1 package DBIx::Class::Storage::DBI;
2 # -*- mode: cperl; cperl-indent-level: 2 -*-
4 use base 'DBIx::Class::Storage';
8 use Carp::Clan qw/^DBIx::Class/;
10 use DBIx::Class::SQLAHacks;
11 use DBIx::Class::Storage::DBI::Cursor;
12 use DBIx::Class::Storage::Statistics;
16 __PACKAGE__->mk_group_accessors('simple' =>
17 qw/_connect_info _dbi_connect_info _dbh _sql_maker _sql_maker_opts
18 _conn_pid _conn_tid transaction_depth _dbh_autocommit savepoints/
21 # the values for these accessors are picked out (and deleted) from
22 # the attribute hashref passed to connect_info
23 my @storage_options = qw/
24 on_connect_do on_disconnect_do disable_sth_caching unsafe auto_savepoint
26 __PACKAGE__->mk_group_accessors('simple' => @storage_options);
29 # default cursor class, overridable in connect_info attributes
30 __PACKAGE__->cursor_class('DBIx::Class::Storage::DBI::Cursor');
32 __PACKAGE__->mk_group_accessors('inherited' => qw/sql_maker_class/);
33 __PACKAGE__->sql_maker_class('DBIx::Class::SQLAHacks');
38 DBIx::Class::Storage::DBI - DBI storage handler
42 my $schema = MySchema->connect('dbi:SQLite:my.db');
44 $schema->storage->debug(1);
45 $schema->dbh_do("DROP TABLE authors");
47 $schema->resultset('Book')->search({
48 written_on => $schema->storage->datetime_parser(DateTime->now)
53 This class represents the connection to an RDBMS via L<DBI>. See
54 L<DBIx::Class::Storage> for general information. This pod only
55 documents DBI-specific methods and behaviors.
62 my $new = shift->next::method(@_);
64 $new->transaction_depth(0);
65 $new->_sql_maker_opts({});
66 $new->{savepoints} = [];
67 $new->{_in_dbh_do} = 0;
75 This method is normally called by L<DBIx::Class::Schema/connection>, which
76 encapsulates its argument list in an arrayref before passing them here.
78 The argument list may contain:
84 The same 4-element argument set one would normally pass to
85 L<DBI/connect>, optionally followed by
86 L<extra attributes|/DBIx::Class specific connection attributes>
87 recognized by DBIx::Class:
89 $connect_info_args = [ $dsn, $user, $password, \%dbi_attributes?, \%extra_attributes? ];
93 A single code reference which returns a connected
94 L<DBI database handle|DBI/connect> optionally followed by
95 L<extra attributes|/DBIx::Class specific connection attributes> recognized
98 $connect_info_args = [ sub { DBI->connect (...) }, \%extra_attributes? ];
102 A single hashref with all the attributes and the dsn/user/password
105 $connect_info_args = [{
113 This is particularly useful for L<Catalyst> based applications, allowing the
114 following config (L<Config::General> style):
119 dsn dbi:mysql:database=test
128 Please note that the L<DBI> docs recommend that you always explicitly
129 set C<AutoCommit> to either I<0> or I<1>. L<DBIx::Class> further
130 recommends that it be set to I<1>, and that you perform transactions
131 via our L<DBIx::Class::Schema/txn_do> method. L<DBIx::Class> will set it
132 to I<1> if you do not do explicitly set it to zero. This is the default
133 for most DBDs. See L</DBIx::Class and AutoCommit> for details.
135 =head3 DBIx::Class specific connection attributes
137 In addition to the standard L<DBI|DBI/ATTRIBUTES_COMMON_TO_ALL_HANDLES>
138 L<connection|DBI/Database_Handle_Attributes> attributes, DBIx::Class recognizes
139 the following connection options. These options can be mixed in with your other
140 L<DBI> connection attributes, or placed in a seperate hashref
141 (C<\%extra_attributes>) as shown above.
143 Every time C<connect_info> is invoked, any previous settings for
144 these options will be cleared before setting the new ones, regardless of
145 whether any options are specified in the new C<connect_info>.
152 Specifies things to do immediately after connecting or re-connecting to
153 the database. Its value may contain:
159 This contains one SQL statement to execute.
161 =item an array reference
163 This contains SQL statements to execute in order. Each element contains
164 a string or a code reference that returns a string.
166 =item a code reference
168 This contains some code to execute. Unlike code references within an
169 array reference, its return value is ignored.
173 =item on_disconnect_do
175 Takes arguments in the same form as L</on_connect_do> and executes them
176 immediately before disconnecting from the database.
178 Note, this only runs if you explicitly call L</disconnect> on the
181 =item disable_sth_caching
183 If set to a true value, this option will disable the caching of
184 statement handles via L<DBI/prepare_cached>.
188 Sets the limit dialect. This is useful for JDBC-bridge among others
189 where the remote SQL-dialect cannot be determined by the name of the
190 driver alone. See also L<SQL::Abstract::Limit>.
194 Specifies what characters to use to quote table and column names. If
195 you use this you will want to specify L</name_sep> as well.
197 C<quote_char> expects either a single character, in which case is it
198 is placed on either side of the table/column name, or an arrayref of length
199 2 in which case the table/column name is placed between the elements.
201 For example under MySQL you should use C<< quote_char => '`' >>, and for
202 SQL Server you should use C<< quote_char => [qw/[ ]/] >>.
206 This only needs to be used in conjunction with C<quote_char>, and is used to
207 specify the charecter that seperates elements (schemas, tables, columns) from
208 each other. In most cases this is simply a C<.>.
210 The consequences of not supplying this value is that L<SQL::Abstract>
211 will assume DBIx::Class' uses of aliases to be complete column
212 names. The output will look like I<"me.name"> when it should actually
217 This Storage driver normally installs its own C<HandleError>, sets
218 C<RaiseError> and C<ShowErrorStatement> on, and sets C<PrintError> off on
219 all database handles, including those supplied by a coderef. It does this
220 so that it can have consistent and useful error behavior.
222 If you set this option to a true value, Storage will not do its usual
223 modifications to the database handle's attributes, and instead relies on
224 the settings in your connect_info DBI options (or the values you set in
225 your connection coderef, in the case that you are connecting via coderef).
227 Note that your custom settings can cause Storage to malfunction,
228 especially if you set a C<HandleError> handler that suppresses exceptions
229 and/or disable C<RaiseError>.
233 If this option is true, L<DBIx::Class> will use savepoints when nesting
234 transactions, making it possible to recover from failure in the inner
235 transaction without having to abort all outer transactions.
239 Use this argument to supply a cursor class other than the default
240 L<DBIx::Class::Storage::DBI::Cursor>.
244 Some real-life examples of arguments to L</connect_info> and
245 L<DBIx::Class::Schema/connect>
247 # Simple SQLite connection
248 ->connect_info([ 'dbi:SQLite:./foo.db' ]);
251 ->connect_info([ sub { DBI->connect(...) } ]);
253 # A bit more complicated
260 { quote_char => q{"}, name_sep => q{.} },
264 # Equivalent to the previous example
270 { AutoCommit => 1, quote_char => q{"}, name_sep => q{.} },
274 # Same, but with hashref as argument
275 # See parse_connect_info for explanation
278 dsn => 'dbi:Pg:dbname=foo',
280 password => 'my_pg_password',
287 # Subref + DBIx::Class-specific connection options
290 sub { DBI->connect(...) },
294 on_connect_do => ['SET search_path TO myschema,otherschema,public'],
295 disable_sth_caching => 1,
305 my ($self, $info_arg) = @_;
307 return $self->_connect_info if !$info_arg;
309 my @args = @$info_arg; # take a shallow copy for further mutilation
310 $self->_connect_info([@args]); # copy for _connect_info
313 # combine/pre-parse arguments depending on invocation style
316 if (ref $args[0] eq 'CODE') { # coderef with optional \%extra_attributes
317 %attrs = %{ $args[1] || {} };
320 elsif (ref $args[0] eq 'HASH') { # single hashref (i.e. Catalyst config)
321 %attrs = %{$args[0]};
323 for (qw/password user dsn/) {
324 unshift @args, delete $attrs{$_};
327 else { # otherwise assume dsn/user/password + \%attrs + \%extra_attrs
329 % { $args[3] || {} },
330 % { $args[4] || {} },
332 @args = @args[0,1,2];
335 # Kill sql_maker/_sql_maker_opts, so we get a fresh one with only
336 # the new set of options
337 $self->_sql_maker(undef);
338 $self->_sql_maker_opts({});
341 for my $storage_opt (@storage_options, 'cursor_class') { # @storage_options is declared at the top of the module
342 if(my $value = delete $attrs{$storage_opt}) {
343 $self->$storage_opt($value);
346 for my $sql_maker_opt (qw/limit_dialect quote_char name_sep/) {
347 if(my $opt_val = delete $attrs{$sql_maker_opt}) {
348 $self->_sql_maker_opts->{$sql_maker_opt} = $opt_val;
353 %attrs = () if (ref $args[0] eq 'CODE'); # _connect() never looks past $args[0] in this case
355 $self->_dbi_connect_info([@args, keys %attrs ? \%attrs : ()]);
356 $self->_connect_info;
361 This method is deprecated in favour of setting via L</connect_info>.
366 Arguments: ($subref | $method_name), @extra_coderef_args?
368 Execute the given $subref or $method_name using the new exception-based
369 connection management.
371 The first two arguments will be the storage object that C<dbh_do> was called
372 on and a database handle to use. Any additional arguments will be passed
373 verbatim to the called subref as arguments 2 and onwards.
375 Using this (instead of $self->_dbh or $self->dbh) ensures correct
376 exception handling and reconnection (or failover in future subclasses).
378 Your subref should have no side-effects outside of the database, as
379 there is the potential for your subref to be partially double-executed
380 if the database connection was stale/dysfunctional.
384 my @stuff = $schema->storage->dbh_do(
386 my ($storage, $dbh, @cols) = @_;
387 my $cols = join(q{, }, @cols);
388 $dbh->selectrow_array("SELECT $cols FROM foo");
399 my $dbh = $self->_dbh;
401 return $self->$code($dbh, @_) if $self->{_in_dbh_do}
402 || $self->{transaction_depth};
404 local $self->{_in_dbh_do} = 1;
407 my $want_array = wantarray;
410 $self->_verify_pid if $dbh;
412 $self->_populate_dbh;
417 @result = $self->$code($dbh, @_);
419 elsif(defined $want_array) {
420 $result[0] = $self->$code($dbh, @_);
423 $self->$code($dbh, @_);
428 if(!$exception) { return $want_array ? @result : $result[0] }
430 $self->throw_exception($exception) if $self->connected;
432 # We were not connected - reconnect and retry, but let any
433 # exception fall right through this time
434 $self->_populate_dbh;
435 $self->$code($self->_dbh, @_);
438 # This is basically a blend of dbh_do above and DBIx::Class::Storage::txn_do.
439 # It also informs dbh_do to bypass itself while under the direction of txn_do,
440 # via $self->{_in_dbh_do} (this saves some redundant eval and errorcheck, etc)
445 ref $coderef eq 'CODE' or $self->throw_exception
446 ('$coderef must be a CODE reference');
448 return $coderef->(@_) if $self->{transaction_depth} && ! $self->auto_savepoint;
450 local $self->{_in_dbh_do} = 1;
453 my $want_array = wantarray;
458 $self->_verify_pid if $self->_dbh;
459 $self->_populate_dbh if !$self->_dbh;
463 @result = $coderef->(@_);
465 elsif(defined $want_array) {
466 $result[0] = $coderef->(@_);
475 if(!$exception) { return $want_array ? @result : $result[0] }
477 if($tried++ > 0 || $self->connected) {
478 eval { $self->txn_rollback };
479 my $rollback_exception = $@;
480 if($rollback_exception) {
481 my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
482 $self->throw_exception($exception) # propagate nested rollback
483 if $rollback_exception =~ /$exception_class/;
485 $self->throw_exception(
486 "Transaction aborted: ${exception}. "
487 . "Rollback failed: ${rollback_exception}"
490 $self->throw_exception($exception)
493 # We were not connected, and was first try - reconnect and retry
495 $self->_populate_dbh;
501 Our C<disconnect> method also performs a rollback first if the
502 database is not in C<AutoCommit> mode.
509 if( $self->connected ) {
510 my $connection_do = $self->on_disconnect_do;
511 $self->_do_connection_actions($connection_do) if ref($connection_do);
513 $self->_dbh->rollback unless $self->_dbh_autocommit;
514 $self->_dbh->disconnect;
520 =head2 with_deferred_fk_checks
524 =item Arguments: C<$coderef>
526 =item Return Value: The return value of $coderef
530 Storage specific method to run the code ref with FK checks deferred or
531 in MySQL's case disabled entirely.
535 # Storage subclasses should override this
536 sub with_deferred_fk_checks {
537 my ($self, $sub) = @_;
545 if(my $dbh = $self->_dbh) {
546 if(defined $self->_conn_tid && $self->_conn_tid != threads->tid) {
553 return 0 if !$self->_dbh;
555 return ($dbh->FETCH('Active') && $dbh->ping);
561 # handle pid changes correctly
562 # NOTE: assumes $self->_dbh is a valid $dbh
566 return if defined $self->_conn_pid && $self->_conn_pid == $$;
568 $self->_dbh->{InactiveDestroy} = 1;
575 sub ensure_connected {
578 unless ($self->connected) {
579 $self->_populate_dbh;
585 Returns the dbh - a data base handle of class L<DBI>.
592 $self->ensure_connected;
596 sub _sql_maker_args {
599 return ( bindtype=>'columns', array_datatypes => 1, limit_dialect => $self->dbh, %{$self->_sql_maker_opts} );
604 unless ($self->_sql_maker) {
605 my $sql_maker_class = $self->sql_maker_class;
606 $self->_sql_maker($sql_maker_class->new( $self->_sql_maker_args ));
608 return $self->_sql_maker;
615 my @info = @{$self->_dbi_connect_info || []};
616 $self->_dbh($self->_connect(@info));
618 $self->_conn_pid($$);
619 $self->_conn_tid(threads->tid) if $INC{'threads.pm'};
621 $self->_determine_driver;
623 # Always set the transaction depth on connect, since
624 # there is no transaction in progress by definition
625 $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
627 my $connection_do = $self->on_connect_do;
628 $self->_do_connection_actions($connection_do) if $connection_do;
631 sub _determine_driver {
634 if (ref $self eq 'DBIx::Class::Storage::DBI') {
637 if ($self->_dbh) { # we are connected
638 $driver = $self->_dbh->{Driver}{Name};
640 # try to use dsn to not require being connected, the driver may still
641 # force a connection in _rebless to determine version
642 ($driver) = $self->_dbi_connect_info->[0] =~ /dbi:([^:]+):/i;
645 if ($self->load_optional_class("DBIx::Class::Storage::DBI::${driver}")) {
646 bless $self, "DBIx::Class::Storage::DBI::${driver}";
652 sub _do_connection_actions {
654 my $connection_do = shift;
656 if (!ref $connection_do) {
657 $self->_do_query($connection_do);
659 elsif (ref $connection_do eq 'ARRAY') {
660 $self->_do_query($_) foreach @$connection_do;
662 elsif (ref $connection_do eq 'CODE') {
663 $connection_do->($self);
666 $self->throw_exception (sprintf ("Don't know how to process conection actions of type '%s'", ref $connection_do) );
673 my ($self, $action) = @_;
675 if (ref $action eq 'CODE') {
676 $action = $action->($self);
677 $self->_do_query($_) foreach @$action;
680 # Most debuggers expect ($sql, @bind), so we need to exclude
681 # the attribute hash which is the second argument to $dbh->do
682 # furthermore the bind values are usually to be presented
683 # as named arrayref pairs, so wrap those here too
684 my @do_args = (ref $action eq 'ARRAY') ? (@$action) : ($action);
685 my $sql = shift @do_args;
686 my $attrs = shift @do_args;
687 my @bind = map { [ undef, $_ ] } @do_args;
689 $self->_query_start($sql, @bind);
690 $self->_dbh->do($sql, $attrs, @do_args);
691 $self->_query_end($sql, @bind);
698 my ($self, @info) = @_;
700 $self->throw_exception("You failed to provide any connection info")
703 my ($old_connect_via, $dbh);
705 if ($INC{'Apache/DBI.pm'} && $ENV{MOD_PERL}) {
706 $old_connect_via = $DBI::connect_via;
707 $DBI::connect_via = 'connect';
711 if(ref $info[0] eq 'CODE') {
715 $dbh = DBI->connect(@info);
718 if($dbh && !$self->unsafe) {
719 my $weak_self = $self;
720 Scalar::Util::weaken($weak_self);
721 $dbh->{HandleError} = sub {
723 $weak_self->throw_exception("DBI Exception: $_[0]");
726 croak ("DBI Exception: $_[0]");
729 $dbh->{ShowErrorStatement} = 1;
730 $dbh->{RaiseError} = 1;
731 $dbh->{PrintError} = 0;
735 $DBI::connect_via = $old_connect_via if $old_connect_via;
737 $self->throw_exception("DBI Connection failed: " . ($@||$DBI::errstr))
740 $self->_dbh_autocommit($dbh->{AutoCommit});
746 my ($self, $name) = @_;
748 $name = $self->_svp_generate_name
749 unless defined $name;
751 $self->throw_exception ("You can't use savepoints outside a transaction")
752 if $self->{transaction_depth} == 0;
754 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
755 unless $self->can('_svp_begin');
757 push @{ $self->{savepoints} }, $name;
759 $self->debugobj->svp_begin($name) if $self->debug;
761 return $self->_svp_begin($name);
765 my ($self, $name) = @_;
767 $self->throw_exception ("You can't use savepoints outside a transaction")
768 if $self->{transaction_depth} == 0;
770 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
771 unless $self->can('_svp_release');
774 $self->throw_exception ("Savepoint '$name' does not exist")
775 unless grep { $_ eq $name } @{ $self->{savepoints} };
777 # Dig through the stack until we find the one we are releasing. This keeps
778 # the stack up to date.
781 do { $svp = pop @{ $self->{savepoints} } } while $svp ne $name;
783 $name = pop @{ $self->{savepoints} };
786 $self->debugobj->svp_release($name) if $self->debug;
788 return $self->_svp_release($name);
792 my ($self, $name) = @_;
794 $self->throw_exception ("You can't use savepoints outside a transaction")
795 if $self->{transaction_depth} == 0;
797 $self->throw_exception ("Your Storage implementation doesn't support savepoints")
798 unless $self->can('_svp_rollback');
801 # If they passed us a name, verify that it exists in the stack
802 unless(grep({ $_ eq $name } @{ $self->{savepoints} })) {
803 $self->throw_exception("Savepoint '$name' does not exist!");
806 # Dig through the stack until we find the one we are releasing. This keeps
807 # the stack up to date.
808 while(my $s = pop(@{ $self->{savepoints} })) {
809 last if($s eq $name);
811 # Add the savepoint back to the stack, as a rollback doesn't remove the
812 # named savepoint, only everything after it.
813 push(@{ $self->{savepoints} }, $name);
815 # We'll assume they want to rollback to the last savepoint
816 $name = $self->{savepoints}->[-1];
819 $self->debugobj->svp_rollback($name) if $self->debug;
821 return $self->_svp_rollback($name);
824 sub _svp_generate_name {
827 return 'savepoint_'.scalar(@{ $self->{'savepoints'} });
832 $self->ensure_connected();
833 if($self->{transaction_depth} == 0) {
834 $self->debugobj->txn_begin()
836 # this isn't ->_dbh-> because
837 # we should reconnect on begin_work
838 # for AutoCommit users
839 $self->dbh->begin_work;
840 } elsif ($self->auto_savepoint) {
843 $self->{transaction_depth}++;
848 if ($self->{transaction_depth} == 1) {
849 my $dbh = $self->_dbh;
850 $self->debugobj->txn_commit()
853 $self->{transaction_depth} = 0
854 if $self->_dbh_autocommit;
856 elsif($self->{transaction_depth} > 1) {
857 $self->{transaction_depth}--;
859 if $self->auto_savepoint;
865 my $dbh = $self->_dbh;
867 if ($self->{transaction_depth} == 1) {
868 $self->debugobj->txn_rollback()
870 $self->{transaction_depth} = 0
871 if $self->_dbh_autocommit;
874 elsif($self->{transaction_depth} > 1) {
875 $self->{transaction_depth}--;
876 if ($self->auto_savepoint) {
882 die DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION->new;
887 my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
888 $error =~ /$exception_class/ and $self->throw_exception($error);
889 # ensure that a failed rollback resets the transaction depth
890 $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
891 $self->throw_exception($error);
895 # This used to be the top-half of _execute. It was split out to make it
896 # easier to override in NoBindVars without duping the rest. It takes up
897 # all of _execute's args, and emits $sql, @bind.
898 sub _prep_for_execute {
899 my ($self, $op, $extra_bind, $ident, $args) = @_;
901 if( Scalar::Util::blessed($ident) && $ident->isa("DBIx::Class::ResultSource") ) {
902 $ident = $ident->from();
905 my ($sql, @bind) = $self->sql_maker->$op($ident, @$args);
908 map { ref $_ eq 'ARRAY' ? $_ : [ '!!dummy', $_ ] } @$extra_bind)
910 return ($sql, \@bind);
913 sub _fix_bind_params {
914 my ($self, @bind) = @_;
916 ### Turn @bind from something like this:
917 ### ( [ "artist", 1 ], [ "cdid", 1, 3 ] )
919 ### ( "'1'", "'1'", "'3'" )
922 if ( defined( $_ && $_->[1] ) ) {
923 map { qq{'$_'}; } @{$_}[ 1 .. $#$_ ];
930 my ( $self, $sql, @bind ) = @_;
932 if ( $self->debug ) {
933 @bind = $self->_fix_bind_params(@bind);
935 $self->debugobj->query_start( $sql, @bind );
940 my ( $self, $sql, @bind ) = @_;
942 if ( $self->debug ) {
943 @bind = $self->_fix_bind_params(@bind);
944 $self->debugobj->query_end( $sql, @bind );
949 my ($self, $dbh, $op, $extra_bind, $ident, $bind_attributes, @args) = @_;
951 my ($sql, $bind) = $self->_prep_for_execute($op, $extra_bind, $ident, \@args);
953 $self->_query_start( $sql, @$bind );
955 my $sth = $self->sth($sql,$op);
957 my $placeholder_index = 1;
959 foreach my $bound (@$bind) {
961 my($column_name, @data) = @$bound;
963 if ($bind_attributes) {
964 $attributes = $bind_attributes->{$column_name}
965 if defined $bind_attributes->{$column_name};
968 foreach my $data (@data) {
970 $data = $ref && $ref ne 'ARRAY' ? ''.$data : $data; # stringify args (except arrayrefs)
972 $sth->bind_param($placeholder_index, $data, $attributes);
973 $placeholder_index++;
977 # Can this fail without throwing an exception anyways???
978 my $rv = $sth->execute();
979 $self->throw_exception($sth->errstr) if !$rv;
981 $self->_query_end( $sql, @$bind );
983 return (wantarray ? ($rv, $sth, @$bind) : $rv);
988 $self->dbh_do('_dbh_execute', @_)
992 my ($self, $source, $to_insert) = @_;
994 my $ident = $source->from;
995 my $bind_attributes = $self->source_bind_attributes($source);
997 my $updated_cols = {};
999 $self->ensure_connected;
1000 foreach my $col ( $source->columns ) {
1001 if ( !defined $to_insert->{$col} ) {
1002 my $col_info = $source->column_info($col);
1004 if ( $col_info->{auto_nextval} ) {
1005 $updated_cols->{$col} = $to_insert->{$col} = $self->_sequence_fetch( 'nextval', $col_info->{sequence} || $self->_dbh_get_autoinc_seq($self->dbh, $source) );
1010 $self->_execute('insert' => [], $source, $bind_attributes, $to_insert);
1012 return $updated_cols;
1015 ## Still not quite perfect, and EXPERIMENTAL
1016 ## Currently it is assumed that all values passed will be "normal", i.e. not
1017 ## scalar refs, or at least, all the same type as the first set, the statement is
1018 ## only prepped once.
1020 my ($self, $source, $cols, $data) = @_;
1022 my $table = $source->from;
1023 @colvalues{@$cols} = (0..$#$cols);
1024 my ($sql, @bind) = $self->sql_maker->insert($table, \%colvalues);
1026 $self->_query_start( $sql, @bind );
1027 my $sth = $self->sth($sql);
1029 # @bind = map { ref $_ ? ''.$_ : $_ } @bind; # stringify args
1031 ## This must be an arrayref, else nothing works!
1032 my $tuple_status = [];
1034 ## Get the bind_attributes, if any exist
1035 my $bind_attributes = $self->source_bind_attributes($source);
1037 ## Bind the values and execute
1038 my $placeholder_index = 1;
1040 foreach my $bound (@bind) {
1042 my $attributes = {};
1043 my ($column_name, $data_index) = @$bound;
1045 if( $bind_attributes ) {
1046 $attributes = $bind_attributes->{$column_name}
1047 if defined $bind_attributes->{$column_name};
1050 my @data = map { $_->[$data_index] } @$data;
1052 $sth->bind_param_array( $placeholder_index, [@data], $attributes );
1053 $placeholder_index++;
1055 my $rv = eval { $sth->execute_array({ArrayTupleStatus => $tuple_status}) };
1058 ++$i while $i <= $#$tuple_status && !ref $tuple_status->[$i];
1060 $self->throw_exception($sth->errstr || "Unexpected populate error: $err")
1061 if ($i > $#$tuple_status);
1063 require Data::Dumper;
1064 local $Data::Dumper::Terse = 1;
1065 local $Data::Dumper::Indent = 1;
1066 local $Data::Dumper::Useqq = 1;
1067 local $Data::Dumper::Quotekeys = 0;
1069 $self->throw_exception(sprintf "%s for populate slice:\n%s",
1070 $tuple_status->[$i][1],
1071 Data::Dumper::Dumper(
1072 { map { $cols->[$_] => $data->[$i][$_] } (0 .. $#$cols) }
1076 $self->throw_exception($sth->errstr) if !$rv;
1078 $self->_query_end( $sql, @bind );
1079 return (wantarray ? ($rv, $sth, @bind) : $rv);
1083 my $self = shift @_;
1084 my $source = shift @_;
1085 my $bind_attributes = $self->source_bind_attributes($source);
1087 return $self->_execute('update' => [], $source, $bind_attributes, @_);
1092 my $self = shift @_;
1093 my $source = shift @_;
1095 my $bind_attrs = $self->source_bind_attributes($source);
1097 return $self->_execute('delete' => [], $source, $bind_attrs, @_);
1100 # We were sent here because the $rs contains a complex search
1101 # which will require a subquery to select the correct rows
1102 # (i.e. joined or limited resultsets)
1104 # Genarating a single PK column subquery is trivial and supported
1105 # by all RDBMS. However if we have a multicolumn PK, things get ugly.
1106 # Look at _multipk_update_delete()
1107 sub subq_update_delete {
1109 my ($rs, $op, $values) = @_;
1111 my $rsrc = $rs->result_source;
1113 # we already check this, but double check naively just in case. Should be removed soon
1114 my $sel = $rs->_resolved_attrs->{select};
1115 $sel = [ $sel ] unless ref $sel eq 'ARRAY';
1116 my @pcols = $rsrc->primary_columns;
1117 if (@$sel != @pcols) {
1118 $self->throw_exception (
1119 'Subquery update/delete can not be called on resultsets selecting a'
1120 .' number of columns different than the number of primary keys'
1127 $op eq 'update' ? $values : (),
1128 { $pcols[0] => { -in => $rs->as_query } },
1133 return $self->_multipk_update_delete (@_);
1137 # ANSI SQL does not provide a reliable way to perform a multicol-PK
1138 # resultset update/delete involving subqueries. So by default resort
1139 # to simple (and inefficient) delete_all style per-row opearations,
1140 # while allowing specific storages to override this with a faster
1143 sub _multipk_update_delete {
1144 return shift->_per_row_update_delete (@_);
1147 # This is the default loop used to delete/update rows for multi PK
1148 # resultsets, and used by mysql exclusively (because it can't do anything
1151 # We do not use $row->$op style queries, because resultset update/delete
1152 # is not expected to cascade (this is what delete_all/update_all is for).
1154 # There should be no race conditions as the entire operation is rolled
1157 sub _per_row_update_delete {
1159 my ($rs, $op, $values) = @_;
1161 my $rsrc = $rs->result_source;
1162 my @pcols = $rsrc->primary_columns;
1164 my $guard = $self->txn_scope_guard;
1166 # emulate the return value of $sth->execute for non-selects
1167 my $row_cnt = '0E0';
1169 my $subrs_cur = $rs->cursor;
1170 while (my @pks = $subrs_cur->next) {
1173 for my $i (0.. $#pcols) {
1174 $cond->{$pcols[$i]} = $pks[$i];
1179 $op eq 'update' ? $values : (),
1193 my $sql_maker = $self->sql_maker;
1194 local $sql_maker->{for};
1195 return $self->_execute($self->_select_args(@_));
1199 my ($self, $ident, $select, $condition, $attrs) = @_;
1200 my $order = $attrs->{order_by};
1202 my $for = delete $attrs->{for};
1203 my $sql_maker = $self->sql_maker;
1204 $sql_maker->{for} = $for;
1206 my @in_order_attrs = qw/group_by having _virtual_order_by/;
1207 if (List::Util::first { exists $attrs->{$_} } (@in_order_attrs) ) {
1210 ? (order_by => $order)
1213 ( map { $_ => $attrs->{$_} } (@in_order_attrs) )
1217 # the reason this is so contrived is because we have several tables in
1218 # from, each with its own set of bindattrs
1220 if ( Scalar::Util::blessed($ident) && $ident->isa("DBIx::Class::ResultSource") ) {
1221 $alias2source->{$ident->alias} = $ident;
1223 elsif (ref $ident eq 'ARRAY') {
1227 if (ref $_ eq 'HASH') {
1230 if (ref $_ eq 'ARRAY' and ref $_->[0] eq 'HASH') {
1234 $alias2source->{$tabinfo->{-alias}} = $tabinfo->{-result_source}
1235 if ($tabinfo->{-result_source});
1239 my $bind_attrs = {};
1240 for my $alias (keys %$alias2source) {
1241 my $bindtypes = $self->source_bind_attributes ($alias2source->{$alias}) || {};
1242 for my $col (keys %$bindtypes) {
1244 my $fqcn = join ('.', $alias, $col);
1245 $bind_attrs->{$fqcn} = $bindtypes->{$col} if $bindtypes->{$col};
1247 # so that unqualified searches can be bound too
1248 $bind_attrs->{$col} = $bind_attrs->{$fqcn} if $alias eq 'me';
1252 # This would be the point to deflate anything found in $condition
1253 # (and leave $attrs->{bind} intact). Problem is - inflators historically
1254 # expect a row object. And all we have is a resultsource (it is trivial
1255 # to extract deflator coderefs via $alias2source above).
1257 # I don't see a way forward other than changing the way deflators are
1258 # invoked, and that's just bad...
1260 my @args = ('select', $attrs->{bind}, $ident, $bind_attrs, $select, $condition, $order);
1261 if ($attrs->{software_limit} ||
1262 $sql_maker->_default_limit_syntax eq "GenericSubQ") {
1263 $attrs->{software_limit} = 1;
1265 $self->throw_exception("rows attribute must be positive if present")
1266 if (defined($attrs->{rows}) && !($attrs->{rows} > 0));
1268 # MySQL actually recommends this approach. I cringe.
1269 $attrs->{rows} = 2**48 if not defined $attrs->{rows} and defined $attrs->{offset};
1270 push @args, $attrs->{rows}, $attrs->{offset};
1276 my ($self, $source, $attrs) = @_;
1278 my $tmp_attrs = { %$attrs };
1280 # take off any column specs, any pagers, record_filter is cdbi, and no point of ordering a count
1281 delete $tmp_attrs->{$_} for (qw/columns +columns select +select as +as rows offset page pager order_by record_filter/);
1283 $tmp_attrs->{select} = { count => '*' };
1284 $tmp_attrs->{as} = [qw/count/];
1286 my $tmp_rs = $source->resultset_class->new($source, $tmp_attrs);
1287 my ($count) = $tmp_rs->cursor->next;
1289 # if the offset/rows attributes are still present, we did not use
1290 # a subquery, so we need to make the calculations in software
1291 $count -= $attrs->{offset} if $attrs->{offset};
1292 $count = $attrs->{rows} if $attrs->{rows} and $attrs->{rows} < $count;
1293 $count = 0 if ($count < 0);
1299 my ($self, $source, $attrs) = @_;
1301 # copy for the subquery, we need to do some adjustments to it too
1302 my $sub_attrs = { %$attrs };
1304 # these can not go in the subquery, and there is no point of ordering it
1305 delete $sub_attrs->{$_} for qw/prefetch collapse select +select as +as columns +columns order_by/;
1307 # if we prefetch, we group_by primary keys only as this is what we would get out of the rs via ->next/->all
1308 # simply deleting group_by suffices, as the code below will re-fill it
1309 # Note: we check $attrs, as $sub_attrs has collapse deleted
1310 if (ref $attrs->{collapse} and keys %{$attrs->{collapse}} ) {
1311 delete $sub_attrs->{group_by};
1314 $sub_attrs->{columns} = $sub_attrs->{group_by} ||= [ map { "$attrs->{alias}.$_" } ($source->primary_columns) ];
1317 count_subq => $source->resultset_class->new ($source, $sub_attrs )->as_query
1320 # the subquery replaces this
1321 delete $attrs->{$_} for qw/where bind prefetch collapse distinct group_by having having_bind rows offset page pager/;
1323 return $self->count ($source, $attrs);
1326 sub source_bind_attributes {
1327 my ($self, $source) = @_;
1329 my $bind_attributes;
1330 foreach my $column ($source->columns) {
1332 my $data_type = $source->column_info($column)->{data_type} || '';
1333 $bind_attributes->{$column} = $self->bind_attribute_by_data_type($data_type)
1337 return $bind_attributes;
1344 =item Arguments: $ident, $select, $condition, $attrs
1348 Handle a SQL select statement.
1354 my ($ident, $select, $condition, $attrs) = @_;
1355 return $self->cursor_class->new($self, \@_, $attrs);
1360 my ($rv, $sth, @bind) = $self->_select(@_);
1361 my @row = $sth->fetchrow_array;
1362 my @nextrow = $sth->fetchrow_array if @row;
1363 if(@row && @nextrow) {
1364 carp "Query returned more than one row. SQL that returns multiple rows is DEPRECATED for ->find and ->single";
1366 # Need to call finish() to work round broken DBDs
1375 =item Arguments: $sql
1379 Returns a L<DBI> sth (statement handle) for the supplied SQL.
1384 my ($self, $dbh, $sql) = @_;
1386 # 3 is the if_active parameter which avoids active sth re-use
1387 my $sth = $self->disable_sth_caching
1388 ? $dbh->prepare($sql)
1389 : $dbh->prepare_cached($sql, {}, 3);
1391 # XXX You would think RaiseError would make this impossible,
1392 # but apparently that's not true :(
1393 $self->throw_exception($dbh->errstr) if !$sth;
1399 my ($self, $sql) = @_;
1400 $self->dbh_do('_dbh_sth', $sql);
1403 sub _dbh_columns_info_for {
1404 my ($self, $dbh, $table) = @_;
1406 if ($dbh->can('column_info')) {
1409 my ($schema,$tab) = $table =~ /^(.+?)\.(.+)$/ ? ($1,$2) : (undef,$table);
1410 my $sth = $dbh->column_info( undef,$schema, $tab, '%' );
1412 while ( my $info = $sth->fetchrow_hashref() ){
1414 $column_info{data_type} = $info->{TYPE_NAME};
1415 $column_info{size} = $info->{COLUMN_SIZE};
1416 $column_info{is_nullable} = $info->{NULLABLE} ? 1 : 0;
1417 $column_info{default_value} = $info->{COLUMN_DEF};
1418 my $col_name = $info->{COLUMN_NAME};
1419 $col_name =~ s/^\"(.*)\"$/$1/;
1421 $result{$col_name} = \%column_info;
1424 return \%result if !$@ && scalar keys %result;
1428 my $sth = $dbh->prepare($self->sql_maker->select($table, undef, \'1 = 0'));
1430 my @columns = @{$sth->{NAME_lc}};
1431 for my $i ( 0 .. $#columns ){
1433 $column_info{data_type} = $sth->{TYPE}->[$i];
1434 $column_info{size} = $sth->{PRECISION}->[$i];
1435 $column_info{is_nullable} = $sth->{NULLABLE}->[$i] ? 1 : 0;
1437 if ($column_info{data_type} =~ m/^(.*?)\((.*?)\)$/) {
1438 $column_info{data_type} = $1;
1439 $column_info{size} = $2;
1442 $result{$columns[$i]} = \%column_info;
1446 foreach my $col (keys %result) {
1447 my $colinfo = $result{$col};
1448 my $type_num = $colinfo->{data_type};
1450 if(defined $type_num && $dbh->can('type_info')) {
1451 my $type_info = $dbh->type_info($type_num);
1452 $type_name = $type_info->{TYPE_NAME} if $type_info;
1453 $colinfo->{data_type} = $type_name if $type_name;
1460 sub columns_info_for {
1461 my ($self, $table) = @_;
1462 $self->dbh_do('_dbh_columns_info_for', $table);
1465 =head2 last_insert_id
1467 Return the row id of the last insert.
1471 sub _dbh_last_insert_id {
1472 # All Storage's need to register their own _dbh_last_insert_id
1473 # the old SQLite-based method was highly inappropriate
1476 my $class = ref $self;
1477 $self->throw_exception (<<EOE);
1479 No _dbh_last_insert_id() method found in $class.
1480 Since the method of obtaining the autoincrement id of the last insert
1481 operation varies greatly between different databases, this method must be
1482 individually implemented for every storage class.
1486 sub last_insert_id {
1488 $self->dbh_do('_dbh_last_insert_id', @_);
1493 Returns the database driver name.
1497 sub sqlt_type { shift->dbh->{Driver}->{Name} }
1499 =head2 bind_attribute_by_data_type
1501 Given a datatype from column info, returns a database specific bind
1502 attribute for C<< $dbh->bind_param($val,$attribute) >> or nothing if we will
1503 let the database planner just handle it.
1505 Generally only needed for special case column types, like bytea in postgres.
1509 sub bind_attribute_by_data_type {
1513 =head2 create_ddl_dir (EXPERIMENTAL)
1517 =item Arguments: $schema \@databases, $version, $directory, $preversion, \%sqlt_args
1521 Creates a SQL file based on the Schema, for each of the specified
1522 database engines in C<\@databases> in the given directory.
1523 (note: specify L<SQL::Translator> names, not L<DBI> driver names).
1525 Given a previous version number, this will also create a file containing
1526 the ALTER TABLE statements to transform the previous schema into the
1527 current one. Note that these statements may contain C<DROP TABLE> or
1528 C<DROP COLUMN> statements that can potentially destroy data.
1530 The file names are created using the C<ddl_filename> method below, please
1531 override this method in your schema if you would like a different file
1532 name format. For the ALTER file, the same format is used, replacing
1533 $version in the name with "$preversion-$version".
1535 See L<SQL::Translator/METHODS> for a list of values for C<\%sqlt_args>.
1536 The most common value for this would be C<< { add_drop_table => 1 } >>
1537 to have the SQL produced include a C<DROP TABLE> statement for each table
1538 created. For quoting purposes supply C<quote_table_names> and
1539 C<quote_field_names>.
1541 If no arguments are passed, then the following default values are assumed:
1545 =item databases - ['MySQL', 'SQLite', 'PostgreSQL']
1547 =item version - $schema->schema_version
1549 =item directory - './'
1551 =item preversion - <none>
1555 By default, C<\%sqlt_args> will have
1557 { add_drop_table => 1, ignore_constraint_names => 1, ignore_index_names => 1 }
1559 merged with the hash passed in. To disable any of those features, pass in a
1560 hashref like the following
1562 { ignore_constraint_names => 0, # ... other options }
1565 Note that this feature is currently EXPERIMENTAL and may not work correctly
1566 across all databases, or fully handle complex relationships.
1568 WARNING: Please check all SQL files created, before applying them.
1572 sub create_ddl_dir {
1573 my ($self, $schema, $databases, $version, $dir, $preversion, $sqltargs) = @_;
1575 if(!$dir || !-d $dir) {
1576 carp "No directory given, using ./\n";
1579 $databases ||= ['MySQL', 'SQLite', 'PostgreSQL'];
1580 $databases = [ $databases ] if(ref($databases) ne 'ARRAY');
1582 my $schema_version = $schema->schema_version || '1.x';
1583 $version ||= $schema_version;
1586 add_drop_table => 1,
1587 ignore_constraint_names => 1,
1588 ignore_index_names => 1,
1592 $self->throw_exception(q{Can't create a ddl file without SQL::Translator 0.09003: '}
1593 . $self->_check_sqlt_message . q{'})
1594 if !$self->_check_sqlt_version;
1596 my $sqlt = SQL::Translator->new( $sqltargs );
1598 $sqlt->parser('SQL::Translator::Parser::DBIx::Class');
1599 my $sqlt_schema = $sqlt->translate({ data => $schema })
1600 or $self->throw_exception ($sqlt->error);
1602 foreach my $db (@$databases) {
1604 $sqlt->{schema} = $sqlt_schema;
1605 $sqlt->producer($db);
1608 my $filename = $schema->ddl_filename($db, $version, $dir);
1609 if (-e $filename && ($version eq $schema_version )) {
1610 # if we are dumping the current version, overwrite the DDL
1611 carp "Overwriting existing DDL file - $filename";
1615 my $output = $sqlt->translate;
1617 carp("Failed to translate to $db, skipping. (" . $sqlt->error . ")");
1620 if(!open($file, ">$filename")) {
1621 $self->throw_exception("Can't open $filename for writing ($!)");
1624 print $file $output;
1627 next unless ($preversion);
1629 require SQL::Translator::Diff;
1631 my $prefilename = $schema->ddl_filename($db, $preversion, $dir);
1632 if(!-e $prefilename) {
1633 carp("No previous schema file found ($prefilename)");
1637 my $difffile = $schema->ddl_filename($db, $version, $dir, $preversion);
1639 carp("Overwriting existing diff file - $difffile");
1645 my $t = SQL::Translator->new($sqltargs);
1650 or $self->throw_exception ($t->error);
1652 my $out = $t->translate( $prefilename )
1653 or $self->throw_exception ($t->error);
1655 $source_schema = $t->schema;
1657 $source_schema->name( $prefilename )
1658 unless ( $source_schema->name );
1661 # The "new" style of producers have sane normalization and can support
1662 # diffing a SQL file against a DBIC->SQLT schema. Old style ones don't
1663 # And we have to diff parsed SQL against parsed SQL.
1664 my $dest_schema = $sqlt_schema;
1666 unless ( "SQL::Translator::Producer::$db"->can('preprocess_schema') ) {
1667 my $t = SQL::Translator->new($sqltargs);
1672 or $self->throw_exception ($t->error);
1674 my $out = $t->translate( $filename )
1675 or $self->throw_exception ($t->error);
1677 $dest_schema = $t->schema;
1679 $dest_schema->name( $filename )
1680 unless $dest_schema->name;
1683 my $diff = SQL::Translator::Diff::schema_diff($source_schema, $db,
1687 if(!open $file, ">$difffile") {
1688 $self->throw_exception("Can't write to $difffile ($!)");
1696 =head2 deployment_statements
1700 =item Arguments: $schema, $type, $version, $directory, $sqlt_args
1704 Returns the statements used by L</deploy> and L<DBIx::Class::Schema/deploy>.
1706 The L<SQL::Translator> (not L<DBI>) database driver name can be explicitly
1707 provided in C<$type>, otherwise the result of L</sqlt_type> is used as default.
1709 C<$directory> is used to return statements from files in a previously created
1710 L</create_ddl_dir> directory and is optional. The filenames are constructed
1711 from L<DBIx::Class::Schema/ddl_filename>, the schema name and the C<$version>.
1713 If no C<$directory> is specified then the statements are constructed on the
1714 fly using L<SQL::Translator> and C<$version> is ignored.
1716 See L<SQL::Translator/METHODS> for a list of values for C<$sqlt_args>.
1720 sub deployment_statements {
1721 my ($self, $schema, $type, $version, $dir, $sqltargs) = @_;
1722 # Need to be connected to get the correct sqlt_type
1723 $self->ensure_connected() unless $type;
1724 $type ||= $self->sqlt_type;
1725 $version ||= $schema->schema_version || '1.x';
1727 my $filename = $schema->ddl_filename($type, $version, $dir);
1731 open($file, "<$filename")
1732 or $self->throw_exception("Can't open $filename ($!)");
1735 return join('', @rows);
1738 $self->throw_exception(q{Can't deploy without SQL::Translator 0.09003: '}
1739 . $self->_check_sqlt_message . q{'})
1740 if !$self->_check_sqlt_version;
1742 require SQL::Translator::Parser::DBIx::Class;
1743 eval qq{use SQL::Translator::Producer::${type}};
1744 $self->throw_exception($@) if $@;
1746 # sources needs to be a parser arg, but for simplicty allow at top level
1748 $sqltargs->{parser_args}{sources} = delete $sqltargs->{sources}
1749 if exists $sqltargs->{sources};
1751 my $tr = SQL::Translator->new(%$sqltargs);
1752 SQL::Translator::Parser::DBIx::Class::parse( $tr, $schema );
1753 return "SQL::Translator::Producer::${type}"->can('produce')->($tr);
1757 my ($self, $schema, $type, $sqltargs, $dir) = @_;
1760 return if($line =~ /^--/);
1762 # next if($line =~ /^DROP/m);
1763 return if($line =~ /^BEGIN TRANSACTION/m);
1764 return if($line =~ /^COMMIT/m);
1765 return if $line =~ /^\s+$/; # skip whitespace only
1766 $self->_query_start($line);
1768 $self->dbh->do($line); # shouldn't be using ->dbh ?
1771 carp qq{$@ (running "${line}")};
1773 $self->_query_end($line);
1775 my @statements = $self->deployment_statements($schema, $type, undef, $dir, { %{ $sqltargs || {} }, no_comments => 1 } );
1776 if (@statements > 1) {
1777 foreach my $statement (@statements) {
1778 $deploy->( $statement );
1781 elsif (@statements == 1) {
1782 foreach my $line ( split(";\n", $statements[0])) {
1788 =head2 datetime_parser
1790 Returns the datetime parser class
1794 sub datetime_parser {
1796 return $self->{datetime_parser} ||= do {
1797 $self->ensure_connected;
1798 $self->build_datetime_parser(@_);
1802 =head2 datetime_parser_type
1804 Defines (returns) the datetime parser class - currently hardwired to
1805 L<DateTime::Format::MySQL>
1809 sub datetime_parser_type { "DateTime::Format::MySQL"; }
1811 =head2 build_datetime_parser
1813 See L</datetime_parser>
1817 sub build_datetime_parser {
1819 my $type = $self->datetime_parser_type(@_);
1821 $self->throw_exception("Couldn't load ${type}: $@") if $@;
1826 my $_check_sqlt_version; # private
1827 my $_check_sqlt_message; # private
1828 sub _check_sqlt_version {
1829 return $_check_sqlt_version if defined $_check_sqlt_version;
1830 eval 'use SQL::Translator "0.09003"';
1831 $_check_sqlt_message = $@ || '';
1832 $_check_sqlt_version = !$@;
1835 sub _check_sqlt_message {
1836 _check_sqlt_version if !defined $_check_sqlt_message;
1837 $_check_sqlt_message;
1841 =head2 is_replicating
1843 A boolean that reports if a particular L<DBIx::Class::Storage::DBI> is set to
1844 replicate from a master database. Default is undef, which is the result
1845 returned by databases that don't support replication.
1849 sub is_replicating {
1854 =head2 lag_behind_master
1856 Returns a number that represents a certain amount of lag behind a master db
1857 when a given storage is replicating. The number is database dependent, but
1858 starts at zero and increases with the amount of lag. Default in undef
1862 sub lag_behind_master {
1868 return if !$self->_dbh;
1877 =head2 DBIx::Class and AutoCommit
1879 DBIx::Class can do some wonderful magic with handling exceptions,
1880 disconnections, and transactions when you use C<< AutoCommit => 1 >>
1881 combined with C<txn_do> for transaction support.
1883 If you set C<< AutoCommit => 0 >> in your connect info, then you are always
1884 in an assumed transaction between commits, and you're telling us you'd
1885 like to manage that manually. A lot of the magic protections offered by
1886 this module will go away. We can't protect you from exceptions due to database
1887 disconnects because we don't know anything about how to restart your
1888 transactions. You're on your own for handling all sorts of exceptional
1889 cases if you choose the C<< AutoCommit => 0 >> path, just as you would
1896 Matt S. Trout <mst@shadowcatsystems.co.uk>
1898 Andy Grundman <andy@hybridized.org>
1902 You may distribute this code under the same terms as Perl itself.