X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FGitHTTPBackend.pm;h=4d2816a00c4b92b8a9b562906b9d944f567eadaa;hb=3d41aa23f35501ca92aab8aa42980fa73f7fa74f;hp=9464cb499a3f0c4744b8b33eec1559e415da4887;hpb=64aea34d06f71828b0bdd6ae177b9bcf22d752b4;p=public-inbox.git diff --git a/lib/PublicInbox/GitHTTPBackend.pm b/lib/PublicInbox/GitHTTPBackend.pm index 9464cb49..4d2816a0 100644 --- a/lib/PublicInbox/GitHTTPBackend.pm +++ b/lib/PublicInbox/GitHTTPBackend.pm @@ -1,4 +1,4 @@ -# Copyright (C) 2016 all contributors +# Copyright (C) 2016-2018 all contributors # License: AGPL-3.0+ # when no endpoints match, fallback to this and serve a static file @@ -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 @@ -23,7 +27,7 @@ my @binary = qw! objects/pack/pack-[a-f0-9]{40}\.(?:pack|idx) !; -our $ANY = join('|', @binary, @text); +our $ANY = join('|', @binary, @text, 'git-upload-pack'); my $BIN = join('|', @binary); my $TEXT = join('|', @text); @@ -40,15 +44,20 @@ sub r ($;$) { } sub serve { - my ($cgi, $git, $path) = @_; + my ($env, $git, $path) = @_; + + # XXX compatibility... ugh, can we stop supporting this? + $git = PublicInbox::Git->new($git) unless ref($git); - my $service = $cgi->query_parameters->get('service') || ''; - if ($service =~ /\Agit-\w+-pack\z/ || $path =~ /\Agit-\w+-pack\z/) { - my $ok = serve_smart($cgi, $git, $path); + # Documentation/technical/http-protocol.txt in git.git + # requires one and exactly one query parameter: + if ($env->{QUERY_STRING} =~ /\Aservice=git-\w+-pack\z/ || + $path =~ /\Agit-\w+-pack\z/) { + my $ok = serve_smart($env, $git, $path); return $ok if $ok; } - serve_dumb($cgi, $git, $path); + serve_dumb($env, $git, $path); } sub err ($@) { @@ -62,15 +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 = $t)), + 'Cache-Control', 'public, max-age=31536000'; +} + sub serve_dumb { - my ($cgi, $git, $path) = @_; + my ($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; @@ -78,11 +101,9 @@ sub serve_dumb { return r(404); } - my $f = "$git->{git_dir}/$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 $env = $cgi->{env}; + my $size = -s _; # TODO: If-Modified-Since and Last-Modified? open my $in, '<', $f or return r(404); @@ -90,7 +111,7 @@ sub serve_dumb { my $code = 200; push @h, 'Content-Type', $type; if (($env->{HTTP_RANGE} || '') =~ /\bbytes=(\d*)-(\d*)\z/) { - ($code, $len) = prepare_range($cgi, $in, \@h, $1, $2, $size); + ($code, $len) = prepare_range($env, $in, \@h, $1, $2, $size); if ($code == 416) { push @h, 'Content-Range', "bytes */$size"; return [ 416, \@h, [] ]; @@ -118,7 +139,7 @@ sub serve_dumb { } sub prepare_range { - my ($cgi, $in, $h, $beg, $end, $size) = @_; + my ($env, $in, $h, $beg, $end, $size) = @_; my $code = 200; my $len = $size; if ($beg eq '') { @@ -152,7 +173,7 @@ sub prepare_range { push @$h, "bytes $beg-$end/$size"; # FIXME: Plack::Middleware::Deflater bug? - $cgi->{env}->{'psgix.no-compress'} = 1; + $env->{'psgix.no-compress'} = 1; } } ($code, $len); @@ -160,8 +181,7 @@ sub prepare_range { # returns undef if 403 so it falls back to dumb HTTP sub serve_smart { - my ($cgi, $git, $path) = @_; - my $env = $cgi->{env}; + my ($env, $git, $path) = @_; my $in = $env->{'psgi.input'}; my $fd = eval { fileno($in) }; unless (defined $fd && $fd >= 0) { @@ -179,17 +199,16 @@ sub serve_smart { my $val = $env->{$name}; $env{$name} = $val if defined $val; } + 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 { - $rpipe = undef; - if (my $err = $x->finish) { + if (my $err = $qsp->finish) { err($env, "git http-backend ($git_dir): $err"); - drop_client($env); } $fh->close if $fh; # async-only }; @@ -201,33 +220,31 @@ sub serve_smart { my $r = sysread($rpipe, $buf, 1024, length($buf)); return if !defined($r) && ($!{EINTR} || $!{EAGAIN}); return r(500, 'http-backend error') unless $r; - $r = parse_cgi_headers(\$buf) or return; - $r->[0] == 403 ? serve_dumb($cgi, $git, $path) : $r; + $r = parse_cgi_headers(\$buf) or return; # incomplete headers + $r->[0] == 403 ? serve_dumb($env, $git, $path) : $r; }; my $res; - my $async = $env->{'pi-httpd.async'}; - my $io = $env->{'psgix.io'}; + my $async = $env->{'pi-httpd.async'}; # XXX unstable API my $cb = sub { my $r = $rd_hdr->() or return; $rd_hdr = undef; if (scalar(@$r) == 3) { # error: - $async->close if $async; - return $res->($r); - } - if ($async) { + if ($async) { + $async->close; # calls rpipe->close + } else { + $rpipe->close; + $end->(); + } + $res->($r); + } elsif ($async) { $fh = $res->($r); - return $async->async_pass($io, $fh, \$buf); + $async->async_pass($env->{'psgix.io'}, $fh, \$buf); + } else { # for synchronous PSGI servers + require PublicInbox::GetlineBody; + $r->[2] = PublicInbox::GetlineBody->new($rpipe, $end, + $buf); + $res->($r); } - - # for synchronous PSGI servers - $r->[2] = Plack::Util::inline_object( - close => $end, - getline => sub { - my $ret = $buf; - $buf = undef; - defined $ret ? $ret : $rpipe->getline; - }); - $res->($r); }; sub { ($res) = @_; @@ -236,7 +253,7 @@ 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) { @@ -250,7 +267,11 @@ 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 $buf; while (1) { @@ -259,11 +280,22 @@ sub input_to_file { err($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; }