-# Copyright (C) 2016-2019 all contributors <meta@public-inbox.org>
+# Copyright (C) all contributors <meta@public-inbox.org>
# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
# when no endpoints match, fallback to this and serve a static file
# or smart HTTP. This is our wrapper for git-http-backend(1)
package PublicInbox::GitHTTPBackend;
use strict;
-use warnings;
+use v5.10.1;
use Fcntl qw(:seek);
-use IO::Handle;
+use IO::Handle; # ->flush
use HTTP::Date qw(time2str);
-use HTTP::Status qw(status_message);
use PublicInbox::Qspawn;
use PublicInbox::Tmpfile;
-use PublicInbox::WwwStatic;
+use PublicInbox::WwwStatic qw(r @NO_CACHE);
+use Carp ();
# 32 is same as the git-daemon connection limit
my $default_limiter = PublicInbox::Qspawn::Limiter->new(32);
objects/info/(?:http-alternates|alternates|packs)
cloneurl description];
-my @binary = qw!
- objects/[a-f0-9]{2}/[a-f0-9]{38}
- objects/pack/pack-[a-f0-9]{40}\.(?:pack|idx)
- !;
+my @binary = ('objects/[a-f0-9]{2}/[a-f0-9]{38,62}',
+ 'objects/pack/pack-[a-f0-9]{40,64}\.(?:pack|idx)');
our $ANY = join('|', @binary, @text, 'git-upload-pack');
-my $BIN = join('|', @binary);
my $TEXT = join('|', @text);
-my @no_cache = ('Expires', 'Fri, 01 Jan 1980 00:00:00 GMT',
- 'Pragma', 'no-cache',
- 'Cache-Control', 'no-cache, max-age=0, must-revalidate');
-
-sub r ($;$) {
- my ($code, $msg) = @_;
- $msg ||= status_message($code);
- my $len = length($msg);
- [ $code, [qw(Content-Type text/plain Content-Length), $len, @no_cache],
- [$msg] ]
-}
-
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-[A-Za-z0-9_]+-pack\z/ ||
serve_dumb($env, $git, $path);
}
-sub err ($@) {
- my ($env, @msg) = @_;
- $env->{'psgi.errors'}->print(@msg, "\n");
-}
+sub ucarp { Carp::carp(@_); undef }
my $prev = 0;
my $exp;
my $h = [];
my $type;
- if ($path =~ m!\Aobjects/[a-f0-9]{2}/[a-f0-9]{38}\z!) {
+ if ($path =~ m!\Aobjects/[a-f0-9]{2}/[a-f0-9]{38,62}\z!) {
$type = 'application/x-git-loose-object';
cache_one_year($h);
- } elsif ($path =~ m!\Aobjects/pack/pack-[a-f0-9]{40}\.pack\z!) {
+ } elsif ($path =~ m!\Aobjects/pack/pack-[a-f0-9]{40,64}\.pack\z!) {
$type = 'application/x-git-packed-objects';
cache_one_year($h);
- } elsif ($path =~ m!\Aobjects/pack/pack-[a-f0-9]{40}\.idx\z!) {
+ } elsif ($path =~ m!\Aobjects/pack/pack-[a-f0-9]{40,64}\.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;
+ push @$h, @NO_CACHE;
} else {
return r(404);
}
$path = "$git->{git_dir}/$path";
- PublicInbox::WwwStatic::response($env, $h, $path, $type) // r(404);
+ PublicInbox::WwwStatic::response($env, $h, $path, $type);
}
sub git_parse_hdr { # {parse_hdr} for Qspawn
foreach my $name (qw(QUERY_STRING
REMOTE_USER REMOTE_ADDR
HTTP_CONTENT_ENCODING
+ HTTP_GIT_PROTOCOL
CONTENT_TYPE
SERVER_PROTOCOL
REQUEST_METHOD)) {
my $input = $env->{'psgi.input'};
my $fd = eval { fileno($input) };
- if (defined $fd && $fd >= 0) {
- return { 0 => $fd };
- }
+ return { 0 => $fd } if (defined $fd && $fd >= 0);
my $id = "git-http.input.$env->{REMOTE_ADDR}:$env->{REMOTE_PORT}";
- my $in = tmpfile($id);
- unless (defined $in) {
- err($env, "could not open temporary file: $!");
- return;
- }
+ my $in = tmpfile($id) // return ucarp("tmpfile: $!");
my $buf;
while (1) {
- my $r = $input->read($buf, 8192);
- unless (defined $r) {
- err($env, "error reading input: $!");
- return;
- }
+ my $r = $input->read($buf, 8192) // return ucarp("read $!");
last if $r == 0;
- unless (print $in $buf) {
- err($env, "error writing temporary file: $!");
- return;
- }
+ print $in $buf // return ucarp("print: $!");
}
# ensure it's visible to git-http-backend(1):
- unless ($in->flush) {
- err($env, "error writing temporary file: $!");
- return;
- }
- unless (defined(sysseek($in, 0, SEEK_SET))) {
- err($env, "error seeking temporary file: $!");
- return;
- }
+ $in->flush // return ucarp("flush: $!");
+ sysseek($in, 0, SEEK_SET) // return ucarp($env, "seek: $!");
{ 0 => $in };
}
-sub parse_cgi_headers {
- my ($r, $bref) = @_;
+sub parse_cgi_headers { # {parse_hdr} for Qspawn
+ my ($r, $bref, $ctx) = @_;
return r(500) unless defined $r && $r >= 0;
$$bref =~ s/\A(.*?)\r?\n\r?\n//s or return $r == 0 ? r(500) : undef;
my $h = $1;
push @h, $k, $v;
}
}
- [ $code, \@h ]
+
+ # fallback to WwwCoderepo if cgit 404s. Duplicating $ctx prevents
+ # ->finalize from the current Qspawn from using qspawn.wcb.
+ # This makes qspawn skip ->async_pass and causes
+ # PublicInbox::HTTPD::Async::event_step to close shortly after
+ if ($code == 404 && $ctx->{www} && !$ctx->{_coderepo_tried}++) {
+ my $wcb = delete $ctx->{env}->{'qspawn.wcb'};
+ $ctx->{env}->{'plack.skip-deflater'} = 1; # prevent 2x gzip
+ $ctx->{env}->{'qspawn.fallback'} = $code;
+ my $res = $ctx->{www}->coderepo->srv($ctx);
+ # for ->psgi_return_init_cb
+ $ctx->{env}->{'qspawn.wcb'} = $wcb;
+ $res; # CODE or ARRAY ref
+ } else {
+ [ $code, \@h ]
+ }
}
1;