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 use Cwd;
  11
  12 check_valid_sql();
  13
  14 my @old = ('../tables.sql');
  15 my $new = 'tables.sql';
  16 my @xfile;
  17
  18 ## Read in exceptions and other metadata
  19 my %ok;
  20 while (<DATA>) {
  21         next unless /^(\w+)\s*:\s*([^#]+)/;
  22         my ($name,$val) = ($1,$2);
  23         chomp $val;
  24         if ($name eq 'RENAME') {
  25                 die "Invalid rename\n" unless $val =~ /(\w+)\s+(\w+)/;
  26                 $ok{OLD}{$1} = $2;
  27                 $ok{NEW}{$2} = $1;
  28                 next;
  29         }
  30         if ($name eq 'XFILE') {
  31                 push @xfile, $val;
  32                 next;
  33         }
  34         for (split /\s+/ => $val) {
  35                 $ok{$name}{$_} = 0;
  36         }
  37 }
  38
  39 my $datatype = join '|' => qw(
  40 bool
  41 tinyint int bigint real float
  42 tinytext mediumtext text char varchar varbinary binary
  43 timestamp datetime
  44 tinyblob mediumblob blob
  45 );
  46 $datatype .= q{|ENUM\([\"\w, ]+\)};
  47 $datatype = qr{($datatype)};
  48
  49 my $typeval = qr{(\(\d+\))?};
  50
  51 my $typeval2 = qr{ signed| unsigned| binary| NOT NULL| NULL| auto_increment| default ['\-\d\w"]+| REFERENCES .+CASCADE};
  52
  53 my $indextype = join '|' => qw(INDEX KEY FULLTEXT), 'PRIMARY KEY', 'UNIQUE INDEX', 'UNIQUE KEY';
  54 $indextype = qr{$indextype};
  55
  56 my $engine = qr{TYPE|ENGINE};
  57
  58 my $tabletype = qr{InnoDB|MyISAM|HEAP|HEAP MAX_ROWS=\d+|InnoDB MAX_ROWS=\d+ AVG_ROW_LENGTH=\d+};
  59
  60 my $charset = qr{utf8|binary};
  61
  62 open my $newfh, '<', $new or die qq{Could not open $new: $!\n};
  63
  64
  65 my ($table,%old);
  66
  67 ## Read in the xfiles
  68 my %xinfo;
  69 for my $xfile (@xfile) {
  70         print "Loading $xfile\n";
  71         my $info = parse_sql($xfile);
  72         for (keys %$info) {
  73                 $xinfo{$_} = $info->{$_};
  74         }
  75 }
  76
  77 for my $oldfile (@old) {
  78         print "Loading $oldfile\n";
  79         my $info = parse_sql($oldfile);
  80         for (keys %xinfo) {
  81                 $info->{$_} = $xinfo{$_};
  82         }
  83         $old{$oldfile} = $info;
  84 }
  85
  86 sub parse_sql {
  87
  88         my $oldfile = shift;
  89
  90         open my $oldfh, '<', $oldfile or die qq{Could not open $oldfile: $!\n};
  91
  92         my %info;
  93         while (<$oldfh>) {
  94                 next if /^\s*\-\-/ or /^\s+$/;
  95                 s/\s*\-\- [\w ]+$//;
  96                 chomp;
  97
  98                 if (/CREATE\s*TABLE/i) {
  99                         m{^CREATE TABLE /\*\$wgDBprefix\*/(\w+) \($}
 100                                 or die qq{Invalid CREATE TABLE at line $. of $oldfile\n};
 101                         $table = $1;
 102                         $info{$table}{name}=$table;
 103                 }
 104                 elsif (m{^\) /\*\$wgDBTableOptions\*/}) {
 105                         $info{$table}{engine} = 'ENGINE';
 106                         $info{$table}{type} = 'variable';
 107                 }
 108                 elsif (/^\) ($engine)=($tabletype);$/) {
 109                         $info{$table}{engine}=$1;
 110                         $info{$table}{type}=$2;
 111                 }
 112                 elsif (/^\) ($engine)=($tabletype), DEFAULT CHARSET=($charset);$/) {
 113                         $info{$table}{engine}=$1;
 114                         $info{$table}{type}=$2;
 115                         $info{$table}{charset}=$3;
 116                 }
 117                 elsif (/^  (\w+) $datatype$typeval$typeval2{0,3},?$/) {
 118                         $info{$table}{column}{$1} = $2;
 119                         my $extra = $3 || '';
 120                         $info{$table}{columnfull}{$1} = "$2$extra";
 121                 }
 122                 elsif (/^  ($indextype)(?: (\w+))? \(([\w, \(\)]+)\),?$/) {
 123                         $info{$table}{lc $1.'_name'} = $2 ? $2 : '';
 124                         $info{$table}{lc $1.'pk_target'} = $3;
 125                 }
 126                 else {
 127                         die "Cannot parse line $. of $oldfile:\n$_\n";
 128                 }
 129
 130         }
 131         close $oldfh or die qq{Could not close "$oldfile": $!\n};
 132
 133         return \%info;
 134
 135 } ## end of parse_sql
 136
 137 ## Read in the parser test information
 138 my $parsefile = '../parserTests.inc';
 139 open my $pfh, '<', $parsefile or die qq{Could not open "$parsefile": $!\n};
 140 my $stat = 0;
 141 my %ptable;
 142 while (<$pfh>) {
 143         if (!$stat) {
 144                 if (/function listTables/) {
 145                         $stat = 1;
 146                 }
 147                 next;
 148         }
 149         $ptable{$1}=2 while m{'(\w+)'}g;
 150         last if /\);/;
 151 }
 152 close $pfh or die qq{Could not close "$parsefile": $!\n};
 153
 154 my $OK_NOT_IN_PTABLE = '
 155 filearchive
 156 logging
 157 profiling
 158 querycache_info
 159 searchindex
 160 trackbacks
 161 transcache
 162 user_newtalk
 163 updatelog
 164 ';
 165
 166 ## Make sure all tables in main tables.sql are accounted for in the parsertest.
 167 for my $table (sort keys %{$old{'../tables.sql'}}) {
 168         $ptable{$table}++;
 169         next if $ptable{$table} > 2;
 170         next if $OK_NOT_IN_PTABLE =~ /\b$table\b/;
 171         print qq{Table "$table" is in the schema, but not used inside of parserTest.inc\n};
 172 }
 173 ## Any that are used in ptables but no longer exist in the schema?
 174 for my $table (sort grep { $ptable{$_} == 2 } keys %ptable) {
 175         print qq{Table "$table" ($ptable{$table}) used in parserTest.inc, but not found in schema\n};
 176 }
 177
 178 for my $oldfile (@old) {
 179
 180 ## Begin non-standard indent
 181
 182 ## MySQL sanity checks
 183 for my $table (sort keys %{$old{$oldfile}}) {
 184         my $t = $old{$oldfile}{$table};
 185         if ($t->{engine} eq 'TYPE') {
 186                 die "Invalid engine for $oldfile: $t->{engine}\n" unless $t->{name} eq 'profiling';
 187         }
 188         my $charset = $t->{charset} || '';
 189         if ($oldfile !~ /binary/ and $charset eq 'binary') {
 190                 die "Invalid charset for $oldfile: $charset\n";
 191         }
 192 }
 193
 194 my $dtype = join '|' => qw(
 195 SMALLINT INTEGER BIGINT NUMERIC SERIAL
 196 TEXT CHAR VARCHAR
 197 BYTEA
 198 TIMESTAMPTZ
 199 CIDR
 200 );
 201 $dtype = qr{($dtype)};
 202 my %new;
 203 my ($infunction,$inview,$inrule,$lastcomma) = (0,0,0,0);
 204 seek $newfh, 0, 0;
 205 while (<$newfh>) {
 206         next if /^\s*\-\-/ or /^\s*$/;
 207         s/\s*\-\- [\w ']+$//;
 208         next if /^BEGIN;/ or /^SET / or /^COMMIT;/;
 209         next if /^CREATE SEQUENCE/;
 210         next if /^CREATE(?: UNIQUE)? INDEX/;
 211         next if /^CREATE FUNCTION/;
 212         next if /^CREATE TRIGGER/ or /^  FOR EACH ROW/;
 213         next if /^INSERT INTO/ or /^  VALUES \(/;
 214         next if /^ALTER TABLE/;
 215         chomp;
 216
 217         if (/^\$mw\$;?$/) {
 218                 $infunction = $infunction ? 0 : 1;
 219                 next;
 220         }
 221         next if $infunction;
 222
 223         next if /^CREATE VIEW/ and $inview = 1;
 224         if ($inview) {
 225                 /;$/ and $inview = 0;
 226                 next;
 227         }
 228
 229         next if /^CREATE RULE/ and $inrule = 1;
 230         if ($inrule) {
 231                 /;$/ and $inrule = 0;
 232                 next;
 233         }
 234
 235         if (/^CREATE TABLE "?(\w+)"? \($/) {
 236                 $table = $1;
 237                 $new{$table}{name}=$table;
 238                 $lastcomma = 1;
 239         }
 240         elsif (/^\);$/) {
 241                 if ($lastcomma) {
 242                         warn "Stray comma before line $.\n";
 243                 }
 244         }
 245         elsif (/^  (\w+) +$dtype.*?(,?)(?: --.*)?$/) {
 246                 $new{$table}{column}{$1} = $2;
 247                 if (!$lastcomma) {
 248                         print "Missing comma before line $. of $new\n";
 249                 }
 250                 $lastcomma = $3 ? 1 : 0;
 251         }
 252         else {
 253                 die "Cannot parse line $. of $new:\n$_\n";
 254         }
 255 }
 256
 257 ## Which column types are okay to map from mysql to postgres?
 258 my $COLMAP = q{
 259 ## INTS:
 260 tinyint SMALLINT
 261 int INTEGER SERIAL
 262 bigint BIGINT
 263 real NUMERIC
 264 float NUMERIC
 265
 266 ## TEXT:
 267 varchar(15) TEXT
 268 varchar(32) TEXT
 269 varchar(70) TEXT
 270 varchar(255) TEXT
 271 varchar TEXT
 272 text TEXT
 273 tinytext TEXT
 274 ENUM TEXT
 275
 276 ## TIMESTAMPS:
 277 varbinary(14) TIMESTAMPTZ
 278 binary(14) TIMESTAMPTZ
 279 datetime TIMESTAMPTZ
 280 timestamp TIMESTAMPTZ
 281
 282 ## BYTEA:
 283 mediumblob BYTEA
 284
 285 ## OTHER:
 286 bool SMALLINT # Sigh
 287
 288 };
 289 ## Allow specific exceptions to the above
 290 my $COLMAPOK = q{
 291 ## User inputted text strings:
 292 ar_comment      tinyblob       TEXT
 293 fa_description  tinyblob       TEXT
 294 img_description tinyblob       TEXT
 295 ipb_reason      tinyblob       TEXT
 296 log_action      varbinary(10)  TEXT
 297 oi_description  tinyblob       TEXT
 298 rev_comment     tinyblob       TEXT
 299 rc_log_action   varbinary(255) TEXT
 300 rc_log_type     varbinary(255) TEXT
 301
 302 ## Simple text-only strings:
 303 ar_flags          tinyblob       TEXT
 304 fa_minor_mime     varbinary(32)  TEXT
 305 fa_storage_group  varbinary(16)  TEXT # Just 'deleted' for now, should stay plain text
 306 fa_storage_key    varbinary(64)  TEXT # sha1 plus text extension
 307 ipb_address       tinyblob       TEXT # IP address or username
 308 ipb_range_end     tinyblob       TEXT # hexadecimal
 309 ipb_range_start   tinyblob       TEXT # hexadecimal
 310 img_minor_mime    varbinary(32)  TEXT
 311 img_sha1          varbinary(32)  TEXT
 312 job_cmd           varbinary(60)  TEXT # Should we limit to 60 as well?
 313 keyname           varbinary(255) TEXT # No tablename prefix (objectcache)
 314 ll_lang           varbinary(20)  TEXT # Language code
 315 log_params        blob           TEXT # LF separated list of args
 316 log_type          varbinary(10)  TEXT
 317 oi_minor_mime     varbinary(32)  TEXT
 318 oi_sha1           varbinary(32)  TEXT
 319 old_flags         tinyblob       TEXT
 320 old_text          mediumblob     TEXT
 321 pp_propname       varbinary(60)  TEXT
 322 pp_value          blob           TEXT
 323 page_restrictions tinyblob       TEXT # CSV string
 324 pf_server         varchar(30)    TEXT
 325 pr_level          varbinary(60)  TEXT
 326 pr_type           varbinary(60)  TEXT
 327 pt_create_perm    varbinary(60)  TEXT
 328 pt_reason         tinyblob       TEXT
 329 qc_type           varbinary(32)  TEXT
 330 qcc_type          varbinary(32)  TEXT
 331 qci_type          varbinary(32)  TEXT
 332 rc_params         blob           TEXT
 333 rlc_to_blob       blob           TEXT
 334 ug_group          varbinary(16)  TEXT
 335 user_email_token  binary(32)     TEXT
 336 user_ip           varbinary(40)  TEXT
 337 user_newpassword  tinyblob       TEXT
 338 user_options      blob           TEXT
 339 user_password     tinyblob       TEXT
 340 user_token        binary(32)     TEXT
 341
 342 ## Text URLs:
 343 el_index blob           TEXT
 344 el_to    blob           TEXT
 345 iw_url   blob           TEXT
 346 tb_url   blob           TEXT
 347 tc_url   varbinary(255) TEXT
 348
 349 ## Deprecated or not yet used:
 350 ar_text     mediumblob TEXT
 351 job_params  blob       TEXT
 352 log_deleted tinyint    INTEGER # Not used yet, but keep it INTEGER for safety
 353 rc_type     tinyint    CHAR
 354
 355 ## Number tweaking:
 356 fa_bits   int SMALLINT # bits per pixel
 357 fa_height int SMALLINT
 358 fa_width  int SMALLINT # Hope we don't see an image this wide...
 359 hc_id     int BIGINT   # Odd that site_stats is all bigint...
 360 img_bits  int SMALLINT # bits per image should stay sane
 361 oi_bits   int SMALLINT
 362
 363 ## True binary fields, usually due to gzdeflate and/or serialize:
 364 math_inputhash  varbinary(16) BYTEA
 365 math_outputhash varbinary(16) BYTEA
 366
 367 ## Namespaces: not need for such a high range
 368 ar_namespace     int SMALLINT
 369 job_namespace    int SMALLINT
 370 log_namespace    int SMALLINT
 371 page_namespace   int SMALLINT
 372 pl_namespace     int SMALLINT
 373 pt_namespace     int SMALLINT
 374 qc_namespace     int SMALLINT
 375 rc_namespace     int SMALLINT
 376 rd_namespace     int SMALLINT
 377 rlc_to_namespace int SMALLINT
 378 tl_namespace     int SMALLINT
 379 wl_namespace     int SMALLINT
 380
 381 ## Easy enough to change if a wiki ever does grow this big:
 382 ss_active_users  bigint INTEGER
 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_valid_sql {
 463
 464         ## Check for a few common problems in most php files
 465
 466         my $olddir = getcwd();
 467         chdir("../..");
 468         for my $basedir (qw/includes extensions/) {
 469                 scan_dir($basedir);
 470         }
 471         chdir $olddir;
 472
 473         return;
 474
 475 } ## end of check_valid_sql
 476
 477
 478 sub scan_dir {
 479
 480         my $dir = shift;
 481
 482         opendir my $dh, $dir or die qq{Could not opendir $dir: $!\n};
 483         #print "Scanning $dir...\n";
 484         for my $file (grep { -f "$dir/$_" and /\.php$/ } readdir $dh) {
 485                 find_problems("$dir/$file");
 486         }
 487         rewinddir $dh;
 488         for my $subdir (grep { -d "$dir/$_" and ! /\./ } readdir $dh) {
 489                 scan_dir("$dir/$subdir");
 490         }
 491         closedir $dh or die qq{Closedir failed: $!\n};
 492         return;
 493
 494 } ## end of scan_dir
 495
 496 sub find_problems {
 497
 498         my $file = shift;
 499         open my $fh, '<', $file or die qq{Could not open "$file": $!\n};
 500         while (<$fh>) {
 501                 if (/FORCE INDEX/ and $file !~ /Database\w*\.php/) {
 502                         warn "Found FORCE INDEX string at line $. of $file\n";
 503                 }
 504                 if (/REPLACE INTO/ and $file !~ /Database\w*\.php/) {
 505                         warn "Found REPLACE INTO string at line $. of $file\n";
 506                 }
 507                 if (/\bIF\s*\(/ and $file !~ /DatabaseMySQL\.php/) {
 508                         warn "Found IF string at line $. of $file\n";
 509                 }
 510                 if (/\bCONCAT\b/ and $file !~ /Database\w*\.php/) {
 511                         warn "Found CONCAT string at line $. of $file\n";
 512                 }
 513                 if (/\bGROUP\s+BY\s*\d\b/i and $file !~ /Database\w*\.php/) {
 514                         warn "Found GROUP BY # at line $. of $file\n";
 515                 }
 516         }
 517         close $fh or die qq{Could not close "$file": $!\n};
 518         return;
 519
 520 } ## end of find_problems
 521
 522
 523 __DATA__
 524 ## Known exceptions
 525 OLD: searchindex          ## We use tsearch2 directly on the page table instead
 526 RENAME: user mwuser       ## Reserved word causing lots of problems
 527 RENAME: text pagecontent  ## Reserved word
 528 NEW: mediawiki_version    ## Just us, for now
 529 XFILE: ../archives/patch-profiling.sql