maintenance/postgres/compare_schemas.pl

   1 #!/usr/bin/perl
   2
   3 ## Rough check that the base and postgres "tables.sql" are in sync
   4 ## Should be run from maintenance/postgres
   5
   6 use strict;
   7 use warnings;
   8 use Data::Dumper;
   9
  10 my @old = ("../tables.sql", "../mysql5/tables.sql");
  11 my $new = "tables.sql";
  12 my @xfile;
  13
  14 ## Read in exceptions and other metadata
  15 my %ok;
  16 while (<DATA>) {
  17         next unless /^(\w+)\s*:\s*([^#]+)/;
  18         my ($name,$val) = ($1,$2);
  19         chomp $val;
  20         if ($name eq 'RENAME') {
  21                 die "Invalid rename\n" unless $val =~ /(\w+)\s+(\w+)/;
  22                 $ok{OLD}{$1} = $2;
  23                 $ok{NEW}{$2} = $1;
  24                 next;
  25         }
  26         if ($name eq 'XFILE') {
  27                 push @xfile, $val;
  28                 next;
  29         }
  30         for (split(/\s+/ => $val)) {
  31                 $ok{$name}{$_} = 0;
  32         }
  33 }
  34
  35 my $datatype = join '|' => qw(
  36 bool
  37 tinyint int bigint real float
  38 tinytext mediumtext text char varchar varbinary
  39 timestamp datetime
  40 tinyblob mediumblob blob
  41 );
  42 $datatype .= q{|ENUM\([\"\w, ]+\)};
  43 $datatype = qr{($datatype)};
  44
  45 my $typeval = qr{(\(\d+\))?};
  46
  47 my $typeval2 = qr{ unsigned| binary| NOT NULL| NULL| auto_increment| default ['\-\d\w"]+| REFERENCES .+CASCADE};
  48
  49 my $indextype = join '|' => qw(INDEX KEY FULLTEXT), "PRIMARY KEY", "UNIQUE INDEX", "UNIQUE KEY";
  50 $indextype = qr{$indextype};
  51
  52 my $engine = qr{TYPE|ENGINE};
  53
  54 my $tabletype = qr{InnoDB|MyISAM|HEAP|HEAP MAX_ROWS=\d+};
  55
  56 my $charset = qr{utf8};
  57
  58
  59 open my $newfh, "<", $new or die qq{Could not open $new: $!\n};
  60
  61
  62 my ($table,%old);
  63
  64 ## Read in the xfiles
  65 my %xinfo;
  66 for my $xfile (@xfile) {
  67         print "Loading $xfile\n";
  68         my $info = &parse_sql($xfile);
  69         for (keys %$info) {
  70                 $xinfo{$_} = $info->{$_};
  71         }
  72 }
  73
  74 for my $oldfile (@old) {
  75         print "Loading $oldfile\n";
  76         my $info = &parse_sql($oldfile);
  77         for (keys %xinfo) {
  78                 $info->{$_} = $xinfo{$_};
  79         }
  80         $old{$oldfile} = $info;
  81 }
  82
  83 sub parse_sql {
  84
  85         my $oldfile = shift;
  86
  87         open my $oldfh, "<", $oldfile or die qq{Could not open $oldfile: $!\n};
  88
  89         my %info;
  90         while (<$oldfh>) {
  91                 next if /^\s*\-\-/ or /^\s+$/;
  92                 s/\s*\-\- [\w ]+$//;
  93                 chomp;
  94
  95                 if (/CREATE\s*TABLE/i) {
  96                         m{^CREATE TABLE /\*\$wgDBprefix\*/(\w+) \($}
  97                                 or die qq{Invalid CREATE TABLE at line $. of $oldfile\n};
  98                         $table = $1;
  99                         $info{$table}{name}=$table;
 100                 }
 101                 elsif (/^\) ($engine)=($tabletype);$/) {
 102                         $info{$table}{engine}=$1;
 103                         $info{$table}{type}=$2;
 104                 }
 105                 elsif (/^\) ($engine)=($tabletype), DEFAULT CHARSET=($charset);$/) {
 106                         $info{$table}{engine}=$1;
 107                         $info{$table}{type}=$2;
 108                         $info{$table}{charset}=$3;
 109                 }
 110                 elsif (/^  (\w+) $datatype$typeval$typeval2{0,3},?$/) {
 111                         $info{$table}{column}{$1} = $2;
 112                 }
 113                 elsif (/^  ($indextype)(?: (\w+))? \(([\w, \(\)]+)\),?$/) {
 114                         $info{$table}{lc $1."_name"} = $2 ? $2 : "";
 115                         $info{$table}{lc $1."pk_target"} = $3;
 116                 }
 117                 else {
 118                         die "Cannot parse line $. of $oldfile:\n$_\n";
 119                 }
 120
 121         }
 122         close $oldfh;
 123
 124         return \%info;
 125
 126 } ## end of parse_sql
 127
 128 for my $oldfile (@old) {
 129
 130 ## Begin non-standard indent
 131
 132 ## MySQL sanity checks
 133 for my $table (sort keys %{$old{$oldfile}}) {
 134         my $t = $old{$oldfile}{$table};
 135         if (($oldfile =~ /5/ and $t->{engine} ne 'ENGINE')
 136                 or
 137                 ($oldfile !~ /5/ and $t->{engine} ne 'TYPE')) {
 138                 die "Invalid engine for $oldfile: $t->{engine}\n" unless $t->{name} eq 'profiling';
 139         }
 140 }
 141
 142 my $dtype = join '|' => qw(
 143 SMALLINT INTEGER BIGINT NUMERIC SERIAL
 144 TEXT CHAR VARCHAR
 145 BYTEA
 146 TIMESTAMPTZ
 147 CIDR
 148 );
 149 $dtype = qr{($dtype)};
 150 my %new;
 151 my ($infunction,$inview,$inrule) = (0,0,0);
 152 seek $newfh, 0, 0;
 153 while (<$newfh>) {
 154         next if /^\s*\-\-/ or /^\s*$/;
 155         s/\s*\-\- [\w ']+$//;
 156         next if /^BEGIN;/ or /^SET / or /^COMMIT;/;
 157         next if /^CREATE SEQUENCE/;
 158         next if /^CREATE(?: UNIQUE)? INDEX/;
 159         next if /^CREATE FUNCTION/;
 160         next if /^CREATE TRIGGER/ or /^  FOR EACH ROW/;
 161         next if /^INSERT INTO/ or /^  VALUES \(/;
 162         next if /^ALTER TABLE/;
 163         chomp;
 164
 165         if (/^\$mw\$;?$/) {
 166                 $infunction = $infunction ? 0 : 1;
 167                 next;
 168         }
 169         next if $infunction;
 170
 171         next if /^CREATE VIEW/ and $inview = 1;
 172         if ($inview) {
 173                 /;$/ and $inview = 0;
 174                 next;
 175         }
 176
 177         next if /^CREATE RULE/ and $inrule = 1;
 178         if ($inrule) {
 179                 /;$/ and $inrule = 0;
 180                 next;
 181         }
 182
 183         if (/^CREATE TABLE "?(\w+)"? \($/) {
 184                 $table = $1;
 185                 $new{$table}{name}=$table;
 186         }
 187         elsif (/^\);$/) {
 188         }
 189         elsif (/^  (\w+) +$dtype/) {
 190                 $new{$table}{column}{$1} = $2;
 191         }
 192         else {
 193                 die "Cannot parse line $. of $new:\n$_\n";
 194         }
 195 }
 196
 197 ## Old but not new
 198 for my $t (sort keys %{$old{$oldfile}}) {
 199         if (!exists $new{$t} and !exists $ok{OLD}{$t}) {
 200                 print "Table not in $new: $t\n";
 201                 next;
 202         }
 203         next if exists $ok{OLD}{$t} and !$ok{OLD}{$t};
 204         my $newt = exists $ok{OLD}{$t} ? $ok{OLD}{$t} : $t;
 205         my $oldcol = $old{$oldfile}{$t}{column};
 206         my $newcol = $new{$newt}{column};
 207         for my $c (keys %$oldcol) {
 208                 if (!exists $newcol->{$c}) {
 209                         print "Column $t.$c not in new\n";
 210                         next;
 211                 }
 212         }
 213         for my $c (keys %$newcol) {
 214                 if (!exists $oldcol->{$c}) {
 215                         print "Column $t.$c not in old\n";
 216                         next;
 217                 }
 218         }
 219 }
 220 ## New but not old:
 221 for (sort keys %new) {
 222         if (!exists $old{$oldfile}{$_} and !exists $ok{NEW}{$_}) {
 223                 print "Not in old: $_\n";
 224                 next;
 225         }
 226 }
 227
 228
 229 } ## end each file to be parsed
 230
 231
 232 __DATA__
 233 ## Known exceptions
 234 OLD: searchindex          ## We use tsearch2 directly on the page table instead
 235 OLD: archive              ## This is a view due to the char(14) timestamp hack
 236 RENAME: user mwuser       ## Reserved word causing lots of problems
 237 RENAME: text pagecontent  ## Reserved word
 238 NEW: archive2             ## The real archive table
 239 NEW: mediawiki_version    ## Just us, for now
 240 XFILE: ../archives/patch-profiling.sql