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(mid_clean mid_mime);
15 use PublicInbox::ContentId qw(content_id);
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...
65 my $mid = mid_clean(mid_mime($mime));
66 my $num = $self->{skel}->{mm}->mid_insert($mid);
67 if (!defined($num)) { # mid is already known
68 $self->done; # ensure all subprocesses are done writing
70 my $existing = $self->lookup_content($mime);
71 warn "<$mid> resent\n" if $existing;
72 return if $existing; # easy, don't store duplicates
74 # reuse NNTP article number?
75 warn "<$mid> reused for mismatched content\n";
77 $num = $self->{skel}->{mm}->num_for($mid);
80 my $im = $self->importer;
81 my $cmt = $im->add($mime);
82 $cmt = $im->get_mark($cmt);
83 my $oid = $im->{last_object_id};
84 my ($len, $msgref) = @{$im->{last_object}};
86 my $nparts = $self->{partitions};
87 my $part = $num % $nparts;
88 my $idx = $self->idx_part($part);
89 $idx->index_raw($len, $msgref, $num, $oid);
90 my $n = $self->{transact_bytes} += $len;
91 if ($n > (PublicInbox::SearchIdx::BATCH_BYTES * $nparts)) {
99 my ($self, $part) = @_;
100 $self->{idx_parts}->[$part];
106 return if $self->{idx_parts};
107 my $ibx = $self->{-inbox};
109 # do not leak read-only FDs to child processes, we only have these
110 # FDs for duplicate detection so they should not be
111 # frequently activated.
112 delete $ibx->{$_} foreach (qw(git mm search));
114 # first time initialization, first we create the skeleton pipe:
115 my $skel = $self->{skel} = PublicInbox::SearchIdxSkeleton->new($self);
117 # need to create all parts before initializing msgmap FD
118 my $max = $self->{partitions} - 1;
119 my $idx = $self->{idx_parts} = [];
120 for my $i (0..$max) {
121 push @$idx, PublicInbox::SearchIdxPart->new($self, $i, $skel);
124 # Now that all subprocesses are up, we can open the FD for SQLite:
125 $skel->_msgmap_init->{dbh}->begin_work;
129 my ($self, $mime, $msg) = @_;
130 my $existing = $self->lookup_content($mime) or return;
132 # don't touch ghosts or already junked messages
133 return unless $existing->type eq 'mail';
135 # always write removals to the current (latest) git repo since
136 # we process chronologically
137 my $im = $self->importer;
138 my ($cmt, undef) = $im->remove($mime, $msg);
139 $cmt = $im->get_mark($cmt);
140 $self->unindex_msg($existing, $cmt);
145 my $im = delete $self->{im};
146 $im->done if $im; # PublicInbox::Import::done
147 $self->searchidx_checkpoint(0);
152 my $im = $self->{im};
153 $im->checkpoint if $im; # PublicInbox::Import::checkpoint
154 $self->searchidx_checkpoint(1);
157 sub searchidx_checkpoint {
158 my ($self, $more) = @_;
160 # order matters, we can only close {skel} after all partitions
161 # are done because the partitions also write to {skel}
162 if (my $parts = $self->{idx_parts}) {
163 foreach my $idx (@$parts) {
164 $idx->remote_commit; # propagates commit to skel
165 $idx->remote_close unless $more;
167 delete $self->{idx_parts} unless $more;
170 if (my $skel = $self->{skel}) {
171 my $dbh = $skel->{mm}->{dbh};
176 $skel->remote_commit; # XXX should be unnecessary...
178 delete $self->{skel};
181 $self->{transact_bytes} = 0;
185 my ($self, $new) = @_;
186 my $pfx = "$self->{-inbox}->{mainrepo}/git";
187 my $git_dir = "$pfx/$new.git";
188 die "$git_dir exists\n" if -e $git_dir;
189 my @cmd = (qw(git init --bare -q), $git_dir);
190 PublicInbox::Import::run_die(\@cmd);
191 @cmd = (qw/git config/, "--file=$git_dir/config",
192 'repack.writeBitmaps', 'true');
193 PublicInbox::Import::run_die(\@cmd);
195 my $all = "$self->{-inbox}->{mainrepo}/all.git";
197 @cmd = (qw(git init --bare -q), $all);
198 PublicInbox::Import::run_die(\@cmd);
201 my $alt = "$all/objects/info/alternates";
202 my $new_obj_dir = "../../git/$new.git/objects";
205 open(my $fh, '<', $alt) or die "open < $alt: $!\n";
206 %alts = map { chomp; $_ => 1 } (<$fh>);
208 return $git_dir if $alts{$new_obj_dir};
209 open my $fh, '>>', $alt or die "open >> $alt: $!\n";
210 print $fh "$new_obj_dir\n" or die "print >> $alt: $!\n";
211 close $fh or die "close $alt: $!\n";
217 my $im = $self->{im};
219 if ($im->{bytes_added} < $self->{rotate_bytes}) {
224 $self->searchidx_checkpoint(1);
226 my $git_dir = $self->git_init(++$self->{max_git});
227 my $git = PublicInbox::Git->new($git_dir);
228 return $self->import_init($git, 0);
234 my $pfx = "$self->{-inbox}->{mainrepo}/git";
236 foreach my $git_dir (glob("$pfx/*.git")) {
237 $git_dir =~ m!/(\d+)\.git\z! or next;
245 if (defined $latest) {
246 my $git = PublicInbox::Git->new($latest);
247 my $packed_bytes = $git->packed_bytes;
248 if ($packed_bytes >= $self->{rotate_bytes}) {
251 $self->{max_git} = $max;
252 return $self->import_init($git, $packed_bytes);
255 $self->{max_git} = $new;
256 $latest = $self->git_init($new);
257 $self->import_init(PublicInbox::Git->new($latest), 0);
261 my ($self, $git, $packed_bytes) = @_;
262 my $im = PublicInbox::Import->new($git, undef, undef, $self->{-inbox});
263 $im->{bytes_added} = int($packed_bytes / $PACKING_FACTOR);
264 $im->{want_object_id} = 1;
265 $im->{ssoma_lock} = 0;
266 $im->{path_type} = 'v2';
271 my ($self, $mime) = @_;
272 my $ibx = $self->{-inbox};
274 my $srch = $ibx->search;
275 my $cid = content_id($mime);
277 my $mid = mid_mime($mime);
278 $srch->each_smsg_by_mid($mid, sub {
281 my $msg = $ibx->msg_by_smsg($smsg);
282 if (!defined($msg)) {
283 warn "broken smsg for $mid\n";
286 my $cur = PublicInbox::MIME->new($msg);
287 if (content_id($cur) eq $cid) {
288 $smsg->{mime} = $cur;
290 return 0; # break out of loop
299 if (my $parts = $self->{idx_parts}) {
300 $_->atfork_child foreach @$parts;
302 if (my $im = $self->{im}) {