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| PRIMARY KEY| 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                         if (m{^CREATE TABLE /\*_\*/(\w+) \($}) {
 100                                 $table = $1;
 101                         }
 102                         elsif (m{^CREATE TABLE /\*\$wgDBprefix\*/(\w+) \($}) {
 103                                 $table = $1;
 104                         }
 105                         else {
 106                                 die qq{Invalid CREATE TABLE at line $. of $oldfile\n};
 107                         }
 108                         $info{$table}{name}=$table;
 109                 }
 110                 elsif (m{^\) /\*\$wgDBTableOptions\*/}) {
 111                         $info{$table}{engine} = 'ENGINE';
 112                         $info{$table}{type} = 'variable';
 113                 }
 114                 elsif (/^\) ($engine)=($tabletype);$/) {
 115                         $info{$table}{engine}=$1;
 116                         $info{$table}{type}=$2;
 117                 }
 118                 elsif (/^\) ($engine)=($tabletype), DEFAULT CHARSET=($charset);$/) {
 119                         $info{$table}{engine}=$1;
 120                         $info{$table}{type}=$2;
 121                         $info{$table}{charset}=$3;
 122                 }
 123                 elsif (/^  (\w+) $datatype$typeval$typeval2{0,4},?$/) {
 124                         $info{$table}{column}{$1} = $2;
 125                         my $extra = $3 || '';
 126                         $info{$table}{columnfull}{$1} = "$2$extra";
 127                 }
 128                 elsif (m{^  UNIQUE KEY (\w+) \((.+?)\)}) {
 129                 }
 130                 elsif (m{^CREATE (?:UNIQUE )?(?:FULLTEXT )?INDEX /\*i\*/(\w+) ON /\*_\*/(\w+) \((.+?)\);}) {
 131                 }
 132                 elsif (m{^\s*PRIMARY KEY \([\w,]+\)}) {
 133                 }
 134                 else {
 135                         die "Cannot parse line $. of $oldfile:\n$_\n";
 136                 }
 137
 138         }
 139         close $oldfh or die qq{Could not close "$oldfile": $!\n};
 140
 141         return \%info;
 142
 143 } ## end of parse_sql
 144
 145 for my $oldfile (@old) {
 146
 147 ## Begin non-standard indent
 148
 149 ## MySQL sanity checks
 150 for my $table (sort keys %{$old{$oldfile}}) {
 151         my $t = $old{$oldfile}{$table};
 152         if ($t->{engine} eq 'TYPE') {
 153                 die "Invalid engine for $oldfile: $t->{engine}\n" unless $t->{name} eq 'profiling';
 154         }
 155         my $charset = $t->{charset} || '';
 156         if ($oldfile !~ /binary/ and $charset eq 'binary') {
 157                 die "Invalid charset for $oldfile: $charset\n";
 158         }
 159 }
 160
 161 my $dtype = join '|' => qw(
 162 SMALLINT INTEGER BIGINT NUMERIC SERIAL
 163 TEXT CHAR VARCHAR
 164 BYTEA
 165 TIMESTAMPTZ
 166 CIDR
 167 );
 168 $dtype = qr{($dtype)};
 169 my %new;
 170 my ($infunction,$inview,$inrule,$lastcomma) = (0,0,0,0);
 171 seek $newfh, 0, 0;
 172 while (<$newfh>) {
 173         next if /^\s*\-\-/ or /^\s*$/;
 174         s/\s*\-\- [\w ']+$//;
 175         next if /^BEGIN;/ or /^SET / or /^COMMIT;/;
 176         next if /^CREATE SEQUENCE/;
 177         next if /^CREATE(?: UNIQUE)? INDEX/;
 178         next if /^CREATE FUNCTION/;
 179         next if /^CREATE TRIGGER/ or /^  FOR EACH ROW/;
 180         next if /^INSERT INTO/ or /^  VALUES \(/;
 181         next if /^ALTER TABLE/;
 182         next if /^DROP SEQUENCE/;
 183         next if /^DROP FUNCTION/;
 184
 185
 186         chomp;
 187
 188         if (/^\$mw\$;?$/) {
 189                 $infunction = $infunction ? 0 : 1;
 190                 next;
 191         }
 192         next if $infunction;
 193
 194         next if /^CREATE VIEW/ and $inview = 1;
 195         if ($inview) {
 196                 /;$/ and $inview = 0;
 197                 next;
 198         }
 199
 200         next if /^CREATE RULE/ and $inrule = 1;
 201         if ($inrule) {
 202                 /;$/ and $inrule = 0;
 203                 next;
 204         }
 205
 206         if (/^CREATE TABLE "?(\w+)"? \($/) {
 207                 $table = $1;
 208                 $new{$table}{name}=$table;
 209                 $lastcomma = 1;
 210         }
 211         elsif (/^\);$/) {
 212                 if ($lastcomma) {
 213                         warn "Stray comma before line $.\n";
 214                 }
 215         }
 216         elsif (/^  (\w+) +$dtype.*?(,?)(?: --.*)?$/) {
 217                 $new{$table}{column}{$1} = $2;
 218                 if (!$lastcomma) {
 219                         print "Missing comma before line $. of $new\n";
 220                 }
 221                 $lastcomma = $3 ? 1 : 0;
 222         }
 223         elsif (m{^\s*PRIMARY KEY \([\w,]+\)}) {
 224                 $lastcomma = 0;
 225         }
 226         else {
 227                 die "Cannot parse line $. of $new:\n$_\n";
 228         }
 229 }
 230
 231 ## Which column types are okay to map from mysql to postgres?
 232 my $COLMAP = q{
 233 ## INTS:
 234 tinyint SMALLINT
 235 int INTEGER SERIAL
 236 bigint BIGINT
 237 real NUMERIC
 238 float NUMERIC
 239
 240 ## TEXT:
 241 varchar(15) TEXT
 242 varchar(32) TEXT
 243 varchar(70) TEXT
 244 varchar(255) TEXT
 245 varchar TEXT
 246 text TEXT
 247 tinytext TEXT
 248 ENUM TEXT
 249
 250 ## TIMESTAMPS:
 251 varbinary(14) TIMESTAMPTZ
 252 binary(14) TIMESTAMPTZ
 253 datetime TIMESTAMPTZ
 254 timestamp TIMESTAMPTZ
 255
 256 ## BYTEA:
 257 mediumblob BYTEA
 258
 259 ## OTHER:
 260 bool SMALLINT # Sigh
 261
 262 };
 263 ## Allow specific exceptions to the above
 264 my $COLMAPOK = q{
 265 ## User inputted text strings:
 266 ar_comment      tinyblob       TEXT
 267 fa_description  tinyblob       TEXT
 268 img_description tinyblob       TEXT
 269 ipb_reason      tinyblob       TEXT
 270 log_action      varbinary(32)  TEXT
 271 log_type        varbinary(32)  TEXT
 272 oi_description  tinyblob       TEXT
 273 rev_comment     tinyblob       TEXT
 274 rc_log_action   varbinary(255) TEXT
 275 rc_log_type     varbinary(255) TEXT
 276
 277 ## Simple text-only strings:
 278 ar_flags          tinyblob       TEXT
 279 cl_collation      varbinary(32)  TEXT
 280 cl_sortkey        varbinary(230) TEXT
 281 ct_params         blob           TEXT
 282 fa_minor_mime     varbinary(100) TEXT
 283 fa_storage_group  varbinary(16)  TEXT # Just 'deleted' for now, should stay plain text
 284 fa_storage_key    varbinary(64)  TEXT # sha1 plus text extension
 285 ipb_address       tinyblob       TEXT # IP address or username
 286 ipb_range_end     tinyblob       TEXT # hexadecimal
 287 ipb_range_start   tinyblob       TEXT # hexadecimal
 288 img_minor_mime    varbinary(100) TEXT
 289 lc_lang           varbinary(32)  TEXT
 290 lc_value          varbinary(32)  TEXT
 291 img_sha1          varbinary(32)  TEXT
 292 iw_wikiid         varchar(64)    TEXT
 293 job_cmd           varbinary(60)  TEXT # Should we limit to 60 as well?
 294 keyname           varbinary(255) TEXT # No tablename prefix (objectcache)
 295 ll_lang           varbinary(20)  TEXT # Language code
 296 lc_value          mediumblob     TEXT
 297 log_params        blob           TEXT # LF separated list of args
 298 log_type          varbinary(10)  TEXT
 299 ls_field          varbinary(32)  TEXT
 300 md_deps           mediumblob     TEXT # JSON
 301 mr_blob           mediumblob     TEXT # JSON
 302 mr_lang           varbinary(32)  TEXT
 303 oi_minor_mime     varbinary(100) TEXT
 304 oi_sha1           varbinary(32)  TEXT
 305 old_flags         tinyblob       TEXT
 306 old_text          mediumblob     TEXT
 307 pp_propname       varbinary(60)  TEXT
 308 pp_value          blob           TEXT
 309 page_restrictions tinyblob       TEXT # CSV string
 310 pf_server         varchar(30)    TEXT
 311 pr_level          varbinary(60)  TEXT
 312 pr_type           varbinary(60)  TEXT
 313 pt_create_perm    varbinary(60)  TEXT
 314 pt_reason         tinyblob       TEXT
 315 qc_type           varbinary(32)  TEXT
 316 qcc_type          varbinary(32)  TEXT
 317 qci_type          varbinary(32)  TEXT
 318 rc_params         blob           TEXT
 319 rlc_to_blob       blob           TEXT
 320 ts_tags           blob           TEXT
 321 ug_group          varbinary(16)  TEXT
 322 ul_value          blob           TEXT
 323 up_property       varbinary(32)  TEXT
 324 up_value          blob           TEXT
 325 user_email_token  binary(32)     TEXT
 326 user_ip           varbinary(40)  TEXT
 327 user_newpassword  tinyblob       TEXT
 328 user_options      blob           TEXT
 329 user_password     tinyblob       TEXT
 330 user_token        binary(32)     TEXT
 331 iwl_prefix      varbinary(20)  TEXT
 332
 333 ## Text URLs:
 334 el_index blob           TEXT
 335 el_to    blob           TEXT
 336 iw_api   blob           TEXT
 337 iw_url   blob           TEXT
 338 tb_url   blob           TEXT
 339 tc_url   varbinary(255) TEXT
 340
 341 ## Deprecated or not yet used:
 342 ar_text     mediumblob TEXT
 343 job_params  blob       TEXT
 344 log_deleted tinyint    INTEGER # Not used yet, but keep it INTEGER for safety
 345 rc_type     tinyint    CHAR
 346
 347 ## Number tweaking:
 348 fa_bits   int SMALLINT # bits per pixel
 349 fa_height int SMALLINT
 350 fa_width  int SMALLINT # Hope we don't see an image this wide...
 351 hc_id     int BIGINT   # Odd that site_stats is all bigint...
 352 img_bits  int SMALLINT # bits per image should stay sane
 353 oi_bits   int SMALLINT
 354
 355 ## True binary fields, usually due to gzdeflate and/or serialize:
 356 math_inputhash  varbinary(16) BYTEA
 357 math_outputhash varbinary(16) BYTEA
 358
 359 ## Namespaces: not need for such a high range
 360 ar_namespace     int SMALLINT
 361 job_namespace    int SMALLINT
 362 log_namespace    int SMALLINT
 363 page_namespace   int SMALLINT
 364 pl_namespace     int SMALLINT
 365 pt_namespace     int SMALLINT
 366 qc_namespace     int SMALLINT
 367 rc_namespace     int SMALLINT
 368 rd_namespace     int SMALLINT
 369 rlc_to_namespace int SMALLINT
 370 tl_namespace     int SMALLINT
 371 wl_namespace     int SMALLINT
 372
 373 ## Easy enough to change if a wiki ever does grow this big:
 374 ss_active_users  bigint INTEGER
 375 ss_good_articles bigint INTEGER
 376 ss_total_edits   bigint INTEGER
 377 ss_total_pages   bigint INTEGER
 378 ss_total_views   bigint INTEGER
 379 ss_users         bigint INTEGER
 380
 381 ## True IP - keep an eye on these, coders tend to make textual assumptions
 382 rc_ip varbinary(40) CIDR # Want to keep an eye on this
 383
 384 ## Others:
 385 tc_time int TIMESTAMPTZ
 386
 387
 388 };
 389
 390 my %colmap;
 391 for (split /\n/ => $COLMAP) {
 392         next unless /^\w/;
 393         s/(.*?)#.*/$1/;
 394         my ($col,@maps) = split / +/, $_;
 395         for (@maps) {
 396                 $colmap{$col}{$_} = 1;
 397         }
 398 }
 399
 400 my %colmapok;
 401 for (split /\n/ => $COLMAPOK) {
 402         next unless /^\w/;
 403         my ($col,$old,$new) = split / +/, $_;
 404         $colmapok{$col}{$old}{$new} = 1;
 405 }
 406
 407 ## Old but not new
 408 for my $t (sort keys %{$old{$oldfile}}) {
 409         if (!exists $new{$t} and !exists $ok{OLD}{$t}) {
 410                 print "Table not in $new: $t\n";
 411                 next;
 412         }
 413         next if exists $ok{OLD}{$t} and !$ok{OLD}{$t};
 414         my $newt = exists $ok{OLD}{$t} ? $ok{OLD}{$t} : $t;
 415         my $oldcol = $old{$oldfile}{$t}{column};
 416         my $oldcolfull = $old{$oldfile}{$t}{columnfull};
 417         my $newcol = $new{$newt}{column};
 418         for my $c (keys %$oldcol) {
 419                 if (!exists $newcol->{$c}) {
 420                         print "Column $t.$c not in $new\n";
 421                         next;
 422                 }
 423         }
 424         for my $c (sort keys %$newcol) {
 425                 if (!exists $oldcol->{$c}) {
 426                         print "Column $t.$c not in $oldfile\n";
 427                         next;
 428                 }
 429                 ## Column types (roughly) match up?
 430                 my $new = $newcol->{$c};
 431                 my $old = $oldcolfull->{$c};
 432
 433                 ## Known exceptions:
 434                 next if exists $colmapok{$c}{$old}{$new};
 435
 436                 $old =~ s/ENUM.*/ENUM/;
 437                 if (! exists $colmap{$old}{$new}) {
 438                         print "Column types for $t.$c do not match: $old does not map to $new\n";
 439                 }
 440         }
 441 }
 442 ## New but not old:
 443 for (sort keys %new) {
 444         if (!exists $old{$oldfile}{$_} and !exists $ok{NEW}{$_}) {
 445                 print "Not in $oldfile: $_\n";
 446                 next;
 447         }
 448 }
 449
 450
 451 } ## end each file to be parsed
 452
 453
 454 sub check_valid_sql {
 455
 456         ## Check for a few common problems in most php files
 457
 458         my $olddir = getcwd();
 459         chdir("../..");
 460         for my $basedir (qw/includes extensions/) {
 461                 scan_dir($basedir);
 462         }
 463         chdir $olddir;
 464
 465         return;
 466
 467 } ## end of check_valid_sql
 468
 469
 470 sub scan_dir {
 471
 472         my $dir = shift;
 473
 474         opendir my $dh, $dir or die qq{Could not opendir $dir: $!\n};
 475         #print "Scanning $dir...\n";
 476         for my $file (grep { -f "$dir/$_" and /\.php$/ } readdir $dh) {
 477                 find_problems("$dir/$file");
 478         }
 479         rewinddir $dh;
 480         for my $subdir (grep { -d "$dir/$_" and ! /\./ } readdir $dh) {
 481                 scan_dir("$dir/$subdir");
 482         }
 483         closedir $dh or die qq{Closedir failed: $!\n};
 484         return;
 485
 486 } ## end of scan_dir
 487
 488 sub find_problems {
 489
 490         my $file = shift;
 491         open my $fh, '<', $file or die qq{Could not open "$file": $!\n};
 492         my $lastline = '';
 493         my $inarray = 0;
 494         while (<$fh>) {
 495                 if (/FORCE INDEX/ and $file !~ /Database\w*\.php/) {
 496                         warn "Found FORCE INDEX string at line $. of $file\n";
 497                 }
 498                 if (/REPLACE INTO/ and $file !~ /Database\w*\.php/) {
 499                         warn "Found REPLACE INTO string at line $. of $file\n";
 500                 }
 501                 if (/\bIF\s*\(/ and $file !~ /DatabaseMySQL\.php/) {
 502                         warn "Found IF string at line $. of $file\n";
 503                 }
 504                 if (/\bCONCAT\b/ and $file !~ /Database\w*\.php/) {
 505                         warn "Found CONCAT string at line $. of $file\n";
 506                 }
 507                 if (/\bGROUP\s+BY\s*\d\b/i and $file !~ /Database\w*\.php/) {
 508                         warn "Found GROUP BY # at line $. of $file\n";
 509                 }
 510                 if (/wfGetDB\s*\(\s+\)/io) {
 511                         warn "wfGETDB is missing parameters at line $. of $file\n";
 512                 }
 513                 if (/=\s*array\s*\(\s*$/) {
 514                         $inarray = 1;
 515                         next;
 516                 }
 517                 if ($inarray) {
 518                         if (/\s*\);\s*$/) {
 519                                 $inarray = 0;
 520                                 next;
 521                         }
 522                         next if ! /\w/ or /array\(\s*$/ or /^\s*#/ or m{^\s*//};
 523                         if (! /,/) {
 524                                 my $nextline = <$fh>;
 525                                 last if ! defined $nextline;
 526                                 if ($nextline =~ /^\s*\)[;,]/) {
 527                                         $inarray = 0;
 528                                         next;
 529                                 }
 530                                 #warn "Array is missing a comma? Line $. of $file\n";
 531                         }
 532                 }
 533         }
 534         close $fh or die qq{Could not close "$file": $!\n};
 535         return;
 536
 537 } ## end of find_problems
 538
 539
 540 __DATA__
 541 ## Known exceptions
 542 OLD: searchindex          ## We use tsearch2 directly on the page table instead
 543 RENAME: user mwuser       ## Reserved word causing lots of problems
 544 RENAME: text pagecontent  ## Reserved word
 545 XFILE: ../archives/patch-profiling.sql