1 # Copyright (C) 2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
3 # based on notmuch, but with no concept of folders, files or flags
5 # Indexes mail with Xapian and our (SQLite-based) ::Msgmap for use
6 # with the web and NNTP interfaces. This index maintains thread
7 # relationships for use by Mail::Thread. This writes to the search
9 package PublicInbox::SearchIdx;
12 use Fcntl qw(:flock :DEFAULT);
14 use Email::MIME::ContentType;
15 $Email::MIME::ContentType::STRICT_PARAMS = 0;
16 use base qw(PublicInbox::Search);
17 use PublicInbox::MID qw/mid_clean id_compress mid_mime/;
18 use PublicInbox::MsgIter;
20 use POSIX qw(strftime);
21 require PublicInbox::Git;
22 *xpfx = *PublicInbox::Search::xpfx;
24 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
28 OLD_PERM_EVERYBODY => 2,
30 PERM_EVERYBODY => 0664,
34 my ($class, $inbox, $creat) = @_;
38 $git_dir = $inbox->{mainrepo};
39 $altid = $inbox->{altid};
41 require PublicInbox::AltId;
43 PublicInbox::AltId->new($inbox, $_);
47 require Search::Xapian::WritableDatabase;
48 my $self = bless { git_dir => $git_dir, -altid => $altid }, $class;
49 my $perm = $self->_git_config_perm;
50 my $umask = _umask_for($perm);
51 $self->{umask} = $umask;
52 $self->{lock_path} = "$git_dir/ssoma.lock";
53 $self->{git} = PublicInbox::Git->new($git_dir);
54 $self->{creat} = ($creat || 0) == 1;
60 my $xdb = delete $self->{xdb} or croak 'not acquired';
62 _lock_release($self) if $self->{creat};
68 croak 'already acquired' if $self->{xdb};
69 my $dir = PublicInbox::Search->xdir($self->{git_dir});
70 my $flag = Search::Xapian::DB_OPEN;
74 File::Path::mkpath($dir);
75 $self->{batch_size} = 100;
76 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
78 $self->{xdb} = Search::Xapian::WritableDatabase->new($dir, $flag);
81 # we only acquire the flock if creating or reindexing;
82 # PublicInbox::Import already has the lock on its own.
85 croak 'already locked' if $self->{lockfh};
86 sysopen(my $lockfh, $self->{lock_path}, O_WRONLY|O_CREAT) or
87 die "failed to open lock $self->{lock_path}: $!\n";
88 flock($lockfh, LOCK_EX) or die "lock failed: $!\n";
89 $self->{lockfh} = $lockfh;
94 my $lockfh = delete $self->{lockfh} or croak 'not locked';
95 flock($lockfh, LOCK_UN) or die "unlock failed: $!\n";
96 close $lockfh or die "close failed: $!\n";
100 my ($doc, $col, $num) = @_;
101 $num = Search::Xapian::sortable_serialise($num);
102 $doc->add_value($col, $num);
106 my ($self, $mime, $bytes, $num, $blob) = @_; # mime = Email::MIME object
107 my $db = $self->{xdb};
109 my ($doc_id, $old_tid);
110 my $mid = mid_clean(mid_mime($mime));
111 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
114 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
115 my $smsg = $self->lookup_message($mid);
117 # convert a ghost to a regular message
118 # it will also clobber any existing regular message
119 $doc_id = $smsg->doc_id;
120 $old_tid = $smsg->thread_id;
122 $smsg = PublicInbox::SearchMsg->new($mime);
123 my $doc = $smsg->{doc};
124 $doc->add_term(xpfx('mid') . $mid);
126 my $subj = $smsg->subject;
128 my $path = $self->subject_path($subj);
129 $doc->add_term(xpfx('path') . id_compress($path));
133 add_val($doc, &PublicInbox::Search::TS, $ts);
136 add_val($doc, &PublicInbox::Search::NUM, $num);
139 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
141 add_val($doc, &PublicInbox::Search::LINES,
142 $mime->body_raw =~ tr!\n!\n!);
144 my $yyyymmdd = strftime('%Y%m%d', gmtime($ts));
145 $doc->add_value(&PublicInbox::Search::YYYYMMDD, $yyyymmdd);
147 my $tg = $self->term_generator;
149 $tg->set_document($doc);
150 $tg->index_text($subj, 1, 'S') if $subj;
151 $tg->increase_termpos;
152 $tg->index_text($subj) if $subj;
153 $tg->increase_termpos;
155 $tg->index_text($smsg->from);
156 $tg->increase_termpos;
158 msg_iter($mime, sub {
159 my ($part, $depth, @idx) = @{$_[0]};
160 my $ct = $part->content_type || $ct_msg;
162 # account for filter bugs...
163 $ct =~ m!\btext/plain\b!i or return;
166 my $body = $part->body;
168 my @lines = split(/\n/, $body);
169 while (defined(my $l = shift @lines)) {
177 $tg->index_text(join("\n", @quot), 0);
179 $tg->increase_termpos;
182 $tg->index_text(join("\n", @orig));
184 $tg->increase_termpos;
188 link_message($self, $smsg, $old_tid);
189 $tg->index_text($mid, 1);
190 $doc->set_data($smsg->to_doc_data($blob));
192 if (my $altid = $self->{-altid}) {
193 foreach my $alt (@$altid) {
194 my $id = $alt->mid2alt($mid);
195 next unless defined $id;
196 $doc->add_term($alt->{xprefix} . $id);
199 if (defined $doc_id) {
200 $db->replace_document($doc_id, $doc);
202 $doc_id = $db->add_document($doc);
207 warn "failed to index message <$mid>: $@\n";
213 # returns deleted doc_id on success, undef on missing
215 my ($self, $mid) = @_;
216 my $db = $self->{xdb};
218 $mid = mid_clean($mid);
221 $doc_id = $self->find_unique_doc_id('mid', $mid);
222 $db->delete_document($doc_id) if defined $doc_id;
226 warn "failed to remove message <$mid>: $@\n";
232 sub term_generator { # write-only
235 my $tg = $self->{term_generator};
238 $tg = Search::Xapian::TermGenerator->new;
239 $tg->set_stemmer($self->stemmer);
241 $self->{term_generator} = $tg;
244 # increments last_thread_id counter
245 # returns a 64-bit integer represented as a hex string
248 my $db = $self->{xdb};
249 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
251 $db->set_metadata('last_thread_id', ++$last_thread_id);
257 my ($self, $smsg, $old_tid) = @_;
258 my $doc = $smsg->{doc};
259 my $mid = $smsg->mid;
260 my $mime = $smsg->mime;
261 my $hdr = $mime->header_obj;
262 my $refs = $hdr->header_raw('References');
263 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
264 if (my $irt = $hdr->header_raw('In-Reply-To')) {
265 # last References should be $irt
266 # we will de-dupe later
267 push @refs, mid_clean($irt);
272 my %uniq = ($mid => 1);
273 my @orig_refs = @refs;
276 # prevent circular references via References: here:
277 foreach my $ref (@orig_refs) {
278 if (length($ref) > MAX_MID_SIZE) {
279 warn "References: <$ref> too long, ignoring\n";
287 $smsg->{references} = '<'.join('> <', @refs).'>';
289 # first ref *should* be the thread root,
290 # but we can never trust clients to do the right thing
291 my $ref = shift @refs;
292 $tid = $self->_resolve_mid_to_tid($ref);
293 $self->merge_threads($tid, $old_tid) if defined $old_tid;
295 # the rest of the refs should point to this tid:
296 foreach $ref (@refs) {
297 my $ptid = $self->_resolve_mid_to_tid($ref);
298 merge_threads($self, $tid, $ptid);
301 $tid = $self->next_thread_id;
303 $doc->add_term(xpfx('thread') . $tid);
307 my ($self, $mime, $bytes, $num, $blob) = @_;
308 $self->add_message($mime, $bytes, $num, $blob);
312 my ($self, $mime) = @_;
313 my $mid = eval { mid_clean(mid_mime($mime)) };
314 $self->remove_message($mid) if defined $mid;
318 my ($self, $mime) = @_;
319 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
323 my ($self, $mime) = @_;
324 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
328 my ($self, $mime, $bytes, $blob) = @_;
329 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
330 index_blob($self, $mime, $bytes, $num, $blob);
334 my ($self, $mime) = @_;
335 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
336 unindex_blob($self, $mime);
340 my ($self, $mime, $bytes, $blob) = @_;
341 my $num = index_mm($self, $mime);
342 index_blob($self, $mime, $bytes, $num, $blob);
346 my ($self, $mime) = @_;
347 unindex_blob($self, $mime);
348 unindex_mm($self, $mime);
352 my ($git, $blob, $sizeref) = @_;
354 my $str = $git->cat_file($blob, $sizeref);
355 # fixup bugs from import:
356 $$str =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
357 Email::MIME->new($str);
363 my ($self, $opts) = @_;
364 with_umask($self, sub { $self->_index_sync($opts) });
368 my ($self, $log, $add_cb, $del_cb, $batch_cb) = @_;
369 my $hex = '[a-f0-9]';
370 my $h40 = $hex .'{40}';
371 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
372 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
373 my $git = $self->{git};
376 my $max = $self->{batch_size}; # may be undef
379 while (defined($line = <$log>)) {
380 if ($line =~ /$addmsg/o) {
382 my $mime = do_cat_mail($git, $blob, \$bytes) or next;
383 $add_cb->($self, $mime, $bytes, $blob);
384 } elsif ($line =~ /$delmsg/o) {
386 my $mime = do_cat_mail($git, $blob) or next;
387 $del_cb->($self, $mime);
388 } elsif ($line =~ /^commit ($h40)/o) {
389 if (defined $max && --$max <= 0) {
390 $max = $self->{batch_size};
391 $batch_cb->($latest, 1);
396 $batch_cb->($latest, 0);
402 require PublicInbox::Msgmap;
403 PublicInbox::Msgmap->new($self->{git_dir}, 1);
408 my ($self, $range) = @_;
409 $self->{git}->popen(qw/log --reverse --no-notes --no-color
410 --raw -r --no-abbrev/, $range);
413 # indexes all unindexed messages
415 my ($self, $opts) = @_;
416 my $tip = $opts->{ref} || 'HEAD';
417 my $reindex = $opts->{reindex};
418 my ($mkey, $last_commit, $lx, $xlog);
419 $self->{git}->batch_prepare;
420 my $xdb = _xdb_acquire($self);
421 $xdb->begin_transaction;
424 $mkey = 'last_commit';
425 $last_commit = $xdb->get_metadata('last_commit');
429 $mkey = undef if $last_commit ne '';
431 $xdb->cancel_transaction;
432 $xdb = _xdb_release($self);
434 # ensure we leak no FDs to "git log"
435 my $range = $lx eq '' ? $tip : "$lx..$tip";
436 $xlog = _git_log($self, $range);
438 $xdb = _xdb_acquire($self);
439 $xdb->begin_transaction;
440 } while ($xdb->get_metadata('last_commit') ne $last_commit);
442 my $mm = _msgmap_init($self);
443 my $dbh = $mm->{dbh} if $mm;
446 my ($commit, $more) = @_;
448 $mm->last_commit($commit) if $commit;
452 $xdb->set_metadata($mkey, $commit) if $mkey && $commit;
453 $xdb->commit_transaction;
454 $xdb = _xdb_release($self);
456 # let another process do some work... <
459 $xdb = _xdb_acquire($self);
460 $xdb->begin_transaction;
462 $dbh->begin_work if $dbh;
468 my $lm = $mm->last_commit || '';
470 # Common case is the indexes are synced,
471 # we only need to run git-log once:
472 rlog($self, $xlog, *index_both, *unindex_both, $cb);
474 # Uncommon case, msgmap and xapian are out-of-sync
475 # do not care for performance (but git is fast :>)
476 # This happens if we have to reindex Xapian since
477 # msgmap is a frozen format and our Xapian format
479 my $r = $lm eq '' ? $tip : "$lm..$tip";
481 # first, ensure msgmap is up-to-date:
482 my $mkey_prev = $mkey;
483 $mkey = undef; # ignore xapian, for now
484 my $mlog = _git_log($self, $r);
486 rlog($self, $mlog, *index_mm, *unindex_mm, $cb);
487 $mm_only = $mlog = undef;
489 # now deal with Xapian
492 rlog($self, $xlog, *index_mm2, *unindex_mm2, $cb);
495 # user didn't install DBD::SQLite and DBI
496 rlog($self, $xlog, *index_blob, *unindex_blob, $cb);
500 # this will create a ghost as necessary
501 sub _resolve_mid_to_tid {
502 my ($self, $mid) = @_;
504 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
509 my ($self, $mid) = @_;
511 my $tid = $self->next_thread_id;
512 my $doc = Search::Xapian::Document->new;
513 $doc->add_term(xpfx('mid') . $mid);
514 $doc->add_term(xpfx('thread') . $tid);
515 $doc->add_term(xpfx('type') . 'ghost');
517 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
518 $self->{xdb}->add_document($doc);
524 my ($self, $winner_tid, $loser_tid) = @_;
525 return if $winner_tid == $loser_tid;
526 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
527 my $thread_pfx = xpfx('thread');
528 my $db = $self->{xdb};
530 for (; $head != $tail; $head->inc) {
531 my $docid = $head->get_docid;
532 my $doc = $db->get_document($docid);
533 $doc->remove_term($thread_pfx . $loser_tid);
534 $doc->add_term($thread_pfx . $winner_tid);
535 $db->replace_document($docid, $doc);
539 sub _read_git_config_perm {
541 my @cmd = qw(config core.sharedRepository);
542 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
545 chomp $perm if defined $perm;
549 sub _git_config_perm {
551 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
552 return PERM_GROUP if (!defined($perm) || $perm eq '');
553 return PERM_UMASK if ($perm eq 'umask');
554 return PERM_GROUP if ($perm eq 'group');
555 if ($perm =~ /\A(?:all|world|everybody)\z/) {
556 return PERM_EVERYBODY;
558 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
559 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
562 return PERM_UMASK if ($i == PERM_UMASK);
563 return PERM_GROUP if ($i == OLD_PERM_GROUP);
564 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
566 if (($i & 0600) != 0600) {
567 die "core.sharedRepository mode invalid: ".
568 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
574 my ($perm) = @_; # _git_config_perm return value
576 return umask if $rv == 0;
578 # set +x bit if +r or +w were set
579 $rv |= 0100 if ($rv & 0600);
580 $rv |= 0010 if ($rv & 0060);
581 $rv |= 0001 if ($rv & 0006);
586 my ($self, $cb) = @_;
587 my $old = umask $self->{umask};
588 my $rv = eval { $cb->() };
596 # order matters for unlocking
597 $_[0]->{xdb} = undef;
598 $_[0]->{lockfh} = undef;