]> git.donarmstrong.com Git - debbugs.git/blob - Debbugs/DB/Load.pm
keep done/owner/submitter NULL if there is no submitter/done/owner
[debbugs.git] / Debbugs / DB / Load.pm
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>.
5
6 package Debbugs::DB::Load;
7
8 =head1 NAME
9
10 Debbugs::DB::Load -- Utility routines for loading the database
11
12 =head1 SYNOPSIS
13
14
15 =head1 DESCRIPTION
16
17
18 =head1 BUGS
19
20 None known.
21
22 =cut
23
24 use warnings;
25 use strict;
26 use vars qw($VERSION $DEBUG %EXPORT_TAGS @EXPORT_OK @EXPORT);
27 use base qw(Exporter);
28
29 BEGIN{
30      ($VERSION) = q$Revision$ =~ /^Revision:\s+([^\s+])/;
31      $DEBUG = 0 unless defined $DEBUG;
32
33      @EXPORT = ();
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)],
38                     );
39      @EXPORT_OK = ();
40      Exporter::export_ok_tags(keys %EXPORT_TAGS);
41      $EXPORT_TAGS{all} = [@EXPORT_OK];
42 }
43
44 use Params::Validate qw(validate_with :types);
45 use List::MoreUtils qw(natatime);
46
47 use Debbugs::Status qw(read_bug split_status_fields);
48 use Debbugs::DB;
49 use DateTime;
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;
54 use Carp;
55
56 =head2 Bug loading
57
58 Routines to load bug; exported with :load_bug
59
60 =over
61
62 =item load_bug
63
64      load_bug(db => $schema,
65               data => split_status_fields($data),
66               tags => \%tags,
67               queue => \%queue);
68
69 Loads a bug's metadata into the database. (Does not load any messages)
70
71 =over
72
73 =item db -- Debbugs::DB object
74
75 =item data -- Bug data (from read_bug) which has been split with split_status_fields
76
77 =item tags -- tag cache (hashref); optional
78
79 =item queue -- queue of operations to perform after bug is loaded; optional.
80
81 =back
82
83 =cut
84
85 sub load_bug {
86     my %param = validate_with(params => \@_,
87                               spec => {db => {type => OBJECT,
88                                              },
89                                        data => {type => HASHREF,
90                                                 optional => 1,
91                                                },
92                                        bug => {type => SCALAR,
93                                                optional => 1,
94                                               },
95                                        tags => {type => HASHREF,
96                                                 default => sub {return {}},
97                                                 optional => 1},
98                                        severities => {type => HASHREF,
99                                                       default => sub {return {}},
100                                                       optional => 1,
101                                                      },
102                                        queue => {type => HASHREF,
103                                                  optional => 1},
104                                       });
105     my $s = $param{db};
106     if (not exists $param{data} and not exists $param{bug}) {
107         croak "One of data or bug must be provided to load_bug";
108     }
109     if (not exists $param{data}) {
110         $param{data} = read_bug(bug => $param{bug});
111     }
112     my $data = $param{data};
113     my $tags = $param{tags};
114     my $queue = $param{queue};
115     my $severities = $param{severities};
116     my $can_queue = 1;
117     if (not defined $queue) {
118         $can_queue = 0;
119         $queue = {};
120     }
121     my %tags;
122     my $s_data = split_status_fields($data);
123     for my $tag (make_list($s_data->{keywords})) {
124         next unless defined $tag and length $tag;
125         # this allows for invalid tags. But we'll use this to try to
126         # find those bugs and clean them up
127         if (not exists $tags->{$tag}) {
128             $tags->{$tag} = $s->resultset('Tag')->
129             find_or_create({tag => $tag});
130         }
131         $tags{$tag} = $tags->{$tag};
132     }
133     my $severity = length($data->{severity}) ? $data->{severity} :
134         $config{default_severity};
135     if (not exists $severities->{$severity}) {
136         $severities->{$severity} =
137             $s->resultset('Severity')->
138             find_or_create({severity => $severity},
139                           );
140     }
141     $severity = $severities->{$severity};
142     my $bug =
143         {id => $data->{bug_num},
144          creation => DateTime->from_epoch(epoch => $data->{date}),
145          log_modified => DateTime->from_epoch(epoch => $data->{log_modified}),
146          last_modified => DateTime->from_epoch(epoch => $data->{last_modified}),
147          archived => $data->{archived},
148          (defined $data->{unarchived} and length($data->{unarchived}))?
149          (unarchived => DateTime->from_epoch(epoch => $data->{unarchived})):(),
150          forwarded => $data->{forwarded} // '',
151          summary => $data->{summary} // '',
152          outlook => $data->{outlook} // '',
153          subject => $data->{subject} // '',
154          done_full => $data->{done} // '',
155          severity => $severity,
156          owner_full => $data->{owner} // '',
157          submitter_full => $data->{originator} // '',
158         };
159     my %addr_map =
160         (done => 'done',
161          owner => 'owner',
162          submitter => 'originator',
163         );
164     for my $addr_type (keys %addr_map) {
165         $bug->{$addr_type} = undef;
166         next unless defined $data->{$addr_map{$addr_type}} and
167             length($data->{$addr_map{$addr_type}});
168         $bug->{$addr_type} =
169             $s->resultset('Correspondent')->
170             get_correspondent_id($data->{$addr_map{$addr_type}})
171     }
172     my $b = $s->resultset('Bug')->update_or_create($bug) or
173         die "Unable to update or create bug $bug->{id}";
174      $s->txn_do(sub {
175                    for my $ff (qw(found fixed)) {
176                        my @elements = $s->resultset('BugVer')->search({bug => $data->{bug_num},
177                                                                        found  => $ff eq 'found'?1:0,
178                                                                       });
179                        my %elements_to_delete = map {($elements[$_]->ver_string(),
180                                                       $elements[$_])} 0..$#elements;
181                        my %elements_to_add;
182                        my @elements_to_keep;
183                        for my $version (@{$data->{"${ff}_versions"}}) {
184                            if (exists $elements_to_delete{$version}) {
185                                push @elements_to_keep,$version;
186                            } else {
187                                $elements_to_add{$version} = 1;
188                            }
189                        }
190                        for my $version (@elements_to_keep) {
191                            delete $elements_to_delete{$version};
192                        }
193                        for my $element (keys %elements_to_delete) {
194                            $elements_to_delete{$element}->delete();
195                        }
196                        for my $element (keys %elements_to_add) {
197                            # find source package and source version id
198                            my $ne = $s->resultset('BugVer')->new_result({bug => $data->{bug_num},
199                                                                          ver_string => $element,
200                                                                          found => $ff eq 'found'?1:0,
201                                                                         }
202                                                                        );
203                            if (my ($src_pkg,$src_ver) = $element =~ m{^([^\/]+)/(.+)$}) {
204                                my $src_pkg_e = $s->resultset('SrcPkg')->single({pkg => $src_pkg});
205                                if (defined $src_pkg_e) {
206                                    $ne->src_pkg($src_pkg_e->id());
207                                    my $src_ver_e = $s->resultset('SrcVer')->single({src_pkg => $src_pkg_e->id(),
208                                                                                     ver => $src_ver
209                                                                                    });
210                                    $ne->src_ver($src_ver_e->id()) if defined $src_ver_e;
211                                }
212                            }
213                            $ne->insert();
214                        }
215                    }
216                });
217     ### set bug tags
218     $s->txn_do(sub {$b->set_tags([values %tags ] )});
219     # because these bugs reference other bugs which might not exist
220     # yet, we can't handle them until we've loaded all bugs. queue
221     # them up.
222     for my $merge_block (qw(merged block)) {
223         my $data_key = $merge_block;
224         $data_key .= 'with' if $merge_block eq 'merged';
225         if (@{$data->{$data_key}||[]}) {
226             my $count = $s->resultset('Bug')->search({id => [@{$data->{$data_key}}]})->count();
227             if ($count == @{$data->{$data_key}}) {
228                 handle_load_bug_queue(db=>$s,
229                                       queue => {$merge_block,
230                                                {$data->{bug_num},[@{$data->{$data_key}}]}
231                                                });
232             } else {
233                 $queue->{$merge_block}{$data->{bug_num}} = [@{$data->{$data_key}}];
234             }
235         }
236     }
237
238     if (not $can_queue and keys %{$queue}) {
239         handle_load_bug_queue(db => $s,queue => $queue);
240     }
241
242     # still need to handle merges, versions, etc.
243 }
244
245 =item handle_load_bug_queue
246
247      handle_load_bug_queue(db => $schema,queue => $queue);
248
249 Handles a queue of operations created by load bug. [These operations
250 are used to handle cases where a bug referenced by a loaded bug may
251 not exist yet. In cases where the bugs should exist, the queue is
252 cleared automatically by load_bug if queue is undefined.
253
254 =cut
255
256 sub handle_load_bug_queue{
257     my %param = validate_with(params => \@_,
258                               spec => {db => {type => OBJECT,
259                                              },
260                                        queue => {type => HASHREF,
261                                                 },
262                                       });
263     my $s = $param{db};
264     my $queue = $param{queue};
265     my %queue_types =
266         (merged => {set => 'BugMerged',
267                     columns => [qw(bug merged)],
268                     bug => 'bug',
269                    },
270          blocks => {set => 'BugBlock',
271                     columns => [qw(bug blocks)],
272                     bug => 'bug',
273                    },
274         );
275     for my $queue_type (keys %queue_types) {
276         for my $bug (%{$queue->{$queue_type}}) {
277             my $qt = $queue_types{$queue_type};
278             $s->txn_do(sub {
279                            $s->resultset($qt->{set})->search({$qt->{bug},$bug})->delete();
280                            $s->populate($qt->{set},[[@{$qt->{columns}}],
281                                                     map {[$bug,$_]} @{$queue->{$queue_type}{$bug}}]) if
282                                @{$queue->{$queue_type}{$bug}//[]};
283                        }
284                       );
285         }
286     }
287 }
288
289 =item load_bug_log -- load bug logs
290
291        load_bug_log(db  => $s,
292                     bug => $bug);
293
294
295 =over
296
297 =item db -- database 
298
299 =item bug -- bug whose log should be loaded
300
301 =back
302
303 =cut
304
305 sub load_bug_log {
306     my %param = validate_with(params => \@_,
307                               spec => {db => {type => OBJECT,
308                                              },
309                                        bug => {type => SCALAR,
310                                               },
311                                        queue => {type => HASHREF,
312                                                  optional => 1},
313                                       });
314     my $s = $param{db};
315     my $msg_num=0;
316     my %seen_msg_ids;
317     my $log = Debbugs::Log->new(bug_num => $param{bug}) or
318         die "Unable to open log for $param{bug} for reading: $!";
319     while (my $record = $log->read_record()) {
320         next unless $record->{type} eq 'incoming-recv';
321         my ($msg_id) = $record->{text} =~ /^Message-Id:\s+<(.+)>/im;
322         next if defined $msg_id and exists $seen_msg_ids{$msg_id};
323         $seen_msg_ids{$msg_id} = 1 if defined $msg_id;
324         next if defined $msg_id and $msg_id =~ /handler\..+\.ack(?:info)?\@/;
325         my $entity = parse_to_mime_entity($record);
326         # search for a message with this message id in the database
327         $msg_id = $entity->head->get('Message-Id:');
328         $msg_id =~ s/^\s*\<//;
329         $msg_id =~ s/>\s*$//;
330         # check to see if the subject, to, and from match. if so, it's
331         # probably the same message.
332         my $subject = decode_rfc1522($entity->head->get('Subject:'));
333         $subject =~ s/\n(?:(\s)\s*|\s*$)/$1/g;
334         my $to = decode_rfc1522($entity->head->get('To:'));
335         $to =~ s/\n(?:(\s)\s*|\s*$)/$1/g;
336         my $from = decode_rfc1522($entity->head->get('From:'));
337         $from =~ s/\n(?:(\s)\s*|\s*$)/$1/g;
338         my $m = $s->resultset('Message')->
339             find({msgid => $msg_id,
340                   from_complete => $from,
341                   to_complete => $to,
342                   subject => $subject
343                  });
344         if (not defined $m) {
345             # if not, create a new message
346             $m = $s->resultset('Message')->
347                 find_or_create({msgid => $msg_id,
348                                 from_complete => $from,
349                                 to_complete => $to,
350                                 subject => $subject
351                                });
352             eval {
353                 $m->sent_date(DateTime::Format::Mail->
354                               parse_datetime($entity->head->get('Date:',0)));
355             };
356             my $spam = $entity->head->get('X-Spam-Status:',0);
357             if ($spam=~ /score=([\d\.]+)/) {
358                 $m->spam_score($1);
359             }
360             my %corr;
361             @{$corr{from}} = getparsedaddrs($from);
362             @{$corr{to}} = getparsedaddrs($to);
363             @{$corr{cc}} = getparsedaddrs($entity->head->get('Cc:'));
364             # add correspondents if necessary
365             my @cors;
366             for my $type (keys %corr) {
367                 for my $addr (@{$corr{$type}}) {
368                     push @cors,
369                         {correspondent => $s->resultset('Correspondent')->
370                          get_correspondent_id($addr),
371                          correspondent_type => $type,
372                         };
373                 }
374             }
375             $m->update();
376             $s->txn_do(sub {
377                            $m->message_correspondents()->delete();
378                            $m->add_to_message_correspondents(@cors);
379                        }
380                       );
381         }
382         my $recv;
383         if ($entity->head->get('Received:',0)
384             =~ /via spool by (\S+)/) {
385             $recv = $s->resultset('Correspondent')->
386                 get_correspondent_id($1);
387             $m->add_to_message_correspondents({correspondent=>$recv,
388                                                correspondent_type => 'recv'});
389         }
390         # link message to bugs if necessary
391         $m->find_or_create_related('bug_messages',
392                                   {bug=>$param{bug},
393                                    message_number => $msg_num});
394     }
395
396 }
397
398 =back
399
400 =head2 Debinfo
401
402 Commands to handle src and package version loading from debinfo files
403
404 =over
405
406 =item load_debinfo
407
408      load_debinfo($schema,$binname, $binver, $binarch, $srcname, $srcver);
409
410
411
412 =cut
413
414 sub load_debinfo {
415     my ($schema,$binname, $binver, $binarch, $srcname, $srcver) = @_;
416     my $sp = $schema->resultset('SrcPkg')->find_or_create({pkg => $srcname});
417     my $sv = $schema->resultset('SrcVer')->find_or_create({src_pkg=>$sp->id(),
418                                                            ver => $srcver});
419     my $arch = $schema->resultset('Arch')->find_or_create({arch => $binarch});
420     my $bp = $schema->resultset('BinPkg')->find_or_create({pkg => $binname});
421     $schema->resultset('BinVer')->find_or_create({bin_pkg_id => $bp->id(),
422                                                   src_ver_id => $sv->id(),
423                                                   arch_id    => $arch->id(),
424                                                   ver        => $binver,
425                                                  });
426 }
427
428
429 =back
430
431 =head Packages
432
433 =over
434
435 =item load_package
436
437      load_package($schema,$suite,$component,$arch,$pkg)
438
439 =cut
440
441 sub load_packages {
442     my ($schema,$suite,$pkgs,$p) = @_;
443     my $suite_id = $schema->resultset('Suite')->
444         find_or_create({codename => $suite})->id;
445     my %maint_cache;
446     my %arch_cache;
447     my %source_cache;
448     my $src_max_last_modified = $schema->resultset('SrcAssociation')->
449         search_rs({suite => $suite_id},
450                  {order_by => {-desc => ['me.modified']},
451                   rows => 1,
452                   page => 1
453                  }
454                  )->single();
455     my $bin_max_last_modified = $schema->resultset('BinAssociation')->
456         search_rs({suite => $suite_id},
457                  {order_by => {-desc => ['me.modified']},
458                   rows => 1,
459                   page => 1
460                  }
461                  )->single();
462     my %maints;
463     my %sources;
464     my %bins;
465     for my $pkg_tuple (@{$pkgs}) {
466         my ($arch,$component,$pkg) = @{$pkg_tuple};
467         $maints{$pkg->{Maintainer}} = $pkg->{Maintainer};
468         if ($arch eq 'source') {
469             my $source = $pkg->{Package};
470             my $source_ver = $pkg->{Version};
471             $sources{$source}{$source_ver} = $pkg->{Maintainer};
472         } else {
473             my $source = $pkg->{Source} // $pkg->{Package};
474             my $source_ver = $pkg->{Version};
475             if ($source =~ /^\s*(\S+) \(([^\)]+)\)\s*$/) {
476                 ($source,$source_ver) = ($1,$2);
477             }
478             $sources{$source}{$source_ver} = $pkg->{Maintainer};
479             $bins{$arch}{$pkg->{Package}} =
480                {arch => $arch,
481                 bin => $pkg->{Package},
482                 bin_ver => $pkg->{Version},
483                 src_ver => $source_ver,
484                 source  => $source,
485                 maint   => $pkg->{Maintainer},
486                };
487         }
488     }
489     # Retrieve and Insert new maintainers
490     my $maints =
491         $schema->resultset('Maintainer')->
492         get_maintainers(keys %maints);
493     my $archs =
494         $schema->resultset('Arch')->
495         get_archs(keys %bins);
496     # We want all of the source package/versions which are in this suite to
497     # start with
498     my @sa_to_add;
499     my @sa_to_del;
500     my %included_sa;
501     # Calculate which source packages are no longer in this suite
502     for my $s ($schema->resultset('SrcPkg')->
503                src_pkg_and_ver_in_suite($suite)) {
504         if (not exists $sources{$s->{pkg}} or
505             not exists $sources{$s->{pkg}}{$s->{src_vers}{ver}}
506            ) {
507             push @sa_to_del,
508                 $s->{src_associations}{id};
509         }
510         $included_sa{$s->{pkg}}{$s->{src_vers}} = 1;
511     }
512     # Calculate which source packages are newly in this suite
513     for my $s (keys %sources) {
514         for my $v (keys %{$sources{$s}}) {
515             if (not exists $included_sa{$s} and
516                 not $included_sa{$s}{$v}) {
517                 push @sa_to_add,
518                     [$s,$v,$sources{$s}{$v}];
519             } else {
520                 $p->update() if defined $p;
521             }
522         }
523     }
524     # add new source packages
525     my $it = natatime 100, @sa_to_add;
526     while (my @v = $it->()) {
527         $schema->txn_do(
528             sub {
529                 for my $svm (@_) {
530                     my $s_id = $schema->resultset('SrcPkg')->
531                         get_src_pkg_id($svm->[0]);
532                     my $sv_id = $schema->resultset('SrcVer')->
533                         get_src_ver_id($s_id,$svm->[1],$maints->{$svm->[2]});
534                     $schema->resultset('SrcAssociation')->
535                         insert_suite_src_ver_association($suite_id,$sv_id);
536                 }
537             },
538                         @v
539                        );
540         $p->update($p->last_update()+
541                    scalar @v) if defined $p;
542     }
543     # remove associations for packages not in this suite
544     if (@sa_to_del) {
545         $schema->resultset('SrcAssociation')->
546             search_rs({id => \@sa_to_del})->delete();
547     }
548     # update packages in this suite to have a modification time of now
549     $schema->resultset('SrcAssociation')->
550         search_rs({suite => $suite_id})->
551         update({modified => 'NOW()'});
552     ## Handle binary packages
553     my @bin_to_del;
554     my @bin_to_add;
555     my %included_bin;
556     # calculate which binary packages are no longer in this suite
557     for my $b ($schema->resultset('BinPkg')->
558                bin_pkg_and_ver_in_suite($suite)) {
559         if (not exists $bins{$b->{arch}{arch}} or
560             not exists $bins{$b->{arch}{arch}}{$b->{pkg}} or
561             ($bins{$b->{arch}{arch}}{$b->{pkg}}{bin_ver} ne
562              $b->{bin_vers}{ver}
563             )
564            ) {
565             push @bin_to_del,
566                 $b->{bin_associations}{id};
567         }
568         $included_bin{$b->{arch}{arch}}{$b->{pkg}} =
569             $b->{bin_vers}{ver};
570     }
571     # calculate which binary packages are newly in this suite
572     for my $a (keys %bins) {
573         for my $pkg (keys %{$bins{$a}}) {
574             if (not exists $included_bin{$a} or
575                 not exists $included_bin{$a}{$pkg} or
576                 $bins{$a}{$pkg}{bin_ver} ne
577                 $included_bin{$a}{$pkg}) {
578                 push @bin_to_add,
579                     $bins{$a}{$pkg};
580             } else {
581                 $p->update() if defined $p;
582             }
583         }
584     }
585     $it = natatime 100, @bin_to_add;
586     while (my @v = $it->()) {
587         $schema->txn_do(
588         sub {
589             for my $bvm (@_) {
590                 my $s_id = $schema->resultset('SrcPkg')->
591                     get_src_pkg_id($bvm->{source});
592                 my $sv_id = $schema->resultset('SrcVer')->
593                     get_src_ver_id($s_id,$bvm->{src_ver},$maints->{$bvm->{maint}});
594                 my $b_id = $schema->resultset('BinPkg')->
595                     get_bin_pkg_id($bvm->{bin});
596                 my $bv_id = $schema->resultset('BinVer')->
597                     get_bin_ver_id($b_id,$bvm->{bin_ver},
598                                    $archs->{$bvm->{arch}},$sv_id);
599                 $schema->resultset('BinAssociation')->
600                     insert_suite_bin_ver_association($suite_id,$bv_id);
601             }
602         },
603                         @v
604                        );
605         $p->update($p->last_update()+
606                    scalar @v) if defined $p;
607     }
608     if (@bin_to_del) {
609         $schema->resultset('BinAssociation')->
610             search_rs({id => \@bin_to_del})->delete();
611     }
612     $schema->resultset('BinAssociation')->
613         search_rs({suite => $suite_id})->
614         update({modified => 'NOW()'});
615
616 }
617
618
619 =back
620
621 =cut
622
623 =head Suites
624
625 =over
626
627 =item load_suite
628
629      load_suite($schema,$codename,$suite,$version,$active);
630
631 =cut
632
633 sub load_suite {
634     my ($schema,$codename,$suite,$version,$active) = @_;
635     if (ref($codename)) {
636         ($codename,$suite,$version) =
637             @{$codename}{qw(Codename Suite Version)};
638         $active = 1;
639     }
640     my $s = $schema->resultset('Suite')->find_or_create({codename => $codename});
641     $s->suite_name($suite);
642     $s->version($version);
643     $s->active($active);
644     $s->update();
645     return $s;
646
647 }
648
649 =back
650
651 =cut
652
653 1;
654
655
656 __END__