Music Product Ltd.'}, - { - join => {artist => 'record_label'}, - prefetch => 'artist', - } - ); - -... will work, searching on the record label's name, but only -prefetching the C. + my $rs = $schema->resultset('Artist')->search({}, { + join => { cds => ['genre', 'tracks' ] }, + collapse => 1, + '+columns' => [ + (map + { +{ "cds.$_" => "cds.$_" } } + $schema->source('Artist')->related_source('cds')->columns + ), + (map + { +{ "cds.genre.$_" => "genre.$_" } } + $schema->source('Artist')->related_source('cds')->related_source('genre')->columns + ), + (map + { +{ "cds.tracks.$_" => "tracks.$_" } } + $schema->source('Artist')->related_source('cds')->related_source('tracks')->columns + ), + ], + }); -=head3 Using L with L / L / L / L +Both producing the following SQL: + + SELECT me.artistid, me.name, me.rank, me.charfield, + cds.cdid, cds.artist, cds.title, cds.year, cds.genreid, cds.single_track, + genre.genreid, genre.name, + tracks.trackid, tracks.cd, tracks.position, tracks.title, tracks.last_updated_on, tracks.last_updated_at + FROM artist me + LEFT JOIN cd cds + ON cds.artist = me.artistid + LEFT JOIN genre genre + ON genre.genreid = cds.genreid + LEFT JOIN track tracks + ON tracks.cd = cds.cdid + ORDER BY me.artistid + +While L implies a L, it is ok to mix the two together, as +the arguments are properly merged and generally do the right thing. For +example, you may want to do the following: + + my $artists_and_cds_without_genre = $schema->resultset('Artist')->search( + { 'genre.genreid' => undef }, + { + join => { cds => 'genre' }, + prefetch => 'cds', + } + ); -L implies a L/L with the fields of the -prefetched relations. So given: +Which generates the following SQL: - my $rs = $schema->resultset('CD')->search( - undef, - { - select => ['cd.title'], - as => ['cd_title'], - prefetch => 'artist', - } - ); + SELECT me.artistid, me.name, me.rank, me.charfield, + cds.cdid, cds.artist, cds.title, cds.year, cds.genreid, cds.single_track + FROM artist me + LEFT JOIN cd cds + ON cds.artist = me.artistid + LEFT JOIN genre genre + ON genre.genreid = cds.genreid + WHERE genre.genreid IS NULL + ORDER BY me.artistid -The L becomes: C<'cd.title', 'artist.*'> and the L -becomes: C<'cd_title', 'artist.*'>. +For a more in-depth discussion, see L. -=head3 CAVEATS - -Prefetch does a lot of deep magic. As such, it may not behave exactly -as you might expect. +=head2 alias =over 4 -=item * - -Prefetch uses the L to populate the prefetched relationships. This -may or may not be what you want. - -=item * - -If you specify a condition on a prefetched relationship, ONLY those -rows that match the prefetched condition will be fetched into that relationship. -This means that adding prefetch to a search() B what is returned by -traversing a relationship. So, if you have C<< Artist->has_many(CDs) >> and you do - - my $artist_rs = $schema->resultset('Artist')->search({ - 'cds.year' => 2008, - }, { - join => 'cds', - }); +=item Value: $source_alias - my $count = $artist_rs->first->cds->count; +=back - my $artist_rs_prefetch = $artist_rs->search( {}, { prefetch => 'cds' } ); +Sets the source alias for the query. Normally, this defaults to C, but +nested search queries (sub-SELECTs) might need specific aliases set to +reference inner queries. For example: - my $prefetch_count = $artist_rs_prefetch->first->cds->count; + my $q = $rs + ->related_resultset('CDs') + ->related_resultset('Tracks') + ->search({ + 'track.id' => { -ident => 'none_search.id' }, + }) + ->as_query; - cmp_ok( $count, '==', $prefetch_count, "Counts should be the same" ); + my $ids = $self->search({ + -not_exists => $q, + }, { + alias => 'none_search', + group_by => 'none_search.id', + })->get_column('id')->as_query; -that cmp_ok() may or may not pass depending on the datasets involved. This -behavior may or may not survive the 0.09 transition. + $self->search({ id => { -in => $ids } }) -=back +This attribute is directly tied to L. =head2 page @@ -4312,7 +4396,7 @@ attribute, this setting is ignored and an appropriate warning is issued. Adds to the WHERE clause. # only return rows WHERE deleted IS NULL for all searches - __PACKAGE__->resultset_attributes({ where => { deleted => undef } }); ) + __PACKAGE__->resultset_attributes({ where => { deleted => undef } }); Can be overridden by passing C<< { where => undef } >> as an attribute to a resultset. @@ -4343,13 +4427,189 @@ L. =over 4 -=item Value: ( 'update' | 'shared' ) +=item Value: ( 'update' | 'shared' | \$scalar ) =back Set to 'update' for a SELECT ... FOR UPDATE or 'shared' for a SELECT -... FOR SHARED. +... FOR SHARED. If \$scalar is passed, this is taken directly and embedded in the +query. -=cut +=head1 PREFETCHING + +DBIx::Class supports arbitrary related data prefetching from multiple related +sources. Any combination of relationship types and column sets are supported. +If L is requested, there is an additional requirement of +selecting enough data to make every individual object uniquely identifiable. + +Here are some more involved examples, based on the following relationship map: + + # Assuming: + My::Schema::CD->belongs_to( artist => 'My::Schema::Artist' ); + My::Schema::CD->might_have( liner_note => 'My::Schema::LinerNotes' ); + My::Schema::CD->has_many( tracks => 'My::Schema::Track' ); + + My::Schema::Artist->belongs_to( record_label => 'My::Schema::RecordLabel' ); + + My::Schema::Track->has_many( guests => 'My::Schema::Guest' ); + + + + my $rs = $schema->resultset('Tag')->search( + undef, + { + prefetch => { + cd => 'artist' + } + } + ); + +The initial search results in SQL like the following: + + SELECT tag.*, cd.*, artist.* FROM tag + JOIN cd ON tag.cd = cd.cdid + JOIN artist ON cd.artist = artist.artistid + +L has no need to go back to the database when we access the +C or C relationships, which saves us two SQL statements in this +case. + +Simple prefetches will be joined automatically, so there is no need +for a C attribute in the above search. + +The L attribute can be used with any of the relationship types +and multiple prefetches can be specified together. Below is a more complex +example that prefetches a CD's artist, its liner notes (if present), +the cover image, the tracks on that CD, and the guests on those +tracks. + + my $rs = $schema->resultset('CD')->search( + undef, + { + prefetch => [ + { artist => 'record_label'}, # belongs_to => belongs_to + 'liner_note', # might_have + 'cover_image', # has_one + { tracks => 'guests' }, # has_many => has_many + ] + } + ); + +This will produce SQL like the following: + + SELECT cd.*, artist.*, record_label.*, liner_note.*, cover_image.*, + tracks.*, guests.* + FROM cd me + JOIN artist artist + ON artist.artistid = me.artistid + JOIN record_label record_label + ON record_label.labelid = artist.labelid + LEFT JOIN track tracks + ON tracks.cdid = me.cdid + LEFT JOIN guest guests + ON guests.trackid = track.trackid + LEFT JOIN liner_notes liner_note + ON liner_note.cdid = me.cdid + JOIN cd_artwork cover_image + ON cover_image.cdid = me.cdid + ORDER BY tracks.cd + +Now the C, C, C, C, +C, and C of the CD will all be available through the +relationship accessors without the need for additional queries to the +database. + +=head3 CAVEATS + +Prefetch does a lot of deep magic. As such, it may not behave exactly +as you might expect. + +=over 4 + +=item * + +Prefetch uses the L to populate the prefetched relationships. This +may or may not be what you want. + +=item * + +If you specify a condition on a prefetched relationship, ONLY those +rows that match the prefetched condition will be fetched into that relationship. +This means that adding prefetch to a search() B what is returned by +traversing a relationship. So, if you have C<< Artist->has_many(CDs) >> and you do + + my $artist_rs = $schema->resultset('Artist')->search({ + 'cds.year' => 2008, + }, { + join => 'cds', + }); + + my $count = $artist_rs->first->cds->count; + + my $artist_rs_prefetch = $artist_rs->search( {}, { prefetch => 'cds' } ); + + my $prefetch_count = $artist_rs_prefetch->first->cds->count; + + cmp_ok( $count, '==', $prefetch_count, "Counts should be the same" ); + +That cmp_ok() may or may not pass depending on the datasets involved. This +behavior may or may not survive the 0.09 transition. + +=back + +=head1 DBIC BIND VALUES + +Because DBIC may need more information to bind values than just the column name +and value itself, it uses a special format for both passing and receiving bind +values. Each bind value should be composed of an arrayref of +C<< [ \%args => $val ] >>. The format of C<< \%args >> is currently: + +=over 4 + +=item dbd_attrs + +If present (in any form), this is what is being passed directly to bind_param. +Note that different DBD's expect different bind args. (e.g. DBD::SQLite takes +a single numerical type, while DBD::Pg takes a hashref if bind options.) + +If this is specified, all other bind options described below are ignored. + +=item sqlt_datatype + +If present, this is used to infer the actual bind attribute by passing to +C<< $resolved_storage->bind_attribute_by_data_type() >>. Defaults to the +"data_type" from the L. + +Note that the data type is somewhat freeform (hence the sqlt_ prefix); +currently drivers are expected to "Do the Right Thing" when given a common +datatype name. (Not ideal, but that's what we got at this point.) + +=item sqlt_size + +Currently used to correctly allocate buffers for bind_param_inout(). +Defaults to "size" from the L, +or to a sensible value based on the "data_type". + +=item dbic_colname + +Used to fill in missing sqlt_datatype and sqlt_size attributes (if they are +explicitly specified they are never overriden). Also used by some weird DBDs, +where the column name should be available at bind_param time (e.g. Oracle). + +=back + +For backwards compatibility and convenience, the following shortcuts are +supported: + + [ $name => $val ] === [ { dbic_colname => $name }, $val ] + [ \$dt => $val ] === [ { sqlt_datatype => $dt }, $val ] + [ undef, $val ] === [ {}, $val ] + +=head1 AUTHOR AND CONTRIBUTORS + +See L and L in DBIx::Class + +=head1 LICENSE + +You may distribute this code under the same terms as Perl itself. -1;