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;
7 use parent qw(PublicInbox::IPC);
9 use PublicInbox::Spawn qw(popen_rd run_die spawn);
10 use PublicInbox::Admin;
12 use PublicInbox::LeiCurl;
13 use PublicInbox::LeiMirror;
16 sub new { bless {}, __PACKAGE__ }
20 my @cmd; # (git --git-dir=...) to be added by caller
21 $opt->{$_} = $lei->{$_} for (0..2);
22 # we support "-c $key=$val" for arbitrary git config options
23 # e.g.: git -c http.proxy=socks5h://127.0.0.1:9050
24 push(@cmd, '-c', $_) for @{$lei->{opt}->{c} // []};
26 push @cmd, '-q' if $lei->{opt}->{quiet};
27 push @cmd, '-v' if $lei->{opt}->{verbose};
33 my $rn = $lei->{opt}->{'try-remote'} // [ 'origin', '_grokmirror' ];
35 my $cmd = [ qw(git config), "remote.$r.url" ];
36 my $fh = popen_rd($cmd, undef, { -C => $dir, 2 => $lei->{2} });
45 # PSGI mount prefixes and manifest.js.gz prefixes don't always align...
46 # TODO: remove, handle multi-inbox fetch
47 sub deduce_epochs ($$) {
49 my ($v1_ent, @v2_epochs);
53 $v1_ent = $m->{$path};
54 @v2_epochs = grep(m!\A\Q$path\E/git/[0-9]+\.git\z!, keys %$m);
55 } while (!defined($v1_ent) && !@v2_epochs &&
56 $path =~ s!\A(/[^/]+)/!/! and $path_pfx .= $1);
57 ($path_pfx, $v1_ent ? $path : undef, @v2_epochs);
60 sub do_manifest ($$$) {
61 my ($lei, $dir, $ibx_uri) = @_;
62 my $muri = URI->new("$ibx_uri/manifest.js.gz");
63 my $ft = File::Temp->new(TEMPLATE => 'm-XXXX',
64 UNLINK => 1, DIR => $dir, SUFFIX => '.tmp');
65 my $fn = $ft->filename;
66 my $mf = "$dir/manifest.js.gz";
67 my $m0; # current manifest.js.gz contents
68 if (open my $fh, '<', $mf) {
70 PublicInbox::LeiMirror::decode_manifest($fh, $mf, $mf)
74 my ($bn) = ($fn =~ m!/([^/]+)\z!);
75 my $curl_cmd = $lei->{curl}->for_uri($lei, $muri, qw(-R -o), $bn);
76 my $opt = { -C => $dir };
77 $opt->{$_} = $lei->{$_} for (0..2);
78 my $cerr = PublicInbox::LeiMirror::run_reap($lei, $curl_cmd, $opt);
80 return [ 404, $muri ] if ($cerr >> 8) == 22; # 404 Missing
81 $lei->child_error($cerr, "@$curl_cmd failed");
85 PublicInbox::LeiMirror::decode_manifest($ft, $fn, $muri);
86 } or return [ 404, $muri ];
89 # filter out unchanged entries. We check modified, too, since
90 # fingerprints are SHA-1, so there's a teeny chance they'll collide
91 while (my ($k, $v0) = each %{$m0 // {}}) {
92 my $cur = $m1->{$k} // next;
93 my $f0 = $v0->{fingerprint} // next;
94 my $f1 = $cur->{fingerprint} // next;
95 my $t0 = $v0->{modified} // next;
96 my $t1 = $cur->{modified} // next;
97 delete($mdiff->{$k}) if $f0 eq $f1 && $t0 == $t1;
99 unless (keys %$mdiff) {
100 $lei->child_error(127 << 8) if $lei->{opt}->{'exit-code'};
103 my (undef, $v1_path, @v2_epochs) =
104 deduce_epochs($mdiff, $ibx_uri->path);
105 [ 200, $muri, $v1_path, \@v2_epochs, $ft, $mf, $m1 ];
108 sub get_fingerprint2 {
111 my $rd = popen_rd([qw(git show-ref)], undef, { -C => $git_dir });
112 Digest::SHA::sha256(do { local $/; <$rd> });
115 sub writable_dir ($) {
117 return unless -d $dir && -w _;
119 $st[2] & 0222; # any writable bits set? (in case of root)
122 sub do_fetch { # main entry point
123 my ($cls, $lei, $cd) = @_;
125 $lei->{curl} //= PublicInbox::LeiCurl->new($lei) or return;
126 my $dir = PublicInbox::Admin::resolve_inboxdir($cd, \$ibx_ver);
127 my ($ibx_uri, @git_dir, @epochs, $mg, @new_epoch, $skip);
129 my $url = remote_url($lei, $dir) //
130 die "E: $dir missing remote.*.url\n";
131 $ibx_uri = URI->new($url);
133 require PublicInbox::MultiGit;
134 $mg = PublicInbox::MultiGit->new($dir, 'all.git', 'git');
135 @epochs = $mg->git_epochs;
136 my ($git_url, $epoch);
137 for my $nr (@epochs) { # try newest epoch, first
138 my $edir = "$dir/git/$nr.git";
139 if (!writable_dir($edir)) {
143 next if defined $git_url;
144 if (defined(my $url = remote_url($lei, $edir))) {
148 warn "W: $edir missing remote.*.url\n";
149 my $pid = spawn([qw(git config -l)], undef,
150 { 1 => $lei->{2}, 2 => $lei->{2} });
152 $lei->child_error($?) if $?;
155 @epochs = grep { !$skip->{$_} } @epochs if $skip;
156 $skip //= {}; # makes code below easier
157 $git_url or die "Unable to determine git URL\n";
158 my $inbox_url = $git_url;
159 $inbox_url =~ s!/git/$epoch(?:\.git)?/?\z!! or
160 $inbox_url =~ s!/$epoch(?:\.git)?/?\z!! or die <<EOM;
161 Unable to infer inbox URL from <$git_url>
163 $ibx_uri = URI->new($inbox_url);
165 PublicInbox::LeiMirror::write_makefile($dir, $ibx_ver);
166 $lei->qerr("# inbox URL: $ibx_uri/");
167 my $res = do_manifest($lei, $dir, $ibx_uri) or return;
168 my ($code, $muri, $v1_path, $v2_epochs, $ft, $mf, $m1) = @$res;
170 # any pre-manifest.js.gz instances running? Just fetch all
171 # existing ones and unconditionally try cloning the next
172 $v2_epochs = [ map { "$dir/git/$_.git" } @epochs ];
174 my $n = $epochs[-1] + 1;
175 push @$v2_epochs, "$dir/git/$n.git" if !$skip->{$n};
178 $code == 200 or die "BUG unexpected code $code\n";
182 defined($v1_path) and warn <<EOM;
183 E: got v1 `$v1_path' when expecting v2 epoch(s) in <$muri>, WTF?
185 @git_dir = map { "$dir/git/$_.git" } sort { $a <=> $b } map {
186 my ($nr) = (m!/([0-9]+)\.git\z!g);
187 $skip->{$nr} ? () : $nr;
189 if ($m1 && scalar keys %$skip) {
190 my $re = join('|', keys %$skip);
191 my @del = grep(m!/git/$re\.git\z!, keys %$m1);
198 # n.b. this expects all epochs are from the same host
199 my $torsocks = $lei->{curl}->torsocks($lei, $muri);
200 my $fp2 = $lei->{opt}->{'exit-code'} ? [] : undef;
202 for my $d (@git_dir) {
204 my $opt = {}; # for spawn
206 $fp2->[0] = get_fingerprint2($d) if $fp2;
207 $cmd = [ @$torsocks, 'git', "--git-dir=$d",
208 fetch_args($lei, $opt) ];
210 my $e_uri = $ibx_uri->clone;
211 my ($epath) = ($d =~ m!(/git/[0-9]+\.git)\z!);
213 die "BUG: $d is not an epoch to clone\n";
214 $e_uri->path($ibx_uri->path.$epath);
216 PublicInbox::LeiMirror::clone_cmd($lei, $opt),
218 push @new_epoch, substr($epath, 5, -4) + 0;
221 my $cerr = PublicInbox::LeiMirror::run_reap($lei, $cmd, $opt);
222 # do not bail on clone failure if we didn't have a manifest
223 if ($cerr && ($code == 200 || -d $d)) {
224 $lei->child_error($cerr, "@$cmd failed");
228 $fp2->[1] = get_fingerprint2($d);
229 $xit = 0 if $fp2->[0] ne $fp2->[1];
232 for my $i (@new_epoch) { $mg->epoch_cfg_set($i) }
234 PublicInbox::LeiMirror::dump_manifest($m1 => $ft) if $mculled;
235 PublicInbox::LeiMirror::ft_rename($ft, $mf, 0666);
237 $lei->child_error($xit << 8) if $fp2 && $xit;