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')->find_or_create({pkg => $versions[$i][0]});
356 $sp = $src_pkgs{$versions[$i][0]};
357 # There's probably something wrong if the source package
358 # doesn't exist, but we'll skip it for now
359 next unless defined $sp;
360 my $sv = $s->resultset('SrcVer')->find({src_pkg=>$sp->id(),
361 ver => $versions[$i][1],
363 if (defined $ancestor_sv and defined $sv and not defined $sv->based_on()) {
364 $sv->update({based_on => $ancestor_sv->id()})
374 my ($options,$opts,$p,$config,$argv) = @_;
376 my @files = @{$argv};
380 local $/ = "\0" if $opts->{0};
382 s/\n$// unless $opts->{0};
383 s/\0$// if $opts->{0};
388 return unless @files;
389 my $s = db_connect($options);
390 $p->target(scalar @files) if $p;
391 my $it = natatime 100, @files;
392 while (my @v = $it->()) {
396 my $fh = IO::File->new($file,'r') or
397 die "Unable to open $file for reading: $!";
398 my $f_stat = stat($file);
399 my $ct_date = DateTime->from_epoch(epoch => $f_stat->ctime);
402 next unless length $_;
403 my ($binname, $binver, $binarch, $srcname, $srcver) = split;
404 # if $srcver is not defined, this is probably a broken
405 # .debinfo file [they were causing #686106, see commit
406 # 49c85ab8 in dak.] Basically, $binarch didn't get put into
407 # the file, so we'll fudge it from the filename.
408 if (not defined $srcver) {
409 ($srcname,$srcver) = ($binarch,$srcname);
410 ($binarch) = $file =~ /_([^\.]+)\.debinfo/;
413 [$binname,$binver,$binarch,$srcname,$srcver,$ct_date];
418 for my $di (@debinfos) {
419 Debbugs::DB::Load::load_debinfo($s,@{$di}[0..5],\%cache);
422 $p->update($p->last_update()+@v) if $p;
427 sub add_maintainers {
428 my ($options,$opts,$p,$config,$argv) = @_;
430 my $s = db_connect($options);
431 my $maintainers = getsourcemaintainers();
433 ## get all of the maintainers, and add the missing ones
434 my $maints = $s->resultset('Maintainer')->
435 get_maintainers(values %{$maintainers});
437 my @svs = $s->resultset('SrcVer')->
438 search({maintainer => undef
441 group_by => 'me.src_pkg, src_pkg.pkg',
442 result_class => 'DBIx::Class::ResultClass::HashRefInflator',
443 columns => [qw(me.src_pkg src_pkg.pkg)],
446 $p->target(2+@svs) if $p;
449 if (exists $maintainers->{$sv->{src_pkg}{pkg}}) {
450 my $pkg = $sv->{src_pkg}{pkg};
451 my $maint = $maints->
452 {$maintainers->{$pkg}};
453 $s->txn_do(sub {$s->resultset('SrcVer')->
454 search({maintainer => undef,
455 'src_pkg.pkg' => $pkg
458 )->update({maintainer => $maint})
466 sub add_configuration {
467 my ($options,$opts,$p,$config,$argv) = @_;
469 my $s = db_connect($options);
474 for my $tag (@{$config{tags}}) {
476 $s->resultset('Tag')->find_or_create({tag => $tag});
479 for my $tag ($s->resultset('Tag')->search_rs()->all()) {
480 next if exists $tags{$tag->tag};
488 for my $sev_name (($config{default_severity},@{$config{severity_list}})) {
489 # add all severitites
490 my $sev = $s->resultset('Severity')->find_or_create({severity => $sev_name});
491 # mark strong severities
492 if (grep {$_ eq $sev_name} @{$config{strong_severities}}) {
495 $sev->ordering($order);
498 $sev_names{$sev_name} = 1;
500 # mark obsolete severities
501 for my $sev ($s->resultset('Severity')->search_rs()->all()) {
502 next if exists $sev_names{$sev->severity()};
509 my ($options,$opts,$p,$config,$argv) = @_;
512 my $s = db_connect($options);
513 my $dist_dir = IO::Dir->new($opts->{ftpdists});
515 grep { $_ !~ /^\./ and
516 -d $opts->{ftpdists}.'/'.$_ and
517 not -l $opts->{ftpdists}.'/'.$_
519 while (my $dist = shift @dist_names) {
520 my $dist_dir = $opts->{ftpdists}.'/'.$dist;
521 my ($dist_info,$package_files) =
522 read_release_file($dist_dir.'/Release');
523 load_suite($s,$dist_info);
528 my ($options,$opts,$p,$config,$argv) = @_;
530 chdir($config->{spool_dir}) or
531 die "chdir $config->{spool_dir} failed: $!";
533 my $verbose = $options->{debug};
535 my $initialdir = "db-h";
537 if (defined $argv->[0] and $argv->[0] eq "archive") {
538 $initialdir = "archive";
540 my $s = db_connect($options);
544 my $start_time = time;
546 walk_bugs([(@{$argv}?@{$argv} : $initialdir)],
552 my $stat = stat(getbugcomponent($bug,'log',$initialdir));
553 if (not defined $stat) {
554 print STDERR "Unable to stat $bug $!\n";
557 if ($options{quick}) {
558 my $rs = $s->resultset('Bug')->search({bug=>$bug})->single();
559 next if defined $rs and $stat->mtime <= $rs->last_modified()->epoch();
562 load_bug_log(db => $s,
566 die "failure while trying to load bug log $bug\n$@";
572 my ($options,$opts,$p,$config,$argv) = @_;
574 my $dist_dir = IO::Dir->new($opts->{ftpdists});
576 grep { $_ !~ /^\./ and
577 -d $opts->{ftpdists}.'/'.$_ and
578 not -l $opts->{ftpdists}.'/'.$_
581 while (my $dist = shift @dist_names) {
582 my $dist_dir = $opts->{ftpdists}.'/'.$dist;
583 my ($dist_info,$package_files) =
584 read_release_file($dist_dir.'/Release');
585 $s_p{$dist_info->{Codename}} = $package_files;
588 for my $suite (keys %s_p) {
589 for my $component (keys %{$s_p{$suite}}) {
590 $tot += scalar keys %{$s_p{$suite}{$component}};
593 $p->target($tot) if $p;
596 my $tot_suites = scalar keys %s_p;
598 my $completed_pkgs=0;
599 # parse packages files
600 for my $suite (keys %s_p) {
602 for my $component (keys %{$s_p{$suite}}) {
603 my @archs = keys %{$s_p{$suite}{$component}};
604 if (grep {$_ eq 'source'} @archs) {
605 @archs = ('source',grep {$_ ne 'source'} @archs);
607 for my $arch (@archs) {
608 my $pfh = open_compressed_file($s_p{$suite}{$component}{$arch}) or
609 die "Unable to open $s_p{$suite}{$component}{$arch} for reading: $!";
611 local $/ = ''; # paragraph mode
614 for my $field (qw(Package Maintainer Version Source)) {
615 /^\Q$field\E: (.*)/m;
618 next unless defined $pkg{Package} and
619 defined $pkg{Version};
620 push @pkgs,[$arch,$component,\%pkg];
624 my $s = db_connect($options);
628 $p->target($avg_pkgs*($tot_suites-$done_suites-1)+
629 $completed_pkgs+@pkgs) if $p;
634 $avg_pkgs=($avg_pkgs*$done_suites + @pkgs)/($done_suites+1);
635 $completed_pkgs += @pkgs;
641 sub handle_subcommand_arguments {
642 my ($argv,$args) = @_;
644 Getopt::Long::GetOptionsFromArray($argv,
649 for my $arg (keys %{$args}) {
650 next unless $args->{$arg};
651 my $r_arg = $arg; # real argument name
652 $r_arg =~ s/[=\|].+//g;
653 if (not defined $subopt->{$r_arg}) {
654 push @usage_errors, "You must give a $r_arg option";
657 pod2usage(join("\n",@usage_errors)) if @usage_errors;
662 my ($subcommand,$config,$options) = @_;
663 if (not lockpid($config->{spool_dir}.'/lock/debbugs-loadsql-$subcommand')) {
664 if ($options->{quick}) {
665 # If this is a quick run, just exit
666 print STDERR "Another debbugs-loadsql is running; stopping\n" if $options->{verbose};
669 print STDERR "Another debbugs-loadsql is running; stopping\n";
676 # connect to the database; figure out how to handle errors
678 my $s = Debbugs::DB->connect($options->{service}) or
679 die "Unable to connect to database: ";
682 sub read_release_file {
685 my $rfh = open_compressed_file($file) or
686 die "Unable to open $file for reading: $!";
692 if (s/^(\S+):\s*//) {
693 if ($1 eq 'SHA1'or $1 eq 'SHA256') {
700 my ($sha,$size,$f) = split /\s+/,$_;
701 next unless $f =~ /(?:Packages|Sources)(?:\.gz|\.xz)$/;
702 next unless $f =~ m{^([^/]+)/([^/]+)/([^/]+)$};
703 my ($component,$arch,$package_source) = ($1,$2,$3);
704 $arch =~ s/binary-//;
705 next if exists $p_f{$component}{$arch};
706 $p_f{$component}{$arch} = File::Spec->catfile(dirname($file),$f);
709 return (\%dist_info,\%p_f);
713 my ($dirs,$p,$what,$verbose,$sub,$n) = @_;
715 my $tot_dirs = @dirs;
717 my $avg_subfiles = 0;
718 my $completed_files = 0;
720 while (my $dir = shift @dirs) {
721 printf "Doing dir %s ...\n", $dir if $verbose;
723 opendir(DIR, "$dir/.") or die "opendir $dir: $!";
724 my @subdirs = readdir(DIR);
727 my @list = map { m/^(\d+)\.$what$/?($1):() } @subdirs;
729 push @dirs, map { m/^(\d+)$/ && -d "$dir/$1"?("$dir/$1"):() } @subdirs;
731 if ($avg_subfiles == 0) {
732 $avg_subfiles = @list;
735 $p->target($avg_subfiles*($tot_dirs-$done_dirs)+$completed_files+@list) if $p;
736 $avg_subfiles = ($avg_subfiles * $done_dirs + @list) / ($done_dirs+1);
739 my $it = natatime $n,@list;
740 while (my @bugs = $it->()) {
742 $completed_files += scalar @bugs;
743 $p->update($completed_files) if $p;
744 print "Up to $completed_files bugs...\n"
745 if ($completed_files % 100 == 0 && $verbose);