summaryrefslogtreecommitdiffstats
path: root/Bugzilla/DB/Mysql.pm
blob: 640cf89ec34b864fffdb7e86f684744cacbe5f72 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
#
# This Source Code Form is "Incompatible With Secondary Licenses", as
# defined by the Mozilla Public License, v. 2.0.

=head1 NAME

Bugzilla::DB::Mysql - Bugzilla database compatibility layer for MySQL

=head1 DESCRIPTION

This module overrides methods of the Bugzilla::DB module with MySQL specific
implementation. It is instantiated by the Bugzilla::DB module and should never
be used directly.

For interface details see L<Bugzilla::DB> and L<DBI>.

=cut

package Bugzilla::DB::Mysql;

use 5.10.1;
use Moo;

extends qw(Bugzilla::DB);

use Bugzilla::Constants;
use Bugzilla::Install::Util qw(install_string);
use Bugzilla::Util;
use Bugzilla::Error;
use Bugzilla::DB::Schema::Mysql;

use List::Util qw(max any);
use Text::ParseWords;
use Carp;

# This is how many comments of MAX_COMMENT_LENGTH we expect on a single bug.
# In reality, you could have a LOT more comments than this, because
# MAX_COMMENT_LENGTH is big.
use constant MAX_COMMENTS => 50;

use constant FULLTEXT_OR => '|';

sub BUILDARGS {
  my ($class, $params) = @_;
  my ($user, $pass, $host, $dbname, $port, $sock)
    = @$params{qw(db_user db_pass db_host db_name db_port db_sock)};

  # construct the DSN from the parameters we got
  my $dsn = "dbi:mysql:host=$host;database=$dbname";
  $dsn .= ";port=$port"         if $port;
  $dsn .= ";mysql_socket=$sock" if $sock;

  my %attrs = (mysql_enable_utf8 => 1);

  return {dsn => $dsn, user => $user, pass => $pass, attrs => \%attrs};
}

sub on_dbi_connected {
  my ($class, $dbh) = @_;

  # This makes sure that if the tables are encoded as UTF-8, we
  # return their data correctly.
  my $charset = $class->utf8_charset;
  my $collate = $class->utf8_collate;
  $dbh->do("SET NAMES $charset COLLATE $collate");

  # Bug 321645 - disable MySQL strict mode, if set
  my ($var, $sql_mode)
    = $dbh->selectrow_array("SHOW VARIABLES LIKE 'sql\\_mode'");

  if ($sql_mode) {

    # STRICT_TRANS_TABLE or STRICT_ALL_TABLES enable MySQL strict mode,
    # causing bug 321645. TRADITIONAL sets these modes (among others) as
    # well, so it has to be stipped as well
    my $new_sql_mode = join(",",
      grep { $_ !~ /^STRICT_(?:TRANS|ALL)_TABLES|TRADITIONAL$/ }
        split(/,/, $sql_mode));

    if ($sql_mode ne $new_sql_mode) {
      $dbh->do("SET SESSION sql_mode = ?", undef, $new_sql_mode);
    }
  }

  # Allow large GROUP_CONCATs (largely for inserting comments
  # into bugs_fulltext).
  $dbh->do('SET SESSION group_concat_max_len = 128000000');
}

# when last_insert_id() is supported on MySQL by lowest DBI/DBD version
# required by Bugzilla, this implementation can be removed.
sub bz_last_key {
  my ($self) = @_;

  my ($last_insert_id) = $self->selectrow_array('SELECT LAST_INSERT_ID()');

  return $last_insert_id;
}

sub sql_group_concat {
  my ($self, $column, $separator, $sort) = @_;
  $separator = $self->quote(', ') if !defined $separator;
  $sort = 1 if !defined $sort;
  if ($sort) {
    my $sort_order = $column;
    $sort_order =~ s/^DISTINCT\s+//i;
    $column = "$column ORDER BY $sort_order";
  }
  return "GROUP_CONCAT($column SEPARATOR $separator)";
}

sub sql_regexp {
  my ($self, $expr, $pattern, $nocheck, $real_pattern) = @_;
  $real_pattern ||= $pattern;

  $self->bz_check_regexp($real_pattern) if !$nocheck;

  return "$expr REGEXP $pattern";
}

sub sql_not_regexp {
  my ($self, $expr, $pattern, $nocheck, $real_pattern) = @_;
  $real_pattern ||= $pattern;

  $self->bz_check_regexp($real_pattern) if !$nocheck;

  return "$expr NOT REGEXP $pattern";
}

sub sql_limit {
  my ($self, $limit, $offset) = @_;

  if (defined($offset)) {
    return "LIMIT $offset, $limit";
  }
  else {
    return "LIMIT $limit";
  }
}

sub sql_string_concat {
  my ($self, @params) = @_;

  return 'CONCAT(' . join(', ', @params) . ')';
}

sub sql_fulltext_search {
  my ($self, $column, $text) = @_;

  # Add the boolean mode modifier if the search string contains
  # boolean operators at the start or end of a word.
  my $mode = '';
  if ($text =~ /(?:^|\W)[+\-<>~"()]/ || $text =~ /[()"*](?:$|\W)/) {
    $mode = 'IN BOOLEAN MODE';

    my @terms = split(quotemeta(FULLTEXT_OR), $text);
    foreach my $term (@terms) {

      # quote un-quoted compound words
      my @words = grep {defined} quotewords('[\s()]+', 'delimiters', $term);
      foreach my $word (@words) {

        # match words that have non-word chars in the middle of them
        if ($word =~ /\w\W+\w/ && $word !~ m/"/) {
          $word = '"' . $word . '"';

          # match words that contain only boolean operators
        }
        elsif ($word =~ /^[\+\-\<\>\~\*]+$/) {
          $word = '"' . $word . '"';
        }
      }
      $term = join('', @words);
    }
    $text = join(FULLTEXT_OR, @terms);
  }

  # quote the text for use in the MATCH AGAINST expression
  $text = $self->quote($text);

  # untaint the text, since it's safe to use now that we've quoted it
  trick_taint($text);

  return "MATCH($column) AGAINST($text $mode)";
}

sub sql_istring {
  my ($self, $string) = @_;

  return $string;
}

sub sql_from_days {
  my ($self, $days) = @_;

  return "FROM_DAYS($days)";
}

sub sql_to_days {
  my ($self, $date) = @_;

  return "TO_DAYS($date)";
}

sub sql_date_format {
  my ($self, $date, $format) = @_;

  $format = "%Y.%m.%d %H:%i:%s" if !$format;

  return "DATE_FORMAT($date, " . $self->quote($format) . ")";
}

sub sql_date_math {
  my ($self, $date, $operator, $interval, $units) = @_;

  return "$date $operator INTERVAL $interval $units";
}

sub sql_iposition {
  my ($self, $fragment, $text) = @_;
  return "INSTR($text, $fragment)";
}

sub sql_position {
  my ($self, $fragment, $text) = @_;

  return "INSTR(CAST($text AS BINARY), CAST($fragment AS BINARY))";
}

sub sql_group_by {
  my ($self, $needed_columns, $optional_columns) = @_;

  # MySQL allows you to specify the minimal subset of columns to get
  # a unique result. While it does allow specifying all columns as
  # ANSI SQL requires, according to MySQL documentation, the fewer
  # columns you specify, the faster the query runs.
  return "GROUP BY $needed_columns";
}

sub bz_explain {
  my ($self, $sql) = @_;
  my $sth = $self->prepare("EXPLAIN $sql");
  $sth->execute();
  my $columns       = $sth->{'NAME'};
  my $lengths       = $sth->{'mysql_max_length'};
  my $format_string = '|';
  my $i             = 0;
  foreach my $column (@$columns) {

    # Sometimes the column name is longer than the contents.
    my $length = max($lengths->[$i], length($column));
    $format_string .= ' %-' . $length . 's |';
    $i++;
  }

  my $first_row = sprintf($format_string, @$columns);
  my @explain_rows = ($first_row, '-' x length($first_row));
  while (my $row = $sth->fetchrow_arrayref) {
    my @fixed = map { defined $_ ? $_ : 'NULL' } @$row;
    push(@explain_rows, sprintf($format_string, @fixed));
  }

  return join("\n", @explain_rows);
}

sub _bz_get_initial_schema {
  my ($self) = @_;
  return $self->_bz_build_schema_from_disk();
}

#####################################################################
# Database Setup
#####################################################################

sub bz_check_server_version {
  my $self = shift;

  my $lc = Bugzilla->localconfig;
  if (lc(Bugzilla->localconfig->{db_name}) eq 'mysql') {
    die "It is not safe to run Bugzilla inside a database named 'mysql'.\n"
      . " Please pick a different value for \$db_name in localconfig.\n";
  }

  $self->SUPER::bz_check_server_version(@_);
}

sub bz_setup_database {
  my ($self) = @_;

  # The "comments" field of the bugs_fulltext table could easily exceed
  # MySQL's default max_allowed_packet. Also, MySQL should never have
  # a max_allowed_packet smaller than our max_attachment_size. So, we
  # warn the user here if max_allowed_packet is too small.
  my $min_max_allowed = MAX_COMMENTS * MAX_COMMENT_LENGTH;
  my (undef, $current_max_allowed)
    = $self->selectrow_array(q{SHOW VARIABLES LIKE 'max\_allowed\_packet'});

  # This parameter is not yet defined when the DB is being built for
  # the very first time. The code below still works properly, however,
  # because the default maxattachmentsize is smaller than $min_max_allowed.
  my $max_attachment = (Bugzilla->params->{'maxattachmentsize'} || 0) * 1024;
  my $needed_max_allowed = max($min_max_allowed, $max_attachment);
  if ($current_max_allowed < $needed_max_allowed) {
    warn install_string('max_allowed_packet',
      {current => $current_max_allowed, needed => $needed_max_allowed})
      . "\n";
  }

  # Make sure the installation has InnoDB turned on, or we're going to be
  # doing silly things like making foreign keys on MyISAM tables, which is
  # hard to fix later. We do this up here because none of the code below
  # works if InnoDB is off. (Particularly if we've already converted the
  # tables to InnoDB.)
  my %engines = @{$self->selectcol_arrayref('SHOW ENGINES', {Columns => [1, 2]})};
  if (!$engines{InnoDB} || $engines{InnoDB} !~ /^(YES|DEFAULT)$/) {
    die install_string('mysql_innodb_disabled');
  }

  if ($self->utf8_charset eq 'utf8mb4') {
    my %global = map {@$_}
      @{$self->selectall_arrayref(q(SHOW GLOBAL VARIABLES LIKE 'innodb_%'))};
    my $utf8mb4_supported
      = $global{innodb_file_format} eq 'Barracuda'
      && $global{innodb_file_per_table} eq 'ON'
      && $global{innodb_large_prefix} eq 'ON';

    die install_string('mysql_innodb_settings') unless $utf8mb4_supported;

    my $tables = $self->selectall_arrayref('SHOW TABLE STATUS');
    foreach my $table (@$tables) {
      my ($table, undef, undef, $row_format) = @$table;
      my $new_row_format = $self->default_row_format($table);
      next if $new_row_format =~ /compact/i;
      if (lc($new_row_format) ne lc($row_format)) {
        print install_string(
          'mysql_row_format_conversion', {table => $table, format => $new_row_format}
          ),
          "\n";
        $self->do(sprintf 'ALTER TABLE %s ROW_FORMAT=%s', $table, $new_row_format);
      }
    }
  }

  my ($sd_index_deleted, $longdescs_index_deleted);
  my @tables = $self->bz_table_list_real();

  # We want to convert tables to InnoDB, but it's possible that they have
  # fulltext indexes on them, and conversion will fail unless we remove
  # the indexes.
  if (grep($_ eq 'bugs', @tables) and !grep($_ eq 'bugs_fulltext', @tables)) {
    if ($self->bz_index_info_real('bugs', 'short_desc')) {
      $self->bz_drop_index_raw('bugs', 'short_desc');
    }
    if ($self->bz_index_info_real('bugs', 'bugs_short_desc_idx')) {
      $self->bz_drop_index_raw('bugs', 'bugs_short_desc_idx');
      $sd_index_deleted = 1;    # Used for later schema cleanup.
    }
  }
  if (grep($_ eq 'longdescs', @tables) and !grep($_ eq 'bugs_fulltext', @tables))
  {
    if ($self->bz_index_info_real('longdescs', 'thetext')) {
      $self->bz_drop_index_raw('longdescs', 'thetext');
    }
    if ($self->bz_index_info_real('longdescs', 'longdescs_thetext_idx')) {
      $self->bz_drop_index_raw('longdescs', 'longdescs_thetext_idx');
      $longdescs_index_deleted = 1;    # For later schema cleanup.
    }
  }

  # Upgrade tables from MyISAM to InnoDB
  my $db_name       = Bugzilla->localconfig->{db_name};
  my $myisam_tables = $self->selectcol_arrayref(
    'SELECT TABLE_NAME FROM information_schema.TABLES
          WHERE TABLE_SCHEMA = ? AND ENGINE = ?', undef, $db_name, 'MyISAM'
  );

  if (scalar @$myisam_tables) {
    print "Bugzilla now uses the InnoDB storage engine in MySQL for",
      " most tables.\nConverting tables to InnoDB:\n";
    foreach my $table (@$myisam_tables) {
      print "Converting table $table... ";
      $self->do("ALTER TABLE $table ENGINE = InnoDB");
      print "done.\n";
    }
  }

  # Versions of Bugzilla before the existence of Bugzilla::DB::Schema did
  # not provide explicit names for the table indexes. This means
  # that our upgrades will not be reliable, because we look for the name
  # of the index, not what fields it is on, when doing upgrades.
  # (using the name is much better for cross-database compatibility
  # and general reliability). It's also very important that our
  # Schema object be consistent with what is on the disk.
  #
  # While we're at it, we also fix some inconsistent index naming
  # from the original checkin of Bugzilla::DB::Schema.

  # We check for the existence of a particular "short name" index that
  # has existed at least since Bugzilla 2.8, and probably earlier.
  # For fixing the inconsistent naming of Schema indexes,
  # we also check for one of those inconsistently-named indexes.
  if (
    grep($_ eq 'bugs', @tables)
    && ( $self->bz_index_info_real('bugs', 'assigned_to')
      || $self->bz_index_info_real('flags', 'flags_bidattid_idx'))
    )
  {

    # This is a check unrelated to the indexes, to see if people are
    # upgrading from 2.18 or below, but somehow have a bz_schema table
    # already. This only happens if they have done a mysqldump into
    # a database without doing a DROP DATABASE first.
    # We just do the check here since this check is a reliable way
    # of telling that we are upgrading from a version pre-2.20.
    if (grep($_ eq 'bz_schema', $self->bz_table_list_real())) {
      die install_string('bz_schema_exists_before_220');
    }

    my $bug_count = $self->selectrow_array("SELECT COUNT(*) FROM bugs");

    # We estimate one minute for each 3000 bugs, plus 3 minutes just
    # to handle basic MySQL stuff.
    my $rename_time = int($bug_count / 3000) + 3;

    # And 45 minutes for every 15,000 attachments, per some experiments.
    my ($attachment_count)
      = $self->selectrow_array("SELECT COUNT(*) FROM attachments");
    $rename_time += int(($attachment_count * 45) / 15000);

    # If we're going to take longer than 5 minutes, we let the user know
    # and allow them to abort.
    if ($rename_time > 5) {
      print "\n", install_string('mysql_index_renaming', {minutes => $rename_time});

      # Wait 45 seconds for them to respond.
      sleep(45) unless Bugzilla->installation_answers->{NO_PAUSE};
    }
    print "Renaming indexes...\n";

    # We can't be interrupted, because of how the "if"
    # works above.
    local $SIG{INT}  = 'IGNORE';
    local $SIG{TERM} = 'IGNORE';
    local $SIG{PIPE} = 'IGNORE';

    # Certain indexes had names in Schema that did not easily conform
    # to a standard. We store those names here, so that they
    # can be properly renamed.
    # Also, sometimes an old mysqldump would incorrectly rename
    # unique indexes to "PRIMARY", so we address that here, also.
    my $bad_names = {

      # 'when' is a possible leftover from Bugzillas before 2.8
      bugs_activity =>
        ['when', 'bugs_activity_bugid_idx', 'bugs_activity_bugwhen_idx'],
      cc                 => ['PRIMARY'],
      longdescs          => ['longdescs_bugid_idx', 'longdescs_bugwhen_idx'],
      flags              => ['flags_bidattid_idx'],
      flaginclusions     => ['flaginclusions_tpcid_idx'],
      flagexclusions     => ['flagexclusions_tpc_id_idx'],
      keywords           => ['PRIMARY'],
      milestones         => ['PRIMARY'],
      profiles_activity  => ['profiles_activity_when_idx'],
      group_control_map  => ['group_control_map_gid_idx', 'PRIMARY'],
      user_group_map     => ['PRIMARY'],
      group_group_map    => ['PRIMARY'],
      email_setting      => ['PRIMARY'],
      bug_group_map      => ['PRIMARY'],
      category_group_map => ['PRIMARY'],
      watch              => ['PRIMARY'],
      namedqueries       => ['PRIMARY'],
      series_data        => ['PRIMARY'],

      # series_categories is dealt with below, not here.
    };

    # The series table is broken and needs to have one index
    # dropped before we begin the renaming, because it had a
    # useless index on it that would cause a naming conflict here.
    if (grep($_ eq 'series', @tables)) {
      my $dropname;

      # This is what the bad index was called before Schema.
      if ($self->bz_index_info_real('series', 'creator_2')) {
        $dropname = 'creator_2';
      }

      # This is what the bad index is called in Schema.
      elsif ($self->bz_index_info_real('series', 'series_creator_idx')) {
        $dropname = 'series_creator_idx';
      }
      $self->bz_drop_index_raw('series', $dropname) if $dropname;
    }

    # The email_setting table also had the same problem.
    if (grep($_ eq 'email_setting', @tables)
      && $self->bz_index_info_real('email_setting', 'email_settings_user_id_idx'))
    {
      $self->bz_drop_index_raw('email_setting', 'email_settings_user_id_idx');
    }

    # Go through all the tables.
    foreach my $table (@tables) {

      # Will contain the names of old indexes as keys, and the
      # definition of the new indexes as a value. The values
      # include an extra hash key, NAME, with the new name of
      # the index.
      my %rename_indexes;

      # And go through all the columns on each table.
      my @columns = $self->bz_table_columns_real($table);

      # We also want to fix the silly naming of unique indexes
      # that happened when we first checked-in Bugzilla::DB::Schema.
      if ($table eq 'series_categories') {

        # The series_categories index had a nonstandard name.
        push(@columns, 'series_cats_unique_idx');
      }
      elsif ($table eq 'email_setting') {

        # The email_setting table had a similar problem.
        push(@columns, 'email_settings_unique_idx');
      }
      else {
        push(@columns, "${table}_unique_idx");
      }

      # And this is how we fix the other inconsistent Schema naming.
      push(@columns, @{$bad_names->{$table}}) if (exists $bad_names->{$table});
      foreach my $column (@columns) {

        # If we have an index named after this column, it's an
        # old-style-name index.
        if (my $index = $self->bz_index_info_real($table, $column)) {

          # Fix the name to fit in with the new naming scheme.
          $index->{NAME} = $table . "_" . $index->{FIELDS}->[0] . "_idx";
          print "Renaming index $column to " . $index->{NAME} . "...\n";
          $rename_indexes{$column} = $index;
        }    # if
      }    # foreach column

      my @rename_sql
        = $self->_bz_schema->get_rename_indexes_ddl($table, %rename_indexes);
      $self->do($_) foreach (@rename_sql);

    }    # foreach table
  }    # if old-name indexes

  # If there are no tables, but the DB isn't utf8 and it should be,
  # then we should alter the database to be utf8. We know it should be
  # if the utf8 parameter is true or there are no params at all.
  # This kind of situation happens when people create the database
  # themselves, and if we don't do this they will get the big
  # scary WARNING statement about conversion to UTF8.
  unless ($self->bz_db_is_utf8) {
    $self->_alter_db_charset_to_utf8();
  }

  # And now we create the tables and the Schema object.
  $self->SUPER::bz_setup_database();

  if ($sd_index_deleted) {
    $self->_bz_real_schema->delete_index('bugs', 'bugs_short_desc_idx');
    $self->_bz_store_real_schema;
  }
  if ($longdescs_index_deleted) {
    $self->_bz_real_schema->delete_index('longdescs', 'longdescs_thetext_idx');
    $self->_bz_store_real_schema;
  }

  # 2005-09-24 - bugreport@peshkin.net, bug 307602
  # Make sure that default 4G table limit is overridden
  my $attach_data_create = $self->selectrow_array(
    'SELECT CREATE_OPTIONS FROM information_schema.TABLES
          WHERE TABLE_SCHEMA = ? AND TABLE_NAME = ?', undef, $db_name, 'attach_data'
  );
  if ($attach_data_create !~ /MAX_ROWS/i) {
    print "Converting attach_data maximum size to 100G...\n";
    $self->do(
      "ALTER TABLE attach_data
                   AVG_ROW_LENGTH=1000000,
                   MAX_ROWS=100000"
    );
  }

  # Convert the database to UTF-8 if the utf8 parameter is on.
  # We check if any table isn't utf8, because lots of crazy
  # partial-conversion situations can happen, and this handles anything
  # that could come up (including having the DB charset be utf8 but not
  # the table charsets.
  #
  # TABLE_COLLATION IS NOT NULL prevents us from trying to convert views.
  my $charset         = $self->utf8_charset;
  my $collate         = $self->utf8_collate;
  my $non_utf8_tables = $self->selectrow_array(
    "SELECT 1 FROM information_schema.TABLES
          WHERE TABLE_SCHEMA = ? AND TABLE_COLLATION IS NOT NULL
                AND TABLE_COLLATION != ?
          LIMIT 1", undef, $db_name, $collate
  );

  if (Bugzilla->params->{'utf8'} && $non_utf8_tables) {
    print "\n", install_string('mysql_utf8_conversion');

    if (!Bugzilla->installation_answers->{NO_PAUSE}) {
      if (Bugzilla->installation_mode == INSTALLATION_MODE_NON_INTERACTIVE) {
        die install_string('continue_without_answers'), "\n";
      }
      else {
        print "\n         " . install_string('enter_or_ctrl_c');
        getc;
      }
    }

    print
      "Converting table storage format to $charset (collate $collate). This may take a while.\n";
    foreach my $table ($self->bz_table_list_real) {
      my $info_sth = $self->prepare("SHOW FULL COLUMNS FROM $table");
      $info_sth->execute();
      my (@binary_sql, @utf8_sql);
      while (my $column = $info_sth->fetchrow_hashref) {

        # Our conversion code doesn't work on enum fields, but they
        # all go away later in checksetup anyway.
        next if $column->{Type} =~ /enum/i;

        # If this particular column isn't stored in utf-8
        if ( $column->{Collation}
          && $column->{Collation} ne 'NULL'
          && $column->{Collation} ne $collate)
        {
          my $name = $column->{Field};

          print "$table.$name needs to be converted to $charset (collate $collate)...\n";

          # These will be automatically re-created at the end
          # of checksetup.
          $self->bz_drop_related_fks($table, $name);

          my $col_info = $self->bz_column_info_real($table, $name);

          # CHANGE COLUMN doesn't take PRIMARY KEY
          delete $col_info->{PRIMARYKEY};
          my $sql_def = $self->_bz_schema->get_type_ddl($col_info);

          # We don't want MySQL to actually try to *convert*
          # from our current charset to UTF-8, we just want to
          # transfer the bytes directly. This is how we do that.

          # The CHARACTER SET part of the definition has to come
          # right after the type, which will always come first.
          my ($binary, $utf8) = ($sql_def, $sql_def);
          my $type = $self->_bz_schema->convert_type($col_info->{TYPE});
          $binary =~ s/(\Q$type\E)/$1 CHARACTER SET binary/;
          $utf8 =~ s/(\Q$type\E)/$1 CHARACTER SET $charset COLLATE $collate/;
          push(@binary_sql, "MODIFY COLUMN $name $binary");
          push(@utf8_sql,   "MODIFY COLUMN $name $utf8");
        }
      }    # foreach column

      if (@binary_sql) {
        my %indexes = %{$self->bz_table_indexes($table)};
        foreach my $index_name (keys %indexes) {
          my $index = $indexes{$index_name};
          if ($index->{TYPE} and $index->{TYPE} eq 'FULLTEXT') {
            $self->bz_drop_index($table, $index_name);
          }
          else {
            delete $indexes{$index_name};
          }
        }

        print "Converting the $table table to UTF-8...\n";
        my $bin = "ALTER TABLE $table " . join(', ', @binary_sql);
        my $utf = "ALTER TABLE $table "
          . join(', ', @utf8_sql, "DEFAULT CHARACTER SET $charset COLLATE $collate");
        $self->do($bin);
        $self->do($utf);

        # Re-add any removed FULLTEXT indexes.
        foreach my $index (keys %indexes) {
          $self->bz_add_index($table, $index, $indexes{$index});
        }
      }
      else {
        $self->do("ALTER TABLE $table DEFAULT CHARACTER SET $charset COLLATE $collate");
      }

    }    # foreach my $table (@tables)
  }

  # Sometimes you can have a situation where all the tables are utf8,
  # but the database isn't. (This tends to happen when you've done
  # a mysqldump.) So we have this change outside of the above block,
  # so that it just happens silently if no actual *table* conversion
  # needs to happen.
  unless ($self->bz_db_is_utf8) {
    $self->_alter_db_charset_to_utf8();
  }

  $self->_fix_defaults();

  # Bug 451735 highlighted a bug in bz_drop_index() which didn't
  # check for FKs before trying to delete an index. Consequently,
  # the series_creator_idx index was considered to be deleted
  # despite it was still present in the DB. That's why we have to
  # force the deletion, bypassing the DB schema.
  if (!$self->bz_index_info('series', 'series_category_idx')) {
    if (!$self->bz_index_info('series', 'series_creator_idx')
      && $self->bz_index_info_real('series', 'series_creator_idx'))
    {
      foreach my $column (qw(creator category subcategory name)) {
        $self->bz_drop_related_fks('series', $column);
      }
      $self->bz_drop_index_raw('series', 'series_creator_idx');
    }
  }
}

# When you import a MySQL 3/4 mysqldump into MySQL 5, columns that
# aren't supposed to have defaults will have defaults. This is only
# a minor issue, but it makes our tests fail, and it's good to keep
# the DB actually consistent with what DB::Schema thinks the database
# looks like. So we remove defaults from columns that aren't supposed
# to have them
sub _fix_defaults {
  my $self = shift;
  my $maj_version = substr($self->bz_server_version, 0, 1);
  return if $maj_version < 5;

  # The oldest column that could have this problem is bugs.assigned_to,
  # so if it doesn't have the problem, we just skip doing this entirely.
  my $assi_def = $self->_bz_raw_column_info('bugs', 'assigned_to');
  my $assi_default = $assi_def->{COLUMN_DEF};

  # This "ne ''" thing is necessary because _raw_column_info seems to
  # return COLUMN_DEF as an empty string for columns that don't have
  # a default.
  return unless (defined $assi_default && $assi_default ne '');

  my %fix_columns;
  foreach my $table ($self->_bz_real_schema->get_table_list()) {
    foreach my $column ($self->bz_table_columns($table)) {
      my $abs_def = $self->bz_column_info($table, $column);

      # BLOB/TEXT columns never have defaults
      next if $abs_def->{TYPE} =~ /BLOB|TEXT/i;
      if (!defined $abs_def->{DEFAULT}) {

        # Get the exact default from the database without any
        # "fixing" by bz_column_info_real.
        my $raw_info = $self->_bz_raw_column_info($table, $column);
        my $raw_default = $raw_info->{COLUMN_DEF};
        if (defined $raw_default) {
          if ($raw_default eq '') {

            # Only (var)char columns can have empty strings as
            # defaults, so if we got an empty string for some
            # other default type, then it's bogus.
            next unless $abs_def->{TYPE} =~ /char/i;
            $raw_default = "''";
          }
          $fix_columns{$table} ||= [];
          push(@{$fix_columns{$table}}, $column);
          print "$table.$column has incorrect DB default: $raw_default\n";
        }
      }
    }    # foreach $column
  }    # foreach $table

  print "Fixing defaults...\n";
  foreach my $table (reverse sort keys %fix_columns) {
    my @alters = map("ALTER COLUMN $_ DROP DEFAULT", @{$fix_columns{$table}});
    my $sql = "ALTER TABLE $table " . join(',', @alters);
    $self->do($sql);
  }
}

sub utf8_charset {
  return 'utf8' unless Bugzilla->params->{'utf8'};
  return Bugzilla->params->{'utf8'} eq 'utf8mb4' ? 'utf8mb4' : 'utf8';
}

sub utf8_collate {
  my $charset = utf8_charset();
  if ($charset eq 'utf8') {
    return 'utf8_general_ci';
  }
  elsif ($charset eq 'utf8mb4') {
    return 'utf8mb4_unicode_520_ci';
  }
  else {
    croak "invalid charset: $charset";
  }
}

sub default_row_format {
  my ($class, $table) = @_;
  my $charset = utf8_charset();
  if ($charset eq 'utf8') {
    return 'Compact';
  }
  elsif ($charset eq 'utf8mb4') {
    my @no_compress = qw(
      bug_user_last_visit
      cc
      email_rates
      logincookies
      token_data
      tokens
      ts_error
      ts_exitstatus
      ts_funcmap
      ts_job
      ts_note
      user_request_log
      votes
    );
    return 'Dynamic' if any { $table eq $_ } @no_compress;
    return 'Compressed';
  }
  else {
    croak "invalid charset: $charset";
  }
}

sub _alter_db_charset_to_utf8 {
  my $self    = shift;
  my $db_name = Bugzilla->localconfig->{db_name};
  my $charset = $self->utf8_charset;
  my $collate = $self->utf8_collate;
  $self->do("ALTER DATABASE $db_name CHARACTER SET $charset COLLATE $collate");
}

sub bz_db_is_utf8 {
  my $self = shift;
  my $db_charset
    = $self->selectrow_arrayref("SHOW VARIABLES LIKE 'character_set_database'");

  # First column holds the variable name, second column holds the value.
  my $charset = $self->utf8_charset;
  return $db_charset->[1] eq $charset ? 1 : 0;
}


sub bz_enum_initial_values {
  my ($self) = @_;
  my %enum_values = %{$self->ENUM_DEFAULTS};

  # Get a complete description of the 'bugs' table; with DBD::MySQL
  # there isn't a column-by-column way of doing this.  Could use
  # $dbh->column_info, but it would go slower and we would have to
  # use the undocumented mysql_type_name accessor to get the type
  # of each row.
  my $sth = $self->prepare("DESCRIBE bugs");
  $sth->execute();

  # Look for the particular columns we are interested in.
  while (my ($thiscol, $thistype) = $sth->fetchrow_array()) {
    if (defined $enum_values{$thiscol}) {

      # this is a column of interest.
      my @value_list;
      if ($thistype and ($thistype =~ /^enum\(/)) {

        # it has an enum type; get the set of values.
        while ($thistype =~ /'([^']*)'(.*)/) {
          push(@value_list, $1);
          $thistype = $2;
        }
      }
      if (@value_list) {

        # record the enum values found.
        $enum_values{$thiscol} = \@value_list;
      }
    }
  }

  return \%enum_values;
}

#####################################################################
# MySQL-specific Database-Reading Methods
#####################################################################

=begin private

=head1 MYSQL-SPECIFIC DATABASE-READING METHODS

These methods read information about the database from the disk,
instead of from a Schema object. They are only reliable for MySQL
(see bug 285111 for the reasons why not all DBs use/have functions
like this), but that's OK because we only need them for
backwards-compatibility anyway, for versions of Bugzilla before 2.20.

=over 4

=item C<bz_column_info_real($table, $column)>

 Description: Returns an abstract column definition for a column
              as it actually exists on disk in the database.
 Params:      $table - The name of the table the column is on.
              $column - The name of the column you want info about.
 Returns:     An abstract column definition.
              If the column does not exist, returns undef.

=cut

sub bz_column_info_real {
  my ($self, $table, $column) = @_;
  my $col_data = $self->_bz_raw_column_info($table, $column);
  return $self->_bz_schema->column_info_to_column($col_data);
}

sub _bz_raw_column_info {
  my ($self, $table, $column) = @_;

  # DBD::mysql does not support selecting a specific column,
  # so we have to get all the columns on the table and find
  # the one we want.
  my $info_sth = $self->column_info(undef, undef, $table, '%');

  # Don't use fetchall_hashref as there's a Win32 DBI bug (292821)
  my $col_data;
  while ($col_data = $info_sth->fetchrow_hashref) {
    last if $col_data->{'COLUMN_NAME'} eq $column;
  }

  if (!defined $col_data) {
    return undef;
  }
  return $col_data;
}

=item C<bz_index_info_real($table, $index)>

 Description: Returns information about an index on a table in the database.
 Params:      $table = name of table containing the index
              $index = name of an index
 Returns:     An abstract index definition, always in hashref format.
              If the index does not exist, the function returns undef.

=cut

sub bz_index_info_real {
  my ($self, $table, $index) = @_;

  my $sth = $self->prepare("SHOW INDEX FROM $table");
  $sth->execute;

  my @fields;
  my $index_type;

  # $raw_def will be an arrayref containing the following information:
  # 0 = name of the table that the index is on
  # 1 = 0 if unique, 1 if not unique
  # 2 = name of the index
  # 3 = seq_in_index (The order of the current field in the index).
  # 4 = Name of ONE column that the index is on
  # 5 = 'Collation' of the index. Usually 'A'.
  # 6 = Cardinality. Either a number or undef.
  # 7 = sub_part. Usually undef. Sometimes 1.
  # 8 = "packed". Usually undef.
  # 9 = Null. Sometimes undef, sometimes 'YES'.
  # 10 = Index_type. The type of the index. Usually either 'BTREE' or 'FULLTEXT'
  # 11 = 'Comment.' Usually undef.
  while (my $raw_def = $sth->fetchrow_arrayref) {
    if ($raw_def->[2] eq $index) {
      push(@fields, $raw_def->[4]);

      # No index can be both UNIQUE and FULLTEXT, that's why
      # this is written this way.
      $index_type = $raw_def->[1] ? '' : 'UNIQUE';
      $index_type = $raw_def->[10] eq 'FULLTEXT' ? 'FULLTEXT' : $index_type;
    }
  }

  my $retval;
  if (scalar(@fields)) {
    $retval = {FIELDS => \@fields, TYPE => $index_type};
  }
  return $retval;
}

=item C<bz_index_list_real($table)>

 Description: Returns a list of index names on a table in
              the database, as it actually exists on disk.
 Params:      $table - The name of the table you want info about.
 Returns:     An array of index names.

=cut

sub bz_index_list_real {
  my ($self, $table) = @_;
  my $sth = $self->prepare("SHOW INDEX FROM $table");

  # Column 3 of a SHOW INDEX statement contains the name of the index.
  return @{$self->selectcol_arrayref($sth, {Columns => [3]})};
}

#####################################################################
# MySQL-Specific "Schema Builder"
#####################################################################

=back

=head1 MYSQL-SPECIFIC "SCHEMA BUILDER"

MySQL needs to be able to read in a legacy database (from before
Schema existed) and create a Schema object out of it. That's what
this code does.

=end private

=cut

# This sub itself is actually written generically, but the subroutines
# that it depends on are database-specific. In particular, the
# bz_column_info_real function would be very difficult to create
# properly for any other DB besides MySQL.
sub _bz_build_schema_from_disk {
  my ($self) = @_;

  my $schema = $self->_bz_schema->get_empty_schema();

  my @tables = $self->bz_table_list_real();
  if (@tables) {
    print "Building Schema object from database...\n";
  }
  foreach my $table (@tables) {
    $schema->add_table($table);
    my @columns = $self->bz_table_columns_real($table);
    foreach my $column (@columns) {
      my $type_info = $self->bz_column_info_real($table, $column);
      $schema->set_column($table, $column, $type_info);
    }

    my @indexes = $self->bz_index_list_real($table);
    foreach my $index (@indexes) {
      unless ($index eq 'PRIMARY') {
        my $index_info = $self->bz_index_info_real($table, $index);
        ($index_info = $index_info->{FIELDS}) if (!$index_info->{TYPE});
        $schema->set_index($table, $index, $index_info);
      }
    }
  }

  return $schema;
}

1;