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 $Email::MIME::ContentType::STRICT_PARAMS = 0;
16 use base qw(PublicInbox::Search);
17 use PublicInbox::MID qw/mid_clean id_compress mid_mime/;
18 use PublicInbox::MsgIter;
19 require PublicInbox::Git;
20 *xpfx = *PublicInbox::Search::xpfx;
22 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
26 OLD_PERM_EVERYBODY => 2,
28 PERM_EVERYBODY => 0664,
32 my ($class, $git_dir, $writable) = @_;
33 my $dir = PublicInbox::Search->xdir($git_dir);
34 require Search::Xapian::WritableDatabase;
35 my $flag = Search::Xapian::DB_OPEN;
36 my $self = bless { git_dir => $git_dir }, $class;
37 my $perm = $self->_git_config_perm;
38 my $umask = _umask_for($perm);
39 $self->{umask} = $umask;
40 $self->{lock_path} = "$git_dir/ssoma.lock";
41 $self->{xdb} = $self->with_umask(sub {
44 File::Path::mkpath($dir);
45 $self->{batch_size} = 100;
46 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
49 Search::Xapian::WritableDatabase->new($dir, $flag);
56 my $xdb = delete $self->{xdb};
57 $xdb->commit_transaction;
65 my $dir = PublicInbox::Search->xdir($self->{git_dir});
66 my $flag = Search::Xapian::DB_OPEN;
67 my $xdb = Search::Xapian::WritableDatabase->new($dir, $flag);
68 $xdb->begin_transaction;
74 sysopen(my $lockfh, $self->{lock_path}, O_WRONLY|O_CREAT) or
75 die "failed to open lock $self->{lock_path}: $!\n";
76 flock($lockfh, LOCK_EX) or die "lock failed: $!\n";
77 $self->{lockfh} = $lockfh;
82 my $lockfh = delete $self->{lockfh};
83 flock($lockfh, LOCK_UN) or die "unlock failed: $!\n";
84 close $lockfh or die "close failed: $!\n";
88 my ($doc, $col, $num) = @_;
89 $num = Search::Xapian::sortable_serialise($num);
90 $doc->add_value($col, $num);
94 my ($self, $mime, $bytes, $num, $blob) = @_; # mime = Email::MIME object
95 my $db = $self->{xdb};
97 my ($doc_id, $old_tid);
98 my $mid = mid_clean(mid_mime($mime));
99 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
102 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
103 my $smsg = $self->lookup_message($mid);
105 # convert a ghost to a regular message
106 # it will also clobber any existing regular message
107 $doc_id = $smsg->doc_id;
108 $old_tid = $smsg->thread_id;
110 $smsg = PublicInbox::SearchMsg->new($mime);
111 my $doc = $smsg->{doc};
112 $doc->add_term(xpfx('mid') . $mid);
114 my $subj = $smsg->subject;
116 my $path = $self->subject_path($subj);
117 $doc->add_term(xpfx('path') . id_compress($path));
120 add_val($doc, &PublicInbox::Search::TS, $smsg->ts);
123 add_val($doc, &PublicInbox::Search::NUM, $num);
126 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
128 add_val($doc, &PublicInbox::Search::LINES,
129 $mime->body_raw =~ tr!\n!\n!);
131 my $tg = $self->term_generator;
133 $tg->set_document($doc);
134 $tg->index_text($subj, 1, 'S') if $subj;
135 $tg->increase_termpos;
136 $tg->index_text($subj) if $subj;
137 $tg->increase_termpos;
139 $tg->index_text($smsg->from);
140 $tg->increase_termpos;
142 msg_iter($mime, sub {
143 my ($part, $depth, @idx) = @{$_[0]};
144 my $ct = $part->content_type || $ct_msg;
146 # account for filter bugs...
147 $ct =~ m!\btext/plain\b!i or return;
150 my $body = $part->body;
152 my @lines = split(/\n/, $body);
153 while (defined(my $l = shift @lines)) {
161 $tg->index_text(join("\n", @quot), 0);
163 $tg->increase_termpos;
166 $tg->index_text(join("\n", @orig));
168 $tg->increase_termpos;
172 link_message($self, $smsg, $old_tid);
173 $doc->set_data($smsg->to_doc_data($blob));
174 if (defined $doc_id) {
175 $db->replace_document($doc_id, $doc);
177 $doc_id = $db->add_document($doc);
182 warn "failed to index message <$mid>: $@\n";
188 # returns deleted doc_id on success, undef on missing
190 my ($self, $mid) = @_;
191 my $db = $self->{xdb};
193 $mid = mid_clean($mid);
196 $doc_id = $self->find_unique_doc_id('mid', $mid);
197 $db->delete_document($doc_id) if defined $doc_id;
201 warn "failed to remove message <$mid>: $@\n";
207 sub term_generator { # write-only
210 my $tg = $self->{term_generator};
213 $tg = Search::Xapian::TermGenerator->new;
214 $tg->set_stemmer($self->stemmer);
216 $self->{term_generator} = $tg;
219 # increments last_thread_id counter
220 # returns a 64-bit integer represented as a hex string
223 my $db = $self->{xdb};
224 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
226 $db->set_metadata('last_thread_id', ++$last_thread_id);
232 my ($self, $smsg, $old_tid) = @_;
233 my $doc = $smsg->{doc};
234 my $mid = $smsg->mid;
235 my $mime = $smsg->mime;
236 my $hdr = $mime->header_obj;
237 my $refs = $hdr->header_raw('References');
238 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
239 if (my $irt = $hdr->header_raw('In-Reply-To')) {
240 # last References should be $irt
241 # we will de-dupe later
242 push @refs, mid_clean($irt);
247 my %uniq = ($mid => 1);
248 my @orig_refs = @refs;
251 # prevent circular references via References: here:
252 foreach my $ref (@orig_refs) {
253 if (length($ref) > MAX_MID_SIZE) {
254 warn "References: <$ref> too long, ignoring\n";
262 $smsg->{references} = '<'.join('> <', @refs).'>';
264 # first ref *should* be the thread root,
265 # but we can never trust clients to do the right thing
266 my $ref = shift @refs;
267 $tid = $self->_resolve_mid_to_tid($ref);
268 $self->merge_threads($tid, $old_tid) if defined $old_tid;
270 # the rest of the refs should point to this tid:
271 foreach $ref (@refs) {
272 my $ptid = $self->_resolve_mid_to_tid($ref);
273 merge_threads($self, $tid, $ptid);
276 $tid = $self->next_thread_id;
278 $doc->add_term(xpfx('thread') . $tid);
282 my ($self, $git, $mime, $bytes, $num, $blob) = @_;
283 $self->add_message($mime, $bytes, $num, $blob);
287 my ($self, $git, $mime) = @_;
288 my $mid = eval { mid_clean(mid_mime($mime)) };
289 $self->remove_message($mid) if defined $mid;
293 my ($self, $git, $mime) = @_;
294 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
298 my ($self, $git, $mime) = @_;
299 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
303 my ($self, $git, $mime, $bytes, $blob) = @_;
304 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
305 index_blob($self, $git, $mime, $bytes, $num, $blob);
309 my ($self, $git, $mime) = @_;
310 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
311 unindex_blob($self, $git, $mime);
315 my ($self, $git, $mime, $bytes, $blob) = @_;
316 my $num = index_mm($self, $git, $mime);
317 index_blob($self, $git, $mime, $bytes, $num, $blob);
321 my ($self, $git, $mime) = @_;
322 unindex_blob($self, $git, $mime);
323 unindex_mm($self, $git, $mime);
327 my ($git, $blob, $sizeref) = @_;
329 my $str = $git->cat_file($blob, $sizeref);
330 # fixup bugs from import:
331 $$str =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
332 Email::MIME->new($str);
338 my ($self, $opts) = @_;
339 with_umask($self, sub { $self->_index_sync($opts) });
343 my ($self, $range, $add_cb, $del_cb, $batch_cb) = @_;
344 my $hex = '[a-f0-9]';
345 my $h40 = $hex .'{40}';
346 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
347 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
348 my $git = PublicInbox::Git->new($self->{git_dir});
349 my $log = $git->popen(qw/log --reverse --no-notes --no-color
350 --raw -r --no-abbrev/, $range);
353 my $max = $self->{batch_size}; # may be undef
356 while (defined($line = <$log>)) {
357 if ($line =~ /$addmsg/o) {
359 my $mime = do_cat_mail($git, $blob, \$bytes) or next;
360 $add_cb->($self, $git, $mime, $bytes, $blob);
361 } elsif ($line =~ /$delmsg/o) {
363 my $mime = do_cat_mail($git, $blob) or next;
364 $del_cb->($self, $git, $mime);
365 } elsif ($line =~ /^commit ($h40)/o) {
366 if (defined $max && --$max <= 0) {
367 $max = $self->{batch_size};
368 $batch_cb->($latest, 1);
373 $batch_cb->($latest, 0);
376 # indexes all unindexed messages
378 my ($self, $opts) = @_;
379 my $tip = $opts->{ref} || 'HEAD';
380 my $mm = $self->{mm} = eval {
381 require PublicInbox::Msgmap;
382 PublicInbox::Msgmap->new($self->{git_dir}, 1);
384 my $xdb = $self->{xdb};
385 $xdb->begin_transaction;
386 my $reindex = $opts->{reindex};
387 my $mkey = 'last_commit';
388 my $last_commit = $xdb->get_metadata($mkey);
389 my $lx = $last_commit;
392 $mkey = undef if $last_commit ne '';
396 my ($commit, $more) = @_;
397 $xdb->set_metadata($mkey, $commit) if $mkey && $commit;
399 $mm->last_commit($commit) if $commit;
403 # let another process do some work...
404 $dbh->begin_work if $dbh && $more;
405 $xdb = _xdb_acquire($self);
408 my $range = $lx eq '' ? $tip : "$lx..$tip";
412 my $lm = $mm->last_commit || '';
414 # Common case is the indexes are synced,
415 # we only need to run git-log once:
416 rlog($self, $range, *index_both, *unindex_both, $cb);
418 # Uncommon case, msgmap and xapian are out-of-sync
419 # do not care for performance (but git is fast :>)
420 # This happens if we have to reindex Xapian since
421 # msgmap is a frozen format and our Xapian format
423 my $r = $lm eq '' ? $tip : "$lm..$tip";
425 # first, ensure msgmap is up-to-date:
426 my $mkey_prev = $mkey;
427 $mkey = undef; # ignore xapian, for now
428 rlog($self, $r, *index_mm, *unindex_mm, $cb);
430 # now deal with Xapian
433 rlog($self, $range, *index_mm2, *unindex_mm2, $cb);
436 # user didn't install DBD::SQLite and DBI
437 rlog($self, $range, *index_blob, *unindex_blob, $cb);
441 # this will create a ghost as necessary
442 sub _resolve_mid_to_tid {
443 my ($self, $mid) = @_;
445 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
450 my ($self, $mid) = @_;
452 my $tid = $self->next_thread_id;
453 my $doc = Search::Xapian::Document->new;
454 $doc->add_term(xpfx('mid') . $mid);
455 $doc->add_term(xpfx('thread') . $tid);
456 $doc->add_term(xpfx('type') . 'ghost');
458 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
459 $self->{xdb}->add_document($doc);
465 my ($self, $winner_tid, $loser_tid) = @_;
466 return if $winner_tid == $loser_tid;
467 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
468 my $thread_pfx = xpfx('thread');
469 my $db = $self->{xdb};
471 for (; $head != $tail; $head->inc) {
472 my $docid = $head->get_docid;
473 my $doc = $db->get_document($docid);
474 $doc->remove_term($thread_pfx . $loser_tid);
475 $doc->add_term($thread_pfx . $winner_tid);
476 $db->replace_document($docid, $doc);
480 sub _read_git_config_perm {
482 my @cmd = qw(config core.sharedRepository);
483 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
486 chomp $perm if defined $perm;
490 sub _git_config_perm {
492 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
493 return PERM_GROUP if (!defined($perm) || $perm eq '');
494 return PERM_UMASK if ($perm eq 'umask');
495 return PERM_GROUP if ($perm eq 'group');
496 if ($perm =~ /\A(?:all|world|everybody)\z/) {
497 return PERM_EVERYBODY;
499 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
500 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
503 return PERM_UMASK if ($i == PERM_UMASK);
504 return PERM_GROUP if ($i == OLD_PERM_GROUP);
505 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
507 if (($i & 0600) != 0600) {
508 die "core.sharedRepository mode invalid: ".
509 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
515 my ($perm) = @_; # _git_config_perm return value
517 return umask if $rv == 0;
519 # set +x bit if +r or +w were set
520 $rv |= 0100 if ($rv & 0600);
521 $rv |= 0010 if ($rv & 0060);
522 $rv |= 0001 if ($rv & 0006);
527 my ($self, $cb) = @_;
528 my $old = umask $self->{umask};
529 my $rv = eval { $cb->() };
537 # order matters for unlocking
538 $_[0]->{xdb} = undef;
539 $_[0]->{lockfh} = undef;