use v5.10.1;
use parent qw(PublicInbox::IPC);
use URI ();
-use PublicInbox::Spawn qw(popen_rd);
+use PublicInbox::Spawn qw(popen_rd run_die);
use PublicInbox::Admin;
use PublicInbox::LEI;
use PublicInbox::LeiCurl;
use PublicInbox::LeiMirror;
-use IO::Uncompress::Gunzip qw(gunzip $GunzipError);
use File::Temp ();
sub new { bless {}, __PACKAGE__ }
-sub fetch_cmd ($$) {
+sub fetch_args ($$) {
my ($lei, $opt) = @_;
- my @cmd = qw(git);
+ my @cmd; # (git --git-dir=...) to be added by caller
$opt->{$_} = $lei->{$_} for (0..2);
# we support "-c $key=$val" for arbitrary git config options
# e.g.: git -c http.proxy=socks5h://127.0.0.1:9050
my $fh = popen_rd($cmd, undef, { -C => $dir, 2 => $lei->{2} });
my $url = <$fh>;
close $fh or return;
- chomp $url;
+ $url =~ s!/*\n!!s;
$url;
}
-sub do_fetch {
- my ($cls, $lei, $cd) = @_;
- my $ibx_ver;
- my $curl = PublicInbox::LeiCurl->new($lei) or return;
- my $dir = PublicInbox::Admin::resolve_inboxdir($cd, \$ibx_ver);
- if ($ibx_ver == 1) {
- my $url = remote_url($lei, $dir) //
- die "E: $dir missing remote.origin.url\n";
- my $uri = URI->new($url);
- my $torsocks = $curl->torsocks($lei, $uri);
- my $opt = { -C => $dir };
- my $cmd = [ @$torsocks, fetch_cmd($lei, $opt) ];
- my $cerr = PublicInbox::LeiMirror::run_reap($lei, $cmd, $opt);
- $lei->child_error($cerr, "@$cmd failed") if $cerr;
- return;
- }
- # v2:
- opendir my $dh, "$dir/git" or die "opendir $dir/git: $!";
- my @epochs = sort { $b <=> $a } map { substr($_, 0, -4) + 0 }
- grep(/\A[0-9]+\.git\z/, readdir($dh));
- my ($git_url, $epoch);
- for my $nr (@epochs) { # try newest epoch, first
- my $edir = "$dir/git/$nr.git";
- if (defined(my $url = remote_url($lei, $edir))) {
- $git_url = $url;
- $epoch = $nr;
- last;
- } else {
- warn "W: $edir missing remote.origin.url\n";
- }
- }
- $git_url or die "Unable to determine git URL\n";
- my $inbox_url = $git_url;
- $inbox_url =~ s!/git/$epoch(?:\.git)?/?\z!! or
- $inbox_url =~ s!/$epoch(?:\.git)?/?\z!! or die <<EOM;
-Unable to infer inbox URL from <$git_url>
-EOM
- $lei->qerr("# inbox URL: $inbox_url/");
- my $muri = URI->new("$inbox_url/manifest.js.gz");
- my $ft = File::Temp->new(TEMPLATE => 'manifest-XXXX',
- UNLINK => 1, DIR => $dir);
+sub do_manifest ($$$) {
+ my ($lei, $dir, $ibx_uri) = @_;
+ my $muri = URI->new("$ibx_uri/manifest.js.gz");
+ my $ft = File::Temp->new(TEMPLATE => 'm-XXXX',
+ UNLINK => 1, DIR => $dir, SUFFIX => '.tmp');
my $fn = $ft->filename;
- my @opt = (qw(-R -o), $fn);
my $mf = "$dir/manifest.js.gz";
my $m0; # current manifest.js.gz contents
if (open my $fh, '<', $mf) {
PublicInbox::LeiMirror::decode_manifest($fh, $mf, $mf)
};
$lei->err($@) if $@;
- push @opt, '-z', $mf if defined($m0);
}
- my $curl_cmd = $curl->for_uri($lei, $muri, @opt);
- my $opt = {};
+ my ($bn) = ($fn =~ m!/([^/]+)\z!);
+ my $curl_cmd = $lei->{curl}->for_uri($lei, $muri, qw(-R -o), $bn);
+ my $opt = { -C => $dir };
$opt->{$_} = $lei->{$_} for (0..2);
my $cerr = PublicInbox::LeiMirror::run_reap($lei, $curl_cmd, $opt);
- return $lei->child_error($cerr, "@$curl_cmd failed") if $cerr;
- return if !-s $ft; # 304 Not Modified via curl -z
-
+ if ($cerr) {
+ return [ 404 ] if ($cerr >> 8) == 22; # 404 Missing
+ $lei->child_error($cerr, "@$curl_cmd failed");
+ return;
+ }
my $m1 = PublicInbox::LeiMirror::decode_manifest($ft, $fn, $muri);
my $mdiff = { %$m1 };
- # filter out unchanged entries
+ # filter out unchanged entries. We check modified, too, since
+ # fingerprints are SHA-1, so there's a teeny chance they'll collide
while (my ($k, $v0) = each %{$m0 // {}}) {
my $cur = $m1->{$k} // next;
my $f0 = $v0->{fingerprint} // next;
my $t1 = $cur->{modified} // next;
delete($mdiff->{$k}) if $f0 eq $f1 && $t0 == $t1;
}
- my $ibx_uri = URI->new("$inbox_url/");
- my ($path_pfx, $v1_bare, @v2_epochs) =
+ unless (keys %$mdiff) {
+ $lei->child_error(127 << 8) if $lei->{opt}->{'exit-code'};
+ return;
+ }
+ my (undef, $v1_path, @v2_epochs) =
PublicInbox::LeiMirror::deduce_epochs($mdiff, $ibx_uri->path);
- defined($v1_bare) and die <<EOM;
-E: got v1 `$v1_bare' when expecting v2 epoch(s) in <$muri>, WTF?
-EOM
- my @epoch_nr = sort { $a <=> $b }
- map { my ($nr) = (m!/([0-9]+)\.git\z!g) } @v2_epochs;
+ [ 200, $v1_path, \@v2_epochs, $muri, $ft, $mf ];
+}
+
+sub get_fingerprint2 {
+ my ($git_dir) = @_;
+ require Digest::SHA;
+ my $rd = popen_rd([qw(git show-ref)], undef, { -C => $git_dir });
+ Digest::SHA::sha256(do { local $/; <$rd> });
+}
+sub do_fetch { # main entry point
+ my ($cls, $lei, $cd) = @_;
+ my $ibx_ver;
+ $lei->{curl} //= PublicInbox::LeiCurl->new($lei) or return;
+ my $dir = PublicInbox::Admin::resolve_inboxdir($cd, \$ibx_ver);
+ my ($ibx_uri, @git_dir, @epochs, $mg, @new_epoch, $skip);
+ if ($ibx_ver == 1) {
+ my $url = remote_url($lei, $dir) //
+ die "E: $dir missing remote.origin.url\n";
+ $ibx_uri = URI->new($url);
+ } else { # v2:
+ require PublicInbox::MultiGit;
+ $mg = PublicInbox::MultiGit->new($dir, 'all.git', 'git');
+ my @epochs = $mg->git_epochs;
+ my ($git_url, $epoch);
+ for my $nr (@epochs) { # try newest epoch, first
+ my $edir = "$dir/git/$nr.git";
+ unless (-d $edir && -w _) { # must be writable dir
+ $skip->{$nr} = 1;
+ next;
+ }
+ if (defined(my $url = remote_url($lei, $edir))) {
+ $git_url = $url;
+ $epoch = $nr;
+ last;
+ } else {
+ warn "W: $edir missing remote.origin.url\n";
+ }
+ }
+ @epochs = grep { !$skip->{$_} } @epochs if $skip;
+ $skip //= {}; # makes code below easier
+ $git_url or die "Unable to determine git URL\n";
+ my $inbox_url = $git_url;
+ $inbox_url =~ s!/git/$epoch(?:\.git)?/?\z!! or
+ $inbox_url =~ s!/$epoch(?:\.git)?/?\z!! or die <<EOM;
+Unable to infer inbox URL from <$git_url>
+EOM
+ $ibx_uri = URI->new($inbox_url);
+ }
+ PublicInbox::LeiMirror::write_makefile($dir, $ibx_ver);
+ $lei->qerr("# inbox URL: $ibx_uri/");
+ my $res = do_manifest($lei, $dir, $ibx_uri) or return;
+ my ($code, $v1_path, $v2_epochs, $muri, $ft, $mf) = @$res;
+ if ($code == 404) {
+ # any pre-manifest.js.gz instances running? Just fetch all
+ # existing ones and unconditionally try cloning the next
+ $v2_epochs = [ map { "$dir/git/$_.git" } @epochs ];
+ if (@epochs) {
+ my $n = $epochs[-1] + 1;
+ push @$v2_epochs, "$dir/git/$n.git" if !$skip->{$n};
+ }
+ } else {
+ $code == 200 or die "BUG unexpected code $code\n";
+ }
+ if ($ibx_ver == 2) {
+ defined($v1_path) and warn <<EOM;
+E: got v1 `$v1_path' when expecting v2 epoch(s) in <$muri>, WTF?
+EOM
+ @git_dir = map { "$dir/git/$_.git" } sort { $a <=> $b } map {
+ my ($nr) = (m!/([0-9]+)\.git\z!g);
+ $skip->{$nr} ? () : $nr;
+ } @$v2_epochs;
+ } else {
+ $git_dir[0] = $dir;
+ }
# n.b. this expects all epochs are from the same host
- my $torsocks = $curl->torsocks($lei, $muri);
- for my $nr (@epoch_nr) {
- my $dir = "$dir/git/$nr.git";
+ my $torsocks = $lei->{curl}->torsocks($lei, $muri);
+ my $fp2 = $lei->{opt}->{'exit-code'} ? [] : undef;
+ my $xit = 127;
+ for my $d (@git_dir) {
my $cmd;
- my $opt = {};
- if (-d $dir) {
- $opt->{-C} = $dir;
- $cmd = [ @$torsocks, fetch_cmd($lei, $opt) ];
+ my $opt = {}; # for spawn
+ if (-d $d) {
+ $fp2->[0] = get_fingerprint2($d) if $fp2;
+ $cmd = [ @$torsocks, 'git', "--git-dir=$d",
+ fetch_args($lei, $opt) ];
} else {
my $e_uri = $ibx_uri->clone;
- $e_uri->path($ibx_uri->path."git/$nr.git");
+ my ($epath) = ($d =~ m!(/git/[0-9]+\.git)\z!);
+ defined($epath) or
+ die "BUG: $d is not an epoch to clone\n";
+ $e_uri->path($ibx_uri->path.$epath);
$cmd = [ @$torsocks,
PublicInbox::LeiMirror::clone_cmd($lei, $opt),
- $$e_uri, $dir ];
+ $$e_uri, $d];
+ push @new_epoch, substr($epath, 5, -4) + 0;
+ $xit = 0;
}
my $cerr = PublicInbox::LeiMirror::run_reap($lei, $cmd, $opt);
- return $lei->child_error($cerr, "@$cmd failed") if $cerr;
+ # do not bail on clone failure if we didn't have a manifest
+ if ($cerr && ($code == 200 || -d $d)) {
+ $lei->child_error($cerr, "@$cmd failed");
+ return;
+ }
+ if ($fp2 && $xit) {
+ $fp2->[1] = get_fingerprint2($d);
+ $xit = 0 if $fp2->[0] ne $fp2->[1];
+ }
+ }
+ for my $i (@new_epoch) { $mg->epoch_cfg_set($i) }
+ if ($ft) {
+ my $fn = $ft->filename;
+ rename($fn, $mf) or die "E: rename($fn, $mf): $!\n";
+ $ft->unlink_on_destroy(0);
}
- rename($fn, $mf) or die "E: rename($fn, $mf): $!\n";
- $ft->unlink_on_destroy(0);
+ $lei->child_error($xit << 8) if $fp2 && $xit;
}
1;