1 # Copyright (C) 2016 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # git fast-import-based ssoma-mda MDA replacement
5 # This is only ever run by public-inbox-mda and public-inbox-learn,
6 # not the WWW or NNTP code which only requires read-only access.
7 package PublicInbox::Import;
10 use Fcntl qw(:flock :DEFAULT);
11 use PublicInbox::Spawn qw(spawn);
12 use PublicInbox::MID qw(mid_mime mid2path);
13 use PublicInbox::Address;
16 my ($class, $git, $name, $email, $inbox) = @_;
19 ident => "$name <$email>",
21 ref => 'refs/heads/master',
26 # idempotent start function
30 return ($self->{in}, $self->{out}) if $self->{pid};
32 my ($in_r, $in_w, $out_r, $out_w);
33 pipe($in_r, $in_w) or die "pipe failed: $!";
34 pipe($out_r, $out_w) or die "pipe failed: $!";
35 my $git = $self->{git};
36 my $git_dir = $git->{git_dir};
37 my $lockpath = "$git_dir/ssoma.lock";
38 sysopen(my $lockfh, $lockpath, O_WRONLY|O_CREAT) or
39 die "failed to open lock $lockpath: $!";
41 # wait for other processes to be done
42 flock($lockfh, LOCK_EX) or die "lock failed: $!\n";
44 chomp($self->{tip} = $git->qx(qw(rev-parse --revs-only), $self->{ref}));
46 my @cmd = ('git', "--git-dir=$git_dir", qw(fast-import
47 --quiet --done --date-format=rfc2822));
48 my $rdr = { 0 => fileno($out_r), 1 => fileno($in_w) };
49 my $pid = spawn(\@cmd, undef, $rdr);
50 die "spawn fast-import failed: $!" unless defined $pid;
53 $self->{out} = $out_w;
54 $self->{lockfh} = $lockfh;
60 sub wfail () { die "write to fast-import failed: $!" }
64 my $day = qw(Sun Mon Tue Wed Thu Fri Sat)[$t[6]];
65 my $mon = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec)[$t[4]];
67 sprintf('%s, %2d %s %d %02d:%02d:%02d +0000',
68 $day, $t[3], $mon, $t[5] + 1900, $t[2], $t[1], $t[0]);
73 my $b = $mime->body_raw;
74 $b =~ s/(\r?\n)+\z//s;
78 sub _check_path ($$$$) {
79 my ($r, $w, $tip, $path) = @_;
81 print $w "ls $tip $path\n" or wfail;
83 defined(my $info = <$r>) or die "EOF from fast-import: $!";
84 $info =~ /\Amissing / ? undef : $info;
87 # returns undef on non-existent
88 # ('MISMATCH', msg) on mismatch
89 # (:MARK, msg) on success
91 my ($self, $mime) = @_; # mime = Email::MIME
93 my $mid = mid_mime($mime);
94 my $path = mid2path($mid);
96 my ($r, $w) = $self->gfi_start;
97 my $tip = $self->{tip};
98 my $info = _check_path($r, $w, $tip, $path) or return ('MISSING',undef);
99 $info =~ m!\A100644 blob ([a-f0-9]{40})\t!s or die "not blob: $info";
102 print $w "cat-blob $blob\n" or wfail;
105 defined $info or die "EOF from fast-import / cat-blob: $!";
106 $info =~ /\A[a-f0-9]{40} blob (\d+)\n\z/ or
107 die "unexpected cat-blob response: $info";
113 $n = read($r, $buf, $left, $offset);
114 defined($n) or die "read cat-blob failed: $!";
115 $n == 0 and die 'fast-export (cat-blob) died';
119 $n = read($r, my $lf, 1);
120 defined($n) or die "read final byte of cat-blob failed: $!";
121 die "bad read on final byte: <$lf>" if $lf ne "\n";
122 my $cur = Email::MIME->new($buf);
123 my $cur_s = $cur->header('Subject');
124 $cur_s = '' unless defined $cur_s;
125 my $cur_m = $mime->header('Subject');
126 $cur_m = '' unless defined $cur_m;
127 if ($cur_s ne $cur_m || norm_body($cur) ne norm_body($mime)) {
128 return ('MISMATCH', $cur);
131 my $ref = $self->{ref};
132 my $commit = $self->{mark}++;
133 my $parent = $tip =~ /\A:/ ? $tip : undef;
135 print $w "reset $ref\n" or wfail;
137 my $ident = $self->{ident};
139 print $w "commit $ref\nmark :$commit\n",
140 "author $ident $now\n",
141 "committer $ident $now\n",
143 'from ', ($parent ? $parent : $tip), "\n" or wfail;
144 print $w "D $path\n\n" or wfail;
146 (($self->{tip} = ":$commit"), $cur);
149 # returns undef on duplicate
151 my ($self, $mime, $check_cb) = @_; # mime = Email::MIME
153 my $from = $mime->header('From');
154 my ($email) = PublicInbox::Address::emails($from);
155 my ($name) = PublicInbox::Address::names($from);
156 # git gets confused with:
157 # "'A U Thor <u@example.com>' via foo" <foo@example.com>
159 # <CAD0k6qSUYANxbjjbE4jTW4EeVwOYgBD=bXkSu=akiYC_CB7Ffw@mail.gmail.com>
162 my $date = $mime->header('Date');
163 my $subject = $mime->header('Subject');
164 $subject = '(no subject)' unless defined $subject;
165 my $mid = mid_mime($mime);
166 my $path = mid2path($mid);
168 my ($r, $w) = $self->gfi_start;
169 my $tip = $self->{tip};
170 _check_path($r, $w, $tip, $path) and return;
172 # kill potentially confusing/misleading headers
173 $mime->header_set($_) for qw(bytes lines content-length status);
175 $mime = $check_cb->($mime) or return;
178 $mime = $mime->as_string;
179 my $blob = $self->{mark}++;
180 print $w "blob\nmark :$blob\ndata ", length($mime), "\n" or wfail;
181 print $w $mime, "\n" or wfail;
182 my $ref = $self->{ref};
183 my $commit = $self->{mark}++;
184 my $parent = $tip =~ /\A:/ ? $tip : undef;
187 print $w "reset $ref\n" or wfail;
190 # quiet down wide character warnings:
191 binmode $w, ':utf8' or die "binmode :utf8 failed: $!";
192 print $w "commit $ref\nmark :$commit\n",
193 "author $name <$email> $date\n",
194 "committer $self->{ident} ", now2822(), "\n",
195 "data ", (bytes::length($subject) + 1), "\n",
196 $subject, "\n\n" or wfail;
197 binmode $w, ':raw' or die "binmode :raw failed: $!";
200 print $w 'from ', ($parent ? $parent : $tip), "\n" or wfail;
202 print $w "M 100644 :$blob $path\n\n" or wfail;
204 $self->{tip} = ":$commit";
208 my ($cmd, $env) = @_;
209 my $pid = spawn($cmd, $env, undef);
210 defined $pid or die "spawning ".join(' ', @$cmd)." failed: $!";
211 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
212 $? == 0 or die join(' ', @$cmd) . " failed: $?\n";
217 my $w = delete $self->{out} or return;
218 my $r = delete $self->{in} or die 'BUG: missing {in} when done';
219 print $w "done\n" or wfail;
220 my $pid = delete $self->{pid} or die 'BUG: missing {pid} when done';
221 waitpid($pid, 0) == $pid or die 'fast-import did not finish';
222 $? == 0 or die "fast-import failed: $?";
223 my $nchg = delete $self->{nchg};
225 # for compatibility with existing ssoma installations
226 # we can probably remove this entirely by 2020
227 my $git_dir = $self->{git}->{git_dir};
228 # XXX: change the following scope to: if (-e $index) # in 2018 or so..
229 my @cmd = ('git', "--git-dir=$git_dir");
230 if ($nchg && !$ENV{FAST}) {
231 my $index = "$git_dir/ssoma.index";
232 my $env = { GIT_INDEX_FILE => $index };
233 run_die([@cmd, qw(read-tree -m -v -i), $self->{ref}], $env);
236 run_die([@cmd, 'update-server-info'], undef);
238 require PublicInbox::SearchIdx;
239 my $inbox = $self->{inbox} || $git_dir;
240 my $s = PublicInbox::SearchIdx->new($inbox);
241 $s->index_sync({ ref => $self->{ref} });
243 run_die([@cmd, qw(gc --auto)], undef);
246 my $lockfh = delete $self->{lockfh} or die "BUG: not locked: $!";
247 flock($lockfh, LOCK_UN) or die "unlock failed: $!";
248 close $lockfh or die "close lock failed: $!";
257 PublicInbox::Import - message importer for public-inbox
266 use PublicInbox::Git;
267 use PublicInbox::Import;
269 chomp(my $git_dir = `git rev-parse --git-dir`);
270 $git_dir or die "GIT_DIR= must be specified\n";
271 my $git = PublicInbox::Git->new($git_dir);
272 my @committer = ('inbox', 'inbox@example.org');
273 my $im = PublicInbox::Import->new($git, @committer);
276 my $message = "From: <u\@example.org>\n".
277 "Subject: test message \n" .
278 "Date: Thu, 01 Jan 1970 00:00:00 +0000\n" .
279 "Message-ID: <m\@example.org>\n".
281 my $parsed = Email::MIME->new($message);
282 my $ret = $im->add($parsed);
285 $parsed->header_obj->header_raw('Message-ID'), "\n";
287 print "imported at mark $ret\n";
291 # to remove a message
292 my $junk = Email::MIME->new($message);
293 my ($mark, $orig) = $im->remove($junk);
294 if ($mark eq 'MISSING') {
296 } elsif ($mark eq 'MISMATCH') {
297 print "Message exists but does not match\n\n",
298 $orig->as_string, "\n",;
300 print "removed at mark $mark\n\n",
301 $orig->as_string, "\n";
307 An importer and remover for public-inboxes which takes L<Email::MIME>
308 messages as input and stores them in a ssoma repository as
309 documented in L<https://ssoma.public-inbox.org/ssoma_repository.txt>,
310 except it does not allow duplicate Message-IDs.
312 It requires L<git(1)> and L<git-fast-import(1)> to be installed.
320 my $im = PublicInbox::Import->new($git, @committer);
322 Initialize a new PublicInbox::Import object.
326 my $parsed = Email::MIME->new($message);
329 Adds a message to to the git repository. This will acquire
330 C<$GIT_DIR/ssoma.lock> and start L<git-fast-import(1)> if necessary.
332 Messages added will not be visible to other processes until L</done>
333 is called, but L</remove> may be called on them.
337 my $junk = Email::MIME->new($message);
338 my ($code, $orig) = $im->remove($junk);
340 Removes a message from the repository. On success, it returns
341 a ':'-prefixed numeric code representing the git-fast-import
342 mark and the original messages as an Email::MIME object.
343 If the message could not be found, the code is "MISSING"
344 and the original message is undef. If there is a mismatch where
345 the "Message-ID" is matched but the subject and body do not match,
346 the returned code is "MISMATCH" and the conflicting message
351 Finalizes the L<git-fast-import(1)> and unlocks the repository.
352 Calling this is required to finalize changes to a repository.
360 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
362 The mail archives are hosted at L<https://public-inbox.org/meta/>
366 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
368 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>