1package DBIx::Class::Storage::DBI;
2# -*- mode: cperl; cperl-indent-level: 2 -*-
3
4use strict;
5use warnings;
6
7use base qw/DBIx::Class::Storage::DBIHacks DBIx::Class::Storage/;
8use mro 'c3';
9
10use Carp::Clan qw/^DBIx::Class/;
11use DBI;
12use DBIx::Class::Storage::DBI::Cursor;
13use DBIx::Class::Storage::Statistics;
14use Scalar::Util();
15use List::Util();
16use Data::Dumper::Concise();
17use Sub::Name ();
18
19__PACKAGE__->mk_group_accessors('simple' =>
20  qw/_connect_info _dbi_connect_info _dbh _sql_maker _sql_maker_opts _conn_pid
21     _conn_tid transaction_depth _dbh_autocommit _driver_determined savepoints/
22);
23
24# the values for these accessors are picked out (and deleted) from
25# the attribute hashref passed to connect_info
26my @storage_options = qw/
27  on_connect_call on_disconnect_call on_connect_do on_disconnect_do
28  disable_sth_caching unsafe auto_savepoint
29/;
30__PACKAGE__->mk_group_accessors('simple' => @storage_options);
31
32
33# default cursor class, overridable in connect_info attributes
34__PACKAGE__->cursor_class('DBIx::Class::Storage::DBI::Cursor');
35
36__PACKAGE__->mk_group_accessors('inherited' => qw/sql_maker_class/);
37__PACKAGE__->sql_maker_class('DBIx::Class::SQLAHacks');
38
39
40# Each of these methods need _determine_driver called before itself
41# in order to function reliably. This is a purely DRY optimization
42my @rdbms_specific_methods = qw/
43  deployment_statements
44  sqlt_type
45  build_datetime_parser
46  datetime_parser_type
47
48  insert
49  insert_bulk
50  update
51  delete
52  select
53  select_single
54/;
55
56for my $meth (@rdbms_specific_methods) {
57
58  my $orig = __PACKAGE__->can ($meth)
59    or next;
60
61  no strict qw/refs/;
62  no warnings qw/redefine/;
63  *{__PACKAGE__ ."::$meth"} = Sub::Name::subname $meth => sub {
64    if (not $_[0]->_driver_determined) {
65      $_[0]->_determine_driver;
66      goto $_[0]->can($meth);
67    }
68    $orig->(@_);
69  };
70}
71
72
73=head1 NAME
74
75DBIx::Class::Storage::DBI - DBI storage handler
76
77=head1 SYNOPSIS
78
79  my $schema = MySchema->connect('dbi:SQLite:my.db');
80
81  $schema->storage->debug(1);
82
83  my @stuff = $schema->storage->dbh_do(
84    sub {
85      my ($storage, $dbh, @args) = @_;
86      $dbh->do("DROP TABLE authors");
87    },
88    @column_list
89  );
90
91  $schema->resultset('Book')->search({
92     written_on => $schema->storage->datetime_parser(DateTime->now)
93  });
94
95=head1 DESCRIPTION
96
97This class represents the connection to an RDBMS via L<DBI>.  See
98L<DBIx::Class::Storage> for general information.  This pod only
99documents DBI-specific methods and behaviors.
100
101=head1 METHODS
102
103=cut
104
105sub new {
106  my $new = shift->next::method(@_);
107
108  $new->transaction_depth(0);
109  $new->_sql_maker_opts({});
110  $new->{savepoints} = [];
111  $new->{_in_dbh_do} = 0;
112  $new->{_dbh_gen} = 0;
113
114  $new;
115}
116
117=head2 connect_info
118
119This method is normally called by L<DBIx::Class::Schema/connection>, which
120encapsulates its argument list in an arrayref before passing them here.
121
122The argument list may contain:
123
124=over
125
126=item *
127
128The same 4-element argument set one would normally pass to
129L<DBI/connect>, optionally followed by
130L<extra attributes|/DBIx::Class specific connection attributes>
131recognized by DBIx::Class:
132
133  $connect_info_args = [ $dsn, $user, $password, \%dbi_attributes?, \%extra_attributes? ];
134
135=item *
136
137A single code reference which returns a connected
138L<DBI database handle|DBI/connect> optionally followed by
139L<extra attributes|/DBIx::Class specific connection attributes> recognized
140by DBIx::Class:
141
142  $connect_info_args = [ sub { DBI->connect (...) }, \%extra_attributes? ];
143
144=item *
145
146A single hashref with all the attributes and the dsn/user/password
147mixed together:
148
149  $connect_info_args = [{
150    dsn => $dsn,
151    user => $user,
152    password => $pass,
153    %dbi_attributes,
154    %extra_attributes,
155  }];
156
157  $connect_info_args = [{
158    dbh_maker => sub { DBI->connect (...) },
159    %dbi_attributes,
160    %extra_attributes,
161  }];
162
163This is particularly useful for L<Catalyst> based applications, allowing the
164following config (L<Config::General> style):
165
166  <Model::DB>
167    schema_class   App::DB
168    <connect_info>
169      dsn          dbi:mysql:database=test
170      user         testuser
171      password     TestPass
172      AutoCommit   1
173    </connect_info>
174  </Model::DB>
175
176The C<dsn>/C<user>/C<password> combination can be substituted by the
177C<dbh_maker> key whose value is a coderef that returns a connected
178L<DBI database handle|DBI/connect>
179
180=back
181
182Please note that the L<DBI> docs recommend that you always explicitly
183set C<AutoCommit> to either I<0> or I<1>.  L<DBIx::Class> further
184recommends that it be set to I<1>, and that you perform transactions
185via our L<DBIx::Class::Schema/txn_do> method.  L<DBIx::Class> will set it
186to I<1> if you do not do explicitly set it to zero.  This is the default
187for most DBDs. See L</DBIx::Class and AutoCommit> for details.
188
189=head3 DBIx::Class specific connection attributes
190
191In addition to the standard L<DBI|DBI/ATTRIBUTES_COMMON_TO_ALL_HANDLES>
192L<connection|DBI/Database_Handle_Attributes> attributes, DBIx::Class recognizes
193the following connection options. These options can be mixed in with your other
194L<DBI> connection attributes, or placed in a separate hashref
195(C<\%extra_attributes>) as shown above.
196
197Every time C<connect_info> is invoked, any previous settings for
198these options will be cleared before setting the new ones, regardless of
199whether any options are specified in the new C<connect_info>.
200
201
202=over
203
204=item on_connect_do
205
206Specifies things to do immediately after connecting or re-connecting to
207the database.  Its value may contain:
208
209=over
210
211=item a scalar
212
213This contains one SQL statement to execute.
214
215=item an array reference
216
217This contains SQL statements to execute in order.  Each element contains
218a string or a code reference that returns a string.
219
220=item a code reference
221
222This contains some code to execute.  Unlike code references within an
223array reference, its return value is ignored.
224
225=back
226
227=item on_disconnect_do
228
229Takes arguments in the same form as L</on_connect_do> and executes them
230immediately before disconnecting from the database.
231
232Note, this only runs if you explicitly call L</disconnect> on the
233storage object.
234
235=item on_connect_call
236
237A more generalized form of L</on_connect_do> that calls the specified
238C<connect_call_METHOD> methods in your storage driver.
239
240  on_connect_do => 'select 1'
241
242is equivalent to:
243
244  on_connect_call => [ [ do_sql => 'select 1' ] ]
245
246Its values may contain:
247
248=over
249
250=item a scalar
251
252Will call the C<connect_call_METHOD> method.
253
254=item a code reference
255
256Will execute C<< $code->($storage) >>
257
258=item an array reference
259
260Each value can be a method name or code reference.
261
262=item an array of arrays
263
264For each array, the first item is taken to be the C<connect_call_> method name
265or code reference, and the rest are parameters to it.
266
267=back
268
269Some predefined storage methods you may use:
270
271=over
272
273=item do_sql
274
275Executes a SQL string or a code reference that returns a SQL string. This is
276what L</on_connect_do> and L</on_disconnect_do> use.
277
278It can take:
279
280=over
281
282=item a scalar
283
284Will execute the scalar as SQL.
285
286=item an arrayref
287
288Taken to be arguments to L<DBI/do>, the SQL string optionally followed by the
289attributes hashref and bind values.
290
291=item a code reference
292
293Will execute C<< $code->($storage) >> and execute the return array refs as
294above.
295
296=back
297
298=item datetime_setup
299
300Execute any statements necessary to initialize the database session to return
301and accept datetime/timestamp values used with
302L<DBIx::Class::InflateColumn::DateTime>.
303
304Only necessary for some databases, see your specific storage driver for
305implementation details.
306
307=back
308
309=item on_disconnect_call
310
311Takes arguments in the same form as L</on_connect_call> and executes them
312immediately before disconnecting from the database.
313
314Calls the C<disconnect_call_METHOD> methods as opposed to the
315C<connect_call_METHOD> methods called by L</on_connect_call>.
316
317Note, this only runs if you explicitly call L</disconnect> on the
318storage object.
319
320=item disable_sth_caching
321
322If set to a true value, this option will disable the caching of
323statement handles via L<DBI/prepare_cached>.
324
325=item limit_dialect
326
327Sets the limit dialect. This is useful for JDBC-bridge among others
328where the remote SQL-dialect cannot be determined by the name of the
329driver alone. See also L<SQL::Abstract::Limit>.
330
331=item quote_char
332
333Specifies what characters to use to quote table and column names. If
334you use this you will want to specify L</name_sep> as well.
335
336C<quote_char> expects either a single character, in which case is it
337is placed on either side of the table/column name, or an arrayref of length
3382 in which case the table/column name is placed between the elements.
339
340For example under MySQL you should use C<< quote_char => '`' >>, and for
341SQL Server you should use C<< quote_char => [qw/[ ]/] >>.
342
343=item name_sep
344
345This only needs to be used in conjunction with C<quote_char>, and is used to
346specify the character that separates elements (schemas, tables, columns) from
347each other. In most cases this is simply a C<.>.
348
349The consequences of not supplying this value is that L<SQL::Abstract>
350will assume DBIx::Class' uses of aliases to be complete column
351names. The output will look like I<"me.name"> when it should actually
352be I<"me"."name">.
353
354=item unsafe
355
356This Storage driver normally installs its own C<HandleError>, sets
357C<RaiseError> and C<ShowErrorStatement> on, and sets C<PrintError> off on
358all database handles, including those supplied by a coderef.  It does this
359so that it can have consistent and useful error behavior.
360
361If you set this option to a true value, Storage will not do its usual
362modifications to the database handle's attributes, and instead relies on
363the settings in your connect_info DBI options (or the values you set in
364your connection coderef, in the case that you are connecting via coderef).
365
366Note that your custom settings can cause Storage to malfunction,
367especially if you set a C<HandleError> handler that suppresses exceptions
368and/or disable C<RaiseError>.
369
370=item auto_savepoint
371
372If this option is true, L<DBIx::Class> will use savepoints when nesting
373transactions, making it possible to recover from failure in the inner
374transaction without having to abort all outer transactions.
375
376=item cursor_class
377
378Use this argument to supply a cursor class other than the default
379L<DBIx::Class::Storage::DBI::Cursor>.
380
381=back
382
383Some real-life examples of arguments to L</connect_info> and
384L<DBIx::Class::Schema/connect>
385
386  # Simple SQLite connection
387  ->connect_info([ 'dbi:SQLite:./foo.db' ]);
388
389  # Connect via subref
390  ->connect_info([ sub { DBI->connect(...) } ]);
391
392  # Connect via subref in hashref
393  ->connect_info([{
394    dbh_maker => sub { DBI->connect(...) },
395    on_connect_do => 'alter session ...',
396  }]);
397
398  # A bit more complicated
399  ->connect_info(
400    [
401      'dbi:Pg:dbname=foo',
402      'postgres',
403      'my_pg_password',
404      { AutoCommit => 1 },
405      { quote_char => q{"}, name_sep => q{.} },
406    ]
407  );
408
409  # Equivalent to the previous example
410  ->connect_info(
411    [
412      'dbi:Pg:dbname=foo',
413      'postgres',
414      'my_pg_password',
415      { AutoCommit => 1, quote_char => q{"}, name_sep => q{.} },
416    ]
417  );
418
419  # Same, but with hashref as argument
420  # See parse_connect_info for explanation
421  ->connect_info(
422    [{
423      dsn         => 'dbi:Pg:dbname=foo',
424      user        => 'postgres',
425      password    => 'my_pg_password',
426      AutoCommit  => 1,
427      quote_char  => q{"},
428      name_sep    => q{.},
429    }]
430  );
431
432  # Subref + DBIx::Class-specific connection options
433  ->connect_info(
434    [
435      sub { DBI->connect(...) },
436      {
437          quote_char => q{`},
438          name_sep => q{@},
439          on_connect_do => ['SET search_path TO myschema,otherschema,public'],
440          disable_sth_caching => 1,
441      },
442    ]
443  );
444
445
446
447=cut
448
449sub connect_info {
450  my ($self, $info) = @_;
451
452  return $self->_connect_info if !$info;
453
454  $self->_connect_info($info); # copy for _connect_info
455
456  $info = $self->_normalize_connect_info($info)
457    if ref $info eq 'ARRAY';
458
459  for my $storage_opt (keys %{ $info->{storage_options} }) {
460    my $value = $info->{storage_options}{$storage_opt};
461
462    $self->$storage_opt($value);
463  }
464
465  # Kill sql_maker/_sql_maker_opts, so we get a fresh one with only
466  #  the new set of options
467  $self->_sql_maker(undef);
468  $self->_sql_maker_opts({});
469
470  for my $sql_maker_opt (keys %{ $info->{sql_maker_options} }) {
471    my $value = $info->{sql_maker_options}{$sql_maker_opt};
472
473    $self->_sql_maker_opts->{$sql_maker_opt} = $value;
474  }
475
476  my %attrs = (
477    %{ $self->_default_dbi_connect_attributes || {} },
478    %{ $info->{attributes} || {} },
479  );
480
481  my @args = @{ $info->{arguments} };
482
483  $self->_dbi_connect_info([@args,
484    %attrs && !(ref $args[0] eq 'CODE') ? \%attrs : ()]);
485
486  return $self->_connect_info;
487}
488
489sub _normalize_connect_info {
490  my ($self, $info_arg) = @_;
491  my %info;
492
493  my @args = @$info_arg;  # take a shallow copy for further mutilation
494
495  # combine/pre-parse arguments depending on invocation style
496
497  my %attrs;
498  if (ref $args[0] eq 'CODE') {     # coderef with optional \%extra_attributes
499    %attrs = %{ $args[1] || {} };
500    @args = $args[0];
501  }
502  elsif (ref $args[0] eq 'HASH') { # single hashref (i.e. Catalyst config)
503    %attrs = %{$args[0]};
504    @args = ();
505    if (my $code = delete $attrs{dbh_maker}) {
506      @args = $code;
507
508      my @ignored = grep { delete $attrs{$_} } (qw/dsn user password/);
509      if (@ignored) {
510        carp sprintf (
511            'Attribute(s) %s in connect_info were ignored, as they can not be applied '
512          . "to the result of 'dbh_maker'",
513
514          join (', ', map { "'$_'" } (@ignored) ),
515        );
516      }
517    }
518    else {
519      @args = delete @attrs{qw/dsn user password/};
520    }
521  }
522  else {                # otherwise assume dsn/user/password + \%attrs + \%extra_attrs
523    %attrs = (
524      % { $args[3] || {} },
525      % { $args[4] || {} },
526    );
527    @args = @args[0,1,2];
528  }
529
530  $info{arguments} = \@args;
531
532  my @storage_opts = grep exists $attrs{$_},
533    @storage_options, 'cursor_class';
534
535  @{ $info{storage_options} }{@storage_opts} =
536    delete @attrs{@storage_opts} if @storage_opts;
537
538  my @sql_maker_opts = grep exists $attrs{$_},
539    qw/limit_dialect quote_char name_sep/;
540
541  @{ $info{sql_maker_options} }{@sql_maker_opts} =
542    delete @attrs{@sql_maker_opts} if @sql_maker_opts;
543
544  $info{attributes} = \%attrs if %attrs;
545
546  return \%info;
547}
548
549sub _default_dbi_connect_attributes {
550  return {
551    AutoCommit => 1,
552    RaiseError => 1,
553    PrintError => 0,
554  };
555}
556
557=head2 on_connect_do
558
559This method is deprecated in favour of setting via L</connect_info>.
560
561=cut
562
563=head2 on_disconnect_do
564
565This method is deprecated in favour of setting via L</connect_info>.
566
567=cut
568
569sub _parse_connect_do {
570  my ($self, $type) = @_;
571
572  my $val = $self->$type;
573  return () if not defined $val;
574
575  my @res;
576
577  if (not ref($val)) {
578    push @res, [ 'do_sql', $val ];
579  } elsif (ref($val) eq 'CODE') {
580    push @res, $val;
581  } elsif (ref($val) eq 'ARRAY') {
582    push @res, map { [ 'do_sql', $_ ] } @$val;
583  } else {
584    $self->throw_exception("Invalid type for $type: ".ref($val));
585  }
586
587  return \@res;
588}
589
590=head2 dbh_do
591
592Arguments: ($subref | $method_name), @extra_coderef_args?
593
594Execute the given $subref or $method_name using the new exception-based
595connection management.
596
597The first two arguments will be the storage object that C<dbh_do> was called
598on and a database handle to use.  Any additional arguments will be passed
599verbatim to the called subref as arguments 2 and onwards.
600
601Using this (instead of $self->_dbh or $self->dbh) ensures correct
602exception handling and reconnection (or failover in future subclasses).
603
604Your subref should have no side-effects outside of the database, as
605there is the potential for your subref to be partially double-executed
606if the database connection was stale/dysfunctional.
607
608Example:
609
610  my @stuff = $schema->storage->dbh_do(
611    sub {
612      my ($storage, $dbh, @cols) = @_;
613      my $cols = join(q{, }, @cols);
614      $dbh->selectrow_array("SELECT $cols FROM foo");
615    },
616    @column_list
617  );
618
619=cut
620
621sub dbh_do {
622  my $self = shift;
623  my $code = shift;
624
625  my $dbh = $self->_get_dbh;
626
627  return $self->$code($dbh, @_) if $self->{_in_dbh_do}
628      || $self->{transaction_depth};
629
630  local $self->{_in_dbh_do} = 1;
631
632  my @result;
633  my $want_array = wantarray;
634
635  eval {
636
637    if($want_array) {
638        @result = $self->$code($dbh, @_);
639    }
640    elsif(defined $want_array) {
641        $result[0] = $self->$code($dbh, @_);
642    }
643    else {
644        $self->$code($dbh, @_);
645    }
646  };
647
648  # ->connected might unset $@ - copy
649  my $exception = $@;
650  if(!$exception) { return $want_array ? @result : $result[0] }
651
652  $self->throw_exception($exception) if $self->connected;
653
654  # We were not connected - reconnect and retry, but let any
655  #  exception fall right through this time
656  carp "Retrying $code after catching disconnected exception: $exception"
657    if $ENV{DBIC_DBIRETRY_DEBUG};
658  $self->_populate_dbh;
659  $self->$code($self->_dbh, @_);
660}
661
662# This is basically a blend of dbh_do above and DBIx::Class::Storage::txn_do.
663# It also informs dbh_do to bypass itself while under the direction of txn_do,
664#  via $self->{_in_dbh_do} (this saves some redundant eval and errorcheck, etc)
665sub txn_do {
666  my $self = shift;
667  my $coderef = shift;
668
669  ref $coderef eq 'CODE' or $self->throw_exception
670    ('$coderef must be a CODE reference');
671
672  return $coderef->(@_) if $self->{transaction_depth} && ! $self->auto_savepoint;
673
674  local $self->{_in_dbh_do} = 1;
675
676  my @result;
677  my $want_array = wantarray;
678
679  my $tried = 0;
680  while(1) {
681    eval {
682      $self->_get_dbh;
683
684      $self->txn_begin;
685      if($want_array) {
686          @result = $coderef->(@_);
687      }
688      elsif(defined $want_array) {
689          $result[0] = $coderef->(@_);
690      }
691      else {
692          $coderef->(@_);
693      }
694      $self->txn_commit;
695    };
696
697    # ->connected might unset $@ - copy
698    my $exception = $@;
699    if(!$exception) { return $want_array ? @result : $result[0] }
700
701    if($tried++ || $self->connected) {
702      eval { $self->txn_rollback };
703      my $rollback_exception = $@;
704      if($rollback_exception) {
705        my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
706        $self->throw_exception($exception)  # propagate nested rollback
707          if $rollback_exception =~ /$exception_class/;
708
709        $self->throw_exception(
710          "Transaction aborted: ${exception}. "
711          . "Rollback failed: ${rollback_exception}"
712        );
713      }
714      $self->throw_exception($exception)
715    }
716
717    # We were not connected, and was first try - reconnect and retry
718    # via the while loop
719    carp "Retrying $coderef after catching disconnected exception: $exception"
720      if $ENV{DBIC_DBIRETRY_DEBUG};
721    $self->_populate_dbh;
722  }
723}
724
725=head2 disconnect
726
727Our C<disconnect> method also performs a rollback first if the
728database is not in C<AutoCommit> mode.
729
730=cut
731
732sub disconnect {
733  my ($self) = @_;
734
735  if( $self->_dbh ) {
736    my @actions;
737
738    push @actions, ( $self->on_disconnect_call || () );
739    push @actions, $self->_parse_connect_do ('on_disconnect_do');
740
741    $self->_do_connection_actions(disconnect_call_ => $_) for @actions;
742
743    $self->_dbh_rollback unless $self->_dbh_autocommit;
744
745    $self->_dbh->disconnect;
746    $self->_dbh(undef);
747    $self->{_dbh_gen}++;
748  }
749}
750
751=head2 with_deferred_fk_checks
752
753=over 4
754
755=item Arguments: C<$coderef>
756
757=item Return Value: The return value of $coderef
758
759=back
760
761Storage specific method to run the code ref with FK checks deferred or
762in MySQL's case disabled entirely.
763
764=cut
765
766# Storage subclasses should override this
767sub with_deferred_fk_checks {
768  my ($self, $sub) = @_;
769  $sub->();
770}
771
772=head2 connected
773
774=over
775
776=item Arguments: none
777
778=item Return Value: 1|0
779
780=back
781
782Verifies that the current database handle is active and ready to execute
783an SQL statement (e.g. the connection did not get stale, server is still
784answering, etc.) This method is used internally by L</dbh>.
785
786=cut
787
788sub connected {
789  my $self = shift;
790  return 0 unless $self->_seems_connected;
791
792  #be on the safe side
793  local $self->_dbh->{RaiseError} = 1;
794
795  return $self->_ping;
796}
797
798sub _seems_connected {
799  my $self = shift;
800
801  my $dbh = $self->_dbh
802    or return 0;
803
804  if(defined $self->_conn_tid && $self->_conn_tid != threads->tid) {
805    $self->_dbh(undef);
806    $self->{_dbh_gen}++;
807    return 0;
808  }
809  else {
810    $self->_verify_pid;
811    return 0 if !$self->_dbh;
812  }
813
814  return $dbh->FETCH('Active');
815}
816
817sub _ping {
818  my $self = shift;
819
820  my $dbh = $self->_dbh or return 0;
821
822  return $dbh->ping;
823}
824
825# handle pid changes correctly
826#  NOTE: assumes $self->_dbh is a valid $dbh
827sub _verify_pid {
828  my ($self) = @_;
829
830  return if defined $self->_conn_pid && $self->_conn_pid == $$;
831
832  $self->_dbh->{InactiveDestroy} = 1;
833  $self->_dbh(undef);
834  $self->{_dbh_gen}++;
835
836  return;
837}
838
839sub ensure_connected {
840  my ($self) = @_;
841
842  unless ($self->connected) {
843    $self->_populate_dbh;
844  }
845}
846
847=head2 dbh
848
849Returns a C<$dbh> - a data base handle of class L<DBI>. The returned handle
850is guaranteed to be healthy by implicitly calling L</connected>, and if
851necessary performing a reconnection before returning. Keep in mind that this
852is very B<expensive> on some database engines. Consider using L<dbh_do>
853instead.
854
855=cut
856
857sub dbh {
858  my ($self) = @_;
859
860  if (not $self->_dbh) {
861    $self->_populate_dbh;
862  } else {
863    $self->ensure_connected;
864  }
865  return $self->_dbh;
866}
867
868# this is the internal "get dbh or connect (don't check)" method
869sub _get_dbh {
870  my $self = shift;
871  $self->_verify_pid if $self->_dbh;
872  $self->_populate_dbh unless $self->_dbh;
873  return $self->_dbh;
874}
875
876sub _sql_maker_args {
877    my ($self) = @_;
878
879    return (
880      bindtype=>'columns',
881      array_datatypes => 1,
882      limit_dialect => $self->_get_dbh,
883      %{$self->_sql_maker_opts}
884    );
885}
886
887sub sql_maker {
888  my ($self) = @_;
889  unless ($self->_sql_maker) {
890    my $sql_maker_class = $self->sql_maker_class;
891    $self->ensure_class_loaded ($sql_maker_class);
892    $self->_sql_maker($sql_maker_class->new( $self->_sql_maker_args ));
893  }
894  return $self->_sql_maker;
895}
896
897# nothing to do by default
898sub _rebless {}
899sub _init {}
900
901sub _populate_dbh {
902  my ($self) = @_;
903
904  my @info = @{$self->_dbi_connect_info || []};
905  $self->_dbh(undef); # in case ->connected failed we might get sent here
906  $self->_dbh($self->_connect(@info));
907
908  $self->_conn_pid($$);
909  $self->_conn_tid(threads->tid) if $INC{'threads.pm'};
910
911  $self->_determine_driver;
912
913  # Always set the transaction depth on connect, since
914  #  there is no transaction in progress by definition
915  $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
916
917  $self->_run_connection_actions unless $self->{_in_determine_driver};
918}
919
920sub _run_connection_actions {
921  my $self = shift;
922  my @actions;
923
924  push @actions, ( $self->on_connect_call || () );
925  push @actions, $self->_parse_connect_do ('on_connect_do');
926
927  $self->_do_connection_actions(connect_call_ => $_) for @actions;
928}
929
930sub _determine_driver {
931  my ($self) = @_;
932
933  if ((not $self->_driver_determined) && (not $self->{_in_determine_driver})) {
934    my $started_connected = 0;
935    local $self->{_in_determine_driver} = 1;
936
937    if (ref($self) eq __PACKAGE__) {
938      my $driver;
939      if ($self->_dbh) { # we are connected
940        $driver = $self->_dbh->{Driver}{Name};
941        $started_connected = 1;
942      } else {
943        # if connect_info is a CODEREF, we have no choice but to connect
944        if (ref $self->_dbi_connect_info->[0] &&
945            Scalar::Util::reftype($self->_dbi_connect_info->[0]) eq 'CODE') {
946          $self->_populate_dbh;
947          $driver = $self->_dbh->{Driver}{Name};
948        }
949        else {
950          # try to use dsn to not require being connected, the driver may still
951          # force a connection in _rebless to determine version
952          ($driver) = $self->_dbi_connect_info->[0] =~ /dbi:([^:]+):/i;
953        }
954      }
955
956      my $storage_class = "DBIx::Class::Storage::DBI::${driver}";
957      if ($self->load_optional_class($storage_class)) {
958        mro::set_mro($storage_class, 'c3');
959        bless $self, $storage_class;
960        $self->_rebless();
961      }
962    }
963
964    $self->_driver_determined(1);
965
966    $self->_init; # run driver-specific initializations
967
968    $self->_run_connection_actions
969        if !$started_connected && defined $self->_dbh;
970  }
971}
972
973sub _do_connection_actions {
974  my $self          = shift;
975  my $method_prefix = shift;
976  my $call          = shift;
977
978  if (not ref($call)) {
979    my $method = $method_prefix . $call;
980    $self->$method(@_);
981  } elsif (ref($call) eq 'CODE') {
982    $self->$call(@_);
983  } elsif (ref($call) eq 'ARRAY') {
984    if (ref($call->[0]) ne 'ARRAY') {
985      $self->_do_connection_actions($method_prefix, $_) for @$call;
986    } else {
987      $self->_do_connection_actions($method_prefix, @$_) for @$call;
988    }
989  } else {
990    $self->throw_exception (sprintf ("Don't know how to process conection actions of type '%s'", ref($call)) );
991  }
992
993  return $self;
994}
995
996sub connect_call_do_sql {
997  my $self = shift;
998  $self->_do_query(@_);
999}
1000
1001sub disconnect_call_do_sql {
1002  my $self = shift;
1003  $self->_do_query(@_);
1004}
1005
1006# override in db-specific backend when necessary
1007sub connect_call_datetime_setup { 1 }
1008
1009sub _do_query {
1010  my ($self, $action) = @_;
1011
1012  if (ref $action eq 'CODE') {
1013    $action = $action->($self);
1014    $self->_do_query($_) foreach @$action;
1015  }
1016  else {
1017    # Most debuggers expect ($sql, @bind), so we need to exclude
1018    # the attribute hash which is the second argument to $dbh->do
1019    # furthermore the bind values are usually to be presented
1020    # as named arrayref pairs, so wrap those here too
1021    my @do_args = (ref $action eq 'ARRAY') ? (@$action) : ($action);
1022    my $sql = shift @do_args;
1023    my $attrs = shift @do_args;
1024    my @bind = map { [ undef, $_ ] } @do_args;
1025
1026    $self->_query_start($sql, @bind);
1027    $self->_get_dbh->do($sql, $attrs, @do_args);
1028    $self->_query_end($sql, @bind);
1029  }
1030
1031  return $self;
1032}
1033
1034sub _connect {
1035  my ($self, @info) = @_;
1036
1037  $self->throw_exception("You failed to provide any connection info")
1038    if !@info;
1039
1040  my ($old_connect_via, $dbh);
1041
1042  if ($INC{'Apache/DBI.pm'} && $ENV{MOD_PERL}) {
1043    $old_connect_via = $DBI::connect_via;
1044    $DBI::connect_via = 'connect';
1045  }
1046
1047  eval {
1048    if(ref $info[0] eq 'CODE') {
1049       $dbh = $info[0]->();
1050    }
1051    else {
1052       $dbh = DBI->connect(@info);
1053    }
1054
1055    if($dbh && !$self->unsafe) {
1056      my $weak_self = $self;
1057      Scalar::Util::weaken($weak_self);
1058      $dbh->{HandleError} = sub {
1059          if ($weak_self) {
1060            $weak_self->throw_exception("DBI Exception: $_[0]");
1061          }
1062          else {
1063            # the handler may be invoked by something totally out of
1064            # the scope of DBIC
1065            croak ("DBI Exception: $_[0]");
1066          }
1067      };
1068      $dbh->{ShowErrorStatement} = 1;
1069      $dbh->{RaiseError} = 1;
1070      $dbh->{PrintError} = 0;
1071    }
1072  };
1073
1074  $DBI::connect_via = $old_connect_via if $old_connect_via;
1075
1076  $self->throw_exception("DBI Connection failed: " . ($@||$DBI::errstr))
1077    if !$dbh || $@;
1078
1079  $self->_dbh_autocommit($dbh->{AutoCommit});
1080
1081  $dbh;
1082}
1083
1084sub svp_begin {
1085  my ($self, $name) = @_;
1086
1087  $name = $self->_svp_generate_name
1088    unless defined $name;
1089
1090  $self->throw_exception ("You can't use savepoints outside a transaction")
1091    if $self->{transaction_depth} == 0;
1092
1093  $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1094    unless $self->can('_svp_begin');
1095
1096  push @{ $self->{savepoints} }, $name;
1097
1098  $self->debugobj->svp_begin($name) if $self->debug;
1099
1100  return $self->_svp_begin($name);
1101}
1102
1103sub svp_release {
1104  my ($self, $name) = @_;
1105
1106  $self->throw_exception ("You can't use savepoints outside a transaction")
1107    if $self->{transaction_depth} == 0;
1108
1109  $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1110    unless $self->can('_svp_release');
1111
1112  if (defined $name) {
1113    $self->throw_exception ("Savepoint '$name' does not exist")
1114      unless grep { $_ eq $name } @{ $self->{savepoints} };
1115
1116    # Dig through the stack until we find the one we are releasing.  This keeps
1117    # the stack up to date.
1118    my $svp;
1119
1120    do { $svp = pop @{ $self->{savepoints} } } while $svp ne $name;
1121  } else {
1122    $name = pop @{ $self->{savepoints} };
1123  }
1124
1125  $self->debugobj->svp_release($name) if $self->debug;
1126
1127  return $self->_svp_release($name);
1128}
1129
1130sub svp_rollback {
1131  my ($self, $name) = @_;
1132
1133  $self->throw_exception ("You can't use savepoints outside a transaction")
1134    if $self->{transaction_depth} == 0;
1135
1136  $self->throw_exception ("Your Storage implementation doesn't support savepoints")
1137    unless $self->can('_svp_rollback');
1138
1139  if (defined $name) {
1140      # If they passed us a name, verify that it exists in the stack
1141      unless(grep({ $_ eq $name } @{ $self->{savepoints} })) {
1142          $self->throw_exception("Savepoint '$name' does not exist!");
1143      }
1144
1145      # Dig through the stack until we find the one we are releasing.  This keeps
1146      # the stack up to date.
1147      while(my $s = pop(@{ $self->{savepoints} })) {
1148          last if($s eq $name);
1149      }
1150      # Add the savepoint back to the stack, as a rollback doesn't remove the
1151      # named savepoint, only everything after it.
1152      push(@{ $self->{savepoints} }, $name);
1153  } else {
1154      # We'll assume they want to rollback to the last savepoint
1155      $name = $self->{savepoints}->[-1];
1156  }
1157
1158  $self->debugobj->svp_rollback($name) if $self->debug;
1159
1160  return $self->_svp_rollback($name);
1161}
1162
1163sub _svp_generate_name {
1164    my ($self) = @_;
1165
1166    return 'savepoint_'.scalar(@{ $self->{'savepoints'} });
1167}
1168
1169sub txn_begin {
1170  my $self = shift;
1171
1172  # this means we have not yet connected and do not know the AC status
1173  # (e.g. coderef $dbh)
1174  $self->ensure_connected if (! defined $self->_dbh_autocommit);
1175
1176  if($self->{transaction_depth} == 0) {
1177    $self->debugobj->txn_begin()
1178      if $self->debug;
1179    $self->_dbh_begin_work;
1180  }
1181  elsif ($self->auto_savepoint) {
1182    $self->svp_begin;
1183  }
1184  $self->{transaction_depth}++;
1185}
1186
1187sub _dbh_begin_work {
1188  my $self = shift;
1189
1190  # if the user is utilizing txn_do - good for him, otherwise we need to
1191  # ensure that the $dbh is healthy on BEGIN.
1192  # We do this via ->dbh_do instead of ->dbh, so that the ->dbh "ping"
1193  # will be replaced by a failure of begin_work itself (which will be
1194  # then retried on reconnect)
1195  if ($self->{_in_dbh_do}) {
1196    $self->_dbh->begin_work;
1197  } else {
1198    $self->dbh_do(sub { $_[1]->begin_work });
1199  }
1200}
1201
1202sub txn_commit {
1203  my $self = shift;
1204  if ($self->{transaction_depth} == 1) {
1205    $self->debugobj->txn_commit()
1206      if ($self->debug);
1207    $self->_dbh_commit;
1208    $self->{transaction_depth} = 0
1209      if $self->_dbh_autocommit;
1210  }
1211  elsif($self->{transaction_depth} > 1) {
1212    $self->{transaction_depth}--;
1213    $self->svp_release
1214      if $self->auto_savepoint;
1215  }
1216}
1217
1218sub _dbh_commit {
1219  my $self = shift;
1220  my $dbh  = $self->_dbh
1221    or $self->throw_exception('cannot COMMIT on a disconnected handle');
1222  $dbh->commit;
1223}
1224
1225sub txn_rollback {
1226  my $self = shift;
1227  my $dbh = $self->_dbh;
1228  eval {
1229    if ($self->{transaction_depth} == 1) {
1230      $self->debugobj->txn_rollback()
1231        if ($self->debug);
1232      $self->{transaction_depth} = 0
1233        if $self->_dbh_autocommit;
1234      $self->_dbh_rollback;
1235    }
1236    elsif($self->{transaction_depth} > 1) {
1237      $self->{transaction_depth}--;
1238      if ($self->auto_savepoint) {
1239        $self->svp_rollback;
1240        $self->svp_release;
1241      }
1242    }
1243    else {
1244      die DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION->new;
1245    }
1246  };
1247  if ($@) {
1248    my $error = $@;
1249    my $exception_class = "DBIx::Class::Storage::NESTED_ROLLBACK_EXCEPTION";
1250    $error =~ /$exception_class/ and $self->throw_exception($error);
1251    # ensure that a failed rollback resets the transaction depth
1252    $self->{transaction_depth} = $self->_dbh_autocommit ? 0 : 1;
1253    $self->throw_exception($error);
1254  }
1255}
1256
1257sub _dbh_rollback {
1258  my $self = shift;
1259  my $dbh  = $self->_dbh
1260    or $self->throw_exception('cannot ROLLBACK on a disconnected handle');
1261  $dbh->rollback;
1262}
1263
1264# This used to be the top-half of _execute.  It was split out to make it
1265#  easier to override in NoBindVars without duping the rest.  It takes up
1266#  all of _execute's args, and emits $sql, @bind.
1267sub _prep_for_execute {
1268  my ($self, $op, $extra_bind, $ident, $args) = @_;
1269
1270  if( Scalar::Util::blessed($ident) && $ident->isa("DBIx::Class::ResultSource") ) {
1271    $ident = $ident->from();
1272  }
1273
1274  my ($sql, @bind) = $self->sql_maker->$op($ident, @$args);
1275
1276  unshift(@bind,
1277    map { ref $_ eq 'ARRAY' ? $_ : [ '!!dummy', $_ ] } @$extra_bind)
1278      if $extra_bind;
1279  return ($sql, \@bind);
1280}
1281
1282
1283sub _fix_bind_params {
1284    my ($self, @bind) = @_;
1285
1286    ### Turn @bind from something like this:
1287    ###   ( [ "artist", 1 ], [ "cdid", 1, 3 ] )
1288    ### to this:
1289    ###   ( "'1'", "'1'", "'3'" )
1290    return
1291        map {
1292            if ( defined( $_ && $_->[1] ) ) {
1293                map { qq{'$_'}; } @{$_}[ 1 .. $#$_ ];
1294            }
1295            else { q{'NULL'}; }
1296        } @bind;
1297}
1298
1299sub _query_start {
1300    my ( $self, $sql, @bind ) = @_;
1301
1302    if ( $self->debug ) {
1303        @bind = $self->_fix_bind_params(@bind);
1304
1305        $self->debugobj->query_start( $sql, @bind );
1306    }
1307}
1308
1309sub _query_end {
1310    my ( $self, $sql, @bind ) = @_;
1311
1312    if ( $self->debug ) {
1313        @bind = $self->_fix_bind_params(@bind);
1314        $self->debugobj->query_end( $sql, @bind );
1315    }
1316}
1317
1318sub _dbh_execute {
1319  my ($self, $dbh, $op, $extra_bind, $ident, $bind_attributes, @args) = @_;
1320
1321  my ($sql, $bind) = $self->_prep_for_execute($op, $extra_bind, $ident, \@args);
1322
1323  $self->_query_start( $sql, @$bind );
1324
1325  my $sth = $self->sth($sql,$op);
1326
1327  my $placeholder_index = 1;
1328
1329  foreach my $bound (@$bind) {
1330    my $attributes = {};
1331    my($column_name, @data) = @$bound;
1332
1333    if ($bind_attributes) {
1334      $attributes = $bind_attributes->{$column_name}
1335      if defined $bind_attributes->{$column_name};
1336    }
1337
1338    foreach my $data (@data) {
1339      my $ref = ref $data;
1340      $data = $ref && $ref ne 'ARRAY' ? ''.$data : $data; # stringify args (except arrayrefs)
1341
1342      $sth->bind_param($placeholder_index, $data, $attributes);
1343      $placeholder_index++;
1344    }
1345  }
1346
1347  # Can this fail without throwing an exception anyways???
1348  my $rv = $sth->execute();
1349  $self->throw_exception($sth->errstr) if !$rv;
1350
1351  $self->_query_end( $sql, @$bind );
1352
1353  return (wantarray ? ($rv, $sth, @$bind) : $rv);
1354}
1355
1356sub _execute {
1357    my $self = shift;
1358    $self->dbh_do('_dbh_execute', @_);  # retry over disconnects
1359}
1360
1361sub insert {
1362  my ($self, $source, $to_insert) = @_;
1363
1364  my $ident = $source->from;
1365  my $bind_attributes = $self->source_bind_attributes($source);
1366
1367  my $updated_cols = {};
1368
1369  foreach my $col ( $source->columns ) {
1370    if ( !defined $to_insert->{$col} ) {
1371      my $col_info = $source->column_info($col);
1372
1373      if ( $col_info->{auto_nextval} ) {
1374        $updated_cols->{$col} = $to_insert->{$col} = $self->_sequence_fetch(
1375          'nextval',
1376          $col_info->{sequence} ||
1377            $self->_dbh_get_autoinc_seq($self->_get_dbh, $source)
1378        );
1379      }
1380    }
1381  }
1382
1383  $self->_execute('insert' => [], $source, $bind_attributes, $to_insert);
1384
1385  return $updated_cols;
1386}
1387
1388## Currently it is assumed that all values passed will be "normal", i.e. not
1389## scalar refs, or at least, all the same type as the first set, the statement is
1390## only prepped once.
1391sub insert_bulk {
1392  my ($self, $source, $cols, $data) = @_;
1393
1394  my %colvalues;
1395  @colvalues{@$cols} = (0..$#$cols);
1396
1397  for my $i (0..$#$cols) {
1398    my $first_val = $data->[0][$i];
1399    next unless ref $first_val eq 'SCALAR';
1400
1401    $colvalues{ $cols->[$i] } = $first_val;
1402  }
1403
1404  # check for bad data and stringify stringifiable objects
1405  my $bad_slice = sub {
1406    my ($msg, $col_idx, $slice_idx) = @_;
1407    $self->throw_exception(sprintf "%s for column '%s' in populate slice:\n%s",
1408      $msg,
1409      $cols->[$col_idx],
1410      do {
1411        local $Data::Dumper::Maxdepth = 1; # don't dump objects, if any
1412        Data::Dumper::Concise::Dumper({
1413          map { $cols->[$_] => $data->[$slice_idx][$_] } (0 .. $#$cols)
1414        }),
1415      }
1416    );
1417  };
1418
1419  for my $datum_idx (0..$#$data) {
1420    my $datum = $data->[$datum_idx];
1421
1422    for my $col_idx (0..$#$cols) {
1423      my $val            = $datum->[$col_idx];
1424      my $sqla_bind      = $colvalues{ $cols->[$col_idx] };
1425      my $is_literal_sql = (ref $sqla_bind) eq 'SCALAR';
1426
1427      if ($is_literal_sql) {
1428        if (not ref $val) {
1429          $bad_slice->('bind found where literal SQL expected', $col_idx, $datum_idx);
1430        }
1431        elsif ((my $reftype = ref $val) ne 'SCALAR') {
1432          $bad_slice->("$reftype reference found where literal SQL expected",
1433            $col_idx, $datum_idx);
1434        }
1435        elsif ($$val ne $$sqla_bind){
1436          $bad_slice->("inconsistent literal SQL value, expecting: '$$sqla_bind'",
1437            $col_idx, $datum_idx);
1438        }
1439      }
1440      elsif (my $reftype = ref $val) {
1441        require overload;
1442        if (overload::Method($val, '""')) {
1443          $datum->[$col_idx] = "".$val;
1444        }
1445        else {
1446          $bad_slice->("$reftype reference found where bind expected",
1447            $col_idx, $datum_idx);
1448        }
1449      }
1450    }
1451  }
1452
1453  my ($sql, $bind) = $self->_prep_for_execute (
1454    'insert', undef, $source, [\%colvalues]
1455  );
1456  my @bind = @$bind;
1457
1458  my $empty_bind = 1 if (not @bind) &&
1459    (grep { ref $_ eq 'SCALAR' } values %colvalues) == @$cols;
1460
1461  if ((not @bind) && (not $empty_bind)) {
1462    $self->throw_exception(
1463      'Cannot insert_bulk without support for placeholders'
1464    );
1465  }
1466
1467  # neither _execute_array, nor _execute_inserts_with_no_binds are
1468  # atomic (even if _execute _array is a single call). Thus a safety
1469  # scope guard
1470  my $guard = $self->txn_scope_guard unless $self->{transaction_depth} != 0;
1471
1472  $self->_query_start( $sql, ['__BULK__'] );
1473  my $sth = $self->sth($sql);
1474  my $rv = do {
1475    if ($empty_bind) {
1476      # bind_param_array doesn't work if there are no binds
1477      $self->_dbh_execute_inserts_with_no_binds( $sth, scalar @$data );
1478    }
1479    else {
1480#      @bind = map { ref $_ ? ''.$_ : $_ } @bind; # stringify args
1481      $self->_execute_array( $source, $sth, \@bind, $cols, $data );
1482    }
1483  };
1484
1485  $self->_query_end( $sql, ['__BULK__'] );
1486
1487
1488  $guard->commit if $guard;
1489
1490  return (wantarray ? ($rv, $sth, @bind) : $rv);
1491}
1492
1493sub _execute_array {
1494  my ($self, $source, $sth, $bind, $cols, $data, @extra) = @_;
1495
1496  ## This must be an arrayref, else nothing works!
1497  my $tuple_status = [];
1498
1499  ## Get the bind_attributes, if any exist
1500  my $bind_attributes = $self->source_bind_attributes($source);
1501
1502  ## Bind the values and execute
1503  my $placeholder_index = 1;
1504
1505  foreach my $bound (@$bind) {
1506
1507    my $attributes = {};
1508    my ($column_name, $data_index) = @$bound;
1509
1510    if( $bind_attributes ) {
1511      $attributes = $bind_attributes->{$column_name}
1512      if defined $bind_attributes->{$column_name};
1513    }
1514
1515    my @data = map { $_->[$data_index] } @$data;
1516
1517    $sth->bind_param_array( $placeholder_index, [@data], $attributes );
1518    $placeholder_index++;
1519  }
1520
1521  my $rv = eval {
1522    $self->_dbh_execute_array($sth, $tuple_status, @extra);
1523  };
1524  my $err = $@ || $sth->errstr;
1525
1526# Statement must finish even if there was an exception.
1527  eval { $sth->finish };
1528  $err = $@ unless $err;
1529
1530  if ($err) {
1531    my $i = 0;
1532    ++$i while $i <= $#$tuple_status && !ref $tuple_status->[$i];
1533
1534    $self->throw_exception("Unexpected populate error: $err")
1535      if ($i > $#$tuple_status);
1536
1537    $self->throw_exception(sprintf "%s for populate slice:\n%s",
1538      ($tuple_status->[$i][1] || $err),
1539      Data::Dumper::Concise::Dumper({
1540        map { $cols->[$_] => $data->[$i][$_] } (0 .. $#$cols)
1541      }),
1542    );
1543  }
1544  return $rv;
1545}
1546
1547sub _dbh_execute_array {
1548    my ($self, $sth, $tuple_status, @extra) = @_;
1549
1550    return $sth->execute_array({ArrayTupleStatus => $tuple_status});
1551}
1552
1553sub _dbh_execute_inserts_with_no_binds {
1554  my ($self, $sth, $count) = @_;
1555
1556  eval {
1557    my $dbh = $self->_get_dbh;
1558    local $dbh->{RaiseError} = 1;
1559    local $dbh->{PrintError} = 0;
1560
1561    $sth->execute foreach 1..$count;
1562  };
1563  my $exception = $@;
1564
1565# Make sure statement is finished even if there was an exception.
1566  eval { $sth->finish };
1567  $exception = $@ unless $exception;
1568
1569  $self->throw_exception($exception) if $exception;
1570
1571  return $count;
1572}
1573
1574sub update {
1575  my ($self, $source, @args) = @_;
1576
1577  my $bind_attrs = $self->source_bind_attributes($source);
1578
1579  return $self->_execute('update' => [], $source, $bind_attrs, @args);
1580}
1581
1582
1583sub delete {
1584  my ($self, $source, @args) = @_;
1585
1586  my $bind_attrs = $self->source_bind_attributes($source);
1587
1588  return $self->_execute('delete' => [], $source, $bind_attrs, @args);
1589}
1590
1591# We were sent here because the $rs contains a complex search
1592# which will require a subquery to select the correct rows
1593# (i.e. joined or limited resultsets, or non-introspectable conditions)
1594#
1595# Generating a single PK column subquery is trivial and supported
1596# by all RDBMS. However if we have a multicolumn PK, things get ugly.
1597# Look at _multipk_update_delete()
1598sub _subq_update_delete {
1599  my $self = shift;
1600  my ($rs, $op, $values) = @_;
1601
1602  my $rsrc = $rs->result_source;
1603
1604  # quick check if we got a sane rs on our hands
1605  my @pcols = $rsrc->_pri_cols;
1606
1607  my $sel = $rs->_resolved_attrs->{select};
1608  $sel = [ $sel ] unless ref $sel eq 'ARRAY';
1609
1610  if (
1611      join ("\x00", map { join '.', $rs->{attrs}{alias}, $_ } sort @pcols)
1612        ne
1613      join ("\x00", sort @$sel )
1614  ) {
1615    $self->throw_exception (
1616      '_subq_update_delete can not be called on resultsets selecting columns other than the primary keys'
1617    );
1618  }
1619
1620  if (@pcols == 1) {
1621    return $self->$op (
1622      $rsrc,
1623      $op eq 'update' ? $values : (),
1624      { $pcols[0] => { -in => $rs->as_query } },
1625    );
1626  }
1627
1628  else {
1629    return $self->_multipk_update_delete (@_);
1630  }
1631}
1632
1633# ANSI SQL does not provide a reliable way to perform a multicol-PK
1634# resultset update/delete involving subqueries. So by default resort
1635# to simple (and inefficient) delete_all style per-row opearations,
1636# while allowing specific storages to override this with a faster
1637# implementation.
1638#
1639sub _multipk_update_delete {
1640  return shift->_per_row_update_delete (@_);
1641}
1642
1643# This is the default loop used to delete/update rows for multi PK
1644# resultsets, and used by mysql exclusively (because it can't do anything
1645# else).
1646#
1647# We do not use $row->$op style queries, because resultset update/delete
1648# is not expected to cascade (this is what delete_all/update_all is for).
1649#
1650# There should be no race conditions as the entire operation is rolled
1651# in a transaction.
1652#
1653sub _per_row_update_delete {
1654  my $self = shift;
1655  my ($rs, $op, $values) = @_;
1656
1657  my $rsrc = $rs->result_source;
1658  my @pcols = $rsrc->_pri_cols;
1659
1660  my $guard = $self->txn_scope_guard;
1661
1662  # emulate the return value of $sth->execute for non-selects
1663  my $row_cnt = '0E0';
1664
1665  my $subrs_cur = $rs->cursor;
1666  my @all_pk = $subrs_cur->all;
1667  for my $pks ( @all_pk) {
1668
1669    my $cond;
1670    for my $i (0.. $#pcols) {
1671      $cond->{$pcols[$i]} = $pks->[$i];
1672    }
1673
1674    $self->$op (
1675      $rsrc,
1676      $op eq 'update' ? $values : (),
1677      $cond,
1678    );
1679
1680    $row_cnt++;
1681  }
1682
1683  $guard->commit;
1684
1685  return $row_cnt;
1686}
1687
1688sub _select {
1689  my $self = shift;
1690
1691  # localization is neccessary as
1692  # 1) there is no infrastructure to pass this around before SQLA2
1693  # 2) _select_args sets it and _prep_for_execute consumes it
1694  my $sql_maker = $self->sql_maker;
1695  local $sql_maker->{_dbic_rs_attrs};
1696
1697  return $self->_execute($self->_select_args(@_));
1698}
1699
1700sub _select_args_to_query {
1701  my $self = shift;
1702
1703  # localization is neccessary as
1704  # 1) there is no infrastructure to pass this around before SQLA2
1705  # 2) _select_args sets it and _prep_for_execute consumes it
1706  my $sql_maker = $self->sql_maker;
1707  local $sql_maker->{_dbic_rs_attrs};
1708
1709  # my ($op, $bind, $ident, $bind_attrs, $select, $cond, $order, $rows, $offset)
1710  #  = $self->_select_args($ident, $select, $cond, $attrs);
1711  my ($op, $bind, $ident, $bind_attrs, @args) =
1712    $self->_select_args(@_);
1713
1714  # my ($sql, $prepared_bind) = $self->_prep_for_execute($op, $bind, $ident, [ $select, $cond, $order, $rows, $offset ]);
1715  my ($sql, $prepared_bind) = $self->_prep_for_execute($op, $bind, $ident, \@args);
1716  $prepared_bind ||= [];
1717
1718  return wantarray
1719    ? ($sql, $prepared_bind, $bind_attrs)
1720    : \[ "($sql)", @$prepared_bind ]
1721  ;
1722}
1723
1724sub _select_args {
1725  my ($self, $ident, $select, $where, $attrs) = @_;
1726
1727  my ($alias2source, $rs_alias) = $self->_resolve_ident_sources ($ident);
1728
1729  my $sql_maker = $self->sql_maker;
1730  $sql_maker->{_dbic_rs_attrs} = {
1731    %$attrs,
1732    select => $select,
1733    from => $ident,
1734    where => $where,
1735    $rs_alias && $alias2source->{$rs_alias}
1736      ? ( _source_handle => $alias2source->{$rs_alias}->handle )
1737      : ()
1738    ,
1739  };
1740
1741  # calculate bind_attrs before possible $ident mangling
1742  my $bind_attrs = {};
1743  for my $alias (keys %$alias2source) {
1744    my $bindtypes = $self->source_bind_attributes ($alias2source->{$alias}) || {};
1745    for my $col (keys %$bindtypes) {
1746
1747      my $fqcn = join ('.', $alias, $col);
1748      $bind_attrs->{$fqcn} = $bindtypes->{$col} if $bindtypes->{$col};
1749
1750      # Unqialified column names are nice, but at the same time can be
1751      # rather ambiguous. What we do here is basically go along with
1752      # the loop, adding an unqualified column slot to $bind_attrs,
1753      # alongside the fully qualified name. As soon as we encounter
1754      # another column by that name (which would imply another table)
1755      # we unset the unqualified slot and never add any info to it
1756      # to avoid erroneous type binding. If this happens the users
1757      # only choice will be to fully qualify his column name
1758
1759      if (exists $bind_attrs->{$col}) {
1760        $bind_attrs->{$col} = {};
1761      }
1762      else {
1763        $bind_attrs->{$col} = $bind_attrs->{$fqcn};
1764      }
1765    }
1766  }
1767
1768  # adjust limits
1769  if (
1770    $attrs->{software_limit}
1771      ||
1772    $sql_maker->_default_limit_syntax eq "GenericSubQ"
1773  ) {
1774    $attrs->{software_limit} = 1;
1775  }
1776  else {
1777    $self->throw_exception("rows attribute must be positive if present")
1778      if (defined($attrs->{rows}) && !($attrs->{rows} > 0));
1779
1780    # MySQL actually recommends this approach.  I cringe.
1781    $attrs->{rows} = 2**48 if not defined $attrs->{rows} and defined $attrs->{offset};
1782  }
1783
1784  my @limit;
1785
1786  # see if we need to tear the prefetch apart otherwise delegate the limiting to the
1787  # storage, unless software limit was requested
1788  if (
1789    #limited has_many
1790    ( $attrs->{rows} && keys %{$attrs->{collapse}} )
1791       ||
1792    # limited prefetch with RNO subqueries
1793    (
1794      $attrs->{rows}
1795        &&
1796      $sql_maker->limit_dialect eq 'RowNumberOver'
1797        &&
1798      $attrs->{_prefetch_select}
1799        &&
1800      @{$attrs->{_prefetch_select}}
1801    )
1802      ||
1803    # grouped prefetch
1804    ( $attrs->{group_by}
1805        &&
1806      @{$attrs->{group_by}}
1807        &&
1808      $attrs->{_prefetch_select}
1809        &&
1810      @{$attrs->{_prefetch_select}}
1811    )
1812  ) {
1813    ($ident, $select, $where, $attrs)
1814      = $self->_adjust_select_args_for_complex_prefetch ($ident, $select, $where, $attrs);
1815  }
1816
1817  elsif (
1818    ($attrs->{rows} || $attrs->{offset})
1819      &&
1820    $sql_maker->limit_dialect eq 'RowNumberOver'
1821      &&
1822    (ref $ident eq 'ARRAY' && @$ident > 1)  # indicates a join
1823      &&
1824    scalar $self->_parse_order_by ($attrs->{order_by})
1825  ) {
1826    # the RNO limit dialect above mangles the SQL such that the join gets lost
1827    # wrap a subquery here
1828
1829    push @limit, delete @{$attrs}{qw/rows offset/};
1830
1831    my $subq = $self->_select_args_to_query (
1832      $ident,
1833      $select,
1834      $where,
1835      $attrs,
1836    );
1837
1838    $ident = {
1839      -alias => $attrs->{alias},
1840      -source_handle => $ident->[0]{-source_handle},
1841      $attrs->{alias} => $subq,
1842    };
1843
1844    # all part of the subquery now
1845    delete @{$attrs}{qw/order_by group_by having/};
1846    $where = undef;
1847  }
1848
1849  elsif (! $attrs->{software_limit} ) {
1850    push @limit, $attrs->{rows}, $attrs->{offset};
1851  }
1852
1853  # try to simplify the joinmap further (prune unreferenced type-single joins)
1854  $ident = $self->_prune_unused_joins ($ident, $select, $where, $attrs);
1855
1856###
1857  # This would be the point to deflate anything found in $where
1858  # (and leave $attrs->{bind} intact). Problem is - inflators historically
1859  # expect a row object. And all we have is a resultsource (it is trivial
1860  # to extract deflator coderefs via $alias2source above).
1861  #
1862  # I don't see a way forward other than changing the way deflators are
1863  # invoked, and that's just bad...
1864###
1865
1866  my $order = { map
1867    { $attrs->{$_} ? ( $_ => $attrs->{$_} ) : ()  }
1868    (qw/order_by group_by having/ )
1869  };
1870
1871  return ('select', $attrs->{bind}, $ident, $bind_attrs, $select, $where, $order, @limit);
1872}
1873
1874# Returns a counting SELECT for a simple count
1875# query. Abstracted so that a storage could override
1876# this to { count => 'firstcol' } or whatever makes
1877# sense as a performance optimization
1878sub _count_select {
1879  #my ($self, $source, $rs_attrs) = @_;
1880  return { count => '*' };
1881}
1882
1883# Returns a SELECT which will end up in the subselect
1884# There may or may not be a group_by, as the subquery
1885# might have been called to accomodate a limit
1886#
1887# Most databases would be happy with whatever ends up
1888# here, but some choke in various ways.
1889#
1890sub _subq_count_select {
1891  my ($self, $source, $rs_attrs) = @_;
1892
1893  if (my $groupby = $rs_attrs->{group_by}) {
1894
1895    my $avail_columns = $self->_resolve_column_info ($rs_attrs->{from});
1896
1897    my $sel_index;
1898    for my $sel (@{$rs_attrs->{select}}) {
1899      if (ref $sel eq 'HASH' and $sel->{-as}) {
1900        $sel_index->{$sel->{-as}} = $sel;
1901      }
1902    }
1903
1904    my @selection;
1905    for my $g_part (@$groupby) {
1906      if (ref $g_part or $avail_columns->{$g_part}) {
1907        push @selection, $g_part;
1908      }
1909      elsif ($sel_index->{$g_part}) {
1910        push @selection, $sel_index->{$g_part};
1911      }
1912      else {
1913        $self->throw_exception ("group_by criteria '$g_part' not contained within current resultset source(s)");
1914      }
1915    }
1916
1917    return \@selection;
1918  }
1919
1920  my @pcols = map { join '.', $rs_attrs->{alias}, $_ } ($source->primary_columns);
1921  return @pcols ? \@pcols : [ 1 ];
1922}
1923
1924sub source_bind_attributes {
1925  my ($self, $source) = @_;
1926
1927  my $bind_attributes;
1928  foreach my $column ($source->columns) {
1929
1930    my $data_type = $source->column_info($column)->{data_type} || '';
1931    $bind_attributes->{$column} = $self->bind_attribute_by_data_type($data_type)
1932     if $data_type;
1933  }
1934
1935  return $bind_attributes;
1936}
1937
1938=head2 select
1939
1940=over 4
1941
1942=item Arguments: $ident, $select, $condition, $attrs
1943
1944=back
1945
1946Handle a SQL select statement.
1947
1948=cut
1949
1950sub select {
1951  my $self = shift;
1952  my ($ident, $select, $condition, $attrs) = @_;
1953  return $self->cursor_class->new($self, \@_, $attrs);
1954}
1955
1956sub select_single {
1957  my $self = shift;
1958  my ($rv, $sth, @bind) = $self->_select(@_);
1959  my @row = $sth->fetchrow_array;
1960  my @nextrow = $sth->fetchrow_array if @row;
1961  if(@row && @nextrow) {
1962    carp "Query returned more than one row.  SQL that returns multiple rows is DEPRECATED for ->find and ->single";
1963  }
1964  # Need to call finish() to work round broken DBDs
1965  $sth->finish();
1966  return @row;
1967}
1968
1969=head2 sth
1970
1971=over 4
1972
1973=item Arguments: $sql
1974
1975=back
1976
1977Returns a L<DBI> sth (statement handle) for the supplied SQL.
1978
1979=cut
1980
1981sub _dbh_sth {
1982  my ($self, $dbh, $sql) = @_;
1983
1984  # 3 is the if_active parameter which avoids active sth re-use
1985  my $sth = $self->disable_sth_caching
1986    ? $dbh->prepare($sql)
1987    : $dbh->prepare_cached($sql, {}, 3);
1988
1989  # XXX You would think RaiseError would make this impossible,
1990  #  but apparently that's not true :(
1991  $self->throw_exception($dbh->errstr) if !$sth;
1992
1993  $sth;
1994}
1995
1996sub sth {
1997  my ($self, $sql) = @_;
1998  $self->dbh_do('_dbh_sth', $sql);  # retry over disconnects
1999}
2000
2001sub _dbh_columns_info_for {
2002  my ($self, $dbh, $table) = @_;
2003
2004  if ($dbh->can('column_info')) {
2005    my %result;
2006    eval {
2007      my ($schema,$tab) = $table =~ /^(.+?)\.(.+)$/ ? ($1,$2) : (undef,$table);
2008      my $sth = $dbh->column_info( undef,$schema, $tab, '%' );
2009      $sth->execute();
2010      while ( my $info = $sth->fetchrow_hashref() ){
2011        my %column_info;
2012        $column_info{data_type}   = $info->{TYPE_NAME};
2013        $column_info{size}      = $info->{COLUMN_SIZE};
2014        $column_info{is_nullable}   = $info->{NULLABLE} ? 1 : 0;
2015        $column_info{default_value} = $info->{COLUMN_DEF};
2016        my $col_name = $info->{COLUMN_NAME};
2017        $col_name =~ s/^\"(.*)\"$/$1/;
2018
2019        $result{$col_name} = \%column_info;
2020      }
2021    };
2022    return \%result if !$@ && scalar keys %result;
2023  }
2024
2025  my %result;
2026  my $sth = $dbh->prepare($self->sql_maker->select($table, undef, \'1 = 0'));
2027  $sth->execute;
2028  my @columns = @{$sth->{NAME_lc}};
2029  for my $i ( 0 .. $#columns ){
2030    my %column_info;
2031    $column_info{data_type} = $sth->{TYPE}->[$i];
2032    $column_info{size} = $sth->{PRECISION}->[$i];
2033    $column_info{is_nullable} = $sth->{NULLABLE}->[$i] ? 1 : 0;
2034
2035    if ($column_info{data_type} =~ m/^(.*?)\((.*?)\)$/) {
2036      $column_info{data_type} = $1;
2037      $column_info{size}    = $2;
2038    }
2039
2040    $result{$columns[$i]} = \%column_info;
2041  }
2042  $sth->finish;
2043
2044  foreach my $col (keys %result) {
2045    my $colinfo = $result{$col};
2046    my $type_num = $colinfo->{data_type};
2047    my $type_name;
2048    if(defined $type_num && $dbh->can('type_info')) {
2049      my $type_info = $dbh->type_info($type_num);
2050      $type_name = $type_info->{TYPE_NAME} if $type_info;
2051      $colinfo->{data_type} = $type_name if $type_name;
2052    }
2053  }
2054
2055  return \%result;
2056}
2057
2058sub columns_info_for {
2059  my ($self, $table) = @_;
2060  $self->_dbh_columns_info_for ($self->_get_dbh, $table);
2061}
2062
2063=head2 last_insert_id
2064
2065Return the row id of the last insert.
2066
2067=cut
2068
2069sub _dbh_last_insert_id {
2070    my ($self, $dbh, $source, $col) = @_;
2071
2072    my $id = eval { $dbh->last_insert_id (undef, undef, $source->name, $col) };
2073
2074    return $id if defined $id;
2075
2076    my $class = ref $self;
2077    $self->throw_exception ("No storage specific _dbh_last_insert_id() method implemented in $class, and the generic DBI::last_insert_id() failed");
2078}
2079
2080sub last_insert_id {
2081  my $self = shift;
2082  $self->_dbh_last_insert_id ($self->_dbh, @_);
2083}
2084
2085=head2 _native_data_type
2086
2087=over 4
2088
2089=item Arguments: $type_name
2090
2091=back
2092
2093This API is B<EXPERIMENTAL>, will almost definitely change in the future, and
2094currently only used by L<::AutoCast|DBIx::Class::Storage::DBI::AutoCast> and
2095L<::Sybase::ASE|DBIx::Class::Storage::DBI::Sybase::ASE>.
2096
2097The default implementation returns C<undef>, implement in your Storage driver if
2098you need this functionality.
2099
2100Should map types from other databases to the native RDBMS type, for example
2101C<VARCHAR2> to C<VARCHAR>.
2102
2103Types with modifiers should map to the underlying data type. For example,
2104C<INTEGER AUTO_INCREMENT> should become C<INTEGER>.
2105
2106Composite types should map to the container type, for example
2107C<ENUM(foo,bar,baz)> becomes C<ENUM>.
2108
2109=cut
2110
2111sub _native_data_type {
2112  #my ($self, $data_type) = @_;
2113  return undef
2114}
2115
2116# Check if placeholders are supported at all
2117sub _placeholders_supported {
2118  my $self = shift;
2119  my $dbh  = $self->_get_dbh;
2120
2121  # some drivers provide a $dbh attribute (e.g. Sybase and $dbh->{syb_dynamic_supported})
2122  # but it is inaccurate more often than not
2123  eval {
2124    local $dbh->{PrintError} = 0;
2125    local $dbh->{RaiseError} = 1;
2126    $dbh->do('select ?', {}, 1);
2127  };
2128  return $@ ? 0 : 1;
2129}
2130
2131# Check if placeholders bound to non-string types throw exceptions
2132#
2133sub _typeless_placeholders_supported {
2134  my $self = shift;
2135  my $dbh  = $self->_get_dbh;
2136
2137  eval {
2138    local $dbh->{PrintError} = 0;
2139    local $dbh->{RaiseError} = 1;
2140    # this specifically tests a bind that is NOT a string
2141    $dbh->do('select 1 where 1 = ?', {}, 1);
2142  };
2143  return $@ ? 0 : 1;
2144}
2145
2146=head2 sqlt_type
2147
2148Returns the database driver name.
2149
2150=cut
2151
2152sub sqlt_type {
2153  shift->_get_dbh->{Driver}->{Name};
2154}
2155
2156=head2 bind_attribute_by_data_type
2157
2158Given a datatype from column info, returns a database specific bind
2159attribute for C<< $dbh->bind_param($val,$attribute) >> or nothing if we will
2160let the database planner just handle it.
2161
2162Generally only needed for special case column types, like bytea in postgres.
2163
2164=cut
2165
2166sub bind_attribute_by_data_type {
2167    return;
2168}
2169
2170=head2 is_datatype_numeric
2171
2172Given a datatype from column_info, returns a boolean value indicating if
2173the current RDBMS considers it a numeric value. This controls how
2174L<DBIx::Class::Row/set_column> decides whether to mark the column as
2175dirty - when the datatype is deemed numeric a C<< != >> comparison will
2176be performed instead of the usual C<eq>.
2177
2178=cut
2179
2180sub is_datatype_numeric {
2181  my ($self, $dt) = @_;
2182
2183  return 0 unless $dt;
2184
2185  return $dt =~ /^ (?:
2186    numeric | int(?:eger)? | (?:tiny|small|medium|big)int | dec(?:imal)? | real | float | double (?: \s+ precision)? | (?:big)?serial
2187  ) $/ix;
2188}
2189
2190
2191=head2 create_ddl_dir
2192
2193=over 4
2194
2195=item Arguments: $schema \@databases, $version, $directory, $preversion, \%sqlt_args
2196
2197=back
2198
2199Creates a SQL file based on the Schema, for each of the specified
2200database engines in C<\@databases> in the given directory.
2201(note: specify L<SQL::Translator> names, not L<DBI> driver names).
2202
2203Given a previous version number, this will also create a file containing
2204the ALTER TABLE statements to transform the previous schema into the
2205current one. Note that these statements may contain C<DROP TABLE> or
2206C<DROP COLUMN> statements that can potentially destroy data.
2207
2208The file names are created using the C<ddl_filename> method below, please
2209override this method in your schema if you would like a different file
2210name format. For the ALTER file, the same format is used, replacing
2211$version in the name with "$preversion-$version".
2212
2213See L<SQL::Translator/METHODS> for a list of values for C<\%sqlt_args>.
2214The most common value for this would be C<< { add_drop_table => 1 } >>
2215to have the SQL produced include a C<DROP TABLE> statement for each table
2216created. For quoting purposes supply C<quote_table_names> and
2217C<quote_field_names>.
2218
2219If no arguments are passed, then the following default values are assumed:
2220
2221=over 4
2222
2223=item databases  - ['MySQL', 'SQLite', 'PostgreSQL']
2224
2225=item version    - $schema->schema_version
2226
2227=item directory  - './'
2228
2229=item preversion - <none>
2230
2231=back
2232
2233By default, C<\%sqlt_args> will have
2234
2235 { add_drop_table => 1, ignore_constraint_names => 1, ignore_index_names => 1 }
2236
2237merged with the hash passed in. To disable any of those features, pass in a
2238hashref like the following
2239
2240 { ignore_constraint_names => 0, # ... other options }
2241
2242
2243WARNING: You are strongly advised to check all SQL files created, before applying
2244them.
2245
2246=cut
2247
2248sub create_ddl_dir {
2249  my ($self, $schema, $databases, $version, $dir, $preversion, $sqltargs) = @_;
2250
2251  if(!$dir || !-d $dir) {
2252    carp "No directory given, using ./\n";
2253    $dir = "./";
2254  }
2255  $databases ||= ['MySQL', 'SQLite', 'PostgreSQL'];
2256  $databases = [ $databases ] if(ref($databases) ne 'ARRAY');
2257
2258  my $schema_version = $schema->schema_version || '1.x';
2259  $version ||= $schema_version;
2260
2261  $sqltargs = {
2262    add_drop_table => 1,
2263    ignore_constraint_names => 1,
2264    ignore_index_names => 1,
2265    %{$sqltargs || {}}
2266  };
2267
2268  unless (DBIx::Class::Optional::Dependencies->req_ok_for ('deploy')) {
2269    $self->throw_exception("Can't create a ddl file without " . DBIx::Class::Optional::Dependencies->req_missing_for ('deploy') );
2270  }
2271
2272  my $sqlt = SQL::Translator->new( $sqltargs );
2273
2274  $sqlt->parser('SQL::Translator::Parser::DBIx::Class');
2275  my $sqlt_schema = $sqlt->translate({ data => $schema })
2276    or $self->throw_exception ($sqlt->error);
2277
2278  foreach my $db (@$databases) {
2279    $sqlt->reset();
2280    $sqlt->{schema} = $sqlt_schema;
2281    $sqlt->producer($db);
2282
2283    my $file;
2284    my $filename = $schema->ddl_filename($db, $version, $dir);
2285    if (-e $filename && ($version eq $schema_version )) {
2286      # if we are dumping the current version, overwrite the DDL
2287      carp "Overwriting existing DDL file - $filename";
2288      unlink($filename);
2289    }
2290
2291    my $output = $sqlt->translate;
2292    if(!$output) {
2293      carp("Failed to translate to $db, skipping. (" . $sqlt->error . ")");
2294      next;
2295    }
2296    if(!open($file, ">$filename")) {
2297      $self->throw_exception("Can't open $filename for writing ($!)");
2298      next;
2299    }
2300    print $file $output;
2301    close($file);
2302
2303    next unless ($preversion);
2304
2305    require SQL::Translator::Diff;
2306
2307    my $prefilename = $schema->ddl_filename($db, $preversion, $dir);
2308    if(!-e $prefilename) {
2309      carp("No previous schema file found ($prefilename)");
2310      next;
2311    }
2312
2313    my $difffile = $schema->ddl_filename($db, $version, $dir, $preversion);
2314    if(-e $difffile) {
2315      carp("Overwriting existing diff file - $difffile");
2316      unlink($difffile);
2317    }
2318
2319    my $source_schema;
2320    {
2321      my $t = SQL::Translator->new($sqltargs);
2322      $t->debug( 0 );
2323      $t->trace( 0 );
2324
2325      $t->parser( $db )
2326        or $self->throw_exception ($t->error);
2327
2328      my $out = $t->translate( $prefilename )
2329        or $self->throw_exception ($t->error);
2330
2331      $source_schema = $t->schema;
2332
2333      $source_schema->name( $prefilename )
2334        unless ( $source_schema->name );
2335    }
2336
2337    # The "new" style of producers have sane normalization and can support
2338    # diffing a SQL file against a DBIC->SQLT schema. Old style ones don't
2339    # And we have to diff parsed SQL against parsed SQL.
2340    my $dest_schema = $sqlt_schema;
2341
2342    unless ( "SQL::Translator::Producer::$db"->can('preprocess_schema') ) {
2343      my $t = SQL::Translator->new($sqltargs);
2344      $t->debug( 0 );
2345      $t->trace( 0 );
2346
2347      $t->parser( $db )
2348        or $self->throw_exception ($t->error);
2349
2350      my $out = $t->translate( $filename )
2351        or $self->throw_exception ($t->error);
2352
2353      $dest_schema = $t->schema;
2354
2355      $dest_schema->name( $filename )
2356        unless $dest_schema->name;
2357    }
2358
2359    my $diff = SQL::Translator::Diff::schema_diff($source_schema, $db,
2360                                                  $dest_schema,   $db,
2361                                                  $sqltargs
2362                                                 );
2363    if(!open $file, ">$difffile") {
2364      $self->throw_exception("Can't write to $difffile ($!)");
2365      next;
2366    }
2367    print $file $diff;
2368    close($file);
2369  }
2370}
2371
2372=head2 deployment_statements
2373
2374=over 4
2375
2376=item Arguments: $schema, $type, $version, $directory, $sqlt_args
2377
2378=back
2379
2380Returns the statements used by L</deploy> and L<DBIx::Class::Schema/deploy>.
2381
2382The L<SQL::Translator> (not L<DBI>) database driver name can be explicitly
2383provided in C<$type>, otherwise the result of L</sqlt_type> is used as default.
2384
2385C<$directory> is used to return statements from files in a previously created
2386L</create_ddl_dir> directory and is optional. The filenames are constructed
2387from L<DBIx::Class::Schema/ddl_filename>, the schema name and the C<$version>.
2388
2389If no C<$directory> is specified then the statements are constructed on the
2390fly using L<SQL::Translator> and C<$version> is ignored.
2391
2392See L<SQL::Translator/METHODS> for a list of values for C<$sqlt_args>.
2393
2394=cut
2395
2396sub deployment_statements {
2397  my ($self, $schema, $type, $version, $dir, $sqltargs) = @_;
2398  $type ||= $self->sqlt_type;
2399  $version ||= $schema->schema_version || '1.x';
2400  $dir ||= './';
2401  my $filename = $schema->ddl_filename($type, $version, $dir);
2402  if(-f $filename)
2403  {
2404      my $file;
2405      open($file, "<$filename")
2406        or $self->throw_exception("Can't open $filename ($!)");
2407      my @rows = <$file>;
2408      close($file);
2409      return join('', @rows);
2410  }
2411
2412  unless (DBIx::Class::Optional::Dependencies->req_ok_for ('deploy') ) {
2413    $self->throw_exception("Can't deploy without a ddl_dir or " . DBIx::Class::Optional::Dependencies->req_missing_for ('deploy') );
2414  }
2415
2416  # sources needs to be a parser arg, but for simplicty allow at top level
2417  # coming in
2418  $sqltargs->{parser_args}{sources} = delete $sqltargs->{sources}
2419      if exists $sqltargs->{sources};
2420
2421  my $tr = SQL::Translator->new(
2422    producer => "SQL::Translator::Producer::${type}",
2423    %$sqltargs,
2424    parser => 'SQL::Translator::Parser::DBIx::Class',
2425    data => $schema,
2426  );
2427
2428  my @ret;
2429  my $wa = wantarray;
2430  if ($wa) {
2431    @ret = $tr->translate;
2432  }
2433  else {
2434    $ret[0] = $tr->translate;
2435  }
2436
2437  $self->throw_exception( 'Unable to produce deployment statements: ' . $tr->error)
2438    unless (@ret && defined $ret[0]);
2439
2440  return $wa ? @ret : $ret[0];
2441}
2442
2443sub deploy {
2444  my ($self, $schema, $type, $sqltargs, $dir) = @_;
2445  my $deploy = sub {
2446    my $line = shift;
2447    return if($line =~ /^--/);
2448    return if(!$line);
2449    # next if($line =~ /^DROP/m);
2450    return if($line =~ /^BEGIN TRANSACTION/m);
2451    return if($line =~ /^COMMIT/m);
2452    return if $line =~ /^\s+$/; # skip whitespace only
2453    $self->_query_start($line);
2454    eval {
2455      # do a dbh_do cycle here, as we need some error checking in
2456      # place (even though we will ignore errors)
2457      $self->dbh_do (sub { $_[1]->do($line) });
2458    };
2459    if ($@) {
2460      carp qq{$@ (running "${line}")};
2461    }
2462    $self->_query_end($line);
2463  };
2464  my @statements = $schema->deployment_statements($type, undef, $dir, { %{ $sqltargs || {} }, no_comments => 1 } );
2465  if (@statements > 1) {
2466    foreach my $statement (@statements) {
2467      $deploy->( $statement );
2468    }
2469  }
2470  elsif (@statements == 1) {
2471    foreach my $line ( split(";\n", $statements[0])) {
2472      $deploy->( $line );
2473    }
2474  }
2475}
2476
2477=head2 datetime_parser
2478
2479Returns the datetime parser class
2480
2481=cut
2482
2483sub datetime_parser {
2484  my $self = shift;
2485  return $self->{datetime_parser} ||= do {
2486    $self->build_datetime_parser(@_);
2487  };
2488}
2489
2490=head2 datetime_parser_type
2491
2492Defines (returns) the datetime parser class - currently hardwired to
2493L<DateTime::Format::MySQL>
2494
2495=cut
2496
2497sub datetime_parser_type { "DateTime::Format::MySQL"; }
2498
2499=head2 build_datetime_parser
2500
2501See L</datetime_parser>
2502
2503=cut
2504
2505sub build_datetime_parser {
2506  my $self = shift;
2507  my $type = $self->datetime_parser_type(@_);
2508  $self->ensure_class_loaded ($type);
2509  return $type;
2510}
2511
2512
2513=head2 is_replicating
2514
2515A boolean that reports if a particular L<DBIx::Class::Storage::DBI> is set to
2516replicate from a master database.  Default is undef, which is the result
2517returned by databases that don't support replication.
2518
2519=cut
2520
2521sub is_replicating {
2522    return;
2523
2524}
2525
2526=head2 lag_behind_master
2527
2528Returns a number that represents a certain amount of lag behind a master db
2529when a given storage is replicating.  The number is database dependent, but
2530starts at zero and increases with the amount of lag. Default in undef
2531
2532=cut
2533
2534sub lag_behind_master {
2535    return;
2536}
2537
2538=head2 relname_to_table_alias
2539
2540=over 4
2541
2542=item Arguments: $relname, $join_count
2543
2544=back
2545
2546L<DBIx::Class> uses L<DBIx::Class::Relationship> names as table aliases in
2547queries.
2548
2549This hook is to allow specific L<DBIx::Class::Storage> drivers to change the
2550way these aliases are named.
2551
2552The default behavior is C<"$relname_$join_count" if $join_count > 1>, otherwise
2553C<"$relname">.
2554
2555=cut
2556
2557sub relname_to_table_alias {
2558  my ($self, $relname, $join_count) = @_;
2559
2560  my $alias = ($join_count && $join_count > 1 ?
2561    join('_', $relname, $join_count) : $relname);
2562
2563  return $alias;
2564}
2565
2566sub DESTROY {
2567  my $self = shift;
2568
2569  $self->_verify_pid if $self->_dbh;
2570
2571  # some databases need this to stop spewing warnings
2572  if (my $dbh = $self->_dbh) {
2573    local $@;
2574    eval {
2575      %{ $dbh->{CachedKids} } = ();
2576      $dbh->disconnect;
2577    };
2578  }
2579
2580  $self->_dbh(undef);
2581}
2582
25831;
2584
2585=head1 USAGE NOTES
2586
2587=head2 DBIx::Class and AutoCommit
2588
2589DBIx::Class can do some wonderful magic with handling exceptions,
2590disconnections, and transactions when you use C<< AutoCommit => 1 >>
2591(the default) combined with C<txn_do> for transaction support.
2592
2593If you set C<< AutoCommit => 0 >> in your connect info, then you are always
2594in an assumed transaction between commits, and you're telling us you'd
2595like to manage that manually.  A lot of the magic protections offered by
2596this module will go away.  We can't protect you from exceptions due to database
2597disconnects because we don't know anything about how to restart your
2598transactions.  You're on your own for handling all sorts of exceptional
2599cases if you choose the C<< AutoCommit => 0 >> path, just as you would
2600be with raw DBI.
2601
2602
2603=head1 AUTHORS
2604
2605Matt S. Trout <mst@shadowcatsystems.co.uk>
2606
2607Andy Grundman <andy@hybridized.org>
2608
2609=head1 LICENSE
2610
2611You may distribute this code under the same terms as Perl itself.
2612
2613=cut
2614