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 ENOENT);
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): $!");
159 my ($in_r, $p) = popen_rd(\@cmd, undef, $rdr);
160 awaitpid($self->{$pid} = $p, undef);
161 $self->{"$pid.owner"} = $$;
162 $out_w->autoflush(1);
163 if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ
164 fcntl($out_w, 1031, 4096);
165 fcntl($in_r, 1031, 4096) if $batch eq '--batch-check';
168 $self->{$out} = $out_w;
169 $self->{$in} = $in_r;
172 sub poll_in ($) { IO::Poll::_poll($RDTIMEO, fileno($_[0]), my $ev = POLLIN) }
175 my ($fh, $rbuf, $len) = @_;
176 my $left = $len - length($$rbuf);
179 $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf));
182 } elsif (defined($r)) { # EOF
185 next if ($! == EAGAIN and poll_in($fh));
186 next if $! == EINTR; # may be set by sysread or poll_in
187 return; # unrecoverable error
190 my $no_pad = substr($$rbuf, 0, $len, '');
194 sub my_readline ($$) {
195 my ($fh, $rbuf) = @_;
197 if ((my $n = index($$rbuf, "\n")) >= 0) {
198 return substr($$rbuf, 0, $n + 1, '');
200 my $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf))
203 # return whatever's left on EOF
204 return substr($$rbuf, 0, length($$rbuf)+1, '') if defined($r);
206 next if ($! == EAGAIN and poll_in($fh));
207 next if $! == EINTR; # may be set by sysread or poll_in
208 return; # unrecoverable error
212 sub cat_async_retry ($$) {
213 my ($self, $inflight) = @_;
215 # {inflight} may be non-existent, but if it isn't we delete it
216 # here to prevent cleanup() from waiting:
217 delete $self->{inflight};
220 $self->{inflight} = $inflight;
221 batch_prepare($self);
223 for (my $i = 0; $i < @$inflight; $i += 3) {
224 $buf .= "$inflight->[$i]\n";
226 $self->{out}->blocking(1); # brand new pipe, should never block
227 print { $self->{out} } $buf or $self->fail("write error: $!");
228 $self->{out}->blocking(0);
229 my $req = shift @$inflight;
230 unshift(@$inflight, \$req); # \$ref to indicate retried
232 cat_async_step($self, $inflight); # take one step
235 sub cat_async_step ($$) {
236 my ($self, $inflight) = @_;
237 die 'BUG: inflight empty or odd' if scalar(@$inflight) < 3;
238 my ($req, $cb, $arg) = @$inflight[0, 1, 2];
239 my $rbuf = delete($self->{rbuf}) // \(my $new = '');
240 my ($bref, $oid, $type, $size);
241 my $head = my_readline($self->{in}, $rbuf);
242 # ->fail may be called via Gcf2Client.pm
243 if ($head =~ /^([0-9a-f]{40,}) (\S+) ([0-9]+)$/) {
244 ($oid, $type, $size) = ($1, $2, $3 + 0);
245 $bref = my_read($self->{in}, $rbuf, $size + 1) or
246 $self->fail(defined($bref) ? 'read EOF' : "read: $!");
247 chop($$bref) eq "\n" or $self->fail('LF missing after blob');
248 } elsif ($head =~ s/ missing\n//s) {
250 # ref($req) indicates it's already been retried
251 # -gcf2 retries internally, so it never hits this path:
252 if (!ref($req) && !$in_cleanup && $self->alternates_changed) {
253 return cat_async_retry($self, $inflight);
256 $oid = ref($req) ? $$req : $req if $oid eq '';
258 my $err = $! ? " ($!)" : '';
259 $self->fail("bad result from async cat-file: $head$err");
261 $self->{rbuf} = $rbuf if $$rbuf ne '';
262 splice(@$inflight, 0, 3); # don't retry $cb on ->fail
263 eval { $cb->($bref, $oid, $type, $size, $arg) };
264 async_err($self, $req, $oid, $@, 'cat') if $@;
267 sub cat_async_wait ($) {
269 my $inflight = $self->{inflight} or return;
270 while (scalar(@$inflight)) {
271 cat_async_step($self, $inflight);
275 sub batch_prepare ($) {
276 _bidi_pipe($_[0], qw(--batch in out pid));
280 my ($bref, $oid, $type, $size, $result) = @_;
281 @$result = ($bref, $oid, $type, $size);
285 my ($self, $oid) = @_;
287 cat_async($self, $oid, \&_cat_file_cb, $result);
288 cat_async_wait($self);
289 wantarray ? @$result : $result->[0];
292 sub check_async_step ($$) {
293 my ($self, $inflight_c) = @_;
294 die 'BUG: inflight empty or odd' if scalar(@$inflight_c) < 3;
295 my ($req, $cb, $arg) = @$inflight_c[0, 1, 2];
296 my $rbuf = delete($self->{rbuf_c}) // \(my $new = '');
297 chomp(my $line = my_readline($self->{in_c}, $rbuf));
298 my ($hex, $type, $size) = split(/ /, $line);
300 # Future versions of git.git may have type=ambiguous, but for now,
301 # we must handle 'dangling' below (and maybe some other oddball
303 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
304 if ($hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop') {
305 my $ret = my_read($self->{in_c}, $rbuf, $type + 1);
306 $self->fail(defined($ret) ? 'read EOF' : "read: $!") if !$ret;
308 $self->{rbuf_c} = $rbuf if $$rbuf ne '';
309 splice(@$inflight_c, 0, 3); # don't retry $cb on ->fail
310 eval { $cb->($hex, $type, $size, $arg, $self) };
311 async_err($self, $req, $hex, $@, 'check') if $@;
314 sub check_async_wait ($) {
316 my $inflight_c = $self->{inflight_c} or return;
317 while (scalar(@$inflight_c)) {
318 check_async_step($self, $inflight_c);
322 sub check_async_begin ($) {
324 cleanup($self) if alternates_changed($self);
325 _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c));
326 die 'BUG: already in async check' if $self->{inflight_c};
327 $self->{inflight_c} = [];
331 my ($self, $out, $buf, $read_step, $inflight) = @_;
332 $read_step->($self, $inflight) while @$inflight >= MAX_INFLIGHT;
334 my $w = syswrite($out, $buf);
336 return if $w == length($buf);
337 substr($buf, 0, $w, ''); # sv_chop
338 } elsif ($! != EAGAIN) {
339 $self->fail("write: $!");
341 $read_step->($self, $inflight);
345 sub check_async ($$$$) {
346 my ($self, $oid, $cb, $arg) = @_;
347 my $inflight_c = $self->{inflight_c} // check_async_begin($self);
348 write_all($self, $self->{out_c}, $oid."\n",
349 \&check_async_step, $inflight_c);
350 push(@$inflight_c, $oid, $cb, $arg);
353 sub _check_cb { # check_async callback
354 my ($hex, $type, $size, $result) = @_;
355 @$result = ($hex, $type, $size);
359 my ($self, $oid) = @_;
361 check_async($self, $oid, \&_check_cb, $result);
362 check_async_wait($self);
363 my ($hex, $type, $size) = @$result;
365 # Future versions of git.git may show 'ambiguous', but for now,
366 # we must handle 'dangling' below (and maybe some other oddball
368 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
369 return if $type eq 'missing' || $type eq 'ambiguous';
370 return if $hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop';
371 ($hex, $type, $size);
375 my ($self, $rbuf, $in, $out, $pid, $err) = @_;
376 delete @$self{($rbuf, $in, $out)};
377 delete $self->{$err} if $err; # `err_c'
379 # GitAsyncCat::event_step may delete {$pid}
380 my $p = delete($self->{$pid}) // return;
381 awaitpid($p) if $$ == $self->{"$pid.owner"};
384 sub async_abort ($) {
386 while (scalar(@{$self->{inflight_c} // []}) ||
387 scalar(@{$self->{inflight} // []})) {
388 for my $c ('', '_c') {
389 my $q = $self->{"inflight$c"} or next;
391 my ($req, $cb, $arg) = splice(@$q, 0, 3);
392 $req = $$req if ref($req);
393 $req =~ s/ .*//; # drop git_dir for Gcf2Client
394 eval { $cb->(undef, $req, undef, undef, $arg) };
395 warn "E: (in abort) $req: $@" if $@;
397 delete $self->{"inflight$c"};
398 delete $self->{"rbuf$c"};
404 sub fail { # may be augmented in subclasses
405 my ($self, $msg) = @_;
407 croak(ref($self) . ' ' . ($self->{git_dir} // '') . ": $msg");
410 sub async_err ($$$$$) {
411 my ($self, $req, $oid, $err, $action) = @_;
412 $req = $$req if ref($req); # retried
413 my $msg = "E: $action $req ($oid): $err";
414 $async_warn ? carp($msg) : $self->fail($msg);
417 # $git->popen(qw(show f00)); # or
418 # $git->popen(qw(show f00), { GIT_CONFIG => ... }, { 2 => ... });
420 my ($self, $cmd) = splice(@_, 0, 2);
421 $cmd = [ 'git', "--git-dir=$self->{git_dir}",
422 ref($cmd) ? @$cmd : ($cmd, grep { defined && !ref } @_) ];
423 popen_rd($cmd, grep { !defined || ref } @_); # env and opt
426 # same args as popen above
431 close $fh; # caller should check $?
436 close $fh; # caller should check $?
444 substr($_, length('--max-age='), -1)
445 } $self->qx('rev-parse', map { "--since=$_" } @_);
448 # check_async and cat_async may trigger the other, so ensure they're
449 # both completely done by using this:
450 sub async_wait_all ($) {
452 while (scalar(@{$self->{inflight_c} // []}) ||
453 scalar(@{$self->{inflight} // []})) {
454 check_async_wait($self);
455 cat_async_wait($self);
459 # returns true if there are pending "git cat-file" processes
461 my ($self, $lazy) = @_;
462 return 1 if $lazy && (scalar(@{$self->{inflight_c} // []}) ||
463 scalar(@{$self->{inflight} // []}));
464 local $in_cleanup = 1;
465 delete $self->{async_cat};
466 delete $self->{async_chk};
467 async_wait_all($self);
468 delete $self->{inflight};
469 delete $self->{inflight_c};
470 _destroy($self, qw(rbuf in out pid));
471 _destroy($self, qw(rbuf_c in_c out_c pid_c err_c));
475 # assuming a well-maintained repo, this should be a somewhat
476 # accurate estimation of its size
477 # TODO: show this in the WWW UI as a hint to potential cloners
481 my $pack_dir = git_path($self, 'objects/pack');
482 foreach my $p (bsd_glob("$pack_dir/*.pack", GLOB_NOSORT)) {
488 sub DESTROY { cleanup(@_) }
491 # don't show full FS path, basename should be OK:
492 $_[0]->{nick} // ($_[0]->{git_dir} =~ m!/([^/]+?)(?:/*\.git/*)?\z! ?
496 sub host_prefix_url ($$) {
497 my ($env, $url) = @_;
498 return $url if index($url, '//') >= 0;
499 my $host_port = $env->{HTTP_HOST} //
500 "$env->{SERVER_NAME}:$env->{SERVER_PORT}";
501 my $sn = $env->{SCRIPT_NAME} // '';
502 "$env->{'psgi.url_scheme'}://$host_port$sn/$url";
505 sub base_url { # for coderepos, PSGI-only
506 my ($self, $env) = @_; # env - PSGI env
507 my $nick = $self->{nick} // return undef;
508 my $url = host_prefix_url($env, '');
509 # for mount in Plack::Builder
510 $url .= '/' if substr($url, -1, 1) ne '/';
517 my ($self, $env) = @_;
518 if (my $urls = $self->{cgit_url}) {
519 map { host_prefix_url($env, $_) } @$urls;
521 (base_url($self, $env) // '???');
525 sub cat_async_begin {
527 cleanup($self) if $self->alternates_changed;
528 $self->batch_prepare;
529 die 'BUG: already in async' if $self->{inflight};
530 $self->{inflight} = [];
533 sub cat_async ($$$;$) {
534 my ($self, $oid, $cb, $arg) = @_;
535 my $inflight = $self->{inflight} // cat_async_begin($self);
536 write_all($self, $self->{out}, $oid."\n", \&cat_async_step, $inflight);
537 push(@$inflight, $oid, $cb, $arg);
540 # returns the modified time of a git repo, same as the "modified" field
541 # of a grokmirror manifest
543 # committerdate:unix is git 2.9.4+ (2017-05-05), so using raw instead
544 my $fh = popen($_[0], qw[for-each-ref --sort=-committerdate
545 --format=%(committerdate:raw) --count=1]);
546 (split(/ /, <$fh> // time))[0] + 0; # integerize for JSON
551 open(my $fh, '<', $path) or return '';
557 my $desc = try_cat($_[0]);
560 $desc =~ s/\s+/ /smg;
561 $desc eq '' ? undef : $desc;
565 cat_desc("$_[0]->{git_dir}/description") // 'Unnamed repository';
569 my ($self, $env) = @_;
570 $self->{cloneurl} // do {
571 my @urls = split(/\s+/s, try_cat("$self->{git_dir}/cloneurl"));
572 scalar(@urls) ? ($self->{cloneurl} = \@urls) : undef;
573 } // [ substr(base_url($self, $env), 0, -1) ];
576 # for grokmirror, which doesn't read gitweb.description
577 # templates/hooks--update.sample and git-multimail in git.git
578 # only match "Unnamed repository", not the full contents of
579 # templates/this--description in git.git
581 my ($self, $epoch, $default_desc) = @_;
582 my $fh = $self->popen('show-ref');
583 my $dig = Digest::SHA->new(1);
584 while (read($fh, my $buf, 65536)) {
587 close $fh or return; # empty, uninitialized git repo
588 undef $fh; # for open, below
589 my $git_dir = $self->{git_dir};
591 fingerprint => $dig->hexdigest,
593 modified => modified($self),
595 chomp(my $owner = $self->qx('config', 'gitweb.owner'));
596 utf8::decode($owner);
597 $ent->{owner} = $owner eq '' ? undef : $owner;
598 my $desc = description($self);
599 if (defined $epoch && index($desc, 'Unnamed repository') == 0) {
600 $desc = "$default_desc [epoch $epoch]";
602 $ent->{description} = $desc;
603 if (open($fh, '<', "$git_dir/objects/info/alternates")) {
604 # n.b.: GitPython doesn't seem to handle comments or C-quoted
605 # strings like native git does; and we don't for now, either.
607 chomp(my @alt = <$fh>);
609 # grokmirror only supports 1 alternate for "reference",
610 if (scalar(@alt) == 1) {
611 my $objdir = "$git_dir/objects";
612 my $ref = File::Spec->rel2abs($alt[0], $objdir);
613 $ref =~ s!/[^/]+/?\z!!; # basename
614 $ent->{reference} = $ref;
620 # returns true if there are pending cat-file processes
621 sub cleanup_if_unlinked {
623 return cleanup($self, 1) if $^O ne 'linux';
624 # Linux-specific /proc/$PID/maps access
625 # TODO: support this inside git.git
627 for my $fld (qw(pid pid_c)) {
628 my $pid = $self->{$fld} // next;
629 open my $fh, '<', "/proc/$pid/maps" or return cleanup($self, 1);
631 # n.b. we do not restart for unlinked multi-pack-index
632 # since it's not too huge, and the startup cost may
634 /\.(?:idx|pack) \(deleted\)$/ and
635 return cleanup($self, 1);
648 PublicInbox::Git - git wrapper
656 use PublicInbox::Git;
657 chomp(my $git_dir = `git rev-parse --git-dir`);
658 $git_dir or die "GIT_DIR= must be specified\n";
659 my $git = PublicInbox::Git->new($git_dir);
663 Unstable API outside of the L</new> method.
664 It requires L<git(1)> to be installed.
672 my $git = PublicInbox::Git->new($git_dir);
674 Initialize a new PublicInbox::Git object for use with L<PublicInbox::Import>
675 This is the only public API method we support. Everything else
676 in this module is subject to change.
680 L<Git>, L<PublicInbox::Import>
684 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
686 The mail archives are hosted at L<https://public-inbox.org/meta/>
690 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
692 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>