]> git.donarmstrong.com Git - debbugs.git/blob - Debbugs/DB/Load.pm
load_packages now only changes rows it has to
[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 Carp;
53
54 =head2 Bug loading
55
56 Routines to load bug; exported with :load_bug
57
58 =over
59
60 =item load_bug
61
62      load_bug(db => $schema,
63               data => split_status_fields($data),
64               tags => \%tags,
65               queue => \%queue);
66
67 Loads a bug's metadata into the database. (Does not load any messages)
68
69 =over
70
71 =item db -- Debbugs::DB object
72
73 =item data -- Bug data (from read_bug) which has been split with split_status_fields
74
75 =item tags -- tag cache (hashref); optional
76
77 =item queue -- queue of operations to perform after bug is loaded; optional.
78
79 =back
80
81 =cut
82
83 sub load_bug {
84     my %param = validate_with(params => \@_,
85                               spec => {db => {type => OBJECT,
86                                              },
87                                        data => {type => HASHREF,
88                                                 optional => 1,
89                                                },
90                                        bug => {type => SCALAR,
91                                                optional => 1,
92                                               },
93                                        tags => {type => HASHREF,
94                                                 default => sub {return {}},
95                                                 optional => 1},
96                                        severities => {type => HASHREF,
97                                                       default => sub {return {}},
98                                                       optional => 1,
99                                                      },
100                                        queue => {type => HASHREF,
101                                                  optional => 1},
102                                       });
103     my $s = $param{db};
104     if (not exists $param{data} and not exists $param{bug}) {
105         croak "One of data or bug must be provided to load_bug";
106     }
107     if (not exists $param{data}) {
108         $param{data} = read_bug(bug => $param{bug});
109     }
110     my $data = $param{data};
111     my $tags = $param{tags};
112     my $queue = $param{queue};
113     my $severities = $param{severities};
114     my $can_queue = 1;
115     if (not defined $queue) {
116         $can_queue = 0;
117         $queue = {};
118     }
119     my %tags;
120     my $s_data = split_status_fields($data);
121     for my $tag (make_list($s_data->{keywords})) {
122         next unless defined $tag and length $tag;
123         # this allows for invalid tags. But we'll use this to try to
124         # find those bugs and clean them up
125         if (not exists $tags->{$tag}) {
126             $tags->{$tag} = $s->resultset('Tag')->
127             find_or_create({tag => $tag});
128         }
129         $tags{$tag} = $tags->{$tag};
130     }
131     my $severity = length($data->{severity}) ? $data->{severity} : $config{default_severity};
132     if (exists $severities->{$severity}) {
133         $severity = $severities->{$severity};
134     } else {
135         $severity = $s->resultset('Severity')->
136             find_or_create({severity => $severity});
137     }
138     my $bug =
139         {id => $data->{bug_num},
140          creation => DateTime->from_epoch(epoch => $data->{date}),
141          log_modified => DateTime->from_epoch(epoch => $data->{log_modified}),
142          last_modified => DateTime->from_epoch(epoch => $data->{last_modified}),
143          archived => $data->{archived},
144          (defined $data->{unarchived} and length($data->{unarchived}))?(unarchived => DateTime->from_epoch(epoch => $data->{unarchived})):(),
145          forwarded => $data->{forwarded} // '',
146          summary => $data->{summary} // '',
147          outlook => $data->{outlook} // '',
148          subject => $data->{subject} // '',
149          done_full => $data->{done} // '',
150          severity => $severity,
151          owner_full => $data->{owner} // '',
152          submitter_full => $data->{originator} // '',
153         };
154     my %addr_map =
155         (done => 'done',
156          owner => 'owner',
157          submitter => 'originator',
158         );
159     for my $addr_type (keys %addr_map) {
160         my @addrs = getparsedaddrs($data->{$addr_map{$addr_type}} // '');
161         next unless @addrs;
162         $bug->{$addr_type} = $s->resultset('Correspondent')->find_or_create({addr => lc($addrs[0]->address())});
163         # insert the full name as well
164         my $full_name = $addrs[0]->phrase();
165         $full_name =~ s/^\"|\"$//g;
166         $full_name =~ s/^\s+|\s+$//g;
167         if (length $full_name) {
168             $bug->{$addr_type}->
169                 update_or_create_related('correspondent_full_names',
170                                         {full_name=>$full_name,
171                                          last_seen => 'NOW()'});
172         }
173     }
174     my $b = $s->resultset('Bug')->update_or_create($bug) or
175         die "Unable to update or create bug $bug->{id}";
176      $s->txn_do(sub {
177                    for my $ff (qw(found fixed)) {
178                        my @elements = $s->resultset('BugVer')->search({bug => $data->{bug_num},
179                                                                        found  => $ff eq 'found'?1:0,
180                                                                       });
181                        my %elements_to_delete = map {($elements[$_]->ver_string(),$elements[$_])} 0..$#elements;
182                        my %elements_to_add;
183                        my @elements_to_keep;
184                        for my $version (@{$data->{"${ff}_versions"}}) {
185                            if (exists $elements_to_delete{$version}) {
186                                push @elements_to_keep,$version;
187                            } else {
188                                $elements_to_add{$version} = 1;
189                            }
190                        }
191                        for my $version (@elements_to_keep) {
192                            delete $elements_to_delete{$version};
193                        }
194                        for my $element (keys %elements_to_delete) {
195                            $elements_to_delete{$element}->delete();
196                        }
197                        for my $element (keys %elements_to_add) {
198                            # find source package and source version id
199                            my $ne = $s->resultset('BugVer')->new_result({bug => $data->{bug_num},
200                                                                          ver_string => $element,
201                                                                          found => $ff eq 'found'?1:0,
202                                                                         }
203                                                                        );
204                            if (my ($src_pkg,$src_ver) = $element =~ m{^([^\/]+)/(.+)$}) {
205                                my $src_pkg_e = $s->resultset('SrcPkg')->single({pkg => $src_pkg});
206                                if (defined $src_pkg_e) {
207                                    $ne->src_pkg($src_pkg_e->id());
208                                    my $src_ver_e = $s->resultset('SrcVer')->single({src_pkg => $src_pkg_e->id(),
209                                                                                     ver => $src_ver
210                                                                                    });
211                                    $ne->src_ver($src_ver_e->id()) if defined $src_ver_e;
212                                }
213                            }
214                            $ne->insert();
215                        }
216                    }
217                });
218     $s->txn_do(sub {
219                    my $t = $s->resultset('BugTag')->search({bug => $data->{bug_num}});
220                    $t->delete() if defined $t;
221                    $s->populate(BugTag => [[qw(bug tag)], map {[$data->{bug_num}, $_->id()]} values %tags]);
222                });
223     # because these bugs reference other bugs which might not exist
224     # yet, we can't handle them until we've loaded all bugs. queue
225     # them up.
226     for my $merge_block (qw(merged block)) {
227         my $data_key = $merge_block;
228         $data_key .= 'with' if $merge_block eq 'merged';
229         if (@{$data->{$data_key}||[]}) {
230             my $count = $s->resultset('Bug')->search({id => [@{$data->{$data_key}}]})->count();
231             if ($count == @{$data->{$data_key}}) {
232                 handle_load_bug_queue(db=>$s,
233                                       queue => {$merge_block,
234                                                {$data->{bug_num},[@{$data->{$data_key}}]}
235                                                });
236             } else {
237                 $queue->{$merge_block}{$data->{bug_num}} = [@{$data->{$data_key}}];
238             }
239         }
240     }
241
242     if (not $can_queue and keys %{$queue}) {
243         handle_load_bug_queue(db => $s,queue => $queue);
244     }
245
246     # still need to handle merges, versions, etc.
247 }
248
249 =item handle_load_bug_queue
250
251      handle_load_bug_queue(db => $schema,queue => $queue);
252
253 Handles a queue of operations created by load bug. [These operations
254 are used to handle cases where a bug referenced by a loaded bug may
255 not exist yet. In cases where the bugs should exist, the queue is
256 cleared automatically by load_bug if queue is undefined.
257
258 =cut
259
260 sub handle_load_bug_queue{
261     my %param = validate_with(params => \@_,
262                               spec => {db => {type => OBJECT,
263                                              },
264                                        queue => {type => HASHREF,
265                                                 },
266                                       });
267     my $s = $param{db};
268     my $queue = $param{queue};
269     my %queue_types =
270         (merged => {set => 'BugMerged',
271                     columns => [qw(bug merged)],
272                     bug => 'bug',
273                    },
274          blocks => {set => 'BugBlock',
275                     columns => [qw(bug blocks)],
276                     bug => 'bug',
277                    },
278         );
279     for my $queue_type (keys %queue_types) {
280         for my $bug (%{$queue->{$queue_type}}) {
281             my $qt = $queue_types{$queue_type};
282             $s->txn_do(sub {
283                            $s->resultset($qt->{set})->search({$qt->{bug},$bug})->delete();
284                            $s->populate($qt->{set},[[@{$qt->{columns}}],
285                                                     map {[$bug,$_]} @{$queue->{$queue_type}{$bug}}]) if
286                                @{$queue->{$queue_type}{$bug}//[]};
287                        }
288                       );
289         }
290     }
291 }
292
293 =item load_bug_log -- load bug logs
294
295        load_bug_log(db  => $s,
296                     bug => $bug);
297
298
299 =over
300
301 =item db -- database 
302
303 =item bug -- bug whose log should be loaded
304
305 =back
306
307 =cut
308
309 sub load_bug_log {
310     my %param = validate_with(params => \@_,
311                               spec => {db => {type => OBJECT,
312                                              },
313                                        bug => {type => SCALAR,
314                                               },
315                                        queue => {type => HASHREF,
316                                                  optional => 1},
317                                       });
318     my $s = $param{db};
319     my $msg_num=0;
320     my %seen_msg_ids;
321     my $log = Debbugs::Log->new(bug_num => $param{bug}) or
322         die "Unable to open log for $param{bug} for reading: $!";
323     while (my $record = $log->read_record()) {
324         next unless $record->{type} eq 'incoming-recv';
325         my ($msg_id) = $record->{text} =~ /^Message-Id:\s+<(.+)>/im;
326         next if defined $msg_id and exists $seen_msg_ids{$msg_id};
327         $seen_msg_ids{$msg_id} = 1 if defined $msg_id;
328         next if defined $msg_id and $msg_id =~ /handler\..+\.ack(?:info)?\@/;
329         my $message = parse($record->{text});
330         # search for a message with this message id in the database
331         
332         # check to see if the subject, to, and from match. if so, it's
333         # probably the same message.
334
335         # if not, create a new message
336
337         # add correspondents if necessary
338
339         # link message to bugs if necessary
340
341     }
342
343 }
344
345 =back
346
347 =head2 Debinfo
348
349 Commands to handle src and package version loading from debinfo files
350
351 =over
352
353 =item load_debinfo
354
355      load_debinfo($schema,$binname, $binver, $binarch, $srcname, $srcver);
356
357
358
359 =cut
360
361 sub load_debinfo {
362     my ($schema,$binname, $binver, $binarch, $srcname, $srcver) = @_;
363     my $sp = $schema->resultset('SrcPkg')->find_or_create({pkg => $srcname});
364     my $sv = $schema->resultset('SrcVer')->find_or_create({src_pkg=>$sp->id(),
365                                                            ver => $srcver});
366     my $arch = $schema->resultset('Arch')->find_or_create({arch => $binarch});
367     my $bp = $schema->resultset('BinPkg')->find_or_create({pkg => $binname});
368     $schema->resultset('BinVer')->find_or_create({bin_pkg_id => $bp->id(),
369                                                   src_ver_id => $sv->id(),
370                                                   arch_id    => $arch->id(),
371                                                   ver        => $binver,
372                                                  });
373 }
374
375
376 =back
377
378 =head Packages
379
380 =over
381
382 =item load_package
383
384      load_package($schema,$suite,$component,$arch,$pkg)
385
386 =cut
387
388 sub load_packages {
389     my ($schema,$suite,$pkgs,$p) = @_;
390     my $suite_id = $schema->resultset('Suite')->
391         find_or_create({codename => $suite})->id;
392     my %maint_cache;
393     my %arch_cache;
394     my %source_cache;
395     my $src_max_last_modified = $schema->resultset('SrcAssociation')->
396         search_rs({suite => $suite_id},
397                  {order_by => {-desc => ['me.modified']},
398                   rows => 1,
399                   page => 1
400                  }
401                  )->single();
402     my $bin_max_last_modified = $schema->resultset('BinAssociation')->
403         search_rs({suite => $suite_id},
404                  {order_by => {-desc => ['me.modified']},
405                   rows => 1,
406                   page => 1
407                  }
408                  )->single();
409     my %maints;
410     my %sources;
411     my %bins;
412     for my $pkg_tuple (@{$pkgs}) {
413         my ($arch,$component,$pkg) = @{$pkg_tuple};
414         $maints{$pkg->{Maintainer}} = $pkg->{Maintainer};
415         if ($arch eq 'source') {
416             my $source = $pkg->{Package};
417             my $source_ver = $pkg->{Version};
418             $sources{$source}{$source_ver} = $pkg->{Maintainer};
419         } else {
420             my $source = $pkg->{Source} // $pkg->{Package};
421             my $source_ver = $pkg->{Version};
422             if ($source =~ /^\s*(\S+) \(([^\)]+)\)\s*$/) {
423                 ($source,$source_ver) = ($1,$2);
424             }
425             $sources{$source}{$source_ver} = $pkg->{Maintainer};
426             $bins{$arch}{$pkg->{Package}} =
427                {arch => $arch,
428                 bin => $pkg->{Package},
429                 bin_ver => $pkg->{Version},
430                 src_ver => $source_ver,
431                 source  => $source,
432                 maint   => $pkg->{Maintainer},
433                };
434         }
435     }
436     # Retrieve and Insert new maintainers
437     my $maints =
438         $schema->resultset('Maintainer')->
439         get_maintainers(keys %maints);
440     my $archs =
441         $schema->resultset('Arch')->
442         get_archs(keys %bins);
443     # We want all of the source package/versions which are in this suite to
444     # start with
445     my @sa_to_add;
446     my @sa_to_del;
447     my %included_sa;
448     # Calculate which source packages are no longer in this suite
449     for my $s ($schema->resultset('SrcPkg')->
450                src_pkg_and_ver_in_suite($suite)) {
451         if (not exists $sources{$s->{pkg}} or
452             not exists $sources{$s->{pkg}}{$s->{src_vers}{ver}}
453            ) {
454             push @sa_to_del,
455                 $s->{src_associations}{id};
456         }
457         $included_sa{$s->{pkg}}{$s->{src_vers}} = 1;
458     }
459     # Calculate which source packages are newly in this suite
460     for my $s (keys %sources) {
461         for my $v (keys %{$sources{$s}}) {
462             if (not exists $included_sa{$s} and
463                 not $included_sa{$s}{$v}) {
464                 push @sa_to_add,
465                     [$s,$v,$sources{$s}{$v}];
466             } else {
467                 $p->update() if defined $p;
468             }
469         }
470     }
471     # add new source packages
472     my $it = natatime 100, @sa_to_add;
473     while (my @v = $it->()) {
474         $schema->txn_do(
475             sub {
476                 for my $svm (@_) {
477                     my $s_id = $schema->resultset('SrcPkg')->
478                         get_src_pkg_id($svm->[0]);
479                     my $sv_id = $schema->resultset('SrcVer')->
480                         get_src_ver_id($s_id,$svm->[1],$maints->{$svm->[2]});
481                     $schema->resultset('SrcAssociation')->
482                         insert_suite_src_ver_association($suite_id,$sv_id);
483                 }
484             },
485                         @v
486                        );
487         $p->update($p->last_update()+
488                    scalar @v) if defined $p;
489     }
490     # remove associations for packages not in this suite
491     if (@sa_to_del) {
492         $schema->resultset('SrcAssociation')->
493             search_rs({id => \@sa_to_del})->delete();
494     }
495     # update packages in this suite to have a modification time of now
496     $schema->resultset('SrcAssociation')->
497         search_rs({suite => $suite_id})->
498         update({modified => 'NOW()'});
499     ## Handle binary packages
500     my @bin_to_del;
501     my @bin_to_add;
502     my %included_bin;
503     # calculate which binary packages are no longer in this suite
504     for my $b ($schema->resultset('BinPkg')->
505                bin_pkg_and_ver_in_suite($suite)) {
506         if (not exists $bins{$b->{arch}{arch}} or
507             not exists $bins{$b->{arch}{arch}}{$b->{pkg}} or
508             ($bins{$b->{arch}{arch}}{$b->{pkg}}{bin_ver} ne
509              $b->{bin_vers}{ver}
510             )
511            ) {
512             push @bin_to_del,
513                 $b->{bin_associations}{id};
514         }
515         $included_bin{$b->{arch}{arch}}{$b->{pkg}} =
516             $b->{bin_vers}{ver};
517     }
518     # calculate which binary packages are newly in this suite
519     for my $a (keys %bins) {
520         for my $pkg (keys %{$bins{$a}}) {
521             if (not exists $included_bin{$a} or
522                 not exists $included_bin{$a}{$pkg} or
523                 $bins{$a}{$pkg}{bin_ver} ne
524                 $included_bin{$a}{$pkg}) {
525                 push @bin_to_add,
526                     $bins{$a}{$pkg};
527             } else {
528                 $p->update() if defined $p;
529             }
530         }
531     }
532     $it = natatime 100, @bin_to_add;
533     while (my @v = $it->()) {
534         $schema->txn_do(
535         sub {
536             for my $bvm (@_) {
537                 my $s_id = $schema->resultset('SrcPkg')->
538                     get_src_pkg_id($bvm->{source});
539                 my $sv_id = $schema->resultset('SrcVer')->
540                     get_src_ver_id($s_id,$bvm->{src_ver},$maints->{$bvm->{maint}});
541                 my $b_id = $schema->resultset('BinPkg')->
542                     get_bin_pkg_id($bvm->{bin});
543                 my $bv_id = $schema->resultset('BinVer')->
544                     get_bin_ver_id($b_id,$bvm->{bin_ver},
545                                    $archs->{$bvm->{arch}},$sv_id);
546                 $schema->resultset('BinAssociation')->
547                     insert_suite_bin_ver_association($suite_id,$bv_id);
548             }
549         },
550                         @v
551                        );
552         $p->update($p->last_update()+
553                    scalar @v) if defined $p;
554     }
555     if (@bin_to_del) {
556         $schema->resultset('BinAssociation')->
557             search_rs({id => \@bin_to_del})->delete();
558     }
559     $schema->resultset('BinAssociation')->
560         search_rs({suite => $suite_id})->
561         update({modified => 'NOW()'});
562
563 }
564
565
566 =back
567
568 =cut
569
570 =head Suites
571
572 =over
573
574 =item load_suite
575
576      load_suite($schema,$codename,$suite,$version,$active);
577
578 =cut
579
580 sub load_suite {
581     my ($schema,$codename,$suite,$version,$active) = @_;
582     if (ref($codename)) {
583         ($codename,$suite,$version) =
584             @{$codename}{qw(Codename Suite Version)};
585         $active = 1;
586     }
587     my $s = $schema->resultset('Suite')->find_or_create({codename => $codename});
588     $s->suite_name($suite);
589     $s->version($version);
590     $s->active($active);
591     $s->update();
592     return $s;
593
594 }
595
596 =back
597
598 =cut
599
600 1;
601
602
603 __END__