1 # Copyright (C) 2014-2021 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 use constant MAX_INFLIGHT => (POSIX::PIPE_BUF * 3) /
32 65; # SHA-256 hex size + "\n" in preparation for git using non-SHA1
45 my %ESC_GIT = map { $GIT_ESC{$_} => $_ } keys %GIT_ESC;
48 # unquote pathnames used by git, see quote.c::unquote_c_style.c in git.git
50 return $_[0] unless ($_[0] =~ /\A"(.*)"\z/);
52 $_[0] =~ s!\\([\\"abfnrtv]|[0-3][0-7]{2})!$GIT_ESC{$1}//chr(oct($1))!ge;
57 if ($_[0] =~ s/([\\"\a\b\f\n\r\t\013]|[^[:print:]])/
58 '\\'.($ESC_GIT{$1}||sprintf("%03o",ord($1)))/egs) {
65 my ($class, $git_dir) = @_;
67 $git_dir =~ s!/*\z!!s;
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: $!");
116 pipe(my ($out_r, $out_w)) or $self->fail("pipe failed: $!");
117 my $rdr = { 0 => $out_r };
118 my $gd = $self->{git_dir};
119 if ($gd =~ s!/([^/]+/[^/]+)\z!/!) {
123 my @cmd = (qw(git), "--git-dir=$gd",
124 qw(-c core.abbrev=40 cat-file), $batch);
126 my $id = "git.$self->{git_dir}$batch.err";
127 my $fh = tmpfile($id) or $self->fail("tmpfile($id): $!");
131 my ($in_r, $p) = popen_rd(\@cmd, undef, $rdr);
133 $self->{"$pid.owner"} = $$;
134 $out_w->autoflush(1);
135 if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ
136 fcntl($out_w, 1031, 4096);
137 fcntl($in_r, 1031, 4096) if $batch eq '--batch-check';
139 $self->{$out} = $out_w;
140 $self->{$in} = $in_r;
143 sub poll_in ($) { IO::Poll::_poll($RDTIMEO, fileno($_[0]), my $ev = POLLIN) }
146 my ($fh, $rbuf, $len) = @_;
147 my $left = $len - length($$rbuf);
150 $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf));
153 } elsif (defined($r)) { # EOF
156 next if ($! == EAGAIN and poll_in($fh));
157 next if $! == EINTR; # may be set by sysread or poll_in
158 return; # unrecoverable error
161 \substr($$rbuf, 0, $len, '');
164 sub my_readline ($$) {
165 my ($fh, $rbuf) = @_;
167 if ((my $n = index($$rbuf, "\n")) >= 0) {
168 return substr($$rbuf, 0, $n + 1, '');
170 my $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf))
173 # return whatever's left on EOF
174 return substr($$rbuf, 0, length($$rbuf)+1, '') if defined($r);
176 next if ($! == EAGAIN and poll_in($fh));
177 next if $! == EINTR; # may be set by sysread or poll_in
178 return; # unrecoverable error
182 sub cat_async_retry ($$$$$) {
183 my ($self, $inflight, $req, $cb, $arg) = @_;
185 # {inflight} may be non-existent, but if it isn't we delete it
186 # here to prevent cleanup() from waiting:
187 delete $self->{inflight};
190 $self->{inflight} = $inflight;
191 batch_prepare($self);
193 for (my $i = 0; $i < @$inflight; $i += 3) {
194 $buf .= "$inflight->[$i]\n";
196 print { $self->{out} } $buf or $self->fail("write error: $!");
197 unshift(@$inflight, \$req, $cb, $arg); # \$ref to indicate retried
199 cat_async_step($self, $inflight); # take one step
202 sub cat_async_step ($$) {
203 my ($self, $inflight) = @_;
204 die 'BUG: inflight empty or odd' if scalar(@$inflight) < 3;
205 my ($req, $cb, $arg) = splice(@$inflight, 0, 3);
206 my $rbuf = delete($self->{rbuf}) // \(my $new = '');
207 my ($bref, $oid, $type, $size);
208 my $head = my_readline($self->{in}, $rbuf);
209 # ->fail may be called via Gcf2Client.pm
210 if ($head =~ /^([0-9a-f]{40,}) (\S+) ([0-9]+)$/) {
211 ($oid, $type, $size) = ($1, $2, $3 + 0);
212 $bref = my_read($self->{in}, $rbuf, $size + 1) or
213 $self->fail(defined($bref) ? 'read EOF' : "read: $!");
214 chop($$bref) eq "\n" or $self->fail('LF missing after blob');
215 } elsif ($head =~ s/ missing\n//s) {
217 # ref($req) indicates it's already been retried
218 # -gcf2 retries internally, so it never hits this path:
219 if (!ref($req) && !$in_cleanup && $self->alternates_changed) {
220 return cat_async_retry($self, $inflight,
224 $oid = ref($req) ? $$req : $req if $oid eq '';
226 my $err = $! ? " ($!)" : '';
227 $self->fail("bad result from async cat-file: $head$err");
229 $self->{rbuf} = $rbuf if $$rbuf ne '';
230 eval { $cb->($bref, $oid, $type, $size, $arg) };
231 async_err($self, $req, $oid, $@, 'cat') if $@;
234 sub cat_async_wait ($) {
236 my $inflight = $self->{inflight} or return;
237 while (scalar(@$inflight)) {
238 cat_async_step($self, $inflight);
242 sub batch_prepare ($) {
243 _bidi_pipe($_[0], qw(--batch in out pid));
247 my ($bref, $oid, $type, $size, $result) = @_;
248 @$result = ($bref, $oid, $type, $size);
252 my ($self, $oid) = @_;
254 cat_async($self, $oid, \&_cat_file_cb, $result);
255 cat_async_wait($self);
256 wantarray ? @$result : $result->[0];
259 sub check_async_step ($$) {
260 my ($self, $inflight_c) = @_;
261 die 'BUG: inflight empty or odd' if scalar(@$inflight_c) < 3;
262 my ($req, $cb, $arg) = splice(@$inflight_c, 0, 3);
263 my $rbuf = delete($self->{rbuf_c}) // \(my $new = '');
264 chomp(my $line = my_readline($self->{in_c}, $rbuf));
265 my ($hex, $type, $size) = split(/ /, $line);
267 # Future versions of git.git may have type=ambiguous, but for now,
268 # we must handle 'dangling' below (and maybe some other oddball
270 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
271 if ($hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop') {
272 my $ret = my_read($self->{in_c}, $rbuf, $type + 1);
273 $self->fail(defined($ret) ? 'read EOF' : "read: $!") if !$ret;
275 $self->{rbuf_c} = $rbuf if $$rbuf ne '';
276 eval { $cb->($hex, $type, $size, $arg, $self) };
277 async_err($self, $req, $hex, $@, 'check') if $@;
280 sub check_async_wait ($) {
282 my $inflight_c = $self->{inflight_c} or return;
283 while (scalar(@$inflight_c)) {
284 check_async_step($self, $inflight_c);
288 sub check_async_begin ($) {
290 cleanup($self) if alternates_changed($self);
291 _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c));
292 die 'BUG: already in async check' if $self->{inflight_c};
293 $self->{inflight_c} = [];
296 sub check_async ($$$$) {
297 my ($self, $oid, $cb, $arg) = @_;
298 my $inflight_c = $self->{inflight_c} // check_async_begin($self);
299 while (scalar(@$inflight_c) >= MAX_INFLIGHT) {
300 check_async_step($self, $inflight_c);
302 print { $self->{out_c} } $oid, "\n" or $self->fail("write error: $!");
303 push(@$inflight_c, $oid, $cb, $arg);
306 sub _check_cb { # check_async callback
307 my ($hex, $type, $size, $result) = @_;
308 @$result = ($hex, $type, $size);
312 my ($self, $oid) = @_;
314 check_async($self, $oid, \&_check_cb, $result);
315 check_async_wait($self);
316 my ($hex, $type, $size) = @$result;
318 # Future versions of git.git may show 'ambiguous', but for now,
319 # we must handle 'dangling' below (and maybe some other oddball
321 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
322 return if $type eq 'missing' || $type eq 'ambiguous';
323 return if $hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop';
324 ($hex, $type, $size);
328 my ($self, $rbuf, $in, $out, $pid, $err) = @_;
329 delete @$self{($rbuf, $in, $out)};
330 delete $self->{$err} if $err; # `err_c'
332 # GitAsyncCat::event_step may delete {pid}
333 my $p = delete $self->{$pid} or return;
334 dwaitpid($p) if $$ == $self->{"$pid.owner"};
337 sub async_abort ($) {
339 while (scalar(@{$self->{inflight_c} // []}) ||
340 scalar(@{$self->{inflight} // []})) {
341 for my $c ('', '_c') {
342 my $q = $self->{"inflight$c"};
344 my ($req, $cb, $arg) = splice(@$q, 0, 3);
345 $req = $$req if ref($req);
346 $req =~ s/ .*//; # drop git_dir for Gcf2Client
347 eval { $cb->(undef, $req, undef, undef, $arg) };
348 warn "E: (in abort) $req: $@" if $@;
350 delete $self->{"inflight$c"};
351 delete $self->{"rbuf$c"};
357 sub fail { # may be augmented in subclasses
358 my ($self, $msg) = @_;
360 croak(ref($self) . ' ' . ($self->{git_dir} // '') . ": $msg");
363 sub async_err ($$$$$) {
364 my ($self, $req, $oid, $err, $action) = @_;
365 $req = $$req if ref($req); # retried
366 my $msg = "E: $action $req ($oid): $err";
367 $async_warn ? carp($msg) : $self->fail($msg);
370 # $git->popen(qw(show f00)); # or
371 # $git->popen(qw(show f00), { GIT_CONFIG => ... }, { 2 => ... });
373 my ($self, $cmd) = splice(@_, 0, 2);
374 $cmd = [ 'git', "--git-dir=$self->{git_dir}",
375 ref($cmd) ? @$cmd : ($cmd, grep { defined && !ref } @_) ];
376 popen_rd($cmd, grep { !defined || ref } @_); # env and opt
379 # same args as popen above
384 close $fh; # caller should check $?
389 close $fh; # caller should check $?
397 substr($_, length('--max-age='), -1)
398 } $self->qx('rev-parse', map { "--since=$_" } @_);
401 # check_async and cat_async may trigger the other, so ensure they're
402 # both completely done by using this:
403 sub async_wait_all ($) {
405 while (scalar(@{$self->{inflight_c} // []}) ||
406 scalar(@{$self->{inflight} // []})) {
407 check_async_wait($self);
408 cat_async_wait($self);
412 # returns true if there are pending "git cat-file" processes
414 my ($self, $lazy) = @_;
415 return 1 if $lazy && (scalar(@{$self->{inflight_c} // []}) ||
416 scalar(@{$self->{inflight} // []}));
417 local $in_cleanup = 1;
418 delete $self->{async_cat};
419 async_wait_all($self);
420 delete $self->{inflight};
421 delete $self->{inflight_c};
422 _destroy($self, qw(rbuf in out pid));
423 _destroy($self, qw(rbuf_c in_c out_c pid_c err_c));
427 # assuming a well-maintained repo, this should be a somewhat
428 # accurate estimation of its size
429 # TODO: show this in the WWW UI as a hint to potential cloners
433 my $pack_dir = git_path($self, 'objects/pack');
434 foreach my $p (bsd_glob("$pack_dir/*.pack", GLOB_NOSORT)) {
440 sub DESTROY { cleanup(@_) }
445 # don't show full FS path, basename should be OK:
446 if ($self->{git_dir} =~ m!/([^/]+)(?:/*\.git/*)?\z!) {
449 wantarray ? ($ret) : $ret;
452 sub host_prefix_url ($$) {
453 my ($env, $url) = @_;
454 return $url if index($url, '//') >= 0;
455 my $scheme = $env->{'psgi.url_scheme'};
456 my $host_port = $env->{HTTP_HOST} //
457 "$env->{SERVER_NAME}:$env->{SERVER_PORT}";
458 "$scheme://$host_port". ($env->{SCRIPT_NAME} || '/') . $url;
462 my ($self, $env) = @_;
463 if (my $urls = $self->{cgit_url}) {
464 return map { host_prefix_url($env, $_) } @$urls;
469 sub cat_async_begin {
471 cleanup($self) if $self->alternates_changed;
472 $self->batch_prepare;
473 die 'BUG: already in async' if $self->{inflight};
474 $self->{inflight} = [];
477 sub cat_async ($$$;$) {
478 my ($self, $oid, $cb, $arg) = @_;
479 my $inflight = $self->{inflight} // cat_async_begin($self);
480 while (scalar(@$inflight) >= MAX_INFLIGHT) {
481 cat_async_step($self, $inflight);
483 print { $self->{out} } $oid, "\n" or $self->fail("write error: $!");
484 push(@$inflight, $oid, $cb, $arg);
487 # returns the modified time of a git repo, same as the "modified" field
488 # of a grokmirror manifest
490 # committerdate:unix is git 2.9.4+ (2017-05-05), so using raw instead
491 my $fh = popen($_[0], qw[for-each-ref --sort=-committerdate
492 --format=%(committerdate:raw) --count=1]);
493 (split(/ /, <$fh> // time))[0] + 0; # integerize for JSON
496 # for grokmirror, which doesn't read gitweb.description
497 # templates/hooks--update.sample and git-multimail in git.git
498 # only match "Unnamed repository", not the full contents of
499 # templates/this--description in git.git
501 my ($self, $epoch, $default_desc) = @_;
502 my $fh = $self->popen('show-ref');
503 my $dig = Digest::SHA->new(1);
504 while (read($fh, my $buf, 65536)) {
507 close $fh or return; # empty, uninitialized git repo
508 undef $fh; # for open, below
509 my $git_dir = $self->{git_dir};
511 fingerprint => $dig->hexdigest,
513 modified => modified($self),
515 chomp(my $owner = $self->qx('config', 'gitweb.owner'));
516 utf8::decode($owner);
517 $ent->{owner} = $owner eq '' ? undef : $owner;
519 if (open($fh, '<', "$git_dir/description")) {
521 chomp($desc = <$fh>);
524 $desc = 'Unnamed repository' if $desc eq '';
525 if (defined $epoch && $desc =~ /\AUnnamed repository/) {
526 $desc = "$default_desc [epoch $epoch]";
528 $ent->{description} = $desc;
529 if (open($fh, '<', "$git_dir/objects/info/alternates")) {
530 # n.b.: GitPython doesn't seem to handle comments or C-quoted
531 # strings like native git does; and we don't for now, either.
533 chomp(my @alt = <$fh>);
535 # grokmirror only supports 1 alternate for "reference",
536 if (scalar(@alt) == 1) {
537 my $objdir = "$git_dir/objects";
538 my $ref = File::Spec->rel2abs($alt[0], $objdir);
539 $ref =~ s!/[^/]+/?\z!!; # basename
540 $ent->{reference} = $ref;
546 # returns true if there are pending cat-file processes
547 sub cleanup_if_unlinked {
549 return cleanup($self, 1) if $^O ne 'linux';
550 # Linux-specific /proc/$PID/maps access
551 # TODO: support this inside git.git
553 for my $fld (qw(pid pid_c)) {
554 my $pid = $self->{$fld} // next;
555 open my $fh, '<', "/proc/$pid/maps" or return cleanup($self, 1);
557 # n.b. we do not restart for unlinked multi-pack-index
558 # since it's not too huge, and the startup cost may
560 /\.(?:idx|pack) \(deleted\)$/ and
561 return cleanup($self, 1);
574 PublicInbox::Git - git wrapper
582 use PublicInbox::Git;
583 chomp(my $git_dir = `git rev-parse --git-dir`);
584 $git_dir or die "GIT_DIR= must be specified\n";
585 my $git = PublicInbox::Git->new($git_dir);
589 Unstable API outside of the L</new> method.
590 It requires L<git(1)> to be installed.
598 my $git = PublicInbox::Git->new($git_dir);
600 Initialize a new PublicInbox::Git object for use with L<PublicInbox::Import>
601 This is the only public API method we support. Everything else
602 in this module is subject to change.
606 L<Git>, L<PublicInbox::Import>
610 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
612 The mail archives are hosted at L<https://public-inbox.org/meta/>
616 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
618 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>