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);
15 my ($class, $git, $name, $email, $inbox) = @_;
18 ident => "$name <$email>",
20 ref => 'refs/heads/master',
25 # idempotent start function
29 return ($self->{in}, $self->{out}) if $self->{pid};
31 my ($in_r, $in_w, $out_r, $out_w);
32 pipe($in_r, $in_w) or die "pipe failed: $!";
33 pipe($out_r, $out_w) or die "pipe failed: $!";
34 my $git = $self->{git};
35 my $git_dir = $git->{git_dir};
36 my $lockpath = "$git_dir/ssoma.lock";
37 sysopen(my $lockfh, $lockpath, O_WRONLY|O_CREAT) or
38 die "failed to open lock $lockpath: $!";
40 # wait for other processes to be done
41 flock($lockfh, LOCK_EX) or die "lock failed: $!\n";
43 chomp($self->{tip} = $git->qx(qw(rev-parse --revs-only), $self->{ref}));
45 my @cmd = ('git', "--git-dir=$git_dir", qw(fast-import
46 --quiet --done --date-format=rfc2822));
47 my $rdr = { 0 => fileno($out_r), 1 => fileno($in_w) };
48 my $pid = spawn(\@cmd, undef, $rdr);
49 die "spawn fast-import failed: $!" unless defined $pid;
52 $self->{out} = $out_w;
53 $self->{lockfh} = $lockfh;
59 sub wfail () { die "write to fast-import failed: $!" }
63 my $day = qw(Sun Mon Tue Wed Thu Fri Sat)[$t[6]];
64 my $mon = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec)[$t[4]];
66 sprintf('%s, %2d %s %d %02d:%02d:%02d +0000',
67 $day, $t[3], $mon, $t[5] + 1900, $t[2], $t[1], $t[0]);
72 my $b = $mime->body_raw;
73 $b =~ s/(\r?\n)+\z//s;
77 # returns undef on non-existent
78 # ('MISMATCH', msg) on mismatch
79 # (:MARK, msg) on success
81 my ($self, $mime) = @_; # mime = Email::MIME
83 my $mid = mid_mime($mime);
84 my $path = mid2path($mid);
86 my ($r, $w) = $self->gfi_start;
87 my $tip = $self->{tip};
88 return ('MISSING', undef) if $tip eq '';
90 print $w "ls $tip $path\n" or wfail;
93 defined $check or die "EOF from fast-import / ls: $!";
94 return ('MISSING', undef) if $check =~ /\Amissing /;
95 $check =~ m!\A100644 blob ([a-f0-9]{40})\t!s or die "not blob: $check";
97 print $w "cat-blob $blob\n" or wfail;
99 defined $check or die "EOF from fast-import / cat-blob: $!";
100 $check =~ /\A[a-f0-9]{40} blob (\d+)\n\z/ or
101 die "unexpected cat-blob response: $check";
107 $n = read($r, $buf, $left, $offset);
108 defined($n) or die "read cat-blob failed: $!";
109 $n == 0 and die 'fast-export (cat-blob) died';
113 $n = read($r, my $lf, 1);
114 defined($n) or die "read final byte of cat-blob failed: $!";
115 die "bad read on final byte: <$lf>" if $lf ne "\n";
116 my $cur = Email::MIME->new($buf);
117 my $cur_s = $cur->header('Subject');
118 $cur_s = '' unless defined $cur_s;
119 my $cur_m = $mime->header('Subject');
120 $cur_m = '' unless defined $cur_m;
121 if ($cur_s ne $cur_m || norm_body($cur) ne norm_body($mime)) {
122 return ('MISMATCH', $cur);
125 my $ref = $self->{ref};
126 my $commit = $self->{mark}++;
127 my $parent = $tip =~ /\A:/ ? $tip : undef;
129 print $w "reset $ref\n" or wfail;
131 my $ident = $self->{ident};
133 print $w "commit $ref\nmark :$commit\n",
134 "author $ident $now\n",
135 "committer $ident $now\n",
137 'from ', ($parent ? $parent : $tip), "\n" or wfail;
138 print $w "D $path\n\n" or wfail;
140 (($self->{tip} = ":$commit"), $cur);
143 # returns undef on duplicate
145 my ($self, $mime, $check_cb) = @_; # mime = Email::MIME
147 my $from = $mime->header('From');
148 my ($email) = ($from =~ /([^<\s]+\@[^>\s]+)/g);
150 $name =~ s/\s*\S+\@\S+\s*\z//;
151 # git gets confused with:
152 # "'A U Thor <u@example.com>' via foo" <foo@example.com>
154 # <CAD0k6qSUYANxbjjbE4jTW4EeVwOYgBD=bXkSu=akiYC_CB7Ffw@mail.gmail.com>
155 $name =~ tr/<>// and $name = $email;
157 my $date = $mime->header('Date');
158 my $subject = $mime->header('Subject');
159 $subject = '(no subject)' unless defined $subject;
160 my $mid = mid_mime($mime);
161 my $path = mid2path($mid);
163 my ($r, $w) = $self->gfi_start;
164 my $tip = $self->{tip};
166 print $w "ls $tip $path\n" or wfail;
169 defined $check or die "EOF from fast-import: $!";
170 return unless $check =~ /\Amissing /;
173 # kill potentially confusing/misleading headers
174 $mime->header_set($_) for qw(bytes lines content-length status);
176 $mime = $check_cb->($mime) or return;
179 $mime = $mime->as_string;
180 my $blob = $self->{mark}++;
181 print $w "blob\nmark :$blob\ndata ", length($mime), "\n" or wfail;
182 print $w $mime, "\n" or wfail;
183 my $ref = $self->{ref};
184 my $commit = $self->{mark}++;
185 my $parent = $tip =~ /\A:/ ? $tip : undef;
188 print $w "reset $ref\n" or wfail;
191 # quiet down wide character warnings:
192 binmode $w, ':utf8' or die "binmode :utf8 failed: $!";
193 print $w "commit $ref\nmark :$commit\n",
194 "author $name <$email> $date\n",
195 "committer $self->{ident} ", now2822(), "\n",
196 "data ", (bytes::length($subject) + 1), "\n",
197 $subject, "\n\n" or wfail;
198 binmode $w, ':raw' or die "binmode :raw failed: $!";
201 print $w 'from ', ($parent ? $parent : $tip), "\n" or wfail;
203 print $w "M 100644 :$blob $path\n\n" or wfail;
205 $self->{tip} = ":$commit";
210 my $w = delete $self->{out} or return;
211 my $r = delete $self->{in} or die 'BUG: missing {in} when done';
212 print $w "done\n" or wfail;
213 my $pid = delete $self->{pid} or die 'BUG: missing {pid} when done';
214 waitpid($pid, 0) == $pid or die 'fast-import did not finish';
215 $? == 0 or die "fast-import failed: $?";
216 my $nchg = delete $self->{nchg};
218 # for compatibility with existing ssoma installations
219 # we can probably remove this entirely by 2020
220 my $git_dir = $self->{git}->{git_dir};
221 # XXX: change the following scope to: if (-e $index) # in 2018 or so..
222 my @cmd = ('git', "--git-dir=$git_dir");
223 if ($nchg && !$ENV{FAST}) {
224 my $index = "$git_dir/ssoma.index";
225 my $env = { GIT_INDEX_FILE => $index };
226 my @rt = (@cmd, qw(read-tree -m -v -i), $self->{ref});
227 $pid = spawn(\@rt, $env, undef);
228 defined $pid or die "spawn read-tree failed: $!";
229 waitpid($pid, 0) == $pid or die 'read-tree did not finish';
230 $? == 0 or die "failed to update $git_dir/ssoma.index: $?\n";
233 $pid = spawn([@cmd, 'update-server-info'], undef, undef);
234 defined $pid or die "spawn update-server-info failed: $!\n";
235 waitpid($pid, 0) == $pid or
236 die 'update-server-info did not finish';
237 $? == 0 or die "failed to update-server-info: $?\n";
240 require PublicInbox::SearchIdx;
241 my $inbox = $self->{inbox} || $git_dir;
242 my $s = PublicInbox::SearchIdx->new($inbox);
243 $s->index_sync({ ref => $self->{ref} });
247 my $lockfh = delete $self->{lockfh} or die "BUG: not locked: $!";
248 flock($lockfh, LOCK_UN) or die "unlock failed: $!";
249 close $lockfh or die "close lock failed: $!";
258 PublicInbox::Import - message importer for public-inbox
267 use PublicInbox::Git;
268 use PublicInbox::Import;
270 chomp(my $git_dir = `git rev-parse --git-dir`);
271 $git_dir or die "GIT_DIR= must be specified\n";
272 my $git = PublicInbox::Git->new($git_dir);
273 my @committer = ('inbox', 'inbox@example.org');
274 my $im = PublicInbox::Import->new($git, @committer);
277 my $message = "From: <u\@example.org>\n".
278 "Subject: test message \n" .
279 "Date: Thu, 01 Jan 1970 00:00:00 +0000\n" .
280 "Message-ID: <m\@example.org>\n".
282 my $parsed = Email::MIME->new($message);
283 my $ret = $im->add($parsed);
286 $parsed->header_obj->header_raw('Message-ID'), "\n";
288 print "imported at mark $ret\n";
292 # to remove a message
293 my $junk = Email::MIME->new($message);
294 my ($mark, $orig) = $im->remove($junk);
295 if ($mark eq 'MISSING') {
297 } elsif ($mark eq 'MISMATCH') {
298 print "Message exists but does not match\n\n",
299 $orig->as_string, "\n",;
301 print "removed at mark $mark\n\n",
302 $orig->as_string, "\n";
308 An importer and remover for public-inboxes which takes L<Email::MIME>
309 messages as input and stores them in a ssoma repository as
310 documented in L<https://ssoma.public-inbox.org/ssoma_repository.txt>,
311 except it does not allow duplicate Message-IDs.
313 It requires L<git(1)> and L<git-fast-import(1)> to be installed.
321 my $im = PublicInbox::Import->new($git, @committer);
323 Initialize a new PublicInbox::Import object.
327 my $parsed = Email::MIME->new($message);
330 Adds a message to to the git repository. This will acquire
331 C<$GIT_DIR/ssoma.lock> and start L<git-fast-import(1)> if necessary.
333 Messages added will not be visible to other processes until L</done>
334 is called, but L</remove> may be called on them.
338 my $junk = Email::MIME->new($message);
339 my ($code, $orig) = $im->remove($junk);
341 Removes a message from the repository. On success, it returns
342 a ':'-prefixed numeric code representing the git-fast-import
343 mark and the original messages as an Email::MIME object.
344 If the message could not be found, the code is "MISSING"
345 and the original message is undef. If there is a mismatch where
346 the "Message-ID" is matched but the subject and body do not match,
347 the returned code is "MISMATCH" and the conflicting message
352 Finalizes the L<git-fast-import(1)> and unlocks the repository.
353 Calling this is required to finalize changes to a repository.
361 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
363 The mail archives are hosted at L<https://public-inbox.org/meta/>
367 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
369 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>