1 # Copyright (C) all contributors <meta@public-inbox.org>
2 # License: GPLv2 or later <https://www.gnu.org/licenses/gpl-2.0.txt>
4 # Used to read files from a git repository without excessive forking.
5 # Used in our web interfaces as well as our -nntpd server.
6 # This is based on code in Git.pm which is GPLv2+, but modified to avoid
7 # dependence on environment variables for compatibility with mod_perl.
8 # There are also API changes to simplify our usage and data set.
9 package PublicInbox::Git;
12 use parent qw(Exporter);
14 use IO::Handle; # ->autoflush
15 use Errno qw(EINTR EAGAIN);
16 use File::Glob qw(bsd_glob GLOB_NOSORT);
18 use Time::HiRes qw(stat);
19 use PublicInbox::Spawn qw(popen_rd which);
20 use PublicInbox::Tmpfile;
21 use IO::Poll qw(POLLIN);
22 use Carp qw(croak carp);
24 use PublicInbox::DS qw(awaitpid);
25 our @EXPORT_OK = qw(git_unquote git_quote);
26 our $PIPE_BUFSIZ = 65536; # Linux default
28 our $RDTIMEO = 60_000; # milliseconds
29 our $async_warn; # true in read-only daemons
31 # 512: POSIX PIPE_BUF minimum (see pipe(7))
32 # 3: @$inflight is flattened [ $OID, $cb, $arg ]
33 # 65: SHA-256 hex size + "\n" in preparation for git using non-SHA1
34 use constant MAX_INFLIGHT => 512 * 3 / 65;
47 my %ESC_GIT = map { $GIT_ESC{$_} => $_ } keys %GIT_ESC;
49 # unquote pathnames used by git, see quote.c::unquote_c_style.c in git.git
51 return $_[0] unless ($_[0] =~ /\A"(.*)"\z/);
53 $_[0] =~ s!\\([\\"abfnrtv]|[0-3][0-7]{2})!$GIT_ESC{$1}//chr(oct($1))!ge;
58 if ($_[0] =~ s/([\\"\a\b\f\n\r\t\013]|[^[:print:]])/
59 '\\'.($ESC_GIT{$1}||sprintf("%03o",ord($1)))/egs) {
66 my ($class, $git_dir) = @_;
68 $git_dir =~ s!/*\z!!s;
69 # may contain {-tmp} field for File::Temp::Dir
70 bless { git_dir => $git_dir }, $class
74 my ($self, $path) = @_;
75 $self->{-git_path}->{$path} //= do {
77 chomp(my $str = $self->qx(qw(rev-parse --git-path), $path));
79 # git prior to 2.5.0 did not understand --git-path
80 if ($str eq "--git-path\n$path") {
81 $str = "$self->{git_dir}/$path";
87 sub alternates_changed {
89 my $alt = git_path($self, 'objects/info/alternates');
90 my @st = stat($alt) or return 0;
92 # can't rely on 'q' on some 32-bit builds, but `d' works
93 my $st = pack('dd', $st[10], $st[7]); # 10: ctime, 7: size
94 return 0 if ($self->{alt_st} // '') eq $st;
95 $self->{alt_st} = $st; # always a true value
99 $_[0]->{object_format} //= do {
100 my $fmt = $_[0]->qx(qw(config extensions.objectformat));
101 $fmt eq "sha256\n" ? \'sha256' : \undef;
107 my $fh = $self->{err_c} or return;
108 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
109 defined(sysread($fh, my $buf, -s $fh)) or
110 $self->fail("sysread failed: $!");
115 my ($self, $batch, $in, $out, $pid, $err) = @_;
117 if (defined $err) { # "err_c"
118 my $fh = $self->{$err};
119 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
120 truncate($fh, 0) or $self->fail("truncate failed: $!");
125 state $EXE_ST = ''; # pack('dd', st_ctime, st_size);
126 my $exe = which('git') // die "git not found in $ENV{PATH}";
127 my @st = stat($exe) or die "stat: $!";
128 my $st = pack('dd', $st[10], $st[7]);
130 if ($st ne $EXE_ST) {
131 my $rd = popen_rd([ $exe, '--version' ]);
132 my $v = readline($rd);
133 $v =~ /\b([0-9]+(?:\.[0-9]+){2})/ or die
134 "$exe --version output: $v # unparseable";
135 my @v = split(/\./, $1, 3);
136 $VER = ($v[0] << 24) | ($v[1] << 16) | $v[2];
140 # git 2.31.0+ supports -c core.abbrev=no, don't bother with
141 # core.abbrev=64 since not many releases had SHA-256 prior to 2.31
142 my $abbr = $VER < (2 << 24 | 31 << 16) ? 40 : 'no';
144 pipe(my ($out_r, $out_w)) or $self->fail("pipe failed: $!");
145 my $rdr = { 0 => $out_r, pgid => 0 };
146 my $gd = $self->{git_dir};
147 if ($gd =~ s!/([^/]+/[^/]+)\z!/!) {
151 my @cmd = ($exe, "--git-dir=$gd", '-c', "core.abbrev=$abbr",
154 my $id = "git.$self->{git_dir}$batch.err";
155 my $fh = tmpfile($id) or $self->fail("tmpfile($id): $!");
159 # see lib/PublicInbox/ProcessPipe.pm for why we don't use that here
160 my ($in_r, $p) = popen_rd(\@cmd, undef, $rdr);
161 awaitpid($self->{$pid} = $p, undef);
162 $self->{"$pid.owner"} = $$;
163 $out_w->autoflush(1);
164 if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ
165 fcntl($out_w, 1031, 4096);
166 fcntl($in_r, 1031, 4096) if $batch eq '--batch-check';
169 $self->{$out} = $out_w;
170 $self->{$in} = $in_r;
173 sub poll_in ($) { IO::Poll::_poll($RDTIMEO, fileno($_[0]), my $ev = POLLIN) }
176 my ($fh, $rbuf, $len) = @_;
177 my $left = $len - length($$rbuf);
180 $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf));
183 } elsif (defined($r)) { # EOF
186 next if ($! == EAGAIN and poll_in($fh));
187 next if $! == EINTR; # may be set by sysread or poll_in
188 return; # unrecoverable error
191 my $no_pad = substr($$rbuf, 0, $len, '');
195 sub my_readline ($$) {
196 my ($fh, $rbuf) = @_;
198 if ((my $n = index($$rbuf, "\n")) >= 0) {
199 return substr($$rbuf, 0, $n + 1, '');
201 my $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf))
204 # return whatever's left on EOF
205 return substr($$rbuf, 0, length($$rbuf)+1, '') if defined($r);
207 next if ($! == EAGAIN and poll_in($fh));
208 next if $! == EINTR; # may be set by sysread or poll_in
209 return; # unrecoverable error
213 sub cat_async_retry ($$) {
214 my ($self, $inflight) = @_;
216 # {inflight} may be non-existent, but if it isn't we delete it
217 # here to prevent cleanup() from waiting:
218 delete $self->{inflight};
221 $self->{inflight} = $inflight;
222 batch_prepare($self);
224 for (my $i = 0; $i < @$inflight; $i += 3) {
225 $buf .= "$inflight->[$i]\n";
227 $self->{out}->blocking(1); # brand new pipe, should never block
228 print { $self->{out} } $buf or $self->fail("write error: $!");
229 $self->{out}->blocking(0);
230 my $req = shift @$inflight;
231 unshift(@$inflight, \$req); # \$ref to indicate retried
233 cat_async_step($self, $inflight); # take one step
236 sub cat_async_step ($$) {
237 my ($self, $inflight) = @_;
238 die 'BUG: inflight empty or odd' if scalar(@$inflight) < 3;
239 my ($req, $cb, $arg) = @$inflight[0, 1, 2];
240 my $rbuf = delete($self->{rbuf}) // \(my $new = '');
241 my ($bref, $oid, $type, $size);
242 my $head = my_readline($self->{in}, $rbuf);
243 # ->fail may be called via Gcf2Client.pm
244 if ($head =~ /^([0-9a-f]{40,}) (\S+) ([0-9]+)$/) {
245 ($oid, $type, $size) = ($1, $2, $3 + 0);
246 $bref = my_read($self->{in}, $rbuf, $size + 1) or
247 $self->fail(defined($bref) ? 'read EOF' : "read: $!");
248 chop($$bref) eq "\n" or $self->fail('LF missing after blob');
249 } elsif ($head =~ s/ missing\n//s) {
251 # ref($req) indicates it's already been retried
252 # -gcf2 retries internally, so it never hits this path:
253 if (!ref($req) && !$in_cleanup && $self->alternates_changed) {
254 return cat_async_retry($self, $inflight);
257 $oid = ref($req) ? $$req : $req if $oid eq '';
259 my $err = $! ? " ($!)" : '';
260 $self->fail("bad result from async cat-file: $head$err");
262 $self->{rbuf} = $rbuf if $$rbuf ne '';
263 splice(@$inflight, 0, 3); # don't retry $cb on ->fail
264 eval { $cb->($bref, $oid, $type, $size, $arg) };
265 async_err($self, $req, $oid, $@, 'cat') if $@;
268 sub cat_async_wait ($) {
270 my $inflight = $self->{inflight} or return;
271 while (scalar(@$inflight)) {
272 cat_async_step($self, $inflight);
276 sub batch_prepare ($) {
277 _bidi_pipe($_[0], qw(--batch in out pid));
281 my ($bref, $oid, $type, $size, $result) = @_;
282 @$result = ($bref, $oid, $type, $size);
286 my ($self, $oid) = @_;
288 cat_async($self, $oid, \&_cat_file_cb, $result);
289 cat_async_wait($self);
290 wantarray ? @$result : $result->[0];
293 sub check_async_step ($$) {
294 my ($self, $inflight_c) = @_;
295 die 'BUG: inflight empty or odd' if scalar(@$inflight_c) < 3;
296 my ($req, $cb, $arg) = @$inflight_c[0, 1, 2];
297 my $rbuf = delete($self->{rbuf_c}) // \(my $new = '');
298 chomp(my $line = my_readline($self->{in_c}, $rbuf));
299 my ($hex, $type, $size) = split(/ /, $line);
301 # Future versions of git.git may have type=ambiguous, but for now,
302 # we must handle 'dangling' below (and maybe some other oddball
304 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
305 if ($hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop') {
306 my $ret = my_read($self->{in_c}, $rbuf, $type + 1);
307 $self->fail(defined($ret) ? 'read EOF' : "read: $!") if !$ret;
309 $self->{rbuf_c} = $rbuf if $$rbuf ne '';
310 splice(@$inflight_c, 0, 3); # don't retry $cb on ->fail
311 eval { $cb->($hex, $type, $size, $arg, $self) };
312 async_err($self, $req, $hex, $@, 'check') if $@;
315 sub check_async_wait ($) {
317 my $inflight_c = $self->{inflight_c} or return;
318 while (scalar(@$inflight_c)) {
319 check_async_step($self, $inflight_c);
323 sub check_async_begin ($) {
325 cleanup($self) if alternates_changed($self);
326 _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c));
327 die 'BUG: already in async check' if $self->{inflight_c};
328 $self->{inflight_c} = [];
332 my ($self, $out, $buf, $read_step, $inflight) = @_;
333 $read_step->($self, $inflight) while @$inflight >= MAX_INFLIGHT;
335 my $w = syswrite($out, $buf);
337 return if $w == length($buf);
338 substr($buf, 0, $w, ''); # sv_chop
339 } elsif ($! != EAGAIN) {
340 $self->fail("write: $!");
342 $read_step->($self, $inflight);
346 sub check_async ($$$$) {
347 my ($self, $oid, $cb, $arg) = @_;
348 my $inflight_c = $self->{inflight_c} // check_async_begin($self);
349 write_all($self, $self->{out_c}, $oid."\n",
350 \&check_async_step, $inflight_c);
351 push(@$inflight_c, $oid, $cb, $arg);
354 sub _check_cb { # check_async callback
355 my ($hex, $type, $size, $result) = @_;
356 @$result = ($hex, $type, $size);
360 my ($self, $oid) = @_;
362 check_async($self, $oid, \&_check_cb, $result);
363 check_async_wait($self);
364 my ($hex, $type, $size) = @$result;
366 # Future versions of git.git may show 'ambiguous', but for now,
367 # we must handle 'dangling' below (and maybe some other oddball
369 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
370 return if $type eq 'missing' || $type eq 'ambiguous';
371 return if $hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop';
372 ($hex, $type, $size);
376 my ($self, $pid, @rest) = @_; # rest = rbuf, in, out, err
377 my ($p) = delete @$self{($pid, @rest)};
379 # GitAsyncCat::event_step may delete {$pid}
380 awaitpid($p) if defined($p) && $$ == $self->{"$pid.owner"};
383 sub async_abort ($) {
385 while (scalar(@{$self->{inflight_c} // []}) ||
386 scalar(@{$self->{inflight} // []})) {
387 for my $c ('', '_c') {
388 my $q = $self->{"inflight$c"} or next;
390 my ($req, $cb, $arg) = splice(@$q, 0, 3);
391 $req = $$req if ref($req);
392 $req =~ s/ .*//; # drop git_dir for Gcf2Client
393 eval { $cb->(undef, $req, undef, undef, $arg) };
394 warn "E: (in abort) $req: $@" if $@;
396 delete $self->{"inflight$c"};
397 delete $self->{"rbuf$c"};
403 sub fail { # may be augmented in subclasses
404 my ($self, $msg) = @_;
406 croak(ref($self) . ' ' . ($self->{git_dir} // '') . ": $msg");
409 sub async_err ($$$$$) {
410 my ($self, $req, $oid, $err, $action) = @_;
411 $req = $$req if ref($req); # retried
412 my $msg = "E: $action $req ($oid): $err";
413 $async_warn ? carp($msg) : $self->fail($msg);
416 # $git->popen(qw(show f00)); # or
417 # $git->popen(qw(show f00), { GIT_CONFIG => ... }, { 2 => ... });
419 my ($self, $cmd) = splice(@_, 0, 2);
420 $cmd = [ 'git', "--git-dir=$self->{git_dir}",
421 ref($cmd) ? @$cmd : ($cmd, grep { defined && !ref } @_) ];
422 popen_rd($cmd, grep { !defined || ref } @_); # env and opt
425 # same args as popen above
430 close $fh; # caller should check $?
435 close $fh; # caller should check $?
443 substr($_, length('--max-age='), -1)
444 } $self->qx('rev-parse', map { "--since=$_" } @_);
447 # check_async and cat_async may trigger the other, so ensure they're
448 # both completely done by using this:
449 sub async_wait_all ($) {
451 while (scalar(@{$self->{inflight_c} // []}) ||
452 scalar(@{$self->{inflight} // []})) {
453 check_async_wait($self);
454 cat_async_wait($self);
458 # returns true if there are pending "git cat-file" processes
460 my ($self, $lazy) = @_;
461 return 1 if $lazy && (scalar(@{$self->{inflight_c} // []}) ||
462 scalar(@{$self->{inflight} // []}));
463 local $in_cleanup = 1;
464 delete $self->{async_cat};
465 delete $self->{async_chk};
466 async_wait_all($self);
467 delete $self->{inflight};
468 delete $self->{inflight_c};
469 _destroy($self, qw(pid rbuf in out));
470 _destroy($self, qw(pid_c rbuf_c in_c out_c err_c));
474 # assuming a well-maintained repo, this should be a somewhat
475 # accurate estimation of its size
476 # TODO: show this in the WWW UI as a hint to potential cloners
480 my $pack_dir = git_path($self, 'objects/pack');
481 foreach my $p (bsd_glob("$pack_dir/*.pack", GLOB_NOSORT)) {
487 sub DESTROY { cleanup(@_) }
490 # don't show full FS path, basename should be OK:
491 $_[0]->{nick} // ($_[0]->{git_dir} =~ m!/([^/]+?)(?:/*\.git/*)?\z! ?
495 sub host_prefix_url ($$) {
496 my ($env, $url) = @_;
497 return $url if index($url, '//') >= 0;
498 my $host_port = $env->{HTTP_HOST} //
499 "$env->{SERVER_NAME}:$env->{SERVER_PORT}";
500 my $sn = $env->{SCRIPT_NAME} // '';
501 "$env->{'psgi.url_scheme'}://$host_port$sn/$url";
504 sub base_url { # for coderepos, PSGI-only
505 my ($self, $env) = @_; # env - PSGI env
506 my $nick = $self->{nick} // return undef;
507 my $url = host_prefix_url($env, '');
508 # for mount in Plack::Builder
509 $url .= '/' if substr($url, -1, 1) ne '/';
516 my ($self, $env) = @_;
517 if (my $urls = $self->{cgit_url}) {
518 map { host_prefix_url($env, $_) } @$urls;
520 (base_url($self, $env) // '???');
524 sub cat_async_begin {
526 cleanup($self) if $self->alternates_changed;
527 $self->batch_prepare;
528 die 'BUG: already in async' if $self->{inflight};
529 $self->{inflight} = [];
532 sub cat_async ($$$;$) {
533 my ($self, $oid, $cb, $arg) = @_;
534 my $inflight = $self->{inflight} // cat_async_begin($self);
535 write_all($self, $self->{out}, $oid."\n", \&cat_async_step, $inflight);
536 push(@$inflight, $oid, $cb, $arg);
539 # returns the modified time of a git repo, same as the "modified" field
540 # of a grokmirror manifest
542 # committerdate:unix is git 2.9.4+ (2017-05-05), so using raw instead
543 my $fh = popen($_[0], qw[for-each-ref --sort=-committerdate
544 --format=%(committerdate:raw) --count=1]);
545 (split(/ /, <$fh> // time))[0] + 0; # integerize for JSON
550 open(my $fh, '<', $path) or return '';
556 my $desc = try_cat($_[0]);
559 $desc =~ s/\s+/ /smg;
560 $desc eq '' ? undef : $desc;
564 cat_desc("$_[0]->{git_dir}/description") // 'Unnamed repository';
568 my ($self, $env) = @_;
569 $self->{cloneurl} // do {
570 my @urls = split(/\s+/s, try_cat("$self->{git_dir}/cloneurl"));
571 scalar(@urls) ? ($self->{cloneurl} = \@urls) : undef;
572 } // [ substr(base_url($self, $env), 0, -1) ];
575 # for grokmirror, which doesn't read gitweb.description
576 # templates/hooks--update.sample and git-multimail in git.git
577 # only match "Unnamed repository", not the full contents of
578 # templates/this--description in git.git
580 my ($self, $epoch, $default_desc) = @_;
581 my $fh = $self->popen('show-ref');
582 my $dig = Digest::SHA->new(1);
583 while (read($fh, my $buf, 65536)) {
586 close $fh or return; # empty, uninitialized git repo
587 undef $fh; # for open, below
588 my $git_dir = $self->{git_dir};
590 fingerprint => $dig->hexdigest,
592 modified => modified($self),
594 chomp(my $owner = $self->qx('config', 'gitweb.owner'));
595 utf8::decode($owner);
596 $ent->{owner} = $owner eq '' ? undef : $owner;
597 my $desc = description($self);
598 if (defined $epoch && index($desc, 'Unnamed repository') == 0) {
599 $desc = "$default_desc [epoch $epoch]";
601 $ent->{description} = $desc;
602 if (open($fh, '<', "$git_dir/objects/info/alternates")) {
603 # n.b.: GitPython doesn't seem to handle comments or C-quoted
604 # strings like native git does; and we don't for now, either.
606 chomp(my @alt = <$fh>);
608 # grokmirror only supports 1 alternate for "reference",
609 if (scalar(@alt) == 1) {
610 my $objdir = "$git_dir/objects";
611 my $ref = File::Spec->rel2abs($alt[0], $objdir);
612 $ref =~ s!/[^/]+/?\z!!; # basename
613 $ent->{reference} = $ref;
619 # returns true if there are pending cat-file processes
620 sub cleanup_if_unlinked {
622 return cleanup($self, 1) if $^O ne 'linux';
623 # Linux-specific /proc/$PID/maps access
624 # TODO: support this inside git.git
626 for my $fld (qw(pid pid_c)) {
627 my $pid = $self->{$fld} // next;
628 open my $fh, '<', "/proc/$pid/maps" or return cleanup($self, 1);
630 # n.b. we do not restart for unlinked multi-pack-index
631 # since it's not too huge, and the startup cost may
633 /\.(?:idx|pack) \(deleted\)$/ and
634 return cleanup($self, 1);
647 PublicInbox::Git - git wrapper
655 use PublicInbox::Git;
656 chomp(my $git_dir = `git rev-parse --git-dir`);
657 $git_dir or die "GIT_DIR= must be specified\n";
658 my $git = PublicInbox::Git->new($git_dir);
662 Unstable API outside of the L</new> method.
663 It requires L<git(1)> to be installed.
671 my $git = PublicInbox::Git->new($git_dir);
673 Initialize a new PublicInbox::Git object for use with L<PublicInbox::Import>
674 This is the only public API method we support. Everything else
675 in this module is subject to change.
679 L<Git>, L<PublicInbox::Import>
683 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
685 The mail archives are hosted at L<https://public-inbox.org/meta/>
689 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
691 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>