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_compressed/;
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 $flag = Search::Xapian::DB_OPEN;
26 File::Path::mkpath($dir);
27 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
29 my $self = bless { git_dir => $git_dir }, $class;
30 my $umask = _umask_for($self->_git_config_perm);
31 my $old_umask = umask $umask;
32 my $db = eval { Search::Xapian::WritableDatabase->new($dir, $flag) };
41 my ($self, $mime) = @_; # mime = Email::MIME object
42 my $db = $self->{xdb};
45 my $mid_orig = mid_clean($mime->header_obj->header('Message-ID'));
46 my $mid = mid_compressed($mid_orig);
48 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
51 my $smsg = $self->lookup_message($mid);
55 $smsg->ensure_metadata;
56 # convert a ghost to a regular message
57 # it will also clobber any existing regular message
61 my $type = xpfx('type');
63 $doc->remove_term($type . 'ghost');
67 # probably does not exist:
68 eval { $doc->remove_term($type . 'mail') };
69 $doc->add_term($type . 'mail');
71 $smsg = PublicInbox::SearchMsg->new($mime);
73 $doc->add_term(xpfx('mid') . $mid);
76 my $subj = $smsg->subject;
79 $doc->add_term(xpfx('subject') . $subj);
81 my $path = $self->subject_path($subj);
82 $doc->add_term(xpfx('path') . mid_compressed($path));
85 my $from = $smsg->from_name;
86 my $date = $smsg->date;
87 my $ts = Search::Xapian::sortable_serialise($smsg->ts);
88 $doc->add_value(PublicInbox::Search::TS, $ts);
90 my $tg = $self->term_generator;
92 $tg->set_document($doc);
93 $tg->index_text($subj, 1, 'S') if $subj;
94 $tg->increase_termpos;
95 $tg->index_text($subj) if $subj;
96 $tg->increase_termpos;
98 $tg->index_text($smsg->from->format);
99 $tg->increase_termpos;
101 $mime->walk_parts(sub {
103 return if $part->subparts; # walk_parts already recurses
104 my $ct = $part->content_type || $ct_msg;
106 # account for filter bugs...
107 $ct =~ m!\btext/plain\b!i or return;
110 my $body = $part->body;
112 my @lines = split(/\n/, $body);
113 while (defined(my $l = shift @lines)) {
121 $tg->index_text(join("\n", @quot), 0);
123 $tg->increase_termpos;
126 $tg->index_text(join("\n", @orig));
128 $tg->increase_termpos;
133 $doc_id = $smsg->doc_id;
134 $self->link_message($smsg, 0);
135 $doc->set_data($smsg->to_doc_data);
136 $db->replace_document($doc_id, $doc);
138 $self->link_message($smsg, 0);
139 $doc->set_data($smsg->to_doc_data);
140 $doc_id = $db->add_document($doc);
145 warn "failed to index message <$mid_orig>: $@\n";
151 # returns deleted doc_id on success, undef on missing
153 my ($self, $mid_orig) = @_;
154 my $db = $self->{xdb};
156 $mid_orig = mid_clean($mid_orig);
157 my $mid = mid_compressed($mid_orig);
160 $doc_id = $self->find_unique_doc_id('mid', $mid);
161 $db->delete_document($doc_id) if defined $doc_id;
165 warn "failed to remove message <$mid_orig>: $@\n";
171 sub term_generator { # write-only
174 my $tg = $self->{term_generator};
177 $tg = Search::Xapian::TermGenerator->new;
178 $tg->set_stemmer($self->stemmer);
180 $self->{term_generator} = $tg;
183 sub next_doc_id { $_[0]->{xdb}->get_lastdocid + 1 }
185 # increments last_thread_id counter
186 # returns a 64-bit integer represented as a hex string
189 my $db = $self->{xdb};
190 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
192 $db->set_metadata('last_thread_id', ++$last_thread_id);
198 my ($self, $smsg, $is_ghost) = @_;
201 $smsg->ensure_metadata;
203 $self->link_message_to_parents($smsg);
207 sub link_message_to_parents {
208 my ($self, $smsg) = @_;
209 my $doc = $smsg->{doc};
210 my $mid = mid_compressed($smsg->mid);
211 my $mime = $smsg->mime;
212 my $refs = $mime->header_obj->header('References');
213 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
214 my $irt = $mime->header_obj->header('In-Reply-To');
216 $irt = mid_compressed(mid_clean($irt));
218 # maybe some crazies will try to make a circular reference:
222 # last References should be $irt
223 # we will de-dupe later
230 my @crefs = map { mid_compressed($_) } @refs;
231 my %uniq = ($mid => 1);
233 # prevent circular references via References: here:
235 foreach my $ref (@crefs) {
242 $doc->add_term(xpfx('inreplyto') . $irt) if defined $irt;
243 $smsg->{references_sorted} = '<'.join('><', @refs).'>';
245 my $ref_pfx = xpfx('references');
247 # first ref *should* be the thread root,
248 # but we can never trust clients to do the right thing
249 my $ref = shift @refs;
250 $doc->add_term($ref_pfx . $ref);
251 $tid = $self->_resolve_mid_to_tid($ref);
253 # the rest of the refs should point to this tid:
254 foreach $ref (@refs) {
255 $doc->add_term($ref_pfx . $ref);
256 my $ptid = $self->_resolve_mid_to_tid($ref);
258 $self->merge_threads($tid, $ptid);
262 $tid = $self->next_thread_id;
264 $doc->add_term(xpfx('thread') . $tid);
268 my ($self, $git, $blob) = @_;
269 my $mime = do_cat_mail($git, $blob) or return;
270 eval { $self->add_message($mime) };
271 warn "W: index_blob $blob: $@\n" if $@;
275 my ($self, $git, $blob) = @_;
276 my $mime = do_cat_mail($git, $blob) or return;
277 my $mid = $mime->header_obj->header('Message-ID');
278 eval { $self->remove_message($mid) } if defined $mid;
279 warn "W: unindex_blob $blob: $@\n" if $@;
283 my ($git, $blob) = @_;
285 my $str = $git->cat_file($blob);
286 Email::MIME->new($str);
291 # indexes all unindexed messages
293 my ($self, $head) = @_;
294 require PublicInbox::GitCatFile;
295 my $db = $self->{xdb};
296 my $hex = '[a-f0-9]';
297 my $h40 = $hex .'{40}';
298 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
299 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
302 $db->begin_transaction;
304 my $git = PublicInbox::GitCatFile->new($self->{git_dir});
306 my $latest = $db->get_metadata('last_commit');
307 my $range = length $latest ? "$latest..$head" : $head;
310 # get indexed messages
311 my @cmd = ('git', "--git-dir=$self->{git_dir}", "log",
312 qw/--reverse --no-notes --no-color --raw -r
313 --no-abbrev/, $range);
314 my $pid = open(my $log, '-|', @cmd) or
315 die('open` '.join(' ', @cmd) . " pipe failed: $!\n");
317 while (my $line = <$log>) {
318 if ($line =~ /$addmsg/o) {
319 $self->index_blob($git, $1);
320 } elsif ($line =~ /$delmsg/o) {
321 $self->unindex_blob($git, $1);
322 } elsif ($line =~ /^commit ($h40)/o) {
327 $db->set_metadata('last_commit', $latest) if defined $latest;
330 warn "indexing failed: $@\n";
331 $db->cancel_transaction;
333 $db->commit_transaction;
337 # this will create a ghost as necessary
338 sub _resolve_mid_to_tid {
339 my ($self, $mid) = @_;
341 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
346 my ($self, $mid, $tid) = @_;
348 $mid = mid_compressed($mid);
349 $tid = $self->next_thread_id unless defined $tid;
351 my $doc = Search::Xapian::Document->new;
352 $doc->add_term(xpfx('mid') . $mid);
353 $doc->add_term(xpfx('thread') . $tid);
354 $doc->add_term(xpfx('type') . 'ghost');
356 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
357 $self->link_message($smsg, 1);
358 $self->{xdb}->add_document($doc);
364 my ($self, $winner_tid, $loser_tid) = @_;
365 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
366 my $thread_pfx = xpfx('thread');
367 my $db = $self->{xdb};
369 for (; $head != $tail; $head->inc) {
370 my $docid = $head->get_docid;
371 my $doc = $db->get_document($docid);
372 $doc->remove_term($thread_pfx . $loser_tid);
373 $doc->add_term($thread_pfx . $winner_tid);
374 $db->replace_document($docid, $doc);
378 sub _read_git_config_perm {
380 my @cmd = ('git', "--git-dir=$self->{git_dir}",
381 qw(config core.sharedRepository));
382 my $pid = open(my $fh, '-|', @cmd) or
383 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
386 chomp $perm if defined $perm;
390 sub _git_config_perm {
392 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
393 return PERM_GROUP if (!defined($perm) || !length($perm));
394 return PERM_UMASK if ($perm eq 'umask');
395 return PERM_GROUP if ($perm eq 'group');
396 if ($perm =~ /\A(?:all|world|everybody)\z/) {
397 return PERM_EVERYBODY;
399 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
400 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
403 return PERM_UMASK if ($i == PERM_UMASK);
404 return PERM_GROUP if ($i == OLD_PERM_GROUP);
405 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
407 if (($i & 0600) != 0600) {
408 die "core.sharedRepository mode invalid: ".
409 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
415 my ($perm) = @_; # _git_config_perm return value
417 return umask if $rv == 0;
419 # set +x bit if +r or +w were set
420 $rv |= 0100 if ($rv & 0600);
421 $rv |= 0010 if ($rv & 0060);
422 $rv |= 0001 if ($rv & 0006);