1 # Copyright (C) all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
3 # Wrapper to "git fetch" remote public-inboxes
4 package PublicInbox::Fetch;
6 use parent qw(PublicInbox::IPC);
8 use PublicInbox::Spawn qw(popen_rd run_die spawn);
9 use PublicInbox::Admin;
11 use PublicInbox::LeiCurl;
12 use PublicInbox::LeiMirror;
15 sub new { bless {}, __PACKAGE__ }
19 my $rn = $lei->{opt}->{'try-remote'} // [ 'origin', '_grokmirror' ];
21 my $cmd = [ qw(git config), "remote.$r.url" ];
22 my $fh = popen_rd($cmd, undef, { -C => $dir, 2 => $lei->{2} });
31 # PSGI mount prefixes and manifest.js.gz prefixes don't always align...
32 # TODO: remove, handle multi-inbox fetch
33 sub deduce_epochs ($$) {
35 my ($v1_ent, @v2_epochs);
39 $v1_ent = $m->{$path};
40 @v2_epochs = grep(m!\A\Q$path\E/git/[0-9]+\.git\z!, keys %$m);
41 } while (!defined($v1_ent) && !@v2_epochs &&
42 $path =~ s!\A(/[^/]+)/!/! and $path_pfx .= $1);
43 ($path_pfx, $v1_ent ? $path : undef, @v2_epochs);
46 sub do_manifest ($$$) {
47 my ($lei, $dir, $ibx_uri) = @_;
48 my $muri = URI->new("$ibx_uri/manifest.js.gz");
49 my $ft = File::Temp->new(TEMPLATE => 'm-XXXX',
50 UNLINK => 1, DIR => $dir, SUFFIX => '.tmp');
51 my $mf = "$dir/manifest.js.gz";
52 my $m0; # current manifest.js.gz contents
53 if (open my $fh, '<', $mf) {
55 PublicInbox::LeiMirror::decode_manifest($fh, $mf, $mf)
59 my ($bn) = ($ft->filename =~ m!/([^/]+)\z!);
60 my $curl_cmd = $lei->{curl}->for_uri($lei, $muri, qw(-R -o), $bn);
61 my $opt = { -C => $dir };
62 $opt->{$_} = $lei->{$_} for (0..2);
63 my $cerr = PublicInbox::LeiMirror::run_reap($lei, $curl_cmd, $opt);
65 return [ 404, $muri ] if ($cerr >> 8) == 22; # 404 Missing
66 $lei->child_error($cerr, "@$curl_cmd failed");
70 PublicInbox::LeiMirror::decode_manifest($ft, $ft, $muri);
71 } or return [ 404, $muri ];
74 # filter out unchanged entries. We check modified, too, since
75 # fingerprints are SHA-1, so there's a teeny chance they'll collide
76 while (my ($k, $v0) = each %{$m0 // {}}) {
77 my $cur = $m1->{$k} // next;
78 my $f0 = $v0->{fingerprint} // next;
79 my $f1 = $cur->{fingerprint} // next;
80 my $t0 = $v0->{modified} // next;
81 my $t1 = $cur->{modified} // next;
82 delete($mdiff->{$k}) if $f0 eq $f1 && $t0 == $t1;
84 unless (keys %$mdiff) {
85 $lei->child_error(127 << 8) if $lei->{opt}->{'exit-code'};
88 my (undef, $v1_path, @v2_epochs) =
89 deduce_epochs($mdiff, $ibx_uri->path);
90 [ 200, $muri, $v1_path, \@v2_epochs, $ft, $mf, $m1 ];
93 sub get_fingerprint2 {
95 require PublicInbox::SHA;
96 my $rd = popen_rd([qw(git show-ref)], undef, { -C => $git_dir });
97 PublicInbox::SHA::sha256(do { local $/; <$rd> });
100 sub writable_dir ($) {
102 return unless -d $dir && -w _;
104 $st[2] & 0222; # any writable bits set? (in case of root)
107 sub do_fetch { # main entry point
108 my ($cls, $lei, $cd) = @_;
110 $lei->{curl} //= PublicInbox::LeiCurl->new($lei) or return;
111 my $dir = PublicInbox::Admin::resolve_inboxdir($cd, \$ibx_ver);
112 my ($ibx_uri, @git_dir, @epochs, $mg, @new_epoch, $skip);
114 my $url = remote_url($lei, $dir) //
115 die "E: $dir missing remote.*.url\n";
116 $ibx_uri = URI->new($url);
118 require PublicInbox::MultiGit;
119 $mg = PublicInbox::MultiGit->new($dir, 'all.git', 'git');
120 @epochs = $mg->git_epochs;
121 my ($git_url, $epoch);
122 for my $nr (@epochs) { # try newest epoch, first
123 my $edir = "$dir/git/$nr.git";
124 if (!writable_dir($edir)) {
128 next if defined $git_url;
129 if (defined(my $url = remote_url($lei, $edir))) {
133 warn "W: $edir missing remote.*.url\n";
134 my $pid = spawn([qw(git config -l)], undef,
135 { 1 => $lei->{2}, 2 => $lei->{2} });
137 $lei->child_error($?) if $?;
140 @epochs = grep { !$skip->{$_} } @epochs if $skip;
141 $skip //= {}; # makes code below easier
142 $git_url or die "Unable to determine git URL\n";
143 my $inbox_url = $git_url;
144 $inbox_url =~ s!/git/$epoch(?:\.git)?/?\z!! or
145 $inbox_url =~ s!/$epoch(?:\.git)?/?\z!! or die <<EOM;
146 Unable to infer inbox URL from <$git_url>
148 $ibx_uri = URI->new($inbox_url);
150 PublicInbox::LeiMirror::write_makefile($dir, $ibx_ver);
151 $lei->qerr("# inbox URL: $ibx_uri/");
152 my $res = do_manifest($lei, $dir, $ibx_uri) or return;
153 my ($code, $muri, $v1_path, $v2_epochs, $ft, $mf, $m1) = @$res;
155 # any pre-manifest.js.gz instances running? Just fetch all
156 # existing ones and unconditionally try cloning the next
157 $v2_epochs = [ map { "$dir/git/$_.git" } @epochs ];
159 my $n = $epochs[-1] + 1;
160 push @$v2_epochs, "$dir/git/$n.git" if !$skip->{$n};
163 $code == 200 or die "BUG unexpected code $code\n";
167 defined($v1_path) and warn <<EOM;
168 E: got v1 `$v1_path' when expecting v2 epoch(s) in <$muri>, WTF?
170 @git_dir = map { "$dir/git/$_.git" } sort { $a <=> $b } map {
171 my ($nr) = (m!/([0-9]+)\.git\z!g);
172 $skip->{$nr} ? () : $nr;
174 if ($m1 && scalar keys %$skip) {
175 my $re = join('|', keys %$skip);
176 my @del = grep(m!/git/$re\.git\z!, keys %$m1);
183 # n.b. this expects all epochs are from the same host
184 my $torsocks = $lei->{curl}->torsocks($lei, $muri);
185 my $fp2 = $lei->{opt}->{'exit-code'} ? [] : undef;
187 for my $d (@git_dir) {
189 my $opt = {}; # for spawn
191 $fp2->[0] = get_fingerprint2($d) if $fp2;
192 $cmd = [ @$torsocks, 'git', "--git-dir=$d",
193 PublicInbox::LeiMirror::fetch_args($lei, $opt)];
195 my $e_uri = $ibx_uri->clone;
196 my ($epath) = ($d =~ m!(/git/[0-9]+\.git)\z!);
198 die "BUG: $d is not an epoch to clone\n";
199 $e_uri->path($ibx_uri->path.$epath);
201 PublicInbox::LeiMirror::clone_cmd($lei, $opt),
203 push @new_epoch, substr($epath, 5, -4) + 0;
206 my $cerr = PublicInbox::LeiMirror::run_reap($lei, $cmd, $opt);
207 # do not bail on clone failure if we didn't have a manifest
208 if ($cerr && ($code == 200 || -d $d)) {
209 $lei->child_error($cerr, "@$cmd failed");
213 $fp2->[1] = get_fingerprint2($d);
214 $xit = 0 if $fp2->[0] ne $fp2->[1];
217 for my $i (@new_epoch) { $mg->epoch_cfg_set($i) }
219 PublicInbox::LeiMirror::dump_manifest($m1 => $ft) if $mculled;
220 PublicInbox::LeiMirror::ft_rename($ft, $mf, 0666);
222 $lei->child_error($xit << 8) if $fp2 && $xit;