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;
17 __PACKAGE__->mk_group_accessors('simple' =>
18 qw/_connect_info _dbi_connect_info _dbh _sql_maker _sql_maker_opts _conn_pid
19 _conn_tid transaction_depth _dbh_autocommit _driver_determined savepoints/
22 # the values for these accessors are picked out (and deleted) from
23 # the attribute hashref passed to connect_info
24 my @storage_options = qw/
25 on_connect_call on_disconnect_call on_connect_do on_disconnect_do
26 disable_sth_caching unsafe auto_savepoint
28 __PACKAGE__->mk_group_accessors('simple' => @storage_options);
31 # default cursor class, overridable in connect_info attributes
32 __PACKAGE__->cursor_class('DBIx::Class::Storage::DBI::Cursor');
34 __PACKAGE__->mk_group_accessors('inherited' => qw/sql_maker_class/);
35 __PACKAGE__->sql_maker_class('DBIx::Class::SQLAHacks');
40 DBIx::Class::Storage::DBI - DBI storage handler
44 my $schema = MySchema->connect('dbi:SQLite:my.db');
46 $schema->storage->debug(1);
47 $schema->dbh_do("DROP TABLE authors");
49 $schema->resultset('Book')->search({
50 written_on => $schema->storage->datetime_parser(DateTime->now)
55 This class represents the connection to an RDBMS via L<DBI>. See
56 L<DBIx::Class::Storage> for general information. This pod only
57 documents DBI-specific methods and behaviors.
64 my $new = shift->next::method(@_);
66 $new->transaction_depth(0);
67 $new->_sql_maker_opts({});
68 $new->{savepoints} = [];
69 $new->{_in_dbh_do} = 0;
77 This method is normally called by L<DBIx::Class::Schema/connection>, which
78 encapsulates its argument list in an arrayref before passing them here.
80 The argument list may contain:
86 The same 4-element argument set one would normally pass to
87 L<DBI/connect>, optionally followed by
88 L<extra attributes|/DBIx::Class specific connection attributes>
89 recognized by DBIx::Class:
91 $connect_info_args = [ $dsn, $user, $password, \%dbi_attributes?, \%extra_attributes? ];
95 A single code reference which returns a connected
96 L<DBI database handle|DBI/connect> optionally followed by
97 L<extra attributes|/DBIx::Class specific connection attributes> recognized
100 $connect_info_args = [ sub { DBI->connect (...) }, \%extra_attributes? ];
104 A single hashref with all the attributes and the dsn/user/password
107 $connect_info_args = [{
115 This is particularly useful for L<Catalyst> based applications, allowing the
116 following config (L<Config::General> style):
121 dsn dbi:mysql:database=test
130 Please note that the L<DBI> docs recommend that you always explicitly
131 set C<AutoCommit> to either I<0> or I<1>. L<DBIx::Class> further
132 recommends that it be set to I<1>, and that you perform transactions
133 via our L<DBIx::Class::Schema/txn_do> method. L<DBIx::Class> will set it
134 to I<1> if you do not do explicitly set it to zero. This is the default
135 for most DBDs. See L</DBIx::Class and AutoCommit> for details.
137 =head3 DBIx::Class specific connection attributes
139 In addition to the standard L<DBI|DBI/ATTRIBUTES_COMMON_TO_ALL_HANDLES>
140 L<connection|DBI/Database_Handle_Attributes> attributes, DBIx::Class recognizes
141 the following connection options. These options can be mixed in with your other
142 L<DBI> connection attributes, or placed in a seperate hashref
143 (C<\%extra_attributes>) as shown above.
145 Every time C<connect_info> is invoked, any previous settings for
146 these options will be cleared before setting the new ones, regardless of
147 whether any options are specified in the new C<connect_info>.
154 Specifies things to do immediately after connecting or re-connecting to
155 the database. Its value may contain:
161 This contains one SQL statement to execute.
163 =item an array reference
165 This contains SQL statements to execute in order. Each element contains
166 a string or a code reference that returns a string.
168 =item a code reference
170 This contains some code to execute. Unlike code references within an
171 array reference, its return value is ignored.
175 =item on_disconnect_do
177 Takes arguments in the same form as L</on_connect_do> and executes them
178 immediately before disconnecting from the database.
180 Note, this only runs if you explicitly call L</disconnect> on the
183 =item on_connect_call
185 A more generalized form of L</on_connect_do> that calls the specified
186 C<connect_call_METHOD> methods in your storage driver.
188 on_connect_do => 'select 1'
192 on_connect_call => [ [ do_sql => 'select 1' ] ]
194 Its values may contain:
200 Will call the C<connect_call_METHOD> method.
202 =item a code reference
204 Will execute C<< $code->($storage) >>
206 =item an array reference
208 Each value can be a method name or code reference.
210 =item an array of arrays
212 For each array, the first item is taken to be the C<connect_call_> method name
213 or code reference, and the rest are parameters to it.
217 Some predefined storage methods you may use:
223 Executes a SQL string or a code reference that returns a SQL string. This is
224 what L</on_connect_do> and L</on_disconnect_do> use.
232 Will execute the scalar as SQL.
236 Taken to be arguments to L<DBI/do>, the SQL string optionally followed by the
237 attributes hashref and bind values.
239 =item a code reference
241 Will execute C<< $code->($storage) >> and execute the return array refs as
248 Execute any statements necessary to initialize the database session to return
249 and accept datetime/timestamp values used with
250 L<DBIx::Class::InflateColumn::DateTime>.
252 Only necessary for some databases, see your specific storage driver for
253 implementation details.
257 =item on_disconnect_call
259 Takes arguments in the same form as L</on_connect_call> and executes them
260 immediately before disconnecting from the database.
262 Calls the C<disconnect_call_METHOD> methods as opposed to the
263 C<connect_call_METHOD> methods called by L</on_connect_call>.
265 Note, this only runs if you explicitly call L</disconnect> on the
268 =item disable_sth_caching
270 If set to a true value, this option will disable the caching of
271 statement handles via L<DBI/prepare_cached>.
275 Sets the limit dialect. This is useful for JDBC-bridge among others
276 where the remote SQL-dialect cannot be determined by the name of the
277 driver alone. See also L<SQL::Abstract::Limit>.
281 Specifies what characters to use to quote table and column names. If
282 you use this you will want to specify L</name_sep> as well.
284 C<quote_char> expects either a single character, in which case is it
285 is placed on either side of the table/column name, or an arrayref of length
286 2 in which case the table/column name is placed between the elements.
288 For example under MySQL you should use C<< quote_char => '`' >>, and for
289 SQL Server you should use C<< quote_char => [qw/[ ]/] >>.
293 This only needs to be used in conjunction with C<quote_char>, and is used to
294 specify the charecter that seperates elements (schemas, tables, columns) from
295 each other. In most cases this is simply a C<.>.
297 The consequences of not supplying this value is that L<SQL::Abstract>
298 will assume DBIx::Class' uses of aliases to be complete column
299 names. The output will look like I<"me.name"> when it should actually
304 This Storage driver normally installs its own C<HandleError>, sets
305 C<RaiseError> and C<ShowErrorStatement> on, and sets C<PrintError> off on
306 all database handles, including those supplied by a coderef. It does this
307 so that it can have consistent and useful error behavior.
309 If you set this option to a true value, Storage will not do its usual
310 modifications to the database handle's attributes, and instead relies on
311 the settings in your connect_info DBI options (or the values you set in
312 your connection coderef, in the case that you are connecting via coderef).
314 Note that your custom settings can cause Storage to malfunction,
315 especially if you set a C<HandleError> handler that suppresses exceptions
316 and/or disable C<RaiseError>.
320 If this option is true, L<DBIx::Class> will use savepoints when nesting
321 transactions, making it possible to recover from failure in the inner
322 transaction without having to abort all outer transactions.
326 Use this argument to supply a cursor class other than the default
327 L<DBIx::Class::Storage::DBI::Cursor>.
331 Some real-life examples of arguments to L</connect_info> and
332 L<DBIx::Class::Schema/connect>
334 # Simple SQLite connection
335 ->connect_info([ 'dbi:SQLite:./foo.db' ]);
338 ->connect_info([ sub { DBI->connect(...) } ]);
340 # A bit more complicated
347 { quote_char => q{"}, name_sep => q{.} },
351 # Equivalent to the previous example
357 { AutoCommit => 1, quote_char => q{"}, name_sep => q{.} },
361 # Same, but with hashref as argument
362 # See parse_connect_info for explanation
365 dsn => 'dbi:Pg:dbname=foo',
367 password => 'my_pg_password',
374 # Subref + DBIx::Class-specific connection options
377 sub { DBI->connect(...) },
381 on_connect_do => ['SET search_path TO myschema,otherschema,public'],
382 disable_sth_caching => 1,
392 my ($self, $info_arg) = @_;
394 return $self->_connect_info if !$info_arg;
396 my @args = @$info_arg; # take a shallow copy for further mutilation
397 $self->_connect_info([@args]); # copy for _connect_info
400 # combine/pre-parse arguments depending on invocation style
403 if (ref $args[0] eq 'CODE') { # coderef with optional \%extra_attributes
404 %attrs = %{ $args[1] || {} };
407 elsif (ref $args[0] eq 'HASH') { # single hashref (i.e. Catalyst config)
408 %attrs = %{$args[0]};
410 for (qw/password user dsn/) {
411 unshift @args, delete $attrs{$_};
414 else { # otherwise assume dsn/user/password + \%attrs + \%extra_attrs
416 % { $args[3] || {} },
417 % { $args[4] || {} },
419 @args = @args[0,1,2];
422 # Kill sql_maker/_sql_maker_opts, so we get a fresh one with only
423 # the new set of options
424 $self->_sql_maker(undef);
425 $self->_sql_maker_opts({});
428 for my $storage_opt (@storage_options, 'cursor_class') { # @storage_options is declared at the top of the module
429 if(my $value = delete $attrs{$storage_opt}) {
430 $self->$storage_opt($value);
433 for my $sql_maker_opt (qw/limit_dialect quote_char name_sep/) {
434 if(my $opt_val = delete $attrs{$sql_maker_opt}) {
435 $self->_sql_maker_opts->{$sql_maker_opt} = $opt_val;
440 if (ref $args[0] eq 'CODE') {
441 # _connect() never looks past $args[0] in this case
445 %{ $self->_default_dbi_connect_attributes || {} },
450 $self->_dbi_connect_info([@args, keys %attrs ? \%attrs : ()]);
451 $self->_connect_info;
454 sub _default_dbi_connect_attributes {
464 This method is deprecated in favour of setting via L</connect_info>.
468 =head2 on_disconnect_do
470 This method is deprecated in favour of setting via L</connect_info>.
474 sub _parse_connect_do {
475 my ($self, $type) = @_;
477 my $val = $self->$type;
478 return () if not defined $val;
483 push @res, [ 'do_sql', $val ];
484 } elsif (ref($val) eq 'CODE') {
486 } elsif (ref($val) eq 'ARRAY') {
487 push @res, map { [ 'do_sql', $_ ] } @$val;
489 $self->throw_exception("Invalid type for $type: ".ref($val));
497 Arguments: ($subref | $method_name), @extra_coderef_args?
499 Execute the given $subref or $method_name using the new exception-based
500 connection management.
502 The first two arguments will be the storage object that C<dbh_do> was called
503 on and a database handle to use. Any additional arguments will be passed
504 verbatim to the called subref as arguments 2 and onwards.
506 Using this (instead of $self->_dbh or $self->dbh) ensures correct
507 exception handling and reconnection (or failover in future subclasses).
509 Your subref should have no side-effects outside of the database, as
510 there is the potential for your subref to be partially double-executed
511 if the database connection was stale/dysfunctional.
515 my @stuff = $schema->storage->dbh_do(
517 my ($storage, $dbh, @cols) = @_;
518 my $cols = join(q{, }, @cols);
519 $dbh->selectrow_array("SELECT $cols FROM foo");
530 my $dbh = $self->_dbh;
532 return $self->$code($dbh, @_) if $self->{_in_dbh_do}
533 || $self->{transaction_depth};
535 local $self->{_in_dbh_do} = 1;
538 my $want_array = wantarray;
541 $self->_verify_pid if $dbh;
543 $self->_populate_dbh;
548 @result = $self->$code($dbh, @_);
550 elsif(defined $want_array) {
551 $result[0] = $self->$code($dbh, @_);
554 $self->$code($dbh, @_);
558 # ->connected might unset $@ - copy
560 if(!$exception) { return $want_array ? @result : $result[0] }
562 $self->throw_exception($exception) if $self->connected;
564 # We were not connected - reconnect and retry, but let any
565 # exception fall right through this time
566 carp "Retrying $code after catching disconnected exception: $exception"
567 if $ENV{DBIC_DBIRETRY_DEBUG};
568 $self->_populate_dbh;
569 $self->$code($self->_dbh, @_);
572 # This is basically a blend of dbh_do above and DBIx::Class::Storage::txn_do.
573 # It also informs dbh_do to bypass itself while under the direction of txn_do,
574 # via $self->{_in_dbh_do} (this saves some redundant eval and errorcheck, etc)
579 ref $coderef eq 'CODE' or $self->throw_exception
580 ('$coderef must be a CODE reference');
582 return $coderef->(@_) if $self->{transaction_depth} && ! $self->auto_savepoint;
584 local $self->{_in_dbh_do} = 1;
587 my $want_array = wantarray;
592 $self->_verify_pid if $self->_dbh;
593 $self->_populate_dbh if !$self->_dbh;
597 @result = $coderef->(@_);
599 elsif(defined $want_array) {
600 $result[0] = $coderef->(@_);
608 # ->connected might unset $@ - copy
610 if(!$exception) { return $want_array ? @result : $result[0] }
612 if($tried++ || $self->connected) {
613 eval { $self->txn_rollback };
614 my $rollback_exception = $@;
615 if($rollback_exception) {
616 my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
617 $self->throw_exception($exception) # propagate nested rollback
618 if $rollback_exception =~ /$exception_class/;
620 $self->throw_exception(
621 "Transaction aborted: ${exception}. "
622 . "Rollback failed: ${rollback_exception}"
625 $self->throw_exception($exception)
628 # We were not connected, and was first try - reconnect and retry
630 carp "Retrying $coderef after catching disconnected exception: $exception"
631 if $ENV{DBIC_DBIRETRY_DEBUG};
632 $self->_populate_dbh;
638 Our C<disconnect> method also performs a rollback first if the
639 database is not in C<AutoCommit> mode.
649 push @actions, ( $self->on_disconnect_call || () );
650 push @actions, $self->_parse_connect_do ('on_disconnect_do');
652 $self->_do_connection_actions(disconnect_call_ => $_) for @actions;
654 $self->_dbh_rollback unless $self->_dbh_autocommit;
656 $self->_dbh->disconnect;
662 =head2 with_deferred_fk_checks
666 =item Arguments: C<$coderef>
668 =item Return Value: The return value of $coderef
672 Storage specific method to run the code ref with FK checks deferred or
673 in MySQL's case disabled entirely.
677 # Storage subclasses should override this
678 sub with_deferred_fk_checks {
679 my ($self, $sub) = @_;
688 =item Arguments: none
690 =item Return Value: 1|0
694 Verifies that the the current database handle is active and ready to execute
695 an SQL statement (i.e. the connection did not get stale, server is still
696 answering, etc.) This method is used internally by L</dbh>.
702 return 0 unless $self->_seems_connected;
705 local $self->_dbh->{RaiseError} = 1;
710 sub _seems_connected {
713 my $dbh = $self->_dbh
716 if(defined $self->_conn_tid && $self->_conn_tid != threads->tid) {
723 return 0 if !$self->_dbh;
726 return $dbh->FETCH('Active');
732 my $dbh = $self->_dbh or return 0;
737 # handle pid changes correctly
738 # NOTE: assumes $self->_dbh is a valid $dbh
742 return if defined $self->_conn_pid && $self->_conn_pid == $$;
744 $self->_dbh->{InactiveDestroy} = 1;
751 sub ensure_connected {
754 unless ($self->connected) {
755 $self->_populate_dbh;
761 Returns a C<$dbh> - a data base handle of class L<DBI>. The returned handle
762 is guaranteed to be healthy by implicitly calling L</connected>, and if
763 necessary performing a reconnection before returning. Keep in mind that this
764 is very B<expensive> on some database engines. Consider using L<dbh_do>
772 if (not $self->_dbh) {
773 $self->_populate_dbh;
775 $self->ensure_connected;
780 # this is the internal "get dbh or connect (don't check)" method
783 $self->_populate_dbh unless $self->_dbh;
787 sub _sql_maker_args {
792 array_datatypes => 1,
793 limit_dialect => $self->_get_dbh,
794 %{$self->_sql_maker_opts}
800 unless ($self->_sql_maker) {
801 my $sql_maker_class = $self->sql_maker_class;
802 $self->ensure_class_loaded ($sql_maker_class);
803 $self->_sql_maker($sql_maker_class->new( $self->_sql_maker_args ));
805 return $self->_sql_maker;
813 my @info = @{$self->_dbi_connect_info || []};
814 $self->_dbh(undef); # in case ->connected failed we might get sent here
815 $self->_dbh($self->_connect(@info));
817 $self->_conn_pid($$);
818 $self->_conn_tid(threads->tid) if $INC{'threads.pm'};
820 $self->_determine_driver;
822 # Always set the transaction depth on connect, since
823 # there is no transaction in progress by definition
824 $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
826 $self->_run_connection_actions unless $self->{_in_determine_driver};
829 sub _run_connection_actions {
833 push @actions, ( $self->on_connect_call || () );
834 push @actions, $self->_parse_connect_do ('on_connect_do');
836 $self->_do_connection_actions(connect_call_ => $_) for @actions;
839 sub _determine_driver {
842 if ((not $self->_driver_determined) && (not $self->{_in_determine_driver})) {
843 my $started_unconnected = 0;
844 local $self->{_in_determine_driver} = 1;
846 if (ref($self) eq __PACKAGE__) {
848 if ($self->_dbh) { # we are connected
849 $driver = $self->_dbh->{Driver}{Name};
851 # try to use dsn to not require being connected, the driver may still
852 # force a connection in _rebless to determine version
853 ($driver) = $self->_dbi_connect_info->[0] =~ /dbi:([^:]+):/i;
854 $started_unconnected = 1;
857 my $storage_class = "DBIx::Class::Storage::DBI::${driver}";
858 if ($self->load_optional_class($storage_class)) {
859 mro::set_mro($storage_class, 'c3');
860 bless $self, $storage_class;
865 $self->_driver_determined(1);
867 $self->_run_connection_actions
868 if $started_unconnected && defined $self->_dbh;
872 sub _do_connection_actions {
874 my $method_prefix = shift;
877 if (not ref($call)) {
878 my $method = $method_prefix . $call;
880 } elsif (ref($call) eq 'CODE') {
882 } elsif (ref($call) eq 'ARRAY') {
883 if (ref($call->[0]) ne 'ARRAY') {
884 $self->_do_connection_actions($method_prefix, $_) for @$call;
886 $self->_do_connection_actions($method_prefix, @$_) for @$call;
889 $self->throw_exception (sprintf ("Don't know how to process conection actions of type '%s'", ref($call)) );
895 sub connect_call_do_sql {
897 $self->_do_query(@_);
900 sub disconnect_call_do_sql {
902 $self->_do_query(@_);
905 # override in db-specific backend when necessary
906 sub connect_call_datetime_setup { 1 }
909 my ($self, $action) = @_;
911 if (ref $action eq 'CODE') {
912 $action = $action->($self);
913 $self->_do_query($_) foreach @$action;
916 # Most debuggers expect ($sql, @bind), so we need to exclude
917 # the attribute hash which is the second argument to $dbh->do
918 # furthermore the bind values are usually to be presented
919 # as named arrayref pairs, so wrap those here too
920 my @do_args = (ref $action eq 'ARRAY') ? (@$action) : ($action);
921 my $sql = shift @do_args;
922 my $attrs = shift @do_args;
923 my @bind = map { [ undef, $_ ] } @do_args;
925 $self->_query_start($sql, @bind);
926 $self->_dbh->do($sql, $attrs, @do_args);
927 $self->_query_end($sql, @bind);
934 my ($self, @info) = @_;
936 $self->throw_exception("You failed to provide any connection info")
939 my ($old_connect_via, $dbh);
941 if ($INC{'Apache/DBI.pm'} && $ENV{MOD_PERL}) {
942 $old_connect_via = $DBI::connect_via;
943 $DBI::connect_via = 'connect';
947 if(ref $info[0] eq 'CODE') {
951 $dbh = DBI->connect(@info);
954 if($dbh && !$self->unsafe) {
955 my $weak_self = $self;
956 Scalar::Util::weaken($weak_self);
957 $dbh->{HandleError} = sub {
959 $weak_self->throw_exception("DBI Exception: $_[0]");
962 croak ("DBI Exception: $_[0]");
965 $dbh->{ShowErrorStatement} = 1;
966 $dbh->{RaiseError} = 1;
967 $dbh->{PrintError} = 0;
971 $DBI::connect_via = $old_connect_via if $old_connect_via;
973 $self->throw_exception("DBI Connection failed: " . ($@||$DBI::errstr))
976 $self->_dbh_autocommit($dbh->{AutoCommit});
982 my ($self, $name) = @_;
984 $name = $self->_svp_generate_name
985 unless defined $name;
987 $self->throw_exception ("You can't use savepoints outside a transaction")
988 if $self->{transaction_depth} == 0;
990 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
991 unless $self->can('_svp_begin');
993 push @{ $self->{savepoints} }, $name;
995 $self->debugobj->svp_begin($name) if $self->debug;
997 return $self->_svp_begin($name);
1001 my ($self, $name) = @_;
1003 $self->throw_exception ("You can't use savepoints outside a transaction")
1004 if $self->{transaction_depth} == 0;
1006 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1007 unless $self->can('_svp_release');
1009 if (defined $name) {
1010 $self->throw_exception ("Savepoint '$name' does not exist")
1011 unless grep { $_ eq $name } @{ $self->{savepoints} };
1013 # Dig through the stack until we find the one we are releasing. This keeps
1014 # the stack up to date.
1017 do { $svp = pop @{ $self->{savepoints} } } while $svp ne $name;
1019 $name = pop @{ $self->{savepoints} };
1022 $self->debugobj->svp_release($name) if $self->debug;
1024 return $self->_svp_release($name);
1028 my ($self, $name) = @_;
1030 $self->throw_exception ("You can't use savepoints outside a transaction")
1031 if $self->{transaction_depth} == 0;
1033 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1034 unless $self->can('_svp_rollback');
1036 if (defined $name) {
1037 # If they passed us a name, verify that it exists in the stack
1038 unless(grep({ $_ eq $name } @{ $self->{savepoints} })) {
1039 $self->throw_exception("Savepoint '$name' does not exist!");
1042 # Dig through the stack until we find the one we are releasing. This keeps
1043 # the stack up to date.
1044 while(my $s = pop(@{ $self->{savepoints} })) {
1045 last if($s eq $name);
1047 # Add the savepoint back to the stack, as a rollback doesn't remove the
1048 # named savepoint, only everything after it.
1049 push(@{ $self->{savepoints} }, $name);
1051 # We'll assume they want to rollback to the last savepoint
1052 $name = $self->{savepoints}->[-1];
1055 $self->debugobj->svp_rollback($name) if $self->debug;
1057 return $self->_svp_rollback($name);
1060 sub _svp_generate_name {
1063 return 'savepoint_'.scalar(@{ $self->{'savepoints'} });
1068 if($self->{transaction_depth} == 0) {
1069 $self->debugobj->txn_begin()
1071 $self->_dbh_begin_work;
1073 elsif ($self->auto_savepoint) {
1076 $self->{transaction_depth}++;
1079 sub _dbh_begin_work {
1081 if ($self->{_in_dbh_do}) {
1082 $self->_dbh->begin_work;
1084 $self->dbh_do(sub { $_[1]->begin_work });
1090 if ($self->{transaction_depth} == 1) {
1091 my $dbh = $self->_dbh;
1092 $self->debugobj->txn_commit()
1095 $self->{transaction_depth} = 0
1096 if $self->_dbh_autocommit;
1098 elsif($self->{transaction_depth} > 1) {
1099 $self->{transaction_depth}--;
1101 if $self->auto_savepoint;
1107 $self->_dbh->commit;
1112 my $dbh = $self->_dbh;
1114 if ($self->{transaction_depth} == 1) {
1115 $self->debugobj->txn_rollback()
1117 $self->{transaction_depth} = 0
1118 if $self->_dbh_autocommit;
1119 $self->_dbh_rollback;
1121 elsif($self->{transaction_depth} > 1) {
1122 $self->{transaction_depth}--;
1123 if ($self->auto_savepoint) {
1124 $self->svp_rollback;
1129 die DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION->new;
1134 my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
1135 $error =~ /$exception_class/ and $self->throw_exception($error);
1136 # ensure that a failed rollback resets the transaction depth
1137 $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
1138 $self->throw_exception($error);
1144 $self->_dbh->rollback;
1147 # This used to be the top-half of _execute. It was split out to make it
1148 # easier to override in NoBindVars without duping the rest. It takes up
1149 # all of _execute's args, and emits $sql, @bind.
1150 sub _prep_for_execute {
1151 my ($self, $op, $extra_bind, $ident, $args) = @_;
1153 if( Scalar::Util::blessed($ident) && $ident->isa("DBIx::Class::ResultSource") ) {
1154 $ident = $ident->from();
1157 my ($sql, @bind) = $self->sql_maker->$op($ident, @$args);
1160 map { ref $_ eq 'ARRAY' ? $_ : [ '!!dummy', $_ ] } @$extra_bind)
1162 return ($sql, \@bind);
1166 sub _fix_bind_params {
1167 my ($self, @bind) = @_;
1169 ### Turn @bind from something like this:
1170 ### ( [ "artist", 1 ], [ "cdid", 1, 3 ] )
1172 ### ( "'1'", "'1'", "'3'" )
1175 if ( defined( $_ && $_->[1] ) ) {
1176 map { qq{'$_'}; } @{$_}[ 1 .. $#$_ ];
1183 my ( $self, $sql, @bind ) = @_;
1185 if ( $self->debug ) {
1186 @bind = $self->_fix_bind_params(@bind);
1188 $self->debugobj->query_start( $sql, @bind );
1193 my ( $self, $sql, @bind ) = @_;
1195 if ( $self->debug ) {
1196 @bind = $self->_fix_bind_params(@bind);
1197 $self->debugobj->query_end( $sql, @bind );
1202 my ($self, $dbh, $op, $extra_bind, $ident, $bind_attributes, @args) = @_;
1204 my ($sql, $bind) = $self->_prep_for_execute($op, $extra_bind, $ident, \@args);
1206 $self->_query_start( $sql, @$bind );
1208 my $sth = $self->sth($sql,$op);
1210 my $placeholder_index = 1;
1212 foreach my $bound (@$bind) {
1213 my $attributes = {};
1214 my($column_name, @data) = @$bound;
1216 if ($bind_attributes) {
1217 $attributes = $bind_attributes->{$column_name}
1218 if defined $bind_attributes->{$column_name};
1221 foreach my $data (@data) {
1222 my $ref = ref $data;
1223 $data = $ref && $ref ne 'ARRAY' ? ''.$data : $data; # stringify args (except arrayrefs)
1225 $sth->bind_param($placeholder_index, $data, $attributes);
1226 $placeholder_index++;
1230 # Can this fail without throwing an exception anyways???
1231 my $rv = $sth->execute();
1232 $self->throw_exception($sth->errstr) if !$rv;
1234 $self->_query_end( $sql, @$bind );
1236 return (wantarray ? ($rv, $sth, @$bind) : $rv);
1241 $self->dbh_do('_dbh_execute', @_)
1245 my ($self, $source, $to_insert) = @_;
1247 # redispatch to insert method of storage we reblessed into, if necessary
1248 if (not $self->_driver_determined) {
1249 $self->_determine_driver;
1250 goto $self->can('insert');
1253 my $ident = $source->from;
1254 my $bind_attributes = $self->source_bind_attributes($source);
1256 my $updated_cols = {};
1258 foreach my $col ( $source->columns ) {
1259 if ( !defined $to_insert->{$col} ) {
1260 my $col_info = $source->column_info($col);
1262 if ( $col_info->{auto_nextval} ) {
1263 $updated_cols->{$col} = $to_insert->{$col} = $self->_sequence_fetch(
1265 $col_info->{sequence} ||
1266 $self->_dbh_get_autoinc_seq($self->_get_dbh, $source)
1272 $self->_execute('insert' => [], $source, $bind_attributes, $to_insert);
1274 return $updated_cols;
1277 ## Still not quite perfect, and EXPERIMENTAL
1278 ## Currently it is assumed that all values passed will be "normal", i.e. not
1279 ## scalar refs, or at least, all the same type as the first set, the statement is
1280 ## only prepped once.
1282 my ($self, $source, $cols, $data) = @_;
1284 my $table = $source->from;
1285 @colvalues{@$cols} = (0..$#$cols);
1286 my ($sql, @bind) = $self->sql_maker->insert($table, \%colvalues);
1288 $self->_determine_driver;
1290 $self->_query_start( $sql, @bind );
1291 my $sth = $self->sth($sql);
1293 # @bind = map { ref $_ ? ''.$_ : $_ } @bind; # stringify args
1295 ## This must be an arrayref, else nothing works!
1296 my $tuple_status = [];
1298 ## Get the bind_attributes, if any exist
1299 my $bind_attributes = $self->source_bind_attributes($source);
1301 ## Bind the values and execute
1302 my $placeholder_index = 1;
1304 foreach my $bound (@bind) {
1306 my $attributes = {};
1307 my ($column_name, $data_index) = @$bound;
1309 if( $bind_attributes ) {
1310 $attributes = $bind_attributes->{$column_name}
1311 if defined $bind_attributes->{$column_name};
1314 my @data = map { $_->[$data_index] } @$data;
1316 $sth->bind_param_array( $placeholder_index, [@data], $attributes );
1317 $placeholder_index++;
1319 my $rv = eval { $sth->execute_array({ArrayTupleStatus => $tuple_status}) };
1322 ++$i while $i <= $#$tuple_status && !ref $tuple_status->[$i];
1324 $self->throw_exception($sth->errstr || "Unexpected populate error: $err")
1325 if ($i > $#$tuple_status);
1327 require Data::Dumper;
1328 local $Data::Dumper::Terse = 1;
1329 local $Data::Dumper::Indent = 1;
1330 local $Data::Dumper::Useqq = 1;
1331 local $Data::Dumper::Quotekeys = 0;
1333 $self->throw_exception(sprintf "%s for populate slice:\n%s",
1334 $tuple_status->[$i][1],
1335 Data::Dumper::Dumper(
1336 { map { $cols->[$_] => $data->[$i][$_] } (0 .. $#$cols) }
1340 $self->throw_exception($sth->errstr) if !$rv;
1342 $self->_query_end( $sql, @bind );
1343 return (wantarray ? ($rv, $sth, @bind) : $rv);
1347 my ($self, $source, @args) = @_;
1349 # redispatch to update method of storage we reblessed into, if necessary
1350 if (not $self->_driver_determined) {
1351 $self->_determine_driver;
1352 goto $self->can('update');
1355 my $bind_attributes = $self->source_bind_attributes($source);
1357 return $self->_execute('update' => [], $source, $bind_attributes, @args);
1362 my $self = shift @_;
1363 my $source = shift @_;
1364 $self->_determine_driver;
1365 my $bind_attrs = $self->source_bind_attributes($source);
1367 return $self->_execute('delete' => [], $source, $bind_attrs, @_);
1370 # We were sent here because the $rs contains a complex search
1371 # which will require a subquery to select the correct rows
1372 # (i.e. joined or limited resultsets)
1374 # Genarating a single PK column subquery is trivial and supported
1375 # by all RDBMS. However if we have a multicolumn PK, things get ugly.
1376 # Look at _multipk_update_delete()
1377 sub _subq_update_delete {
1379 my ($rs, $op, $values) = @_;
1381 my $rsrc = $rs->result_source;
1383 # we already check this, but double check naively just in case. Should be removed soon
1384 my $sel = $rs->_resolved_attrs->{select};
1385 $sel = [ $sel ] unless ref $sel eq 'ARRAY';
1386 my @pcols = $rsrc->primary_columns;
1387 if (@$sel != @pcols) {
1388 $self->throw_exception (
1389 'Subquery update/delete can not be called on resultsets selecting a'
1390 .' number of columns different than the number of primary keys'
1397 $op eq 'update' ? $values : (),
1398 { $pcols[0] => { -in => $rs->as_query } },
1403 return $self->_multipk_update_delete (@_);
1407 # ANSI SQL does not provide a reliable way to perform a multicol-PK
1408 # resultset update/delete involving subqueries. So by default resort
1409 # to simple (and inefficient) delete_all style per-row opearations,
1410 # while allowing specific storages to override this with a faster
1413 sub _multipk_update_delete {
1414 return shift->_per_row_update_delete (@_);
1417 # This is the default loop used to delete/update rows for multi PK
1418 # resultsets, and used by mysql exclusively (because it can't do anything
1421 # We do not use $row->$op style queries, because resultset update/delete
1422 # is not expected to cascade (this is what delete_all/update_all is for).
1424 # There should be no race conditions as the entire operation is rolled
1427 sub _per_row_update_delete {
1429 my ($rs, $op, $values) = @_;
1431 my $rsrc = $rs->result_source;
1432 my @pcols = $rsrc->primary_columns;
1434 my $guard = $self->txn_scope_guard;
1436 # emulate the return value of $sth->execute for non-selects
1437 my $row_cnt = '0E0';
1439 my $subrs_cur = $rs->cursor;
1440 while (my @pks = $subrs_cur->next) {
1443 for my $i (0.. $#pcols) {
1444 $cond->{$pcols[$i]} = $pks[$i];
1449 $op eq 'update' ? $values : (),
1464 # localization is neccessary as
1465 # 1) there is no infrastructure to pass this around before SQLA2
1466 # 2) _select_args sets it and _prep_for_execute consumes it
1467 my $sql_maker = $self->sql_maker;
1468 local $sql_maker->{_dbic_rs_attrs};
1470 return $self->_execute($self->_select_args(@_));
1473 sub _select_args_to_query {
1476 # localization is neccessary as
1477 # 1) there is no infrastructure to pass this around before SQLA2
1478 # 2) _select_args sets it and _prep_for_execute consumes it
1479 my $sql_maker = $self->sql_maker;
1480 local $sql_maker->{_dbic_rs_attrs};
1482 # my ($op, $bind, $ident, $bind_attrs, $select, $cond, $order, $rows, $offset)
1483 # = $self->_select_args($ident, $select, $cond, $attrs);
1484 my ($op, $bind, $ident, $bind_attrs, @args) =
1485 $self->_select_args(@_);
1487 # my ($sql, $prepared_bind) = $self->_prep_for_execute($op, $bind, $ident, [ $select, $cond, $order, $rows, $offset ]);
1488 my ($sql, $prepared_bind) = $self->_prep_for_execute($op, $bind, $ident, \@args);
1489 $prepared_bind ||= [];
1492 ? ($sql, $prepared_bind, $bind_attrs)
1493 : \[ "($sql)", @$prepared_bind ]
1498 my ($self, $ident, $select, $where, $attrs) = @_;
1500 my ($alias2source, $rs_alias) = $self->_resolve_ident_sources ($ident);
1502 my $sql_maker = $self->sql_maker;
1503 $sql_maker->{_dbic_rs_attrs} = {
1509 ? ( _source_handle => $alias2source->{$rs_alias}->handle )
1514 # calculate bind_attrs before possible $ident mangling
1515 my $bind_attrs = {};
1516 for my $alias (keys %$alias2source) {
1517 my $bindtypes = $self->source_bind_attributes ($alias2source->{$alias}) || {};
1518 for my $col (keys %$bindtypes) {
1520 my $fqcn = join ('.', $alias, $col);
1521 $bind_attrs->{$fqcn} = $bindtypes->{$col} if $bindtypes->{$col};
1523 # Unqialified column names are nice, but at the same time can be
1524 # rather ambiguous. What we do here is basically go along with
1525 # the loop, adding an unqualified column slot to $bind_attrs,
1526 # alongside the fully qualified name. As soon as we encounter
1527 # another column by that name (which would imply another table)
1528 # we unset the unqualified slot and never add any info to it
1529 # to avoid erroneous type binding. If this happens the users
1530 # only choice will be to fully qualify his column name
1532 if (exists $bind_attrs->{$col}) {
1533 $bind_attrs->{$col} = {};
1536 $bind_attrs->{$col} = $bind_attrs->{$fqcn};
1543 $attrs->{software_limit}
1545 $sql_maker->_default_limit_syntax eq "GenericSubQ"
1547 $attrs->{software_limit} = 1;
1550 $self->throw_exception("rows attribute must be positive if present")
1551 if (defined($attrs->{rows}) && !($attrs->{rows} > 0));
1553 # MySQL actually recommends this approach. I cringe.
1554 $attrs->{rows} = 2**48 if not defined $attrs->{rows} and defined $attrs->{offset};
1559 # see if we need to tear the prefetch apart (either limited has_many or grouped prefetch)
1560 # otherwise delegate the limiting to the storage, unless software limit was requested
1562 ( $attrs->{rows} && keys %{$attrs->{collapse}} )
1564 ( $attrs->{group_by} && @{$attrs->{group_by}} &&
1565 $attrs->{_prefetch_select} && @{$attrs->{_prefetch_select}} )
1567 ($ident, $select, $where, $attrs)
1568 = $self->_adjust_select_args_for_complex_prefetch ($ident, $select, $where, $attrs);
1570 elsif (! $attrs->{software_limit} ) {
1571 push @limit, $attrs->{rows}, $attrs->{offset};
1575 # This would be the point to deflate anything found in $where
1576 # (and leave $attrs->{bind} intact). Problem is - inflators historically
1577 # expect a row object. And all we have is a resultsource (it is trivial
1578 # to extract deflator coderefs via $alias2source above).
1580 # I don't see a way forward other than changing the way deflators are
1581 # invoked, and that's just bad...
1585 { $attrs->{$_} ? ( $_ => $attrs->{$_} ) : () }
1586 (qw/order_by group_by having/ )
1589 return ('select', $attrs->{bind}, $ident, $bind_attrs, $select, $where, $order, @limit);
1593 # This is the code producing joined subqueries like:
1594 # SELECT me.*, other.* FROM ( SELECT me.* FROM ... ) JOIN other ON ...
1596 sub _adjust_select_args_for_complex_prefetch {
1597 my ($self, $from, $select, $where, $attrs) = @_;
1599 $self->throw_exception ('Nothing to prefetch... how did we get here?!')
1600 if not @{$attrs->{_prefetch_select}};
1602 $self->throw_exception ('Complex prefetches are not supported on resultsets with a custom from attribute')
1603 if (ref $from ne 'ARRAY' || ref $from->[0] ne 'HASH' || ref $from->[1] ne 'ARRAY');
1606 # generate inner/outer attribute lists, remove stuff that doesn't apply
1607 my $outer_attrs = { %$attrs };
1608 delete $outer_attrs->{$_} for qw/where bind rows offset group_by having/;
1610 my $inner_attrs = { %$attrs };
1611 delete $inner_attrs->{$_} for qw/for collapse _prefetch_select _collapse_order_by select as/;
1614 # bring over all non-collapse-induced order_by into the inner query (if any)
1615 # the outer one will have to keep them all
1616 delete $inner_attrs->{order_by};
1617 if (my $ord_cnt = @{$outer_attrs->{order_by}} - @{$outer_attrs->{_collapse_order_by}} ) {
1618 $inner_attrs->{order_by} = [
1619 @{$outer_attrs->{order_by}}[ 0 .. $ord_cnt - 1]
1624 # generate the inner/outer select lists
1625 # for inside we consider only stuff *not* brought in by the prefetch
1626 # on the outside we substitute any function for its alias
1627 my $outer_select = [ @$select ];
1628 my $inner_select = [];
1629 for my $i (0 .. ( @$outer_select - @{$outer_attrs->{_prefetch_select}} - 1) ) {
1630 my $sel = $outer_select->[$i];
1632 if (ref $sel eq 'HASH' ) {
1633 $sel->{-as} ||= $attrs->{as}[$i];
1634 $outer_select->[$i] = join ('.', $attrs->{alias}, ($sel->{-as} || "inner_column_$i") );
1637 push @$inner_select, $sel;
1640 # normalize a copy of $from, so it will be easier to work with further
1641 # down (i.e. promote the initial hashref to an AoH)
1643 $from->[0] = [ $from->[0] ];
1644 my %original_join_info = map { $_->[0]{-alias} => $_->[0] } (@$from);
1647 # decide which parts of the join will remain in either part of
1648 # the outer/inner query
1650 # First we compose a list of which aliases are used in restrictions
1651 # (i.e. conditions/order/grouping/etc). Since we do not have
1652 # introspectable SQLA, we fall back to ugly scanning of raw SQL for
1653 # WHERE, and for pieces of ORDER BY in order to determine which aliases
1654 # need to appear in the resulting sql.
1655 # It may not be very efficient, but it's a reasonable stop-gap
1656 # Also unqualified column names will not be considered, but more often
1657 # than not this is actually ok
1659 # In the same loop we enumerate part of the selection aliases, as
1660 # it requires the same sqla hack for the time being
1661 my ($restrict_aliases, $select_aliases, $prefetch_aliases);
1663 # produce stuff unquoted, so it can be scanned
1664 my $sql_maker = $self->sql_maker;
1665 local $sql_maker->{quote_char};
1666 my $sep = $self->_sql_maker_opts->{name_sep} || '.';
1669 my $non_prefetch_select_sql = $sql_maker->_recurse_fields ($inner_select);
1670 my $prefetch_select_sql = $sql_maker->_recurse_fields ($outer_attrs->{_prefetch_select});
1671 my $where_sql = $sql_maker->where ($where);
1672 my $group_by_sql = $sql_maker->_order_by({
1673 map { $_ => $inner_attrs->{$_} } qw/group_by having/
1675 my @non_prefetch_order_by_chunks = (map
1676 { ref $_ ? $_->[0] : $_ }
1677 $sql_maker->_order_by_chunks ($inner_attrs->{order_by})
1681 for my $alias (keys %original_join_info) {
1682 my $seen_re = qr/\b $alias $sep/x;
1684 for my $piece ($where_sql, $group_by_sql, @non_prefetch_order_by_chunks ) {
1685 if ($piece =~ $seen_re) {
1686 $restrict_aliases->{$alias} = 1;
1690 if ($non_prefetch_select_sql =~ $seen_re) {
1691 $select_aliases->{$alias} = 1;
1694 if ($prefetch_select_sql =~ $seen_re) {
1695 $prefetch_aliases->{$alias} = 1;
1701 # Add any non-left joins to the restriction list (such joins are indeed restrictions)
1702 for my $j (values %original_join_info) {
1703 my $alias = $j->{-alias} or next;
1704 $restrict_aliases->{$alias} = 1 if (
1705 (not $j->{-join_type})
1707 ($j->{-join_type} !~ /^left (?: \s+ outer)? $/xi)
1711 # mark all join parents as mentioned
1712 # (e.g. join => { cds => 'tracks' } - tracks will need to bring cds too )
1713 for my $collection ($restrict_aliases, $select_aliases) {
1714 for my $alias (keys %$collection) {
1715 $collection->{$_} = 1
1716 for (@{ $original_join_info{$alias}{-join_path} || [] });
1720 # construct the inner $from for the subquery
1721 my %inner_joins = (map { %{$_ || {}} } ($restrict_aliases, $select_aliases) );
1723 for my $j (@$from) {
1724 push @inner_from, $j if $inner_joins{$j->[0]{-alias}};
1727 # if a multi-type join was needed in the subquery ("multi" is indicated by
1728 # presence in {collapse}) - add a group_by to simulate the collapse in the subq
1729 unless ($inner_attrs->{group_by}) {
1730 for my $alias (keys %inner_joins) {
1732 # the dot comes from some weirdness in collapse
1733 # remove after the rewrite
1734 if ($attrs->{collapse}{".$alias"}) {
1735 $inner_attrs->{group_by} ||= $inner_select;
1741 # demote the inner_from head
1742 $inner_from[0] = $inner_from[0][0];
1744 # generate the subquery
1745 my $subq = $self->_select_args_to_query (
1752 my $subq_joinspec = {
1753 -alias => $attrs->{alias},
1754 -source_handle => $inner_from[0]{-source_handle},
1755 $attrs->{alias} => $subq,
1758 # Generate the outer from - this is relatively easy (really just replace
1759 # the join slot with the subquery), with a major caveat - we can not
1760 # join anything that is non-selecting (not part of the prefetch), but at
1761 # the same time is a multi-type relationship, as it will explode the result.
1763 # There are two possibilities here
1764 # - either the join is non-restricting, in which case we simply throw it away
1765 # - it is part of the restrictions, in which case we need to collapse the outer
1766 # result by tackling yet another group_by to the outside of the query
1768 # so first generate the outer_from, up to the substitution point
1770 while (my $j = shift @$from) {
1771 if ($j->[0]{-alias} eq $attrs->{alias}) { # time to swap
1776 last; # we'll take care of what's left in $from below
1779 push @outer_from, $j;
1783 # see what's left - throw away if not selecting/restricting
1784 # also throw in a group_by if restricting to guard against
1785 # cross-join explosions
1787 while (my $j = shift @$from) {
1788 my $alias = $j->[0]{-alias};
1790 if ($select_aliases->{$alias} || $prefetch_aliases->{$alias}) {
1791 push @outer_from, $j;
1793 elsif ($restrict_aliases->{$alias}) {
1794 push @outer_from, $j;
1796 # FIXME - this should be obviated by SQLA2, as I'll be able to
1797 # have restrict_inner and restrict_outer... or something to that
1798 # effect... I think...
1800 # FIXME2 - I can't find a clean way to determine if a particular join
1801 # is a multi - instead I am just treating everything as a potential
1802 # explosive join (ribasushi)
1804 # if (my $handle = $j->[0]{-source_handle}) {
1805 # my $rsrc = $handle->resolve;
1806 # ... need to bail out of the following if this is not a multi,
1807 # as it will be much easier on the db ...
1809 $outer_attrs->{group_by} ||= $outer_select;
1814 # demote the outer_from head
1815 $outer_from[0] = $outer_from[0][0];
1817 # This is totally horrific - the $where ends up in both the inner and outer query
1818 # Unfortunately not much can be done until SQLA2 introspection arrives, and even
1819 # then if where conditions apply to the *right* side of the prefetch, you may have
1820 # to both filter the inner select (e.g. to apply a limit) and then have to re-filter
1821 # the outer select to exclude joins you didin't want in the first place
1823 # OTOH it can be seen as a plus: <ash> (notes that this query would make a DBA cry ;)
1824 return (\@outer_from, $outer_select, $where, $outer_attrs);
1827 sub _resolve_ident_sources {
1828 my ($self, $ident) = @_;
1830 my $alias2source = {};
1833 # the reason this is so contrived is that $ident may be a {from}
1834 # structure, specifying multiple tables to join
1835 if ( Scalar::Util::blessed($ident) && $ident->isa("DBIx::Class::ResultSource") ) {
1836 # this is compat mode for insert/update/delete which do not deal with aliases
1837 $alias2source->{me} = $ident;
1840 elsif (ref $ident eq 'ARRAY') {
1844 if (ref $_ eq 'HASH') {
1846 $rs_alias = $tabinfo->{-alias};
1848 if (ref $_ eq 'ARRAY' and ref $_->[0] eq 'HASH') {
1852 $alias2source->{$tabinfo->{-alias}} = $tabinfo->{-source_handle}->resolve
1853 if ($tabinfo->{-source_handle});
1857 return ($alias2source, $rs_alias);
1860 # Takes $ident, \@column_names
1862 # returns { $column_name => \%column_info, ... }
1863 # also note: this adds -result_source => $rsrc to the column info
1866 # my $col_sources = $self->_resolve_column_info($ident, @column_names);
1867 sub _resolve_column_info {
1868 my ($self, $ident, $colnames) = @_;
1869 my ($alias2src, $root_alias) = $self->_resolve_ident_sources($ident);
1871 my $sep = $self->_sql_maker_opts->{name_sep} || '.';
1874 my (%return, %seen_cols);
1876 # compile a global list of column names, to be able to properly
1877 # disambiguate unqualified column names (if at all possible)
1878 for my $alias (keys %$alias2src) {
1879 my $rsrc = $alias2src->{$alias};
1880 for my $colname ($rsrc->columns) {
1881 push @{$seen_cols{$colname}}, $alias;
1886 foreach my $col (@$colnames) {
1887 my ($alias, $colname) = $col =~ m/^ (?: ([^$sep]+) $sep)? (.+) $/x;
1890 # see if the column was seen exactly once (so we know which rsrc it came from)
1891 if ($seen_cols{$colname} and @{$seen_cols{$colname}} == 1) {
1892 $alias = $seen_cols{$colname}[0];
1899 my $rsrc = $alias2src->{$alias};
1900 $return{$col} = $rsrc && {
1901 %{$rsrc->column_info($colname)},
1902 -result_source => $rsrc,
1903 -source_alias => $alias,
1910 # Returns a counting SELECT for a simple count
1911 # query. Abstracted so that a storage could override
1912 # this to { count => 'firstcol' } or whatever makes
1913 # sense as a performance optimization
1915 #my ($self, $source, $rs_attrs) = @_;
1916 return { count => '*' };
1919 # Returns a SELECT which will end up in the subselect
1920 # There may or may not be a group_by, as the subquery
1921 # might have been called to accomodate a limit
1923 # Most databases would be happy with whatever ends up
1924 # here, but some choke in various ways.
1926 sub _subq_count_select {
1927 my ($self, $source, $rs_attrs) = @_;
1928 return $rs_attrs->{group_by} if $rs_attrs->{group_by};
1930 my @pcols = map { join '.', $rs_attrs->{alias}, $_ } ($source->primary_columns);
1931 return @pcols ? \@pcols : [ 1 ];
1935 # Returns an ordered list of column names before they are used
1936 # in a SELECT statement. By default simply returns the list
1939 # This may be overridden in a specific storage when there are
1940 # requirements such as moving BLOB columns to the end of the
1942 sub _order_select_columns {
1943 #my ($self, $source, $columns) = @_;
1947 sub source_bind_attributes {
1948 my ($self, $source) = @_;
1950 my $bind_attributes;
1951 foreach my $column ($source->columns) {
1953 my $data_type = $source->column_info($column)->{data_type} || '';
1954 $bind_attributes->{$column} = $self->bind_attribute_by_data_type($data_type)
1958 return $bind_attributes;
1965 =item Arguments: $ident, $select, $condition, $attrs
1969 Handle a SQL select statement.
1975 my ($ident, $select, $condition, $attrs) = @_;
1976 return $self->cursor_class->new($self, \@_, $attrs);
1981 my ($rv, $sth, @bind) = $self->_select(@_);
1982 my @row = $sth->fetchrow_array;
1983 my @nextrow = $sth->fetchrow_array if @row;
1984 if(@row && @nextrow) {
1985 carp "Query returned more than one row. SQL that returns multiple rows is DEPRECATED for ->find and ->single";
1987 # Need to call finish() to work round broken DBDs
1996 =item Arguments: $sql
2000 Returns a L<DBI> sth (statement handle) for the supplied SQL.
2005 my ($self, $dbh, $sql) = @_;
2007 # 3 is the if_active parameter which avoids active sth re-use
2008 my $sth = $self->disable_sth_caching
2009 ? $dbh->prepare($sql)
2010 : $dbh->prepare_cached($sql, {}, 3);
2012 # XXX You would think RaiseError would make this impossible,
2013 # but apparently that's not true :(
2014 $self->throw_exception($dbh->errstr) if !$sth;
2020 my ($self, $sql) = @_;
2021 $self->dbh_do('_dbh_sth', $sql);
2024 sub _dbh_columns_info_for {
2025 my ($self, $dbh, $table) = @_;
2027 if ($dbh->can('column_info')) {
2030 my ($schema,$tab) = $table =~ /^(.+?)\.(.+)$/ ? ($1,$2) : (undef,$table);
2031 my $sth = $dbh->column_info( undef,$schema, $tab, '%' );
2033 while ( my $info = $sth->fetchrow_hashref() ){
2035 $column_info{data_type} = $info->{TYPE_NAME};
2036 $column_info{size} = $info->{COLUMN_SIZE};
2037 $column_info{is_nullable} = $info->{NULLABLE} ? 1 : 0;
2038 $column_info{default_value} = $info->{COLUMN_DEF};
2039 my $col_name = $info->{COLUMN_NAME};
2040 $col_name =~ s/^\"(.*)\"$/$1/;
2042 $result{$col_name} = \%column_info;
2045 return \%result if !$@ && scalar keys %result;
2049 my $sth = $dbh->prepare($self->sql_maker->select($table, undef, \'1 = 0'));
2051 my @columns = @{$sth->{NAME_lc}};
2052 for my $i ( 0 .. $#columns ){
2054 $column_info{data_type} = $sth->{TYPE}->[$i];
2055 $column_info{size} = $sth->{PRECISION}->[$i];
2056 $column_info{is_nullable} = $sth->{NULLABLE}->[$i] ? 1 : 0;
2058 if ($column_info{data_type} =~ m/^(.*?)\((.*?)\)$/) {
2059 $column_info{data_type} = $1;
2060 $column_info{size} = $2;
2063 $result{$columns[$i]} = \%column_info;
2067 foreach my $col (keys %result) {
2068 my $colinfo = $result{$col};
2069 my $type_num = $colinfo->{data_type};
2071 if(defined $type_num && $dbh->can('type_info')) {
2072 my $type_info = $dbh->type_info($type_num);
2073 $type_name = $type_info->{TYPE_NAME} if $type_info;
2074 $colinfo->{data_type} = $type_name if $type_name;
2081 sub columns_info_for {
2082 my ($self, $table) = @_;
2083 $self->dbh_do('_dbh_columns_info_for', $table);
2086 =head2 last_insert_id
2088 Return the row id of the last insert.
2092 sub _dbh_last_insert_id {
2093 # All Storage's need to register their own _dbh_last_insert_id
2094 # the old SQLite-based method was highly inappropriate
2097 my $class = ref $self;
2098 $self->throw_exception (<<EOE);
2100 No _dbh_last_insert_id() method found in $class.
2101 Since the method of obtaining the autoincrement id of the last insert
2102 operation varies greatly between different databases, this method must be
2103 individually implemented for every storage class.
2107 sub last_insert_id {
2109 $self->dbh_do('_dbh_last_insert_id', @_);
2112 =head2 _native_data_type
2116 =item Arguments: $type_name
2120 This API is B<EXPERIMENTAL>, will almost definitely change in the future, and
2121 currently only used by L<::AutoCast|DBIx::Class::Storage::DBI::AutoCast> and
2122 L<::Sybase|DBIx::Class::Storage::DBI::Sybase>.
2124 The default implementation returns C<undef>, implement in your Storage driver if
2125 you need this functionality.
2127 Should map types from other databases to the native RDBMS type, for example
2128 C<VARCHAR2> to C<VARCHAR>.
2130 Types with modifiers should map to the underlying data type. For example,
2131 C<INTEGER AUTO_INCREMENT> should become C<INTEGER>.
2133 Composite types should map to the container type, for example
2134 C<ENUM(foo,bar,baz)> becomes C<ENUM>.
2138 sub _native_data_type {
2139 #my ($self, $data_type) = @_;
2143 # Check if placeholders are supported at all
2144 sub _placeholders_supported {
2146 my $dbh = $self->_get_dbh;
2148 # some drivers provide a $dbh attribute (e.g. Sybase and $dbh->{syb_dynamic_supported})
2149 # but it is inaccurate more often than not
2151 local $dbh->{PrintError} = 0;
2152 local $dbh->{RaiseError} = 1;
2153 $dbh->do('select ?', {}, 1);
2158 # Check if placeholders bound to non-string types throw exceptions
2160 sub _typeless_placeholders_supported {
2162 my $dbh = $self->_get_dbh;
2165 local $dbh->{PrintError} = 0;
2166 local $dbh->{RaiseError} = 1;
2167 # this specifically tests a bind that is NOT a string
2168 $dbh->do('select 1 where 1 = ?', {}, 1);
2175 Returns the database driver name.
2182 if (not $self->_driver_determined) {
2183 $self->_determine_driver;
2184 goto $self->can ('sqlt_type');
2187 $self->_get_dbh->{Driver}->{Name};
2190 =head2 bind_attribute_by_data_type
2192 Given a datatype from column info, returns a database specific bind
2193 attribute for C<< $dbh->bind_param($val,$attribute) >> or nothing if we will
2194 let the database planner just handle it.
2196 Generally only needed for special case column types, like bytea in postgres.
2200 sub bind_attribute_by_data_type {
2204 =head2 is_datatype_numeric
2206 Given a datatype from column_info, returns a boolean value indicating if
2207 the current RDBMS considers it a numeric value. This controls how
2208 L<DBIx::Class::Row/set_column> decides whether to mark the column as
2209 dirty - when the datatype is deemed numeric a C<< != >> comparison will
2210 be performed instead of the usual C<eq>.
2214 sub is_datatype_numeric {
2215 my ($self, $dt) = @_;
2217 return 0 unless $dt;
2219 return $dt =~ /^ (?:
2220 numeric | int(?:eger)? | (?:tiny|small|medium|big)int | dec(?:imal)? | real | float | double (?: \s+ precision)? | (?:big)?serial
2225 =head2 create_ddl_dir (EXPERIMENTAL)
2229 =item Arguments: $schema \@databases, $version, $directory, $preversion, \%sqlt_args
2233 Creates a SQL file based on the Schema, for each of the specified
2234 database engines in C<\@databases> in the given directory.
2235 (note: specify L<SQL::Translator> names, not L<DBI> driver names).
2237 Given a previous version number, this will also create a file containing
2238 the ALTER TABLE statements to transform the previous schema into the
2239 current one. Note that these statements may contain C<DROP TABLE> or
2240 C<DROP COLUMN> statements that can potentially destroy data.
2242 The file names are created using the C<ddl_filename> method below, please
2243 override this method in your schema if you would like a different file
2244 name format. For the ALTER file, the same format is used, replacing
2245 $version in the name with "$preversion-$version".
2247 See L<SQL::Translator/METHODS> for a list of values for C<\%sqlt_args>.
2248 The most common value for this would be C<< { add_drop_table => 1 } >>
2249 to have the SQL produced include a C<DROP TABLE> statement for each table
2250 created. For quoting purposes supply C<quote_table_names> and
2251 C<quote_field_names>.
2253 If no arguments are passed, then the following default values are assumed:
2257 =item databases - ['MySQL', 'SQLite', 'PostgreSQL']
2259 =item version - $schema->schema_version
2261 =item directory - './'
2263 =item preversion - <none>
2267 By default, C<\%sqlt_args> will have
2269 { add_drop_table => 1, ignore_constraint_names => 1, ignore_index_names => 1 }
2271 merged with the hash passed in. To disable any of those features, pass in a
2272 hashref like the following
2274 { ignore_constraint_names => 0, # ... other options }
2277 Note that this feature is currently EXPERIMENTAL and may not work correctly
2278 across all databases, or fully handle complex relationships.
2280 WARNING: Please check all SQL files created, before applying them.
2284 sub create_ddl_dir {
2285 my ($self, $schema, $databases, $version, $dir, $preversion, $sqltargs) = @_;
2287 if(!$dir || !-d $dir) {
2288 carp "No directory given, using ./\n";
2291 $databases ||= ['MySQL', 'SQLite', 'PostgreSQL'];
2292 $databases = [ $databases ] if(ref($databases) ne 'ARRAY');
2294 my $schema_version = $schema->schema_version || '1.x';
2295 $version ||= $schema_version;
2298 add_drop_table => 1,
2299 ignore_constraint_names => 1,
2300 ignore_index_names => 1,
2304 $self->throw_exception(q{Can't create a ddl file without SQL::Translator 0.09003: '}
2305 . $self->_check_sqlt_message . q{'})
2306 if !$self->_check_sqlt_version;
2308 my $sqlt = SQL::Translator->new( $sqltargs );
2310 $sqlt->parser('SQL::Translator::Parser::DBIx::Class');
2311 my $sqlt_schema = $sqlt->translate({ data => $schema })
2312 or $self->throw_exception ($sqlt->error);
2314 foreach my $db (@$databases) {
2316 $sqlt->{schema} = $sqlt_schema;
2317 $sqlt->producer($db);
2320 my $filename = $schema->ddl_filename($db, $version, $dir);
2321 if (-e $filename && ($version eq $schema_version )) {
2322 # if we are dumping the current version, overwrite the DDL
2323 carp "Overwriting existing DDL file - $filename";
2327 my $output = $sqlt->translate;
2329 carp("Failed to translate to $db, skipping. (" . $sqlt->error . ")");
2332 if(!open($file, ">$filename")) {
2333 $self->throw_exception("Can't open $filename for writing ($!)");
2336 print $file $output;
2339 next unless ($preversion);
2341 require SQL::Translator::Diff;
2343 my $prefilename = $schema->ddl_filename($db, $preversion, $dir);
2344 if(!-e $prefilename) {
2345 carp("No previous schema file found ($prefilename)");
2349 my $difffile = $schema->ddl_filename($db, $version, $dir, $preversion);
2351 carp("Overwriting existing diff file - $difffile");
2357 my $t = SQL::Translator->new($sqltargs);
2362 or $self->throw_exception ($t->error);
2364 my $out = $t->translate( $prefilename )
2365 or $self->throw_exception ($t->error);
2367 $source_schema = $t->schema;
2369 $source_schema->name( $prefilename )
2370 unless ( $source_schema->name );
2373 # The "new" style of producers have sane normalization and can support
2374 # diffing a SQL file against a DBIC->SQLT schema. Old style ones don't
2375 # And we have to diff parsed SQL against parsed SQL.
2376 my $dest_schema = $sqlt_schema;
2378 unless ( "SQL::Translator::Producer::$db"->can('preprocess_schema') ) {
2379 my $t = SQL::Translator->new($sqltargs);
2384 or $self->throw_exception ($t->error);
2386 my $out = $t->translate( $filename )
2387 or $self->throw_exception ($t->error);
2389 $dest_schema = $t->schema;
2391 $dest_schema->name( $filename )
2392 unless $dest_schema->name;
2395 my $diff = SQL::Translator::Diff::schema_diff($source_schema, $db,
2399 if(!open $file, ">$difffile") {
2400 $self->throw_exception("Can't write to $difffile ($!)");
2408 =head2 deployment_statements
2412 =item Arguments: $schema, $type, $version, $directory, $sqlt_args
2416 Returns the statements used by L</deploy> and L<DBIx::Class::Schema/deploy>.
2418 The L<SQL::Translator> (not L<DBI>) database driver name can be explicitly
2419 provided in C<$type>, otherwise the result of L</sqlt_type> is used as default.
2421 C<$directory> is used to return statements from files in a previously created
2422 L</create_ddl_dir> directory and is optional. The filenames are constructed
2423 from L<DBIx::Class::Schema/ddl_filename>, the schema name and the C<$version>.
2425 If no C<$directory> is specified then the statements are constructed on the
2426 fly using L<SQL::Translator> and C<$version> is ignored.
2428 See L<SQL::Translator/METHODS> for a list of values for C<$sqlt_args>.
2432 sub deployment_statements {
2433 my ($self, $schema, $type, $version, $dir, $sqltargs) = @_;
2434 $type ||= $self->sqlt_type;
2435 $version ||= $schema->schema_version || '1.x';
2437 my $filename = $schema->ddl_filename($type, $version, $dir);
2441 open($file, "<$filename")
2442 or $self->throw_exception("Can't open $filename ($!)");
2445 return join('', @rows);
2448 $self->throw_exception(q{Can't deploy without SQL::Translator 0.09003: '}
2449 . $self->_check_sqlt_message . q{'})
2450 if !$self->_check_sqlt_version;
2452 # sources needs to be a parser arg, but for simplicty allow at top level
2454 $sqltargs->{parser_args}{sources} = delete $sqltargs->{sources}
2455 if exists $sqltargs->{sources};
2457 my $tr = SQL::Translator->new(
2458 producer => "SQL::Translator::Producer::${type}",
2460 parser => 'SQL::Translator::Parser::DBIx::Class',
2463 return $tr->translate;
2467 my ($self, $schema, $type, $sqltargs, $dir) = @_;
2470 return if($line =~ /^--/);
2472 # next if($line =~ /^DROP/m);
2473 return if($line =~ /^BEGIN TRANSACTION/m);
2474 return if($line =~ /^COMMIT/m);
2475 return if $line =~ /^\s+$/; # skip whitespace only
2476 $self->_query_start($line);
2478 # do a dbh_do cycle here, as we need some error checking in
2479 # place (even though we will ignore errors)
2480 $self->dbh_do (sub { $_[1]->do($line) });
2483 carp qq{$@ (running "${line}")};
2485 $self->_query_end($line);
2487 my @statements = $self->deployment_statements($schema, $type, undef, $dir, { %{ $sqltargs || {} }, no_comments => 1 } );
2488 if (@statements > 1) {
2489 foreach my $statement (@statements) {
2490 $deploy->( $statement );
2493 elsif (@statements == 1) {
2494 foreach my $line ( split(";\n", $statements[0])) {
2500 =head2 datetime_parser
2502 Returns the datetime parser class
2506 sub datetime_parser {
2508 return $self->{datetime_parser} ||= do {
2509 $self->_populate_dbh unless $self->_dbh;
2510 $self->build_datetime_parser(@_);
2514 =head2 datetime_parser_type
2516 Defines (returns) the datetime parser class - currently hardwired to
2517 L<DateTime::Format::MySQL>
2521 sub datetime_parser_type { "DateTime::Format::MySQL"; }
2523 =head2 build_datetime_parser
2525 See L</datetime_parser>
2529 sub build_datetime_parser {
2531 my $type = $self->datetime_parser_type(@_);
2533 $self->throw_exception("Couldn't load ${type}: $@") if $@;
2538 my $_check_sqlt_version; # private
2539 my $_check_sqlt_message; # private
2540 sub _check_sqlt_version {
2541 return $_check_sqlt_version if defined $_check_sqlt_version;
2542 eval 'use SQL::Translator "0.09003"';
2543 $_check_sqlt_message = $@ || '';
2544 $_check_sqlt_version = !$@;
2547 sub _check_sqlt_message {
2548 _check_sqlt_version if !defined $_check_sqlt_message;
2549 $_check_sqlt_message;
2553 =head2 is_replicating
2555 A boolean that reports if a particular L<DBIx::Class::Storage::DBI> is set to
2556 replicate from a master database. Default is undef, which is the result
2557 returned by databases that don't support replication.
2561 sub is_replicating {
2566 =head2 lag_behind_master
2568 Returns a number that represents a certain amount of lag behind a master db
2569 when a given storage is replicating. The number is database dependent, but
2570 starts at zero and increases with the amount of lag. Default in undef
2574 sub lag_behind_master {
2580 $self->_verify_pid if $self->_dbh;
2582 # some databases need this to stop spewing warnings
2583 if (my $dbh = $self->_dbh) {
2584 eval { $dbh->disconnect };
2594 =head2 DBIx::Class and AutoCommit
2596 DBIx::Class can do some wonderful magic with handling exceptions,
2597 disconnections, and transactions when you use C<< AutoCommit => 1 >>
2598 (the default) combined with C<txn_do> for transaction support.
2600 If you set C<< AutoCommit => 0 >> in your connect info, then you are always
2601 in an assumed transaction between commits, and you're telling us you'd
2602 like to manage that manually. A lot of the magic protections offered by
2603 this module will go away. We can't protect you from exceptions due to database
2604 disconnects because we don't know anything about how to restart your
2605 transactions. You're on your own for handling all sorts of exceptional
2606 cases if you choose the C<< AutoCommit => 0 >> path, just as you would
2612 Matt S. Trout <mst@shadowcatsystems.co.uk>
2614 Andy Grundman <andy@hybridized.org>
2618 You may distribute this code under the same terms as Perl itself.