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;
57 binmode $out_w, ':raw' or die "binmode :raw failed: $!";
58 binmode $in_r, ':raw' or die "binmode :raw failed: $!";
62 sub wfail () { die "write to fast-import failed: $!" }
66 my $day = qw(Sun Mon Tue Wed Thu Fri Sat)[$t[6]];
67 my $mon = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec)[$t[4]];
69 sprintf('%s, %2d %s %d %02d:%02d:%02d +0000',
70 $day, $t[3], $mon, $t[5] + 1900, $t[2], $t[1], $t[0]);
75 my $b = $mime->body_raw;
76 $b =~ s/(\r?\n)+\z//s;
80 sub _check_path ($$$$) {
81 my ($r, $w, $tip, $path) = @_;
83 print $w "ls $tip $path\n" or wfail;
85 defined(my $info = <$r>) or die "EOF from fast-import: $!";
86 $info =~ /\Amissing / ? undef : $info;
89 # returns undef on non-existent
90 # ('MISMATCH', msg) on mismatch
91 # (:MARK, msg) on success
93 my ($self, $mime, $msg) = @_; # mime = Email::MIME
95 my $mid = mid_mime($mime);
96 my $path = mid2path($mid);
98 my ($r, $w) = $self->gfi_start;
99 my $tip = $self->{tip};
100 my $info = _check_path($r, $w, $tip, $path) or return ('MISSING',undef);
101 $info =~ m!\A100644 blob ([a-f0-9]{40})\t!s or die "not blob: $info";
104 print $w "cat-blob $blob\n" or wfail;
107 defined $info or die "EOF from fast-import / cat-blob: $!";
108 $info =~ /\A[a-f0-9]{40} blob (\d+)\n\z/ or
109 die "unexpected cat-blob response: $info";
115 $n = read($r, $buf, $left, $offset);
116 defined($n) or die "read cat-blob failed: $!";
117 $n == 0 and die 'fast-export (cat-blob) died';
121 $n = read($r, my $lf, 1);
122 defined($n) or die "read final byte of cat-blob failed: $!";
123 die "bad read on final byte: <$lf>" if $lf ne "\n";
124 my $cur = PublicInbox::MIME->new($buf);
125 my $cur_s = $cur->header('Subject');
126 $cur_s = '' unless defined $cur_s;
127 my $cur_m = $mime->header('Subject');
128 $cur_m = '' unless defined $cur_m;
129 if ($cur_s ne $cur_m || norm_body($cur) ne norm_body($mime)) {
130 return ('MISMATCH', $cur);
133 my $ref = $self->{ref};
134 my $commit = $self->{mark}++;
135 my $parent = $tip =~ /\A:/ ? $tip : undef;
137 print $w "reset $ref\n" or wfail;
139 my $ident = $self->{ident};
142 my $len = length($msg) + 1;
143 print $w "commit $ref\nmark :$commit\n",
144 "author $ident $now\n",
145 "committer $ident $now\n",
146 "data $len\n$msg\n\n",
147 'from ', ($parent ? $parent : $tip), "\n" or wfail;
148 print $w "D $path\n\n" or wfail;
150 (($self->{tip} = ":$commit"), $cur);
153 # returns undef on duplicate
155 my ($self, $mime, $check_cb) = @_; # mime = Email::MIME
157 my $from = $mime->header('From');
158 my ($email) = PublicInbox::Address::emails($from);
159 my ($name) = PublicInbox::Address::names($from);
160 # git gets confused with:
161 # "'A U Thor <u@example.com>' via foo" <foo@example.com>
163 # <CAD0k6qSUYANxbjjbE4jTW4EeVwOYgBD=bXkSu=akiYC_CB7Ffw@mail.gmail.com>
166 my $date = $mime->header('Date');
167 my $subject = $mime->header('Subject');
168 $subject = '(no subject)' unless defined $subject;
169 my $mid = mid_mime($mime);
170 my $path = mid2path($mid);
172 my ($r, $w) = $self->gfi_start;
173 my $tip = $self->{tip};
174 _check_path($r, $w, $tip, $path) and return;
176 # kill potentially confusing/misleading headers
177 $mime->header_set($_) for qw(bytes lines content-length status);
179 $mime = $check_cb->($mime) or return;
182 $mime = $mime->as_string;
183 my $blob = $self->{mark}++;
184 print $w "blob\nmark :$blob\ndata ", length($mime), "\n" or wfail;
185 print $w $mime, "\n" or wfail;
186 my $ref = $self->{ref};
187 my $commit = $self->{mark}++;
188 my $parent = $tip =~ /\A:/ ? $tip : undef;
191 print $w "reset $ref\n" or wfail;
194 utf8::encode($email);
196 utf8::encode($subject);
197 # quiet down wide character warnings:
198 print $w "commit $ref\nmark :$commit\n",
199 "author $name <$email> $date\n",
200 "committer $self->{ident} ", now2822(), "\n" or wfail;
201 print $w "data ", (length($subject) + 1), "\n",
202 $subject, "\n\n" or wfail;
204 print $w 'from ', ($parent ? $parent : $tip), "\n" or wfail;
206 print $w "M 100644 :$blob $path\n\n" or wfail;
208 $self->{tip} = ":$commit";
212 my ($cmd, $env) = @_;
213 my $pid = spawn($cmd, $env, undef);
214 defined $pid or die "spawning ".join(' ', @$cmd)." failed: $!";
215 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
216 $? == 0 or die join(' ', @$cmd) . " failed: $?\n";
221 my $w = delete $self->{out} or return;
222 my $r = delete $self->{in} or die 'BUG: missing {in} when done';
223 print $w "done\n" or wfail;
224 my $pid = delete $self->{pid} or die 'BUG: missing {pid} when done';
225 waitpid($pid, 0) == $pid or die 'fast-import did not finish';
226 $? == 0 or die "fast-import failed: $?";
227 my $nchg = delete $self->{nchg};
229 # for compatibility with existing ssoma installations
230 # we can probably remove this entirely by 2020
231 my $git_dir = $self->{git}->{git_dir};
232 # XXX: change the following scope to: if (-e $index) # in 2018 or so..
233 my @cmd = ('git', "--git-dir=$git_dir");
234 if ($nchg && !$ENV{FAST}) {
235 my $index = "$git_dir/ssoma.index";
236 my $env = { GIT_INDEX_FILE => $index };
237 run_die([@cmd, qw(read-tree -m -v -i), $self->{ref}], $env);
240 run_die([@cmd, 'update-server-info'], undef);
242 require PublicInbox::SearchIdx;
243 my $inbox = $self->{inbox} || $git_dir;
244 my $s = PublicInbox::SearchIdx->new($inbox);
245 $s->index_sync({ ref => $self->{ref} });
248 eval { run_die([@cmd, qw(gc --auto)], undef) };
251 my $lockfh = delete $self->{lockfh} or die "BUG: not locked: $!";
252 flock($lockfh, LOCK_UN) or die "unlock failed: $!";
253 close $lockfh or die "close lock failed: $!";
262 PublicInbox::Import - message importer for public-inbox
271 use PublicInbox::Git;
272 use PublicInbox::Import;
274 chomp(my $git_dir = `git rev-parse --git-dir`);
275 $git_dir or die "GIT_DIR= must be specified\n";
276 my $git = PublicInbox::Git->new($git_dir);
277 my @committer = ('inbox', 'inbox@example.org');
278 my $im = PublicInbox::Import->new($git, @committer);
281 my $message = "From: <u\@example.org>\n".
282 "Subject: test message \n" .
283 "Date: Thu, 01 Jan 1970 00:00:00 +0000\n" .
284 "Message-ID: <m\@example.org>\n".
286 my $parsed = Email::MIME->new($message);
287 my $ret = $im->add($parsed);
290 $parsed->header_obj->header_raw('Message-ID'), "\n";
292 print "imported at mark $ret\n";
296 # to remove a message
297 my $junk = Email::MIME->new($message);
298 my ($mark, $orig) = $im->remove($junk);
299 if ($mark eq 'MISSING') {
301 } elsif ($mark eq 'MISMATCH') {
302 print "Message exists but does not match\n\n",
303 $orig->as_string, "\n",;
305 print "removed at mark $mark\n\n",
306 $orig->as_string, "\n";
312 An importer and remover for public-inboxes which takes L<Email::MIME>
313 messages as input and stores them in a ssoma repository as
314 documented in L<https://ssoma.public-inbox.org/ssoma_repository.txt>,
315 except it does not allow duplicate Message-IDs.
317 It requires L<git(1)> and L<git-fast-import(1)> to be installed.
325 my $im = PublicInbox::Import->new($git, @committer);
327 Initialize a new PublicInbox::Import object.
331 my $parsed = Email::MIME->new($message);
334 Adds a message to to the git repository. This will acquire
335 C<$GIT_DIR/ssoma.lock> and start L<git-fast-import(1)> if necessary.
337 Messages added will not be visible to other processes until L</done>
338 is called, but L</remove> may be called on them.
342 my $junk = Email::MIME->new($message);
343 my ($code, $orig) = $im->remove($junk);
345 Removes a message from the repository. On success, it returns
346 a ':'-prefixed numeric code representing the git-fast-import
347 mark and the original messages as an Email::MIME object.
348 If the message could not be found, the code is "MISSING"
349 and the original message is undef. If there is a mismatch where
350 the "Message-ID" is matched but the subject and body do not match,
351 the returned code is "MISMATCH" and the conflicting message
356 Finalizes the L<git-fast-import(1)> and unlocks the repository.
357 Calling this is required to finalize changes to a repository.
365 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
367 The mail archives are hosted at L<https://public-inbox.org/meta/>
371 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
373 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>