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,
31 # XXX temporary hack...
32 my $xap_ver = ((Search::Xapian::major_version << 16) |
33 (Search::Xapian::minor_version << 8 ) |
34 Search::Xapian::revision());
35 our $XAP_LOCK_BROKEN = $xap_ver >= 0x010216; # >= 1.2.22
38 my ($class, $git_dir, $writable) = @_;
39 my $dir = PublicInbox::Search->xdir($git_dir);
40 require Search::Xapian::WritableDatabase;
41 my $flag = Search::Xapian::DB_OPEN;
42 my $self = bless { git_dir => $git_dir }, $class;
43 my $perm = $self->_git_config_perm;
44 my $umask = _umask_for($perm);
45 $self->{umask} = $umask;
46 $self->{lock_path} = "$git_dir/ssoma.lock";
47 $self->{git} = PublicInbox::Git->new($git_dir);
48 $self->{xdb} = $self->with_umask(sub {
51 File::Path::mkpath($dir);
52 $self->{batch_size} = 100 unless $XAP_LOCK_BROKEN;
53 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
56 Search::Xapian::WritableDatabase->new($dir, $flag);
63 my $xdb = delete $self->{xdb};
64 $xdb->commit_transaction;
69 my ($self, $more) = @_;
70 my $dir = PublicInbox::Search->xdir($self->{git_dir});
71 my $flag = Search::Xapian::DB_OPEN;
72 my $xdb = Search::Xapian::WritableDatabase->new($dir, $flag);
73 $xdb->begin_transaction if $more;
79 sysopen(my $lockfh, $self->{lock_path}, O_WRONLY|O_CREAT) or
80 die "failed to open lock $self->{lock_path}: $!\n";
81 flock($lockfh, LOCK_EX) or die "lock failed: $!\n";
82 $self->{lockfh} = $lockfh;
87 my $lockfh = delete $self->{lockfh};
88 flock($lockfh, LOCK_UN) or die "unlock failed: $!\n";
89 close $lockfh or die "close failed: $!\n";
93 my ($doc, $col, $num) = @_;
94 $num = Search::Xapian::sortable_serialise($num);
95 $doc->add_value($col, $num);
99 my ($self, $mime, $bytes, $num, $blob) = @_; # mime = Email::MIME object
100 my $db = $self->{xdb};
102 my ($doc_id, $old_tid);
103 my $mid = mid_clean(mid_mime($mime));
104 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
107 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
108 my $smsg = $self->lookup_message($mid);
110 # convert a ghost to a regular message
111 # it will also clobber any existing regular message
112 $doc_id = $smsg->doc_id;
113 $old_tid = $smsg->thread_id;
115 $smsg = PublicInbox::SearchMsg->new($mime);
116 my $doc = $smsg->{doc};
117 $doc->add_term(xpfx('mid') . $mid);
119 my $subj = $smsg->subject;
121 my $path = $self->subject_path($subj);
122 $doc->add_term(xpfx('path') . id_compress($path));
125 add_val($doc, &PublicInbox::Search::TS, $smsg->ts);
128 add_val($doc, &PublicInbox::Search::NUM, $num);
131 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
133 add_val($doc, &PublicInbox::Search::LINES,
134 $mime->body_raw =~ tr!\n!\n!);
136 my $tg = $self->term_generator;
138 $tg->set_document($doc);
139 $tg->index_text($subj, 1, 'S') if $subj;
140 $tg->increase_termpos;
141 $tg->index_text($subj) if $subj;
142 $tg->increase_termpos;
144 $tg->index_text($smsg->from);
145 $tg->increase_termpos;
147 msg_iter($mime, sub {
148 my ($part, $depth, @idx) = @{$_[0]};
149 my $ct = $part->content_type || $ct_msg;
151 # account for filter bugs...
152 $ct =~ m!\btext/plain\b!i or return;
155 my $body = $part->body;
157 my @lines = split(/\n/, $body);
158 while (defined(my $l = shift @lines)) {
166 $tg->index_text(join("\n", @quot), 0);
168 $tg->increase_termpos;
171 $tg->index_text(join("\n", @orig));
173 $tg->increase_termpos;
177 link_message($self, $smsg, $old_tid);
178 $doc->set_data($smsg->to_doc_data($blob));
179 if (defined $doc_id) {
180 $db->replace_document($doc_id, $doc);
182 $doc_id = $db->add_document($doc);
187 warn "failed to index message <$mid>: $@\n";
193 # returns deleted doc_id on success, undef on missing
195 my ($self, $mid) = @_;
196 my $db = $self->{xdb};
198 $mid = mid_clean($mid);
201 $doc_id = $self->find_unique_doc_id('mid', $mid);
202 $db->delete_document($doc_id) if defined $doc_id;
206 warn "failed to remove message <$mid>: $@\n";
212 sub term_generator { # write-only
215 my $tg = $self->{term_generator};
218 $tg = Search::Xapian::TermGenerator->new;
219 $tg->set_stemmer($self->stemmer);
221 $self->{term_generator} = $tg;
224 # increments last_thread_id counter
225 # returns a 64-bit integer represented as a hex string
228 my $db = $self->{xdb};
229 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
231 $db->set_metadata('last_thread_id', ++$last_thread_id);
237 my ($self, $smsg, $old_tid) = @_;
238 my $doc = $smsg->{doc};
239 my $mid = $smsg->mid;
240 my $mime = $smsg->mime;
241 my $hdr = $mime->header_obj;
242 my $refs = $hdr->header_raw('References');
243 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
244 if (my $irt = $hdr->header_raw('In-Reply-To')) {
245 # last References should be $irt
246 # we will de-dupe later
247 push @refs, mid_clean($irt);
252 my %uniq = ($mid => 1);
253 my @orig_refs = @refs;
256 # prevent circular references via References: here:
257 foreach my $ref (@orig_refs) {
258 if (length($ref) > MAX_MID_SIZE) {
259 warn "References: <$ref> too long, ignoring\n";
267 $smsg->{references} = '<'.join('> <', @refs).'>';
269 # first ref *should* be the thread root,
270 # but we can never trust clients to do the right thing
271 my $ref = shift @refs;
272 $tid = $self->_resolve_mid_to_tid($ref);
273 $self->merge_threads($tid, $old_tid) if defined $old_tid;
275 # the rest of the refs should point to this tid:
276 foreach $ref (@refs) {
277 my $ptid = $self->_resolve_mid_to_tid($ref);
278 merge_threads($self, $tid, $ptid);
281 $tid = $self->next_thread_id;
283 $doc->add_term(xpfx('thread') . $tid);
287 my ($self, $mime, $bytes, $num, $blob) = @_;
288 $self->add_message($mime, $bytes, $num, $blob);
292 my ($self, $mime) = @_;
293 my $mid = eval { mid_clean(mid_mime($mime)) };
294 $self->remove_message($mid) if defined $mid;
298 my ($self, $mime) = @_;
299 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
303 my ($self, $mime) = @_;
304 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
308 my ($self, $mime, $bytes, $blob) = @_;
309 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
310 index_blob($self, $mime, $bytes, $num, $blob);
314 my ($self, $mime) = @_;
315 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
316 unindex_blob($self, $mime);
320 my ($self, $mime, $bytes, $blob) = @_;
321 my $num = index_mm($self, $mime);
322 index_blob($self, $mime, $bytes, $num, $blob);
326 my ($self, $mime) = @_;
327 unindex_blob($self, $mime);
328 unindex_mm($self, $mime);
332 my ($git, $blob, $sizeref) = @_;
334 my $str = $git->cat_file($blob, $sizeref);
335 # fixup bugs from import:
336 $$str =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
337 Email::MIME->new($str);
343 my ($self, $opts) = @_;
344 with_umask($self, sub { $self->_index_sync($opts) });
348 my ($self, $range, $add_cb, $del_cb, $batch_cb) = @_;
349 my $hex = '[a-f0-9]';
350 my $h40 = $hex .'{40}';
351 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
352 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
353 my $git = $self->{git};
354 my $log = $git->popen(qw/log --reverse --no-notes --no-color
355 --raw -r --no-abbrev/, $range);
358 my $max = $self->{batch_size}; # may be undef
361 while (defined($line = <$log>)) {
362 if ($line =~ /$addmsg/o) {
364 my $mime = do_cat_mail($git, $blob, \$bytes) or next;
365 $add_cb->($self, $mime, $bytes, $blob);
366 } elsif ($line =~ /$delmsg/o) {
368 my $mime = do_cat_mail($git, $blob) or next;
369 $del_cb->($self, $mime);
370 } elsif ($line =~ /^commit ($h40)/o) {
371 if (defined $max && --$max <= 0) {
372 $max = $self->{batch_size};
373 $batch_cb->($latest, 1);
378 $batch_cb->($latest, 0);
381 # indexes all unindexed messages
383 my ($self, $opts) = @_;
384 my $tip = $opts->{ref} || 'HEAD';
385 my $mm = $self->{mm} = eval {
386 require PublicInbox::Msgmap;
387 PublicInbox::Msgmap->new($self->{git_dir}, 1);
389 my $xdb = $self->{xdb};
390 $xdb->begin_transaction;
391 my $reindex = $opts->{reindex};
392 my $mkey = 'last_commit';
393 my $last_commit = $xdb->get_metadata($mkey);
394 my $lx = $last_commit;
397 $mkey = undef if $last_commit ne '';
401 my ($commit, $more) = @_;
402 $xdb->set_metadata($mkey, $commit) if $mkey && $commit;
404 $mm->last_commit($commit) if $commit;
407 if ($XAP_LOCK_BROKEN) {
408 $xdb->commit_transaction if !$more;
412 _lock_release($self);
414 # let another process do some work...
415 if (!$XAP_LOCK_BROKEN) {
416 _lock_acquire($self);
417 $dbh->begin_work if $dbh && $more;
418 $xdb = _xdb_acquire($self, $more);
422 my $range = $lx eq '' ? $tip : "$lx..$tip";
426 my $lm = $mm->last_commit || '';
428 # Common case is the indexes are synced,
429 # we only need to run git-log once:
430 rlog($self, $range, *index_both, *unindex_both, $cb);
432 # Uncommon case, msgmap and xapian are out-of-sync
433 # do not care for performance (but git is fast :>)
434 # This happens if we have to reindex Xapian since
435 # msgmap is a frozen format and our Xapian format
437 my $r = $lm eq '' ? $tip : "$lm..$tip";
439 # first, ensure msgmap is up-to-date:
440 my $mkey_prev = $mkey;
441 $mkey = undef; # ignore xapian, for now
442 rlog($self, $r, *index_mm, *unindex_mm, $cb);
444 # now deal with Xapian
447 rlog($self, $range, *index_mm2, *unindex_mm2, $cb);
450 # user didn't install DBD::SQLite and DBI
451 rlog($self, $range, *index_blob, *unindex_blob, $cb);
455 # this will create a ghost as necessary
456 sub _resolve_mid_to_tid {
457 my ($self, $mid) = @_;
459 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
464 my ($self, $mid) = @_;
466 my $tid = $self->next_thread_id;
467 my $doc = Search::Xapian::Document->new;
468 $doc->add_term(xpfx('mid') . $mid);
469 $doc->add_term(xpfx('thread') . $tid);
470 $doc->add_term(xpfx('type') . 'ghost');
472 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
473 $self->{xdb}->add_document($doc);
479 my ($self, $winner_tid, $loser_tid) = @_;
480 return if $winner_tid == $loser_tid;
481 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
482 my $thread_pfx = xpfx('thread');
483 my $db = $self->{xdb};
485 for (; $head != $tail; $head->inc) {
486 my $docid = $head->get_docid;
487 my $doc = $db->get_document($docid);
488 $doc->remove_term($thread_pfx . $loser_tid);
489 $doc->add_term($thread_pfx . $winner_tid);
490 $db->replace_document($docid, $doc);
494 sub _read_git_config_perm {
496 my @cmd = qw(config core.sharedRepository);
497 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
500 chomp $perm if defined $perm;
504 sub _git_config_perm {
506 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
507 return PERM_GROUP if (!defined($perm) || $perm eq '');
508 return PERM_UMASK if ($perm eq 'umask');
509 return PERM_GROUP if ($perm eq 'group');
510 if ($perm =~ /\A(?:all|world|everybody)\z/) {
511 return PERM_EVERYBODY;
513 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
514 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
517 return PERM_UMASK if ($i == PERM_UMASK);
518 return PERM_GROUP if ($i == OLD_PERM_GROUP);
519 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
521 if (($i & 0600) != 0600) {
522 die "core.sharedRepository mode invalid: ".
523 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
529 my ($perm) = @_; # _git_config_perm return value
531 return umask if $rv == 0;
533 # set +x bit if +r or +w were set
534 $rv |= 0100 if ($rv & 0600);
535 $rv |= 0010 if ($rv & 0060);
536 $rv |= 0001 if ($rv & 0006);
541 my ($self, $cb) = @_;
542 my $old = umask $self->{umask};
543 my $rv = eval { $cb->() };
551 # order matters for unlocking
552 $_[0]->{xdb} = undef;
553 $_[0]->{lockfh} = undef;