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 $mid = mid_clean($mime->header('Message-ID'));
46 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
49 my $smsg = $self->lookup_message($mid);
53 $smsg->ensure_metadata;
54 # convert a ghost to a regular message
55 # it will also clobber any existing regular message
59 my $type = xpfx('type');
61 $doc->remove_term($type . 'ghost');
65 # probably does not exist:
66 eval { $doc->remove_term($type . 'mail') };
67 $doc->add_term($type . 'mail');
69 $smsg = PublicInbox::SearchMsg->new($mime);
71 $doc->add_term(xpfx('mid') . $mid);
74 my $subj = $smsg->subject;
77 $doc->add_term(xpfx('subject') . $subj);
79 my $path = $self->subject_path($subj);
80 $doc->add_term(xpfx('path') . mid_compress($path));
83 my $ts = Search::Xapian::sortable_serialise($smsg->ts);
84 $doc->add_value(PublicInbox::Search::TS, $ts);
86 my $tg = $self->term_generator;
88 $tg->set_document($doc);
89 $tg->index_text($subj, 1, 'S') if $subj;
90 $tg->increase_termpos;
91 $tg->index_text($subj) if $subj;
92 $tg->increase_termpos;
94 $tg->index_text($smsg->from->format);
95 $tg->increase_termpos;
97 $mime->walk_parts(sub {
99 return if $part->subparts; # walk_parts already recurses
100 my $ct = $part->content_type || $ct_msg;
102 # account for filter bugs...
103 $ct =~ m!\btext/plain\b!i or return;
106 my $body = $part->body;
108 my @lines = split(/\n/, $body);
109 while (defined(my $l = shift @lines)) {
117 $tg->index_text(join("\n", @quot), 0);
119 $tg->increase_termpos;
122 $tg->index_text(join("\n", @orig));
124 $tg->increase_termpos;
129 $doc_id = $smsg->doc_id;
130 $self->link_message($smsg, 0);
131 $doc->set_data($smsg->to_doc_data);
132 $db->replace_document($doc_id, $doc);
134 $self->link_message($smsg, 0);
135 $doc->set_data($smsg->to_doc_data);
136 $doc_id = $db->add_document($doc);
141 warn "failed to index message <$mid>: $@\n";
147 # returns deleted doc_id on success, undef on missing
149 my ($self, $mid) = @_;
150 my $db = $self->{xdb};
152 $mid = mid_clean($mid);
155 $doc_id = $self->find_unique_doc_id('mid', $mid);
156 $db->delete_document($doc_id) if defined $doc_id;
160 warn "failed to remove message <$mid>: $@\n";
166 sub term_generator { # write-only
169 my $tg = $self->{term_generator};
172 $tg = Search::Xapian::TermGenerator->new;
173 $tg->set_stemmer($self->stemmer);
175 $self->{term_generator} = $tg;
178 # increments last_thread_id counter
179 # returns a 64-bit integer represented as a hex string
182 my $db = $self->{xdb};
183 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
185 $db->set_metadata('last_thread_id', ++$last_thread_id);
191 my ($self, $smsg, $is_ghost) = @_;
194 $smsg->ensure_metadata;
196 $self->link_message_to_parents($smsg);
200 sub link_message_to_parents {
201 my ($self, $smsg) = @_;
202 my $doc = $smsg->{doc};
203 my $mid = $smsg->mid;
204 my $mime = $smsg->mime;
205 my $refs = $mime->header('References');
206 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
207 if (my $irt = $mime->header('In-Reply-To')) {
208 # last References should be $irt
209 # we will de-dupe later
210 push @refs, mid_clean($irt);
215 my %uniq = ($mid => 1);
216 my @orig_refs = @refs;
219 # prevent circular references via References: here:
220 foreach my $ref (@orig_refs) {
227 $smsg->{references_sorted} = '<'.join('><', @refs).'>';
229 # first ref *should* be the thread root,
230 # but we can never trust clients to do the right thing
231 my $ref = shift @refs;
232 $tid = $self->_resolve_mid_to_tid($ref);
234 # the rest of the refs should point to this tid:
235 foreach $ref (@refs) {
236 my $ptid = $self->_resolve_mid_to_tid($ref);
238 $self->merge_threads($tid, $ptid);
242 $tid = $self->next_thread_id;
244 $doc->add_term(xpfx('thread') . $tid);
248 my ($self, $git, $mime) = @_;
249 $self->add_message($mime);
253 my ($self, $git, $mime) = @_;
254 my $mid = mid_clean($mime->header('Message-ID'));
255 $self->remove_message($mid) if defined $mid;
259 my ($self, $git, $mime) = @_;
260 $self->{mm}->mid_insert(mid_clean($mime->header('Message-ID')));
264 my ($self, $git, $mime) = @_;
265 $self->{mm}->mid_delete(mid_clean($mime->header('Message-ID')));
269 my ($self, $git, $mime) = @_;
270 index_blob($self, $git, $mime);
271 index_mm($self, $git, $mime);
275 my ($self, $git, $mime) = @_;
276 unindex_blob($self, $git, $mime);
277 unindex_mm($self, $git, $mime);
281 my ($git, $blob) = @_;
283 my $str = $git->cat_file($blob);
284 Email::MIME->new($str);
290 my ($self, $head) = @_;
291 $self->with_umask(sub { $self->_index_sync($head) });
295 my ($self, $range, $add_cb, $del_cb) = @_;
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}$!;
300 my $git_dir = $self->{git_dir};
301 require PublicInbox::GitCatFile;
302 my $git = PublicInbox::GitCatFile->new($git_dir);
303 my @cmd = ('git', "--git-dir=$git_dir", "log",
304 qw/--reverse --no-notes --no-color --raw -r --no-abbrev/,
307 my $pid = open(my $log, '-|', @cmd) or
308 die('open` '.join(' ', @cmd) . " pipe failed: $!\n");
309 while (my $line = <$log>) {
310 if ($line =~ /$addmsg/o) {
311 my $mime = do_cat_mail($git, $1) or next;
312 $add_cb->($self, $git, $mime);
313 } elsif ($line =~ /$delmsg/o) {
314 my $mime = do_cat_mail($git, $1) or next;
315 $del_cb->($self, $git, $mime);
316 } elsif ($line =~ /^commit ($h40)/o) {
324 # indexes all unindexed messages
326 my ($self, $head) = @_;
327 my $db = $self->{xdb};
329 my $mm = $self->{mm} = eval {
330 require PublicInbox::Msgmap;
331 PublicInbox::Msgmap->new($self->{git_dir}, 1);
334 $db->begin_transaction;
335 my $lx = $db->get_metadata('last_commit');
336 my $range = $lx eq '' ? $head : "$lx..$head";
338 $mm->{dbh}->begin_work;
339 my $lm = $mm->last_commit || '';
341 # Common case is the indexes are synced,
342 # we only need to run git-log once:
343 $lx = $self->rlog($range, *index_both, *unindex_both);
346 $db->set_metadata('last_commit', $lx);
347 $mm->last_commit($lx);
350 # dumb case, msgmap and xapian are out-of-sync
351 # do not care for performance:
352 my $r = $lm eq '' ? $head : "$lm..$head";
353 $lm = $self->rlog($r, *index_mm, *unindex_mm);
355 $mm->last_commit($lm) if defined $lm;
360 # user didn't install DBD::SQLite and DBI
362 $lx = $self->rlog($range, *index_blob, *unindex_blob);
363 $db->set_metadata('last_commit', $lx) if defined $lx;
366 $db->cancel_transaction;
367 $mm->{dbh}->rollback if $mm;
369 $db->commit_transaction;
373 # this will create a ghost as necessary
374 sub _resolve_mid_to_tid {
375 my ($self, $mid) = @_;
377 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
382 my ($self, $mid, $tid) = @_;
384 $tid = $self->next_thread_id unless defined $tid;
386 my $doc = Search::Xapian::Document->new;
387 $doc->add_term(xpfx('mid') . $mid);
388 $doc->add_term(xpfx('thread') . $tid);
389 $doc->add_term(xpfx('type') . 'ghost');
391 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
392 $self->link_message($smsg, 1);
393 $self->{xdb}->add_document($doc);
399 my ($self, $winner_tid, $loser_tid) = @_;
400 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
401 my $thread_pfx = xpfx('thread');
402 my $db = $self->{xdb};
404 for (; $head != $tail; $head->inc) {
405 my $docid = $head->get_docid;
406 my $doc = $db->get_document($docid);
407 $doc->remove_term($thread_pfx . $loser_tid);
408 $doc->add_term($thread_pfx . $winner_tid);
409 $db->replace_document($docid, $doc);
413 sub _read_git_config_perm {
415 my @cmd = ('git', "--git-dir=$self->{git_dir}",
416 qw(config core.sharedRepository));
417 my $pid = open(my $fh, '-|', @cmd) or
418 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
421 chomp $perm if defined $perm;
425 sub _git_config_perm {
427 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
428 return PERM_GROUP if (!defined($perm) || $perm eq '');
429 return PERM_UMASK if ($perm eq 'umask');
430 return PERM_GROUP if ($perm eq 'group');
431 if ($perm =~ /\A(?:all|world|everybody)\z/) {
432 return PERM_EVERYBODY;
434 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
435 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
438 return PERM_UMASK if ($i == PERM_UMASK);
439 return PERM_GROUP if ($i == OLD_PERM_GROUP);
440 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
442 if (($i & 0600) != 0600) {
443 die "core.sharedRepository mode invalid: ".
444 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
450 my ($perm) = @_; # _git_config_perm return value
452 return umask if $rv == 0;
454 # set +x bit if +r or +w were set
455 $rv |= 0100 if ($rv & 0600);
456 $rv |= 0010 if ($rv & 0060);
457 $rv |= 0001 if ($rv & 0006);
462 my ($self, $cb) = @_;
463 my $old = umask $self->{umask};
464 my $rv = eval { $cb->() };