1 # Copyright (C) 2014-2020 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);
20 use PublicInbox::Tmpfile;
21 use IO::Poll qw(POLLIN);
24 use PublicInbox::DS qw(dwaitpid);
25 our @EXPORT_OK = qw(git_unquote git_quote);
26 our $PIPE_BUFSIZ = 65536; # Linux default
28 our $RDTIMEO = 60_000; # milliseconds
30 use constant MAX_INFLIGHT =>
31 (($^O eq 'linux' ? 4096 : POSIX::_POSIX_PIPE_BUF()) * 3)
33 65; # SHA-256 hex size + "\n" in preparation for git using non-SHA1
46 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 $_[0] =~ s/\\([\\"abfnrtv])/$GIT_ESC{$1}/g;
54 $_[0] =~ s/\\([0-7]{1,3})/chr(oct($1))/ge;
59 if ($_[0] =~ s/([\\"\a\b\f\n\r\t\013]|[^[:print:]])/
60 '\\'.($ESC_GIT{$1}||sprintf("%0o",ord($1)))/egs) {
67 my ($class, $git_dir) = @_;
68 # may contain {-tmp} field for File::Temp::Dir
69 bless { git_dir => $git_dir, alt_st => '', -git_path => {} }, $class
73 my ($self, $path) = @_;
74 $self->{-git_path}->{$path} ||= do {
76 chomp(my $str = $self->qx(qw(rev-parse --git-path), $path));
78 # git prior to 2.5.0 did not understand --git-path
79 if ($str eq "--git-path\n$path") {
80 $str = "$self->{git_dir}/$path";
86 sub alternates_changed {
88 my $alt = git_path($self, 'objects/info/alternates');
89 my @st = stat($alt) or return 0;
91 # can't rely on 'q' on some 32-bit builds, but `d' works
92 my $st = pack('dd', $st[10], $st[7]); # 10: ctime, 7: size
93 return 0 if $self->{alt_st} eq $st;
94 $self->{alt_st} = $st; # always a true value
99 my $fh = $self->{err_c} or return;
100 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
101 defined(sysread($fh, my $buf, -s $fh)) or
102 $self->fail("sysread failed: $!");
107 my ($self, $batch, $in, $out, $pid, $err) = @_;
109 if (defined $err) { # "err_c"
110 my $fh = $self->{$err};
111 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
112 truncate($fh, 0) or $self->fail("truncate failed: $!");
117 pipe($out_r, $out_w) or $self->fail("pipe failed: $!");
118 my @cmd = (qw(git), "--git-dir=$self->{git_dir}",
119 qw(-c core.abbrev=40 cat-file), $batch);
120 my $redir = { 0 => $out_r };
122 my $id = "git.$self->{git_dir}$batch.err";
123 my $fh = tmpfile($id) or $self->fail("tmpfile($id): $!");
127 my ($in_r, $p) = popen_rd(\@cmd, undef, $redir);
129 $self->{"$pid.owner"} = $$;
130 $out_w->autoflush(1);
131 if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ
132 fcntl($out_w, 1031, 4096);
133 fcntl($in_r, 1031, 4096) if $batch eq '--batch-check';
135 $self->{$out} = $out_w;
136 $self->{$in} = $in_r;
139 sub poll_in ($) { IO::Poll::_poll($RDTIMEO, fileno($_[0]), my $ev = POLLIN) }
142 my ($fh, $rbuf, $len) = @_;
143 my $left = $len - length($$rbuf);
146 $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf));
149 } elsif (defined($r)) { # EOF
152 next if ($! == EAGAIN and poll_in($fh));
153 next if $! == EINTR; # may be set by sysread or poll_in
154 return; # unrecoverable error
157 \substr($$rbuf, 0, $len, '');
160 sub my_readline ($$) {
161 my ($fh, $rbuf) = @_;
163 if ((my $n = index($$rbuf, "\n")) >= 0) {
164 return substr($$rbuf, 0, $n + 1, '');
166 my $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf))
169 # return whatever's left on EOF
170 return substr($$rbuf, 0, length($$rbuf)+1, '') if defined($r);
172 next if ($! == EAGAIN and poll_in($fh));
173 next if $! == EINTR; # may be set by sysread or poll_in
174 return; # unrecoverable error
178 sub cat_async_retry ($$$$$) {
179 my ($self, $inflight, $req, $cb, $arg) = @_;
181 # {inflight} may be non-existent, but if it isn't we delete it
182 # here to prevent cleanup() from waiting:
183 delete $self->{inflight};
186 $self->{inflight} = $inflight;
187 batch_prepare($self);
189 for (my $i = 0; $i < @$inflight; $i += 3) {
190 $buf .= "$inflight->[$i]\n";
192 print { $self->{out} } $buf or $self->fail("write error: $!");
193 unshift(@$inflight, \$req, $cb, $arg); # \$ref to indicate retried
195 cat_async_step($self, $inflight); # take one step
198 sub cat_async_step ($$) {
199 my ($self, $inflight) = @_;
200 die 'BUG: inflight empty or odd' if scalar(@$inflight) < 3;
201 my ($req, $cb, $arg) = splice(@$inflight, 0, 3);
202 my $rbuf = delete($self->{cat_rbuf}) // \(my $new = '');
203 my ($bref, $oid, $type, $size);
204 my $head = my_readline($self->{in}, $rbuf);
205 # ->fail may be called via Gcf2Client.pm
206 if ($head =~ /^([0-9a-f]{40,}) (\S+) ([0-9]+)$/) {
207 ($oid, $type, $size) = ($1, $2, $3 + 0);
208 $bref = my_read($self->{in}, $rbuf, $size + 1) or
209 $self->fail(defined($bref) ? 'read EOF' : "read: $!");
210 chop($$bref) eq "\n" or $self->fail('LF missing after blob');
211 } elsif ($head =~ s/ missing\n//s) {
213 # ref($req) indicates it's already been retried
214 # -gcf2 retries internally, so it never hits this path:
215 if (!ref($req) && !$in_cleanup && $self->alternates_changed) {
216 return cat_async_retry($self, $inflight,
220 $oid = ref($req) ? $$req : $req if $oid eq '';
222 my $err = $! ? " ($!)" : '';
223 $self->fail("bad result from async cat-file: $head$err");
225 $self->{cat_rbuf} = $rbuf if $$rbuf ne '';
226 eval { $cb->($bref, $oid, $type, $size, $arg) };
227 warn "E: $oid: $@\n" if $@;
230 sub cat_async_wait ($) {
232 my $inflight = $self->{inflight} or return;
233 while (scalar(@$inflight)) {
234 cat_async_step($self, $inflight);
238 sub batch_prepare ($) {
239 _bidi_pipe($_[0], qw(--batch in out pid));
243 my ($bref, undef, undef, $size, $result) = @_;
244 @$result = ($bref, $size);
248 my ($self, $oid, $sizeref) = @_;
250 cat_async($self, $oid, \&_cat_file_cb, $result);
251 cat_async_wait($self);
252 $$sizeref = $result->[1] if $sizeref;
256 sub check_async_step ($$) {
257 my ($self, $inflight_c) = @_;
258 die 'BUG: inflight empty or odd' if scalar(@$inflight_c) < 3;
259 my ($req, $cb, $arg) = splice(@$inflight_c, 0, 3);
260 my $rbuf = delete($self->{chk_rbuf}) // \(my $new = '');
261 chomp(my $line = my_readline($self->{in_c}, $rbuf));
262 my ($hex, $type, $size) = split(/ /, $line);
264 # Future versions of git.git may have type=ambiguous, but for now,
265 # we must handle 'dangling' below (and maybe some other oddball
267 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
268 if ($hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop') {
269 my $ret = my_read($self->{in_c}, $rbuf, $type + 1);
270 $self->fail(defined($ret) ? 'read EOF' : "read: $!") if !$ret;
272 $self->{chk_rbuf} = $rbuf if $$rbuf ne '';
273 eval { $cb->($hex, $type, $size, $arg, $self) };
274 warn "E: check($req) $@\n" if $@;
277 sub check_async_wait ($) {
279 my $inflight_c = $self->{inflight_c} or return;
280 while (scalar(@$inflight_c)) {
281 check_async_step($self, $inflight_c);
285 sub check_async_begin ($) {
287 cleanup($self) if alternates_changed($self);
288 _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c));
289 die 'BUG: already in async check' if $self->{inflight_c};
290 $self->{inflight_c} = [];
293 sub check_async ($$$$) {
294 my ($self, $oid, $cb, $arg) = @_;
295 my $inflight_c = $self->{inflight_c} // check_async_begin($self);
296 while (scalar(@$inflight_c) >= MAX_INFLIGHT) {
297 check_async_step($self, $inflight_c);
299 print { $self->{out_c} } $oid, "\n" or $self->fail("write error: $!");
300 push(@$inflight_c, $oid, $cb, $arg);
303 sub _check_cb { # check_async callback
304 my ($hex, $type, $size, $result) = @_;
305 @$result = ($hex, $type, $size);
309 my ($self, $oid) = @_;
311 check_async($self, $oid, \&_check_cb, $result);
312 check_async_wait($self);
313 my ($hex, $type, $size) = @$result;
315 # Future versions of git.git may show 'ambiguous', but for now,
316 # we must handle 'dangling' below (and maybe some other oddball
318 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
319 return if $type eq 'missing' || $type eq 'ambiguous';
320 return if $hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop';
321 ($hex, $type, $size);
325 my ($self, $rbuf, $in, $out, $pid, $err) = @_;
326 delete @$self{($rbuf, $in, $out)};
327 delete $self->{$err} if $err; # `err_c'
329 # GitAsyncCat::event_step may delete {pid}
330 my $p = delete $self->{$pid} or return;
331 dwaitpid($p) if $$ == $self->{"$pid.owner"};
334 sub cat_async_abort ($) {
336 if (my $inflight = $self->{inflight}) {
338 my ($req, $cb, $arg) = splice(@$inflight, 0, 3);
339 $req =~ s/ .*//; # drop git_dir for Gcf2Client
340 eval { $cb->(undef, $req, undef, undef, $arg) };
341 warn "E: $req: $@ (in abort)\n" if $@;
343 delete $self->{cat_rbuf};
344 delete $self->{inflight};
349 sub fail { # may be augmented in subclasses
350 my ($self, $msg) = @_;
351 cat_async_abort($self);
352 croak(ref($self) . ' ' . ($self->{git_dir} // '') . ": $msg");
356 my ($self, @cmd) = @_;
357 @cmd = ('git', "--git-dir=$self->{git_dir}", @cmd);
362 my ($self, @cmd) = @_;
363 my $fh = $self->popen(@cmd);
364 local $/ = wantarray ? "\n" : undef;
368 # check_async and cat_async may trigger the other, so ensure they're
369 # both completely done by using this:
370 sub async_wait_all ($) {
372 while (scalar(@{$self->{inflight_c} // []}) ||
373 scalar(@{$self->{inflight} // []})) {
374 $self->check_async_wait;
375 $self->cat_async_wait;
379 # returns true if there are pending "git cat-file" processes
382 local $in_cleanup = 1;
383 delete $self->{async_cat};
384 async_wait_all($self);
385 delete $self->{inflight};
386 delete $self->{inflight_c};
387 _destroy($self, qw(cat_rbuf in out pid));
388 _destroy($self, qw(chk_rbuf in_c out_c pid_c err_c));
389 !!($self->{pid} || $self->{pid_c});
393 # assuming a well-maintained repo, this should be a somewhat
394 # accurate estimation of its size
395 # TODO: show this in the WWW UI as a hint to potential cloners
399 my $pack_dir = git_path($self, 'objects/pack');
400 foreach my $p (bsd_glob("$pack_dir/*.pack", GLOB_NOSORT)) {
406 sub DESTROY { cleanup(@_) }
411 # don't show full FS path, basename should be OK:
412 if ($self->{git_dir} =~ m!/([^/]+)(?:/\.git)?\z!) {
413 $ret = "/path/to/$1";
415 wantarray ? ($ret) : $ret;
418 sub host_prefix_url ($$) {
419 my ($env, $url) = @_;
420 return $url if index($url, '//') >= 0;
421 my $scheme = $env->{'psgi.url_scheme'};
422 my $host_port = $env->{HTTP_HOST} //
423 "$env->{SERVER_NAME}:$env->{SERVER_PORT}";
424 "$scheme://$host_port". ($env->{SCRIPT_NAME} || '/') . $url;
428 my ($self, $env) = @_;
429 if (my $urls = $self->{cgit_url}) {
430 return map { host_prefix_url($env, $_) } @$urls;
435 sub cat_async_begin {
437 cleanup($self) if $self->alternates_changed;
438 $self->batch_prepare;
439 die 'BUG: already in async' if $self->{inflight};
440 $self->{inflight} = [];
443 sub cat_async ($$$;$) {
444 my ($self, $oid, $cb, $arg) = @_;
445 my $inflight = $self->{inflight} // cat_async_begin($self);
446 while (scalar(@$inflight) >= MAX_INFLIGHT) {
447 cat_async_step($self, $inflight);
449 print { $self->{out} } $oid, "\n" or $self->fail("write error: $!");
450 push(@$inflight, $oid, $cb, $arg);
454 my ($self, $oid, $cb, $arg) = @_;
455 if (my $inflight = $self->{inflight}) {
456 # we could use MAX_INFLIGHT here w/o the halving,
457 # but lets not allow one client to monopolize a git process
458 if (scalar(@$inflight) < int(MAX_INFLIGHT/2)) {
459 print { $self->{out} } $oid, "\n" or
460 $self->fail("write error: $!");
461 return push(@$inflight, $oid, $cb, $arg);
467 sub extract_cmt_time {
468 my ($bref, undef, undef, undef, $modified) = @_;
470 if ($$bref =~ /^committer .*?> ([0-9]+) [\+\-]?[0-9]+/sm) {
471 my $cmt_time = $1 + 0;
472 $$modified = $cmt_time if $cmt_time > $$modified;
476 # returns the modified time of a git repo, same as the "modified" field
477 # of a grokmirror manifest
481 my $fh = popen($self, qw(rev-parse --branches));
483 while (my $oid = <$fh>) {
485 cat_async($self, $oid, \&extract_cmt_time, \$modified);
487 cat_async_wait($self);
491 # for grokmirror, which doesn't read gitweb.description
492 # templates/hooks--update.sample and git-multimail in git.git
493 # only match "Unnamed repository", not the full contents of
494 # templates/this--description in git.git
496 my ($self, $epoch, $default_desc) = @_;
497 my ($fh, $pid) = $self->popen('show-ref');
498 my $dig = Digest::SHA->new(1);
499 while (read($fh, my $buf, 65536)) {
504 return if $?; # empty, uninitialized git repo
505 my $git_dir = $self->{git_dir};
507 fingerprint => $dig->hexdigest,
509 modified => modified($self),
511 chomp(my $owner = $self->qx('config', 'gitweb.owner'));
512 utf8::decode($owner);
513 $ent->{owner} = $owner eq '' ? undef : $owner;
515 if (open($fh, '<', "$git_dir/description")) {
517 chomp($desc = <$fh>);
520 $desc = 'Unnamed repository' if $desc eq '';
521 if (defined $epoch && $desc =~ /\AUnnamed repository/) {
522 $desc = "$default_desc [epoch $epoch]";
524 $ent->{description} = $desc;
525 if (open($fh, '<', "$git_dir/objects/info/alternates")) {
526 # n.b.: GitPython doesn't seem to handle comments or C-quoted
527 # strings like native git does; and we don't for now, either.
529 chomp(my @alt = <$fh>);
531 # grokmirror only supports 1 alternate for "reference",
532 if (scalar(@alt) == 1) {
533 my $objdir = "$git_dir/objects";
534 my $ref = File::Spec->rel2abs($alt[0], $objdir);
535 $ref =~ s!/[^/]+/?\z!!; # basename
536 $ent->{reference} = $ref;
548 PublicInbox::Git - git wrapper
556 use PublicInbox::Git;
557 chomp(my $git_dir = `git rev-parse --git-dir`);
558 $git_dir or die "GIT_DIR= must be specified\n";
559 my $git = PublicInbox::Git->new($git_dir);
563 Unstable API outside of the L</new> method.
564 It requires L<git(1)> to be installed.
572 my $git = PublicInbox::Git->new($git_dir);
574 Initialize a new PublicInbox::Git object for use with L<PublicInbox::Import>
575 This is the only public API method we support. Everything else
576 in this module is subject to change.
580 L<Git>, L<PublicInbox::Import>
584 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
586 The mail archives are hosted at L<https://public-inbox.org/meta/>
590 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
592 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>