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
27 use vars qw($VERSION $DEBUG %EXPORT_TAGS @EXPORT_OK @EXPORT);
28 use base qw(Exporter);
31 ($VERSION) = q$Revision$ =~ /^Revision:\s+([^\s+])/;
32 $DEBUG = 0 unless defined $DEBUG;
35 %EXPORT_TAGS = (load_bug => [qw(load_bug handle_load_bug_queue load_bug_log)],
36 load_debinfo => [qw(load_debinfo)],
37 load_package => [qw(load_packages)],
38 load_suite => [qw(load_suite)],
41 Exporter::export_ok_tags(keys %EXPORT_TAGS);
42 $EXPORT_TAGS{all} = [@EXPORT_OK];
45 use Params::Validate qw(validate_with :types);
46 use List::AllUtils qw(natatime);
48 use Debbugs::Status qw(read_bug split_status_fields);
51 use Debbugs::Common qw(make_list getparsedaddrs);
52 use Debbugs::Config qw(:config);
53 use Debbugs::MIME qw(parse_to_mime_entity decode_rfc1522);
54 use DateTime::Format::Mail;
59 Routines to load bug; exported with :load_bug
65 load_bug(db => $schema,
66 data => split_status_fields($data),
70 Loads a bug's metadata into the database. (Does not load any messages)
74 =item db -- Debbugs::DB object
76 =item data -- Bug data (from read_bug) which has been split with split_status_fields
78 =item tags -- tag cache (hashref); optional
80 =item queue -- queue of operations to perform after bug is loaded; optional.
87 my %param = validate_with(params => \@_,
88 spec => {db => {type => OBJECT,
90 data => {type => HASHREF,
93 bug => {type => SCALAR,
96 tags => {type => HASHREF,
97 default => sub {return {}},
99 severities => {type => HASHREF,
100 default => sub {return {}},
103 queue => {type => HASHREF,
105 packages => {type => HASHREF,
106 default => sub {return {}},
111 if (not exists $param{data} and not exists $param{bug}) {
112 croak "One of data or bug must be provided to load_bug";
114 if (not exists $param{data}) {
115 $param{data} = read_bug(bug => $param{bug});
117 my $data = $param{data};
118 my $tags = $param{tags};
119 my $queue = $param{queue};
120 my $severities = $param{severities};
122 if (not defined $queue) {
127 $data = split_status_fields($data);
128 for my $tag (make_list($data->{keywords})) {
129 next unless defined $tag and length $tag;
130 # this allows for invalid tags. But we'll use this to try to
131 # find those bugs and clean them up
132 if (not exists $tags->{$tag}) {
133 $tags->{$tag} = $s->resultset('Tag')->
134 find_or_create({tag => $tag});
136 $tags{$tag} = $tags->{$tag};
138 my $severity = length($data->{severity}) ? $data->{severity} :
139 $config{default_severity};
140 if (not exists $severities->{$severity}) {
141 $severities->{$severity} =
142 $s->resultset('Severity')->
143 find_or_create({severity => $severity},
146 $severity = $severities->{$severity};
148 {id => $data->{bug_num},
149 creation => DateTime->from_epoch(epoch => $data->{date}),
150 log_modified => DateTime->from_epoch(epoch => $data->{log_modified}),
151 last_modified => DateTime->from_epoch(epoch => $data->{last_modified}),
152 archived => $data->{archived},
153 (defined $data->{unarchived} and length($data->{unarchived}))?
154 (unarchived => DateTime->from_epoch(epoch => $data->{unarchived})):(),
155 forwarded => $data->{forwarded} // '',
156 summary => $data->{summary} // '',
157 outlook => $data->{outlook} // '',
158 subject => $data->{subject} // '',
159 done_full => $data->{done} // '',
160 severity => $severity,
161 owner_full => $data->{owner} // '',
162 submitter_full => $data->{originator} // '',
167 submitter => 'originator',
169 for my $addr_type (keys %addr_map) {
170 $bug->{$addr_type} = undef;
171 next unless defined $data->{$addr_map{$addr_type}} and
172 length($data->{$addr_map{$addr_type}});
174 $s->resultset('Correspondent')->
175 get_correspondent_id($data->{$addr_map{$addr_type}})
177 my $b = $s->resultset('Bug')->update_or_create($bug) or
178 die "Unable to update or create bug $bug->{id}";
180 $b->set_related_packages('binpackages',
181 [grep {defined $_ and
182 length $_ and $_ !~ /^src:/}
183 make_list($data->{package})],
186 $b->set_related_packages('srcpackages',
191 make_list($data->{package})],
194 $b->set_related_packages('affects_binpackages',
195 [grep {defined $_ and
196 length $_ and $_ !~ /^src:/}
197 make_list($data->{affects})
201 $b->set_related_packages('affects_srcpackages',
206 make_list($data->{affects})],
209 for my $ff (qw(found fixed)) {
210 my @elements = $s->resultset('BugVer')->search({bug => $data->{bug_num},
211 found => $ff eq 'found'?1:0,
213 my %elements_to_delete = map {($elements[$_]->ver_string(),
214 $elements[$_])} 0..$#elements;
216 my @elements_to_keep;
217 for my $version (@{$data->{"${ff}_versions"}}) {
218 if (exists $elements_to_delete{$version}) {
219 push @elements_to_keep,$version;
221 $elements_to_add{$version} = 1;
224 for my $version (@elements_to_keep) {
225 delete $elements_to_delete{$version};
227 for my $element (keys %elements_to_delete) {
228 $elements_to_delete{$element}->delete();
230 for my $element (keys %elements_to_add) {
231 # find source package and source version id
232 my $ne = $s->resultset('BugVer')->new_result({bug => $data->{bug_num},
233 ver_string => $element,
234 found => $ff eq 'found'?1:0,
237 if (my ($src_pkg,$src_ver) = $element =~ m{^([^\/]+)/(.+)$}) {
238 my $src_pkg_e = $s->resultset('SrcPkg')->single({pkg => $src_pkg});
239 if (defined $src_pkg_e) {
240 $ne->src_pkg($src_pkg_e->id());
241 my $src_ver_e = $s->resultset('SrcVer')->single({src_pkg => $src_pkg_e->id(),
244 $ne->src_ver($src_ver_e->id()) if defined $src_ver_e;
252 $s->txn_do(sub {$b->set_tags([values %tags ] )});
253 # because these bugs reference other bugs which might not exist
254 # yet, we can't handle them until we've loaded all bugs. queue
256 for my $merge_block (qw(mergedwith blocks)) {
257 if (@{$data->{$merge_block}||[]}) {
258 my $count = $s->resultset('Bug')->
259 search({id => [@{$data->{$merge_block}}]})->
261 # if all of the bugs exist, immediately fix the merge/blocks
262 if ($count == @{$data->{$merge_block}}) {
263 handle_load_bug_queue(db=>$s,
264 queue => {$merge_block,
265 {$data->{bug_num},[@{$data->{$merge_block}}]}
268 $queue->{$merge_block}{$data->{bug_num}} = [@{$data->{$merge_block}}];
273 if (not $can_queue and keys %{$queue}) {
274 handle_load_bug_queue(db => $s,queue => $queue);
277 # still need to handle merges, versions, etc.
280 =item handle_load_bug_queue
282 handle_load_bug_queue(db => $schema,queue => $queue);
284 Handles a queue of operations created by load bug. [These operations
285 are used to handle cases where a bug referenced by a loaded bug may
286 not exist yet. In cases where the bugs should exist, the queue is
287 cleared automatically by load_bug if queue is undefined.
291 sub handle_load_bug_queue{
292 my %param = validate_with(params => \@_,
293 spec => {db => {type => OBJECT,
295 queue => {type => HASHREF,
299 my $queue = $param{queue};
301 (mergedwith => {set => 'BugMerged',
302 columns => [qw(bug merged)],
305 blocks => {set => 'BugBlock',
306 columns => [qw(bug blocks)],
310 for my $queue_type (keys %queue_types) {
311 my $qt = $queue_types{$queue_type};
312 my @bugs = keys %{$queue->{$queue_type}};
314 for my $bug (@bugs) {
317 @{$queue->{$queue_type}{$bug}};
320 $s->resultset($qt->{set})->
321 search({$qt->{bug}=>\@bugs})->delete();
322 $s->resultset($qt->{set})->
323 populate([[@{$qt->{columns}}],
324 @entries]) if @entries;
330 =item load_bug_log -- load bug logs
332 load_bug_log(db => $s,
340 =item bug -- bug whose log should be loaded
347 my %param = validate_with(params => \@_,
348 spec => {db => {type => OBJECT,
350 bug => {type => SCALAR,
352 queue => {type => HASHREF,
358 my $log = Debbugs::Log->new(bug_num => $param{bug}) or
359 die "Unable to open log for $param{bug} for reading: $!";
360 while (my $record = $log->read_record()) {
361 next unless $record->{type} eq 'incoming-recv';
362 my ($msg_id) = $record->{text} =~ /^Message-Id:\s+<(.+)>/im;
363 next if defined $msg_id and exists $seen_msg_ids{$msg_id};
364 $seen_msg_ids{$msg_id} = 1 if defined $msg_id;
365 next if defined $msg_id and $msg_id =~ /handler\..+\.ack(?:info)?\@/;
366 my $entity = parse_to_mime_entity($record);
367 # search for a message with this message id in the database
368 $msg_id = $entity->head->get('Message-Id') //
369 $entity->head->get('Resent-Message-ID') //
371 $msg_id =~ s/^\s*\<//;
372 $msg_id =~ s/>\s*$//;
373 # check to see if the subject, to, and from match. if so, it's
374 # probably the same message.
375 my $subject = decode_rfc1522($entity->head->get('Subject')//'');
376 $subject =~ s/\n(?:(\s)\s*|\s*$)//g;
377 my $to = decode_rfc1522($entity->head->get('To')//'');
378 $to =~ s/\n(?:(\s)\s*|\s*$)//g;
379 my $from = decode_rfc1522($entity->head->get('From')//'');
380 $from =~ s/\n(?:(\s)\s*|\s*$)//g;
381 my $m = $s->resultset('Message')->
382 find({msgid => $msg_id,
383 from_complete => $from,
387 if (not defined $m) {
388 # if not, create a new message
389 $m = $s->resultset('Message')->
390 find_or_create({msgid => $msg_id,
391 from_complete => $from,
396 my $date = DateTime::Format::Mail->
397 parse_datetime($entity->head->get('Date',0));
398 if (abs($date->offset) >= 60 * 60 * 12) {
399 $date = $date->set_time_zone('UTC');
401 $m->sent_date($date);
403 my $spam = $entity->head->get('X-Spam-Status',0)//'';
404 if ($spam=~ /score=([\d\.]+)/) {
408 @{$corr{from}} = getparsedaddrs($from);
409 @{$corr{to}} = getparsedaddrs($to);
410 @{$corr{cc}} = getparsedaddrs($entity->head->get('Cc'));
411 # add correspondents if necessary
413 for my $type (keys %corr) {
414 for my $addr (@{$corr{$type}}) {
415 my $cor = $s->resultset('Correspondent')->
416 get_correspondent_id($addr);
417 next unless defined $cor;
419 {correspondent => $cor,
420 correspondent_type => $type,
426 $m->message_correspondents()->delete();
427 $m->add_to_message_correspondents(@cors) if
433 if ($entity->head->get('Received',0)
434 =~ /via spool by (\S+)/) {
435 $recv = $s->resultset('Correspondent')->
436 get_correspondent_id($1);
437 $m->add_to_message_correspondents({correspondent=>$recv,
438 correspondent_type => 'recv'});
440 # link message to bugs if necessary
441 $m->find_or_create_related('bug_messages',
443 message_number => $msg_num});
452 Commands to handle src and package version loading from debinfo files
458 load_debinfo($schema,$binname, $binver, $binarch, $srcname, $srcver);
465 my ($s,$binname, $binver, $binarch, $srcname, $srcver,$ct_date,$cache) = @_;
468 if (not defined $cache->{sp}{$srcname}) {
469 $cache->{sp}{$srcname} =
470 $s->resultset('SrcPkg')->find_or_create({pkg => $srcname});
472 $sp = $cache->{sp}{$srcname};
473 # update the creation date if the data we have is earlier
474 if (defined $ct_date and
475 (not defined $sp->creation or
476 $ct_date < $sp->creation)) {
477 $sp->creation($ct_date);
478 $sp->last_modified(DateTime->now);
482 if (not defined $cache->{sv}{$srcname}{$srcver}) {
483 $cache->{sv}{$srcname}{$srcver} =
484 $s->resultset('SrcVer')->
485 find_or_create({src_pkg =>$sp->id(),
488 $sv = $cache->{sv}{$srcname}{$srcver};
489 if (defined $ct_date and
490 (not defined $sv->upload_date() or $ct_date < $sv->upload_date())) {
491 $sv->upload_date($ct_date);
495 if (not defined $cache->{arch}{$binarch}) {
496 $cache->{arch}{$binarch} =
497 $s->resultset('Arch')->
498 find_or_create({arch => $binarch},
501 $arch = $cache->{arch}{$binarch};
503 if (not defined $cache->{bp}{$binname}) {
504 $cache->{bp}{$binname} =
505 $s->resultset('BinPkg')->
506 get_bin_pkg_id($binname);
508 $bp = $cache->{bp}{$binname};
509 $s->resultset('BinVer')->
510 get_bin_ver_id($bp,$binver,$arch,$sv->id());
522 load_package($schema,$suite,$component,$arch,$pkg)
527 my ($schema,$suite,$pkgs,$p) = @_;
528 my $suite_id = $schema->resultset('Suite')->
529 find_or_create({codename => $suite})->id;
533 my $src_max_last_modified = $schema->resultset('SrcAssociation')->
534 search_rs({suite => $suite_id},
535 {order_by => {-desc => ['me.modified']},
540 my $bin_max_last_modified = $schema->resultset('BinAssociation')->
541 search_rs({suite => $suite_id},
542 {order_by => {-desc => ['me.modified']},
550 for my $pkg_tuple (@{$pkgs}) {
551 my ($arch,$component,$pkg) = @{$pkg_tuple};
552 $maints{$pkg->{Maintainer}} = $pkg->{Maintainer};
553 if ($arch eq 'source') {
554 my $source = $pkg->{Package};
555 my $source_ver = $pkg->{Version};
556 $sources{$source}{$source_ver} = $pkg->{Maintainer};
558 my $source = $pkg->{Source} // $pkg->{Package};
559 my $source_ver = $pkg->{Version};
560 if ($source =~ /^\s*(\S+) \(([^\)]+)\)\s*$/) {
561 ($source,$source_ver) = ($1,$2);
563 $sources{$source}{$source_ver} = $pkg->{Maintainer};
564 $bins{$arch}{$pkg->{Package}} =
566 bin => $pkg->{Package},
567 bin_ver => $pkg->{Version},
568 src_ver => $source_ver,
570 maint => $pkg->{Maintainer},
574 # Retrieve and Insert new maintainers
576 $schema->resultset('Maintainer')->
577 get_maintainers(keys %maints);
579 $schema->resultset('Arch')->
580 get_archs(keys %bins);
581 # We want all of the source package/versions which are in this suite to
586 # Calculate which source packages are no longer in this suite
587 for my $s ($schema->resultset('SrcPkg')->
588 src_pkg_and_ver_in_suite($suite)) {
589 if (not exists $sources{$s->{pkg}} or
590 not exists $sources{$s->{pkg}}{$s->{src_vers}{ver}}
593 $s->{src_associations}{id};
595 $included_sa{$s->{pkg}}{$s->{src_vers}} = 1;
597 # Calculate which source packages are newly in this suite
598 for my $s (keys %sources) {
599 for my $v (keys %{$sources{$s}}) {
600 if (not exists $included_sa{$s} and
601 not $included_sa{$s}{$v}) {
603 [$s,$v,$sources{$s}{$v}];
605 $p->update() if defined $p;
609 # add new source packages
610 my $it = natatime 100, @sa_to_add;
611 while (my @v = $it->()) {
615 my $s_id = $schema->resultset('SrcPkg')->
616 get_src_pkg_id($svm->[0]);
617 my $sv_id = $schema->resultset('SrcVer')->
618 get_src_ver_id($s_id,$svm->[1],$maints->{$svm->[2]});
619 $schema->resultset('SrcAssociation')->
620 insert_suite_src_ver_association($suite_id,$sv_id);
625 $p->update($p->last_update()+
626 scalar @v) if defined $p;
628 # remove associations for packages not in this suite
630 $it = natatime 1000, @sa_to_del;
631 while (my @v = $it->()) {
634 $schema->resultset('SrcAssociation')->
635 search_rs({id => \@v})->
640 # update packages in this suite to have a modification time of now
641 $schema->resultset('SrcAssociation')->
642 search_rs({suite => $suite_id})->
643 update({modified => 'NOW()'});
644 ## Handle binary packages
648 # calculate which binary packages are no longer in this suite
649 for my $b ($schema->resultset('BinPkg')->
650 bin_pkg_and_ver_in_suite($suite)) {
651 if (not exists $bins{$b->{arch}{arch}} or
652 not exists $bins{$b->{arch}{arch}}{$b->{pkg}} or
653 ($bins{$b->{arch}{arch}}{$b->{pkg}}{bin_ver} ne
658 $b->{bin_associations}{id};
660 $included_bin{$b->{arch}{arch}}{$b->{pkg}} =
663 # calculate which binary packages are newly in this suite
664 for my $a (keys %bins) {
665 for my $pkg (keys %{$bins{$a}}) {
666 if (not exists $included_bin{$a} or
667 not exists $included_bin{$a}{$pkg} or
668 $bins{$a}{$pkg}{bin_ver} ne
669 $included_bin{$a}{$pkg}) {
673 $p->update() if defined $p;
677 $it = natatime 100, @bin_to_add;
678 while (my @v = $it->()) {
682 my $s_id = $schema->resultset('SrcPkg')->
683 get_src_pkg_id($bvm->{source});
684 my $sv_id = $schema->resultset('SrcVer')->
685 get_src_ver_id($s_id,$bvm->{src_ver},$maints->{$bvm->{maint}});
686 my $b_id = $schema->resultset('BinPkg')->
687 get_bin_pkg_id($bvm->{bin});
688 my $bv_id = $schema->resultset('BinVer')->
689 get_bin_ver_id($b_id,$bvm->{bin_ver},
690 $archs->{$bvm->{arch}},$sv_id);
691 $schema->resultset('BinAssociation')->
692 insert_suite_bin_ver_association($suite_id,$bv_id);
697 $p->update($p->last_update()+
698 scalar @v) if defined $p;
701 $it = natatime 1000, @bin_to_del;
702 while (my @v = $it->()) {
705 $schema->resultset('BinAssociation')->
706 search_rs({id => \@v})->
711 $schema->resultset('BinAssociation')->
712 search_rs({suite => $suite_id})->
713 update({modified => 'NOW()'});
728 load_suite($schema,$codename,$suite,$version,$active);
733 my ($schema,$codename,$suite,$version,$active) = @_;
734 if (ref($codename)) {
735 ($codename,$suite,$version) =
736 @{$codename}{qw(Codename Suite Version)};
739 my $s = $schema->resultset('Suite')->find_or_create({codename => $codename});
740 $s->suite_name($suite);
741 $s->version($version);
757 # indent-tabs-mode: nil
758 # cperl-indent-level: 4