X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FGitHTTPBackend.pm;h=4d2816a00c4b92b8a9b562906b9d944f567eadaa;hb=3d41aa23f35501ca92aab8aa42980fa73f7fa74f;hp=4f58c6bbcb82a3e392793c3d765f11d8324fe0cf;hpb=bd3ed82913cd065636525a5dd8bab9edcb3e3cb1;p=public-inbox.git diff --git a/lib/PublicInbox/GitHTTPBackend.pm b/lib/PublicInbox/GitHTTPBackend.pm index 4f58c6bb..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 $TEXT = join('|', @text); @@ -42,6 +46,9 @@ sub r ($;$) { sub serve { my ($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 @@ -203,8 +224,7 @@ sub serve_smart { $r->[0] == 403 ? serve_dumb($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; @@ -215,17 +235,16 @@ sub serve_smart { $rpipe->close; $end->(); } - return $res->($r); - } - if ($async) { + $res->($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 - require PublicInbox::GetlineBody; - $r->[2] = PublicInbox::GetlineBody->new($rpipe, $end, $buf); - $res->($r); }; sub { ($res) = @_; @@ -234,7 +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) { @@ -248,7 +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 $buf; while (1) { @@ -257,11 +280,22 @@ sub input_to_file { err($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; }