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
4 package PublicInbox::SearchIdx;
7 use base qw(PublicInbox::Search);
8 use PublicInbox::MID qw/mid_clean mid_compress/;
9 *xpfx = *PublicInbox::Search::xpfx;
14 OLD_PERM_EVERYBODY => 2,
16 PERM_EVERYBODY => 0664,
20 my ($class, $git_dir, $writable) = @_;
21 my $dir = $class->xdir($git_dir);
22 require Search::Xapian::WritableDatabase;
23 my $flag = Search::Xapian::DB_OPEN;
24 my $self = bless { git_dir => $git_dir }, $class;
25 my $perm = $self->_git_config_perm;
26 my $umask = _umask_for($perm);
27 $self->{umask} = $umask;
28 $self->{xdb} = $self->with_umask(sub {
31 File::Path::mkpath($dir);
32 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
34 Search::Xapian::WritableDatabase->new($dir, $flag);
40 my ($self, $mime) = @_; # mime = Email::MIME object
41 my $db = $self->{xdb};
44 my $mid_orig = mid_clean($mime->header('Message-ID'));
45 my $mid = mid_compress($mid_orig);
47 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
50 my $smsg = $self->lookup_message($mid);
54 $smsg->ensure_metadata;
55 # convert a ghost to a regular message
56 # it will also clobber any existing regular message
60 my $type = xpfx('type');
62 $doc->remove_term($type . 'ghost');
66 # probably does not exist:
67 eval { $doc->remove_term($type . 'mail') };
68 $doc->add_term($type . 'mail');
70 $smsg = PublicInbox::SearchMsg->new($mime);
72 $doc->add_term(xpfx('mid') . $mid);
75 my $subj = $smsg->subject;
78 $doc->add_term(xpfx('subject') . $subj);
80 my $path = $self->subject_path($subj);
81 $doc->add_term(xpfx('path') . mid_compress($path));
84 my $ts = Search::Xapian::sortable_serialise($smsg->ts);
85 $doc->add_value(PublicInbox::Search::TS, $ts);
87 my $tg = $self->term_generator;
89 $tg->set_document($doc);
90 $tg->index_text($subj, 1, 'S') if $subj;
91 $tg->increase_termpos;
92 $tg->index_text($subj) if $subj;
93 $tg->increase_termpos;
95 $tg->index_text($smsg->from->format);
96 $tg->increase_termpos;
98 $mime->walk_parts(sub {
100 return if $part->subparts; # walk_parts already recurses
101 my $ct = $part->content_type || $ct_msg;
103 # account for filter bugs...
104 $ct =~ m!\btext/plain\b!i or return;
107 my $body = $part->body;
109 my @lines = split(/\n/, $body);
110 while (defined(my $l = shift @lines)) {
118 $tg->index_text(join("\n", @quot), 0);
120 $tg->increase_termpos;
123 $tg->index_text(join("\n", @orig));
125 $tg->increase_termpos;
130 $doc_id = $smsg->doc_id;
131 $self->link_message($smsg, 0);
132 $doc->set_data($smsg->to_doc_data);
133 $db->replace_document($doc_id, $doc);
135 $self->link_message($smsg, 0);
136 $doc->set_data($smsg->to_doc_data);
137 $doc_id = $db->add_document($doc);
142 warn "failed to index message <$mid_orig>: $@\n";
148 # returns deleted doc_id on success, undef on missing
150 my ($self, $mid_orig) = @_;
151 my $db = $self->{xdb};
153 $mid_orig = mid_clean($mid_orig);
154 my $mid = mid_compress($mid_orig);
157 $doc_id = $self->find_unique_doc_id('mid', $mid);
158 $db->delete_document($doc_id) if defined $doc_id;
162 warn "failed to remove message <$mid_orig>: $@\n";
168 sub term_generator { # write-only
171 my $tg = $self->{term_generator};
174 $tg = Search::Xapian::TermGenerator->new;
175 $tg->set_stemmer($self->stemmer);
177 $self->{term_generator} = $tg;
180 sub next_doc_id { $_[0]->{xdb}->get_lastdocid + 1 }
182 # increments last_thread_id counter
183 # returns a 64-bit integer represented as a hex string
186 my $db = $self->{xdb};
187 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
189 $db->set_metadata('last_thread_id', ++$last_thread_id);
195 my ($self, $smsg, $is_ghost) = @_;
198 $smsg->ensure_metadata;
200 $self->link_message_to_parents($smsg);
204 sub link_message_to_parents {
205 my ($self, $smsg) = @_;
206 my $doc = $smsg->{doc};
207 my $mid = mid_compress($smsg->mid);
208 my $mime = $smsg->mime;
209 my $refs = $mime->header('References');
210 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
211 my $irt = $mime->header('In-Reply-To');
213 $irt = mid_compress(mid_clean($irt));
215 # maybe some crazies will try to make a circular reference:
219 # last References should be $irt
220 # we will de-dupe later
227 my @crefs = map { mid_compress($_) } @refs;
228 my %uniq = ($mid => 1);
230 # prevent circular references via References: here:
232 foreach my $ref (@crefs) {
239 $smsg->{references_sorted} = '<'.join('><', @refs).'>';
241 # first ref *should* be the thread root,
242 # but we can never trust clients to do the right thing
243 my $ref = shift @refs;
244 $tid = $self->_resolve_mid_to_tid($ref);
246 # the rest of the refs should point to this tid:
247 foreach $ref (@refs) {
248 my $ptid = $self->_resolve_mid_to_tid($ref);
250 $self->merge_threads($tid, $ptid);
254 $tid = $self->next_thread_id;
256 $doc->add_term(xpfx('thread') . $tid);
260 my ($self, $git, $blob) = @_;
261 my $mime = do_cat_mail($git, $blob) or return;
262 eval { $self->add_message($mime) };
263 warn "W: index_blob $blob: $@\n" if $@;
267 my ($self, $git, $blob) = @_;
268 my $mime = do_cat_mail($git, $blob) or return;
269 my $mid = $mime->header('Message-ID');
270 eval { $self->remove_message($mid) } if defined $mid;
271 warn "W: unindex_blob $blob: $@\n" if $@;
275 my ($git, $blob) = @_;
277 my $str = $git->cat_file($blob);
278 Email::MIME->new($str);
284 my ($self, $head) = @_;
285 $self->with_umask(sub { $self->_index_sync($head) });
288 # indexes all unindexed messages
290 my ($self, $head) = @_;
291 require PublicInbox::GitCatFile;
292 my $db = $self->{xdb};
293 my $hex = '[a-f0-9]';
294 my $h40 = $hex .'{40}';
295 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
296 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
299 $db->begin_transaction;
301 my $git = PublicInbox::GitCatFile->new($self->{git_dir});
303 my $latest = $db->get_metadata('last_commit');
304 my $range = $latest eq '' ? $head : "$latest..$head";
307 # get indexed messages
308 my @cmd = ('git', "--git-dir=$self->{git_dir}", "log",
309 qw/--reverse --no-notes --no-color --raw -r
310 --no-abbrev/, $range);
311 my $pid = open(my $log, '-|', @cmd) or
312 die('open` '.join(' ', @cmd) . " pipe failed: $!\n");
314 while (my $line = <$log>) {
315 if ($line =~ /$addmsg/o) {
316 $self->index_blob($git, $1);
317 } elsif ($line =~ /$delmsg/o) {
318 $self->unindex_blob($git, $1);
319 } elsif ($line =~ /^commit ($h40)/o) {
324 $db->set_metadata('last_commit', $latest) if defined $latest;
327 warn "indexing failed: $@\n";
328 $db->cancel_transaction;
330 $db->commit_transaction;
334 # this will create a ghost as necessary
335 sub _resolve_mid_to_tid {
336 my ($self, $mid) = @_;
338 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
343 my ($self, $mid, $tid) = @_;
345 $mid = mid_compress($mid);
346 $tid = $self->next_thread_id unless defined $tid;
348 my $doc = Search::Xapian::Document->new;
349 $doc->add_term(xpfx('mid') . $mid);
350 $doc->add_term(xpfx('thread') . $tid);
351 $doc->add_term(xpfx('type') . 'ghost');
353 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
354 $self->link_message($smsg, 1);
355 $self->{xdb}->add_document($doc);
361 my ($self, $winner_tid, $loser_tid) = @_;
362 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
363 my $thread_pfx = xpfx('thread');
364 my $db = $self->{xdb};
366 for (; $head != $tail; $head->inc) {
367 my $docid = $head->get_docid;
368 my $doc = $db->get_document($docid);
369 $doc->remove_term($thread_pfx . $loser_tid);
370 $doc->add_term($thread_pfx . $winner_tid);
371 $db->replace_document($docid, $doc);
375 sub _read_git_config_perm {
377 my @cmd = ('git', "--git-dir=$self->{git_dir}",
378 qw(config core.sharedRepository));
379 my $pid = open(my $fh, '-|', @cmd) or
380 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
383 chomp $perm if defined $perm;
387 sub _git_config_perm {
389 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
390 return PERM_GROUP if (!defined($perm) || $perm eq '');
391 return PERM_UMASK if ($perm eq 'umask');
392 return PERM_GROUP if ($perm eq 'group');
393 if ($perm =~ /\A(?:all|world|everybody)\z/) {
394 return PERM_EVERYBODY;
396 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
397 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
400 return PERM_UMASK if ($i == PERM_UMASK);
401 return PERM_GROUP if ($i == OLD_PERM_GROUP);
402 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
404 if (($i & 0600) != 0600) {
405 die "core.sharedRepository mode invalid: ".
406 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
412 my ($perm) = @_; # _git_config_perm return value
414 return umask if $rv == 0;
416 # set +x bit if +r or +w were set
417 $rv |= 0100 if ($rv & 0600);
418 $rv |= 0010 if ($rv & 0060);
419 $rv |= 0001 if ($rv & 0006);
424 my ($self, $cb) = @_;
425 my $old = umask $self->{umask};
426 my $rv = eval { $cb->() };