1 package DBIx::Class::ResultSet;
9 use Carp::Clan qw/^DBIx::Class/;
12 use DBIx::Class::ResultSetColumn;
13 use DBIx::Class::ResultSourceHandle;
16 use base qw/DBIx::Class/;
18 __PACKAGE__->mk_group_accessors('simple' => qw/_result_class _source_handle/);
22 DBIx::Class::ResultSet - Responsible for fetching and creating resultset.
26 my $rs = $schema->resultset('User')->search({ registered => 1 });
27 my @rows = $schema->resultset('CD')->search({ year => 2005 })->all();
31 The resultset is also known as an iterator. It is responsible for handling
32 queries that may return an arbitrary number of rows, e.g. via L</search>
33 or a C<has_many> relationship.
35 In the examples below, the following table classes are used:
37 package MyApp::Schema::Artist;
38 use base qw/DBIx::Class/;
39 __PACKAGE__->load_components(qw/Core/);
40 __PACKAGE__->table('artist');
41 __PACKAGE__->add_columns(qw/artistid name/);
42 __PACKAGE__->set_primary_key('artistid');
43 __PACKAGE__->has_many(cds => 'MyApp::Schema::CD');
46 package MyApp::Schema::CD;
47 use base qw/DBIx::Class/;
48 __PACKAGE__->load_components(qw/Core/);
49 __PACKAGE__->table('cd');
50 __PACKAGE__->add_columns(qw/cdid artist title year/);
51 __PACKAGE__->set_primary_key('cdid');
52 __PACKAGE__->belongs_to(artist => 'MyApp::Schema::Artist');
57 When you create a resultset (usually as a result of calling search()), DBIx::Class
58 B<doesn't> make a DB call. Not yet. A resultset is (in simplistic terms) a set of
59 where conditions, join conditions, and other metadata that would be needed to execute
60 a SELECT statement. This has several big implications:
64 =item * You can chain resultsets
66 =item * You can run multiple queries using the same resultset
70 =head2 Chaining resultsets
72 Let's say you've got a query that needs to be run to return some data to the user. But,
73 you have an authorization system in place that prevents certain users from seeing certain
74 information. So, you want to construct the query in one method, but add constraints to it
79 my $request = $self->get_request; # Get a request object somehow.
80 my $schema = $self->get_schema; # Get the DBIC schema object somehow.
82 my $rs = $schema->resultset('some_data')->search({
83 foo => $request->param('foo'),
84 bar => $request->param('bar'),
87 $self->apply_security_policy( $rs );
92 sub apply_security_policy {
101 =head2 Multiple queries
103 Since a resultset hasn't hit the database yet, you can do all sorts of things with it.
105 # Don't hit the DB yet.
106 my $rs = $schema->resultset('some_table')->search({
111 # Each of these hits the DB individually.
112 my $count = $rs->count;
113 my $max_baz = $rs->get_column('baz')->max;
114 my @records = $rs->all;
116 And it's not just limited to SELECT statements.
122 $rs->create({ baz => 20 });
124 That is equivalent to
126 $schema->resultset('some_table')->create({
132 Note that C<get_column()> returns a ResultSetColumn object. This will behave almost
133 exactly like a resultset, except it has methods tuned for working with columns.
137 If a resultset is used in a numeric context it returns the L</count>.
138 However, if it is used in a booleand context it is always true. So if
139 you want to check if a resultset has any results use C<if $rs != 0>.
140 C<if $rs> will always be true.
148 =item Arguments: $source, \%$attrs
150 =item Return Value: $rs
154 The resultset constructor. Takes a source object (usually a
155 L<DBIx::Class::ResultSourceProxy::Table>) and an attribute hash (see
156 L</ATTRIBUTES> below). Does not perform any queries -- these are
157 executed as needed by the other methods.
159 Generally you won't need to construct a resultset manually. You'll
160 automatically get one from e.g. a L</search> called in scalar context:
162 my $rs = $schema->resultset('CD')->search({ title => '100th Window' });
164 IMPORTANT: If called on an object, proxies to new_result instead so
166 my $cd = $schema->resultset('CD')->new({ title => 'Spoon' });
168 will return a CD object, not a ResultSet.
174 return $class->new_result(@_) if ref $class;
176 my ($source, $attrs) = @_;
177 $source = $source->handle
178 unless $source->isa('DBIx::Class::ResultSourceHandle');
179 $attrs = { %{$attrs||{}} };
181 if ($attrs->{page}) {
182 $attrs->{rows} ||= 10;
185 $attrs->{alias} ||= 'me';
187 # Creation of {} and bless separated to mitigate RH perl bug
188 # see https://bugzilla.redhat.com/show_bug.cgi?id=196836
190 _source_handle => $source,
191 cond => $attrs->{where},
200 $attrs->{result_class} || $source->resolve->result_class
210 =item Arguments: $cond, \%attrs?
212 =item Return Value: $resultset (scalar context), @row_objs (list context)
216 my @cds = $cd_rs->search({ year => 2001 }); # "... WHERE year = 2001"
217 my $new_rs = $cd_rs->search({ year => 2005 });
219 my $new_rs = $cd_rs->search([ { year => 2005 }, { year => 2004 } ]);
220 # year = 2005 OR year = 2004
222 If you need to pass in additional attributes but no additional condition,
223 call it as C<search(undef, \%attrs)>.
225 # "SELECT name, artistid FROM $artist_table"
226 my @all_artists = $schema->resultset('Artist')->search(undef, {
227 columns => [qw/name artistid/],
230 For a list of attributes that can be passed to C<search>, see
231 L</ATTRIBUTES>. For more examples of using this function, see
232 L<Searching|DBIx::Class::Manual::Cookbook/Searching>. For a complete
233 documentation for the first argument, see L<SQL::Abstract>.
235 For more help on using joins with search, see L<DBIx::Class::Manual::Joining>.
241 my $rs = $self->search_rs( @_ );
242 return (wantarray ? $rs->all : $rs);
249 =item Arguments: $cond, \%attrs?
251 =item Return Value: $resultset
255 This method does the same exact thing as search() except it will
256 always return a resultset, even in list context.
264 $attrs = pop(@_) if @_ > 1 and ref $_[$#_] eq 'HASH';
265 my $our_attrs = { %{$self->{attrs}} };
266 my $having = delete $our_attrs->{having};
267 my $where = delete $our_attrs->{where};
271 my %safe = (alias => 1, cache => 1);
274 (@_ && defined($_[0])) # @_ == () or (undef)
276 (keys %$attrs # empty attrs or only 'safe' attrs
277 && List::Util::first { !$safe{$_} } keys %$attrs)
279 # no search, effectively just a clone
280 $rows = $self->get_cache;
283 my $new_attrs = { %{$our_attrs}, %{$attrs} };
285 # merge new attrs into inherited
286 foreach my $key (qw/join prefetch +select +as/) {
287 next unless exists $attrs->{$key};
288 $new_attrs->{$key} = $self->_merge_attr($our_attrs->{$key}, $attrs->{$key});
293 (@_ == 1 || ref $_[0] eq "HASH")
295 (ref $_[0] eq 'HASH')
297 (keys %{ $_[0] } > 0)
305 ? $self->throw_exception("Odd number of arguments to search")
312 if (defined $where) {
313 $new_attrs->{where} = (
314 defined $new_attrs->{where}
317 ref $_ eq 'ARRAY' ? [ -or => $_ ] : $_
318 } $where, $new_attrs->{where}
325 $new_attrs->{where} = (
326 defined $new_attrs->{where}
329 ref $_ eq 'ARRAY' ? [ -or => $_ ] : $_
330 } $cond, $new_attrs->{where}
336 if (defined $having) {
337 $new_attrs->{having} = (
338 defined $new_attrs->{having}
341 ref $_ eq 'ARRAY' ? [ -or => $_ ] : $_
342 } $having, $new_attrs->{having}
348 my $rs = (ref $self)->new($self->result_source, $new_attrs);
350 $rs->set_cache($rows);
355 =head2 search_literal
359 =item Arguments: $sql_fragment, @bind_values
361 =item Return Value: $resultset (scalar context), @row_objs (list context)
365 my @cds = $cd_rs->search_literal('year = ? AND title = ?', qw/2001 Reload/);
366 my $newrs = $artist_rs->search_literal('name = ?', 'Metallica');
368 Pass a literal chunk of SQL to be added to the conditional part of the
371 CAVEAT: C<search_literal> is provided for Class::DBI compatibility and should
372 only be used in that context. There are known problems using C<search_literal>
373 in chained queries; it can result in bind values in the wrong order. See
374 L<DBIx::Class::Manual::Cookbook/Searching> and
375 L<DBIx::Class::Manual::FAQ/Searching> for searching techniques that do not
376 require C<search_literal>.
381 my ($self, $cond, @vals) = @_;
382 my $attrs = (ref $vals[$#vals] eq 'HASH' ? { %{ pop(@vals) } } : {});
383 $attrs->{bind} = [ @{$self->{attrs}{bind}||[]}, @vals ];
384 return $self->search(\$cond, $attrs);
391 =item Arguments: @values | \%cols, \%attrs?
393 =item Return Value: $row_object | undef
397 Finds a row based on its primary key or unique constraint. For example, to find
398 a row by its primary key:
400 my $cd = $schema->resultset('CD')->find(5);
402 You can also find a row by a specific unique constraint using the C<key>
403 attribute. For example:
405 my $cd = $schema->resultset('CD')->find('Massive Attack', 'Mezzanine', {
406 key => 'cd_artist_title'
409 Additionally, you can specify the columns explicitly by name:
411 my $cd = $schema->resultset('CD')->find(
413 artist => 'Massive Attack',
414 title => 'Mezzanine',
416 { key => 'cd_artist_title' }
419 If the C<key> is specified as C<primary>, it searches only on the primary key.
421 If no C<key> is specified, it searches on all unique constraints defined on the
422 source for which column data is provided, including the primary key.
424 If your table does not have a primary key, you B<must> provide a value for the
425 C<key> attribute matching one of the unique constraints on the source.
427 In addition to C<key>, L</find> recognizes and applies standard
428 L<resultset attributes|/ATTRIBUTES> in the same way as L</search> does.
430 Note: If your query does not return only one row, a warning is generated:
432 Query returned more than one row
434 See also L</find_or_create> and L</update_or_create>. For information on how to
435 declare unique constraints, see
436 L<DBIx::Class::ResultSource/add_unique_constraint>.
442 my $attrs = (@_ > 1 && ref $_[$#_] eq 'HASH' ? pop(@_) : {});
444 # Default to the primary key, but allow a specific key
445 my @cols = exists $attrs->{key}
446 ? $self->result_source->unique_constraint_columns($attrs->{key})
447 : $self->result_source->primary_columns;
448 $self->throw_exception(
449 "Can't find unless a primary key is defined or unique constraint is specified"
452 # Parse out a hashref from input
454 if (ref $_[0] eq 'HASH') {
455 $input_query = { %{$_[0]} };
457 elsif (@_ == @cols) {
459 @{$input_query}{@cols} = @_;
462 # Compatibility: Allow e.g. find(id => $value)
463 carp "Find by key => value deprecated; please use a hashref instead";
467 my (%related, $info);
469 KEY: foreach my $key (keys %$input_query) {
470 if (ref($input_query->{$key})
471 && ($info = $self->result_source->relationship_info($key))) {
472 my $val = delete $input_query->{$key};
473 next KEY if (ref($val) eq 'ARRAY'); # has_many for multi_create
474 my $rel_q = $self->result_source->resolve_condition(
475 $info->{cond}, $val, $key
477 die "Can't handle OR join condition in find" if ref($rel_q) eq 'ARRAY';
478 @related{keys %$rel_q} = values %$rel_q;
481 if (my @keys = keys %related) {
482 @{$input_query}{@keys} = values %related;
486 # Build the final query: Default to the disjunction of the unique queries,
487 # but allow the input query in case the ResultSet defines the query or the
488 # user is abusing find
489 my $alias = exists $attrs->{alias} ? $attrs->{alias} : $self->{attrs}{alias};
491 if (exists $attrs->{key}) {
492 my @unique_cols = $self->result_source->unique_constraint_columns($attrs->{key});
493 my $unique_query = $self->_build_unique_query($input_query, \@unique_cols);
494 $query = $self->_add_alias($unique_query, $alias);
497 my @unique_queries = $self->_unique_queries($input_query, $attrs);
498 $query = @unique_queries
499 ? [ map { $self->_add_alias($_, $alias) } @unique_queries ]
500 : $self->_add_alias($input_query, $alias);
505 my $rs = $self->search($query, $attrs);
506 if (keys %{$rs->_resolved_attrs->{collapse}}) {
508 carp "Query returned more than one row" if $rs->next;
516 if (keys %{$self->_resolved_attrs->{collapse}}) {
517 my $rs = $self->search($query);
519 carp "Query returned more than one row" if $rs->next;
523 return $self->single($query);
530 # Add the specified alias to the specified query hash. A copy is made so the
531 # original query is not modified.
534 my ($self, $query, $alias) = @_;
536 my %aliased = %$query;
537 foreach my $col (grep { ! m/\./ } keys %aliased) {
538 $aliased{"$alias.$col"} = delete $aliased{$col};
546 # Build a list of queries which satisfy unique constraints.
548 sub _unique_queries {
549 my ($self, $query, $attrs) = @_;
551 my @constraint_names = exists $attrs->{key}
553 : $self->result_source->unique_constraint_names;
555 my $where = $self->_collapse_cond($self->{attrs}{where} || {});
556 my $num_where = scalar keys %$where;
559 foreach my $name (@constraint_names) {
560 my @unique_cols = $self->result_source->unique_constraint_columns($name);
561 my $unique_query = $self->_build_unique_query($query, \@unique_cols);
563 my $num_cols = scalar @unique_cols;
564 my $num_query = scalar keys %$unique_query;
566 my $total = $num_query + $num_where;
567 if ($num_query && ($num_query == $num_cols || $total == $num_cols)) {
568 # The query is either unique on its own or is unique in combination with
569 # the existing where clause
570 push @unique_queries, $unique_query;
574 return @unique_queries;
577 # _build_unique_query
579 # Constrain the specified query hash based on the specified column names.
581 sub _build_unique_query {
582 my ($self, $query, $unique_cols) = @_;
585 map { $_ => $query->{$_} }
586 grep { exists $query->{$_} }
591 =head2 search_related
595 =item Arguments: $rel, $cond, \%attrs?
597 =item Return Value: $new_resultset
601 $new_rs = $cd_rs->search_related('artist', {
605 Searches the specified relationship, optionally specifying a condition and
606 attributes for matching records. See L</ATTRIBUTES> for more information.
611 return shift->related_resultset(shift)->search(@_);
614 =head2 search_related_rs
616 This method works exactly the same as search_related, except that
617 it guarantees a restultset, even in list context.
621 sub search_related_rs {
622 return shift->related_resultset(shift)->search_rs(@_);
629 =item Arguments: none
631 =item Return Value: $cursor
635 Returns a storage-driven cursor to the given resultset. See
636 L<DBIx::Class::Cursor> for more information.
643 my $attrs = { %{$self->_resolved_attrs} };
644 return $self->{cursor}
645 ||= $self->result_source->storage->select($attrs->{from}, $attrs->{select},
646 $attrs->{where},$attrs);
653 =item Arguments: $cond?
655 =item Return Value: $row_object?
659 my $cd = $schema->resultset('CD')->single({ year => 2001 });
661 Inflates the first result without creating a cursor if the resultset has
662 any records in it; if not returns nothing. Used by L</find> as a lean version of
665 While this method can take an optional search condition (just like L</search>)
666 being a fast-code-path it does not recognize search attributes. If you need to
667 add extra joins or similar, call L</search> and then chain-call L</single> on the
668 L<DBIx::Class::ResultSet> returned.
674 As of 0.08100, this method enforces the assumption that the preceeding
675 query returns only one row. If more than one row is returned, you will receive
678 Query returned more than one row
680 In this case, you should be using L</first> or L</find> instead, or if you really
681 know what you are doing, use the L</rows> attribute to explicitly limit the size
689 my ($self, $where) = @_;
691 $self->throw_exception('single() only takes search conditions, no attributes. You want ->search( $cond, $attrs )->single()');
694 my $attrs = { %{$self->_resolved_attrs} };
696 if (defined $attrs->{where}) {
699 [ map { ref $_ eq 'ARRAY' ? [ -or => $_ ] : $_ }
700 $where, delete $attrs->{where} ]
703 $attrs->{where} = $where;
707 # XXX: Disabled since it doesn't infer uniqueness in all cases
708 # unless ($self->_is_unique_query($attrs->{where})) {
709 # carp "Query not guaranteed to return a single row"
710 # . "; please declare your unique constraints or use search instead";
713 my @data = $self->result_source->storage->select_single(
714 $attrs->{from}, $attrs->{select},
715 $attrs->{where}, $attrs
718 return (@data ? ($self->_construct_object(@data))[0] : undef);
723 # Try to determine if the specified query is guaranteed to be unique, based on
724 # the declared unique constraints.
726 sub _is_unique_query {
727 my ($self, $query) = @_;
729 my $collapsed = $self->_collapse_query($query);
730 my $alias = $self->{attrs}{alias};
732 foreach my $name ($self->result_source->unique_constraint_names) {
733 my @unique_cols = map {
735 } $self->result_source->unique_constraint_columns($name);
737 # Count the values for each unique column
738 my %seen = map { $_ => 0 } @unique_cols;
740 foreach my $key (keys %$collapsed) {
741 my $aliased = $key =~ /\./ ? $key : "$alias.$key";
742 next unless exists $seen{$aliased}; # Additional constraints are okay
743 $seen{$aliased} = scalar keys %{ $collapsed->{$key} };
746 # If we get 0 or more than 1 value for a column, it's not necessarily unique
747 return 1 unless grep { $_ != 1 } values %seen;
755 # Recursively collapse the query, accumulating values for each column.
757 sub _collapse_query {
758 my ($self, $query, $collapsed) = @_;
762 if (ref $query eq 'ARRAY') {
763 foreach my $subquery (@$query) {
764 next unless ref $subquery; # -or
765 # warn "ARRAY: " . Dumper $subquery;
766 $collapsed = $self->_collapse_query($subquery, $collapsed);
769 elsif (ref $query eq 'HASH') {
770 if (keys %$query and (keys %$query)[0] eq '-and') {
771 foreach my $subquery (@{$query->{-and}}) {
772 # warn "HASH: " . Dumper $subquery;
773 $collapsed = $self->_collapse_query($subquery, $collapsed);
777 # warn "LEAF: " . Dumper $query;
778 foreach my $col (keys %$query) {
779 my $value = $query->{$col};
780 $collapsed->{$col}{$value}++;
792 =item Arguments: $cond?
794 =item Return Value: $resultsetcolumn
798 my $max_length = $rs->get_column('length')->max;
800 Returns a L<DBIx::Class::ResultSetColumn> instance for a column of the ResultSet.
805 my ($self, $column) = @_;
806 my $new = DBIx::Class::ResultSetColumn->new($self, $column);
814 =item Arguments: $cond, \%attrs?
816 =item Return Value: $resultset (scalar context), @row_objs (list context)
820 # WHERE title LIKE '%blue%'
821 $cd_rs = $rs->search_like({ title => '%blue%'});
823 Performs a search, but uses C<LIKE> instead of C<=> as the condition. Note
824 that this is simply a convenience method retained for ex Class::DBI users.
825 You most likely want to use L</search> with specific operators.
827 For more information, see L<DBIx::Class::Manual::Cookbook>.
833 my $attrs = (@_ > 1 && ref $_[$#_] eq 'HASH' ? pop(@_) : {});
834 my $query = ref $_[0] eq 'HASH' ? { %{shift()} }: {@_};
835 $query->{$_} = { 'like' => $query->{$_} } for keys %$query;
836 return $class->search($query, { %$attrs });
843 =item Arguments: $first, $last
845 =item Return Value: $resultset (scalar context), @row_objs (list context)
849 Returns a resultset or object list representing a subset of elements from the
850 resultset slice is called on. Indexes are from 0, i.e., to get the first
853 my ($one, $two, $three) = $rs->slice(0, 2);
858 my ($self, $min, $max) = @_;
859 my $attrs = {}; # = { %{ $self->{attrs} || {} } };
860 $attrs->{offset} = $self->{attrs}{offset} || 0;
861 $attrs->{offset} += $min;
862 $attrs->{rows} = ($max ? ($max - $min + 1) : 1);
863 return $self->search(undef(), $attrs);
864 #my $slice = (ref $self)->new($self->result_source, $attrs);
865 #return (wantarray ? $slice->all : $slice);
872 =item Arguments: none
874 =item Return Value: $result?
878 Returns the next element in the resultset (C<undef> is there is none).
880 Can be used to efficiently iterate over records in the resultset:
882 my $rs = $schema->resultset('CD')->search;
883 while (my $cd = $rs->next) {
887 Note that you need to store the resultset object, and call C<next> on it.
888 Calling C<< resultset('Table')->next >> repeatedly will always return the
889 first record from the resultset.
895 if (my $cache = $self->get_cache) {
896 $self->{all_cache_position} ||= 0;
897 return $cache->[$self->{all_cache_position}++];
899 if ($self->{attrs}{cache}) {
900 $self->{all_cache_position} = 1;
901 return ($self->all)[0];
903 if ($self->{stashed_objects}) {
904 my $obj = shift(@{$self->{stashed_objects}});
905 delete $self->{stashed_objects} unless @{$self->{stashed_objects}};
909 exists $self->{stashed_row}
910 ? @{delete $self->{stashed_row}}
911 : $self->cursor->next
913 return undef unless (@row);
914 my ($row, @more) = $self->_construct_object(@row);
915 $self->{stashed_objects} = \@more if @more;
919 sub _construct_object {
920 my ($self, @row) = @_;
921 my $info = $self->_collapse_result($self->{_attrs}{as}, \@row);
922 my @new = $self->result_class->inflate_result($self->result_source, @$info);
923 @new = $self->{_attrs}{record_filter}->(@new)
924 if exists $self->{_attrs}{record_filter};
928 sub _collapse_result {
929 my ($self, $as_proto, $row) = @_;
933 # 'foo' => [ undef, 'foo' ]
934 # 'foo.bar' => [ 'foo', 'bar' ]
935 # 'foo.bar.baz' => [ 'foo.bar', 'baz' ]
937 my @construct_as = map { [ (/^(?:(.*)\.)?([^.]+)$/) ] } @$as_proto;
939 my %collapse = %{$self->{_attrs}{collapse}||{}};
943 # if we're doing collapsing (has_many prefetch) we need to grab records
944 # until the PK changes, so fill @pri_index. if not, we leave it empty so
945 # we know we don't have to bother.
947 # the reason for not using the collapse stuff directly is because if you
948 # had for e.g. two artists in a row with no cds, the collapse info for
949 # both would be NULL (undef) so you'd lose the second artist
951 # store just the index so we can check the array positions from the row
952 # without having to contruct the full hash
954 if (keys %collapse) {
955 my %pri = map { ($_ => 1) } $self->result_source->primary_columns;
956 foreach my $i (0 .. $#construct_as) {
957 next if defined($construct_as[$i][0]); # only self table
958 if (delete $pri{$construct_as[$i][1]}) {
959 push(@pri_index, $i);
961 last unless keys %pri; # short circuit (Johnny Five Is Alive!)
965 # no need to do an if, it'll be empty if @pri_index is empty anyway
967 my %pri_vals = map { ($_ => $copy[$_]) } @pri_index;
971 do { # no need to check anything at the front, we always want the first row
975 foreach my $this_as (@construct_as) {
976 $const{$this_as->[0]||''}{$this_as->[1]} = shift(@copy);
979 push(@const_rows, \%const);
981 } until ( # no pri_index => no collapse => drop straight out
984 do { # get another row, stash it, drop out if different PK
986 @copy = $self->cursor->next;
987 $self->{stashed_row} = \@copy;
989 # last thing in do block, counts as true if anything doesn't match
991 # check xor defined first for NULL vs. NOT NULL then if one is
992 # defined the other must be so check string equality
995 (defined $pri_vals{$_} ^ defined $copy[$_])
996 || (defined $pri_vals{$_} && ($pri_vals{$_} ne $copy[$_]))
1001 my $alias = $self->{attrs}{alias};
1008 foreach my $const (@const_rows) {
1009 scalar @const_keys or do {
1010 @const_keys = sort { length($a) <=> length($b) } keys %$const;
1012 foreach my $key (@const_keys) {
1015 my @parts = split(/\./, $key);
1017 my $data = $const->{$key};
1018 foreach my $p (@parts) {
1019 $target = $target->[1]->{$p} ||= [];
1021 if ($cur eq ".${key}" && (my @ckey = @{$collapse{$cur}||[]})) {
1022 # collapsing at this point and on final part
1023 my $pos = $collapse_pos{$cur};
1024 CK: foreach my $ck (@ckey) {
1025 if (!defined $pos->{$ck} || $pos->{$ck} ne $data->{$ck}) {
1026 $collapse_pos{$cur} = $data;
1027 delete @collapse_pos{ # clear all positioning for sub-entries
1028 grep { m/^\Q${cur}.\E/ } keys %collapse_pos
1035 if (exists $collapse{$cur}) {
1036 $target = $target->[-1];
1039 $target->[0] = $data;
1041 $info->[0] = $const->{$key};
1049 =head2 result_source
1053 =item Arguments: $result_source?
1055 =item Return Value: $result_source
1059 An accessor for the primary ResultSource object from which this ResultSet
1066 =item Arguments: $result_class?
1068 =item Return Value: $result_class
1072 An accessor for the class to use when creating row objects. Defaults to
1073 C<< result_source->result_class >> - which in most cases is the name of the
1074 L<"table"|DBIx::Class::Manual::Glossary/"ResultSource"> class.
1079 my ($self, $result_class) = @_;
1080 if ($result_class) {
1081 $self->ensure_class_loaded($result_class);
1082 $self->_result_class($result_class);
1084 $self->_result_class;
1091 =item Arguments: $cond, \%attrs??
1093 =item Return Value: $count
1097 Performs an SQL C<COUNT> with the same query as the resultset was built
1098 with to find the number of elements. If passed arguments, does a search
1099 on the resultset and counts the results of that.
1101 Note: When using C<count> with C<group_by>, L<DBIx::Class> emulates C<GROUP BY>
1102 using C<COUNT( DISTINCT( columns ) )>. Some databases (notably SQLite) do
1103 not support C<DISTINCT> with multiple columns. If you are using such a
1104 database, you should only use columns from the main table in your C<group_by>
1111 return $self->search(@_)->count if @_ and defined $_[0];
1112 return scalar @{ $self->get_cache } if $self->get_cache;
1113 my $count = $self->_count;
1114 return 0 unless $count;
1116 # need to take offset from resolved attrs
1118 $count -= $self->{_attrs}{offset} if $self->{_attrs}{offset};
1119 $count = $self->{attrs}{rows} if
1120 $self->{attrs}{rows} and $self->{attrs}{rows} < $count;
1121 $count = 0 if ($count < 0);
1125 sub _count { # Separated out so pager can get the full count
1127 my $select = { count => '*' };
1129 my $attrs = { %{$self->_resolved_attrs} };
1130 if (my $group_by = delete $attrs->{group_by}) {
1131 delete $attrs->{having};
1132 my @distinct = (ref $group_by ? @$group_by : ($group_by));
1133 # todo: try CONCAT for multi-column pk
1134 my @pk = $self->result_source->primary_columns;
1136 my $alias = $attrs->{alias};
1137 foreach my $column (@distinct) {
1138 if ($column =~ qr/^(?:\Q${alias}.\E)?$pk[0]$/) {
1139 @distinct = ($column);
1145 $select = { count => { distinct => \@distinct } };
1148 $attrs->{select} = $select;
1149 $attrs->{as} = [qw/count/];
1151 # offset, order by and page are not needed to count. record_filter is cdbi
1152 delete $attrs->{$_} for qw/rows offset order_by page pager record_filter/;
1154 my $tmp_rs = (ref $self)->new($self->result_source, $attrs);
1155 my ($count) = $tmp_rs->cursor->next;
1163 =head2 count_literal
1167 =item Arguments: $sql_fragment, @bind_values
1169 =item Return Value: $count
1173 Counts the results in a literal query. Equivalent to calling L</search_literal>
1174 with the passed arguments, then L</count>.
1178 sub count_literal { shift->search_literal(@_)->count; }
1184 =item Arguments: none
1186 =item Return Value: @objects
1190 Returns all elements in the resultset. Called implicitly if the resultset
1191 is returned in list context.
1198 $self->throw_exception("all() doesn't take any arguments, you probably wanted ->search(...)->all()");
1201 return @{ $self->get_cache } if $self->get_cache;
1205 # TODO: don't call resolve here
1206 if (keys %{$self->_resolved_attrs->{collapse}}) {
1207 # if ($self->{attrs}{prefetch}) {
1208 # Using $self->cursor->all is really just an optimisation.
1209 # If we're collapsing has_many prefetches it probably makes
1210 # very little difference, and this is cleaner than hacking
1211 # _construct_object to survive the approach
1212 my @row = $self->cursor->next;
1214 push(@obj, $self->_construct_object(@row));
1215 @row = (exists $self->{stashed_row}
1216 ? @{delete $self->{stashed_row}}
1217 : $self->cursor->next);
1220 @obj = map { $self->_construct_object(@$_) } $self->cursor->all;
1223 $self->set_cache(\@obj) if $self->{attrs}{cache};
1231 =item Arguments: none
1233 =item Return Value: $self
1237 Resets the resultset's cursor, so you can iterate through the elements again.
1243 delete $self->{_attrs} if exists $self->{_attrs};
1244 $self->{all_cache_position} = 0;
1245 $self->cursor->reset;
1253 =item Arguments: none
1255 =item Return Value: $object?
1259 Resets the resultset and returns an object for the first result (if the
1260 resultset returns anything).
1265 return $_[0]->reset->next;
1268 # _cond_for_update_delete
1270 # update/delete require the condition to be modified to handle
1271 # the differing SQL syntax available. This transforms the $self->{cond}
1272 # appropriately, returning the new condition.
1274 sub _cond_for_update_delete {
1275 my ($self, $full_cond) = @_;
1278 $full_cond ||= $self->{cond};
1279 # No-op. No condition, we're updating/deleting everything
1280 return $cond unless ref $full_cond;
1282 if (ref $full_cond eq 'ARRAY') {
1286 foreach my $key (keys %{$_}) {
1288 $hash{$1} = $_->{$key};
1294 elsif (ref $full_cond eq 'HASH') {
1295 if ((keys %{$full_cond})[0] eq '-and') {
1298 my @cond = @{$full_cond->{-and}};
1299 for (my $i = 0; $i < @cond; $i++) {
1300 my $entry = $cond[$i];
1303 if (ref $entry eq 'HASH') {
1304 $hash = $self->_cond_for_update_delete($entry);
1307 $entry =~ /([^.]+)$/;
1308 $hash->{$1} = $cond[++$i];
1311 push @{$cond->{-and}}, $hash;
1315 foreach my $key (keys %{$full_cond}) {
1317 $cond->{$1} = $full_cond->{$key};
1322 $self->throw_exception(
1323 "Can't update/delete on resultset with condition unless hash or array"
1335 =item Arguments: \%values
1337 =item Return Value: $storage_rv
1341 Sets the specified columns in the resultset to the supplied values in a
1342 single query. Return value will be true if the update succeeded or false
1343 if no records were updated; exact type of success value is storage-dependent.
1348 my ($self, $values) = @_;
1349 $self->throw_exception("Values for update must be a hash")
1350 unless ref $values eq 'HASH';
1352 carp( 'WARNING! Currently $rs->update() does not generate proper SQL'
1353 . ' on joined resultsets, and may affect rows well outside of the'
1354 . ' contents of $rs. Use at your own risk' )
1355 if ( $self->{attrs}{seen_join} );
1357 my $cond = $self->_cond_for_update_delete;
1359 return $self->result_source->storage->update(
1360 $self->result_source, $values, $cond
1368 =item Arguments: \%values
1370 =item Return Value: 1
1374 Fetches all objects and updates them one at a time. Note that C<update_all>
1375 will run DBIC cascade triggers, while L</update> will not.
1380 my ($self, $values) = @_;
1381 $self->throw_exception("Values for update must be a hash")
1382 unless ref $values eq 'HASH';
1383 foreach my $obj ($self->all) {
1384 $obj->set_columns($values)->update;
1393 =item Arguments: none
1395 =item Return Value: 1
1399 Deletes the contents of the resultset from its result source. Note that this
1400 will not run DBIC cascade triggers. See L</delete_all> if you need triggers
1401 to run. See also L<DBIx::Class::Row/delete>.
1403 delete may not generate correct SQL for a query with joins or a resultset
1404 chained from a related resultset. In this case it will generate a warning:-
1406 WARNING! Currently $rs->delete() does not generate proper SQL on
1407 joined resultsets, and may delete rows well outside of the contents
1408 of $rs. Use at your own risk
1410 In these cases you may find that delete_all is more appropriate, or you
1411 need to respecify your query in a way that can be expressed without a join.
1417 $self->throw_exception("Delete should not be passed any arguments")
1419 carp( 'WARNING! Currently $rs->delete() does not generate proper SQL'
1420 . ' on joined resultsets, and may delete rows well outside of the'
1421 . ' contents of $rs. Use at your own risk' )
1422 if ( $self->{attrs}{seen_join} );
1423 my $cond = $self->_cond_for_update_delete;
1425 $self->result_source->storage->delete($self->result_source, $cond);
1433 =item Arguments: none
1435 =item Return Value: 1
1439 Fetches all objects and deletes them one at a time. Note that C<delete_all>
1440 will run DBIC cascade triggers, while L</delete> will not.
1446 $_->delete for $self->all;
1454 =item Arguments: \@data;
1458 Accepts either an arrayref of hashrefs or alternatively an arrayref of arrayrefs.
1459 For the arrayref of hashrefs style each hashref should be a structure suitable
1460 forsubmitting to a $resultset->create(...) method.
1462 In void context, C<insert_bulk> in L<DBIx::Class::Storage::DBI> is used
1463 to insert the data, as this is a faster method.
1465 Otherwise, each set of data is inserted into the database using
1466 L<DBIx::Class::ResultSet/create>, and a arrayref of the resulting row
1467 objects is returned.
1469 Example: Assuming an Artist Class that has many CDs Classes relating:
1471 my $Artist_rs = $schema->resultset("Artist");
1473 ## Void Context Example
1474 $Artist_rs->populate([
1475 { artistid => 4, name => 'Manufactured Crap', cds => [
1476 { title => 'My First CD', year => 2006 },
1477 { title => 'Yet More Tweeny-Pop crap', year => 2007 },
1480 { artistid => 5, name => 'Angsty-Whiny Girl', cds => [
1481 { title => 'My parents sold me to a record company' ,year => 2005 },
1482 { title => 'Why Am I So Ugly?', year => 2006 },
1483 { title => 'I Got Surgery and am now Popular', year => 2007 }
1488 ## Array Context Example
1489 my ($ArtistOne, $ArtistTwo, $ArtistThree) = $Artist_rs->populate([
1490 { name => "Artist One"},
1491 { name => "Artist Two"},
1492 { name => "Artist Three", cds=> [
1493 { title => "First CD", year => 2007},
1494 { title => "Second CD", year => 2008},
1498 print $ArtistOne->name; ## response is 'Artist One'
1499 print $ArtistThree->cds->count ## reponse is '2'
1501 For the arrayref of arrayrefs style, the first element should be a list of the
1502 fieldsnames to which the remaining elements are rows being inserted. For
1505 $Arstist_rs->populate([
1506 [qw/artistid name/],
1507 [100, 'A Formally Unknown Singer'],
1508 [101, 'A singer that jumped the shark two albums ago'],
1509 [102, 'An actually cool singer.'],
1512 Please note an important effect on your data when choosing between void and
1513 wantarray context. Since void context goes straight to C<insert_bulk> in
1514 L<DBIx::Class::Storage::DBI> this will skip any component that is overriding
1515 c<insert>. So if you are using something like L<DBIx-Class-UUIDColumns> to
1516 create primary keys for you, you will find that your PKs are empty. In this
1517 case you will have to use the wantarray context in order to create those
1523 my $self = shift @_;
1524 my $data = ref $_[0][0] eq 'HASH'
1525 ? $_[0] : ref $_[0][0] eq 'ARRAY' ? $self->_normalize_populate_args($_[0]) :
1526 $self->throw_exception('Populate expects an arrayref of hashes or arrayref of arrayrefs');
1528 if(defined wantarray) {
1530 foreach my $item (@$data) {
1531 push(@created, $self->create($item));
1535 my ($first, @rest) = @$data;
1537 my @names = grep {!ref $first->{$_}} keys %$first;
1538 my @rels = grep { $self->result_source->has_relationship($_) } keys %$first;
1539 my @pks = $self->result_source->primary_columns;
1541 ## do the belongs_to relationships
1542 foreach my $index (0..$#$data) {
1543 if( grep { !defined $data->[$index]->{$_} } @pks ) {
1544 my @ret = $self->populate($data);
1548 foreach my $rel (@rels) {
1549 next unless $data->[$index]->{$rel} && ref $data->[$index]->{$rel} eq "HASH";
1550 my $result = $self->related_resultset($rel)->create($data->[$index]->{$rel});
1551 my ($reverse) = keys %{$self->result_source->reverse_relationship_info($rel)};
1552 my $related = $result->result_source->resolve_condition(
1553 $result->result_source->relationship_info($reverse)->{cond},
1558 delete $data->[$index]->{$rel};
1559 $data->[$index] = {%{$data->[$index]}, %$related};
1561 push @names, keys %$related if $index == 0;
1565 ## do bulk insert on current row
1566 my @values = map { [ @$_{@names} ] } @$data;
1568 $self->result_source->storage->insert_bulk(
1569 $self->result_source,
1574 ## do the has_many relationships
1575 foreach my $item (@$data) {
1577 foreach my $rel (@rels) {
1578 next unless $item->{$rel} && ref $item->{$rel} eq "ARRAY";
1580 my $parent = $self->find(map {{$_=>$item->{$_}} } @pks)
1581 || $self->throw_exception('Cannot find the relating object.');
1583 my $child = $parent->$rel;
1585 my $related = $child->result_source->resolve_condition(
1586 $parent->result_source->relationship_info($rel)->{cond},
1591 my @rows_to_add = ref $item->{$rel} eq 'ARRAY' ? @{$item->{$rel}} : ($item->{$rel});
1592 my @populate = map { {%$_, %$related} } @rows_to_add;
1594 $child->populate( \@populate );
1600 =head2 _normalize_populate_args ($args)
1602 Private method used by L</populate> to normalize its incoming arguments. Factored
1603 out in case you want to subclass and accept new argument structures to the
1604 L</populate> method.
1608 sub _normalize_populate_args {
1609 my ($self, $data) = @_;
1610 my @names = @{shift(@$data)};
1611 my @results_to_create;
1612 foreach my $datum (@$data) {
1613 my %result_to_create;
1614 foreach my $index (0..$#names) {
1615 $result_to_create{$names[$index]} = $$datum[$index];
1617 push @results_to_create, \%result_to_create;
1619 return \@results_to_create;
1626 =item Arguments: none
1628 =item Return Value: $pager
1632 Return Value a L<Data::Page> object for the current resultset. Only makes
1633 sense for queries with a C<page> attribute.
1639 my $attrs = $self->{attrs};
1640 $self->throw_exception("Can't create pager for non-paged rs")
1641 unless $self->{attrs}{page};
1642 $attrs->{rows} ||= 10;
1643 return $self->{pager} ||= Data::Page->new(
1644 $self->_count, $attrs->{rows}, $self->{attrs}{page});
1651 =item Arguments: $page_number
1653 =item Return Value: $rs
1657 Returns a resultset for the $page_number page of the resultset on which page
1658 is called, where each page contains a number of rows equal to the 'rows'
1659 attribute set on the resultset (10 by default).
1664 my ($self, $page) = @_;
1665 return (ref $self)->new($self->result_source, { %{$self->{attrs}}, page => $page });
1672 =item Arguments: \%vals
1674 =item Return Value: $rowobject
1678 Creates a new row object in the resultset's result class and returns
1679 it. The row is not inserted into the database at this point, call
1680 L<DBIx::Class::Row/insert> to do that. Calling L<DBIx::Class::Row/in_storage>
1681 will tell you whether the row object has been inserted or not.
1683 Passes the hashref of input on to L<DBIx::Class::Row/new>.
1688 my ($self, $values) = @_;
1689 $self->throw_exception( "new_result needs a hash" )
1690 unless (ref $values eq 'HASH');
1693 my $alias = $self->{attrs}{alias};
1696 defined $self->{cond}
1697 && $self->{cond} eq $DBIx::Class::ResultSource::UNRESOLVABLE_CONDITION
1699 %new = %{$self->{attrs}{related_objects}};
1701 $self->throw_exception(
1702 "Can't abstract implicit construct, condition not a hash"
1703 ) if ($self->{cond} && !(ref $self->{cond} eq 'HASH'));
1705 my $collapsed_cond = (
1707 ? $self->_collapse_cond($self->{cond})
1711 # precendence must be given to passed values over values inherited from
1712 # the cond, so the order here is important.
1713 my %implied = %{$self->_remove_alias($collapsed_cond, $alias)};
1714 while( my($col,$value) = each %implied ){
1715 if(ref($value) eq 'HASH' && keys(%$value) && (keys %$value)[0] eq '='){
1716 $new{$col} = $value->{'='};
1719 $new{$col} = $value if $self->_is_deterministic_value($value);
1725 %{ $self->_remove_alias($values, $alias) },
1726 -source_handle => $self->_source_handle,
1727 -result_source => $self->result_source, # DO NOT REMOVE THIS, REQUIRED
1730 return $self->result_class->new(\%new);
1733 # _is_deterministic_value
1735 # Make an effor to strip non-deterministic values from the condition,
1736 # to make sure new_result chokes less
1738 sub _is_deterministic_value {
1741 my $ref_type = ref $value;
1742 return 1 if $ref_type eq '' || $ref_type eq 'SCALAR';
1743 return 1 if Scalar::Util::blessed($value);
1749 # Recursively collapse the condition.
1751 sub _collapse_cond {
1752 my ($self, $cond, $collapsed) = @_;
1756 if (ref $cond eq 'ARRAY') {
1757 foreach my $subcond (@$cond) {
1758 next unless ref $subcond; # -or
1759 # warn "ARRAY: " . Dumper $subcond;
1760 $collapsed = $self->_collapse_cond($subcond, $collapsed);
1763 elsif (ref $cond eq 'HASH') {
1764 if (keys %$cond and (keys %$cond)[0] eq '-and') {
1765 foreach my $subcond (@{$cond->{-and}}) {
1766 # warn "HASH: " . Dumper $subcond;
1767 $collapsed = $self->_collapse_cond($subcond, $collapsed);
1771 # warn "LEAF: " . Dumper $cond;
1772 foreach my $col (keys %$cond) {
1773 my $value = $cond->{$col};
1774 $collapsed->{$col} = $value;
1784 # Remove the specified alias from the specified query hash. A copy is made so
1785 # the original query is not modified.
1788 my ($self, $query, $alias) = @_;
1790 my %orig = %{ $query || {} };
1793 foreach my $key (keys %orig) {
1795 $unaliased{$key} = $orig{$key};
1798 $unaliased{$1} = $orig{$key}
1799 if $key =~ m/^(?:\Q$alias\E\.)?([^.]+)$/;
1809 =item Arguments: none
1811 =item Return Value: \[ $sql, @bind ]
1815 Returns the SQL query and bind vars associated with the invocant.
1817 This is generally used as the RHS for a subquery.
1821 sub as_query { return shift->cursor->as_query(@_) }
1827 =item Arguments: \%vals, \%attrs?
1829 =item Return Value: $rowobject
1833 my $artist = $schema->resultset('Artist')->find_or_new(
1834 { artist => 'fred' }, { key => 'artists' });
1836 $cd->cd_to_producer->find_or_new({ producer => $producer },
1837 { key => 'primary });
1839 Find an existing record from this resultset, based on its primary
1840 key, or a unique constraint. If none exists, instantiate a new result
1841 object and return it. The object will not be saved into your storage
1842 until you call L<DBIx::Class::Row/insert> on it.
1844 You most likely want this method when looking for existing rows using
1845 a unique constraint that is not the primary key, or looking for
1848 If you want objects to be saved immediately, use L</find_or_create> instead.
1850 B<Note>: C<find_or_new> is probably not what you want when creating a
1851 new row in a table that uses primary keys supplied by the
1852 database. Passing in a primary key column with a value of I<undef>
1853 will cause L</find> to attempt to search for a row with a value of
1860 my $attrs = (@_ > 1 && ref $_[$#_] eq 'HASH' ? pop(@_) : {});
1861 my $hash = ref $_[0] eq 'HASH' ? shift : {@_};
1862 my $exists = $self->find($hash, $attrs);
1863 return defined $exists ? $exists : $self->new_result($hash);
1870 =item Arguments: \%vals
1872 =item Return Value: a L<DBIx::Class::Row> $object
1876 Attempt to create a single new row or a row with multiple related rows
1877 in the table represented by the resultset (and related tables). This
1878 will not check for duplicate rows before inserting, use
1879 L</find_or_create> to do that.
1881 To create one row for this resultset, pass a hashref of key/value
1882 pairs representing the columns of the table and the values you wish to
1883 store. If the appropriate relationships are set up, foreign key fields
1884 can also be passed an object representing the foreign row, and the
1885 value will be set to its primary key.
1887 To create related objects, pass a hashref for the value if the related
1888 item is a foreign key relationship (L<DBIx::Class::Relationship/belongs_to>),
1889 and use the name of the relationship as the key. (NOT the name of the field,
1890 necessarily). For C<has_many> and C<has_one> relationships, pass an arrayref
1891 of hashrefs containing the data for each of the rows to create in the foreign
1892 tables, again using the relationship name as the key.
1894 Instead of hashrefs of plain related data (key/value pairs), you may
1895 also pass new or inserted objects. New objects (not inserted yet, see
1896 L</new>), will be inserted into their appropriate tables.
1898 Effectively a shortcut for C<< ->new_result(\%vals)->insert >>.
1900 Example of creating a new row.
1902 $person_rs->create({
1903 name=>"Some Person",
1904 email=>"somebody@someplace.com"
1907 Example of creating a new row and also creating rows in a related C<has_many>
1908 or C<has_one> resultset. Note Arrayref.
1911 { artistid => 4, name => 'Manufactured Crap', cds => [
1912 { title => 'My First CD', year => 2006 },
1913 { title => 'Yet More Tweeny-Pop crap', year => 2007 },
1918 Example of creating a new row and also creating a row in a related
1919 C<belongs_to>resultset. Note Hashref.
1922 title=>"Music for Silly Walks",
1925 name=>"Silly Musician",
1932 my ($self, $attrs) = @_;
1933 $self->throw_exception( "create needs a hashref" )
1934 unless ref $attrs eq 'HASH';
1935 return $self->new_result($attrs)->insert;
1938 =head2 find_or_create
1942 =item Arguments: \%vals, \%attrs?
1944 =item Return Value: $rowobject
1948 $cd->cd_to_producer->find_or_create({ producer => $producer },
1949 { key => 'primary });
1951 Tries to find a record based on its primary key or unique constraints; if none
1952 is found, creates one and returns that instead.
1954 my $cd = $schema->resultset('CD')->find_or_create({
1956 artist => 'Massive Attack',
1957 title => 'Mezzanine',
1961 Also takes an optional C<key> attribute, to search by a specific key or unique
1962 constraint. For example:
1964 my $cd = $schema->resultset('CD')->find_or_create(
1966 artist => 'Massive Attack',
1967 title => 'Mezzanine',
1969 { key => 'cd_artist_title' }
1972 B<Note>: Because find_or_create() reads from the database and then
1973 possibly inserts based on the result, this method is subject to a race
1974 condition. Another process could create a record in the table after
1975 the find has completed and before the create has started. To avoid
1976 this problem, use find_or_create() inside a transaction.
1978 B<Note>: C<find_or_create> is probably not what you want when creating
1979 a new row in a table that uses primary keys supplied by the
1980 database. Passing in a primary key column with a value of I<undef>
1981 will cause L</find> to attempt to search for a row with a value of
1984 See also L</find> and L</update_or_create>. For information on how to declare
1985 unique constraints, see L<DBIx::Class::ResultSource/add_unique_constraint>.
1989 sub find_or_create {
1991 my $attrs = (@_ > 1 && ref $_[$#_] eq 'HASH' ? pop(@_) : {});
1992 my $hash = ref $_[0] eq 'HASH' ? shift : {@_};
1993 my $exists = $self->find($hash, $attrs);
1994 return defined $exists ? $exists : $self->create($hash);
1997 =head2 update_or_create
2001 =item Arguments: \%col_values, { key => $unique_constraint }?
2003 =item Return Value: $rowobject
2007 $resultset->update_or_create({ col => $val, ... });
2009 First, searches for an existing row matching one of the unique constraints
2010 (including the primary key) on the source of this resultset. If a row is
2011 found, updates it with the other given column values. Otherwise, creates a new
2014 Takes an optional C<key> attribute to search on a specific unique constraint.
2017 # In your application
2018 my $cd = $schema->resultset('CD')->update_or_create(
2020 artist => 'Massive Attack',
2021 title => 'Mezzanine',
2024 { key => 'cd_artist_title' }
2027 $cd->cd_to_producer->update_or_create({
2028 producer => $producer,
2035 If no C<key> is specified, it searches on all unique constraints defined on the
2036 source, including the primary key.
2038 If the C<key> is specified as C<primary>, it searches only on the primary key.
2040 See also L</find> and L</find_or_create>. For information on how to declare
2041 unique constraints, see L<DBIx::Class::ResultSource/add_unique_constraint>.
2043 B<Note>: C<update_or_create> is probably not what you want when
2044 looking for a row in a table that uses primary keys supplied by the
2045 database, unless you actually have a key value. Passing in a primary
2046 key column with a value of I<undef> will cause L</find> to attempt to
2047 search for a row with a value of I<NULL>.
2051 sub update_or_create {
2053 my $attrs = (@_ > 1 && ref $_[$#_] eq 'HASH' ? pop(@_) : {});
2054 my $cond = ref $_[0] eq 'HASH' ? shift : {@_};
2056 my $row = $self->find($cond, $attrs);
2058 $row->update($cond);
2062 return $self->create($cond);
2069 =item Arguments: none
2071 =item Return Value: \@cache_objects?
2075 Gets the contents of the cache for the resultset, if the cache is set.
2077 The cache is populated either by using the L</prefetch> attribute to
2078 L</search> or by calling L</set_cache>.
2090 =item Arguments: \@cache_objects
2092 =item Return Value: \@cache_objects
2096 Sets the contents of the cache for the resultset. Expects an arrayref
2097 of objects of the same class as those produced by the resultset. Note that
2098 if the cache is set the resultset will return the cached objects rather
2099 than re-querying the database even if the cache attr is not set.
2101 The contents of the cache can also be populated by using the
2102 L</prefetch> attribute to L</search>.
2107 my ( $self, $data ) = @_;
2108 $self->throw_exception("set_cache requires an arrayref")
2109 if defined($data) && (ref $data ne 'ARRAY');
2110 $self->{all_cache} = $data;
2117 =item Arguments: none
2119 =item Return Value: []
2123 Clears the cache for the resultset.
2128 shift->set_cache(undef);
2131 =head2 related_resultset
2135 =item Arguments: $relationship_name
2137 =item Return Value: $resultset
2141 Returns a related resultset for the supplied relationship name.
2143 $artist_rs = $schema->resultset('CD')->related_resultset('Artist');
2147 sub related_resultset {
2148 my ($self, $rel) = @_;
2150 $self->{related_resultsets} ||= {};
2151 return $self->{related_resultsets}{$rel} ||= do {
2152 my $rel_obj = $self->result_source->relationship_info($rel);
2154 $self->throw_exception(
2155 "search_related: result source '" . $self->result_source->source_name .
2156 "' has no such relationship $rel")
2159 my ($from,$seen) = $self->_resolve_from($rel);
2161 my $join_count = $seen->{$rel};
2162 my $alias = ($join_count > 1 ? join('_', $rel, $join_count) : $rel);
2164 #XXX - temp fix for result_class bug. There likely is a more elegant fix -groditi
2165 my %attrs = %{$self->{attrs}||{}};
2166 delete @attrs{qw(result_class alias)};
2170 if (my $cache = $self->get_cache) {
2171 if ($cache->[0] && $cache->[0]->related_resultset($rel)->get_cache) {
2172 $new_cache = [ map { @{$_->related_resultset($rel)->get_cache} }
2177 my $rel_source = $self->result_source->related_source($rel);
2181 # The reason we do this now instead of passing the alias to the
2182 # search_rs below is that if you wrap/overload resultset on the
2183 # source you need to know what alias it's -going- to have for things
2184 # to work sanely (e.g. RestrictWithObject wants to be able to add
2185 # extra query restrictions, and these may need to be $alias.)
2187 my $attrs = $rel_source->resultset_attributes;
2188 local $attrs->{alias} = $alias;
2190 $rel_source->resultset
2198 where => $self->{cond},
2203 $new->set_cache($new_cache) if $new_cache;
2208 =head2 current_source_alias
2212 =item Arguments: none
2214 =item Return Value: $source_alias
2218 Returns the current table alias for the result source this resultset is built
2219 on, that will be used in the SQL query. Usually it is C<me>.
2221 Currently the source alias that refers to the result set returned by a
2222 L</search>/L</find> family method depends on how you got to the resultset: it's
2223 C<me> by default, but eg. L</search_related> aliases it to the related result
2224 source name (and keeps C<me> referring to the original result set). The long
2225 term goal is to make L<DBIx::Class> always alias the current resultset as C<me>
2226 (and make this method unnecessary).
2228 Thus it's currently necessary to use this method in predefined queries (see
2229 L<DBIx::Class::Manual::Cookbook/Predefined searches>) when referring to the
2230 source alias of the current result set:
2232 # in a result set class
2234 my ($self, $user) = @_;
2236 my $me = $self->current_source_alias;
2238 return $self->search(
2239 "$me.modified" => $user->id,
2245 sub current_source_alias {
2248 return ($self->{attrs} || {})->{alias} || 'me';
2252 my ($self, $extra_join) = @_;
2253 my $source = $self->result_source;
2254 my $attrs = $self->{attrs};
2256 my $from = $attrs->{from}
2257 || [ { $attrs->{alias} => $source->from } ];
2259 my $seen = { %{$attrs->{seen_join}||{}} };
2261 my $join = ($attrs->{join}
2262 ? [ $attrs->{join}, $extra_join ]
2265 # we need to take the prefetch the attrs into account before we
2266 # ->resolve_join as otherwise they get lost - captainL
2267 my $merged = $self->_merge_attr( $join, $attrs->{prefetch} );
2271 ($join ? $source->resolve_join($merged, $attrs->{alias}, $seen) : ()),
2274 return ($from,$seen);
2277 sub _resolved_attrs {
2279 return $self->{_attrs} if $self->{_attrs};
2281 my $attrs = { %{$self->{attrs}||{}} };
2282 my $source = $self->result_source;
2283 my $alias = $attrs->{alias};
2285 $attrs->{columns} ||= delete $attrs->{cols} if exists $attrs->{cols};
2286 if ($attrs->{columns}) {
2287 delete $attrs->{as};
2288 } elsif (!$attrs->{select}) {
2289 $attrs->{columns} = [ $source->columns ];
2294 ? (ref $attrs->{select} eq 'ARRAY'
2295 ? [ @{$attrs->{select}} ]
2296 : [ $attrs->{select} ])
2297 : [ map { m/\./ ? $_ : "${alias}.$_" } @{delete $attrs->{columns}} ]
2301 ? (ref $attrs->{as} eq 'ARRAY'
2302 ? [ @{$attrs->{as}} ]
2304 : [ map { m/^\Q${alias}.\E(.+)$/ ? $1 : $_ } @{$attrs->{select}} ]
2308 if ($adds = delete $attrs->{include_columns}) {
2309 $adds = [$adds] unless ref $adds eq 'ARRAY';
2310 push(@{$attrs->{select}}, @$adds);
2311 push(@{$attrs->{as}}, map { m/([^.]+)$/; $1 } @$adds);
2313 if ($adds = delete $attrs->{'+select'}) {
2314 $adds = [$adds] unless ref $adds eq 'ARRAY';
2315 push(@{$attrs->{select}},
2316 map { /\./ || ref $_ ? $_ : "${alias}.$_" } @$adds);
2318 if (my $adds = delete $attrs->{'+as'}) {
2319 $adds = [$adds] unless ref $adds eq 'ARRAY';
2320 push(@{$attrs->{as}}, @$adds);
2323 $attrs->{from} ||= [ { $self->{attrs}{alias} => $source->from } ];
2325 if (exists $attrs->{join} || exists $attrs->{prefetch}) {
2326 my $join = delete $attrs->{join} || {};
2328 if (defined $attrs->{prefetch}) {
2329 $join = $self->_merge_attr(
2330 $join, $attrs->{prefetch}
2335 $attrs->{from} = # have to copy here to avoid corrupting the original
2338 $source->resolve_join($join, $alias, { %{$attrs->{seen_join}||{}} })
2343 $attrs->{group_by} ||= $attrs->{select} if delete $attrs->{distinct};
2344 if ($attrs->{order_by}) {
2345 $attrs->{order_by} = (ref($attrs->{order_by}) eq 'ARRAY'
2346 ? [ @{$attrs->{order_by}} ]
2347 : [ $attrs->{order_by} ]);
2349 $attrs->{order_by} = [];
2352 my $collapse = $attrs->{collapse} || {};
2353 if (my $prefetch = delete $attrs->{prefetch}) {
2354 $prefetch = $self->_merge_attr({}, $prefetch);
2356 my $seen = { %{ $attrs->{seen_join} || {} } };
2357 foreach my $p (ref $prefetch eq 'ARRAY' ? @$prefetch : ($prefetch)) {
2358 # bring joins back to level of current class
2359 my @prefetch = $source->resolve_prefetch(
2360 $p, $alias, $seen, \@pre_order, $collapse
2362 push(@{$attrs->{select}}, map { $_->[0] } @prefetch);
2363 push(@{$attrs->{as}}, map { $_->[1] } @prefetch);
2365 push(@{$attrs->{order_by}}, @pre_order);
2367 $attrs->{collapse} = $collapse;
2369 if ($attrs->{page}) {
2370 $attrs->{offset} ||= 0;
2371 $attrs->{offset} += ($attrs->{rows} * ($attrs->{page} - 1));
2374 return $self->{_attrs} = $attrs;
2378 my ($self, $attr) = @_;
2380 if (ref $attr eq 'HASH') {
2381 return $self->_rollout_hash($attr);
2382 } elsif (ref $attr eq 'ARRAY') {
2383 return $self->_rollout_array($attr);
2389 sub _rollout_array {
2390 my ($self, $attr) = @_;
2393 foreach my $element (@{$attr}) {
2394 if (ref $element eq 'HASH') {
2395 push( @rolled_array, @{ $self->_rollout_hash( $element ) } );
2396 } elsif (ref $element eq 'ARRAY') {
2397 # XXX - should probably recurse here
2398 push( @rolled_array, @{$self->_rollout_array($element)} );
2400 push( @rolled_array, $element );
2403 return \@rolled_array;
2407 my ($self, $attr) = @_;
2410 foreach my $key (keys %{$attr}) {
2411 push( @rolled_array, { $key => $attr->{$key} } );
2413 return \@rolled_array;
2416 sub _calculate_score {
2417 my ($self, $a, $b) = @_;
2419 if (ref $b eq 'HASH') {
2420 my ($b_key) = keys %{$b};
2421 if (ref $a eq 'HASH') {
2422 my ($a_key) = keys %{$a};
2423 if ($a_key eq $b_key) {
2424 return (1 + $self->_calculate_score( $a->{$a_key}, $b->{$b_key} ));
2429 return ($a eq $b_key) ? 1 : 0;
2432 if (ref $a eq 'HASH') {
2433 my ($a_key) = keys %{$a};
2434 return ($b eq $a_key) ? 1 : 0;
2436 return ($b eq $a) ? 1 : 0;
2442 my ($self, $orig, $import) = @_;
2444 return $import unless defined($orig);
2445 return $orig unless defined($import);
2447 $orig = $self->_rollout_attr($orig);
2448 $import = $self->_rollout_attr($import);
2451 foreach my $import_element ( @{$import} ) {
2452 # find best candidate from $orig to merge $b_element into
2453 my $best_candidate = { position => undef, score => 0 }; my $position = 0;
2454 foreach my $orig_element ( @{$orig} ) {
2455 my $score = $self->_calculate_score( $orig_element, $import_element );
2456 if ($score > $best_candidate->{score}) {
2457 $best_candidate->{position} = $position;
2458 $best_candidate->{score} = $score;
2462 my ($import_key) = ( ref $import_element eq 'HASH' ) ? keys %{$import_element} : ($import_element);
2464 if ($best_candidate->{score} == 0 || exists $seen_keys->{$import_key}) {
2465 push( @{$orig}, $import_element );
2467 my $orig_best = $orig->[$best_candidate->{position}];
2468 # merge orig_best and b_element together and replace original with merged
2469 if (ref $orig_best ne 'HASH') {
2470 $orig->[$best_candidate->{position}] = $import_element;
2471 } elsif (ref $import_element eq 'HASH') {
2472 my ($key) = keys %{$orig_best};
2473 $orig->[$best_candidate->{position}] = { $key => $self->_merge_attr($orig_best->{$key}, $import_element->{$key}) };
2476 $seen_keys->{$import_key} = 1; # don't merge the same key twice
2486 $self->_source_handle($_[0]->handle);
2488 $self->_source_handle->resolve;
2492 =head2 throw_exception
2494 See L<DBIx::Class::Schema/throw_exception> for details.
2498 sub throw_exception {
2500 if (ref $self && $self->_source_handle->schema) {
2501 $self->_source_handle->schema->throw_exception(@_)
2508 # XXX: FIXME: Attributes docs need clearing up
2512 Attributes are used to refine a ResultSet in various ways when
2513 searching for data. They can be passed to any method which takes an
2514 C<\%attrs> argument. See L</search>, L</search_rs>, L</find>,
2517 These are in no particular order:
2523 =item Value: ($order_by | \@order_by)
2527 Which column(s) to order the results by. This is currently passed
2528 through directly to SQL, so you can give e.g. C<year DESC> for a
2529 descending order on the column `year'.
2531 Please note that if you have C<quote_char> enabled (see
2532 L<DBIx::Class::Storage::DBI/connect_info>) you will need to do C<\'year DESC' > to
2533 specify an order. (The scalar ref causes it to be passed as raw sql to the DB,
2534 so you will need to manually quote things as appropriate.)
2536 If your L<SQL::Abstract> version supports it (>=1.50), you can also use
2537 C<{-desc => 'year'}>, which takes care of the quoting for you. This is the
2544 =item Value: \@columns
2548 Shortcut to request a particular set of columns to be retrieved. Adds
2549 C<me.> onto the start of any column without a C<.> in it and sets C<select>
2550 from that, then auto-populates C<as> from C<select> as normal. (You may also
2551 use the C<cols> attribute, as in earlier versions of DBIC.)
2553 =head2 include_columns
2557 =item Value: \@columns
2561 Shortcut to include additional columns in the returned results - for example
2563 $schema->resultset('CD')->search(undef, {
2564 include_columns => ['artist.name'],
2568 would return all CDs and include a 'name' column to the information
2569 passed to object inflation. Note that the 'artist' is the name of the
2570 column (or relationship) accessor, and 'name' is the name of the column
2571 accessor in the related table.
2577 =item Value: \@select_columns
2581 Indicates which columns should be selected from the storage. You can use
2582 column names, or in the case of RDBMS back ends, function or stored procedure
2585 $rs = $schema->resultset('Employee')->search(undef, {
2588 { count => 'employeeid' },
2593 When you use function/stored procedure names and do not supply an C<as>
2594 attribute, the column names returned are storage-dependent. E.g. MySQL would
2595 return a column named C<count(employeeid)> in the above example.
2601 Indicates additional columns to be selected from storage. Works the same as
2602 L</select> but adds columns to the selection.
2610 Indicates additional column names for those added via L</+select>. See L</as>.
2618 =item Value: \@inflation_names
2622 Indicates column names for object inflation. That is, C<as>
2623 indicates the name that the column can be accessed as via the
2624 C<get_column> method (or via the object accessor, B<if one already
2625 exists>). It has nothing to do with the SQL code C<SELECT foo AS bar>.
2627 The C<as> attribute is used in conjunction with C<select>,
2628 usually when C<select> contains one or more function or stored
2631 $rs = $schema->resultset('Employee')->search(undef, {
2634 { count => 'employeeid' }
2636 as => ['name', 'employee_count'],
2639 my $employee = $rs->first(); # get the first Employee
2641 If the object against which the search is performed already has an accessor
2642 matching a column name specified in C<as>, the value can be retrieved using
2643 the accessor as normal:
2645 my $name = $employee->name();
2647 If on the other hand an accessor does not exist in the object, you need to
2648 use C<get_column> instead:
2650 my $employee_count = $employee->get_column('employee_count');
2652 You can create your own accessors if required - see
2653 L<DBIx::Class::Manual::Cookbook> for details.
2655 Please note: This will NOT insert an C<AS employee_count> into the SQL
2656 statement produced, it is used for internal access only. Thus
2657 attempting to use the accessor in an C<order_by> clause or similar
2658 will fail miserably.
2660 To get around this limitation, you can supply literal SQL to your
2661 C<select> attibute that contains the C<AS alias> text, eg:
2663 select => [\'myfield AS alias']
2669 =item Value: ($rel_name | \@rel_names | \%rel_names)
2673 Contains a list of relationships that should be joined for this query. For
2676 # Get CDs by Nine Inch Nails
2677 my $rs = $schema->resultset('CD')->search(
2678 { 'artist.name' => 'Nine Inch Nails' },
2679 { join => 'artist' }
2682 Can also contain a hash reference to refer to the other relation's relations.
2685 package MyApp::Schema::Track;
2686 use base qw/DBIx::Class/;
2687 __PACKAGE__->table('track');
2688 __PACKAGE__->add_columns(qw/trackid cd position title/);
2689 __PACKAGE__->set_primary_key('trackid');
2690 __PACKAGE__->belongs_to(cd => 'MyApp::Schema::CD');
2693 # In your application
2694 my $rs = $schema->resultset('Artist')->search(
2695 { 'track.title' => 'Teardrop' },
2697 join => { cd => 'track' },
2698 order_by => 'artist.name',
2702 You need to use the relationship (not the table) name in conditions,
2703 because they are aliased as such. The current table is aliased as "me", so
2704 you need to use me.column_name in order to avoid ambiguity. For example:
2706 # Get CDs from 1984 with a 'Foo' track
2707 my $rs = $schema->resultset('CD')->search(
2710 'tracks.name' => 'Foo'
2712 { join => 'tracks' }
2715 If the same join is supplied twice, it will be aliased to <rel>_2 (and
2716 similarly for a third time). For e.g.
2718 my $rs = $schema->resultset('Artist')->search({
2719 'cds.title' => 'Down to Earth',
2720 'cds_2.title' => 'Popular',
2722 join => [ qw/cds cds/ ],
2725 will return a set of all artists that have both a cd with title 'Down
2726 to Earth' and a cd with title 'Popular'.
2728 If you want to fetch related objects from other tables as well, see C<prefetch>
2731 For more help on using joins with search, see L<DBIx::Class::Manual::Joining>.
2737 =item Value: ($rel_name | \@rel_names | \%rel_names)
2741 Contains one or more relationships that should be fetched along with
2742 the main query (when they are accessed afterwards the data will
2743 already be available, without extra queries to the database). This is
2744 useful for when you know you will need the related objects, because it
2745 saves at least one query:
2747 my $rs = $schema->resultset('Tag')->search(
2756 The initial search results in SQL like the following:
2758 SELECT tag.*, cd.*, artist.* FROM tag
2759 JOIN cd ON tag.cd = cd.cdid
2760 JOIN artist ON cd.artist = artist.artistid
2762 L<DBIx::Class> has no need to go back to the database when we access the
2763 C<cd> or C<artist> relationships, which saves us two SQL statements in this
2766 Simple prefetches will be joined automatically, so there is no need
2767 for a C<join> attribute in the above search.
2769 C<prefetch> can be used with the following relationship types: C<belongs_to>,
2770 C<has_one> (or if you're using C<add_relationship>, any relationship declared
2771 with an accessor type of 'single' or 'filter'). A more complex example that
2772 prefetches an artists cds, the tracks on those cds, and the tags associted
2773 with that artist is given below (assuming many-to-many from artists to tags):
2775 my $rs = $schema->resultset('Artist')->search(
2779 { cds => 'tracks' },
2780 { artist_tags => 'tags' }
2786 B<NOTE:> If you specify a C<prefetch> attribute, the C<join> and C<select>
2787 attributes will be ignored.
2797 Makes the resultset paged and specifies the page to retrieve. Effectively
2798 identical to creating a non-pages resultset and then calling ->page($page)
2801 If L<rows> attribute is not specified it defualts to 10 rows per page.
2811 Specifes the maximum number of rows for direct retrieval or the number of
2812 rows per page if the page attribute or method is used.
2818 =item Value: $offset
2822 Specifies the (zero-based) row number for the first row to be returned, or the
2823 of the first row of the first page if paging is used.
2829 =item Value: \@columns
2833 A arrayref of columns to group by. Can include columns of joined tables.
2835 group_by => [qw/ column1 column2 ... /]
2841 =item Value: $condition
2845 HAVING is a select statement attribute that is applied between GROUP BY and
2846 ORDER BY. It is applied to the after the grouping calculations have been
2849 having => { 'count(employee)' => { '>=', 100 } }
2855 =item Value: (0 | 1)
2859 Set to 1 to group by all columns.
2865 Adds to the WHERE clause.
2867 # only return rows WHERE deleted IS NULL for all searches
2868 __PACKAGE__->resultset_attributes({ where => { deleted => undef } }); )
2870 Can be overridden by passing C<{ where => undef }> as an attribute
2877 Set to 1 to cache search results. This prevents extra SQL queries if you
2878 revisit rows in your ResultSet:
2880 my $resultset = $schema->resultset('Artist')->search( undef, { cache => 1 } );
2882 while( my $artist = $resultset->next ) {
2886 $rs->first; # without cache, this would issue a query
2888 By default, searches are not cached.
2890 For more examples of using these attributes, see
2891 L<DBIx::Class::Manual::Cookbook>.
2897 =item Value: \@from_clause
2901 The C<from> attribute gives you manual control over the C<FROM> clause of SQL
2902 statements generated by L<DBIx::Class>, allowing you to express custom C<JOIN>
2905 NOTE: Use this on your own risk. This allows you to shoot off your foot!
2907 C<join> will usually do what you need and it is strongly recommended that you
2908 avoid using C<from> unless you cannot achieve the desired result using C<join>.
2909 And we really do mean "cannot", not just tried and failed. Attempting to use
2910 this because you're having problems with C<join> is like trying to use x86
2911 ASM because you've got a syntax error in your C. Trust us on this.
2913 Now, if you're still really, really sure you need to use this (and if you're
2914 not 100% sure, ask the mailing list first), here's an explanation of how this
2917 The syntax is as follows -
2920 { <alias1> => <table1> },
2922 { <alias2> => <table2>, -join_type => 'inner|left|right' },
2923 [], # nested JOIN (optional)
2924 { <table1.column1> => <table2.column2>, ... (more conditions) },
2926 # More of the above [ ] may follow for additional joins
2933 ON <table1.column1> = <table2.column2>
2934 <more joins may follow>
2936 An easy way to follow the examples below is to remember the following:
2938 Anything inside "[]" is a JOIN
2939 Anything inside "{}" is a condition for the enclosing JOIN
2941 The following examples utilize a "person" table in a family tree application.
2942 In order to express parent->child relationships, this table is self-joined:
2944 # Person->belongs_to('father' => 'Person');
2945 # Person->belongs_to('mother' => 'Person');
2947 C<from> can be used to nest joins. Here we return all children with a father,
2948 then search against all mothers of those children:
2950 $rs = $schema->resultset('Person')->search(
2953 alias => 'mother', # alias columns in accordance with "from"
2955 { mother => 'person' },
2958 { child => 'person' },
2960 { father => 'person' },
2961 { 'father.person_id' => 'child.father_id' }
2964 { 'mother.person_id' => 'child.mother_id' }
2971 # SELECT mother.* FROM person mother
2974 # JOIN person father
2975 # ON ( father.person_id = child.father_id )
2977 # ON ( mother.person_id = child.mother_id )
2979 The type of any join can be controlled manually. To search against only people
2980 with a father in the person table, we could explicitly use C<INNER JOIN>:
2982 $rs = $schema->resultset('Person')->search(
2985 alias => 'child', # alias columns in accordance with "from"
2987 { child => 'person' },
2989 { father => 'person', -join_type => 'inner' },
2990 { 'father.id' => 'child.father_id' }
2997 # SELECT child.* FROM person child
2998 # INNER JOIN person father ON child.father_id = father.id
3000 If you need to express really complex joins or you need a subselect, you
3001 can supply literal SQL to C<from> via a scalar reference. In this case
3002 the contents of the scalar will replace the table name asscoiated with the
3005 WARNING: This technique might very well not work as expected on chained
3006 searches - you have been warned.
3008 # Assuming the Event resultsource is defined as:
3010 MySchema::Event->add_columns (
3013 is_auto_increment => 1,
3022 MySchema::Event->set_primary_key ('sequence');
3024 # This will get back the latest event for every location. The column
3025 # selector is still provided by DBIC, all we do is add a JOIN/WHERE
3026 # combo to limit the resultset
3028 $rs = $schema->resultset('Event');
3029 $table = $rs->result_source->name;
3030 $latest = $rs->search (
3033 (SELECT e1.* FROM $table e1
3035 ON e1.location = e2.location
3036 AND e1.sequence < e2.sequence
3037 WHERE e2.sequence is NULL
3042 # Equivalent SQL (with the DBIC chunks added):
3044 SELECT me.sequence, me.location, me.type FROM
3045 (SELECT e1.* FROM events e1
3047 ON e1.location = e2.location
3048 AND e1.sequence < e2.sequence
3049 WHERE e2.sequence is NULL
3056 =item Value: ( 'update' | 'shared' )
3060 Set to 'update' for a SELECT ... FOR UPDATE or 'shared' for a SELECT