/[pgdiff]/pgdiff
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /pgdiff

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.6 - (hide annotations)
Wed Aug 13 13:27:32 2003 UTC (20 years, 7 months ago) by dpavlin
Branch: MAIN
Changes since 1.5: +18 -9 lines
fixed insert

1 dpavlin 1.1 #!/usr/bin/perl -w
2    
3     # Compare data in two different databases
4     #
5     # 2003-08-07 Dobrica Pavlinusic
6     #
7     # somewhat based on ides from
8     # DataDiff 0.1 by Jon D. Frisby, http://www.mrjoy.com
9    
10     use strict;
11     use Getopt::Long;
12     use DBI;
13     use Data::Dumper;
14 dpavlin 1.4 use Pg::Scheme;
15 dpavlin 1.1
16     $| = 1;
17    
18     my ($debug,$verbose) = (0,0);
19     my ($help,$masterhost,$masterport,$masteruser,$masterpassword,
20     $slavehost,$slaveport,$slaveuser,$slavepassword,
21     $masterfile,
22     $slavefile,
23     );
24     my $tables;
25    
26     my $result = GetOptions(
27     "debug!" => \$debug, "verbose!" => \$verbose, "help" => \$help,
28     "masterhost=s" => \$masterhost, "masterport=i" => \$masterport,
29     "masteruser=s" => \$masteruser, "masterpassword=s" => \$masterpassword,
30     "masterfile=s" => \$masterfile,
31     "slavehost=s" => \$slavehost, "slaveport=i" => \$slaveport,
32     "slaveuser=s" => \$slaveuser, "slavepassword=s" => \$slavepassword,
33     "slavefile=s" => \$slavefile,
34     "tables=s" => \$tables,
35     );
36    
37     if (defined($help) || (scalar(@ARGV) < 2)) {
38     print "Usage: $0 [options] masterdb slavedb
39     Options:
40     --masterhost=hostname --masterport=port
41     --masteruser=username --masterpassword=string
42     --masterfile=filename
43     --slavehost=hostname --slaveport=port
44     --slaveuser=username --slavepassword=string
45     --slavefile=filename
46     --tables[s]=table[,table...]
47     ";
48     # exit ((scalar(@ARGV) < 2)? 1:0);
49     exit;
50     }
51    
52     my $master = $ARGV[0] || "master";
53     my $slave = $ARGV[1] || "slave";
54    
55     my $minfo = "dbname=$master";
56     $minfo = "$minfo host=$masterhost" if (defined($masterhost));
57     $minfo = "$minfo port=$masterport" if (defined($masterport));
58    
59     my $sinfo = "dbname=$slave";
60     $sinfo = "$sinfo host=$slavehost" if (defined($slavehost));
61     $sinfo = "$sinfo port=$slaveport" if (defined($slaveport));
62    
63     print "Master connection is $minfo\n" if ($debug);
64     print "Slave connection is $sinfo\n" if ($debug);
65    
66     my $mdbh = DBI->connect("DBI:Pg:$minfo", $masteruser, $masterpassword, { PrintError => 1 });
67     my $sdbh = DBI->connect("DBI:Pg:$sinfo", $slaveuser, $slavepassword, { PrintError => 1 });
68    
69     my ($diff_shema,$diff_data) = (0,0);
70    
71     my $sql;
72    
73     sub debug_sql {
74     return if (! $debug);
75     my $sql = shift;
76     $sql =~ s/[\n\r]/ /gs;
77     $sql =~ s/\s\s+/ /g;
78 dpavlin 1.2 print STDERR "DEBUG: SQL: $sql\n";
79 dpavlin 1.1 }
80    
81     sub debug_row {
82     return if (! $debug);
83     my $row = shift;
84     my @cols = @_;
85     if (! $row) {
86 dpavlin 1.2 print STDERR "DEBUG: ROW data is undef!\n";
87 dpavlin 1.1 return;
88     }
89 dpavlin 1.2 print STDERR "DEBUG: ROW: [",$#cols+1,"] ";
90 dpavlin 1.1 foreach my $col (@cols) {
91     print STDERR "$col:";
92     if ($row->{$col}) {
93     print $row->{$col};
94     } else {
95     print "null";
96     }
97     print " ";
98     }
99     print STDERR "\n";
100     }
101    
102 dpavlin 1.2 sub debug {
103     return if (!$debug);
104     print STDERR "DEBUG: ",@_;
105     }
106    
107 dpavlin 1.1 $verbose = 1 if ($debug);
108    
109 dpavlin 1.4 # init object for scheme in master database
110 dpavlin 1.5 my $mscheme = new Pg::Scheme( 'dbh' => $mdbh, 'DEBUG' => 0 ) || die "can't query schema";
111 dpavlin 1.4
112 dpavlin 1.1 # which tables to compare?
113    
114 dpavlin 1.4 my @tables = $mscheme->list_tables($tables);
115 dpavlin 1.1
116 dpavlin 1.2 debug "Comparing tables: ".join(", ",@tables)."\n";
117 dpavlin 1.1
118     my $cols;
119     my $diff_total = 0;
120    
121     foreach my $table (@tables) {
122    
123 dpavlin 1.4 my $sth;
124 dpavlin 1.1
125     # diff schema
126    
127 dpavlin 1.5 # all colums (for insert)
128     my @cols = @{$mscheme->cols($table)};
129    
130     # colums compared by a=b
131     my @cols_notnull = @{$mscheme->cols_notnull($table)};
132    
133     # colums compared by a=b or a is null and b is null
134     my @cols_null = @{$mscheme->cols_null($table)};
135    
136     # primary key columns
137     my @cols_pk = @{$mscheme->cols_pk($table)};
138    
139     # columns to compare (not in primary key)
140     my @cols_cmp = @{$mscheme->cols_notpk($table)};
141    
142 dpavlin 1.1 my @cols_skip; # skipped columns
143     my @cols_test; # all colums to test (without skipped)
144    
145 dpavlin 1.5 foreach my $row (@{$mscheme->pg_attribute($table)}) {
146 dpavlin 1.1 # attname | format_type | attnotnull | atthasdef | attnum
147    
148     # FIXME: do something with attributes which shouldn't be compared
149     # (date, time, datetime, timestamp)
150     if ($row->{format_type} =~ /(date)|(time)/i) {
151     push @cols_skip,$row->{attname};
152 dpavlin 1.5 } else {
153     push @cols_test,$row->{attname};
154 dpavlin 1.1 }
155    
156     }
157    
158     if ($debug) {
159 dpavlin 1.4 print STDERR "DEBUG: table $table not null cols: (",join(", ",@cols_notnull),")";
160     print STDERR " - null cols: (",join(", ",@cols_null),")" if (@cols_null);
161     print STDERR " - skip cols: (",join(", ",@cols_skip),")" if (@cols_skip);
162 dpavlin 1.1 print STDERR "\n";
163     }
164    
165     # diff data
166    
167     if (! @cols_pk) {
168     print STDERR "can't find PK rows for table '$table' using all\n";
169     @cols_pk = @cols;
170     }
171    
172    
173     if ($verbose) {
174     print "table '$table' using for key: (",join(", ",@cols_pk),") to compare cols: (",join(", ",@cols_cmp),")\n";
175     }
176    
177     # diff data
178    
179     my $msql = "select ".join(",",@cols)." from $table";
180     my $ssql = $msql;
181    
182     sub sql_where {
183     my @cols = @_;
184     my $and = "";
185     my $where = " where ";
186     foreach my $col (@cols) {
187     $where .= "$and$col=?";
188     $and = " and ";
189     }
190     return $where;
191     }
192    
193     sub sql_order {
194     my @cols = @_;
195     my $order = " order by ";
196     my $comma = "";
197     foreach my $col (@cols) {
198     $order .= "$comma$col asc";
199     $comma = ", ";
200     }
201     return $order;
202     }
203    
204     my $order = sql_order(@cols_pk);
205     $msql .= $order;
206 dpavlin 1.2 $ssql .= $order;
207 dpavlin 1.1
208     debug_sql($msql);
209    
210     my $msth = $mdbh->prepare($msql) || die;
211     $msth->execute() || die;
212    
213     my $ssth = $sdbh->prepare($ssql) || die;
214 dpavlin 1.2 $ssth->execute() || die;
215 dpavlin 1.1
216     my $diff_row = 0;
217    
218     my ($mrow,$srow);
219 dpavlin 1.2 # have_*
220     use constant NO_ROW => 0;
221     use constant FETCH_ROW => 1;
222     use constant HAVE_ROW => 2;
223     my ($have_mrow,$have_srow) = (FETCH_ROW,FETCH_ROW);
224    
225     while ($have_mrow != NO_ROW || $have_srow != NO_ROW) {
226    
227     debug "have mrow: $have_mrow srow: $have_srow\n";
228    
229     sub pk_val {
230     my $row = shift || die "need row";
231     my @cols = shift || die "need cols";
232     my @val;
233     foreach my $col (@cols) {
234     push @val,$row->{$col};
235     }
236     return @val;
237     }
238 dpavlin 1.1
239     # fetch row from master
240 dpavlin 1.2 if ($have_mrow == FETCH_ROW) {
241     debug "fetch row from master: $msql\n";
242 dpavlin 1.1 $mrow = $msth->fetchrow_hashref();
243     debug_row($mrow,@cols);
244    
245     if ($mrow) {
246     # fill-in primary key values
247 dpavlin 1.2 $have_mrow = HAVE_ROW;
248 dpavlin 1.1 } else {
249 dpavlin 1.2 $have_mrow = NO_ROW;
250 dpavlin 1.1 }
251     }
252    
253     # fetch row from slave
254 dpavlin 1.2 if ($have_srow == FETCH_ROW) {
255     debug "fetch row from slave: $ssql\n";
256 dpavlin 1.1 $srow = $ssth->fetchrow_hashref();
257     debug_row($srow,@cols);
258     if ($srow) {
259 dpavlin 1.2 $have_srow = HAVE_ROW;
260 dpavlin 1.1 } else {
261 dpavlin 1.2 $have_srow = NO_ROW;
262 dpavlin 1.1 }
263     }
264    
265 dpavlin 1.2 debug "have mrow: $have_mrow srow: $have_srow\n";
266 dpavlin 1.1
267     # insert into slave database
268     sub sql_insert {
269 dpavlin 1.6 my $dbh = shift @_ || die "need dbh";
270 dpavlin 1.1 my $table = shift @_ || die "need table as argument";
271     my $row = shift @_ || die "need row data";
272     my @cols = @_;
273    
274     my $sql = "insert into $table (".join(",",@cols).") values (";
275     my $comma = "";
276     foreach my $col (@cols) {
277 dpavlin 1.6 $sql .= $comma.$dbh->quote($row->{$col});
278 dpavlin 1.1 $comma = ",";
279     }
280     $sql.=")";
281     debug_sql($sql);
282     return $sql;
283     }
284    
285     # delete from slave database
286     sub sql_delete {
287 dpavlin 1.6 my $dbh = shift @_ || die "need dbh";
288 dpavlin 1.1 my $table = shift @_ || die "need table as argument";
289     my $row = shift @_ || die "need row as argument";
290     my @cols_pk = @_;
291    
292     my $where = sql_where(@cols_pk);
293    
294 dpavlin 1.6 my $sql = "delete from $table";
295 dpavlin 1.1 foreach my $col (@cols_pk) {
296 dpavlin 1.6 my $val = $dbh->quote($row->{$col}) || die "can't find value in row for col $col";
297 dpavlin 1.1 $where =~ s/\?/$val/;
298     }
299     $sql .= $where;
300     debug_sql($sql);
301     return $sql;
302     }
303    
304 dpavlin 1.3 # update row in slave database
305     sub sql_update {
306 dpavlin 1.6 my $dbh = shift @_ || die "need dbh";
307 dpavlin 1.3 my $table = shift @_ || die "need table as argument";
308     my $col = shift @_ || die "need col to update";
309 dpavlin 1.6 my $row = shift @_ || die "need row";
310     my @cols_pk = @_;
311 dpavlin 1.3
312 dpavlin 1.6 my $sql = "update $table set $col=".$dbh->quote($row->{$col});
313     my $where = sql_where(@cols_pk);
314     foreach my $col (@cols_pk) {
315     my $val = $dbh->quote($row->{$col}) || die "can't find value in row for col $col";
316     $where =~ s/\?/$val/;
317     }
318     $sql .= $where;
319 dpavlin 1.3 debug_sql($sql);
320     return $sql;
321     }
322 dpavlin 1.1 # master slave
323     # 1 = 1 test
324     # 1 < 2 insert mrow
325     # 2 > 1 delete srow
326     # 1 = undef insert mrow
327     # undef = 1 delete srow
328    
329     my $pk_same = 1;
330    
331     # check key cols for row
332     foreach my $col (@cols_pk) {
333 dpavlin 1.2 if ( ($have_mrow == NO_ROW && $have_srow == HAVE_ROW) ||
334     ($have_mrow == HAVE_ROW && $have_srow == HAVE_ROW && $mrow->{$col} gt $srow->{$col}) ) {
335 dpavlin 1.1 $diff_row++;
336     $pk_same = 0;
337 dpavlin 1.3 print STDERR "EXTRA row in table '$table' pk: [".join(",",@cols_pk)."] value (".join(",",pk_val($srow,@cols_pk)).")\n" if ($verbose);
338 dpavlin 1.6 print sql_delete($sdbh,$table,$srow,@cols_pk),";\n";
339 dpavlin 1.2 $have_srow = FETCH_ROW;
340 dpavlin 1.1 last;
341 dpavlin 1.2 } elsif ( ($have_mrow == HAVE_ROW && $have_srow == NO_ROW) ||
342     ($have_mrow == HAVE_ROW && $have_srow == HAVE_ROW && $mrow->{$col} lt $srow->{$col}) ) {
343 dpavlin 1.1 $diff_row++;
344     $pk_same = 0;
345 dpavlin 1.3 print STDERR "MISSING row in table '$table' pk: [".join(",",@cols_pk)."] value (".join(",",pk_val($mrow,@cols_pk)).")\n" if ($verbose);
346 dpavlin 1.6 print sql_insert($mdbh,$table,$mrow,@cols),";\n";
347 dpavlin 1.2 $have_mrow = FETCH_ROW;
348 dpavlin 1.1 last;
349     }
350     }
351    
352 dpavlin 1.2 if ($pk_same && $have_mrow == HAVE_ROW && $have_srow == HAVE_ROW) {
353 dpavlin 1.1 # check non-key cols for row
354     foreach my $col (@cols_cmp) {
355     if ($mrow->{$col} ne $srow->{$col}) {
356     $diff_row++;
357 dpavlin 1.2 print STDERR "DIFF in table '$table' row ($col): [".join(",",@cols_pk)."] '$mrow->{$col}' != '$srow->{$col}'\n" if ($verbose);
358 dpavlin 1.6 print sql_update($mdbh,$table,$col,$mrow,@cols_pk),";\n";
359 dpavlin 1.1 }
360     }
361 dpavlin 1.3 $have_mrow = FETCH_ROW;
362     $have_srow = FETCH_ROW;
363 dpavlin 1.1 }
364     }
365    
366     print STDERR "$diff_row differences in table $table\n" if ($verbose && $diff_row > 0);
367     $diff_total += $diff_row;
368     }
369    
370 dpavlin 1.4 if ($verbose) {
371     if ($diff_total == 0) {
372     print STDERR "databases are same";
373     } elsif ($diff_total > 0) {
374     print STDERR "$diff_total differences in all tables\n";
375     } else {
376     die "this shouldn't happend. please report a bug!";
377     }
378     }
379 dpavlin 1.1
380     $mdbh->disconnect();
381     $sdbh->disconnect();

  ViewVC Help
Powered by ViewVC 1.1.26