1 # Copyright (C) 2018 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # This interface wraps and mimics PublicInbox::Import
5 package PublicInbox::V2Writable;
8 use Fcntl qw(:flock :DEFAULT);
9 use PublicInbox::SearchIdxPart;
10 use PublicInbox::SearchIdxSkeleton;
11 use PublicInbox::MIME;
13 use PublicInbox::Import;
14 use PublicInbox::MID qw(mids);
15 use PublicInbox::ContentId qw(content_id content_digest);
16 use PublicInbox::Inbox;
18 # an estimate of the post-packed size to the raw uncompressed size
19 my $PACKING_FACTOR = 0.4;
21 # assume 2 cores if GNU nproc(1) is not available
22 my $NPROC = int($ENV{NPROC} || `nproc 2>/dev/null` || 2);
25 my ($class, $v2ibx, $creat) = @_;
26 my $dir = $v2ibx->{mainrepo} or die "no mainrepo in inbox\n";
30 File::Path::mkpath($dir);
32 die "$dir does not exist\n";
37 im => undef, # PublicInbox::Import
38 xap_rw => undef, # PublicInbox::V2SearchIdx
42 # limit each repo to 1GB or so
43 rotate_bytes => int((1024 * 1024 * 1024) / $PACKING_FACTOR),
48 # returns undef on duplicate or spam
49 # mimics Import::add and wraps it for v2
51 my ($self, $mime, $check_cb) = @_;
55 $mime = $check_cb->($mime) or return;
58 # All pipes (> $^F) known to Perl 5.6+ have FD_CLOEXEC set,
59 # as does SQLite 3.4.1+ (released in 2007-07-20), and
60 # Xapian 1.3.2+ (released 2015-03-15).
61 # For the most part, we can spawn git-fast-import without
62 # leaking FDs to it...
66 my $num = num_for($self, $mime, \$mid0);
67 defined $num or return; # duplicate
68 defined $mid0 or die "BUG: $mid0 undefined\n";
69 my $im = $self->importer;
70 my $cmt = $im->add($mime);
71 $cmt = $im->get_mark($cmt);
72 my $oid = $im->{last_object_id};
73 my ($len, $msgref) = @{$im->{last_object}};
75 my $nparts = $self->{partitions};
76 my $part = $num % $nparts;
77 my $idx = $self->idx_part($part);
78 $idx->index_raw($len, $msgref, $num, $oid, $mid0);
79 my $n = $self->{transact_bytes} += $len;
80 if ($n > (PublicInbox::SearchIdx::BATCH_BYTES * $nparts)) {
88 my ($self, $mime, $mid0) = @_;
89 my $mids = mids($mime->header_obj);
92 my $num = $self->{skel}->{mm}->mid_insert($mid);
93 if (defined $num) { # common case
98 # crap, Message-ID is already known, hope somebody just resent:
99 $self->done; # write barrier, clears $self->{skel}
100 foreach my $m (@$mids) {
101 # read-only lookup now safe to do after above barrier
102 my $existing = $self->lookup_content($mime, $m);
104 warn "<$m> resent\n";
105 return; # easy, don't store duplicates
110 warn "<$mid> reused for mismatched content\n";
113 # try the rest of the mids
114 foreach my $i (1..$#$mids) {
116 $num = $self->{skel}->{mm}->mid_insert($m);
118 warn "alternative <$m> for <$mid> found\n";
124 # none of the existing Message-IDs are good, generate a new one:
125 num_for_harder($self, $mime, $mid0);
129 my ($self, $mime, $mid0) = @_;
131 my $hdr = $mime->header_obj;
132 my $dig = content_digest($mime);
133 $$mid0 = $dig->clone->hexdigest . '@localhost';
134 my $num = $self->{skel}->{mm}->mid_insert($$mid0);
135 unless (defined $num) {
136 # it's hard to spoof the last Received: header
137 my @recvd = $hdr->header_raw('Received');
138 $dig->add("Received: $_") foreach (@recvd);
139 $$mid0 = $dig->clone->hexdigest . '@localhost';
140 $num = $self->{skel}->{mm}->mid_insert($$mid0);
142 # fall back to a random Message-ID and give up determinism:
143 until (defined($num)) {
145 $$mid0 = $dig->clone->hexdigest . '@localhost';
146 warn "using random Message-ID <$$mid0> as fallback\n";
147 $num = $self->{skel}->{mm}->mid_insert($$mid0);
150 my @cur = $hdr->header_raw('Message-Id');
151 $hdr->header_set('Message-Id', "<$$mid0>", @cur);
156 my ($self, $part) = @_;
157 $self->{idx_parts}->[$part];
163 return if $self->{idx_parts};
164 my $ibx = $self->{-inbox};
166 # do not leak read-only FDs to child processes, we only have these
167 # FDs for duplicate detection so they should not be
168 # frequently activated.
169 delete $ibx->{$_} foreach (qw(git mm search));
171 # first time initialization, first we create the skeleton pipe:
172 my $skel = $self->{skel} = PublicInbox::SearchIdxSkeleton->new($self);
174 # need to create all parts before initializing msgmap FD
175 my $max = $self->{partitions} - 1;
176 my $idx = $self->{idx_parts} = [];
177 for my $i (0..$max) {
178 push @$idx, PublicInbox::SearchIdxPart->new($self, $i, $skel);
181 # Now that all subprocesses are up, we can open the FD for SQLite:
182 $skel->_msgmap_init->{dbh}->begin_work;
186 my ($self, $mime, $msg) = @_;
187 my $existing = $self->lookup_content($mime) or return;
189 # don't touch ghosts or already junked messages
190 return unless $existing->type eq 'mail';
192 # always write removals to the current (latest) git repo since
193 # we process chronologically
194 my $im = $self->importer;
195 my ($cmt, undef) = $im->remove($mime, $msg);
196 $cmt = $im->get_mark($cmt);
197 $self->unindex_msg($existing, $cmt);
202 my $im = delete $self->{im};
203 $im->done if $im; # PublicInbox::Import::done
204 $self->searchidx_checkpoint(0);
209 my $im = $self->{im};
210 $im->checkpoint if $im; # PublicInbox::Import::checkpoint
211 $self->searchidx_checkpoint(1);
214 sub searchidx_checkpoint {
215 my ($self, $more) = @_;
217 # order matters, we can only close {skel} after all partitions
218 # are done because the partitions also write to {skel}
219 if (my $parts = $self->{idx_parts}) {
220 foreach my $idx (@$parts) {
221 $idx->remote_commit; # propagates commit to skel
222 $idx->remote_close unless $more;
224 delete $self->{idx_parts} unless $more;
227 if (my $skel = $self->{skel}) {
228 my $dbh = $skel->{mm}->{dbh};
233 $skel->remote_commit; # XXX should be unnecessary...
235 delete $self->{skel};
238 $self->{transact_bytes} = 0;
242 my ($self, $new) = @_;
243 my $pfx = "$self->{-inbox}->{mainrepo}/git";
244 my $git_dir = "$pfx/$new.git";
245 die "$git_dir exists\n" if -e $git_dir;
246 my @cmd = (qw(git init --bare -q), $git_dir);
247 PublicInbox::Import::run_die(\@cmd);
248 @cmd = (qw/git config/, "--file=$git_dir/config",
249 'repack.writeBitmaps', 'true');
250 PublicInbox::Import::run_die(\@cmd);
252 my $all = "$self->{-inbox}->{mainrepo}/all.git";
254 @cmd = (qw(git init --bare -q), $all);
255 PublicInbox::Import::run_die(\@cmd);
258 my $alt = "$all/objects/info/alternates";
259 my $new_obj_dir = "../../git/$new.git/objects";
262 open(my $fh, '<', $alt) or die "open < $alt: $!\n";
263 %alts = map { chomp; $_ => 1 } (<$fh>);
265 return $git_dir if $alts{$new_obj_dir};
266 open my $fh, '>>', $alt or die "open >> $alt: $!\n";
267 print $fh "$new_obj_dir\n" or die "print >> $alt: $!\n";
268 close $fh or die "close $alt: $!\n";
274 my $im = $self->{im};
276 if ($im->{bytes_added} < $self->{rotate_bytes}) {
281 $self->searchidx_checkpoint(1);
283 my $git_dir = $self->git_init(++$self->{max_git});
284 my $git = PublicInbox::Git->new($git_dir);
285 return $self->import_init($git, 0);
291 my $pfx = "$self->{-inbox}->{mainrepo}/git";
293 foreach my $git_dir (glob("$pfx/*.git")) {
294 $git_dir =~ m!/(\d+)\.git\z! or next;
302 if (defined $latest) {
303 my $git = PublicInbox::Git->new($latest);
304 my $packed_bytes = $git->packed_bytes;
305 if ($packed_bytes >= $self->{rotate_bytes}) {
308 $self->{max_git} = $max;
309 return $self->import_init($git, $packed_bytes);
312 $self->{max_git} = $new;
313 $latest = $self->git_init($new);
314 $self->import_init(PublicInbox::Git->new($latest), 0);
318 my ($self, $git, $packed_bytes) = @_;
319 my $im = PublicInbox::Import->new($git, undef, undef, $self->{-inbox});
320 $im->{bytes_added} = int($packed_bytes / $PACKING_FACTOR);
321 $im->{want_object_id} = 1;
322 $im->{ssoma_lock} = 0;
323 $im->{path_type} = 'v2';
328 my ($self, $mime, $mid) = @_;
329 my $ibx = $self->{-inbox};
331 my $srch = $ibx->search;
332 my $cid = content_id($mime);
334 $srch->each_smsg_by_mid($mid, sub {
337 my $msg = $ibx->msg_by_smsg($smsg);
338 if (!defined($msg)) {
339 warn "broken smsg for $mid\n";
342 my $cur = PublicInbox::MIME->new($msg);
343 if (content_id($cur) eq $cid) {
344 $smsg->{mime} = $cur;
346 return 0; # break out of loop
355 if (my $parts = $self->{idx_parts}) {
356 $_->atfork_child foreach @$parts;
358 if (my $im = $self->{im}) {