]> git.donarmstrong.com Git - debbugs.git/blob - Debbugs/DB/Load.pm
the correspondent id was being set improperly
[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} =
166             $s->resultset('Correspondent')->
167             get_correspondent_id($data->{$addr_map{$addr_type}})
168     }
169     my $b = $s->resultset('Bug')->update_or_create($bug) or
170         die "Unable to update or create bug $bug->{id}";
171      $s->txn_do(sub {
172                    for my $ff (qw(found fixed)) {
173                        my @elements = $s->resultset('BugVer')->search({bug => $data->{bug_num},
174                                                                        found  => $ff eq 'found'?1:0,
175                                                                       });
176                        my %elements_to_delete = map {($elements[$_]->ver_string(),
177                                                       $elements[$_])} 0..$#elements;
178                        my %elements_to_add;
179                        my @elements_to_keep;
180                        for my $version (@{$data->{"${ff}_versions"}}) {
181                            if (exists $elements_to_delete{$version}) {
182                                push @elements_to_keep,$version;
183                            } else {
184                                $elements_to_add{$version} = 1;
185                            }
186                        }
187                        for my $version (@elements_to_keep) {
188                            delete $elements_to_delete{$version};
189                        }
190                        for my $element (keys %elements_to_delete) {
191                            $elements_to_delete{$element}->delete();
192                        }
193                        for my $element (keys %elements_to_add) {
194                            # find source package and source version id
195                            my $ne = $s->resultset('BugVer')->new_result({bug => $data->{bug_num},
196                                                                          ver_string => $element,
197                                                                          found => $ff eq 'found'?1:0,
198                                                                         }
199                                                                        );
200                            if (my ($src_pkg,$src_ver) = $element =~ m{^([^\/]+)/(.+)$}) {
201                                my $src_pkg_e = $s->resultset('SrcPkg')->single({pkg => $src_pkg});
202                                if (defined $src_pkg_e) {
203                                    $ne->src_pkg($src_pkg_e->id());
204                                    my $src_ver_e = $s->resultset('SrcVer')->single({src_pkg => $src_pkg_e->id(),
205                                                                                     ver => $src_ver
206                                                                                    });
207                                    $ne->src_ver($src_ver_e->id()) if defined $src_ver_e;
208                                }
209                            }
210                            $ne->insert();
211                        }
212                    }
213                });
214     ### set bug tags
215     $s->txn_do(sub {$b->set_tags([values %tags ] )});
216     # because these bugs reference other bugs which might not exist
217     # yet, we can't handle them until we've loaded all bugs. queue
218     # them up.
219     for my $merge_block (qw(merged block)) {
220         my $data_key = $merge_block;
221         $data_key .= 'with' if $merge_block eq 'merged';
222         if (@{$data->{$data_key}||[]}) {
223             my $count = $s->resultset('Bug')->search({id => [@{$data->{$data_key}}]})->count();
224             if ($count == @{$data->{$data_key}}) {
225                 handle_load_bug_queue(db=>$s,
226                                       queue => {$merge_block,
227                                                {$data->{bug_num},[@{$data->{$data_key}}]}
228                                                });
229             } else {
230                 $queue->{$merge_block}{$data->{bug_num}} = [@{$data->{$data_key}}];
231             }
232         }
233     }
234
235     if (not $can_queue and keys %{$queue}) {
236         handle_load_bug_queue(db => $s,queue => $queue);
237     }
238
239     # still need to handle merges, versions, etc.
240 }
241
242 =item handle_load_bug_queue
243
244      handle_load_bug_queue(db => $schema,queue => $queue);
245
246 Handles a queue of operations created by load bug. [These operations
247 are used to handle cases where a bug referenced by a loaded bug may
248 not exist yet. In cases where the bugs should exist, the queue is
249 cleared automatically by load_bug if queue is undefined.
250
251 =cut
252
253 sub handle_load_bug_queue{
254     my %param = validate_with(params => \@_,
255                               spec => {db => {type => OBJECT,
256                                              },
257                                        queue => {type => HASHREF,
258                                                 },
259                                       });
260     my $s = $param{db};
261     my $queue = $param{queue};
262     my %queue_types =
263         (merged => {set => 'BugMerged',
264                     columns => [qw(bug merged)],
265                     bug => 'bug',
266                    },
267          blocks => {set => 'BugBlock',
268                     columns => [qw(bug blocks)],
269                     bug => 'bug',
270                    },
271         );
272     for my $queue_type (keys %queue_types) {
273         for my $bug (%{$queue->{$queue_type}}) {
274             my $qt = $queue_types{$queue_type};
275             $s->txn_do(sub {
276                            $s->resultset($qt->{set})->search({$qt->{bug},$bug})->delete();
277                            $s->populate($qt->{set},[[@{$qt->{columns}}],
278                                                     map {[$bug,$_]} @{$queue->{$queue_type}{$bug}}]) if
279                                @{$queue->{$queue_type}{$bug}//[]};
280                        }
281                       );
282         }
283     }
284 }
285
286 =item load_bug_log -- load bug logs
287
288        load_bug_log(db  => $s,
289                     bug => $bug);
290
291
292 =over
293
294 =item db -- database 
295
296 =item bug -- bug whose log should be loaded
297
298 =back
299
300 =cut
301
302 sub load_bug_log {
303     my %param = validate_with(params => \@_,
304                               spec => {db => {type => OBJECT,
305                                              },
306                                        bug => {type => SCALAR,
307                                               },
308                                        queue => {type => HASHREF,
309                                                  optional => 1},
310                                       });
311     my $s = $param{db};
312     my $msg_num=0;
313     my %seen_msg_ids;
314     my $log = Debbugs::Log->new(bug_num => $param{bug}) or
315         die "Unable to open log for $param{bug} for reading: $!";
316     while (my $record = $log->read_record()) {
317         next unless $record->{type} eq 'incoming-recv';
318         my ($msg_id) = $record->{text} =~ /^Message-Id:\s+<(.+)>/im;
319         next if defined $msg_id and exists $seen_msg_ids{$msg_id};
320         $seen_msg_ids{$msg_id} = 1 if defined $msg_id;
321         next if defined $msg_id and $msg_id =~ /handler\..+\.ack(?:info)?\@/;
322         my $entity = parse_to_mime_entity($record);
323         # search for a message with this message id in the database
324         $msg_id = $entity->head->get('Message-Id:');
325         $msg_id =~ s/^\s*\<//;
326         $msg_id =~ s/>\s*$//;
327         # check to see if the subject, to, and from match. if so, it's
328         # probably the same message.
329         my $subject = decode_rfc1522($entity->head->get('Subject:'));
330         $subject =~ s/\n(?:(\s)\s*|\s*$)/$1/g;
331         my $to = decode_rfc1522($entity->head->get('To:'));
332         $to =~ s/\n(?:(\s)\s*|\s*$)/$1/g;
333         my $from = decode_rfc1522($entity->head->get('From:'));
334         $from =~ s/\n(?:(\s)\s*|\s*$)/$1/g;
335         my $m = $s->resultset('Message')->
336             find({msgid => $msg_id,
337                   from_complete => $from,
338                   to_complete => $to,
339                   subject => $subject
340                  });
341         if (not defined $m) {
342             # if not, create a new message
343             $m = $s->resultset('Message')->
344                 find_or_create({msgid => $msg_id,
345                                 from_complete => $from,
346                                 to_complete => $to,
347                                 subject => $subject
348                                });
349             eval {
350                 $m->sent_date(DateTime::Format::Mail->
351                               parse_datetime($entity->head->get('Date:',0)));
352             };
353             my $spam = $entity->head->get('X-Spam-Status:',0);
354             if ($spam=~ /score=([\d\.]+)/) {
355                 $m->spam_score($1);
356             }
357             my %corr;
358             @{$corr{from}} = getparsedaddrs($from);
359             @{$corr{to}} = getparsedaddrs($to);
360             @{$corr{cc}} = getparsedaddrs($entity->head->get('Cc:'));
361             # add correspondents if necessary
362             my @cors;
363             for my $type (keys %corr) {
364                 for my $addr (@{$corr{$type}}) {
365                     push @cors,
366                         {correspondent => $s->resultset('Correspondent')->
367                          get_correspondent_id($addr),
368                          correspondent_type => $type,
369                         };
370                 }
371             }
372             $m->update();
373             $s->txn_do(sub {
374                            $m->message_correspondents()->delete();
375                            $m->add_to_message_correspondents(@cors);
376                        }
377                       );
378         }
379         my $recv;
380         if ($entity->head->get('Received:',0)
381             =~ /via spool by (\S+)/) {
382             $recv = $s->resultset('Correspondent')->
383                 get_correspondent_id($1);
384             $m->add_to_message_correspondents({correspondent=>$recv,
385                                                correspondent_type => 'recv'});
386         }
387         # link message to bugs if necessary
388         $m->find_or_create_related('bug_messages',
389                                   {bug=>$param{bug},
390                                    message_number => $msg_num});
391     }
392
393 }
394
395 =back
396
397 =head2 Debinfo
398
399 Commands to handle src and package version loading from debinfo files
400
401 =over
402
403 =item load_debinfo
404
405      load_debinfo($schema,$binname, $binver, $binarch, $srcname, $srcver);
406
407
408
409 =cut
410
411 sub load_debinfo {
412     my ($schema,$binname, $binver, $binarch, $srcname, $srcver) = @_;
413     my $sp = $schema->resultset('SrcPkg')->find_or_create({pkg => $srcname});
414     my $sv = $schema->resultset('SrcVer')->find_or_create({src_pkg=>$sp->id(),
415                                                            ver => $srcver});
416     my $arch = $schema->resultset('Arch')->find_or_create({arch => $binarch});
417     my $bp = $schema->resultset('BinPkg')->find_or_create({pkg => $binname});
418     $schema->resultset('BinVer')->find_or_create({bin_pkg_id => $bp->id(),
419                                                   src_ver_id => $sv->id(),
420                                                   arch_id    => $arch->id(),
421                                                   ver        => $binver,
422                                                  });
423 }
424
425
426 =back
427
428 =head Packages
429
430 =over
431
432 =item load_package
433
434      load_package($schema,$suite,$component,$arch,$pkg)
435
436 =cut
437
438 sub load_packages {
439     my ($schema,$suite,$pkgs,$p) = @_;
440     my $suite_id = $schema->resultset('Suite')->
441         find_or_create({codename => $suite})->id;
442     my %maint_cache;
443     my %arch_cache;
444     my %source_cache;
445     my $src_max_last_modified = $schema->resultset('SrcAssociation')->
446         search_rs({suite => $suite_id},
447                  {order_by => {-desc => ['me.modified']},
448                   rows => 1,
449                   page => 1
450                  }
451                  )->single();
452     my $bin_max_last_modified = $schema->resultset('BinAssociation')->
453         search_rs({suite => $suite_id},
454                  {order_by => {-desc => ['me.modified']},
455                   rows => 1,
456                   page => 1
457                  }
458                  )->single();
459     my %maints;
460     my %sources;
461     my %bins;
462     for my $pkg_tuple (@{$pkgs}) {
463         my ($arch,$component,$pkg) = @{$pkg_tuple};
464         $maints{$pkg->{Maintainer}} = $pkg->{Maintainer};
465         if ($arch eq 'source') {
466             my $source = $pkg->{Package};
467             my $source_ver = $pkg->{Version};
468             $sources{$source}{$source_ver} = $pkg->{Maintainer};
469         } else {
470             my $source = $pkg->{Source} // $pkg->{Package};
471             my $source_ver = $pkg->{Version};
472             if ($source =~ /^\s*(\S+) \(([^\)]+)\)\s*$/) {
473                 ($source,$source_ver) = ($1,$2);
474             }
475             $sources{$source}{$source_ver} = $pkg->{Maintainer};
476             $bins{$arch}{$pkg->{Package}} =
477                {arch => $arch,
478                 bin => $pkg->{Package},
479                 bin_ver => $pkg->{Version},
480                 src_ver => $source_ver,
481                 source  => $source,
482                 maint   => $pkg->{Maintainer},
483                };
484         }
485     }
486     # Retrieve and Insert new maintainers
487     my $maints =
488         $schema->resultset('Maintainer')->
489         get_maintainers(keys %maints);
490     my $archs =
491         $schema->resultset('Arch')->
492         get_archs(keys %bins);
493     # We want all of the source package/versions which are in this suite to
494     # start with
495     my @sa_to_add;
496     my @sa_to_del;
497     my %included_sa;
498     # Calculate which source packages are no longer in this suite
499     for my $s ($schema->resultset('SrcPkg')->
500                src_pkg_and_ver_in_suite($suite)) {
501         if (not exists $sources{$s->{pkg}} or
502             not exists $sources{$s->{pkg}}{$s->{src_vers}{ver}}
503            ) {
504             push @sa_to_del,
505                 $s->{src_associations}{id};
506         }
507         $included_sa{$s->{pkg}}{$s->{src_vers}} = 1;
508     }
509     # Calculate which source packages are newly in this suite
510     for my $s (keys %sources) {
511         for my $v (keys %{$sources{$s}}) {
512             if (not exists $included_sa{$s} and
513                 not $included_sa{$s}{$v}) {
514                 push @sa_to_add,
515                     [$s,$v,$sources{$s}{$v}];
516             } else {
517                 $p->update() if defined $p;
518             }
519         }
520     }
521     # add new source packages
522     my $it = natatime 100, @sa_to_add;
523     while (my @v = $it->()) {
524         $schema->txn_do(
525             sub {
526                 for my $svm (@_) {
527                     my $s_id = $schema->resultset('SrcPkg')->
528                         get_src_pkg_id($svm->[0]);
529                     my $sv_id = $schema->resultset('SrcVer')->
530                         get_src_ver_id($s_id,$svm->[1],$maints->{$svm->[2]});
531                     $schema->resultset('SrcAssociation')->
532                         insert_suite_src_ver_association($suite_id,$sv_id);
533                 }
534             },
535                         @v
536                        );
537         $p->update($p->last_update()+
538                    scalar @v) if defined $p;
539     }
540     # remove associations for packages not in this suite
541     if (@sa_to_del) {
542         $schema->resultset('SrcAssociation')->
543             search_rs({id => \@sa_to_del})->delete();
544     }
545     # update packages in this suite to have a modification time of now
546     $schema->resultset('SrcAssociation')->
547         search_rs({suite => $suite_id})->
548         update({modified => 'NOW()'});
549     ## Handle binary packages
550     my @bin_to_del;
551     my @bin_to_add;
552     my %included_bin;
553     # calculate which binary packages are no longer in this suite
554     for my $b ($schema->resultset('BinPkg')->
555                bin_pkg_and_ver_in_suite($suite)) {
556         if (not exists $bins{$b->{arch}{arch}} or
557             not exists $bins{$b->{arch}{arch}}{$b->{pkg}} or
558             ($bins{$b->{arch}{arch}}{$b->{pkg}}{bin_ver} ne
559              $b->{bin_vers}{ver}
560             )
561            ) {
562             push @bin_to_del,
563                 $b->{bin_associations}{id};
564         }
565         $included_bin{$b->{arch}{arch}}{$b->{pkg}} =
566             $b->{bin_vers}{ver};
567     }
568     # calculate which binary packages are newly in this suite
569     for my $a (keys %bins) {
570         for my $pkg (keys %{$bins{$a}}) {
571             if (not exists $included_bin{$a} or
572                 not exists $included_bin{$a}{$pkg} or
573                 $bins{$a}{$pkg}{bin_ver} ne
574                 $included_bin{$a}{$pkg}) {
575                 push @bin_to_add,
576                     $bins{$a}{$pkg};
577             } else {
578                 $p->update() if defined $p;
579             }
580         }
581     }
582     $it = natatime 100, @bin_to_add;
583     while (my @v = $it->()) {
584         $schema->txn_do(
585         sub {
586             for my $bvm (@_) {
587                 my $s_id = $schema->resultset('SrcPkg')->
588                     get_src_pkg_id($bvm->{source});
589                 my $sv_id = $schema->resultset('SrcVer')->
590                     get_src_ver_id($s_id,$bvm->{src_ver},$maints->{$bvm->{maint}});
591                 my $b_id = $schema->resultset('BinPkg')->
592                     get_bin_pkg_id($bvm->{bin});
593                 my $bv_id = $schema->resultset('BinVer')->
594                     get_bin_ver_id($b_id,$bvm->{bin_ver},
595                                    $archs->{$bvm->{arch}},$sv_id);
596                 $schema->resultset('BinAssociation')->
597                     insert_suite_bin_ver_association($suite_id,$bv_id);
598             }
599         },
600                         @v
601                        );
602         $p->update($p->last_update()+
603                    scalar @v) if defined $p;
604     }
605     if (@bin_to_del) {
606         $schema->resultset('BinAssociation')->
607             search_rs({id => \@bin_to_del})->delete();
608     }
609     $schema->resultset('BinAssociation')->
610         search_rs({suite => $suite_id})->
611         update({modified => 'NOW()'});
612
613 }
614
615
616 =back
617
618 =cut
619
620 =head Suites
621
622 =over
623
624 =item load_suite
625
626      load_suite($schema,$codename,$suite,$version,$active);
627
628 =cut
629
630 sub load_suite {
631     my ($schema,$codename,$suite,$version,$active) = @_;
632     if (ref($codename)) {
633         ($codename,$suite,$version) =
634             @{$codename}{qw(Codename Suite Version)};
635         $active = 1;
636     }
637     my $s = $schema->resultset('Suite')->find_or_create({codename => $codename});
638     $s->suite_name($suite);
639     $s->version($version);
640     $s->active($active);
641     $s->update();
642     return $s;
643
644 }
645
646 =back
647
648 =cut
649
650 1;
651
652
653 __END__