Add in binary type

[lhc/web/wiklou.git] / maintenance / postgres / compare_schemas.pl

diff --git a/maintenance/postgres/compare_schemas.pl b/maintenance/postgres/compare_schemas.pl
index f8305a0..297e3af 100644 (file)
--- a/maintenance/postgres/compare_schemas.pl
+++ b/maintenance/postgres/compare_schemas.pl
@@ -7,8 +7,8 @@ use strict;
 use warnings;
 use Data::Dumper;
 
-my @old = ("../tables.sql", "../mysql5/tables.sql");
-my $new = "tables.sql";
+my @old = ('../tables.sql');
+my $new = 'tables.sql';
 my @xfile;
 
 ## Read in exceptions and other metadata
@@ -27,7 +27,7 @@ while (<DATA>) {
                push @xfile, $val;
                next;
        }
-       for (split(/\s+/ => $val)) {
+       for (split /\s+/ => $val) {
                $ok{$name}{$_} = 0;
        }
 }
@@ -35,7 +35,7 @@ while (<DATA>) {
 my $datatype = join '|' => qw(
 bool
 tinyint int bigint real float
-tinytext mediumtext text char varchar varbinary
+tinytext mediumtext text char varchar varbinary binary
 timestamp datetime
 tinyblob mediumblob blob
 );
@@ -46,17 +46,16 @@ my $typeval = qr{(\(\d+\))?};
 
 my $typeval2 = qr{ unsigned| binary| NOT NULL| NULL| auto_increment| default ['\-\d\w"]+| REFERENCES .+CASCADE};
 
-my $indextype = join '|' => qw(INDEX KEY FULLTEXT), "PRIMARY KEY", "UNIQUE INDEX", "UNIQUE KEY";
+my $indextype = join '|' => qw(INDEX KEY FULLTEXT), 'PRIMARY KEY', 'UNIQUE INDEX', 'UNIQUE KEY';
 $indextype = qr{$indextype};
 
 my $engine = qr{TYPE|ENGINE};
 
-my $tabletype = qr{InnoDB|MyISAM|HEAP|HEAP MAX_ROWS=\d+};
+my $tabletype = qr{InnoDB|MyISAM|HEAP|HEAP MAX_ROWS=\d+|InnoDB MAX_ROWS=\d+ AVG_ROW_LENGTH=\d+};
 
-my $charset = qr{utf8};
+my $charset = qr{utf8|binary};
 
-
-open my $newfh, "<", $new or die qq{Could not open $new: $!\n};
+open my $newfh, '<', $new or die qq{Could not open $new: $!\n};
 
 
 my ($table,%old);
@@ -84,7 +83,7 @@ sub parse_sql {
 
        my $oldfile = shift;
 
-       open my $oldfh, "<", $oldfile or die qq{Could not open $oldfile: $!\n};
+       open my $oldfh, '<', $oldfile or die qq{Could not open $oldfile: $!\n};
 
        my %info;
        while (<$oldfh>) {
@@ -98,6 +97,10 @@ sub parse_sql {
                        $table = $1;
                        $info{$table}{name}=$table;
                }
+               elsif (m#^\) /\*\$wgDBTableOptions\*/#) {
+                       $info{$table}{engine} = 'TYPE';
+                       $info{$table}{type} = 'variable';
+               }
                elsif (/^\) ($engine)=($tabletype);$/) {
                        $info{$table}{engine}=$1;
                        $info{$table}{type}=$2;
@@ -111,8 +114,8 @@ sub parse_sql {
                        $info{$table}{column}{$1} = $2;
                }
                elsif (/^  ($indextype)(?: (\w+))? \(([\w, \(\)]+)\),?$/) {
-                       $info{$table}{lc $1."_name"} = $2 ? $2 : "";
-                       $info{$table}{lc $1."pk_target"} = $3;
+                       $info{$table}{lc $1.'_name'} = $2 ? $2 : '';
+                       $info{$table}{lc $1.'pk_target'} = $3;
                }
                else {
                        die "Cannot parse line $. of $oldfile:\n$_\n";
@@ -137,6 +140,10 @@ for my $table (sort keys %{$old{$oldfile}}) {
                ($oldfile !~ /5/ and $t->{engine} ne 'TYPE')) {
                die "Invalid engine for $oldfile: $t->{engine}\n" unless $t->{name} eq 'profiling';
        }
+       my $charset = $t->{charset} || '';
+       if ($oldfile !~ /binary/ and $charset eq 'binary') {
+               die "Invalid charset for $oldfile: $charset\n";
+       }
 }
 
 my $dtype = join '|' => qw(
@@ -148,7 +155,7 @@ CIDR
 );
 $dtype = qr{($dtype)};
 my %new;
-my ($infunction,$inview,$inrule) = (0,0,0);
+my ($infunction,$inview,$inrule,$lastcomma) = (0,0,0,0);
 seek $newfh, 0, 0;
 while (<$newfh>) {
        next if /^\s*\-\-/ or /^\s*$/;
@@ -183,11 +190,19 @@ while (<$newfh>) {
        if (/^CREATE TABLE "?(\w+)"? \($/) {
                $table = $1;
                $new{$table}{name}=$table;
+               $lastcomma = 1;
        }
        elsif (/^\);$/) {
+               if ($lastcomma) {
+                       warn "Stray comma before line $.\n";
+               }
        }
-       elsif (/^  (\w+) +$dtype/) {
+       elsif (/^  (\w+) +$dtype.*?(,?)(?: --.*)?$/) {
                $new{$table}{column}{$1} = $2;
+               if (!$lastcomma) {
+                       print "Missing comma before line $. of $new\n";
+               }
+               $lastcomma = $3 ? 1 : 0;
        }
        else {
                die "Cannot parse line $. of $new:\n$_\n";
@@ -206,13 +221,13 @@ for my $t (sort keys %{$old{$oldfile}}) {
        my $newcol = $new{$newt}{column};
        for my $c (keys %$oldcol) {
                if (!exists $newcol->{$c}) {
-                       print "Column $t.$c not in new\n";
+                       print "Column $t.$c not in $new\n";
                        next;
                }
        }
        for my $c (keys %$newcol) {
                if (!exists $oldcol->{$c}) {
-                       print "Column $t.$c not in old\n";
+                       print "Column $t.$c not in $oldfile\n";
                        next;
                }
        }
@@ -220,7 +235,7 @@ for my $t (sort keys %{$old{$oldfile}}) {
 ## New but not old:
 for (sort keys %new) {
        if (!exists $old{$oldfile}{$_} and !exists $ok{NEW}{$_}) {
-               print "Not in old: $_\n";
+               print "Not in $oldfile: $_\n";
                next;
        }
 }
@@ -232,9 +247,7 @@ for (sort keys %new) {
 __DATA__
 ## Known exceptions
 OLD: searchindex          ## We use tsearch2 directly on the page table instead
-OLD: archive              ## This is a view due to the char(14) timestamp hack
 RENAME: user mwuser       ## Reserved word causing lots of problems
 RENAME: text pagecontent  ## Reserved word
-NEW: archive2             ## The real archive table
 NEW: mediawiki_version    ## Just us, for now
 XFILE: ../archives/patch-profiling.sql