| 1 |
#!/usr/bin/perl -w
|
| 2 |
# Last-Modified: <Mon Aug 18 14:29:47 2008>
|
| 3 |
|
| 4 |
use strict;
|
| 5 |
use warnings;
|
| 6 |
|
| 7 |
use FindBin '$Bin';
|
| 8 |
|
| 9 |
# We need our own copy of Debbugs::Status for now
|
| 10 |
use lib $Bin, qw{/org/udd.debian.org/mirrors/bugs.debian.org/perl};
|
| 11 |
|
| 12 |
use DBI;
|
| 13 |
use DBI qw{:sql_types};
|
| 14 |
use YAML::Syck;
|
| 15 |
use Time::Local;
|
| 16 |
|
| 17 |
use Debbugs::Bugs qw{get_bugs};
|
| 18 |
use Debbugs::Status qw{read_bug get_bug_status bug_presence};
|
| 19 |
use Debbugs::Packages qw{getpkgsrc};
|
| 20 |
use Debbugs::Config qw{:globals %config};
|
| 21 |
use Debbugs::User;
|
| 22 |
use Mail::Address;
|
| 23 |
#use Debbugs::User qw{read_usertags};
|
| 24 |
|
| 25 |
$YAML::Syck::ImplicitTyping = 1;
|
| 26 |
|
| 27 |
#Used for measuring time
|
| 28 |
our $t;
|
| 29 |
our $timing = 0;
|
| 30 |
our @archs = grep { !/(^m68k$|^kfreebsd|^hurd)/ } @{$config{default_architectures}};
|
| 31 |
|
| 32 |
# Return the list of usernames
|
| 33 |
sub get_bugs_users {
|
| 34 |
my $topdir = "$gSpoolDir/user";
|
| 35 |
my @ret = ();
|
| 36 |
# see Debbugs::User::filefromemail for why 0...6
|
| 37 |
for(my $i = 0; $i < 7; $i++) {
|
| 38 |
my $dir = "$topdir/$i";
|
| 39 |
opendir DIR, $dir or die "Can't open dir $dir: $!";
|
| 40 |
# Replace all occurences of %dd with the corresponding
|
| 41 |
# character represented by dd, where dd is a hexadecimal
|
| 42 |
# number
|
| 43 |
push @ret, map { s/%(..)/chr(hex($1))/ge; $_ } readdir DIR;
|
| 44 |
}
|
| 45 |
return @ret;
|
| 46 |
}
|
| 47 |
|
| 48 |
sub parse_time {
|
| 49 |
if(shift =~ /(\d\d\d\d)-(\d\d)-(\d\d) (\d\d):(\d\d):(\d\d)/) {
|
| 50 |
return ($1, $2, $3, $4, $5, $6);
|
| 51 |
}
|
| 52 |
return undef;
|
| 53 |
}
|
| 54 |
|
| 55 |
|
| 56 |
sub get_db_max_last_modified {
|
| 57 |
my $dbh = shift or die "Argument required";
|
| 58 |
my $sth = $dbh->prepare("SELECT MAX (last_modified) FROM bugs");
|
| 59 |
$sth->execute() or die $!;
|
| 60 |
my $date = $sth->fetchrow_array();
|
| 61 |
if(defined $date) {
|
| 62 |
my ($year, $month, $day, $hour, $minute, $second) = parse_time($date);
|
| 63 |
return timelocal($second, $minute, $hour, $day, $month-1, $year);
|
| 64 |
} else {
|
| 65 |
return 0;
|
| 66 |
}
|
| 67 |
}
|
| 68 |
|
| 69 |
sub get_mtime {
|
| 70 |
return ((stat(shift))[9]);
|
| 71 |
}
|
| 72 |
|
| 73 |
sub get_modified_bugs {
|
| 74 |
my $prune_stamp = shift;
|
| 75 |
die "Argument required" unless defined $prune_stamp;
|
| 76 |
my $top_dir = $gSpoolDir;
|
| 77 |
my @result = ();
|
| 78 |
foreach my $sub (qw(archive db-h)) {
|
| 79 |
my $spool = "$top_dir/$sub";
|
| 80 |
foreach my $subsub (glob "$spool/*") {
|
| 81 |
if( -d $subsub and get_mtime($subsub) > $prune_stamp ) {
|
| 82 |
push @result,
|
| 83 |
map { s{.*/(.*)\.log}{$1}; $_ }
|
| 84 |
grep { get_mtime("$_") > $prune_stamp }
|
| 85 |
glob "$subsub/*.log";
|
| 86 |
}
|
| 87 |
}
|
| 88 |
}
|
| 89 |
return \@result;
|
| 90 |
}
|
| 91 |
|
| 92 |
sub without_duplicates {
|
| 93 |
my %h = ();
|
| 94 |
return (grep { ($h{$_}++ == 0) || 0 } @_);
|
| 95 |
}
|
| 96 |
|
| 97 |
sub run_usertags {
|
| 98 |
my ($config, $source, $dbh) = @_;
|
| 99 |
my %src_config = %{$config->{$source}};
|
| 100 |
my $table = $src_config{'usertags-table'} or die "usertags-table not specified for source $source";
|
| 101 |
our $timing;
|
| 102 |
our $t;
|
| 103 |
|
| 104 |
|
| 105 |
$t = time();
|
| 106 |
# Free usertags table
|
| 107 |
$dbh->do("DELETE FROM $table") or die
|
| 108 |
"Couldn't empty $table: $!";
|
| 109 |
print "Deleting usertags: ",(time() - $t),"s\n" if $timing;
|
| 110 |
$t = time();
|
| 111 |
# read and insert user tags
|
| 112 |
my @users = get_bugs_users();
|
| 113 |
foreach my $user (@users) {
|
| 114 |
#read_usertags(\%tags, $user);
|
| 115 |
my $u = Debbugs::User->new($user);
|
| 116 |
my %tags = %{$u->{tags}};
|
| 117 |
$user = $dbh->quote($user);
|
| 118 |
foreach my $tag (keys %tags) {
|
| 119 |
my $qtag = $dbh->quote($tag);
|
| 120 |
map { $dbh->do("INSERT INTO $table (email, tag, id) VALUES ($user, $qtag, $_)") or die $! } @{$tags{$tag}};
|
| 121 |
}
|
| 122 |
}
|
| 123 |
}
|
| 124 |
|
| 125 |
sub run {
|
| 126 |
my ($config, $source, $dbh) = @_;
|
| 127 |
|
| 128 |
our $t;
|
| 129 |
our $timing;
|
| 130 |
my %src_config = %{$config->{$source}};
|
| 131 |
my $table = $src_config{table};
|
| 132 |
my $archived_table = $src_config{'archived-table'};
|
| 133 |
|
| 134 |
if (!$src_config{debug}) {
|
| 135 |
run_usertags($config, $source, $dbh);
|
| 136 |
print "Inserting usertags: ",(time() - $t),"s\n" if $timing;
|
| 137 |
$t = time();
|
| 138 |
}
|
| 139 |
|
| 140 |
|
| 141 |
my %pkgsrc = %{getpkgsrc()};
|
| 142 |
|
| 143 |
my @modified_bugs;
|
| 144 |
|
| 145 |
if($src_config{archived}) {
|
| 146 |
# some bugs (the unarchived ones) are in both list. exclude them.
|
| 147 |
my %unarchived;
|
| 148 |
foreach my $b (get_bugs()) {
|
| 149 |
$unarchived{$b} = 1;
|
| 150 |
}
|
| 151 |
foreach my $b (get_bugs(archive => 1)) {
|
| 152 |
push(@modified_bugs, $b) if not $unarchived{$b};
|
| 153 |
}
|
| 154 |
} else {
|
| 155 |
@modified_bugs = get_bugs();
|
| 156 |
}
|
| 157 |
my @modified_bugs2;
|
| 158 |
if ($src_config{debug}) {
|
| 159 |
print "Running in debug mode with restricted bug list!!\n";
|
| 160 |
foreach my $b (@modified_bugs) {
|
| 161 |
push(@modified_bugs2, $b) if ($b =~ /58$/);
|
| 162 |
}
|
| 163 |
@modified_bugs = @modified_bugs2;
|
| 164 |
}
|
| 165 |
|
| 166 |
print "Fetching list of ",scalar(@modified_bugs), " bugs to insert: ",(time() - $t),"s\n" if $timing;
|
| 167 |
$t = time();
|
| 168 |
|
| 169 |
foreach my $prefix ($table, $archived_table) {
|
| 170 |
foreach my $postfix (qw{_packages _merged_with _found_in _fixed_in _tags _blocks _blockedby}, '') {
|
| 171 |
my $sth = $dbh->prepare("DELETE FROM $prefix$postfix WHERE id = \$1");
|
| 172 |
map {
|
| 173 |
$sth->execute($_) or die $!;
|
| 174 |
} @modified_bugs;
|
| 175 |
}
|
| 176 |
}
|
| 177 |
print "Deleting bugs: ",(time() - $t),"s\n" if $timing;
|
| 178 |
$t = time();
|
| 179 |
|
| 180 |
my $location = $src_config{archived} ? 'archive' : 'db_h';
|
| 181 |
$table = $src_config{archived} ? $archived_table : $table;
|
| 182 |
# Read all bugs
|
| 183 |
my $insert_bugs_handle = $dbh->prepare("INSERT INTO $table (id, package, source, arrival, status, severity, submitter, submitter_name, submitter_email, owner, owner_name, owner_email, done, done_name, done_email, done_date, title, forwarded, last_modified, affects_oldstable, affects_stable, affects_testing, affects_unstable, affects_experimental) VALUES (\$1, \$2, \$3, \$4::abstime, \$5, \$6, \$7, \$8, \$9, \$10, \$11, \$12, \$13, \$14, \$15, \$16::abstime, \$17, \$18, \$19::abstime, \$20, \$21, \$22, \$23, \$24)");
|
| 184 |
my $insert_bugs_packages_handle = $dbh->prepare("INSERT INTO ${table}_packages (id, package, source) VALUES (\$1, \$2, \$3)");
|
| 185 |
my $insert_bugs_found_handle = $dbh->prepare("INSERT INTO ${table}_found_in (id, version) VALUES (\$1, \$2)");
|
| 186 |
my $insert_bugs_fixed_handle = $dbh->prepare("INSERT INTO ${table}_fixed_in (id, version) VALUES (\$1, \$2)");
|
| 187 |
my $insert_bugs_merged_handle = $dbh->prepare("INSERT INTO ${table}_merged_with (id, merged_with) VALUES (\$1, \$2)");
|
| 188 |
my $insert_bugs_tags_handle = $dbh->prepare("INSERT INTO ${table}_tags (id, tag) VALUES (\$1, \$2)");
|
| 189 |
my $insert_bugs_blocks_handle = $dbh->prepare("INSERT INTO ${table}_blocks (id, blocked) VALUES (\$1, \$2)");
|
| 190 |
my $insert_bugs_blockedby_handle = $dbh->prepare("INSERT INTO ${table}_blockedby (id, blocker) VALUES (\$1, \$2)");
|
| 191 |
$insert_bugs_handle->bind_param(4, undef, SQL_INTEGER);
|
| 192 |
$insert_bugs_handle->bind_param(18, undef, SQL_INTEGER);
|
| 193 |
|
| 194 |
$t = time();
|
| 195 |
foreach my $bug_nr (@modified_bugs) {
|
| 196 |
# Fetch bug using Debbugs
|
| 197 |
# Bugs which were once archived and have been unarchived again will appear in get_bugs(archive => 1).
|
| 198 |
# However, those bugs are not to be found in location 'archive', so we detect them, and skip them
|
| 199 |
my $bug_ref = read_bug(bug => $bug_nr, location => $location) or (print STDERR "Could not read file for bug $bug_nr; skipping\n" and next);
|
| 200 |
# Yeah, great, why does get_bug_status not accept a location?
|
| 201 |
my %bug = %{get_bug_status(bug => $bug_nr, status => $bug_ref)};
|
| 202 |
|
| 203 |
# Convert data where necessary
|
| 204 |
my @found_versions = @{$bug{found_versions}};
|
| 205 |
my @fixed_versions = @{$bug{fixed_versions}};
|
| 206 |
my @tags = split / /, $bug{keywords};
|
| 207 |
|
| 208 |
# log_modified and date are not necessarily set. If they are not available, they
|
| 209 |
# are assumed to be epoch (i.e. bug #4170)
|
| 210 |
map {
|
| 211 |
if($bug{$_}) {
|
| 212 |
$bug{$_} = int($bug{$_});
|
| 213 |
} else {
|
| 214 |
$bug{$_} = 0;
|
| 215 |
}
|
| 216 |
} qw{date log_modified done_date};
|
| 217 |
|
| 218 |
my $srcpkg;
|
| 219 |
if ($bug{package} =~ /^src:(.*)/)
|
| 220 |
{
|
| 221 |
$srcpkg = $1;
|
| 222 |
} else {
|
| 223 |
$srcpkg = exists($pkgsrc{$bug{package}}) ? $pkgsrc{$bug{package}} : $bug{package};
|
| 224 |
}
|
| 225 |
|
| 226 |
# split emails
|
| 227 |
my (@addr, $submitter_name, $submitter_email, $owner_name, $owner_email, $done_name, $done_email);
|
| 228 |
if ($bug{originator}) {
|
| 229 |
@addr = Mail::Address->parse($bug{originator});
|
| 230 |
$submitter_name = $addr[0]->phrase;
|
| 231 |
$submitter_email = $addr[0]->address;
|
| 232 |
} else {
|
| 233 |
$submitter_name = '';
|
| 234 |
$submitter_email = '';
|
| 235 |
}
|
| 236 |
|
| 237 |
if ($bug{owner}) {
|
| 238 |
@addr = Mail::Address->parse($bug{owner});
|
| 239 |
$owner_name = $addr[0]->phrase;
|
| 240 |
$owner_email = $addr[0]->address;
|
| 241 |
} else {
|
| 242 |
$owner_name = '';
|
| 243 |
$owner_email = '';
|
| 244 |
}
|
| 245 |
|
| 246 |
if ($bug{done}) {
|
| 247 |
@addr = Mail::Address->parse($bug{done});
|
| 248 |
$done_name = $addr[0]->phrase;
|
| 249 |
$done_email = $addr[0]->address;
|
| 250 |
} else {
|
| 251 |
$done_name = '';
|
| 252 |
$done_email = '';
|
| 253 |
}
|
| 254 |
|
| 255 |
#Calculate bug presence in distributions
|
| 256 |
my ($present_in_oldstable, $present_in_stable, $present_in_testing, $present_in_unstable, $present_in_experimental);
|
| 257 |
if($src_config{archived}) {
|
| 258 |
$present_in_oldstable = $present_in_stable = $present_in_testing = $present_in_unstable = $present_in_experimental = 'FALSE';
|
| 259 |
} else {
|
| 260 |
$present_in_oldstable =
|
| 261 |
bug_presence(bug => $bug_nr, status => \%bug,
|
| 262 |
dist => 'oldstable',
|
| 263 |
arch => \@archs);
|
| 264 |
$present_in_stable =
|
| 265 |
bug_presence(bug => $bug_nr, status => \%bug,
|
| 266 |
dist => 'stable',
|
| 267 |
arch => \@archs);
|
| 268 |
$present_in_testing =
|
| 269 |
bug_presence(bug => $bug_nr, status => \%bug,
|
| 270 |
dist => 'testing',
|
| 271 |
arch => \@archs);
|
| 272 |
$present_in_unstable =
|
| 273 |
bug_presence(bug => $bug_nr, status => \%bug,
|
| 274 |
dist => 'unstable',
|
| 275 |
arch => \@archs);
|
| 276 |
$present_in_experimental =
|
| 277 |
bug_presence(bug => $bug_nr, status => \%bug,
|
| 278 |
dist => 'experimental',
|
| 279 |
arch => \@archs);
|
| 280 |
|
| 281 |
if(!defined($present_in_oldstable) or !defined($present_in_stable) or !defined($present_in_unstable) or !defined($present_in_testing) or !defined($present_in_experimental)) {
|
| 282 |
print "NUMBER: $bug_nr\n";
|
| 283 |
}
|
| 284 |
|
| 285 |
if(defined($present_in_oldstable) and ($present_in_oldstable eq 'absent' or $present_in_oldstable eq 'fixed')) {
|
| 286 |
$present_in_oldstable = 'FALSE';
|
| 287 |
} else {
|
| 288 |
$present_in_oldstable = 'TRUE';
|
| 289 |
}
|
| 290 |
if(defined($present_in_stable) and ($present_in_stable eq 'absent' or $present_in_stable eq 'fixed')) {
|
| 291 |
$present_in_stable = 'FALSE';
|
| 292 |
} else {
|
| 293 |
$present_in_stable = 'TRUE';
|
| 294 |
}
|
| 295 |
if(defined($present_in_testing) and ($present_in_testing eq 'absent' or $present_in_testing eq 'fixed')) {
|
| 296 |
$present_in_testing = 'FALSE';
|
| 297 |
} else {
|
| 298 |
$present_in_testing = 'TRUE';
|
| 299 |
}
|
| 300 |
if(defined($present_in_unstable) and ($present_in_unstable eq 'absent' or $present_in_unstable eq 'fixed')) {
|
| 301 |
$present_in_unstable = 'FALSE';
|
| 302 |
} else {
|
| 303 |
$present_in_unstable = 'TRUE';
|
| 304 |
}
|
| 305 |
if(defined($present_in_experimental) and ($present_in_experimental eq 'absent' or $present_in_experimental eq 'fixed')) {
|
| 306 |
$present_in_experimental = 'FALSE';
|
| 307 |
} else {
|
| 308 |
$present_in_experimental = 'TRUE';
|
| 309 |
}
|
| 310 |
}
|
| 311 |
|
| 312 |
# Insert data into bugs table
|
| 313 |
$insert_bugs_handle->execute($bug_nr, $bug{package}, $srcpkg, $bug{date}, $bug{pending},
|
| 314 |
$bug{severity}, $bug{originator}, $submitter_name, $submitter_email, $bug{owner},
|
| 315 |
$owner_name, $owner_email, $bug{done}, $done_name, $done_email, $bug{done_date},
|
| 316 |
$bug{subject}, $bug{forwarded}, $bug{log_modified},
|
| 317 |
$present_in_oldstable, $present_in_stable, $present_in_testing, $present_in_unstable, $present_in_experimental) or die $!;
|
| 318 |
|
| 319 |
my $src;
|
| 320 |
foreach my $pkg (keys %{{ map { $_ => 1 } split(/\s*[, ]\s*/, $bug{package})}}) {
|
| 321 |
if ($pkg =~ /^src:(.*)/)
|
| 322 |
{
|
| 323 |
$src = $1;
|
| 324 |
} else {
|
| 325 |
$src = exists($pkgsrc{$pkg}) ? $pkgsrc{$pkg} : $pkg;
|
| 326 |
}
|
| 327 |
$insert_bugs_packages_handle->execute($bug_nr, $pkg, $src) or die $!;
|
| 328 |
}
|
| 329 |
|
| 330 |
# insert data into bug_fixed_in and bug_found_in tables
|
| 331 |
foreach my $version (without_duplicates(@found_versions)) {
|
| 332 |
$insert_bugs_found_handle->execute($bug_nr, $version) or die $!;
|
| 333 |
}
|
| 334 |
foreach my $version (without_duplicates(@fixed_versions)) {
|
| 335 |
$insert_bugs_fixed_handle->execute($bug_nr, $version) or die $!;
|
| 336 |
}
|
| 337 |
foreach my $mergee (without_duplicates(split / /, $bug{mergedwith})) {
|
| 338 |
$insert_bugs_merged_handle->execute($bug_nr, $mergee) or die $!;
|
| 339 |
}
|
| 340 |
foreach my $blocked (without_duplicates(split / /, $bug{blocks})) {
|
| 341 |
$insert_bugs_blocks_handle->execute($bug_nr, $blocked) or die $!;
|
| 342 |
}
|
| 343 |
foreach my $blocker (without_duplicates(split / /, $bug{blockedby})) {
|
| 344 |
$insert_bugs_blockedby_handle->execute($bug_nr, $blocker) or die $!;
|
| 345 |
}
|
| 346 |
foreach my $tag (without_duplicates(@tags)) {
|
| 347 |
$insert_bugs_tags_handle->execute($bug_nr, $tag) or die $!;
|
| 348 |
}
|
| 349 |
}
|
| 350 |
print "Inserting bugs: ",(time() - $t),"s\n" if $timing;
|
| 351 |
|
| 352 |
foreach my $postfix (qw{_packages _merged_with _found_in _fixed_in _tags}, '') {
|
| 353 |
my $sth = $dbh->prepare("ANALYZE $table$postfix");
|
| 354 |
$sth->execute() or die $!;
|
| 355 |
}
|
| 356 |
|
| 357 |
my $sth = $dbh->prepare("ANALYZE ".$src_config{'usertags-table'});
|
| 358 |
$sth->execute() or die $!;
|
| 359 |
|
| 360 |
print "Analyzing bugs: ",(time() - $t),"s\n" if $timing;
|
| 361 |
|
| 362 |
$dbh->commit();
|
| 363 |
print "Committing bugs: ",(time() - $t),"s\n" if $timing;
|
| 364 |
}
|
| 365 |
|
| 366 |
sub main {
|
| 367 |
if(@ARGV != 3) {
|
| 368 |
print STDERR "Usage: $0 <config> <command> <source>\n";
|
| 369 |
exit 1;
|
| 370 |
}
|
| 371 |
|
| 372 |
our $t = time();
|
| 373 |
our $timing;
|
| 374 |
|
| 375 |
my $config = LoadFile($ARGV[0]) or die "Could not load configuration: $!";
|
| 376 |
my $command = $ARGV[1];
|
| 377 |
my $source = $ARGV[2];
|
| 378 |
|
| 379 |
my $dbname = $config->{general}->{dbname};
|
| 380 |
my $dbport;
|
| 381 |
if ($config->{general}->{dbport} ne '') {
|
| 382 |
$dbport = ";port=".$config->{general}->{dbport};
|
| 383 |
} else {
|
| 384 |
$dbport = "";
|
| 385 |
}
|
| 386 |
# Connection to DB
|
| 387 |
my $dbh = DBI->connect("dbi:Pg:dbname=$dbname".$dbport);
|
| 388 |
# We want to commit the transaction as a hole at the end
|
| 389 |
$dbh->{AutoCommit} = 0;
|
| 390 |
$dbh->do('SET CONSTRAINTS ALL DEFERRED');
|
| 391 |
|
| 392 |
if($command eq 'run') {
|
| 393 |
run($config, $source, $dbh);
|
| 394 |
} else {
|
| 395 |
print STDERR "<command> has to be one of run, drop and setup\n";
|
| 396 |
exit(1)
|
| 397 |
}
|
| 398 |
|
| 399 |
}
|
| 400 |
|
| 401 |
main();
|
Parent Directory
| 
Revision Log