1 # This module is part of debbugs, and is released
2 # under the terms of the GPL version 2, or any later version. See the
3 # file README and COPYING for more information.
4 # Copyright 2013 by Don Armstrong <don@donarmstrong.com>.
6 package Debbugs::DB::Load;
10 Debbugs::DB::Load -- Utility routines for loading the database
26 use vars qw($VERSION $DEBUG %EXPORT_TAGS @EXPORT_OK @EXPORT);
27 use base qw(Exporter);
30 ($VERSION) = q$Revision$ =~ /^Revision:\s+([^\s+])/;
31 $DEBUG = 0 unless defined $DEBUG;
34 %EXPORT_TAGS = (load_bug => [qw(load_bug handle_load_bug_queue load_bug_log)],
35 load_debinfo => [qw(load_debinfo)],
36 load_package => [qw(load_packages)],
37 load_suite => [qw(load_suite)],
40 Exporter::export_ok_tags(keys %EXPORT_TAGS);
41 $EXPORT_TAGS{all} = [@EXPORT_OK];
44 use Params::Validate qw(validate_with :types);
45 use List::AllUtils qw(natatime);
47 use Debbugs::Status qw(read_bug split_status_fields);
50 use Debbugs::Common qw(make_list getparsedaddrs);
51 use Debbugs::Config qw(:config);
52 use Debbugs::MIME qw(parse_to_mime_entity decode_rfc1522);
53 use DateTime::Format::Mail;
58 Routines to load bug; exported with :load_bug
64 load_bug(db => $schema,
65 data => split_status_fields($data),
69 Loads a bug's metadata into the database. (Does not load any messages)
73 =item db -- Debbugs::DB object
75 =item data -- Bug data (from read_bug) which has been split with split_status_fields
77 =item tags -- tag cache (hashref); optional
79 =item queue -- queue of operations to perform after bug is loaded; optional.
86 my %param = validate_with(params => \@_,
87 spec => {db => {type => OBJECT,
89 data => {type => HASHREF,
92 bug => {type => SCALAR,
95 tags => {type => HASHREF,
96 default => sub {return {}},
98 severities => {type => HASHREF,
99 default => sub {return {}},
102 queue => {type => HASHREF,
104 packages => {type => HASHREF,
105 default => sub {return {}},
110 if (not exists $param{data} and not exists $param{bug}) {
111 croak "One of data or bug must be provided to load_bug";
113 if (not exists $param{data}) {
114 $param{data} = read_bug(bug => $param{bug});
116 my $data = $param{data};
117 my $tags = $param{tags};
118 my $queue = $param{queue};
119 my $severities = $param{severities};
121 if (not defined $queue) {
126 $data = split_status_fields($data);
127 for my $tag (make_list($data->{keywords})) {
128 next unless defined $tag and length $tag;
129 # this allows for invalid tags. But we'll use this to try to
130 # find those bugs and clean them up
131 if (not exists $tags->{$tag}) {
132 $tags->{$tag} = $s->resultset('Tag')->
133 find_or_create({tag => $tag});
135 $tags{$tag} = $tags->{$tag};
137 my $severity = length($data->{severity}) ? $data->{severity} :
138 $config{default_severity};
139 if (not exists $severities->{$severity}) {
140 $severities->{$severity} =
141 $s->resultset('Severity')->
142 find_or_create({severity => $severity},
145 $severity = $severities->{$severity};
147 {id => $data->{bug_num},
148 creation => DateTime->from_epoch(epoch => $data->{date}),
149 log_modified => DateTime->from_epoch(epoch => $data->{log_modified}),
150 last_modified => DateTime->from_epoch(epoch => $data->{last_modified}),
151 archived => $data->{archived},
152 (defined $data->{unarchived} and length($data->{unarchived}))?
153 (unarchived => DateTime->from_epoch(epoch => $data->{unarchived})):(),
154 forwarded => $data->{forwarded} // '',
155 summary => $data->{summary} // '',
156 outlook => $data->{outlook} // '',
157 subject => $data->{subject} // '',
158 done_full => $data->{done} // '',
159 severity => $severity,
160 owner_full => $data->{owner} // '',
161 submitter_full => $data->{originator} // '',
166 submitter => 'originator',
168 for my $addr_type (keys %addr_map) {
169 $bug->{$addr_type} = undef;
170 next unless defined $data->{$addr_map{$addr_type}} and
171 length($data->{$addr_map{$addr_type}});
173 $s->resultset('Correspondent')->
174 get_correspondent_id($data->{$addr_map{$addr_type}})
176 my $b = $s->resultset('Bug')->update_or_create($bug) or
177 die "Unable to update or create bug $bug->{id}";
179 $b->set_related_packages('binpackages',
180 [grep {defined $_ and
181 length $_ and $_ !~ /^src:/}
182 make_list($data->{package})],
185 $b->set_related_packages('srcpackages',
190 make_list($data->{package})],
193 $b->set_related_packages('affects_binpackages',
194 [grep {defined $_ and
195 length $_ and $_ !~ /^src:/}
196 make_list($data->{affects})
200 $b->set_related_packages('affects_srcpackages',
205 make_list($data->{affects})],
208 for my $ff (qw(found fixed)) {
209 my @elements = $s->resultset('BugVer')->search({bug => $data->{bug_num},
210 found => $ff eq 'found'?1:0,
212 my %elements_to_delete = map {($elements[$_]->ver_string(),
213 $elements[$_])} 0..$#elements;
215 my @elements_to_keep;
216 for my $version (@{$data->{"${ff}_versions"}}) {
217 if (exists $elements_to_delete{$version}) {
218 push @elements_to_keep,$version;
220 $elements_to_add{$version} = 1;
223 for my $version (@elements_to_keep) {
224 delete $elements_to_delete{$version};
226 for my $element (keys %elements_to_delete) {
227 $elements_to_delete{$element}->delete();
229 for my $element (keys %elements_to_add) {
230 # find source package and source version id
231 my $ne = $s->resultset('BugVer')->new_result({bug => $data->{bug_num},
232 ver_string => $element,
233 found => $ff eq 'found'?1:0,
236 if (my ($src_pkg,$src_ver) = $element =~ m{^([^\/]+)/(.+)$}) {
237 my $src_pkg_e = $s->resultset('SrcPkg')->single({pkg => $src_pkg});
238 if (defined $src_pkg_e) {
239 $ne->src_pkg($src_pkg_e->id());
240 my $src_ver_e = $s->resultset('SrcVer')->single({src_pkg => $src_pkg_e->id(),
243 $ne->src_ver($src_ver_e->id()) if defined $src_ver_e;
251 $s->txn_do(sub {$b->set_tags([values %tags ] )});
252 # because these bugs reference other bugs which might not exist
253 # yet, we can't handle them until we've loaded all bugs. queue
255 for my $merge_block (qw(merged block)) {
256 my $data_key = $merge_block;
257 $data_key .= 'with' if $merge_block eq 'merged';
258 if (@{$data->{$data_key}||[]}) {
259 my $count = $s->resultset('Bug')->search({id => [@{$data->{$data_key}}]})->count();
260 if ($count == @{$data->{$data_key}}) {
261 handle_load_bug_queue(db=>$s,
262 queue => {$merge_block,
263 {$data->{bug_num},[@{$data->{$data_key}}]}
266 $queue->{$merge_block}{$data->{bug_num}} = [@{$data->{$data_key}}];
271 if (not $can_queue and keys %{$queue}) {
272 handle_load_bug_queue(db => $s,queue => $queue);
275 # still need to handle merges, versions, etc.
278 =item handle_load_bug_queue
280 handle_load_bug_queue(db => $schema,queue => $queue);
282 Handles a queue of operations created by load bug. [These operations
283 are used to handle cases where a bug referenced by a loaded bug may
284 not exist yet. In cases where the bugs should exist, the queue is
285 cleared automatically by load_bug if queue is undefined.
289 sub handle_load_bug_queue{
290 my %param = validate_with(params => \@_,
291 spec => {db => {type => OBJECT,
293 queue => {type => HASHREF,
297 my $queue = $param{queue};
299 (merged => {set => 'BugMerged',
300 columns => [qw(bug merged)],
303 blocks => {set => 'BugBlock',
304 columns => [qw(bug blocks)],
308 for my $queue_type (keys %queue_types) {
309 for my $bug (%{$queue->{$queue_type}}) {
310 my $qt = $queue_types{$queue_type};
312 $s->resultset($qt->{set})->search({$qt->{bug},$bug})->delete();
313 $s->populate($qt->{set},[[@{$qt->{columns}}],
314 map {[$bug,$_]} @{$queue->{$queue_type}{$bug}}]) if
315 @{$queue->{$queue_type}{$bug}//[]};
322 =item load_bug_log -- load bug logs
324 load_bug_log(db => $s,
332 =item bug -- bug whose log should be loaded
339 my %param = validate_with(params => \@_,
340 spec => {db => {type => OBJECT,
342 bug => {type => SCALAR,
344 queue => {type => HASHREF,
350 my $log = Debbugs::Log->new(bug_num => $param{bug}) or
351 die "Unable to open log for $param{bug} for reading: $!";
352 while (my $record = $log->read_record()) {
353 next unless $record->{type} eq 'incoming-recv';
354 my ($msg_id) = $record->{text} =~ /^Message-Id:\s+<(.+)>/im;
355 next if defined $msg_id and exists $seen_msg_ids{$msg_id};
356 $seen_msg_ids{$msg_id} = 1 if defined $msg_id;
357 next if defined $msg_id and $msg_id =~ /handler\..+\.ack(?:info)?\@/;
358 my $entity = parse_to_mime_entity($record);
359 # search for a message with this message id in the database
360 $msg_id = $entity->head->get('Message-Id') //
361 $entity->head->get('Resent-Message-ID') //
363 $msg_id =~ s/^\s*\<//;
364 $msg_id =~ s/>\s*$//;
365 # check to see if the subject, to, and from match. if so, it's
366 # probably the same message.
367 my $subject = decode_rfc1522($entity->head->get('Subject')//'');
368 $subject =~ s/\n(?:(\s)\s*|\s*$)//g;
369 my $to = decode_rfc1522($entity->head->get('To')//'');
370 $to =~ s/\n(?:(\s)\s*|\s*$)//g;
371 my $from = decode_rfc1522($entity->head->get('From')//'');
372 $from =~ s/\n(?:(\s)\s*|\s*$)//g;
373 my $m = $s->resultset('Message')->
374 find({msgid => $msg_id,
375 from_complete => $from,
379 if (not defined $m) {
380 # if not, create a new message
381 $m = $s->resultset('Message')->
382 find_or_create({msgid => $msg_id,
383 from_complete => $from,
388 my $date = DateTime::Format::Mail->
389 parse_datetime($entity->head->get('Date',0));
390 if (abs($date->offset) >= 60 * 60 * 12) {
391 $date = $date->set_time_zone('UTC');
393 $m->sent_date($date);
395 my $spam = $entity->head->get('X-Spam-Status',0)//'';
396 if ($spam=~ /score=([\d\.]+)/) {
400 @{$corr{from}} = getparsedaddrs($from);
401 @{$corr{to}} = getparsedaddrs($to);
402 @{$corr{cc}} = getparsedaddrs($entity->head->get('Cc'));
403 # add correspondents if necessary
405 for my $type (keys %corr) {
406 for my $addr (@{$corr{$type}}) {
407 my $cor = $s->resultset('Correspondent')->
408 get_correspondent_id($addr);
409 next unless defined $cor;
411 {correspondent => $cor,
412 correspondent_type => $type,
418 $m->message_correspondents()->delete();
419 $m->add_to_message_correspondents(@cors) if
425 if ($entity->head->get('Received',0)
426 =~ /via spool by (\S+)/) {
427 $recv = $s->resultset('Correspondent')->
428 get_correspondent_id($1);
429 $m->add_to_message_correspondents({correspondent=>$recv,
430 correspondent_type => 'recv'});
432 # link message to bugs if necessary
433 $m->find_or_create_related('bug_messages',
435 message_number => $msg_num});
444 Commands to handle src and package version loading from debinfo files
450 load_debinfo($schema,$binname, $binver, $binarch, $srcname, $srcver);
457 my ($s,$binname, $binver, $binarch, $srcname, $srcver,$ct_date,$cache) = @_;
460 if (not defined $cache->{sp}{$srcname}) {
461 $cache->{sp}{$srcname} =
462 $s->resultset('SrcPkg')->find_or_create({pkg => $srcname});
464 $sp = $cache->{sp}{$srcname};
465 # update the creation date if the data we have is earlier
466 if (defined $ct_date and
467 (not defined $sp->creation or
468 $ct_date < $sp->creation)) {
469 $sp->creation($ct_date);
470 $sp->last_modified(DateTime->now);
474 if (not defined $cache->{sv}{$srcname}{$srcver}) {
475 $cache->{sv}{$srcname}{$srcver} =
476 $s->resultset('SrcVer')->
477 find_or_create({src_pkg =>$sp->id(),
480 $sv = $cache->{sv}{$srcname}{$srcver};
481 if (defined $ct_date and
482 (not defined $sv->upload_date() or $ct_date < $sv->upload_date())) {
483 $sv->upload_date($ct_date);
487 if (not defined $cache->{arch}{$binarch}) {
488 $cache->{arch}{$binarch} =
489 $s->resultset('Arch')->
490 find_or_create({arch => $binarch},
493 $arch = $cache->{arch}{$binarch};
495 if (not defined $cache->{bp}{$binname}) {
496 $cache->{bp}{$binname} =
497 $s->resultset('BinPkg')->
498 get_bin_pkg_id($binname);
500 $bp = $cache->{bp}{$binname};
501 $s->resultset('BinVer')->
502 get_bin_ver_id($bp,$binver,$arch,$sv->id());
514 load_package($schema,$suite,$component,$arch,$pkg)
519 my ($schema,$suite,$pkgs,$p) = @_;
520 my $suite_id = $schema->resultset('Suite')->
521 find_or_create({codename => $suite})->id;
525 my $src_max_last_modified = $schema->resultset('SrcAssociation')->
526 search_rs({suite => $suite_id},
527 {order_by => {-desc => ['me.modified']},
532 my $bin_max_last_modified = $schema->resultset('BinAssociation')->
533 search_rs({suite => $suite_id},
534 {order_by => {-desc => ['me.modified']},
542 for my $pkg_tuple (@{$pkgs}) {
543 my ($arch,$component,$pkg) = @{$pkg_tuple};
544 $maints{$pkg->{Maintainer}} = $pkg->{Maintainer};
545 if ($arch eq 'source') {
546 my $source = $pkg->{Package};
547 my $source_ver = $pkg->{Version};
548 $sources{$source}{$source_ver} = $pkg->{Maintainer};
550 my $source = $pkg->{Source} // $pkg->{Package};
551 my $source_ver = $pkg->{Version};
552 if ($source =~ /^\s*(\S+) \(([^\)]+)\)\s*$/) {
553 ($source,$source_ver) = ($1,$2);
555 $sources{$source}{$source_ver} = $pkg->{Maintainer};
556 $bins{$arch}{$pkg->{Package}} =
558 bin => $pkg->{Package},
559 bin_ver => $pkg->{Version},
560 src_ver => $source_ver,
562 maint => $pkg->{Maintainer},
566 # Retrieve and Insert new maintainers
568 $schema->resultset('Maintainer')->
569 get_maintainers(keys %maints);
571 $schema->resultset('Arch')->
572 get_archs(keys %bins);
573 # We want all of the source package/versions which are in this suite to
578 # Calculate which source packages are no longer in this suite
579 for my $s ($schema->resultset('SrcPkg')->
580 src_pkg_and_ver_in_suite($suite)) {
581 if (not exists $sources{$s->{pkg}} or
582 not exists $sources{$s->{pkg}}{$s->{src_vers}{ver}}
585 $s->{src_associations}{id};
587 $included_sa{$s->{pkg}}{$s->{src_vers}} = 1;
589 # Calculate which source packages are newly in this suite
590 for my $s (keys %sources) {
591 for my $v (keys %{$sources{$s}}) {
592 if (not exists $included_sa{$s} and
593 not $included_sa{$s}{$v}) {
595 [$s,$v,$sources{$s}{$v}];
597 $p->update() if defined $p;
601 # add new source packages
602 my $it = natatime 100, @sa_to_add;
603 while (my @v = $it->()) {
607 my $s_id = $schema->resultset('SrcPkg')->
608 get_src_pkg_id($svm->[0]);
609 my $sv_id = $schema->resultset('SrcVer')->
610 get_src_ver_id($s_id,$svm->[1],$maints->{$svm->[2]});
611 $schema->resultset('SrcAssociation')->
612 insert_suite_src_ver_association($suite_id,$sv_id);
617 $p->update($p->last_update()+
618 scalar @v) if defined $p;
620 # remove associations for packages not in this suite
622 $it = natatime 1000, @sa_to_del;
623 while (my @v = $it->()) {
626 $schema->resultset('SrcAssociation')->
627 search_rs({id => \@v})->
632 # update packages in this suite to have a modification time of now
633 $schema->resultset('SrcAssociation')->
634 search_rs({suite => $suite_id})->
635 update({modified => 'NOW()'});
636 ## Handle binary packages
640 # calculate which binary packages are no longer in this suite
641 for my $b ($schema->resultset('BinPkg')->
642 bin_pkg_and_ver_in_suite($suite)) {
643 if (not exists $bins{$b->{arch}{arch}} or
644 not exists $bins{$b->{arch}{arch}}{$b->{pkg}} or
645 ($bins{$b->{arch}{arch}}{$b->{pkg}}{bin_ver} ne
650 $b->{bin_associations}{id};
652 $included_bin{$b->{arch}{arch}}{$b->{pkg}} =
655 # calculate which binary packages are newly in this suite
656 for my $a (keys %bins) {
657 for my $pkg (keys %{$bins{$a}}) {
658 if (not exists $included_bin{$a} or
659 not exists $included_bin{$a}{$pkg} or
660 $bins{$a}{$pkg}{bin_ver} ne
661 $included_bin{$a}{$pkg}) {
665 $p->update() if defined $p;
669 $it = natatime 100, @bin_to_add;
670 while (my @v = $it->()) {
674 my $s_id = $schema->resultset('SrcPkg')->
675 get_src_pkg_id($bvm->{source});
676 my $sv_id = $schema->resultset('SrcVer')->
677 get_src_ver_id($s_id,$bvm->{src_ver},$maints->{$bvm->{maint}});
678 my $b_id = $schema->resultset('BinPkg')->
679 get_bin_pkg_id($bvm->{bin});
680 my $bv_id = $schema->resultset('BinVer')->
681 get_bin_ver_id($b_id,$bvm->{bin_ver},
682 $archs->{$bvm->{arch}},$sv_id);
683 $schema->resultset('BinAssociation')->
684 insert_suite_bin_ver_association($suite_id,$bv_id);
689 $p->update($p->last_update()+
690 scalar @v) if defined $p;
693 $it = natatime 1000, @bin_to_del;
694 while (my @v = $it->()) {
697 $schema->resultset('BinAssociation')->
698 search_rs({id => \@v})->
703 $schema->resultset('BinAssociation')->
704 search_rs({suite => $suite_id})->
705 update({modified => 'NOW()'});
720 load_suite($schema,$codename,$suite,$version,$active);
725 my ($schema,$codename,$suite,$version,$active) = @_;
726 if (ref($codename)) {
727 ($codename,$suite,$version) =
728 @{$codename}{qw(Codename Suite Version)};
731 my $s = $schema->resultset('Suite')->find_or_create({codename => $codename});
732 $s->suite_name($suite);
733 $s->version($version);
749 # indent-tabs-mode: nil
750 # cperl-indent-level: 4