2 # debbugs-loadsql is part of debbugs, and is released
3 # under the terms of the GPL version 2, or any later version, at your
4 # option. See the file README and COPYING for more information.
5 # Copyright 2012 by Don Armstrong <don@donarmstrong.com>.
11 use Getopt::Long qw(:config no_ignore_case);
16 debbugs-loadsql -- load debbugs sql database
20 debbugs-loadsql [options] [subcommand]
23 bugs help versions configuration
24 suites logs packages debinfo
26 --quick, -q only load changed things
27 --progress Show progress bar
28 --service, -s service name
29 --sysconfdir, -c postgresql service config dir
30 --spool-dir debbugs spool directory
31 --debug, -d debugging level (Default 0)
32 --help, -h display this help
33 --man, -m display manual
43 Add bugs (subject, number, etc) to the database
45 --preload create all bugs first, then add information
49 Add version descendant information (which version is based on which version) to
54 Add source maintainers to the BTS
58 Add debbugs configuration information (tags, severity, etc)
62 Add suite information from ftp distribution
64 --ftpdists location of FTP mirror
72 Add package information from the ftp archive
74 --ftpdists location of FTP mirror
75 --suites Suite to operate on
79 Add package information from a debinfo file
81 --null -0 names of debinfo files are null separated
89 Only load changed bugs
93 Show progress bar (requires Term::ProgressBar)
97 Postgreql service to use; defaults to debbugs
99 =item B<--sysconfdir,-c>
101 System configuration directory to use; if not set, defaults to the
102 postgresql default. [Operates by setting PGSYSCONFDIR]
106 Debbugs spool directory; defaults to the value configured in the
107 debbugs configuration file.
111 Output more information about what is happening. Probably not useful
112 if you also set --progress.
120 Display brief useage information.
134 use Debbugs::Common (qw(checkpid lockpid get_hashname getparsedaddrs getbugcomponent make_list getsourcemaintainers),
135 qw(hash_slice open_compressed_file),);
136 use Debbugs::Config qw(:config);
137 use Debbugs::Status qw(read_bug split_status_fields);
140 use Debbugs::DB::Load qw(:load_bug :load_package :load_suite);
147 use IO::Uncompress::AnyUncompress;
148 use Encode qw(decode_utf8);
149 use List::MoreUtils qw(natatime);
158 service => $config{debbugs_db},
162 Getopt::Long::Configure('pass_through');
163 GetOptions(\%options,
168 'spool_dir|spool-dir=s',
171 'debug|d+','help|h|?','man|m');
172 Getopt::Long::Configure('default');
174 pod2usage() if $options{help};
175 pod2usage({verbose=>2}) if $options{man};
177 $DEBUG = $options{debug};
180 ('bugs' => {function => \&add_bugs,
181 arguments => {'preload' => 0},
183 'versions' => {function => \&add_versions,
185 'debinfo' => {function => \&add_debinfo,
186 arguments => {'0|null' => 0},
188 'maintainers' => {function => \&add_maintainers,
190 'configuration' => {function => \&add_configuration,
192 'suites' => {function => \&add_suite,
193 arguments => {'ftpdists=s' => 1,
196 'logs' => {function => \&add_logs,
198 'packages' => {function => \&add_packages,
199 arguments => {'ftpdists=s' => 1,
203 'help' => {function => sub {pod2usage({verbose => 2});}}
207 $options{verbose} = $options{verbose} - $options{quiet};
209 if ($options{progress}) {
210 eval "use Term::ProgressBar";
211 push @USAGE_ERRORS, "You asked for a progress bar, but Term::ProgressBar isn't installed" if $@;
215 pod2usage(join("\n",@USAGE_ERRORS)) if @USAGE_ERRORS;
217 if (exists $options{sysconfdir}) {
218 if (not defined $options{sysconfdir} or not length $options{sysconfdir}) {
219 delete $ENV{PGSYSCONFDIR};
221 $ENV{PGSYSCONFDIR} = $options{sysconfdir};
225 if (exists $options{spool_dir} and defined $options{spool_dir}) {
226 $config{spool_dir} = $options{spool_dir};
230 if ($options{progress}) {
231 $prog_bar = eval "Term::ProgressBar->new({count => 1,ETA=>q(linear)})";
232 warn "Unable to initialize progress bar: $@" if not $prog_bar;
236 my ($subcommand) = shift @ARGV;
237 if (not defined $subcommand) {
238 $subcommand = 'help';
239 print STDERR "You must provide a subcommand; displaying usage.\n";
241 } elsif (not exists $subcommands{$subcommand}) {
242 print STDERR "$subcommand is not a valid subcommand; displaying usage.\n";
246 binmode(STDOUT,':encoding(UTF-8)');
247 binmode(STDERR,':encoding(UTF-8)');
250 handle_subcommand_arguments(\@ARGV,$subcommands{$subcommand}{arguments});
251 $subcommands{$subcommand}{function}->(\%options,$opts,$prog_bar,\%config,\@ARGV);
254 my ($options,$opts,$p,$config,$argv) = @_;
255 chdir($config->{spool_dir}) or
256 die "chdir $config->{spool_dir} failed: $!";
258 my $verbose = $options->{debug};
260 my $initialdir = "db-h";
262 if (defined $argv->[0] and $argv->[0] eq "archive") {
263 $initialdir = "archive";
265 my $s = db_connect($options);
269 my $start_time = time;
274 if ($opts->{preload}) {
276 walk_bugs([(@{$argv}?@{$argv} : $initialdir)],
285 $s->resultset('Bug')->quick_insert_bugs(@bugs);
287 walk_bugs([(@{$argv}?@{$argv} : $initialdir)],
294 if ($options{quick}) {
295 for my $bug (@bugs) {
296 my $stat = stat(getbugcomponent($bug,'summary',$initialdir));
297 if (not defined $stat) {
298 print STDERR "Unable to stat $bug $!\n";
301 my $rs = $s->resultset('Bug')->search({id=>$bug})->single();
302 next if defined $rs and $stat->mtime <= $rs->last_modified()->epoch();
303 push @bugs_to_update, $bug;
306 @bugs_to_update = @bugs;
310 for my $bug (@bugs_to_update) {
314 severities => \%severities,
320 die "failure while trying to load bug: $@";
325 handle_load_bug_queue(db => $s,
330 my ($options,$opts,$p,$config,$argv) = @_;
332 my $s = db_connect($options);
334 my @files = @{$argv};
335 $p->target(scalar @files) if $p;
336 for my $file (@files) {
337 my $fh = IO::File->new($file,'r') or
338 die "Unable to open $file for reading: $!";
343 next unless length $_;
344 if (/(\w[-+0-9a-z.]+) \(([^\(\) \t]+)\)/) {
345 push @versions, [$1,$2];
350 for my $i (reverse 0..($#versions)) {
352 if (not defined $src_pkgs{$versions[$i][0]}) {
353 $src_pkgs{$versions[$i][0]} =
354 $s->resultset('SrcPkg')->
355 get_src_pkg_id($versions[$i][0]);
357 $sp = $src_pkgs{$versions[$i][0]};
358 # There's probably something wrong if the source package
359 # doesn't exist, but we'll skip it for now
360 next unless defined $sp;
361 my $sv = $s->resultset('SrcVer')->find({src_pkg=>$sp,
362 ver => $versions[$i][1],
364 if (defined $ancestor_sv and defined $sv and not defined $sv->based_on()) {
365 $sv->update({based_on => $ancestor_sv})
367 $ancestor_sv = $sv->id();
375 my ($options,$opts,$p,$config,$argv) = @_;
377 my @files = @{$argv};
381 local $/ = "\0" if $opts->{0};
383 s/\n$// unless $opts->{0};
384 s/\0$// if $opts->{0};
389 return unless @files;
390 my $s = db_connect($options);
391 $p->target(scalar @files) if $p;
392 my $it = natatime 100, @files;
393 while (my @v = $it->()) {
397 my $fh = IO::File->new($file,'r') or
398 die "Unable to open $file for reading: $!";
399 my $f_stat = stat($file);
400 my $ct_date = DateTime->from_epoch(epoch => $f_stat->ctime);
403 next unless length $_;
404 my ($binname, $binver, $binarch, $srcname, $srcver) = split;
405 # if $srcver is not defined, this is probably a broken
406 # .debinfo file [they were causing #686106, see commit
407 # 49c85ab8 in dak.] Basically, $binarch didn't get put into
408 # the file, so we'll fudge it from the filename.
409 if (not defined $srcver) {
410 ($srcname,$srcver) = ($binarch,$srcname);
411 ($binarch) = $file =~ /_([^\.]+)\.debinfo/;
413 if (not defined $srcver) {
414 print STDERR "malformed debinfo (no srcver): $file\n";
418 [$binname,$binver,$binarch,$srcname,$srcver,$ct_date];
423 for my $di (@debinfos) {
424 Debbugs::DB::Load::load_debinfo($s,@{$di}[0..5],\%cache);
427 $p->update($p->last_update()+@v) if $p;
432 sub add_maintainers {
433 my ($options,$opts,$p,$config,$argv) = @_;
435 my $s = db_connect($options);
436 my $maintainers = getsourcemaintainers();
438 ## get all of the maintainers, and add the missing ones
439 my $maints = $s->resultset('Maintainer')->
440 get_maintainers(values %{$maintainers});
442 my @svs = $s->resultset('SrcVer')->
443 search({maintainer => undef
446 group_by => 'me.src_pkg, src_pkg.pkg',
447 result_class => 'DBIx::Class::ResultClass::HashRefInflator',
448 columns => [qw(me.src_pkg src_pkg.pkg)],
451 $p->target(2+@svs) if $p;
454 if (exists $maintainers->{$sv->{src_pkg}{pkg}}) {
455 my $pkg = $sv->{src_pkg}{pkg};
456 my $maint = $maints->
457 {$maintainers->{$pkg}};
458 $s->txn_do(sub {$s->resultset('SrcVer')->
459 search({maintainer => undef,
460 'src_pkg.pkg' => $pkg
463 )->update({maintainer => $maint})
471 sub add_configuration {
472 my ($options,$opts,$p,$config,$argv) = @_;
474 my $s = db_connect($options);
479 for my $tag (@{$config{tags}}) {
481 $s->resultset('Tag')->find_or_create({tag => $tag});
484 for my $tag ($s->resultset('Tag')->search_rs()->all()) {
485 next if exists $tags{$tag->tag};
493 for my $sev_name (($config{default_severity},@{$config{severity_list}})) {
494 # add all severitites
495 my $sev = $s->resultset('Severity')->find_or_create({severity => $sev_name});
496 # mark strong severities
497 if (grep {$_ eq $sev_name} @{$config{strong_severities}}) {
500 $sev->ordering($order);
503 $sev_names{$sev_name} = 1;
505 # mark obsolete severities
506 for my $sev ($s->resultset('Severity')->search_rs()->all()) {
507 next if exists $sev_names{$sev->severity()};
514 my ($options,$opts,$p,$config,$argv) = @_;
517 my $s = db_connect($options);
518 my $dist_dir = IO::Dir->new($opts->{ftpdists});
520 grep { $_ !~ /^\./ and
521 -d $opts->{ftpdists}.'/'.$_ and
522 not -l $opts->{ftpdists}.'/'.$_
524 while (my $dist = shift @dist_names) {
525 my $dist_dir = $opts->{ftpdists}.'/'.$dist;
526 my ($dist_info,$package_files) =
527 read_release_file($dist_dir.'/Release');
528 load_suite($s,$dist_info);
533 my ($options,$opts,$p,$config,$argv) = @_;
535 chdir($config->{spool_dir}) or
536 die "chdir $config->{spool_dir} failed: $!";
538 my $verbose = $options->{debug};
540 my $initialdir = "db-h";
542 if (defined $argv->[0] and $argv->[0] eq "archive") {
543 $initialdir = "archive";
545 my $s = db_connect($options);
549 my $start_time = time;
551 walk_bugs([(@{$argv}?@{$argv} : $initialdir)],
557 my $stat = stat(getbugcomponent($bug,'log',$initialdir));
558 if (not defined $stat) {
559 print STDERR "Unable to stat $bug $!\n";
562 if ($options{quick}) {
563 my $rs = $s->resultset('Bug')->search({bug=>$bug})->single();
564 next if defined $rs and $stat->mtime <= $rs->last_modified()->epoch();
567 load_bug_log(db => $s,
571 die "failure while trying to load bug log $bug\n$@";
577 my ($options,$opts,$p,$config,$argv) = @_;
579 my $dist_dir = IO::Dir->new($opts->{ftpdists});
581 grep { $_ !~ /^\./ and
582 -d $opts->{ftpdists}.'/'.$_ and
583 not -l $opts->{ftpdists}.'/'.$_
586 while (my $dist = shift @dist_names) {
587 my $dist_dir = $opts->{ftpdists}.'/'.$dist;
588 my ($dist_info,$package_files) =
589 read_release_file($dist_dir.'/Release');
590 $s_p{$dist_info->{Codename}} = $package_files;
593 for my $suite (keys %s_p) {
594 for my $component (keys %{$s_p{$suite}}) {
595 $tot += scalar keys %{$s_p{$suite}{$component}};
598 $p->target($tot) if $p;
601 my $tot_suites = scalar keys %s_p;
603 my $completed_pkgs=0;
604 # parse packages files
605 for my $suite (keys %s_p) {
607 for my $component (keys %{$s_p{$suite}}) {
608 my @archs = keys %{$s_p{$suite}{$component}};
609 if (grep {$_ eq 'source'} @archs) {
610 @archs = ('source',grep {$_ ne 'source'} @archs);
612 for my $arch (@archs) {
613 my $pfh = open_compressed_file($s_p{$suite}{$component}{$arch}) or
614 die "Unable to open $s_p{$suite}{$component}{$arch} for reading: $!";
616 local $/ = ''; # paragraph mode
619 for my $field (qw(Package Maintainer Version Source)) {
620 /^\Q$field\E: (.*)/m;
623 next unless defined $pkg{Package} and
624 defined $pkg{Version};
625 push @pkgs,[$arch,$component,\%pkg];
629 my $s = db_connect($options);
633 $p->target($avg_pkgs*($tot_suites-$done_suites-1)+
634 $completed_pkgs+@pkgs) if $p;
639 $avg_pkgs=($avg_pkgs*$done_suites + @pkgs)/($done_suites+1);
640 $completed_pkgs += @pkgs;
646 sub handle_subcommand_arguments {
647 my ($argv,$args) = @_;
649 Getopt::Long::GetOptionsFromArray($argv,
654 for my $arg (keys %{$args}) {
655 next unless $args->{$arg};
656 my $r_arg = $arg; # real argument name
657 $r_arg =~ s/[=\|].+//g;
658 if (not defined $subopt->{$r_arg}) {
659 push @usage_errors, "You must give a $r_arg option";
662 pod2usage(join("\n",@usage_errors)) if @usage_errors;
667 my ($subcommand,$config,$options) = @_;
668 if (not lockpid($config->{spool_dir}.'/lock/debbugs-loadsql-$subcommand')) {
669 if ($options->{quick}) {
670 # If this is a quick run, just exit
671 print STDERR "Another debbugs-loadsql is running; stopping\n" if $options->{verbose};
674 print STDERR "Another debbugs-loadsql is running; stopping\n";
681 # connect to the database; figure out how to handle errors
683 my $s = Debbugs::DB->connect($options->{service}) or
684 die "Unable to connect to database: ";
687 sub read_release_file {
690 my $rfh = open_compressed_file($file) or
691 die "Unable to open $file for reading: $!";
697 if (s/^(\S+):\s*//) {
698 if ($1 eq 'SHA1'or $1 eq 'SHA256') {
705 my ($sha,$size,$f) = split /\s+/,$_;
706 next unless $f =~ /(?:Packages|Sources)(?:\.gz|\.xz)$/;
707 next unless $f =~ m{^([^/]+)/([^/]+)/([^/]+)$};
708 my ($component,$arch,$package_source) = ($1,$2,$3);
709 $arch =~ s/binary-//;
710 next if exists $p_f{$component}{$arch};
711 $p_f{$component}{$arch} = File::Spec->catfile(dirname($file),$f);
714 return (\%dist_info,\%p_f);
718 my ($dirs,$p,$what,$verbose,$sub,$n) = @_;
720 my $tot_dirs = @dirs;
722 my $avg_subfiles = 0;
723 my $completed_files = 0;
725 while (my $dir = shift @dirs) {
726 printf "Doing dir %s ...\n", $dir if $verbose;
728 opendir(DIR, "$dir/.") or die "opendir $dir: $!";
729 my @subdirs = readdir(DIR);
732 my @list = map { m/^(\d+)\.$what$/?($1):() } @subdirs;
734 push @dirs, map { m/^(\d+)$/ && -d "$dir/$1"?("$dir/$1"):() } @subdirs;
736 if ($avg_subfiles == 0) {
737 $avg_subfiles = @list;
740 $p->target($avg_subfiles*($tot_dirs-$done_dirs)+$completed_files+@list) if $p;
741 $avg_subfiles = ($avg_subfiles * $done_dirs + @list) / ($done_dirs+1);
744 my $it = natatime $n,@list;
745 while (my @bugs = $it->()) {
747 $completed_files += scalar @bugs;
748 $p->update($completed_files) if $p;
749 print "Up to $completed_files bugs...\n"
750 if ($completed_files % 100 == 0 && $verbose);