1 # Copyright (C) all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # "lei add-external --mirror" support (also "public-inbox-clone");
5 package PublicInbox::LeiMirror;
8 use parent qw(PublicInbox::IPC);
9 use IO::Uncompress::Gunzip qw(gunzip $GunzipError);
10 use IO::Compress::Gzip qw(gzip $GzipError);
11 use PublicInbox::Spawn qw(popen_rd spawn run_die);
15 use Fcntl qw(SEEK_SET O_CREAT O_EXCL O_WRONLY);
18 use PublicInbox::Config;
19 use PublicInbox::Inbox;
20 use PublicInbox::LeiCurl;
21 use PublicInbox::OnDestroy;
23 our $LIVE; # pid => callback
25 sub _wq_done_wait { # dwaitpid callback (via wq_eof)
27 my ($mrr, $lei) = @$arg;
28 my $f = "$mrr->{dst}/mirror.done";
30 $lei->child_error($?);
31 } elsif (!$mrr->{dry_run} && !unlink($f)) {
32 warn("unlink($f): $!\n") unless $!{ENOENT};
34 if (!$mrr->{dry_run} && $lei->{cmd} ne 'public-inbox-clone') {
35 # calls _finish_add_external
36 $lei->lazy_cb('add-external', '_finish_'
37 )->($lei, $mrr->{dst});
39 $lei->qerr("# mirrored $mrr->{src} => $mrr->{dst}");
44 # for old installations without manifest.js.gz
47 my $uri = URI->new($self->{src});
48 my $lei = $self->{lei};
49 my $curl = $self->{curl} //= PublicInbox::LeiCurl->new($lei) or return;
50 my $cmd = $curl->for_uri($lei, $uri, '--compressed');
51 my $opt = { 0 => $lei->{0}, 2 => $lei->{2} };
52 my $fh = popen_rd($cmd, undef, $opt);
53 my $html = do { local $/; <$fh> } // die "read(curl $uri): $!";
54 close($fh) or return $lei->child_error($?, "@$cmd failed");
56 # we grep with URL below, we don't want Subject/From headers
57 # making us clone random URLs
58 my @html = split(/<hr>/, $html);
59 my @urls = ($html[-1] =~ m!\bgit clone --mirror ([a-z\+]+://\S+)!g);
60 my $url = $uri->as_string;
61 chop($url) eq '/' or die "BUG: $uri not canonicalized";
63 # since this is for old instances w/o manifest.js.gz, try v1 first
64 return clone_v1($self) if grep(m!\A\Q$url\E/*\z!, @urls);
65 if (my @v2_urls = grep(m!\A\Q$url\E/[0-9]+\z!, @urls)) {
67 my ($n) = (m!/([0-9]+)\z!);
68 $n => [ URI->new($_), '' ]
70 clone_v2($self, \%v2_epochs);
71 reap_live() while keys(%$LIVE);
75 # filter out common URLs served by WWW (e.g /$MSGID/T/)
76 if (@urls && $url =~ s!/+[^/]+\@[^/]+/.*\z!! &&
77 grep(m!\A\Q$url\E/*\z!, @urls)) {
79 E: confused by scraping <$uri>, did you mean <$url>?
83 E: confused by scraping <$uri>, got ambiguous results:
86 die "E: scraping <$uri> revealed nothing\n";
92 $opt->{$_} = $lei->{$_} for (0..2);
93 # we support "-c $key=$val" for arbitrary git config options
94 # e.g.: git -c http.proxy=socks5h://127.0.0.1:9050
95 push(@cmd, '-c', $_) for @{$lei->{opt}->{c} // []};
96 push @cmd, qw(clone --mirror);
97 push @cmd, '-q' if $lei->{opt}->{quiet} ||
98 ($lei->{opt}->{jobs} // 1) > 1;
99 push @cmd, '-v' if $lei->{opt}->{verbose};
100 # XXX any other options to support?
101 # --reference is tricky with multiple epochs...
105 sub ft_rename ($$$) {
106 my ($ft, $dst, $open_mode) = @_;
107 my $fn = $ft->filename;
109 my $mode = @st ? ($st[2] & 07777) : ($open_mode & ~umask);
110 chmod($mode, $ft) or croak "E: chmod $fn: $!";
112 File::Copy::mv($fn, $dst) or croak "E: mv($fn => $ft): $!";
113 $ft->unlink_on_destroy(0);
116 sub _get_txt_start { # non-fatal
117 my ($self, $endpoint, $fini) = @_;
118 my $uri = URI->new($self->{cur_src} // $self->{src});
119 my $lei = $self->{lei};
120 my $path = $uri->path;
121 chop($path) eq '/' or die "BUG: $uri not canonicalized";
122 $uri->path("$path/$endpoint");
123 my $f = (split(m!/!, $endpoint))[-1];
124 my $ft = File::Temp->new(TEMPLATE => "$f-XXXX", TMPDIR => 1);
125 my $opt = { 0 => $lei->{0}, 1 => $lei->{1}, 2 => $lei->{2} };
126 my $cmd = $self->{curl}->for_uri($lei, $uri, qw(--compressed -R -o),
128 my $jobs = $lei->{opt}->{jobs} // 1;
129 reap_live() while keys(%$LIVE) >= $jobs;
130 $lei->qerr("# @$cmd");
131 return if $self->{dry_run};
132 $self->{"-get_txt.$endpoint"} = [ $ft, $cmd, $uri ];
133 $LIVE->{spawn($cmd, undef, $opt)} =
134 [ \&_get_txt_done, $self, $endpoint, $fini ];
137 sub _get_txt_done { # returns true on error (non-fatal), undef on success
138 my ($self, $endpoint) = @_;
139 my ($fh, $cmd, $uri) = @{delete $self->{"-get_txt.$endpoint"}};
141 $? = 0; # don't influence normal lei exit
142 return warn("$uri missing\n") if ($cerr >> 8) == 22;
143 return warn("# @$cmd failed (non-fatal)\n") if $cerr;
144 seek($fh, SEEK_SET, 0) or die "seek: $!";
145 $self->{"mtime.$endpoint"} = (stat($fh))[9];
147 $self->{"txt.$endpoint"} = <$fh>;
151 sub _write_inbox_config {
153 my $buf = delete($self->{'txt._/text/config/raw'}) // return;
154 my $dst = $self->{cur_dst} // $self->{dst};
155 my $f = "$dst/inbox.config.example";
156 open my $fh, '>', $f or die "open($f): $!";
157 print $fh $buf or die "print: $!";
158 chmod(0444 & ~umask, $fh) or die "chmod($f): $!";
159 my $mtime = delete $self->{'mtime._/text/config/raw'};
160 $fh->flush or die "flush($f): $!";
161 if (defined $mtime) {
162 utime($mtime, $mtime, $fh) or die "utime($f): $!";
164 my $cfg = PublicInbox::Config->git_config_dump($f, $self->{lei}->{2});
165 my $ibx = $self->{ibx} = {};
166 for my $sec (grep(/\Apublicinbox\./, @{$cfg->{-section_order}})) {
167 for (qw(address newsgroup nntpmirror)) {
168 $ibx->{$_} = $cfg->{"$sec.$_"};
173 sub set_description ($) {
175 my $dst = $self->{cur_dst} // $self->{dst};
176 my $f = "$dst/description";
177 open my $fh, '+>>', $f or die "open($f): $!";
178 seek($fh, 0, SEEK_SET) or die "seek($f): $!";
179 my $d = do { local $/; <$fh> } // die "read($f): $!";
181 while (defined($d) && ($d =~ m!^\(\$INBOX_DIR/description missing\)! ||
182 $d =~ /^Unnamed repository/ || $d !~ /\S/)) {
183 $d = delete($self->{'txt.description'});
185 $d //= 'mirror of '.($self->{cur_src} // $self->{src})."\n";
186 return if $d eq $orig;
187 seek($fh, 0, SEEK_SET) or die "seek($f): $!";
188 truncate($fh, 0) or die "truncate($f): $!";
189 print $fh $d or die "print($f): $!";
190 close $fh or die "close($f): $!";
193 sub index_cloned_inbox {
194 my ($self, $iv) = @_;
195 my $lei = $self->{lei};
196 eval { set_description($self) };
199 # n.b. public-inbox-clone works w/o (SQLite || Xapian)
200 # lei is useless without Xapian + SQLite
201 if ($lei->{cmd} ne 'public-inbox-clone') {
202 require PublicInbox::InboxWritable;
203 require PublicInbox::Admin;
204 my $ibx = delete($self->{ibx}) // {
205 address => [ 'lei@example.com' ],
208 $ibx->{inboxdir} = $self->{cur_dst} // $self->{dst};
209 PublicInbox::Inbox->new($ibx);
210 PublicInbox::InboxWritable->new($ibx);
212 for my $sw ($lei->index_opt) {
213 my ($k) = ($sw =~ /\A([\w-]+)/);
214 $opt->{$k} = $lei->{opt}->{$k};
216 # force synchronous dwaitpid for v2:
217 local $PublicInbox::DS::in_loop = 0;
218 my $cfg = PublicInbox::Config->new(undef, $lei->{2});
219 my $env = PublicInbox::Admin::index_prepare($opt, $cfg);
220 local %ENV = (%ENV, %$env) if $env;
221 PublicInbox::Admin::progress_prepare($opt, $lei->{2});
222 PublicInbox::Admin::index_inbox($ibx, undef, $opt);
224 return if defined $self->{cur_dst};
225 open my $x, '>', "$self->{dst}/mirror.done"; # for _wq_done_wait
229 my ($lei, $cmd, $opt) = @_;
230 $lei->qerr("# @$cmd");
231 waitpid(spawn($cmd, undef, $opt), 0) // die "waitpid: $!";
233 $? = 0; # don't let it influence normal exit
238 my ($self, $cmd, $opt, $fini) = @_;
239 my $jobs = $self->{lei}->{opt}->{jobs} // 1;
240 reap_live() while keys(%$LIVE) >= $jobs;
241 $self->{lei}->qerr("# @$cmd");
242 return if $self->{dry_run};
243 $LIVE->{spawn($cmd, undef, $opt)} = [ \&reap_clone, $self, $cmd, $fini ]
247 my ($self, $nohang) = @_;
248 my $lei = $self->{lei};
249 my $curl = $self->{curl} //= PublicInbox::LeiCurl->new($lei) or return;
250 my $uri = URI->new($self->{cur_src} // $self->{src});
251 defined($lei->{opt}->{epoch}) and
252 die "$uri is a v1 inbox, --epoch is not supported\n";
253 my $pfx = $curl->torsocks($lei, $uri) or return;
254 my $dst = $self->{cur_dst} // $self->{dst};
255 my $fini = PublicInbox::OnDestroy->new($$, \&v1_done, $self);
256 my $cmd = [ @$pfx, clone_cmd($lei, my $opt = {}), "$uri", $dst ];
257 my $ref = $self->{-ent} ? $self->{-ent}->{reference} : undef;
258 defined($ref) && -e "$self->{dst}$ref" and
259 push @$cmd, '--reference', "$self->{dst}$ref";
260 start_clone($self, $cmd, $opt, $fini);
262 $lei->{opt}->{'inbox-config'} =~ /\A(?:always|v1)\z/s and
263 _get_txt_start($self, '_/text/config/raw', $fini);
264 my $d = $self->{-ent} ? $self->{-ent}->{description} : undef;
265 defined($d) ? ($self->{'txt.description'} = $d) :
266 _get_txt_start($self, 'description', $fini);
268 reap_live() until ($nohang || !keys(%$LIVE)); # for non-manifest clone
271 sub parse_epochs ($$) {
272 my ($opt_epochs, $v2_epochs) = @_; # $epochs "LOW..HIGH"
273 $opt_epochs // return; # undef => all epochs
274 my ($lo, $dotdot, $hi, @extra) = split(/(\.\.)/, $opt_epochs);
275 undef($lo) if ($lo // '') eq '';
276 my $re = qr/\A~?[0-9]+\z/;
277 if (@extra || (($lo // '0') !~ $re) ||
278 (($hi // '0') !~ $re) ||
279 !(grep(defined, $lo, $hi))) {
281 --epoch=$opt_epochs not in the form of `LOW..HIGH', `LOW..', nor `..HIGH'
284 my @n = sort { $a <=> $b } keys %$v2_epochs;
285 for (grep(defined, $lo, $hi)) {
288 "`$_' exceeds maximum available epoch ($n[-1])\n";
290 "`$_' is lower than minimum available epoch ($n[0])\n";
291 } elsif (/\A~([0-9]+)/) {
293 $n[$off] // die "`$_' is out of range\n";
295 } else { die "`$_' not understood\n" }
297 defined($lo) && defined($hi) && $lo > $hi and die
298 "low value (`$lo') exceeds high (`$hi')\n";
299 $lo //= $n[0] if $dotdot;
300 $hi //= $n[-1] if $dotdot;
304 if (defined $v2_epochs->{$_}) {
308 "# epoch $_ is not available (non-fatal, $lo..$hi)\n";
314 sub init_placeholder ($$$) {
315 my ($src, $edst, $ent) = @_;
316 PublicInbox::Import::init_bare($edst);
317 my $f = "$edst/config";
318 open my $fh, '>>', $f or die "open($f): $!";
319 print $fh <<EOM or die "print($f): $!";
322 fetch = +refs/*:refs/*
325 ; This git epoch was created read-only and "public-inbox-fetch"
326 ; will not fetch updates for it unless write permission is added.
327 ; Hint: chmod +w $edst
329 if (defined($ent->{owner})) {
330 print $fh <<EOM or die "print($f): $!";
332 owner = $ent->{owner}
335 close $fh or die "close($f): $!";
336 if (defined $ent->{head}) {
338 open $fh, '>', $f or die "open($f): $!";
339 print $fh $ent->{head}, "\n" or die "print($f): $!";
340 close $fh or die "close($f): $!";
344 sub reap_clone { # async, called via SIGCHLD
345 my ($self, $cmd) = @_;
347 $? = 0; # don't let it influence normal exit
349 kill('TERM', keys %$LIVE);
350 $self->{lei}->child_error($cerr, "@$cmd failed");
354 sub v1_done { # called via OnDestroy
356 return if $self->{dry_run} || !$LIVE;
357 _write_inbox_config($self);
358 my $dst = $self->{cur_dst} // $self->{dst};
359 if (defined(my $o = $self->{-ent} ? $self->{-ent}->{owner} : undef)) {
360 run_die([qw(git config -f), "$dst/config", 'gitweb.owner', $o]);
362 my $o = "$dst/objects";
363 if (open(my $fh, '<', "$o/info/alternates")) {
364 chomp(my @l = <$fh>);
365 for (@l) { $_ = File::Spec->abs2rel($_, $o)."\n" }
366 my $f = File::Temp->new(TEMPLATE => '.XXXX', DIR => "$o/info");
368 $f->flush or die "flush($f): $!";
369 rename($f->filename, "$o/info/alternates") or
370 die "rename($f, $o/info/alternates): $!";
371 $f->unlink_on_destroy(0);
373 write_makefile($dst, 1);
374 index_cloned_inbox($self, 1);
377 sub v2_done { # called via OnDestroy
379 return if $self->{dry_run} || !$LIVE;
380 my $dst = $self->{cur_dst} // $self->{dst};
381 my $lk = bless { lock_path => "$dst/inbox.lock" }, 'PublicInbox::Lock';
382 my $lck = $lk->lock_for_scope($$);
383 _write_inbox_config($self);
384 require PublicInbox::MultiGit;
385 my $mg = PublicInbox::MultiGit->new($dst, 'all.git', 'git');
386 $mg->fill_alternates;
387 for my $i ($mg->git_epochs) { $mg->epoch_cfg_set($i) }
388 my $edst_owner = delete($self->{-owner}) // [];
389 while (@$edst_owner) {
390 my ($edst, $o) = splice(@$edst_owner);
391 run_die [qw(git config -f), "$edst/config", 'gitweb.owner', $o];
393 for my $edst (@{delete($self->{-read_only}) // []}) {
394 my @st = stat($edst) or die "stat($edst): $!";
395 chmod($st[2] & 0555, $edst) or die "chmod(a-w, $edst): $!";
397 write_makefile($dst, 2);
399 index_cloned_inbox($self, 2);
403 my $pid = waitpid(-1, 0) // die "waitpid(-1): $!";
404 if (my $x = delete $LIVE->{$pid}) {
408 warn "reaped unknown PID=$pid ($?)\n";
412 sub clone_v2 ($$;$) {
413 my ($self, $v2_epochs, $m) = @_; # $m => manifest.js.gz hashref
414 my $lei = $self->{lei};
415 my $curl = $self->{curl} //= PublicInbox::LeiCurl->new($lei) or return;
416 my $first_uri = (map { $_->[0] } values %$v2_epochs)[0];
417 my $pfx = $curl->torsocks($lei, $first_uri) or return;
418 my $dst = $self->{cur_dst} // $self->{dst};
419 my $want = parse_epochs($lei->{opt}->{epoch}, $v2_epochs);
420 my $task = $m ? bless { %$self }, __PACKAGE__ : $self;
421 my (@src_edst, @skip);
422 for my $nr (sort { $a <=> $b } keys %$v2_epochs) {
423 my ($uri, $key) = @{$v2_epochs->{$nr}};
424 my $src = $uri->as_string;
426 $src =~ m!/([0-9]+)(?:\.git)?\z! or die <<"";
427 failed to extract epoch number from $src
429 $1 + 0 == $nr or die "BUG: <$uri> miskeyed $1 != $nr";
430 $edst .= "/git/$nr.git";
431 $m->{$key} // die "BUG: `$key' not in manifest.js.gz";
432 if (!$want || $want->{$nr}) {
433 push @src_edst, $src, $edst;
434 my $o = $m->{$key}->{owner};
435 push(@{$task->{-owner}}, $edst, $o) if defined($o);
436 } else { # create a placeholder so users only need to chmod +w
437 init_placeholder($src, $edst, $m->{$key});
438 push @{$task->{-read_only}}, $edst;
442 # filter out the epochs we skipped
443 $self->{-culled_manifest} = 1 if delete(@$m{@skip});
445 (!$self->{dry_run} && !-d $dst) and File::Path::mkpath($dst);
447 require PublicInbox::Lock;
448 my $fini = PublicInbox::OnDestroy->new($$, \&v2_done, $task);
450 $lei->{opt}->{'inbox-config'} =~ /\A(?:always|v2)\z/s and
451 _get_txt_start($task, '_/text/config/raw', $fini);
453 _get_txt_start($task, 'description', $fini);
455 my @cmd = clone_cmd($lei, my $opt = {});
456 while (@src_edst && !$lei->{child_error}) {
457 my $cmd = [ @$pfx, @cmd, splice(@src_edst, 0, 2) ];
458 start_clone($self, $cmd, $opt, $fini);
462 sub decode_manifest ($$$) {
463 my ($fh, $fn, $uri) = @_;
465 my $gz = do { local $/; <$fh> } // die "slurp($fn): $!";
466 gunzip(\$gz => \$js, MultiStream => 1) or
467 die "gunzip($uri): $GunzipError\n";
468 my $m = eval { PublicInbox::Config->json->decode($js) };
469 die "$uri: error decoding `$js': $@\n" if $@;
470 ref($m) eq 'HASH' or die "$uri unknown type: ".ref($m);
474 sub multi_inbox ($$$) {
475 my ($self, $path, $m) = @_;
476 my $incl = $self->{lei}->{opt}->{include};
477 my $excl = $self->{lei}->{opt}->{exclude};
479 # assuming everything not v2 is v1, for now
480 my @v1 = sort grep(!m!.+/git/[0-9]+\.git\z!, keys %$m);
481 my @v2_epochs = sort grep(m!.+/git/[0-9]+\.git\z!, keys %$m);
485 m!\A(/.+)/git/[0-9]+\.git\z! or die "BUG: $_";
486 push @{$v2->{$1}}, $_;
488 my $n = scalar(keys %$v2) + scalar(@v1);
489 my @orig = defined($incl // $excl) ? (keys %$v2, @v1) : ();
491 my $re = '(?:'.join('|', map {
492 $self->{lei}->glob2re($_) // qr/\A\Q$_\E\z/
494 my @gone = delete @$v2{grep(!/$re/, keys %$v2)};
495 delete @$m{map { @$_ } @gone} and $self->{-culled_manifest} = 1;
496 delete @$m{grep(!/$re/, @v1)} and $self->{-culled_manifest} = 1;
497 @v1 = grep(/$re/, @v1);
500 my $re = '(?:'.join('|', map {
501 $self->{lei}->glob2re($_) // qr/\A\Q$_\E\z/
503 my @gone = delete @$v2{grep(/$re/, keys %$v2)};
504 delete @$m{map { @$_ } @gone} and $self->{-culled_manifest} = 1;
505 delete @$m{grep(/$re/, @v1)} and $self->{-culled_manifest} = 1;
506 @v1 = grep(!/$re/, @v1);
508 my $ret; # { v1 => [ ... ], v2 => { "/$inbox_name" => [ epochs ] }}
509 $ret->{v1} = \@v1 if @v1;
510 $ret->{v2} = $v2 if keys %$v2;
511 $ret //= @orig ? "Nothing to clone, available repositories:\n\t".
512 join("\n\t", sort @orig)
513 : "Nothing available to clone\n";
516 # PSGI mount prefixes and manifest.js.gz prefixes don't always align...
518 until (grep(m!\A\Q$$path\E/git/[0-9]+\.git\z!,
519 @v2_epochs) == @v2_epochs) {
520 $$path =~ s!\A(/[^/]+)/!/! or last;
524 while (!defined($m->{$$path}) && $$path =~ s!\A(/[^/]+)/!/!) {
528 ($path_pfx, $n, $ret);
532 my ($self, $todo, $m) = @_;
533 # handle no-dependency repos, first
534 for (@{delete($todo->{''}) // []}) {
536 return if $self->{lei}->{child_error};
538 # resolve references, deepest, first:
539 while (scalar keys %$todo) {
540 for my $x (keys %$todo) {
541 # resolve multi-level references
542 while (defined(my $nxt = $m->{$x}->{reference})) {
543 exists($todo->{$nxt}) or last;
546 my $y = delete $todo->{$x} // next; # already done
549 return if $self->{lei}->{child_error};
551 last; # restart %$todo iteration
556 # FIXME: this gets confused by single inbox instance w/ global manifest.js.gz
559 my $uri = URI->new($self->{src});
560 my $lei = $self->{lei};
561 my $curl = $self->{curl} //= PublicInbox::LeiCurl->new($lei) or return;
562 my $path = $uri->path;
563 chop($path) eq '/' or die "BUG: $uri not canonicalized";
564 $uri->path($path . '/manifest.js.gz');
565 my $ft = File::Temp->new(TEMPLATE => '.manifest-XXXX',
566 UNLINK => 1, TMPDIR => 1, SUFFIX => '.tmp');
567 my $fn = $ft->filename;
568 my $cmd = $curl->for_uri($lei, $uri, '-R', '-o', $fn);
569 my %opt = map { $_ => $lei->{$_} } (0..2);
570 my $cerr = run_reap($lei, $cmd, \%opt);
573 return try_scrape($self) if ($cerr >> 8) == 22; # 404 missing
574 return $lei->child_error($cerr, "@$cmd failed");
576 my $m = eval { decode_manifest($ft, $fn, $uri) };
579 return try_scrape($self);
581 my ($path_pfx, $n, $multi) = multi_inbox($self, \$path, $m);
582 return $lei->child_error(1, $multi) if !ref($multi);
583 my $v2 = delete $multi->{v2};
585 for my $name (sort keys %$v2) {
586 my $epochs = delete $v2->{$name};
587 my %v2_epochs = map {
588 $uri->path($n > 1 ? $path_pfx.$path.$_
590 my ($e) = ("$uri" =~ m!/([0-9]+)\.git\z!);
591 $e // die "no [0-9]+\.git in `$uri'";
592 $e => [ $uri->clone, $_ ];
594 ("$uri" =~ m!\A(.+/)git/[0-9]+\.git\z!) or
595 die "BUG: `$uri' !~ m!/git/[0-9]+.git!";
596 local $self->{cur_src} = $1;
597 local $self->{cur_dst} = $self->{dst};
598 if ($n > 1 && $uri->path =~ m!\A\Q$path_pfx$path\E/(.+)/
599 git/[0-9]+\.git\z!x) {
600 $self->{cur_dst} .= "/$1";
602 index($self->{cur_dst}, "\n") >= 0 and die <<EOM;
603 E: `$self->{cur_dst}' must not contain newline
605 clone_v2($self, \%v2_epochs, $m);
606 return if $self->{lei}->{child_error};
609 if (my $v1 = delete $multi->{v1}) {
610 delete local $lei->{opt}->{epoch} if defined($v2);
611 my $p = $path_pfx.$path;
612 chop($p) if substr($p, -1, 1) eq '/';
615 my %want = map { $_ => 1 } @$v1;
616 for my $name (@$v1) {
617 my $task = bless { %$self }, __PACKAGE__;
618 $task->{-ent} = $m->{$name} //
619 die("BUG: no `$name' in manifest");
620 $task->{cur_src} = "$uri";
621 $task->{cur_dst} = $task->{dst};
623 $task->{cur_dst} .= $name;
624 $task->{cur_src} .= $name;
626 index($task->{cur_dst}, "\n") >= 0 and die <<EOM;
627 E: `$task->{cur_dst}' must not contain newline
629 $task->{cur_src} .= '/';
630 my $dep = $task->{-ent}->{reference} // '';
631 $dep = '' if !$want{$dep};
632 push @{$todo->{$dep}}, $task;
634 clone_all($self, $todo, $m);
636 reap_live() while keys(%$LIVE);
637 return if $self->{lei}->{child_error} || $self->{dry_run};
639 if (delete $self->{-culled_manifest}) { # set by clone_v2/-I/--exclude
640 # write the smaller manifest if epochs were skipped so
641 # users won't have to delete manifest if they +w an
642 # epoch they no longer want to skip
643 my $json = PublicInbox::Config->json->encode($m);
644 my $mtime = (stat($fn))[9];
645 gzip(\$json => $fn) or die "gzip: $GzipError";
646 utime($mtime, $mtime, $fn) or die "utime(..., $fn): $!";
648 ft_rename($ft, "$self->{dst}/manifest.js.gz", 0666);
649 open my $x, '>', "$self->{dst}/mirror.done"; # for _wq_done_wait
652 sub start_clone_url {
654 return try_manifest($self) if $self->{src} =~ m!\Ahttps?://!;
655 die "TODO: non-HTTP/HTTPS clone of $self->{src} not supported, yet";
658 sub do_mirror { # via wq_io_do or public-inbox-clone
660 my $lei = $self->{lei};
661 umask($lei->{client_umask}) if defined $lei->{client_umask};
663 my $ic = $lei->{opt}->{'inbox-config'} //= 'always';
664 $ic =~ /\A(?:v1|v2|always|never)\z/s or die <<"";
665 --inbox-config must be one of `always', `v2', `v1', or `never'
667 my $iv = $lei->{opt}->{'inbox-version'};
670 return clone_v1($self) if $iv == 1;
671 return try_scrape($self) if $iv == 2;
672 die "bad --inbox-version=$iv\n";
674 return start_clone_url($self) if $self->{src} =~ m!://!;
675 die "TODO: cloning local directories not supported, yet";
677 $lei->fail($@) if $@;
681 my ($cls, $lei, $src, $dst) = @_;
682 my $self = bless { src => $src, dst => $dst }, $cls;
684 my ($op_c, $ops) = $lei->workers_start($self, 1);
686 $self->wq_io_do('do_mirror', []);
688 $lei->wait_wq_events($op_c, $ops);
691 sub ipc_atfork_child {
693 $self->{lei}->_lei_atfork_child;
694 $self->SUPER::ipc_atfork_child;
698 my ($dir, $ibx_ver) = @_;
699 my $f = "$dir/Makefile";
700 if (sysopen my $fh, $f, O_CREAT|O_EXCL|O_WRONLY) {
701 print $fh <<EOM or die "print($f) $!";
702 # This is a v$ibx_ver public-inbox, see the public-inbox-v$ibx_ver-format(5)
703 # manpage for more information on the format. This Makefile is
704 # intended as a familiar wrapper for users unfamiliar with
705 # public-inbox-* commands.
707 # See the respective manpages for public-inbox-fetch(1),
708 # public-inbox-index(1), etc for more information on
709 # some of the commands used by this Makefile.
711 # This Makefile will not be modified nor read by public-inbox,
712 # so you may edit it freely with your own convenience targets
713 # and notes. public-inbox-fetch will recreate it if removed.
715 print $fh <<'EOM' or die "print($f): $!";
716 # the default target:
718 @echo Common targets:
719 @echo ' make fetch - fetch from remote git repostorie(s)'
720 @echo ' make update - fetch and update index '
722 @echo Rarely needed targets:
723 @echo ' make reindex - may be needed for new features/bugfixes'
724 @echo ' make compact - rewrite Xapian storage to save space'
729 @if ! public-inbox-fetch --exit-code; \
732 test $$c -eq 127 && exit 0; \
734 elif test -f msgmap.sqlite3 || test -f public-inbox/msgmap.sqlite3; \
736 public-inbox-index; \
738 echo 'public-inbox index not initialized'; \
739 echo 'see public-inbox-index(1) man page'; \
742 public-inbox-index --reindex
746 .PHONY : help fetch update reindex compact
748 close $fh or die "close($f): $!";
750 die "open($f): $!" unless $!{EEXIST};