1 package DBIx::Class::Storage::DBI;
2 # -*- mode: cperl; cperl-indent-level: 2 -*-
7 use base 'DBIx::Class::Storage';
10 use Carp::Clan qw/^DBIx::Class/;
12 use DBIx::Class::Storage::DBI::Cursor;
13 use DBIx::Class::Storage::Statistics;
16 use Data::Dumper::Concise();
18 # what version of sqlt do we require if deploy() without a ddl_dir is invoked
19 # when changing also adjust the corresponding author_require in Makefile.PL
20 my $minimum_sqlt_version = '0.11002';
23 __PACKAGE__->mk_group_accessors('simple' =>
24 qw/_connect_info _dbi_connect_info _dbh _sql_maker _sql_maker_opts _conn_pid
25 _conn_tid transaction_depth _dbh_autocommit _driver_determined savepoints/
28 # the values for these accessors are picked out (and deleted) from
29 # the attribute hashref passed to connect_info
30 my @storage_options = qw/
31 on_connect_call on_disconnect_call on_connect_do on_disconnect_do
32 disable_sth_caching unsafe auto_savepoint
34 __PACKAGE__->mk_group_accessors('simple' => @storage_options);
37 # default cursor class, overridable in connect_info attributes
38 __PACKAGE__->cursor_class('DBIx::Class::Storage::DBI::Cursor');
40 __PACKAGE__->mk_group_accessors('inherited' => qw/sql_maker_class/);
41 __PACKAGE__->sql_maker_class('DBIx::Class::SQLAHacks');
44 # Each of these methods need _determine_driver called before itself
45 # in order to function reliably. This is a purely DRY optimization
46 my @rdbms_specific_methods = qw/
59 for my $meth (@rdbms_specific_methods) {
61 my $orig = __PACKAGE__->can ($meth)
65 no warnings qw/redefine/;
66 *{__PACKAGE__ ."::$meth"} = sub {
67 if (not $_[0]->_driver_determined) {
68 $_[0]->_determine_driver;
69 goto $_[0]->can($meth);
78 DBIx::Class::Storage::DBI - DBI storage handler
82 my $schema = MySchema->connect('dbi:SQLite:my.db');
84 $schema->storage->debug(1);
86 my @stuff = $schema->storage->dbh_do(
88 my ($storage, $dbh, @args) = @_;
89 $dbh->do("DROP TABLE authors");
94 $schema->resultset('Book')->search({
95 written_on => $schema->storage->datetime_parser(DateTime->now)
100 This class represents the connection to an RDBMS via L<DBI>. See
101 L<DBIx::Class::Storage> for general information. This pod only
102 documents DBI-specific methods and behaviors.
109 my $new = shift->next::method(@_);
111 $new->transaction_depth(0);
112 $new->_sql_maker_opts({});
113 $new->{savepoints} = [];
114 $new->{_in_dbh_do} = 0;
115 $new->{_dbh_gen} = 0;
122 This method is normally called by L<DBIx::Class::Schema/connection>, which
123 encapsulates its argument list in an arrayref before passing them here.
125 The argument list may contain:
131 The same 4-element argument set one would normally pass to
132 L<DBI/connect>, optionally followed by
133 L<extra attributes|/DBIx::Class specific connection attributes>
134 recognized by DBIx::Class:
136 $connect_info_args = [ $dsn, $user, $password, \%dbi_attributes?, \%extra_attributes? ];
140 A single code reference which returns a connected
141 L<DBI database handle|DBI/connect> optionally followed by
142 L<extra attributes|/DBIx::Class specific connection attributes> recognized
145 $connect_info_args = [ sub { DBI->connect (...) }, \%extra_attributes? ];
149 A single hashref with all the attributes and the dsn/user/password
152 $connect_info_args = [{
160 $connect_info_args = [{
161 dbh_maker => sub { DBI->connect (...) },
166 This is particularly useful for L<Catalyst> based applications, allowing the
167 following config (L<Config::General> style):
172 dsn dbi:mysql:database=test
179 The C<dsn>/C<user>/C<password> combination can be substituted by the
180 C<dbh_maker> key whose value is a coderef that returns a connected
181 L<DBI database handle|DBI/connect>
185 Please note that the L<DBI> docs recommend that you always explicitly
186 set C<AutoCommit> to either I<0> or I<1>. L<DBIx::Class> further
187 recommends that it be set to I<1>, and that you perform transactions
188 via our L<DBIx::Class::Schema/txn_do> method. L<DBIx::Class> will set it
189 to I<1> if you do not do explicitly set it to zero. This is the default
190 for most DBDs. See L</DBIx::Class and AutoCommit> for details.
192 =head3 DBIx::Class specific connection attributes
194 In addition to the standard L<DBI|DBI/ATTRIBUTES_COMMON_TO_ALL_HANDLES>
195 L<connection|DBI/Database_Handle_Attributes> attributes, DBIx::Class recognizes
196 the following connection options. These options can be mixed in with your other
197 L<DBI> connection attributes, or placed in a seperate hashref
198 (C<\%extra_attributes>) as shown above.
200 Every time C<connect_info> is invoked, any previous settings for
201 these options will be cleared before setting the new ones, regardless of
202 whether any options are specified in the new C<connect_info>.
209 Specifies things to do immediately after connecting or re-connecting to
210 the database. Its value may contain:
216 This contains one SQL statement to execute.
218 =item an array reference
220 This contains SQL statements to execute in order. Each element contains
221 a string or a code reference that returns a string.
223 =item a code reference
225 This contains some code to execute. Unlike code references within an
226 array reference, its return value is ignored.
230 =item on_disconnect_do
232 Takes arguments in the same form as L</on_connect_do> and executes them
233 immediately before disconnecting from the database.
235 Note, this only runs if you explicitly call L</disconnect> on the
238 =item on_connect_call
240 A more generalized form of L</on_connect_do> that calls the specified
241 C<connect_call_METHOD> methods in your storage driver.
243 on_connect_do => 'select 1'
247 on_connect_call => [ [ do_sql => 'select 1' ] ]
249 Its values may contain:
255 Will call the C<connect_call_METHOD> method.
257 =item a code reference
259 Will execute C<< $code->($storage) >>
261 =item an array reference
263 Each value can be a method name or code reference.
265 =item an array of arrays
267 For each array, the first item is taken to be the C<connect_call_> method name
268 or code reference, and the rest are parameters to it.
272 Some predefined storage methods you may use:
278 Executes a SQL string or a code reference that returns a SQL string. This is
279 what L</on_connect_do> and L</on_disconnect_do> use.
287 Will execute the scalar as SQL.
291 Taken to be arguments to L<DBI/do>, the SQL string optionally followed by the
292 attributes hashref and bind values.
294 =item a code reference
296 Will execute C<< $code->($storage) >> and execute the return array refs as
303 Execute any statements necessary to initialize the database session to return
304 and accept datetime/timestamp values used with
305 L<DBIx::Class::InflateColumn::DateTime>.
307 Only necessary for some databases, see your specific storage driver for
308 implementation details.
312 =item on_disconnect_call
314 Takes arguments in the same form as L</on_connect_call> and executes them
315 immediately before disconnecting from the database.
317 Calls the C<disconnect_call_METHOD> methods as opposed to the
318 C<connect_call_METHOD> methods called by L</on_connect_call>.
320 Note, this only runs if you explicitly call L</disconnect> on the
323 =item disable_sth_caching
325 If set to a true value, this option will disable the caching of
326 statement handles via L<DBI/prepare_cached>.
330 Sets the limit dialect. This is useful for JDBC-bridge among others
331 where the remote SQL-dialect cannot be determined by the name of the
332 driver alone. See also L<SQL::Abstract::Limit>.
336 Specifies what characters to use to quote table and column names. If
337 you use this you will want to specify L</name_sep> as well.
339 C<quote_char> expects either a single character, in which case is it
340 is placed on either side of the table/column name, or an arrayref of length
341 2 in which case the table/column name is placed between the elements.
343 For example under MySQL you should use C<< quote_char => '`' >>, and for
344 SQL Server you should use C<< quote_char => [qw/[ ]/] >>.
348 This only needs to be used in conjunction with C<quote_char>, and is used to
349 specify the charecter that seperates elements (schemas, tables, columns) from
350 each other. In most cases this is simply a C<.>.
352 The consequences of not supplying this value is that L<SQL::Abstract>
353 will assume DBIx::Class' uses of aliases to be complete column
354 names. The output will look like I<"me.name"> when it should actually
359 This Storage driver normally installs its own C<HandleError>, sets
360 C<RaiseError> and C<ShowErrorStatement> on, and sets C<PrintError> off on
361 all database handles, including those supplied by a coderef. It does this
362 so that it can have consistent and useful error behavior.
364 If you set this option to a true value, Storage will not do its usual
365 modifications to the database handle's attributes, and instead relies on
366 the settings in your connect_info DBI options (or the values you set in
367 your connection coderef, in the case that you are connecting via coderef).
369 Note that your custom settings can cause Storage to malfunction,
370 especially if you set a C<HandleError> handler that suppresses exceptions
371 and/or disable C<RaiseError>.
375 If this option is true, L<DBIx::Class> will use savepoints when nesting
376 transactions, making it possible to recover from failure in the inner
377 transaction without having to abort all outer transactions.
381 Use this argument to supply a cursor class other than the default
382 L<DBIx::Class::Storage::DBI::Cursor>.
386 Some real-life examples of arguments to L</connect_info> and
387 L<DBIx::Class::Schema/connect>
389 # Simple SQLite connection
390 ->connect_info([ 'dbi:SQLite:./foo.db' ]);
393 ->connect_info([ sub { DBI->connect(...) } ]);
395 # Connect via subref in hashref
397 dbh_maker => sub { DBI->connect(...) },
398 on_connect_do => 'alter session ...',
401 # A bit more complicated
408 { quote_char => q{"}, name_sep => q{.} },
412 # Equivalent to the previous example
418 { AutoCommit => 1, quote_char => q{"}, name_sep => q{.} },
422 # Same, but with hashref as argument
423 # See parse_connect_info for explanation
426 dsn => 'dbi:Pg:dbname=foo',
428 password => 'my_pg_password',
435 # Subref + DBIx::Class-specific connection options
438 sub { DBI->connect(...) },
442 on_connect_do => ['SET search_path TO myschema,otherschema,public'],
443 disable_sth_caching => 1,
453 my ($self, $info_arg) = @_;
455 return $self->_connect_info if !$info_arg;
457 my @args = @$info_arg; # take a shallow copy for further mutilation
458 $self->_connect_info([@args]); # copy for _connect_info
461 # combine/pre-parse arguments depending on invocation style
464 if (ref $args[0] eq 'CODE') { # coderef with optional \%extra_attributes
465 %attrs = %{ $args[1] || {} };
468 elsif (ref $args[0] eq 'HASH') { # single hashref (i.e. Catalyst config)
469 %attrs = %{$args[0]};
471 if (my $code = delete $attrs{dbh_maker}) {
474 my @ignored = grep { delete $attrs{$_} } (qw/dsn user password/);
477 'Attribute(s) %s in connect_info were ignored, as they can not be applied '
478 . "to the result of 'dbh_maker'",
480 join (', ', map { "'$_'" } (@ignored) ),
485 @args = delete @attrs{qw/dsn user password/};
488 else { # otherwise assume dsn/user/password + \%attrs + \%extra_attrs
490 % { $args[3] || {} },
491 % { $args[4] || {} },
493 @args = @args[0,1,2];
496 # Kill sql_maker/_sql_maker_opts, so we get a fresh one with only
497 # the new set of options
498 $self->_sql_maker(undef);
499 $self->_sql_maker_opts({});
502 for my $storage_opt (@storage_options, 'cursor_class') { # @storage_options is declared at the top of the module
503 if(my $value = delete $attrs{$storage_opt}) {
504 $self->$storage_opt($value);
507 for my $sql_maker_opt (qw/limit_dialect quote_char name_sep/) {
508 if(my $opt_val = delete $attrs{$sql_maker_opt}) {
509 $self->_sql_maker_opts->{$sql_maker_opt} = $opt_val;
514 if (ref $args[0] eq 'CODE') {
515 # _connect() never looks past $args[0] in this case
519 %{ $self->_default_dbi_connect_attributes || {} },
524 $self->_dbi_connect_info([@args, keys %attrs ? \%attrs : ()]);
525 $self->_connect_info;
528 sub _default_dbi_connect_attributes {
538 This method is deprecated in favour of setting via L</connect_info>.
542 =head2 on_disconnect_do
544 This method is deprecated in favour of setting via L</connect_info>.
548 sub _parse_connect_do {
549 my ($self, $type) = @_;
551 my $val = $self->$type;
552 return () if not defined $val;
557 push @res, [ 'do_sql', $val ];
558 } elsif (ref($val) eq 'CODE') {
560 } elsif (ref($val) eq 'ARRAY') {
561 push @res, map { [ 'do_sql', $_ ] } @$val;
563 $self->throw_exception("Invalid type for $type: ".ref($val));
571 Arguments: ($subref | $method_name), @extra_coderef_args?
573 Execute the given $subref or $method_name using the new exception-based
574 connection management.
576 The first two arguments will be the storage object that C<dbh_do> was called
577 on and a database handle to use. Any additional arguments will be passed
578 verbatim to the called subref as arguments 2 and onwards.
580 Using this (instead of $self->_dbh or $self->dbh) ensures correct
581 exception handling and reconnection (or failover in future subclasses).
583 Your subref should have no side-effects outside of the database, as
584 there is the potential for your subref to be partially double-executed
585 if the database connection was stale/dysfunctional.
589 my @stuff = $schema->storage->dbh_do(
591 my ($storage, $dbh, @cols) = @_;
592 my $cols = join(q{, }, @cols);
593 $dbh->selectrow_array("SELECT $cols FROM foo");
604 my $dbh = $self->_get_dbh;
606 return $self->$code($dbh, @_) if $self->{_in_dbh_do}
607 || $self->{transaction_depth};
609 local $self->{_in_dbh_do} = 1;
612 my $want_array = wantarray;
617 @result = $self->$code($dbh, @_);
619 elsif(defined $want_array) {
620 $result[0] = $self->$code($dbh, @_);
623 $self->$code($dbh, @_);
627 # ->connected might unset $@ - copy
629 if(!$exception) { return $want_array ? @result : $result[0] }
631 $self->throw_exception($exception) if $self->connected;
633 # We were not connected - reconnect and retry, but let any
634 # exception fall right through this time
635 carp "Retrying $code after catching disconnected exception: $exception"
636 if $ENV{DBIC_DBIRETRY_DEBUG};
637 $self->_populate_dbh;
638 $self->$code($self->_dbh, @_);
641 # This is basically a blend of dbh_do above and DBIx::Class::Storage::txn_do.
642 # It also informs dbh_do to bypass itself while under the direction of txn_do,
643 # via $self->{_in_dbh_do} (this saves some redundant eval and errorcheck, etc)
648 ref $coderef eq 'CODE' or $self->throw_exception
649 ('$coderef must be a CODE reference');
651 return $coderef->(@_) if $self->{transaction_depth} && ! $self->auto_savepoint;
653 local $self->{_in_dbh_do} = 1;
656 my $want_array = wantarray;
665 @result = $coderef->(@_);
667 elsif(defined $want_array) {
668 $result[0] = $coderef->(@_);
676 # ->connected might unset $@ - copy
678 if(!$exception) { return $want_array ? @result : $result[0] }
680 if($tried++ || $self->connected) {
681 eval { $self->txn_rollback };
682 my $rollback_exception = $@;
683 if($rollback_exception) {
684 my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
685 $self->throw_exception($exception) # propagate nested rollback
686 if $rollback_exception =~ /$exception_class/;
688 $self->throw_exception(
689 "Transaction aborted: ${exception}. "
690 . "Rollback failed: ${rollback_exception}"
693 $self->throw_exception($exception)
696 # We were not connected, and was first try - reconnect and retry
698 carp "Retrying $coderef after catching disconnected exception: $exception"
699 if $ENV{DBIC_DBIRETRY_DEBUG};
700 $self->_populate_dbh;
706 Our C<disconnect> method also performs a rollback first if the
707 database is not in C<AutoCommit> mode.
717 push @actions, ( $self->on_disconnect_call || () );
718 push @actions, $self->_parse_connect_do ('on_disconnect_do');
720 $self->_do_connection_actions(disconnect_call_ => $_) for @actions;
722 $self->_dbh_rollback unless $self->_dbh_autocommit;
724 $self->_dbh->disconnect;
730 =head2 with_deferred_fk_checks
734 =item Arguments: C<$coderef>
736 =item Return Value: The return value of $coderef
740 Storage specific method to run the code ref with FK checks deferred or
741 in MySQL's case disabled entirely.
745 # Storage subclasses should override this
746 sub with_deferred_fk_checks {
747 my ($self, $sub) = @_;
755 =item Arguments: none
757 =item Return Value: 1|0
761 Verifies that the the current database handle is active and ready to execute
762 an SQL statement (i.e. the connection did not get stale, server is still
763 answering, etc.) This method is used internally by L</dbh>.
769 return 0 unless $self->_seems_connected;
772 local $self->_dbh->{RaiseError} = 1;
777 sub _seems_connected {
780 my $dbh = $self->_dbh
783 if(defined $self->_conn_tid && $self->_conn_tid != threads->tid) {
790 return 0 if !$self->_dbh;
793 return $dbh->FETCH('Active');
799 my $dbh = $self->_dbh or return 0;
804 # handle pid changes correctly
805 # NOTE: assumes $self->_dbh is a valid $dbh
809 return if defined $self->_conn_pid && $self->_conn_pid == $$;
811 $self->_dbh->{InactiveDestroy} = 1;
818 sub ensure_connected {
821 unless ($self->connected) {
822 $self->_populate_dbh;
828 Returns a C<$dbh> - a data base handle of class L<DBI>. The returned handle
829 is guaranteed to be healthy by implicitly calling L</connected>, and if
830 necessary performing a reconnection before returning. Keep in mind that this
831 is very B<expensive> on some database engines. Consider using L<dbh_do>
839 if (not $self->_dbh) {
840 $self->_populate_dbh;
842 $self->ensure_connected;
847 # this is the internal "get dbh or connect (don't check)" method
850 $self->_verify_pid if $self->_dbh;
851 $self->_populate_dbh unless $self->_dbh;
855 sub _sql_maker_args {
860 array_datatypes => 1,
861 limit_dialect => $self->_get_dbh,
862 %{$self->_sql_maker_opts}
868 unless ($self->_sql_maker) {
869 my $sql_maker_class = $self->sql_maker_class;
870 $self->ensure_class_loaded ($sql_maker_class);
871 $self->_sql_maker($sql_maker_class->new( $self->_sql_maker_args ));
873 return $self->_sql_maker;
876 # nothing to do by default
883 my @info = @{$self->_dbi_connect_info || []};
884 $self->_dbh(undef); # in case ->connected failed we might get sent here
885 $self->_dbh($self->_connect(@info));
887 $self->_conn_pid($$);
888 $self->_conn_tid(threads->tid) if $INC{'threads.pm'};
890 $self->_determine_driver;
892 # Always set the transaction depth on connect, since
893 # there is no transaction in progress by definition
894 $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
896 $self->_run_connection_actions unless $self->{_in_determine_driver};
899 sub _run_connection_actions {
903 push @actions, ( $self->on_connect_call || () );
904 push @actions, $self->_parse_connect_do ('on_connect_do');
906 $self->_do_connection_actions(connect_call_ => $_) for @actions;
909 sub _determine_driver {
912 if ((not $self->_driver_determined) && (not $self->{_in_determine_driver})) {
913 my $started_connected = 0;
914 local $self->{_in_determine_driver} = 1;
916 if (ref($self) eq __PACKAGE__) {
918 if ($self->_dbh) { # we are connected
919 $driver = $self->_dbh->{Driver}{Name};
920 $started_connected = 1;
922 # if connect_info is a CODEREF, we have no choice but to connect
923 if (ref $self->_dbi_connect_info->[0] &&
924 Scalar::Util::reftype($self->_dbi_connect_info->[0]) eq 'CODE') {
925 $self->_populate_dbh;
926 $driver = $self->_dbh->{Driver}{Name};
929 # try to use dsn to not require being connected, the driver may still
930 # force a connection in _rebless to determine version
931 ($driver) = $self->_dbi_connect_info->[0] =~ /dbi:([^:]+):/i;
935 my $storage_class = "DBIx::Class::Storage::DBI::${driver}";
936 if ($self->load_optional_class($storage_class)) {
937 mro::set_mro($storage_class, 'c3');
938 bless $self, $storage_class;
943 $self->_driver_determined(1);
945 $self->_init; # run driver-specific initializations
947 $self->_run_connection_actions
948 if !$started_connected && defined $self->_dbh;
952 sub _do_connection_actions {
954 my $method_prefix = shift;
957 if (not ref($call)) {
958 my $method = $method_prefix . $call;
960 } elsif (ref($call) eq 'CODE') {
962 } elsif (ref($call) eq 'ARRAY') {
963 if (ref($call->[0]) ne 'ARRAY') {
964 $self->_do_connection_actions($method_prefix, $_) for @$call;
966 $self->_do_connection_actions($method_prefix, @$_) for @$call;
969 $self->throw_exception (sprintf ("Don't know how to process conection actions of type '%s'", ref($call)) );
975 sub connect_call_do_sql {
977 $self->_do_query(@_);
980 sub disconnect_call_do_sql {
982 $self->_do_query(@_);
985 # override in db-specific backend when necessary
986 sub connect_call_datetime_setup { 1 }
989 my ($self, $action) = @_;
991 if (ref $action eq 'CODE') {
992 $action = $action->($self);
993 $self->_do_query($_) foreach @$action;
996 # Most debuggers expect ($sql, @bind), so we need to exclude
997 # the attribute hash which is the second argument to $dbh->do
998 # furthermore the bind values are usually to be presented
999 # as named arrayref pairs, so wrap those here too
1000 my @do_args = (ref $action eq 'ARRAY') ? (@$action) : ($action);
1001 my $sql = shift @do_args;
1002 my $attrs = shift @do_args;
1003 my @bind = map { [ undef, $_ ] } @do_args;
1005 $self->_query_start($sql, @bind);
1006 $self->_get_dbh->do($sql, $attrs, @do_args);
1007 $self->_query_end($sql, @bind);
1014 my ($self, @info) = @_;
1016 $self->throw_exception("You failed to provide any connection info")
1019 my ($old_connect_via, $dbh);
1021 if ($INC{'Apache/DBI.pm'} && $ENV{MOD_PERL}) {
1022 $old_connect_via = $DBI::connect_via;
1023 $DBI::connect_via = 'connect';
1027 if(ref $info[0] eq 'CODE') {
1031 $dbh = DBI->connect(@info);
1034 if($dbh && !$self->unsafe) {
1035 my $weak_self = $self;
1036 Scalar::Util::weaken($weak_self);
1037 $dbh->{HandleError} = sub {
1039 $weak_self->throw_exception("DBI Exception: $_[0]");
1042 # the handler may be invoked by something totally out of
1044 croak ("DBI Exception: $_[0]");
1047 $dbh->{ShowErrorStatement} = 1;
1048 $dbh->{RaiseError} = 1;
1049 $dbh->{PrintError} = 0;
1053 $DBI::connect_via = $old_connect_via if $old_connect_via;
1055 $self->throw_exception("DBI Connection failed: " . ($@||$DBI::errstr))
1058 $self->_dbh_autocommit($dbh->{AutoCommit});
1064 my ($self, $name) = @_;
1066 $name = $self->_svp_generate_name
1067 unless defined $name;
1069 $self->throw_exception ("You can't use savepoints outside a transaction")
1070 if $self->{transaction_depth} == 0;
1072 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1073 unless $self->can('_svp_begin');
1075 push @{ $self->{savepoints} }, $name;
1077 $self->debugobj->svp_begin($name) if $self->debug;
1079 return $self->_svp_begin($name);
1083 my ($self, $name) = @_;
1085 $self->throw_exception ("You can't use savepoints outside a transaction")
1086 if $self->{transaction_depth} == 0;
1088 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1089 unless $self->can('_svp_release');
1091 if (defined $name) {
1092 $self->throw_exception ("Savepoint '$name' does not exist")
1093 unless grep { $_ eq $name } @{ $self->{savepoints} };
1095 # Dig through the stack until we find the one we are releasing. This keeps
1096 # the stack up to date.
1099 do { $svp = pop @{ $self->{savepoints} } } while $svp ne $name;
1101 $name = pop @{ $self->{savepoints} };
1104 $self->debugobj->svp_release($name) if $self->debug;
1106 return $self->_svp_release($name);
1110 my ($self, $name) = @_;
1112 $self->throw_exception ("You can't use savepoints outside a transaction")
1113 if $self->{transaction_depth} == 0;
1115 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1116 unless $self->can('_svp_rollback');
1118 if (defined $name) {
1119 # If they passed us a name, verify that it exists in the stack
1120 unless(grep({ $_ eq $name } @{ $self->{savepoints} })) {
1121 $self->throw_exception("Savepoint '$name' does not exist!");
1124 # Dig through the stack until we find the one we are releasing. This keeps
1125 # the stack up to date.
1126 while(my $s = pop(@{ $self->{savepoints} })) {
1127 last if($s eq $name);
1129 # Add the savepoint back to the stack, as a rollback doesn't remove the
1130 # named savepoint, only everything after it.
1131 push(@{ $self->{savepoints} }, $name);
1133 # We'll assume they want to rollback to the last savepoint
1134 $name = $self->{savepoints}->[-1];
1137 $self->debugobj->svp_rollback($name) if $self->debug;
1139 return $self->_svp_rollback($name);
1142 sub _svp_generate_name {
1145 return 'savepoint_'.scalar(@{ $self->{'savepoints'} });
1150 if($self->{transaction_depth} == 0) {
1151 $self->debugobj->txn_begin()
1153 $self->_dbh_begin_work;
1155 elsif ($self->auto_savepoint) {
1158 $self->{transaction_depth}++;
1161 sub _dbh_begin_work {
1164 # if the user is utilizing txn_do - good for him, otherwise we need to
1165 # ensure that the $dbh is healthy on BEGIN.
1166 # We do this via ->dbh_do instead of ->dbh, so that the ->dbh "ping"
1167 # will be replaced by a failure of begin_work itself (which will be
1168 # then retried on reconnect)
1169 if ($self->{_in_dbh_do}) {
1170 $self->_dbh->begin_work;
1172 $self->dbh_do(sub { $_[1]->begin_work });
1178 if ($self->{transaction_depth} == 1) {
1179 $self->debugobj->txn_commit()
1182 $self->{transaction_depth} = 0
1183 if $self->_dbh_autocommit;
1185 elsif($self->{transaction_depth} > 1) {
1186 $self->{transaction_depth}--;
1188 if $self->auto_savepoint;
1194 my $dbh = $self->_dbh
1195 or $self->throw_exception('cannot COMMIT on a disconnected handle');
1201 my $dbh = $self->_dbh;
1203 if ($self->{transaction_depth} == 1) {
1204 $self->debugobj->txn_rollback()
1206 $self->{transaction_depth} = 0
1207 if $self->_dbh_autocommit;
1208 $self->_dbh_rollback;
1210 elsif($self->{transaction_depth} > 1) {
1211 $self->{transaction_depth}--;
1212 if ($self->auto_savepoint) {
1213 $self->svp_rollback;
1218 die DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION->new;
1223 my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
1224 $error =~ /$exception_class/ and $self->throw_exception($error);
1225 # ensure that a failed rollback resets the transaction depth
1226 $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
1227 $self->throw_exception($error);
1233 my $dbh = $self->_dbh
1234 or $self->throw_exception('cannot ROLLBACK on a disconnected handle');
1238 # This used to be the top-half of _execute. It was split out to make it
1239 # easier to override in NoBindVars without duping the rest. It takes up
1240 # all of _execute's args, and emits $sql, @bind.
1241 sub _prep_for_execute {
1242 my ($self, $op, $extra_bind, $ident, $args) = @_;
1244 if( Scalar::Util::blessed($ident) && $ident->isa("DBIx::Class::ResultSource") ) {
1245 $ident = $ident->from();
1248 my ($sql, @bind) = $self->sql_maker->$op($ident, @$args);
1251 map { ref $_ eq 'ARRAY' ? $_ : [ '!!dummy', $_ ] } @$extra_bind)
1253 return ($sql, \@bind);
1257 sub _fix_bind_params {
1258 my ($self, @bind) = @_;
1260 ### Turn @bind from something like this:
1261 ### ( [ "artist", 1 ], [ "cdid", 1, 3 ] )
1263 ### ( "'1'", "'1'", "'3'" )
1266 if ( defined( $_ && $_->[1] ) ) {
1267 map { qq{'$_'}; } @{$_}[ 1 .. $#$_ ];
1274 my ( $self, $sql, @bind ) = @_;
1276 if ( $self->debug ) {
1277 @bind = $self->_fix_bind_params(@bind);
1279 $self->debugobj->query_start( $sql, @bind );
1284 my ( $self, $sql, @bind ) = @_;
1286 if ( $self->debug ) {
1287 @bind = $self->_fix_bind_params(@bind);
1288 $self->debugobj->query_end( $sql, @bind );
1293 my ($self, $dbh, $op, $extra_bind, $ident, $bind_attributes, @args) = @_;
1295 my ($sql, $bind) = $self->_prep_for_execute($op, $extra_bind, $ident, \@args);
1297 $self->_query_start( $sql, @$bind );
1299 my $sth = $self->sth($sql,$op);
1301 my $placeholder_index = 1;
1303 foreach my $bound (@$bind) {
1304 my $attributes = {};
1305 my($column_name, @data) = @$bound;
1307 if ($bind_attributes) {
1308 $attributes = $bind_attributes->{$column_name}
1309 if defined $bind_attributes->{$column_name};
1312 foreach my $data (@data) {
1313 my $ref = ref $data;
1314 $data = $ref && $ref ne 'ARRAY' ? ''.$data : $data; # stringify args (except arrayrefs)
1316 $sth->bind_param($placeholder_index, $data, $attributes);
1317 $placeholder_index++;
1321 # Can this fail without throwing an exception anyways???
1322 my $rv = $sth->execute();
1323 $self->throw_exception($sth->errstr) if !$rv;
1325 $self->_query_end( $sql, @$bind );
1327 return (wantarray ? ($rv, $sth, @$bind) : $rv);
1332 $self->dbh_do('_dbh_execute', @_); # retry over disconnects
1336 my ($self, $source, $to_insert) = @_;
1338 my $ident = $source->from;
1339 my $bind_attributes = $self->source_bind_attributes($source);
1341 my $updated_cols = {};
1343 foreach my $col ( $source->columns ) {
1344 if ( !defined $to_insert->{$col} ) {
1345 my $col_info = $source->column_info($col);
1347 if ( $col_info->{auto_nextval} ) {
1348 $updated_cols->{$col} = $to_insert->{$col} = $self->_sequence_fetch(
1350 $col_info->{sequence} ||
1351 $self->_dbh_get_autoinc_seq($self->_get_dbh, $source)
1357 $self->_execute('insert' => [], $source, $bind_attributes, $to_insert);
1359 return $updated_cols;
1362 ## Still not quite perfect, and EXPERIMENTAL
1363 ## Currently it is assumed that all values passed will be "normal", i.e. not
1364 ## scalar refs, or at least, all the same type as the first set, the statement is
1365 ## only prepped once.
1367 my ($self, $source, $cols, $data) = @_;
1370 @colvalues{@$cols} = (0..$#$cols);
1372 for my $i (0..$#$cols) {
1373 my $first_val = $data->[0][$i];
1374 next unless ref $first_val eq 'SCALAR';
1376 $colvalues{ $cols->[$i] } = $first_val;
1379 # check for bad data and stringify stringifiable objects
1380 my $bad_slice = sub {
1381 my ($msg, $col_idx, $slice_idx) = @_;
1382 $self->throw_exception(sprintf "%s for column '%s' in populate slice:\n%s",
1386 local $Data::Dumper::Maxdepth = 1; # don't dump objects, if any
1387 Data::Dumper::Concise::Dumper({
1388 map { $cols->[$_] => $data->[$slice_idx][$_] } (0 .. $#$cols)
1394 for my $datum_idx (0..$#$data) {
1395 my $datum = $data->[$datum_idx];
1397 for my $col_idx (0..$#$cols) {
1398 my $val = $datum->[$col_idx];
1399 my $sqla_bind = $colvalues{ $cols->[$col_idx] };
1400 my $is_literal_sql = (ref $sqla_bind) eq 'SCALAR';
1402 if ($is_literal_sql) {
1404 $bad_slice->('bind found where literal SQL expected', $col_idx, $datum_idx);
1406 elsif ((my $reftype = ref $val) ne 'SCALAR') {
1407 $bad_slice->("$reftype reference found where literal SQL expected",
1408 $col_idx, $datum_idx);
1410 elsif ($$val ne $$sqla_bind){
1411 $bad_slice->("inconsistent literal SQL value, expecting: '$$sqla_bind'",
1412 $col_idx, $datum_idx);
1415 elsif (my $reftype = ref $val) {
1417 if (overload::Method($val, '""')) {
1418 $datum->[$col_idx] = "".$val;
1421 $bad_slice->("$reftype reference found where bind expected",
1422 $col_idx, $datum_idx);
1428 my ($sql, $bind) = $self->_prep_for_execute (
1429 'insert', undef, $source, [\%colvalues]
1433 my $empty_bind = 1 if (not @bind) &&
1434 (grep { ref $_ eq 'SCALAR' } values %colvalues) == @$cols;
1436 if ((not @bind) && (not $empty_bind)) {
1437 $self->throw_exception(
1438 'Cannot insert_bulk without support for placeholders'
1442 $self->_query_start( $sql, ['__BULK__'] );
1443 my $sth = $self->sth($sql);
1447 # bind_param_array doesn't work if there are no binds
1448 $self->_dbh_execute_inserts_with_no_binds( $sth, scalar @$data );
1451 # @bind = map { ref $_ ? ''.$_ : $_ } @bind; # stringify args
1452 $self->_execute_array( $source, $sth, \@bind, $cols, $data );
1456 $self->_query_end( $sql, ['__BULK__'] );
1458 return (wantarray ? ($rv, $sth, @bind) : $rv);
1461 sub _execute_array {
1462 my ($self, $source, $sth, $bind, $cols, $data, @extra) = @_;
1464 my $guard = $self->txn_scope_guard unless $self->{transaction_depth} != 0;
1466 ## This must be an arrayref, else nothing works!
1467 my $tuple_status = [];
1469 ## Get the bind_attributes, if any exist
1470 my $bind_attributes = $self->source_bind_attributes($source);
1472 ## Bind the values and execute
1473 my $placeholder_index = 1;
1475 foreach my $bound (@$bind) {
1477 my $attributes = {};
1478 my ($column_name, $data_index) = @$bound;
1480 if( $bind_attributes ) {
1481 $attributes = $bind_attributes->{$column_name}
1482 if defined $bind_attributes->{$column_name};
1485 my @data = map { $_->[$data_index] } @$data;
1487 $sth->bind_param_array( $placeholder_index, [@data], $attributes );
1488 $placeholder_index++;
1492 $self->_dbh_execute_array($sth, $tuple_status, @extra);
1494 my $err = $@ || $sth->errstr;
1496 # Statement must finish even if there was an exception.
1497 eval { $sth->finish };
1498 $err = $@ unless $err;
1502 ++$i while $i <= $#$tuple_status && !ref $tuple_status->[$i];
1504 $self->throw_exception("Unexpected populate error: $err")
1505 if ($i > $#$tuple_status);
1507 $self->throw_exception(sprintf "%s for populate slice:\n%s",
1508 ($tuple_status->[$i][1] || $err),
1509 Data::Dumper::Concise::Dumper({
1510 map { $cols->[$_] => $data->[$i][$_] } (0 .. $#$cols)
1515 $guard->commit if $guard;
1520 sub _dbh_execute_array {
1521 my ($self, $sth, $tuple_status, @extra) = @_;
1523 return $sth->execute_array({ArrayTupleStatus => $tuple_status});
1526 sub _dbh_execute_inserts_with_no_binds {
1527 my ($self, $sth, $count) = @_;
1529 my $guard = $self->txn_scope_guard unless $self->{transaction_depth} != 0;
1532 my $dbh = $self->_get_dbh;
1533 local $dbh->{RaiseError} = 1;
1534 local $dbh->{PrintError} = 0;
1536 $sth->execute foreach 1..$count;
1540 # Make sure statement is finished even if there was an exception.
1541 eval { $sth->finish };
1542 $exception = $@ unless $exception;
1544 $self->throw_exception($exception) if $exception;
1546 $guard->commit if $guard;
1552 my ($self, $source, @args) = @_;
1554 my $bind_attributes = $self->source_bind_attributes($source);
1556 return $self->_execute('update' => [], $source, $bind_attributes, @args);
1561 my ($self, $source, @args) = @_;
1563 my $bind_attrs = $self->source_bind_attributes($source);
1565 return $self->_execute('delete' => [], $source, $bind_attrs, @args);
1568 # We were sent here because the $rs contains a complex search
1569 # which will require a subquery to select the correct rows
1570 # (i.e. joined or limited resultsets)
1572 # Genarating a single PK column subquery is trivial and supported
1573 # by all RDBMS. However if we have a multicolumn PK, things get ugly.
1574 # Look at _multipk_update_delete()
1575 sub _subq_update_delete {
1577 my ($rs, $op, $values) = @_;
1579 my $rsrc = $rs->result_source;
1581 # we already check this, but double check naively just in case. Should be removed soon
1582 my $sel = $rs->_resolved_attrs->{select};
1583 $sel = [ $sel ] unless ref $sel eq 'ARRAY';
1584 my @pcols = $rsrc->primary_columns;
1585 if (@$sel != @pcols) {
1586 $self->throw_exception (
1587 'Subquery update/delete can not be called on resultsets selecting a'
1588 .' number of columns different than the number of primary keys'
1595 $op eq 'update' ? $values : (),
1596 { $pcols[0] => { -in => $rs->as_query } },
1601 return $self->_multipk_update_delete (@_);
1605 # ANSI SQL does not provide a reliable way to perform a multicol-PK
1606 # resultset update/delete involving subqueries. So by default resort
1607 # to simple (and inefficient) delete_all style per-row opearations,
1608 # while allowing specific storages to override this with a faster
1611 sub _multipk_update_delete {
1612 return shift->_per_row_update_delete (@_);
1615 # This is the default loop used to delete/update rows for multi PK
1616 # resultsets, and used by mysql exclusively (because it can't do anything
1619 # We do not use $row->$op style queries, because resultset update/delete
1620 # is not expected to cascade (this is what delete_all/update_all is for).
1622 # There should be no race conditions as the entire operation is rolled
1625 sub _per_row_update_delete {
1627 my ($rs, $op, $values) = @_;
1629 my $rsrc = $rs->result_source;
1630 my @pcols = $rsrc->primary_columns;
1632 my $guard = $self->txn_scope_guard;
1634 # emulate the return value of $sth->execute for non-selects
1635 my $row_cnt = '0E0';
1637 my $subrs_cur = $rs->cursor;
1638 while (my @pks = $subrs_cur->next) {
1641 for my $i (0.. $#pcols) {
1642 $cond->{$pcols[$i]} = $pks[$i];
1647 $op eq 'update' ? $values : (),
1662 # localization is neccessary as
1663 # 1) there is no infrastructure to pass this around before SQLA2
1664 # 2) _select_args sets it and _prep_for_execute consumes it
1665 my $sql_maker = $self->sql_maker;
1666 local $sql_maker->{_dbic_rs_attrs};
1668 return $self->_execute($self->_select_args(@_));
1671 sub _select_args_to_query {
1674 # localization is neccessary as
1675 # 1) there is no infrastructure to pass this around before SQLA2
1676 # 2) _select_args sets it and _prep_for_execute consumes it
1677 my $sql_maker = $self->sql_maker;
1678 local $sql_maker->{_dbic_rs_attrs};
1680 # my ($op, $bind, $ident, $bind_attrs, $select, $cond, $order, $rows, $offset)
1681 # = $self->_select_args($ident, $select, $cond, $attrs);
1682 my ($op, $bind, $ident, $bind_attrs, @args) =
1683 $self->_select_args(@_);
1685 # my ($sql, $prepared_bind) = $self->_prep_for_execute($op, $bind, $ident, [ $select, $cond, $order, $rows, $offset ]);
1686 my ($sql, $prepared_bind) = $self->_prep_for_execute($op, $bind, $ident, \@args);
1687 $prepared_bind ||= [];
1690 ? ($sql, $prepared_bind, $bind_attrs)
1691 : \[ "($sql)", @$prepared_bind ]
1696 my ($self, $ident, $select, $where, $attrs) = @_;
1698 my ($alias2source, $rs_alias) = $self->_resolve_ident_sources ($ident);
1700 my $sql_maker = $self->sql_maker;
1701 $sql_maker->{_dbic_rs_attrs} = {
1707 ? ( _source_handle => $alias2source->{$rs_alias}->handle )
1712 # calculate bind_attrs before possible $ident mangling
1713 my $bind_attrs = {};
1714 for my $alias (keys %$alias2source) {
1715 my $bindtypes = $self->source_bind_attributes ($alias2source->{$alias}) || {};
1716 for my $col (keys %$bindtypes) {
1718 my $fqcn = join ('.', $alias, $col);
1719 $bind_attrs->{$fqcn} = $bindtypes->{$col} if $bindtypes->{$col};
1721 # Unqialified column names are nice, but at the same time can be
1722 # rather ambiguous. What we do here is basically go along with
1723 # the loop, adding an unqualified column slot to $bind_attrs,
1724 # alongside the fully qualified name. As soon as we encounter
1725 # another column by that name (which would imply another table)
1726 # we unset the unqualified slot and never add any info to it
1727 # to avoid erroneous type binding. If this happens the users
1728 # only choice will be to fully qualify his column name
1730 if (exists $bind_attrs->{$col}) {
1731 $bind_attrs->{$col} = {};
1734 $bind_attrs->{$col} = $bind_attrs->{$fqcn};
1741 $attrs->{software_limit}
1743 $sql_maker->_default_limit_syntax eq "GenericSubQ"
1745 $attrs->{software_limit} = 1;
1748 $self->throw_exception("rows attribute must be positive if present")
1749 if (defined($attrs->{rows}) && !($attrs->{rows} > 0));
1751 # MySQL actually recommends this approach. I cringe.
1752 $attrs->{rows} = 2**48 if not defined $attrs->{rows} and defined $attrs->{offset};
1757 # see if we need to tear the prefetch apart (either limited has_many or grouped prefetch)
1758 # otherwise delegate the limiting to the storage, unless software limit was requested
1760 ( $attrs->{rows} && keys %{$attrs->{collapse}} )
1762 ( $attrs->{group_by} && @{$attrs->{group_by}} &&
1763 $attrs->{_prefetch_select} && @{$attrs->{_prefetch_select}} )
1765 ($ident, $select, $where, $attrs)
1766 = $self->_adjust_select_args_for_complex_prefetch ($ident, $select, $where, $attrs);
1768 elsif (! $attrs->{software_limit} ) {
1769 push @limit, $attrs->{rows}, $attrs->{offset};
1773 # This would be the point to deflate anything found in $where
1774 # (and leave $attrs->{bind} intact). Problem is - inflators historically
1775 # expect a row object. And all we have is a resultsource (it is trivial
1776 # to extract deflator coderefs via $alias2source above).
1778 # I don't see a way forward other than changing the way deflators are
1779 # invoked, and that's just bad...
1783 { $attrs->{$_} ? ( $_ => $attrs->{$_} ) : () }
1784 (qw/order_by group_by having/ )
1787 return ('select', $attrs->{bind}, $ident, $bind_attrs, $select, $where, $order, @limit);
1791 # This is the code producing joined subqueries like:
1792 # SELECT me.*, other.* FROM ( SELECT me.* FROM ... ) JOIN other ON ...
1794 sub _adjust_select_args_for_complex_prefetch {
1795 my ($self, $from, $select, $where, $attrs) = @_;
1797 $self->throw_exception ('Nothing to prefetch... how did we get here?!')
1798 if not @{$attrs->{_prefetch_select}};
1800 $self->throw_exception ('Complex prefetches are not supported on resultsets with a custom from attribute')
1801 if (ref $from ne 'ARRAY' || ref $from->[0] ne 'HASH' || ref $from->[1] ne 'ARRAY');
1804 # generate inner/outer attribute lists, remove stuff that doesn't apply
1805 my $outer_attrs = { %$attrs };
1806 delete $outer_attrs->{$_} for qw/where bind rows offset group_by having/;
1808 my $inner_attrs = { %$attrs };
1809 delete $inner_attrs->{$_} for qw/for collapse _prefetch_select _collapse_order_by select as/;
1812 # bring over all non-collapse-induced order_by into the inner query (if any)
1813 # the outer one will have to keep them all
1814 delete $inner_attrs->{order_by};
1815 if (my $ord_cnt = @{$outer_attrs->{order_by}} - @{$outer_attrs->{_collapse_order_by}} ) {
1816 $inner_attrs->{order_by} = [
1817 @{$outer_attrs->{order_by}}[ 0 .. $ord_cnt - 1]
1822 # generate the inner/outer select lists
1823 # for inside we consider only stuff *not* brought in by the prefetch
1824 # on the outside we substitute any function for its alias
1825 my $outer_select = [ @$select ];
1826 my $inner_select = [];
1827 for my $i (0 .. ( @$outer_select - @{$outer_attrs->{_prefetch_select}} - 1) ) {
1828 my $sel = $outer_select->[$i];
1830 if (ref $sel eq 'HASH' ) {
1831 $sel->{-as} ||= $attrs->{as}[$i];
1832 $outer_select->[$i] = join ('.', $attrs->{alias}, ($sel->{-as} || "inner_column_$i") );
1835 push @$inner_select, $sel;
1838 # normalize a copy of $from, so it will be easier to work with further
1839 # down (i.e. promote the initial hashref to an AoH)
1841 $from->[0] = [ $from->[0] ];
1842 my %original_join_info = map { $_->[0]{-alias} => $_->[0] } (@$from);
1845 # decide which parts of the join will remain in either part of
1846 # the outer/inner query
1848 # First we compose a list of which aliases are used in restrictions
1849 # (i.e. conditions/order/grouping/etc). Since we do not have
1850 # introspectable SQLA, we fall back to ugly scanning of raw SQL for
1851 # WHERE, and for pieces of ORDER BY in order to determine which aliases
1852 # need to appear in the resulting sql.
1853 # It may not be very efficient, but it's a reasonable stop-gap
1854 # Also unqualified column names will not be considered, but more often
1855 # than not this is actually ok
1857 # In the same loop we enumerate part of the selection aliases, as
1858 # it requires the same sqla hack for the time being
1859 my ($restrict_aliases, $select_aliases, $prefetch_aliases);
1861 # produce stuff unquoted, so it can be scanned
1862 my $sql_maker = $self->sql_maker;
1863 local $sql_maker->{quote_char};
1864 my $sep = $self->_sql_maker_opts->{name_sep} || '.';
1867 my $non_prefetch_select_sql = $sql_maker->_recurse_fields ($inner_select);
1868 my $prefetch_select_sql = $sql_maker->_recurse_fields ($outer_attrs->{_prefetch_select});
1869 my $where_sql = $sql_maker->where ($where);
1870 my $group_by_sql = $sql_maker->_order_by({
1871 map { $_ => $inner_attrs->{$_} } qw/group_by having/
1873 my @non_prefetch_order_by_chunks = (map
1874 { ref $_ ? $_->[0] : $_ }
1875 $sql_maker->_order_by_chunks ($inner_attrs->{order_by})
1879 for my $alias (keys %original_join_info) {
1880 my $seen_re = qr/\b $alias $sep/x;
1882 for my $piece ($where_sql, $group_by_sql, @non_prefetch_order_by_chunks ) {
1883 if ($piece =~ $seen_re) {
1884 $restrict_aliases->{$alias} = 1;
1888 if ($non_prefetch_select_sql =~ $seen_re) {
1889 $select_aliases->{$alias} = 1;
1892 if ($prefetch_select_sql =~ $seen_re) {
1893 $prefetch_aliases->{$alias} = 1;
1899 # Add any non-left joins to the restriction list (such joins are indeed restrictions)
1900 for my $j (values %original_join_info) {
1901 my $alias = $j->{-alias} or next;
1902 $restrict_aliases->{$alias} = 1 if (
1903 (not $j->{-join_type})
1905 ($j->{-join_type} !~ /^left (?: \s+ outer)? $/xi)
1909 # mark all join parents as mentioned
1910 # (e.g. join => { cds => 'tracks' } - tracks will need to bring cds too )
1911 for my $collection ($restrict_aliases, $select_aliases) {
1912 for my $alias (keys %$collection) {
1913 $collection->{$_} = 1
1914 for (@{ $original_join_info{$alias}{-join_path} || [] });
1918 # construct the inner $from for the subquery
1919 my %inner_joins = (map { %{$_ || {}} } ($restrict_aliases, $select_aliases) );
1921 for my $j (@$from) {
1922 push @inner_from, $j if $inner_joins{$j->[0]{-alias}};
1925 # if a multi-type join was needed in the subquery ("multi" is indicated by
1926 # presence in {collapse}) - add a group_by to simulate the collapse in the subq
1927 unless ($inner_attrs->{group_by}) {
1928 for my $alias (keys %inner_joins) {
1930 # the dot comes from some weirdness in collapse
1931 # remove after the rewrite
1932 if ($attrs->{collapse}{".$alias"}) {
1933 $inner_attrs->{group_by} ||= $inner_select;
1939 # demote the inner_from head
1940 $inner_from[0] = $inner_from[0][0];
1942 # generate the subquery
1943 my $subq = $self->_select_args_to_query (
1950 my $subq_joinspec = {
1951 -alias => $attrs->{alias},
1952 -source_handle => $inner_from[0]{-source_handle},
1953 $attrs->{alias} => $subq,
1956 # Generate the outer from - this is relatively easy (really just replace
1957 # the join slot with the subquery), with a major caveat - we can not
1958 # join anything that is non-selecting (not part of the prefetch), but at
1959 # the same time is a multi-type relationship, as it will explode the result.
1961 # There are two possibilities here
1962 # - either the join is non-restricting, in which case we simply throw it away
1963 # - it is part of the restrictions, in which case we need to collapse the outer
1964 # result by tackling yet another group_by to the outside of the query
1966 # so first generate the outer_from, up to the substitution point
1968 while (my $j = shift @$from) {
1969 if ($j->[0]{-alias} eq $attrs->{alias}) { # time to swap
1974 last; # we'll take care of what's left in $from below
1977 push @outer_from, $j;
1981 # see what's left - throw away if not selecting/restricting
1982 # also throw in a group_by if restricting to guard against
1983 # cross-join explosions
1985 while (my $j = shift @$from) {
1986 my $alias = $j->[0]{-alias};
1988 if ($select_aliases->{$alias} || $prefetch_aliases->{$alias}) {
1989 push @outer_from, $j;
1991 elsif ($restrict_aliases->{$alias}) {
1992 push @outer_from, $j;
1994 # FIXME - this should be obviated by SQLA2, as I'll be able to
1995 # have restrict_inner and restrict_outer... or something to that
1996 # effect... I think...
1998 # FIXME2 - I can't find a clean way to determine if a particular join
1999 # is a multi - instead I am just treating everything as a potential
2000 # explosive join (ribasushi)
2002 # if (my $handle = $j->[0]{-source_handle}) {
2003 # my $rsrc = $handle->resolve;
2004 # ... need to bail out of the following if this is not a multi,
2005 # as it will be much easier on the db ...
2007 $outer_attrs->{group_by} ||= $outer_select;
2012 # demote the outer_from head
2013 $outer_from[0] = $outer_from[0][0];
2015 # This is totally horrific - the $where ends up in both the inner and outer query
2016 # Unfortunately not much can be done until SQLA2 introspection arrives, and even
2017 # then if where conditions apply to the *right* side of the prefetch, you may have
2018 # to both filter the inner select (e.g. to apply a limit) and then have to re-filter
2019 # the outer select to exclude joins you didin't want in the first place
2021 # OTOH it can be seen as a plus: <ash> (notes that this query would make a DBA cry ;)
2022 return (\@outer_from, $outer_select, $where, $outer_attrs);
2025 sub _resolve_ident_sources {
2026 my ($self, $ident) = @_;
2028 my $alias2source = {};
2031 # the reason this is so contrived is that $ident may be a {from}
2032 # structure, specifying multiple tables to join
2033 if ( Scalar::Util::blessed($ident) && $ident->isa("DBIx::Class::ResultSource") ) {
2034 # this is compat mode for insert/update/delete which do not deal with aliases
2035 $alias2source->{me} = $ident;
2038 elsif (ref $ident eq 'ARRAY') {
2042 if (ref $_ eq 'HASH') {
2044 $rs_alias = $tabinfo->{-alias};
2046 if (ref $_ eq 'ARRAY' and ref $_->[0] eq 'HASH') {
2050 $alias2source->{$tabinfo->{-alias}} = $tabinfo->{-source_handle}->resolve
2051 if ($tabinfo->{-source_handle});
2055 return ($alias2source, $rs_alias);
2058 # Takes $ident, \@column_names
2060 # returns { $column_name => \%column_info, ... }
2061 # also note: this adds -result_source => $rsrc to the column info
2064 # my $col_sources = $self->_resolve_column_info($ident, @column_names);
2065 sub _resolve_column_info {
2066 my ($self, $ident, $colnames) = @_;
2067 my ($alias2src, $root_alias) = $self->_resolve_ident_sources($ident);
2069 my $sep = $self->_sql_maker_opts->{name_sep} || '.';
2072 my (%return, %seen_cols);
2074 # compile a global list of column names, to be able to properly
2075 # disambiguate unqualified column names (if at all possible)
2076 for my $alias (keys %$alias2src) {
2077 my $rsrc = $alias2src->{$alias};
2078 for my $colname ($rsrc->columns) {
2079 push @{$seen_cols{$colname}}, $alias;
2084 foreach my $col (@$colnames) {
2085 my ($alias, $colname) = $col =~ m/^ (?: ([^$sep]+) $sep)? (.+) $/x;
2088 # see if the column was seen exactly once (so we know which rsrc it came from)
2089 if ($seen_cols{$colname} and @{$seen_cols{$colname}} == 1) {
2090 $alias = $seen_cols{$colname}[0];
2097 my $rsrc = $alias2src->{$alias};
2098 $return{$col} = $rsrc && {
2099 %{$rsrc->column_info($colname)},
2100 -result_source => $rsrc,
2101 -source_alias => $alias,
2108 # Returns a counting SELECT for a simple count
2109 # query. Abstracted so that a storage could override
2110 # this to { count => 'firstcol' } or whatever makes
2111 # sense as a performance optimization
2113 #my ($self, $source, $rs_attrs) = @_;
2114 return { count => '*' };
2117 # Returns a SELECT which will end up in the subselect
2118 # There may or may not be a group_by, as the subquery
2119 # might have been called to accomodate a limit
2121 # Most databases would be happy with whatever ends up
2122 # here, but some choke in various ways.
2124 sub _subq_count_select {
2125 my ($self, $source, $rs_attrs) = @_;
2126 return $rs_attrs->{group_by} if $rs_attrs->{group_by};
2128 my @pcols = map { join '.', $rs_attrs->{alias}, $_ } ($source->primary_columns);
2129 return @pcols ? \@pcols : [ 1 ];
2132 sub source_bind_attributes {
2133 my ($self, $source) = @_;
2135 my $bind_attributes;
2136 foreach my $column ($source->columns) {
2138 my $data_type = $source->column_info($column)->{data_type} || '';
2139 $bind_attributes->{$column} = $self->bind_attribute_by_data_type($data_type)
2143 return $bind_attributes;
2150 =item Arguments: $ident, $select, $condition, $attrs
2154 Handle a SQL select statement.
2160 my ($ident, $select, $condition, $attrs) = @_;
2161 return $self->cursor_class->new($self, \@_, $attrs);
2166 my ($rv, $sth, @bind) = $self->_select(@_);
2167 my @row = $sth->fetchrow_array;
2168 my @nextrow = $sth->fetchrow_array if @row;
2169 if(@row && @nextrow) {
2170 carp "Query returned more than one row. SQL that returns multiple rows is DEPRECATED for ->find and ->single";
2172 # Need to call finish() to work round broken DBDs
2181 =item Arguments: $sql
2185 Returns a L<DBI> sth (statement handle) for the supplied SQL.
2190 my ($self, $dbh, $sql) = @_;
2192 # 3 is the if_active parameter which avoids active sth re-use
2193 my $sth = $self->disable_sth_caching
2194 ? $dbh->prepare($sql)
2195 : $dbh->prepare_cached($sql, {}, 3);
2197 # XXX You would think RaiseError would make this impossible,
2198 # but apparently that's not true :(
2199 $self->throw_exception($dbh->errstr) if !$sth;
2205 my ($self, $sql) = @_;
2206 $self->dbh_do('_dbh_sth', $sql); # retry over disconnects
2209 sub _dbh_columns_info_for {
2210 my ($self, $dbh, $table) = @_;
2212 if ($dbh->can('column_info')) {
2215 my ($schema,$tab) = $table =~ /^(.+?)\.(.+)$/ ? ($1,$2) : (undef,$table);
2216 my $sth = $dbh->column_info( undef,$schema, $tab, '%' );
2218 while ( my $info = $sth->fetchrow_hashref() ){
2220 $column_info{data_type} = $info->{TYPE_NAME};
2221 $column_info{size} = $info->{COLUMN_SIZE};
2222 $column_info{is_nullable} = $info->{NULLABLE} ? 1 : 0;
2223 $column_info{default_value} = $info->{COLUMN_DEF};
2224 my $col_name = $info->{COLUMN_NAME};
2225 $col_name =~ s/^\"(.*)\"$/$1/;
2227 $result{$col_name} = \%column_info;
2230 return \%result if !$@ && scalar keys %result;
2234 my $sth = $dbh->prepare($self->sql_maker->select($table, undef, \'1 = 0'));
2236 my @columns = @{$sth->{NAME_lc}};
2237 for my $i ( 0 .. $#columns ){
2239 $column_info{data_type} = $sth->{TYPE}->[$i];
2240 $column_info{size} = $sth->{PRECISION}->[$i];
2241 $column_info{is_nullable} = $sth->{NULLABLE}->[$i] ? 1 : 0;
2243 if ($column_info{data_type} =~ m/^(.*?)\((.*?)\)$/) {
2244 $column_info{data_type} = $1;
2245 $column_info{size} = $2;
2248 $result{$columns[$i]} = \%column_info;
2252 foreach my $col (keys %result) {
2253 my $colinfo = $result{$col};
2254 my $type_num = $colinfo->{data_type};
2256 if(defined $type_num && $dbh->can('type_info')) {
2257 my $type_info = $dbh->type_info($type_num);
2258 $type_name = $type_info->{TYPE_NAME} if $type_info;
2259 $colinfo->{data_type} = $type_name if $type_name;
2266 sub columns_info_for {
2267 my ($self, $table) = @_;
2268 $self->_dbh_columns_info_for ($self->_get_dbh, $table);
2271 =head2 last_insert_id
2273 Return the row id of the last insert.
2277 sub _dbh_last_insert_id {
2278 # All Storage's need to register their own _dbh_last_insert_id
2279 # the old SQLite-based method was highly inappropriate
2282 my $class = ref $self;
2283 $self->throw_exception (<<EOE);
2285 No _dbh_last_insert_id() method found in $class.
2286 Since the method of obtaining the autoincrement id of the last insert
2287 operation varies greatly between different databases, this method must be
2288 individually implemented for every storage class.
2292 sub last_insert_id {
2294 $self->_dbh_last_insert_id ($self->_dbh, @_);
2297 =head2 _native_data_type
2301 =item Arguments: $type_name
2305 This API is B<EXPERIMENTAL>, will almost definitely change in the future, and
2306 currently only used by L<::AutoCast|DBIx::Class::Storage::DBI::AutoCast> and
2307 L<::Sybase|DBIx::Class::Storage::DBI::Sybase>.
2309 The default implementation returns C<undef>, implement in your Storage driver if
2310 you need this functionality.
2312 Should map types from other databases to the native RDBMS type, for example
2313 C<VARCHAR2> to C<VARCHAR>.
2315 Types with modifiers should map to the underlying data type. For example,
2316 C<INTEGER AUTO_INCREMENT> should become C<INTEGER>.
2318 Composite types should map to the container type, for example
2319 C<ENUM(foo,bar,baz)> becomes C<ENUM>.
2323 sub _native_data_type {
2324 #my ($self, $data_type) = @_;
2328 # Check if placeholders are supported at all
2329 sub _placeholders_supported {
2331 my $dbh = $self->_get_dbh;
2333 # some drivers provide a $dbh attribute (e.g. Sybase and $dbh->{syb_dynamic_supported})
2334 # but it is inaccurate more often than not
2336 local $dbh->{PrintError} = 0;
2337 local $dbh->{RaiseError} = 1;
2338 $dbh->do('select ?', {}, 1);
2343 # Check if placeholders bound to non-string types throw exceptions
2345 sub _typeless_placeholders_supported {
2347 my $dbh = $self->_get_dbh;
2350 local $dbh->{PrintError} = 0;
2351 local $dbh->{RaiseError} = 1;
2352 # this specifically tests a bind that is NOT a string
2353 $dbh->do('select 1 where 1 = ?', {}, 1);
2360 Returns the database driver name.
2365 shift->_get_dbh->{Driver}->{Name};
2368 =head2 bind_attribute_by_data_type
2370 Given a datatype from column info, returns a database specific bind
2371 attribute for C<< $dbh->bind_param($val,$attribute) >> or nothing if we will
2372 let the database planner just handle it.
2374 Generally only needed for special case column types, like bytea in postgres.
2378 sub bind_attribute_by_data_type {
2382 =head2 is_datatype_numeric
2384 Given a datatype from column_info, returns a boolean value indicating if
2385 the current RDBMS considers it a numeric value. This controls how
2386 L<DBIx::Class::Row/set_column> decides whether to mark the column as
2387 dirty - when the datatype is deemed numeric a C<< != >> comparison will
2388 be performed instead of the usual C<eq>.
2392 sub is_datatype_numeric {
2393 my ($self, $dt) = @_;
2395 return 0 unless $dt;
2397 return $dt =~ /^ (?:
2398 numeric | int(?:eger)? | (?:tiny|small|medium|big)int | dec(?:imal)? | real | float | double (?: \s+ precision)? | (?:big)?serial
2403 =head2 create_ddl_dir (EXPERIMENTAL)
2407 =item Arguments: $schema \@databases, $version, $directory, $preversion, \%sqlt_args
2411 Creates a SQL file based on the Schema, for each of the specified
2412 database engines in C<\@databases> in the given directory.
2413 (note: specify L<SQL::Translator> names, not L<DBI> driver names).
2415 Given a previous version number, this will also create a file containing
2416 the ALTER TABLE statements to transform the previous schema into the
2417 current one. Note that these statements may contain C<DROP TABLE> or
2418 C<DROP COLUMN> statements that can potentially destroy data.
2420 The file names are created using the C<ddl_filename> method below, please
2421 override this method in your schema if you would like a different file
2422 name format. For the ALTER file, the same format is used, replacing
2423 $version in the name with "$preversion-$version".
2425 See L<SQL::Translator/METHODS> for a list of values for C<\%sqlt_args>.
2426 The most common value for this would be C<< { add_drop_table => 1 } >>
2427 to have the SQL produced include a C<DROP TABLE> statement for each table
2428 created. For quoting purposes supply C<quote_table_names> and
2429 C<quote_field_names>.
2431 If no arguments are passed, then the following default values are assumed:
2435 =item databases - ['MySQL', 'SQLite', 'PostgreSQL']
2437 =item version - $schema->schema_version
2439 =item directory - './'
2441 =item preversion - <none>
2445 By default, C<\%sqlt_args> will have
2447 { add_drop_table => 1, ignore_constraint_names => 1, ignore_index_names => 1 }
2449 merged with the hash passed in. To disable any of those features, pass in a
2450 hashref like the following
2452 { ignore_constraint_names => 0, # ... other options }
2455 Note that this feature is currently EXPERIMENTAL and may not work correctly
2456 across all databases, or fully handle complex relationships.
2458 WARNING: Please check all SQL files created, before applying them.
2462 sub create_ddl_dir {
2463 my ($self, $schema, $databases, $version, $dir, $preversion, $sqltargs) = @_;
2465 if(!$dir || !-d $dir) {
2466 carp "No directory given, using ./\n";
2469 $databases ||= ['MySQL', 'SQLite', 'PostgreSQL'];
2470 $databases = [ $databases ] if(ref($databases) ne 'ARRAY');
2472 my $schema_version = $schema->schema_version || '1.x';
2473 $version ||= $schema_version;
2476 add_drop_table => 1,
2477 ignore_constraint_names => 1,
2478 ignore_index_names => 1,
2482 $self->throw_exception("Can't create a ddl file without SQL::Translator: " . $self->_sqlt_version_error)
2483 if !$self->_sqlt_version_ok;
2485 my $sqlt = SQL::Translator->new( $sqltargs );
2487 $sqlt->parser('SQL::Translator::Parser::DBIx::Class');
2488 my $sqlt_schema = $sqlt->translate({ data => $schema })
2489 or $self->throw_exception ($sqlt->error);
2491 foreach my $db (@$databases) {
2493 $sqlt->{schema} = $sqlt_schema;
2494 $sqlt->producer($db);
2497 my $filename = $schema->ddl_filename($db, $version, $dir);
2498 if (-e $filename && ($version eq $schema_version )) {
2499 # if we are dumping the current version, overwrite the DDL
2500 carp "Overwriting existing DDL file - $filename";
2504 my $output = $sqlt->translate;
2506 carp("Failed to translate to $db, skipping. (" . $sqlt->error . ")");
2509 if(!open($file, ">$filename")) {
2510 $self->throw_exception("Can't open $filename for writing ($!)");
2513 print $file $output;
2516 next unless ($preversion);
2518 require SQL::Translator::Diff;
2520 my $prefilename = $schema->ddl_filename($db, $preversion, $dir);
2521 if(!-e $prefilename) {
2522 carp("No previous schema file found ($prefilename)");
2526 my $difffile = $schema->ddl_filename($db, $version, $dir, $preversion);
2528 carp("Overwriting existing diff file - $difffile");
2534 my $t = SQL::Translator->new($sqltargs);
2539 or $self->throw_exception ($t->error);
2541 my $out = $t->translate( $prefilename )
2542 or $self->throw_exception ($t->error);
2544 $source_schema = $t->schema;
2546 $source_schema->name( $prefilename )
2547 unless ( $source_schema->name );
2550 # The "new" style of producers have sane normalization and can support
2551 # diffing a SQL file against a DBIC->SQLT schema. Old style ones don't
2552 # And we have to diff parsed SQL against parsed SQL.
2553 my $dest_schema = $sqlt_schema;
2555 unless ( "SQL::Translator::Producer::$db"->can('preprocess_schema') ) {
2556 my $t = SQL::Translator->new($sqltargs);
2561 or $self->throw_exception ($t->error);
2563 my $out = $t->translate( $filename )
2564 or $self->throw_exception ($t->error);
2566 $dest_schema = $t->schema;
2568 $dest_schema->name( $filename )
2569 unless $dest_schema->name;
2572 my $diff = SQL::Translator::Diff::schema_diff($source_schema, $db,
2576 if(!open $file, ">$difffile") {
2577 $self->throw_exception("Can't write to $difffile ($!)");
2585 =head2 deployment_statements
2589 =item Arguments: $schema, $type, $version, $directory, $sqlt_args
2593 Returns the statements used by L</deploy> and L<DBIx::Class::Schema/deploy>.
2595 The L<SQL::Translator> (not L<DBI>) database driver name can be explicitly
2596 provided in C<$type>, otherwise the result of L</sqlt_type> is used as default.
2598 C<$directory> is used to return statements from files in a previously created
2599 L</create_ddl_dir> directory and is optional. The filenames are constructed
2600 from L<DBIx::Class::Schema/ddl_filename>, the schema name and the C<$version>.
2602 If no C<$directory> is specified then the statements are constructed on the
2603 fly using L<SQL::Translator> and C<$version> is ignored.
2605 See L<SQL::Translator/METHODS> for a list of values for C<$sqlt_args>.
2609 sub deployment_statements {
2610 my ($self, $schema, $type, $version, $dir, $sqltargs) = @_;
2611 $type ||= $self->sqlt_type;
2612 $version ||= $schema->schema_version || '1.x';
2614 my $filename = $schema->ddl_filename($type, $version, $dir);
2618 open($file, "<$filename")
2619 or $self->throw_exception("Can't open $filename ($!)");
2622 return join('', @rows);
2625 $self->throw_exception("Can't deploy without either SQL::Translator or a ddl_dir: " . $self->_sqlt_version_error )
2626 if !$self->_sqlt_version_ok;
2628 # sources needs to be a parser arg, but for simplicty allow at top level
2630 $sqltargs->{parser_args}{sources} = delete $sqltargs->{sources}
2631 if exists $sqltargs->{sources};
2633 my $tr = SQL::Translator->new(
2634 producer => "SQL::Translator::Producer::${type}",
2636 parser => 'SQL::Translator::Parser::DBIx::Class',
2640 my $ret = $tr->translate
2641 or $self->throw_exception( 'Unable to produce deployment statements: ' . $tr->error);
2647 my ($self, $schema, $type, $sqltargs, $dir) = @_;
2650 return if($line =~ /^--/);
2652 # next if($line =~ /^DROP/m);
2653 return if($line =~ /^BEGIN TRANSACTION/m);
2654 return if($line =~ /^COMMIT/m);
2655 return if $line =~ /^\s+$/; # skip whitespace only
2656 $self->_query_start($line);
2658 # do a dbh_do cycle here, as we need some error checking in
2659 # place (even though we will ignore errors)
2660 $self->dbh_do (sub { $_[1]->do($line) });
2663 carp qq{$@ (running "${line}")};
2665 $self->_query_end($line);
2667 my @statements = $self->deployment_statements($schema, $type, undef, $dir, { %{ $sqltargs || {} }, no_comments => 1 } );
2668 if (@statements > 1) {
2669 foreach my $statement (@statements) {
2670 $deploy->( $statement );
2673 elsif (@statements == 1) {
2674 foreach my $line ( split(";\n", $statements[0])) {
2680 =head2 datetime_parser
2682 Returns the datetime parser class
2686 sub datetime_parser {
2688 return $self->{datetime_parser} ||= do {
2689 $self->build_datetime_parser(@_);
2693 =head2 datetime_parser_type
2695 Defines (returns) the datetime parser class - currently hardwired to
2696 L<DateTime::Format::MySQL>
2700 sub datetime_parser_type { "DateTime::Format::MySQL"; }
2702 =head2 build_datetime_parser
2704 See L</datetime_parser>
2708 sub build_datetime_parser {
2710 my $type = $self->datetime_parser_type(@_);
2711 $self->ensure_class_loaded ($type);
2716 =head2 is_replicating
2718 A boolean that reports if a particular L<DBIx::Class::Storage::DBI> is set to
2719 replicate from a master database. Default is undef, which is the result
2720 returned by databases that don't support replication.
2724 sub is_replicating {
2729 =head2 lag_behind_master
2731 Returns a number that represents a certain amount of lag behind a master db
2732 when a given storage is replicating. The number is database dependent, but
2733 starts at zero and increases with the amount of lag. Default in undef
2737 sub lag_behind_master {
2741 # SQLT version handling
2743 my $_sqlt_version_ok; # private
2744 my $_sqlt_version_error; # private
2746 sub _sqlt_version_ok {
2747 if (!defined $_sqlt_version_ok) {
2748 eval "use SQL::Translator $minimum_sqlt_version";
2750 $_sqlt_version_ok = 0;
2751 $_sqlt_version_error = $@;
2754 $_sqlt_version_ok = 1;
2757 return $_sqlt_version_ok;
2760 sub _sqlt_version_error {
2761 shift->_sqlt_version_ok unless defined $_sqlt_version_ok;
2762 return $_sqlt_version_error;
2765 sub _sqlt_minimum_version { $minimum_sqlt_version };
2771 $self->_verify_pid if $self->_dbh;
2773 # some databases need this to stop spewing warnings
2774 if (my $dbh = $self->_dbh) {
2776 eval { $dbh->disconnect };
2786 =head2 DBIx::Class and AutoCommit
2788 DBIx::Class can do some wonderful magic with handling exceptions,
2789 disconnections, and transactions when you use C<< AutoCommit => 1 >>
2790 (the default) combined with C<txn_do> for transaction support.
2792 If you set C<< AutoCommit => 0 >> in your connect info, then you are always
2793 in an assumed transaction between commits, and you're telling us you'd
2794 like to manage that manually. A lot of the magic protections offered by
2795 this module will go away. We can't protect you from exceptions due to database
2796 disconnects because we don't know anything about how to restart your
2797 transactions. You're on your own for handling all sorts of exceptional
2798 cases if you choose the C<< AutoCommit => 0 >> path, just as you would
2804 Matt S. Trout <mst@shadowcatsystems.co.uk>
2806 Andy Grundman <andy@hybridized.org>
2810 You may distribute this code under the same terms as Perl itself.