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 spawn);
20 use PublicInbox::Tmpfile;
21 use IO::Poll qw(POLLIN);
22 use Carp qw(croak carp);
24 use PublicInbox::DS qw(dwaitpid);
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;
50 # unquote pathnames used by git, see quote.c::unquote_c_style.c in git.git
52 return $_[0] unless ($_[0] =~ /\A"(.*)"\z/);
54 $_[0] =~ s!\\([\\"abfnrtv]|[0-3][0-7]{2})!$GIT_ESC{$1}//chr(oct($1))!ge;
59 if ($_[0] =~ s/([\\"\a\b\f\n\r\t\013]|[^[:print:]])/
60 '\\'.($ESC_GIT{$1}||sprintf("%03o",ord($1)))/egs) {
67 my ($class, $git_dir) = @_;
69 $git_dir =~ s!/*\z!!s;
70 # may contain {-tmp} field for File::Temp::Dir
71 bless { git_dir => $git_dir }, $class
75 my ($self, $path) = @_;
76 $self->{-git_path}->{$path} //= do {
78 chomp(my $str = $self->qx(qw(rev-parse --git-path), $path));
80 # git prior to 2.5.0 did not understand --git-path
81 if ($str eq "--git-path\n$path") {
82 $str = "$self->{git_dir}/$path";
88 sub alternates_changed {
90 my $alt = git_path($self, 'objects/info/alternates');
91 my @st = stat($alt) or return 0;
93 # can't rely on 'q' on some 32-bit builds, but `d' works
94 my $st = pack('dd', $st[10], $st[7]); # 10: ctime, 7: size
95 return 0 if ($self->{alt_st} // '') eq $st;
96 $self->{alt_st} = $st; # always a true value
100 $_[0]->{object_format} //= do {
101 my $fmt = $_[0]->qx(qw(config extensions.objectformat));
102 $fmt eq "sha256\n" ? \'sha256' : \undef;
108 my $fh = $self->{err_c} or return;
109 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
110 defined(sysread($fh, my $buf, -s $fh)) or
111 $self->fail("sysread failed: $!");
116 my ($self, $batch, $in, $out, $pid, $err) = @_;
118 if (defined $err) { # "err_c"
119 my $fh = $self->{$err};
120 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
121 truncate($fh, 0) or $self->fail("truncate failed: $!");
125 pipe(my ($out_r, $out_w)) or $self->fail("pipe failed: $!");
126 my $rdr = { 0 => $out_r, pgid => 0 };
127 my $gd = $self->{git_dir};
128 if ($gd =~ s!/([^/]+/[^/]+)\z!/!) {
132 my @cmd = (qw(git), "--git-dir=$gd",
133 qw(-c core.abbrev=40 cat-file), $batch);
135 my $id = "git.$self->{git_dir}$batch.err";
136 my $fh = tmpfile($id) or $self->fail("tmpfile($id): $!");
140 my ($in_r, $p) = popen_rd(\@cmd, undef, $rdr);
142 $self->{"$pid.owner"} = $$;
143 $out_w->autoflush(1);
144 if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ
145 fcntl($out_w, 1031, 4096);
146 fcntl($in_r, 1031, 4096) if $batch eq '--batch-check';
149 $self->{$out} = $out_w;
150 $self->{$in} = $in_r;
153 sub poll_in ($) { IO::Poll::_poll($RDTIMEO, fileno($_[0]), my $ev = POLLIN) }
156 my ($fh, $rbuf, $len) = @_;
157 my $left = $len - length($$rbuf);
160 $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf));
163 } elsif (defined($r)) { # EOF
166 next if ($! == EAGAIN and poll_in($fh));
167 next if $! == EINTR; # may be set by sysread or poll_in
168 return; # unrecoverable error
171 my $no_pad = substr($$rbuf, 0, $len, '');
175 sub my_readline ($$) {
176 my ($fh, $rbuf) = @_;
178 if ((my $n = index($$rbuf, "\n")) >= 0) {
179 return substr($$rbuf, 0, $n + 1, '');
181 my $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf))
184 # return whatever's left on EOF
185 return substr($$rbuf, 0, length($$rbuf)+1, '') if defined($r);
187 next if ($! == EAGAIN and poll_in($fh));
188 next if $! == EINTR; # may be set by sysread or poll_in
189 return; # unrecoverable error
193 sub cat_async_retry ($$) {
194 my ($self, $inflight) = @_;
196 # {inflight} may be non-existent, but if it isn't we delete it
197 # here to prevent cleanup() from waiting:
198 delete $self->{inflight};
201 $self->{inflight} = $inflight;
202 batch_prepare($self);
204 for (my $i = 0; $i < @$inflight; $i += 3) {
205 $buf .= "$inflight->[$i]\n";
207 $self->{out}->blocking(1); # brand new pipe, should never block
208 print { $self->{out} } $buf or $self->fail("write error: $!");
209 $self->{out}->blocking(0);
210 my $req = shift @$inflight;
211 unshift(@$inflight, \$req); # \$ref to indicate retried
213 cat_async_step($self, $inflight); # take one step
216 sub cat_async_step ($$) {
217 my ($self, $inflight) = @_;
218 die 'BUG: inflight empty or odd' if scalar(@$inflight) < 3;
219 my ($req, $cb, $arg) = @$inflight[0, 1, 2];
220 my $rbuf = delete($self->{rbuf}) // \(my $new = '');
221 my ($bref, $oid, $type, $size);
222 my $head = my_readline($self->{in}, $rbuf);
223 # ->fail may be called via Gcf2Client.pm
224 if ($head =~ /^([0-9a-f]{40,}) (\S+) ([0-9]+)$/) {
225 ($oid, $type, $size) = ($1, $2, $3 + 0);
226 $bref = my_read($self->{in}, $rbuf, $size + 1) or
227 $self->fail(defined($bref) ? 'read EOF' : "read: $!");
228 chop($$bref) eq "\n" or $self->fail('LF missing after blob');
229 } elsif ($head =~ s/ missing\n//s) {
231 # ref($req) indicates it's already been retried
232 # -gcf2 retries internally, so it never hits this path:
233 if (!ref($req) && !$in_cleanup && $self->alternates_changed) {
234 return cat_async_retry($self, $inflight);
237 $oid = ref($req) ? $$req : $req if $oid eq '';
239 my $err = $! ? " ($!)" : '';
240 $self->fail("bad result from async cat-file: $head$err");
242 $self->{rbuf} = $rbuf if $$rbuf ne '';
243 splice(@$inflight, 0, 3); # don't retry $cb on ->fail
244 eval { $cb->($bref, $oid, $type, $size, $arg) };
245 async_err($self, $req, $oid, $@, 'cat') if $@;
248 sub cat_async_wait ($) {
250 my $inflight = $self->{inflight} or return;
251 while (scalar(@$inflight)) {
252 cat_async_step($self, $inflight);
256 sub batch_prepare ($) {
257 _bidi_pipe($_[0], qw(--batch in out pid));
261 my ($bref, $oid, $type, $size, $result) = @_;
262 @$result = ($bref, $oid, $type, $size);
266 my ($self, $oid) = @_;
268 cat_async($self, $oid, \&_cat_file_cb, $result);
269 cat_async_wait($self);
270 wantarray ? @$result : $result->[0];
273 sub check_async_step ($$) {
274 my ($self, $inflight_c) = @_;
275 die 'BUG: inflight empty or odd' if scalar(@$inflight_c) < 3;
276 my ($req, $cb, $arg) = @$inflight_c[0, 1, 2];
277 my $rbuf = delete($self->{rbuf_c}) // \(my $new = '');
278 chomp(my $line = my_readline($self->{in_c}, $rbuf));
279 my ($hex, $type, $size) = split(/ /, $line);
281 # Future versions of git.git may have type=ambiguous, but for now,
282 # we must handle 'dangling' below (and maybe some other oddball
284 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
285 if ($hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop') {
286 my $ret = my_read($self->{in_c}, $rbuf, $type + 1);
287 $self->fail(defined($ret) ? 'read EOF' : "read: $!") if !$ret;
289 $self->{rbuf_c} = $rbuf if $$rbuf ne '';
290 splice(@$inflight_c, 0, 3); # don't retry $cb on ->fail
291 eval { $cb->($hex, $type, $size, $arg, $self) };
292 async_err($self, $req, $hex, $@, 'check') if $@;
295 sub check_async_wait ($) {
297 my $inflight_c = $self->{inflight_c} or return;
298 while (scalar(@$inflight_c)) {
299 check_async_step($self, $inflight_c);
303 sub check_async_begin ($) {
305 cleanup($self) if alternates_changed($self);
306 _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c));
307 die 'BUG: already in async check' if $self->{inflight_c};
308 $self->{inflight_c} = [];
312 my ($self, $out, $buf, $read_step, $inflight) = @_;
313 $read_step->($self, $inflight) while @$inflight >= MAX_INFLIGHT;
315 my $w = syswrite($out, $buf);
317 return if $w == length($buf);
319 substr($buf, 0, $w, ''); # sv_chop
320 } elsif ($! != EAGAIN) {
321 $self->fail("write: $!");
322 } else { warn "E: $!" }
323 $read_step->($self, $inflight);
327 sub check_async ($$$$) {
328 my ($self, $oid, $cb, $arg) = @_;
329 my $inflight_c = $self->{inflight_c} // check_async_begin($self);
330 write_all($self, $self->{out_c}, $oid."\n",
331 \&check_async_step, $inflight_c);
332 push(@$inflight_c, $oid, $cb, $arg);
335 sub _check_cb { # check_async callback
336 my ($hex, $type, $size, $result) = @_;
337 @$result = ($hex, $type, $size);
341 my ($self, $oid) = @_;
343 check_async($self, $oid, \&_check_cb, $result);
344 check_async_wait($self);
345 my ($hex, $type, $size) = @$result;
347 # Future versions of git.git may show 'ambiguous', but for now,
348 # we must handle 'dangling' below (and maybe some other oddball
350 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
351 return if $type eq 'missing' || $type eq 'ambiguous';
352 return if $hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop';
353 ($hex, $type, $size);
357 my ($self, $rbuf, $in, $out, $pid, $err) = @_;
358 delete @$self{($rbuf, $in, $out)};
359 delete $self->{$err} if $err; # `err_c'
361 # GitAsyncCat::event_step may delete {pid}
362 my $p = delete $self->{$pid} or return;
363 dwaitpid($p) if $$ == $self->{"$pid.owner"};
366 sub async_abort ($) {
368 while (scalar(@{$self->{inflight_c} // []}) ||
369 scalar(@{$self->{inflight} // []})) {
370 for my $c ('', '_c') {
371 my $q = $self->{"inflight$c"} or next;
373 my ($req, $cb, $arg) = splice(@$q, 0, 3);
374 $req = $$req if ref($req);
375 $req =~ s/ .*//; # drop git_dir for Gcf2Client
376 eval { $cb->(undef, $req, undef, undef, $arg) };
377 warn "E: (in abort) $req: $@" if $@;
379 delete $self->{"inflight$c"};
380 delete $self->{"rbuf$c"};
386 sub fail { # may be augmented in subclasses
387 my ($self, $msg) = @_;
389 croak(ref($self) . ' ' . ($self->{git_dir} // '') . ": $msg");
392 sub async_err ($$$$$) {
393 my ($self, $req, $oid, $err, $action) = @_;
394 $req = $$req if ref($req); # retried
395 my $msg = "E: $action $req ($oid): $err";
396 $async_warn ? carp($msg) : $self->fail($msg);
399 # $git->popen(qw(show f00)); # or
400 # $git->popen(qw(show f00), { GIT_CONFIG => ... }, { 2 => ... });
402 my ($self, $cmd) = splice(@_, 0, 2);
403 $cmd = [ 'git', "--git-dir=$self->{git_dir}",
404 ref($cmd) ? @$cmd : ($cmd, grep { defined && !ref } @_) ];
405 popen_rd($cmd, grep { !defined || ref } @_); # env and opt
408 # same args as popen above
413 close $fh; # caller should check $?
418 close $fh; # caller should check $?
426 substr($_, length('--max-age='), -1)
427 } $self->qx('rev-parse', map { "--since=$_" } @_);
430 # check_async and cat_async may trigger the other, so ensure they're
431 # both completely done by using this:
432 sub async_wait_all ($) {
434 while (scalar(@{$self->{inflight_c} // []}) ||
435 scalar(@{$self->{inflight} // []})) {
436 check_async_wait($self);
437 cat_async_wait($self);
441 # returns true if there are pending "git cat-file" processes
443 my ($self, $lazy) = @_;
444 return 1 if $lazy && (scalar(@{$self->{inflight_c} // []}) ||
445 scalar(@{$self->{inflight} // []}));
446 local $in_cleanup = 1;
447 delete $self->{async_cat};
448 delete $self->{async_chk};
449 async_wait_all($self);
450 delete $self->{inflight};
451 delete $self->{inflight_c};
452 _destroy($self, qw(rbuf in out pid));
453 _destroy($self, qw(rbuf_c in_c out_c pid_c err_c));
457 # assuming a well-maintained repo, this should be a somewhat
458 # accurate estimation of its size
459 # TODO: show this in the WWW UI as a hint to potential cloners
463 my $pack_dir = git_path($self, 'objects/pack');
464 foreach my $p (bsd_glob("$pack_dir/*.pack", GLOB_NOSORT)) {
470 sub DESTROY { cleanup(@_) }
473 # don't show full FS path, basename should be OK:
474 $_[0]->{nick} // ($_[0]->{git_dir} =~ m!/([^/]+?)(?:/*\.git/*)?\z! ?
478 sub host_prefix_url ($$) {
479 my ($env, $url) = @_;
480 return $url if index($url, '//') >= 0;
481 my $scheme = $env->{'psgi.url_scheme'};
482 my $host_port = $env->{HTTP_HOST} //
483 "$env->{SERVER_NAME}:$env->{SERVER_PORT}";
484 "$scheme://$host_port". ($env->{SCRIPT_NAME} || '/') . $url;
487 sub base_url { # for coderepos, PSGI-only
488 my ($self, $env) = @_; # env - PSGI env
489 my $url = host_prefix_url($env, '');
490 # for mount in Plack::Builder
491 $url .= '/' if substr($url, -1, 1) ne '/';
492 $url . $self->{nick} . '/';
498 my ($self, $env) = @_;
499 if (my $urls = $self->{cgit_url}) {
500 return map { host_prefix_url($env, $_) } @$urls;
502 (local_nick($self) // '???');
505 sub cat_async_begin {
507 cleanup($self) if $self->alternates_changed;
508 $self->batch_prepare;
509 die 'BUG: already in async' if $self->{inflight};
510 $self->{inflight} = [];
513 sub cat_async ($$$;$) {
514 my ($self, $oid, $cb, $arg) = @_;
515 my $inflight = $self->{inflight} // cat_async_begin($self);
516 write_all($self, $self->{out}, $oid."\n", \&cat_async_step, $inflight);
517 push(@$inflight, $oid, $cb, $arg);
520 # returns the modified time of a git repo, same as the "modified" field
521 # of a grokmirror manifest
523 # committerdate:unix is git 2.9.4+ (2017-05-05), so using raw instead
524 my $fh = popen($_[0], qw[for-each-ref --sort=-committerdate
525 --format=%(committerdate:raw) --count=1]);
526 (split(/ /, <$fh> // time))[0] + 0; # integerize for JSON
531 open(my $fh, '<', $path) or return '';
537 my $desc = try_cat($_[0]);
540 $desc =~ s/\s+/ /smg;
541 $desc eq '' ? undef : $desc;
545 cat_desc("$_[0]->{git_dir}/description") // 'Unnamed repository';
549 my ($self, $env) = @_;
550 $self->{cloneurl} // do {
551 my @urls = split(/\s+/s, try_cat("$self->{git_dir}/cloneurl"));
552 scalar(@urls) ? ($self->{cloneurl} = \@urls) : undef;
553 } // [ substr(base_url($self, $env), 0, -1) ];
556 # for grokmirror, which doesn't read gitweb.description
557 # templates/hooks--update.sample and git-multimail in git.git
558 # only match "Unnamed repository", not the full contents of
559 # templates/this--description in git.git
561 my ($self, $epoch, $default_desc) = @_;
562 my $fh = $self->popen('show-ref');
563 my $dig = Digest::SHA->new(1);
564 while (read($fh, my $buf, 65536)) {
567 close $fh or return; # empty, uninitialized git repo
568 undef $fh; # for open, below
569 my $git_dir = $self->{git_dir};
571 fingerprint => $dig->hexdigest,
573 modified => modified($self),
575 chomp(my $owner = $self->qx('config', 'gitweb.owner'));
576 utf8::decode($owner);
577 $ent->{owner} = $owner eq '' ? undef : $owner;
578 my $desc = description($self);
579 if (defined $epoch && index($desc, 'Unnamed repository') == 0) {
580 $desc = "$default_desc [epoch $epoch]";
582 $ent->{description} = $desc;
583 if (open($fh, '<', "$git_dir/objects/info/alternates")) {
584 # n.b.: GitPython doesn't seem to handle comments or C-quoted
585 # strings like native git does; and we don't for now, either.
587 chomp(my @alt = <$fh>);
589 # grokmirror only supports 1 alternate for "reference",
590 if (scalar(@alt) == 1) {
591 my $objdir = "$git_dir/objects";
592 my $ref = File::Spec->rel2abs($alt[0], $objdir);
593 $ref =~ s!/[^/]+/?\z!!; # basename
594 $ent->{reference} = $ref;
600 # returns true if there are pending cat-file processes
601 sub cleanup_if_unlinked {
603 return cleanup($self, 1) if $^O ne 'linux';
604 # Linux-specific /proc/$PID/maps access
605 # TODO: support this inside git.git
607 for my $fld (qw(pid pid_c)) {
608 my $pid = $self->{$fld} // next;
609 open my $fh, '<', "/proc/$pid/maps" or return cleanup($self, 1);
611 # n.b. we do not restart for unlinked multi-pack-index
612 # since it's not too huge, and the startup cost may
614 /\.(?:idx|pack) \(deleted\)$/ and
615 return cleanup($self, 1);
628 PublicInbox::Git - git wrapper
636 use PublicInbox::Git;
637 chomp(my $git_dir = `git rev-parse --git-dir`);
638 $git_dir or die "GIT_DIR= must be specified\n";
639 my $git = PublicInbox::Git->new($git_dir);
643 Unstable API outside of the L</new> method.
644 It requires L<git(1)> to be installed.
652 my $git = PublicInbox::Git->new($git_dir);
654 Initialize a new PublicInbox::Git object for use with L<PublicInbox::Import>
655 This is the only public API method we support. Everything else
656 in this module is subject to change.
660 L<Git>, L<PublicInbox::Import>
664 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
666 The mail archives are hosted at L<https://public-inbox.org/meta/>
670 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
672 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>