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);
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 $_[0] =~ s!\\([\\"abfnrtv]|[0-3][0-7]{2})!$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) = @_;
67 # may contain {-tmp} field for File::Temp::Dir
68 bless { git_dir => $git_dir, alt_st => '', -git_path => {} }, $class
72 my ($self, $path) = @_;
73 $self->{-git_path}->{$path} ||= do {
75 chomp(my $str = $self->qx(qw(rev-parse --git-path), $path));
77 # git prior to 2.5.0 did not understand --git-path
78 if ($str eq "--git-path\n$path") {
79 $str = "$self->{git_dir}/$path";
85 sub alternates_changed {
87 my $alt = git_path($self, 'objects/info/alternates');
88 my @st = stat($alt) or return 0;
90 # can't rely on 'q' on some 32-bit builds, but `d' works
91 my $st = pack('dd', $st[10], $st[7]); # 10: ctime, 7: size
92 return 0 if $self->{alt_st} eq $st;
93 $self->{alt_st} = $st; # always a true value
98 my $fh = $self->{err_c} or return;
99 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
100 defined(sysread($fh, my $buf, -s $fh)) or
101 $self->fail("sysread failed: $!");
106 my ($self, $batch, $in, $out, $pid, $err) = @_;
108 if (defined $err) { # "err_c"
109 my $fh = $self->{$err};
110 sysseek($fh, 0, 0) or $self->fail("sysseek failed: $!");
111 truncate($fh, 0) or $self->fail("truncate failed: $!");
116 pipe($out_r, $out_w) or $self->fail("pipe failed: $!");
117 my @cmd = (qw(git), "--git-dir=$self->{git_dir}",
118 qw(-c core.abbrev=40 cat-file), $batch);
119 my $redir = { 0 => $out_r };
121 my $id = "git.$self->{git_dir}$batch.err";
122 my $fh = tmpfile($id) or $self->fail("tmpfile($id): $!");
126 my ($in_r, $p) = popen_rd(\@cmd, undef, $redir);
128 $self->{"$pid.owner"} = $$;
129 $out_w->autoflush(1);
130 if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ
131 fcntl($out_w, 1031, 4096);
132 fcntl($in_r, 1031, 4096) if $batch eq '--batch-check';
134 $self->{$out} = $out_w;
135 $self->{$in} = $in_r;
138 sub poll_in ($) { IO::Poll::_poll($RDTIMEO, fileno($_[0]), my $ev = POLLIN) }
141 my ($fh, $rbuf, $len) = @_;
142 my $left = $len - length($$rbuf);
145 $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf));
148 } elsif (defined($r)) { # EOF
151 next if ($! == EAGAIN and poll_in($fh));
152 next if $! == EINTR; # may be set by sysread or poll_in
153 return; # unrecoverable error
156 \substr($$rbuf, 0, $len, '');
159 sub my_readline ($$) {
160 my ($fh, $rbuf) = @_;
162 if ((my $n = index($$rbuf, "\n")) >= 0) {
163 return substr($$rbuf, 0, $n + 1, '');
165 my $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf))
168 # return whatever's left on EOF
169 return substr($$rbuf, 0, length($$rbuf)+1, '') if defined($r);
171 next if ($! == EAGAIN and poll_in($fh));
172 next if $! == EINTR; # may be set by sysread or poll_in
173 return; # unrecoverable error
177 sub cat_async_retry ($$$$$) {
178 my ($self, $inflight, $req, $cb, $arg) = @_;
180 # {inflight} may be non-existent, but if it isn't we delete it
181 # here to prevent cleanup() from waiting:
182 delete $self->{inflight};
185 $self->{inflight} = $inflight;
186 batch_prepare($self);
188 for (my $i = 0; $i < @$inflight; $i += 3) {
189 $buf .= "$inflight->[$i]\n";
191 print { $self->{out} } $buf or $self->fail("write error: $!");
192 unshift(@$inflight, \$req, $cb, $arg); # \$ref to indicate retried
194 cat_async_step($self, $inflight); # take one step
197 sub cat_async_step ($$) {
198 my ($self, $inflight) = @_;
199 die 'BUG: inflight empty or odd' if scalar(@$inflight) < 3;
200 my ($req, $cb, $arg) = splice(@$inflight, 0, 3);
201 my $rbuf = delete($self->{cat_rbuf}) // \(my $new = '');
202 my ($bref, $oid, $type, $size);
203 my $head = my_readline($self->{in}, $rbuf);
204 # ->fail may be called via Gcf2Client.pm
205 if ($head =~ /^([0-9a-f]{40,}) (\S+) ([0-9]+)$/) {
206 ($oid, $type, $size) = ($1, $2, $3 + 0);
207 $bref = my_read($self->{in}, $rbuf, $size + 1) or
208 $self->fail(defined($bref) ? 'read EOF' : "read: $!");
209 chop($$bref) eq "\n" or $self->fail('LF missing after blob');
210 } elsif ($head =~ s/ missing\n//s) {
212 # ref($req) indicates it's already been retried
213 # -gcf2 retries internally, so it never hits this path:
214 if (!ref($req) && !$in_cleanup && $self->alternates_changed) {
215 return cat_async_retry($self, $inflight,
219 $oid = ref($req) ? $$req : $req if $oid eq '';
221 my $err = $! ? " ($!)" : '';
222 $self->fail("bad result from async cat-file: $head$err");
224 $self->{cat_rbuf} = $rbuf if $$rbuf ne '';
225 eval { $cb->($bref, $oid, $type, $size, $arg) };
226 warn "E: $oid: $@\n" if $@;
229 sub cat_async_wait ($) {
231 my $inflight = $self->{inflight} or return;
232 while (scalar(@$inflight)) {
233 cat_async_step($self, $inflight);
237 sub batch_prepare ($) {
238 _bidi_pipe($_[0], qw(--batch in out pid));
242 my ($bref, $oid, $type, $size, $result) = @_;
243 @$result = ($bref, $oid, $type, $size);
247 my ($self, $oid) = @_;
249 cat_async($self, $oid, \&_cat_file_cb, $result);
250 cat_async_wait($self);
251 wantarray ? @$result : $result->[0];
254 sub check_async_step ($$) {
255 my ($self, $inflight_c) = @_;
256 die 'BUG: inflight empty or odd' if scalar(@$inflight_c) < 3;
257 my ($req, $cb, $arg) = splice(@$inflight_c, 0, 3);
258 my $rbuf = delete($self->{chk_rbuf}) // \(my $new = '');
259 chomp(my $line = my_readline($self->{in_c}, $rbuf));
260 my ($hex, $type, $size) = split(/ /, $line);
262 # Future versions of git.git may have type=ambiguous, but for now,
263 # we must handle 'dangling' below (and maybe some other oddball
265 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
266 if ($hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop') {
267 my $ret = my_read($self->{in_c}, $rbuf, $type + 1);
268 $self->fail(defined($ret) ? 'read EOF' : "read: $!") if !$ret;
270 $self->{chk_rbuf} = $rbuf if $$rbuf ne '';
271 eval { $cb->($hex, $type, $size, $arg, $self) };
272 warn "E: check($req) $@\n" if $@;
275 sub check_async_wait ($) {
277 my $inflight_c = $self->{inflight_c} or return;
278 while (scalar(@$inflight_c)) {
279 check_async_step($self, $inflight_c);
283 sub check_async_begin ($) {
285 cleanup($self) if alternates_changed($self);
286 _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c));
287 die 'BUG: already in async check' if $self->{inflight_c};
288 $self->{inflight_c} = [];
291 sub check_async ($$$$) {
292 my ($self, $oid, $cb, $arg) = @_;
293 my $inflight_c = $self->{inflight_c} // check_async_begin($self);
294 while (scalar(@$inflight_c) >= MAX_INFLIGHT) {
295 check_async_step($self, $inflight_c);
297 print { $self->{out_c} } $oid, "\n" or $self->fail("write error: $!");
298 push(@$inflight_c, $oid, $cb, $arg);
301 sub _check_cb { # check_async callback
302 my ($hex, $type, $size, $result) = @_;
303 @$result = ($hex, $type, $size);
307 my ($self, $oid) = @_;
309 check_async($self, $oid, \&_check_cb, $result);
310 check_async_wait($self);
311 my ($hex, $type, $size) = @$result;
313 # Future versions of git.git may show 'ambiguous', but for now,
314 # we must handle 'dangling' below (and maybe some other oddball
316 # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/
317 return if $type eq 'missing' || $type eq 'ambiguous';
318 return if $hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop';
319 ($hex, $type, $size);
323 my ($self, $rbuf, $in, $out, $pid, $err) = @_;
324 delete @$self{($rbuf, $in, $out)};
325 delete $self->{$err} if $err; # `err_c'
327 # GitAsyncCat::event_step may delete {pid}
328 my $p = delete $self->{$pid} or return;
329 dwaitpid($p) if $$ == $self->{"$pid.owner"};
332 sub cat_async_abort ($) {
334 if (my $inflight = $self->{inflight}) {
336 my ($req, $cb, $arg) = splice(@$inflight, 0, 3);
337 $req =~ s/ .*//; # drop git_dir for Gcf2Client
338 eval { $cb->(undef, $req, undef, undef, $arg) };
339 warn "E: $req: $@ (in abort)\n" if $@;
341 delete $self->{cat_rbuf};
342 delete $self->{inflight};
347 sub fail { # may be augmented in subclasses
348 my ($self, $msg) = @_;
349 cat_async_abort($self);
350 croak(ref($self) . ' ' . ($self->{git_dir} // '') . ": $msg");
353 # $git->popen(qw(show f00)); # or
354 # $git->popen(qw(show f00), { GIT_CONFIG => ... }, { 2 => ... });
356 my ($self, $cmd) = splice(@_, 0, 2);
357 $cmd = [ 'git', "--git-dir=$self->{git_dir}",
358 ref($cmd) ? @$cmd : ($cmd, grep { defined && !ref } @_) ];
359 popen_rd($cmd, grep { !defined || ref } @_); # env and opt
362 # same args as popen above
367 close $fh; # caller should check $?
372 close $fh; # caller should check $?
380 substr($_, length('--max-age='), -1)
381 } $self->qx('rev-parse', map { "--since=$_" } @_);
384 # check_async and cat_async may trigger the other, so ensure they're
385 # both completely done by using this:
386 sub async_wait_all ($) {
388 while (scalar(@{$self->{inflight_c} // []}) ||
389 scalar(@{$self->{inflight} // []})) {
390 $self->check_async_wait;
391 $self->cat_async_wait;
395 # returns true if there are pending "git cat-file" processes
398 local $in_cleanup = 1;
399 delete $self->{async_cat};
400 async_wait_all($self);
401 delete $self->{inflight};
402 delete $self->{inflight_c};
403 _destroy($self, qw(cat_rbuf in out pid));
404 _destroy($self, qw(chk_rbuf in_c out_c pid_c err_c));
405 !!($self->{pid} || $self->{pid_c});
409 # assuming a well-maintained repo, this should be a somewhat
410 # accurate estimation of its size
411 # TODO: show this in the WWW UI as a hint to potential cloners
415 my $pack_dir = git_path($self, 'objects/pack');
416 foreach my $p (bsd_glob("$pack_dir/*.pack", GLOB_NOSORT)) {
422 sub DESTROY { cleanup(@_) }
427 # don't show full FS path, basename should be OK:
428 if ($self->{git_dir} =~ m!/([^/]+)(?:/*\.git/*)?\z!) {
431 wantarray ? ($ret) : $ret;
434 sub host_prefix_url ($$) {
435 my ($env, $url) = @_;
436 return $url if index($url, '//') >= 0;
437 my $scheme = $env->{'psgi.url_scheme'};
438 my $host_port = $env->{HTTP_HOST} //
439 "$env->{SERVER_NAME}:$env->{SERVER_PORT}";
440 "$scheme://$host_port". ($env->{SCRIPT_NAME} || '/') . $url;
444 my ($self, $env) = @_;
445 if (my $urls = $self->{cgit_url}) {
446 return map { host_prefix_url($env, $_) } @$urls;
451 sub cat_async_begin {
453 cleanup($self) if $self->alternates_changed;
454 $self->batch_prepare;
455 die 'BUG: already in async' if $self->{inflight};
456 $self->{inflight} = [];
459 sub cat_async ($$$;$) {
460 my ($self, $oid, $cb, $arg) = @_;
461 my $inflight = $self->{inflight} // cat_async_begin($self);
462 while (scalar(@$inflight) >= MAX_INFLIGHT) {
463 cat_async_step($self, $inflight);
465 print { $self->{out} } $oid, "\n" or $self->fail("write error: $!");
466 push(@$inflight, $oid, $cb, $arg);
469 sub extract_cmt_time {
470 my ($bref, undef, undef, undef, $modified) = @_;
472 if ($$bref =~ /^committer .*?> ([0-9]+) [\+\-]?[0-9]+/sm) {
473 my $cmt_time = $1 + 0;
474 $$modified = $cmt_time if $cmt_time > $$modified;
478 # returns the modified time of a git repo, same as the "modified" field
479 # of a grokmirror manifest
483 my $fh = popen($self, qw(rev-parse --branches));
485 while (my $oid = <$fh>) {
487 cat_async($self, $oid, \&extract_cmt_time, \$modified);
489 cat_async_wait($self);
493 # for grokmirror, which doesn't read gitweb.description
494 # templates/hooks--update.sample and git-multimail in git.git
495 # only match "Unnamed repository", not the full contents of
496 # templates/this--description in git.git
498 my ($self, $epoch, $default_desc) = @_;
499 my $fh = $self->popen('show-ref');
500 my $dig = Digest::SHA->new(1);
501 while (read($fh, my $buf, 65536)) {
504 close $fh or return; # empty, uninitialized git repo
505 undef $fh; # for open, below
506 my $git_dir = $self->{git_dir};
508 fingerprint => $dig->hexdigest,
510 modified => modified($self),
512 chomp(my $owner = $self->qx('config', 'gitweb.owner'));
513 utf8::decode($owner);
514 $ent->{owner} = $owner eq '' ? undef : $owner;
516 if (open($fh, '<', "$git_dir/description")) {
518 chomp($desc = <$fh>);
521 $desc = 'Unnamed repository' if $desc eq '';
522 if (defined $epoch && $desc =~ /\AUnnamed repository/) {
523 $desc = "$default_desc [epoch $epoch]";
525 $ent->{description} = $desc;
526 if (open($fh, '<', "$git_dir/objects/info/alternates")) {
527 # n.b.: GitPython doesn't seem to handle comments or C-quoted
528 # strings like native git does; and we don't for now, either.
530 chomp(my @alt = <$fh>);
532 # grokmirror only supports 1 alternate for "reference",
533 if (scalar(@alt) == 1) {
534 my $objdir = "$git_dir/objects";
535 my $ref = File::Spec->rel2abs($alt[0], $objdir);
536 $ref =~ s!/[^/]+/?\z!!; # basename
537 $ent->{reference} = $ref;
549 PublicInbox::Git - git wrapper
557 use PublicInbox::Git;
558 chomp(my $git_dir = `git rev-parse --git-dir`);
559 $git_dir or die "GIT_DIR= must be specified\n";
560 my $git = PublicInbox::Git->new($git_dir);
564 Unstable API outside of the L</new> method.
565 It requires L<git(1)> to be installed.
573 my $git = PublicInbox::Git->new($git_dir);
575 Initialize a new PublicInbox::Git object for use with L<PublicInbox::Import>
576 This is the only public API method we support. Everything else
577 in this module is subject to change.
581 L<Git>, L<PublicInbox::Import>
585 All feedback welcome via plain-text mail to L<mailto:meta@public-inbox.org>
587 The mail archives are hosted at L<https://public-inbox.org/meta/>
591 Copyright (C) 2016 all contributors L<mailto:meta@public-inbox.org>
593 License: AGPL-3.0+ L<http://www.gnu.org/licenses/agpl-3.0.txt>