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 ## Checks a few other things as well...
   6
   7 use strict;
   8 use warnings;
   9 use Data::Dumper;
  10
  11 check_includes_dir();
  12
  13 my @old = ('../tables.sql');
  14 my $new = 'tables.sql';
  15 my @xfile;
  16
  17 ## Read in exceptions and other metadata
  18 my %ok;
  19 while (<DATA>) {
  20         next unless /^(\w+)\s*:\s*([^#]+)/;
  21         my ($name,$val) = ($1,$2);
  22         chomp $val;
  23         if ($name eq 'RENAME') {
  24                 die "Invalid rename\n" unless $val =~ /(\w+)\s+(\w+)/;
  25                 $ok{OLD}{$1} = $2;
  26                 $ok{NEW}{$2} = $1;
  27                 next;
  28         }
  29         if ($name eq 'XFILE') {
  30                 push @xfile, $val;
  31                 next;
  32         }
  33         for (split /\s+/ => $val) {
  34                 $ok{$name}{$_} = 0;
  35         }
  36 }
  37
  38 my $datatype = join '|' => qw(
  39 bool
  40 tinyint int bigint real float
  41 tinytext mediumtext text char varchar varbinary binary
  42 timestamp datetime
  43 tinyblob mediumblob blob
  44 );
  45 $datatype .= q{|ENUM\([\"\w, ]+\)};
  46 $datatype = qr{($datatype)};
  47
  48 my $typeval = qr{(\(\d+\))?};
  49
  50 my $typeval2 = qr{ signed| unsigned| binary| NOT NULL| NULL| auto_increment| default ['\-\d\w"]+| REFERENCES .+CASCADE};
  51
  52 my $indextype = join '|' => qw(INDEX KEY FULLTEXT), 'PRIMARY KEY', 'UNIQUE INDEX', 'UNIQUE KEY';
  53 $indextype = qr{$indextype};
  54
  55 my $engine = qr{TYPE|ENGINE};
  56
  57 my $tabletype = qr{InnoDB|MyISAM|HEAP|HEAP MAX_ROWS=\d+|InnoDB MAX_ROWS=\d+ AVG_ROW_LENGTH=\d+};
  58
  59 my $charset = qr{utf8|binary};
  60
  61 open my $newfh, '<', $new or die qq{Could not open $new: $!\n};
  62
  63
  64 my ($table,%old);
  65
  66 ## Read in the xfiles
  67 my %xinfo;
  68 for my $xfile (@xfile) {
  69         print "Loading $xfile\n";
  70         my $info = parse_sql($xfile);
  71         for (keys %$info) {
  72                 $xinfo{$_} = $info->{$_};
  73         }
  74 }
  75
  76 for my $oldfile (@old) {
  77         print "Loading $oldfile\n";
  78         my $info = parse_sql($oldfile);
  79         for (keys %xinfo) {
  80                 $info->{$_} = $xinfo{$_};
  81         }
  82         $old{$oldfile} = $info;
  83 }
  84
  85 sub parse_sql {
  86
  87         my $oldfile = shift;
  88
  89         open my $oldfh, '<', $oldfile or die qq{Could not open $oldfile: $!\n};
  90
  91         my %info;
  92         while (<$oldfh>) {
  93                 next if /^\s*\-\-/ or /^\s+$/;
  94                 s/\s*\-\- [\w ]+$//;
  95                 chomp;
  96
  97                 if (/CREATE\s*TABLE/i) {
  98                         m{^CREATE TABLE /\*\$wgDBprefix\*/(\w+) \($}
  99                                 or die qq{Invalid CREATE TABLE at line $. of $oldfile\n};
 100                         $table = $1;
 101                         $info{$table}{name}=$table;
 102                 }
 103                 elsif (m{^\) /\*\$wgDBTableOptions\*/}) {
 104                         $info{$table}{engine} = 'TYPE';
 105                         $info{$table}{type} = 'variable';
 106                 }
 107                 elsif (/^\) ($engine)=($tabletype);$/) {
 108                         $info{$table}{engine}=$1;
 109                         $info{$table}{type}=$2;
 110                 }
 111                 elsif (/^\) ($engine)=($tabletype), DEFAULT CHARSET=($charset);$/) {
 112                         $info{$table}{engine}=$1;
 113                         $info{$table}{type}=$2;
 114                         $info{$table}{charset}=$3;
 115                 }
 116                 elsif (/^  (\w+) $datatype$typeval$typeval2{0,3},?$/) {
 117                         $info{$table}{column}{$1} = $2;
 118                         my $extra = $3 || '';
 119                         $info{$table}{columnfull}{$1} = "$2$extra";
 120                 }
 121                 elsif (/^  ($indextype)(?: (\w+))? \(([\w, \(\)]+)\),?$/) {
 122                         $info{$table}{lc $1.'_name'} = $2 ? $2 : '';
 123                         $info{$table}{lc $1.'pk_target'} = $3;
 124                 }
 125                 else {
 126                         die "Cannot parse line $. of $oldfile:\n$_\n";
 127                 }
 128
 129         }
 130         close $oldfh or die qq{Could not close "$oldfile": $!\n};
 131
 132         return \%info;
 133
 134 } ## end of parse_sql
 135
 136 ## Read in the parser test information
 137 my $parsefile = '../parserTests.inc';
 138 open my $pfh, '<', $parsefile or die qq{Could not open "$parsefile": $!\n};
 139 my $stat = 0;
 140 my %ptable;
 141 while (<$pfh>) {
 142         if (!$stat) {
 143                 if (/function listTables/) {
 144                         $stat = 1;
 145                 }
 146                 next;
 147         }
 148         $ptable{$1}=2 while m{'(\w+)'}g;
 149         last if /\);/;
 150 }
 151 close $pfh or die qq{Could not close "$parsefile": $!\n};
 152
 153 my $OK_NOT_IN_PTABLE = '
 154 filearchive
 155 logging
 156 profiling
 157 querycache_info
 158 searchindex
 159 trackbacks
 160 transcache
 161 user_newtalk
 162 updatelog
 163 ';
 164
 165 ## Make sure all tables in main tables.sql are accounted for int the parsertest.
 166 for my $table (sort keys %{$old{'../tables.sql'}}) {
 167         $ptable{$table}++;
 168         next if $ptable{$table} > 2;
 169         next if $OK_NOT_IN_PTABLE =~ /\b$table\b/;
 170         print qq{Table "$table" is in the schema, but not used inside of parserTest.inc\n};
 171 }
 172 ## Any that are used in ptables but no longer exist in the schema?
 173 for my $table (sort grep { $ptable{$_} == 2 } keys %ptable) {
 174         print qq{Table "$table" ($ptable{$table}) used in parserTest.inc, but not found in schema\n};
 175 }
 176
 177 for my $oldfile (@old) {
 178
 179 ## Begin non-standard indent
 180
 181 ## MySQL sanity checks
 182 for my $table (sort keys %{$old{$oldfile}}) {
 183         my $t = $old{$oldfile}{$table};
 184         if (($oldfile =~ /5/ and $t->{engine} ne 'ENGINE')
 185                 or
 186                 ($oldfile !~ /5/ and $t->{engine} ne 'TYPE')) {
 187                 die "Invalid engine for $oldfile: $t->{engine}\n" unless $t->{name} eq 'profiling';
 188         }
 189         my $charset = $t->{charset} || '';
 190         if ($oldfile !~ /binary/ and $charset eq 'binary') {
 191                 die "Invalid charset for $oldfile: $charset\n";
 192         }
 193 }
 194
 195 my $dtype = join '|' => qw(
 196 SMALLINT INTEGER BIGINT NUMERIC SERIAL
 197 TEXT CHAR VARCHAR
 198 BYTEA
 199 TIMESTAMPTZ
 200 CIDR
 201 );
 202 $dtype = qr{($dtype)};
 203 my %new;
 204 my ($infunction,$inview,$inrule,$lastcomma) = (0,0,0,0);
 205 seek $newfh, 0, 0;
 206 while (<$newfh>) {
 207         next if /^\s*\-\-/ or /^\s*$/;
 208         s/\s*\-\- [\w ']+$//;
 209         next if /^BEGIN;/ or /^SET / or /^COMMIT;/;
 210         next if /^CREATE SEQUENCE/;
 211         next if /^CREATE(?: UNIQUE)? INDEX/;
 212         next if /^CREATE FUNCTION/;
 213         next if /^CREATE TRIGGER/ or /^  FOR EACH ROW/;
 214         next if /^INSERT INTO/ or /^  VALUES \(/;
 215         next if /^ALTER TABLE/;
 216         chomp;
 217
 218         if (/^\$mw\$;?$/) {
 219                 $infunction = $infunction ? 0 : 1;
 220                 next;
 221         }
 222         next if $infunction;
 223
 224         next if /^CREATE VIEW/ and $inview = 1;
 225         if ($inview) {
 226                 /;$/ and $inview = 0;
 227                 next;
 228         }
 229
 230         next if /^CREATE RULE/ and $inrule = 1;
 231         if ($inrule) {
 232                 /;$/ and $inrule = 0;
 233                 next;
 234         }
 235
 236         if (/^CREATE TABLE "?(\w+)"? \($/) {
 237                 $table = $1;
 238                 $new{$table}{name}=$table;
 239                 $lastcomma = 1;
 240         }
 241         elsif (/^\);$/) {
 242                 if ($lastcomma) {
 243                         warn "Stray comma before line $.\n";
 244                 }
 245         }
 246         elsif (/^  (\w+) +$dtype.*?(,?)(?: --.*)?$/) {
 247                 $new{$table}{column}{$1} = $2;
 248                 if (!$lastcomma) {
 249                         print "Missing comma before line $. of $new\n";
 250                 }
 251                 $lastcomma = $3 ? 1 : 0;
 252         }
 253         else {
 254                 die "Cannot parse line $. of $new:\n$_\n";
 255         }
 256 }
 257
 258 ## Which column types are okay to map from mysql to postgres?
 259 my $COLMAP = q{
 260 ## INTS:
 261 tinyint SMALLINT
 262 int INTEGER SERIAL
 263 bigint BIGINT
 264 real NUMERIC
 265 float NUMERIC
 266
 267 ## TEXT:
 268 varchar(15) TEXT
 269 varchar(32) TEXT
 270 varchar(70) TEXT
 271 varchar(255) TEXT
 272 varchar TEXT
 273 text TEXT
 274 tinytext TEXT
 275 ENUM TEXT
 276
 277 ## TIMESTAMPS:
 278 varbinary(14) TIMESTAMPTZ
 279 binary(14) TIMESTAMPTZ
 280 datetime TIMESTAMPTZ
 281 timestamp TIMESTAMPTZ
 282
 283 ## BYTEA:
 284 mediumblob BYTEA
 285
 286 ## OTHER:
 287 bool SMALLINT # Sigh
 288
 289 };
 290 ## Allow specific exceptions to the above
 291 my $COLMAPOK = q{
 292 ## User inputted text strings:
 293 ar_comment      tinyblob       TEXT
 294 fa_description  tinyblob       TEXT
 295 img_description tinyblob       TEXT
 296 ipb_reason      tinyblob       TEXT
 297 log_action      varbinary(10)  TEXT
 298 oi_description  tinyblob       TEXT
 299 rev_comment     tinyblob       TEXT
 300 rc_log_action   varbinary(255) TEXT
 301 rc_log_type     varbinary(255) TEXT
 302
 303 ## Simple text-only strings:
 304 ar_flags          tinyblob       TEXT
 305 fa_minor_mime     varbinary(32)  TEXT
 306 fa_storage_group  varbinary(16)  TEXT # Just 'deleted' for now, should stay plain text
 307 fa_storage_key    varbinary(64)  TEXT # sha1 plus text extension
 308 ipb_address       tinyblob       TEXT # IP address or username
 309 ipb_range_end     tinyblob       TEXT # hexadecimal
 310 ipb_range_start   tinyblob       TEXT # hexadecimal
 311 img_minor_mime    varbinary(32)  TEXT
 312 img_sha1          varbinary(32)  TEXT
 313 job_cmd           varbinary(60)  TEXT # Should we limit to 60 as well?
 314 keyname           varbinary(255) TEXT # No tablename prefix (objectcache)
 315 ll_lang           varbinary(20)  TEXT # Language code
 316 log_params        blob           TEXT # LF separated list of args
 317 log_type          varbinary(10)  TEXT
 318 oi_minor_mime     varbinary(32)  TEXT
 319 oi_sha1           varbinary(32)  TEXT
 320 old_flags         tinyblob       TEXT
 321 old_text          mediumblob     TEXT
 322 pp_propname       varbinary(60)  TEXT
 323 pp_value          blob           TEXT
 324 page_restrictions tinyblob       TEXT # CSV string
 325 pf_server         varchar(30)    TEXT
 326 pr_level          varbinary(60)  TEXT
 327 pr_type           varbinary(60)  TEXT
 328 pt_create_perm    varbinary(60)  TEXT
 329 pt_reason         tinyblob       TEXT
 330 qc_type           varbinary(32)  TEXT
 331 qcc_type          varbinary(32)  TEXT
 332 qci_type          varbinary(32)  TEXT
 333 rc_params         blob           TEXT
 334 rlc_to_blob       blob           TEXT
 335 ug_group          varbinary(16)  TEXT
 336 user_email_token  binary(32)     TEXT
 337 user_ip           varbinary(40)  TEXT
 338 user_newpassword  tinyblob       TEXT
 339 user_options      blob           TEXT
 340 user_password     tinyblob       TEXT
 341 user_token        binary(32)     TEXT
 342
 343 ## Text URLs:
 344 el_index blob           TEXT
 345 el_to    blob           TEXT
 346 iw_url   blob           TEXT
 347 tb_url   blob           TEXT
 348 tc_url   varbinary(255) TEXT
 349
 350 ## Deprecated or not yet used:
 351 ar_text     mediumblob TEXT
 352 job_params  blob       TEXT
 353 log_deleted tinyint    INTEGER # Not used yet, but keep it INTEGER for safety
 354 rc_type     tinyint    CHAR
 355
 356 ## Number tweaking:
 357 fa_bits   int SMALLINT # bits per pixel
 358 fa_height int SMALLINT
 359 fa_width  int SMALLINT # Hope we don't see an image this wide...
 360 hc_id     int BIGINT   # Odd that site_stats is all bigint...
 361 img_bits  int SMALLINT # bits per image should stay sane
 362 oi_bits   int SMALLINT
 363
 364 ## True binary fields, usually due to gzdeflate and/or serialize:
 365 math_inputhash  varbinary(16) BYTEA
 366 math_outputhash varbinary(16) BYTEA
 367
 368 ## Namespaces: not need for such a high range
 369 ar_namespace     int SMALLINT
 370 job_namespace    int SMALLINT
 371 log_namespace    int SMALLINT
 372 page_namespace   int SMALLINT
 373 pl_namespace     int SMALLINT
 374 pt_namespace     int SMALLINT
 375 qc_namespace     int SMALLINT
 376 rc_namespace     int SMALLINT
 377 rd_namespace     int SMALLINT
 378 rlc_to_namespace int SMALLINT
 379 tl_namespace     int SMALLINT
 380 wl_namespace     int SMALLINT
 381
 382 ## Easy enough to change if a wiki ever does grow this big:
 383 ss_good_articles bigint INTEGER
 384 ss_total_edits   bigint INTEGER
 385 ss_total_pages   bigint INTEGER
 386 ss_total_views   bigint INTEGER
 387 ss_users         bigint INTEGER
 388
 389 ## True IP - keep an eye on these, coders tend to make textual assumptions
 390 rc_ip varbinary(40) CIDR # Want to keep an eye on this
 391
 392 ## Others:
 393 tc_time int TIMESTAMPTZ
 394
 395
 396 };
 397
 398 my %colmap;
 399 for (split /\n/ => $COLMAP) {
 400         next unless /^\w/;
 401         s/(.*?)#.*/$1/;
 402         my ($col,@maps) = split / +/, $_;
 403         for (@maps) {
 404                 $colmap{$col}{$_} = 1;
 405         }
 406 }
 407
 408 my %colmapok;
 409 for (split /\n/ => $COLMAPOK) {
 410         next unless /^\w/;
 411         my ($col,$old,$new) = split / +/, $_;
 412         $colmapok{$col}{$old}{$new} = 1;
 413 }
 414
 415 ## Old but not new
 416 for my $t (sort keys %{$old{$oldfile}}) {
 417         if (!exists $new{$t} and !exists $ok{OLD}{$t}) {
 418                 print "Table not in $new: $t\n";
 419                 next;
 420         }
 421         next if exists $ok{OLD}{$t} and !$ok{OLD}{$t};
 422         my $newt = exists $ok{OLD}{$t} ? $ok{OLD}{$t} : $t;
 423         my $oldcol = $old{$oldfile}{$t}{column};
 424         my $oldcolfull = $old{$oldfile}{$t}{columnfull};
 425         my $newcol = $new{$newt}{column};
 426         for my $c (keys %$oldcol) {
 427                 if (!exists $newcol->{$c}) {
 428                         print "Column $t.$c not in $new\n";
 429                         next;
 430                 }
 431         }
 432         for my $c (sort keys %$newcol) {
 433                 if (!exists $oldcol->{$c}) {
 434                         print "Column $t.$c not in $oldfile\n";
 435                         next;
 436                 }
 437                 ## Column types (roughly) match up?
 438                 my $new = $newcol->{$c};
 439                 my $old = $oldcolfull->{$c};
 440
 441                 ## Known exceptions:
 442                 next if exists $colmapok{$c}{$old}{$new};
 443
 444                 $old =~ s/ENUM.*/ENUM/;
 445                 if (! exists $colmap{$old}{$new}) {
 446                         print "Column types for $t.$c do not match: $old does not map to $new\n";
 447                 }
 448         }
 449 }
 450 ## New but not old:
 451 for (sort keys %new) {
 452         if (!exists $old{$oldfile}{$_} and !exists $ok{NEW}{$_}) {
 453                 print "Not in $oldfile: $_\n";
 454                 next;
 455         }
 456 }
 457
 458
 459 } ## end each file to be parsed
 460
 461
 462 sub check_includes_dir {
 463
 464         ## Check for some common errors in the files in the includes directory
 465
 466         print "Checking files in includes directory...\n";
 467         my $dir = '../../includes';
 468         opendir my $dh, $dir or die qq{Could not opendir $dir: $!\n};
 469         for my $file (grep { -f "$dir/$_" and /\.php$/ } readdir $dh) {
 470                 $file = "$dir/$file";
 471                 open my $fh, '<', $file or die qq{Could not open "$file": $!\n};
 472                 while (<$fh>) {
 473                         if (/FORCE INDEX/ and $file !~ /Database.php/) {
 474                                 warn "Found FORCE INDEX string at line $. of $file\n";
 475                         }
 476                         if (/REPLACE INTO/ and $file !~ /Database/) {
 477                                 warn "Found REPLACE INTO string at line $. of $file\n";
 478                         }
 479                         if (/\bIF\s*\(/ and $file !~ /Database.php/) {
 480                                 warn "Found IF string at line $. of $file\n";
 481                         }
 482                         if (/\bCONCAT\b/ and $file !~ /Database.php/) {
 483                                 warn "Found CONCAT string at line $. of $file\n";
 484                         }
 485                 }
 486                 close $fh or die qq{Could not close "$file": $!\n};
 487         }
 488         closedir $dh or die qq{Closedir failed?!\n};
 489
 490         return;
 491
 492 } ## end of check_includes_dir
 493
 494 __DATA__
 495 ## Known exceptions
 496 OLD: searchindex          ## We use tsearch2 directly on the page table instead
 497 RENAME: user mwuser       ## Reserved word causing lots of problems
 498 RENAME: text pagecontent  ## Reserved word
 499 NEW: mediawiki_version    ## Just us, for now
 500 XFILE: ../archives/patch-profiling.sql