]> git.donarmstrong.com Git - debbugs.git/blob - Debbugs/DB/Load.pm
switch load_packages to use gzip -dc
[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_package)],
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
46 use Debbugs::Status qw(read_bug split_status_fields);
47 use Debbugs::DB;
48 use DateTime;
49 use Debbugs::Common qw(make_list getparsedaddrs);
50 use Debbugs::Config qw(:config);
51 use Carp;
52
53 =head2 Bug loading
54
55 Routines to load bug; exported with :load_bug
56
57 =over
58
59 =item load_bug
60
61      load_bug(db => $schema,
62               data => split_status_fields($data),
63               tags => \%tags,
64               queue => \%queue);
65
66 Loads a bug's metadata into the database. (Does not load any messages)
67
68 =over
69
70 =item db -- Debbugs::DB object
71
72 =item data -- Bug data (from read_bug) which has been split with split_status_fields
73
74 =item tags -- tag cache (hashref); optional
75
76 =item queue -- queue of operations to perform after bug is loaded; optional.
77
78 =back
79
80 =cut
81
82 sub load_bug {
83     my %param = validate_with(params => \@_,
84                               spec => {db => {type => OBJECT,
85                                              },
86                                        data => {type => HASHREF,
87                                                 optional => 1,
88                                                },
89                                        bug => {type => SCALAR,
90                                                optional => 1,
91                                               },
92                                        tags => {type => HASHREF,
93                                                 default => sub {return {}},
94                                                 optional => 1},
95                                        severities => {type => HASHREF,
96                                                       default => sub {return {}},
97                                                       optional => 1,
98                                                      },
99                                        queue => {type => HASHREF,
100                                                  optional => 1},
101                                       });
102     my $s = $param{db};
103     if (not exists $param{data} and not exists $param{bug}) {
104         croak "One of data or bug must be provided to load_bug";
105     }
106     if (not exists $param{data}) {
107         $param{data} = read_bug(bug => $param{bug});
108     }
109     my $data = $param{data};
110     my $tags = $param{tags};
111     my $queue = $param{queue};
112     my $severities = $param{severities};
113     my $can_queue = 1;
114     if (not defined $queue) {
115         $can_queue = 0;
116         $queue = {};
117     }
118     my %tags;
119     my $s_data = split_status_fields($data);
120     for my $tag (make_list($s_data->{keywords})) {
121         next unless defined $tag and length $tag;
122         # this allows for invalid tags. But we'll use this to try to
123         # find those bugs and clean them up
124         if (not exists $tags->{$tag}) {
125             $tags->{$tag} = $s->resultset('Tag')->
126             find_or_create({tag => $tag});
127         }
128         $tags{$tag} = $tags->{$tag};
129     }
130     my $severity = length($data->{severity}) ? $data->{severity} : $config{default_severity};
131     if (exists $severities->{$severity}) {
132         $severity = $severities->{$severity};
133     } else {
134         $severity = $s->resultset('Severity')->
135             find_or_create({severity => $severity});
136     }
137     my $bug =
138         {id => $data->{bug_num},
139          creation => DateTime->from_epoch(epoch => $data->{date}),
140          log_modified => DateTime->from_epoch(epoch => $data->{log_modified}),
141          last_modified => DateTime->from_epoch(epoch => $data->{last_modified}),
142          archived => $data->{archived},
143          (defined $data->{unarchived} and length($data->{unarchived}))?(unarchived => DateTime->from_epoch(epoch => $data->{unarchived})):(),
144          forwarded => $data->{forwarded} // '',
145          summary => $data->{summary} // '',
146          outlook => $data->{outlook} // '',
147          subject => $data->{subject} // '',
148          done_full => $data->{done} // '',
149          severity => $severity,
150          owner_full => $data->{owner} // '',
151          submitter_full => $data->{originator} // '',
152         };
153     my %addr_map =
154         (done => 'done',
155          owner => 'owner',
156          submitter => 'originator',
157         );
158     for my $addr_type (keys %addr_map) {
159         my @addrs = getparsedaddrs($data->{$addr_map{$addr_type}} // '');
160         next unless @addrs;
161         $bug->{$addr_type} = $s->resultset('Correspondent')->find_or_create({addr => lc($addrs[0]->address())});
162         # insert the full name as well
163         my $full_name = $addrs[0]->phrase();
164         $full_name =~ s/^\"|\"$//g;
165         $full_name =~ s/^\s+|\s+$//g;
166         if (length $full_name) {
167             $bug->{$addr_type}->
168                 update_or_create_related('correspondent_full_names',
169                                         {full_name=>$full_name,
170                                          last_seen => 'NOW()'});
171         }
172     }
173     my $b = $s->resultset('Bug')->update_or_create($bug) or
174         die "Unable to update or create bug $bug->{id}";
175      $s->txn_do(sub {
176                    for my $ff (qw(found fixed)) {
177                        my @elements = $s->resultset('BugVer')->search({bug => $data->{bug_num},
178                                                                        found  => $ff eq 'found'?1:0,
179                                                                       });
180                        my %elements_to_delete = map {($elements[$_]->ver_string(),$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     $s->txn_do(sub {
218                    my $t = $s->resultset('BugTag')->search({bug => $data->{bug_num}});
219                    $t->delete() if defined $t;
220                    $s->populate(BugTag => [[qw(bug tag)], map {[$data->{bug_num}, $_->id()]} values %tags]);
221                });
222     # because these bugs reference other bugs which might not exist
223     # yet, we can't handle them until we've loaded all bugs. queue
224     # them up.
225     for my $merge_block (qw(merged block)) {
226         my $data_key = $merge_block;
227         $data_key .= 'with' if $merge_block eq 'merged';
228         if (@{$data->{$data_key}||[]}) {
229             my $count = $s->resultset('Bug')->search({id => [@{$data->{$data_key}}]})->count();
230             if ($count == @{$data->{$data_key}}) {
231                 handle_load_bug_queue(db=>$s,
232                                       queue => {$merge_block,
233                                                {$data->{bug_num},[@{$data->{$data_key}}]}
234                                                });
235             } else {
236                 $queue->{$merge_block}{$data->{bug_num}} = [@{$data->{$data_key}}];
237             }
238         }
239     }
240
241     if (not $can_queue and keys %{$queue}) {
242         handle_load_bug_queue(db => $s,queue => $queue);
243     }
244
245     # still need to handle merges, versions, etc.
246 }
247
248 =item handle_load_bug_queue
249
250      handle_load_bug_queue(db => $schema,queue => $queue);
251
252 Handles a queue of operations created by load bug. [These operations
253 are used to handle cases where a bug referenced by a loaded bug may
254 not exist yet. In cases where the bugs should exist, the queue is
255 cleared automatically by load_bug if queue is undefined.
256
257 =cut
258
259 sub handle_load_bug_queue{
260     my %param = validate_with(params => \@_,
261                               spec => {db => {type => OBJECT,
262                                              },
263                                        queue => {type => HASHREF,
264                                                 },
265                                       });
266     my $s = $param{db};
267     my $queue = $param{queue};
268     my %queue_types =
269         (merged => {set => 'BugMerged',
270                     columns => [qw(bug merged)],
271                     bug => 'bug',
272                    },
273          blocks => {set => 'BugBlock',
274                     columns => [qw(bug blocks)],
275                     bug => 'bug',
276                    },
277         );
278     for my $queue_type (keys %queue_types) {
279         for my $bug (%{$queue->{$queue_type}}) {
280             my $qt = $queue_types{$queue_type};
281             $s->txn_do(sub {
282                            $s->resultset($qt->{set})->search({$qt->{bug},$bug})->delete();
283                            $s->populate($qt->{set},[[@{$qt->{columns}}],
284                                                     map {[$bug,$_]} @{$queue->{$queue_type}{$bug}}]) if
285                                @{$queue->{$queue_type}{$bug}//[]};
286                        }
287                       );
288         }
289     }
290 }
291
292 =item load_bug_log -- load bug logs
293
294        load_bug_log(db  => $s,
295                     bug => $bug);
296
297
298 =over
299
300 =item db -- database 
301
302 =item bug -- bug whose log should be loaded
303
304 =back
305
306 =cut
307
308 sub load_bug_log {
309     my %param = validate_with(params => \@_,
310                               spec => {db => {type => OBJECT,
311                                              },
312                                        bug => {type => SCALAR,
313                                               },
314                                        queue => {type => HASHREF,
315                                                  optional => 1},
316                                       });
317     my $s = $param{db};
318     my $msg_num=0;
319     my %seen_msg_ids;
320     my $log = Debbugs::Log->new(bug_num => $param{bug}) or
321         die "Unable to open log for $param{bug} for reading: $!";
322     while (my $record = $log->read_record()) {
323         next unless $record->{type} eq 'incoming-recv';
324         my ($msg_id) = $record->{text} =~ /^Message-Id:\s+<(.+)>/im;
325         next if defined $msg_id and exists $seen_msg_ids{$msg_id};
326         $seen_msg_ids{$msg_id} = 1 if defined $msg_id;
327         next if defined $msg_id and $msg_id =~ /handler\..+\.ack(?:info)?\@/;
328         my $message = parse($record->{text});
329         # search for a message with this message id in the database
330         
331         # check to see if the subject, to, and from match. if so, it's
332         # probably the same message.
333
334         # if not, create a new message
335
336         # add correspondents if necessary
337
338         # link message to bugs if necessary
339
340     }
341
342 }
343
344 =back
345
346 =head2 Debinfo
347
348 Commands to handle src and package version loading from debinfo files
349
350 =over
351
352 =item load_debinfo
353
354      load_debinfo($schema,$binname, $binver, $binarch, $srcname, $srcver);
355
356
357
358 =cut
359
360 sub load_debinfo {
361     my ($schema,$binname, $binver, $binarch, $srcname, $srcver) = @_;
362     my $sp = $schema->resultset('SrcPkg')->find_or_create({pkg => $srcname});
363     my $sv = $schema->resultset('SrcVer')->find_or_create({src_pkg=>$sp->id(),
364                                                            ver => $srcver});
365     my $arch = $schema->resultset('Arch')->find_or_create({arch => $binarch});
366     my $bp = $schema->resultset('BinPkg')->find_or_create({pkg => $binname});
367     $schema->resultset('BinVer')->find_or_create({bin_pkg_id => $bp->id(),
368                                                   src_ver_id => $sv->id(),
369                                                   arch_id    => $arch->id(),
370                                                   ver        => $binver,
371                                                  });
372 }
373
374
375 =back
376
377 =head Packages
378
379 =over
380
381 =item load_package
382
383      load_package($schema,$suite,$component,$arch,$pkg)
384
385 =cut
386
387 sub load_packages {
388     my ($schema,$suite,$pkgs,$p) = @_;
389     my $suite_id = $schema->resultset('Suite')->
390         find_or_create({codename => $suite})->id;
391     my %maint_cache;
392     my %arch_cache;
393     my %source_cache;
394     my $src_max_last_modified = $schema->resultset('SrcAssociation')->
395         search_rs({suite => $suite_id},
396                  {order_by => {-desc => ['me.modified']},
397                   rows => 1,
398                   page => 1
399                  }
400                  )->single();
401     my $bin_max_last_modified = $schema->resultset('BinAssociation')->
402         search_rs({suite => $suite_id},
403                  {order_by => {-desc => ['me.modified']},
404                   rows => 1,
405                   page => 1
406                  }
407                  )->single();
408     print STDERR time." handling packages\n";
409     for my $pkg_tuple (@{$pkgs}) {
410         my ($arch,$component,$pkg) = @{$pkg_tuple};
411         $p->update() if $p;
412         if ($arch eq 'source') {
413             my $source = $pkg->{Package};
414             my $source_ver = $pkg->{Version};
415             if (not exists $maint_cache{$pkg->{Maintainer}}) {
416                 my @addrs = getparsedaddrs($pkg->{Maintainer} // '');
417                 if (@addrs) {
418                     my $mc = $schema->resultset('Correspondent')->
419                         find_or_create({addr => lc($addrs[0]->address())},
420                                       {key => 'correspondent_addr_idx'}
421                                       );
422                     my $full_name = $addrs[0]->phrase();
423                     $full_name =~ s/^\"|\"$//g;
424                     $full_name =~ s/^\s+|\s+$//g;
425                     # $sv->discard_changes;
426                     my $maint = $schema->resultset('Maintainer')->
427                         find_or_create({name => $pkg->{Maintainer},
428                                         correspondent => $mc->id},
429                                       {key => 'maintainer_name_idx'},
430                                       );
431                     $mc->find_or_create_related('correspondent_full_names',
432                                                {full_name => $full_name},
433                                                {key => 'correspondent_full_name_correspondent_full_name_idx'}
434                                                );
435                     $mc->update;
436                     $maint_cache{$pkg->{Maintainer}} = $maint;
437                 }
438             }
439             if (not exists $source_cache{$source}{$source_ver}) {
440                 my $sp = $schema->resultset('SrcPkg')->
441                     find_or_create({pkg => $source});
442                 my $sv = $sp->find_or_create_related('src_vers',
443                                                     {ver => $source_ver});
444                 $source_cache{$source}{$source_ver} = $sv;
445                 if (exists $maint_cache{$pkg->{Maintainer}}) {
446                     $source_cache{$source}{$source_ver}->
447                         set_from_related('maintainer',
448                                          $maint_cache{$pkg->{Maintainer}}
449                                         );
450                     $source_cache{$source}{$source_ver}->update;
451                 }
452             }
453             $schema->resultset('SrcAssociation')->
454                 update_or_create({suite => $suite_id,
455                                   source => $source_cache{$source}{$source_ver}->id,
456                                   modified => 'NOW()',
457                                  },
458                                 {key => 'src_associations_source_suite'}
459                                 );
460         } else {
461             my $ar = $schema->resultset('Arch')->
462                 find_or_create(arch => $arch);
463             my $bp = $schema->resultset('BinPkg')->
464                 find_or_create({pkg => $pkg->{Package}});
465             my $source = $pkg->{Source} // $pkg->{Package};
466             my $source_ver = $pkg->{Version};
467             if ($source =~ /^\s*(\S+) \(([^\)]+)\)\s*$/) {
468                 ($source,$source_ver) = ($1,$2);
469             }
470             if (not exists $source_cache{$source}{$source_ver}) {
471                 my $sp = $schema->resultset('SrcPkg')->
472                     find_or_create({pkg => $source});
473                 my $sv = $sp->find_or_create_related('src_vers',
474                                                     {ver => $source_ver});
475                 $source_cache{$source}{$source_ver} = $sv;
476             }
477             my $bv = $bp->find_or_create_related('bin_vers',
478                                                 {ver => $pkg->{Version},
479                                                  src_ver => $source_cache{$source}{$source_ver}->id,
480                                                  arch => $ar->id,
481                                                 });
482             $schema->resultset('BinAssociation')->
483                 update_or_create({suite => $suite_id,
484                                   bin => $bv->id,
485                                   modified => 'NOW()',
486                                  },
487                                 {key => 'bin_associations_bin_suite'}
488                                 );
489         }
490     }
491     print STDERR time." deleting associations\n";
492     # delete old binary associations in this suite which have not recently been
493     # modified
494     $schema->resultset('BinAssociation')->
495         search_rs({suite => $suite_id,
496                    modified => {'<',$bin_max_last_modified->modified()},
497                   }) if defined
498                       $bin_max_last_modified;
499     $schema->resultset('SrcAssociation')->
500         search_rs({suite => $suite_id,
501                    modified => {'<',$src_max_last_modified->modified()},
502                   }) if defined
503                       $src_max_last_modified;
504 }
505
506 =back
507
508 =cut
509
510 =head Suites
511
512 =over
513
514 =item load_suite
515
516      load_suite($schema,$codename,$suite,$version,$active);
517
518 =cut
519
520 sub load_suite {
521     my ($schema,$codename,$suite,$version,$active) = @_;
522     if (ref($codename)) {
523         ($codename,$suite,$version) =
524             @{$codename}{qw(Codename Suite Version)};
525         $active = 1;
526     }
527     my $s = $schema->resultset('Suite')->find_or_create({codename => $codename});
528     $s->suite_name($suite);
529     $s->version($version);
530     $s->active($active);
531     $s->update();
532     return $s;
533
534 }
535
536 =back
537
538 =cut
539
540 1;
541
542
543 __END__