X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FGitHTTPBackend.pm;h=4ad3fd1ee3aca9ef0b3c44e3a9ac6598163ba825;hb=292ca34140489da2c3458e1d45da5a9ae4af540d;hp=b4851920fe04f55b51ede42fa6770b70982cf17e;hpb=52052329acedddc0082487283a5a0390a8c86602;p=public-inbox.git diff --git a/lib/PublicInbox/GitHTTPBackend.pm b/lib/PublicInbox/GitHTTPBackend.pm index b4851920..4ad3fd1e 100644 --- a/lib/PublicInbox/GitHTTPBackend.pm +++ b/lib/PublicInbox/GitHTTPBackend.pm @@ -7,11 +7,15 @@ package PublicInbox::GitHTTPBackend; use strict; use warnings; use Fcntl qw(:seek); -use IO::File; +use IO::Handle; use HTTP::Date qw(time2str); use HTTP::Status qw(status_message); +use Plack::Util; use PublicInbox::Qspawn; +# 32 is same as the git-daemon connection limit +my $default_limiter = PublicInbox::Qspawn::Limiter->new(32); + # n.b. serving "description" and "cloneurl" should be innocuous enough to # not cause problems. serving "config" might... my @text = qw[HEAD info/refs @@ -42,6 +46,9 @@ sub r ($;$) { sub serve { my ($env, $git, $path) = @_; + # XXX compatibility... ugh, can we stop supporting this? + $git = PublicInbox::Git->new($git) unless ref($git); + # Documentation/technical/http-protocol.txt in git.git # requires one and exactly one query parameter: if ($env->{QUERY_STRING} =~ /\Aservice=git-\w+-pack\z/ || @@ -64,15 +71,29 @@ sub drop_client ($) { } } +my $prev = 0; +my $exp; +sub cache_one_year { + my ($h) = @_; + my $t = time + 31536000; + push @$h, 'Expires', $t == $prev ? $exp : ($exp = time2str($prev = $t)), + 'Cache-Control', 'public, max-age=31536000'; +} + sub serve_dumb { my ($env, $git, $path) = @_; my @h; my $type; - if ($path =~ /\A(?:$BIN)\z/o) { - $type = 'application/octet-stream'; - push @h, 'Expires', time2str(time + 31536000); - push @h, 'Cache-Control', 'public, max-age=31536000'; + if ($path =~ m!\Aobjects/[a-f0-9]{2}/[a-f0-9]{38}\z!) { + $type = 'application/x-git-loose-object'; + cache_one_year(\@h); + } elsif ($path =~ m!\Aobjects/pack/pack-[a-f0-9]{40}\.pack\z!) { + $type = 'application/x-git-packed-objects'; + cache_one_year(\@h); + } elsif ($path =~ m!\Aobjects/pack/pack-[a-f0-9]{40}\.idx\z!) { + $type = 'application/x-git-packed-objects-toc'; + cache_one_year(\@h); } elsif ($path =~ /\A(?:$TEXT)\z/o) { $type = 'text/plain'; push @h, @no_cache; @@ -80,10 +101,9 @@ sub serve_dumb { return r(404); } - my $f = (ref $git ? $git->{git_dir} : $git) . '/' . $path; + my $f = $git->{git_dir} . '/' . $path; return r(404) unless -f $f && -r _; # just in case it's a FIFO :P - my @st = stat(_); - my $size = $st[7]; + my $size = -s _; # TODO: If-Modified-Since and Last-Modified? open my $in, '<', $f or return r(404); @@ -179,14 +199,15 @@ sub serve_smart { my $val = $env->{$name}; $env{$name} = $val if defined $val; } - my $git_dir = ref $git ? $git->{git_dir} : $git; + my $limiter = $git->{-httpbackend_limiter} || $default_limiter; + my $git_dir = $git->{git_dir}; $env{GIT_HTTP_EXPORT_ALL} = '1'; $env{PATH_TRANSLATED} = "$git_dir/$path"; - my %rdr = ( 0 => fileno($in) ); - my $x = PublicInbox::Qspawn->new([qw(git http-backend)], \%env, \%rdr); + my $rdr = { 0 => fileno($in) }; + my $qsp = PublicInbox::Qspawn->new([qw(git http-backend)], \%env, $rdr); my ($fh, $rpipe); my $end = sub { - if (my $err = $x->finish) { + if (my $err = $qsp->finish) { err($env, "git http-backend ($git_dir): $err"); } $fh->close if $fh; # async-only @@ -203,7 +224,7 @@ sub serve_smart { $r->[0] == 403 ? serve_dumb($env, $git, $path) : $r; }; my $res; - my $async = $env->{'pi-httpd.async'}; + my $async = $env->{'pi-httpd.async'}; # XXX unstable API my $io = $env->{'psgix.io'}; my $cb = sub { my $r = $rd_hdr->() or return; @@ -234,7 +255,7 @@ sub serve_smart { # holding the input here is a waste of FDs and memory $env->{'psgi.input'} = undef; - $x->start(sub { # may run later, much later... + $qsp->start($limiter, sub { # may run later, much later... ($rpipe) = @_; $in = undef; if ($async) { @@ -248,7 +269,11 @@ sub serve_smart { sub input_to_file { my ($env) = @_; - my $in = IO::File->new_tmpfile; + open(my $in, '+>', undef); + unless (defined $in) { + err($env, "could not open temporary file: $!"); + return; + } my $input = $env->{'psgi.input'}; my $buf; while (1) { @@ -257,11 +282,22 @@ sub input_to_file { err($env, "error reading input: $!"); return; } - last if ($r == 0); - $in->write($buf); + my $off = 0; + while ($r > 0) { + my $w = syswrite($in, $buf, $r, $off); + if (defined $w) { + $r -= $w; + $off += $w; + } else { + err($env, "error writing temporary file: $!"); + return; + } + } + } + unless (defined(sysseek($in, 0, SEEK_SET))) { + err($env, "error seeking temporary file: $!"); + return; } - $in->flush; - $in->sysseek(0, SEEK_SET); return $in; }