1 # Copyright (C) 2014-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Used throughout the project for reading configuration
6 # Note: I hate camelCase; but git-config(1) uses it, but it's better
7 # than alllowercasewithoutunderscores, so use lc('configKey') where
8 # applicable for readability
10 package PublicInbox::Config;
13 use PublicInbox::Inbox;
14 use PublicInbox::Spawn qw(popen_rd);
16 sub _array ($) { ref($_[0]) eq 'ARRAY' ? $_[0] : [ $_[0] ] }
18 # returns key-value pairs of config directives in a hash
19 # if keys may be multi-value, the value is an array ref containing all values
21 my ($class, $file) = @_;
22 $file //= default_file();
24 if (ref($file) eq 'SCALAR') { # used by some tests
25 open my $fh, '<', $file or die; # PerlIO::scalar
26 $self = config_fh_parse($fh, "\n", '=');
28 $self = git_config_dump($file);
32 $self->{-by_addr} = {};
33 $self->{-by_list_id} = {};
34 $self->{-by_name} = {};
35 $self->{-by_newsgroup} = {};
36 $self->{-by_eidx_key} = {};
37 $self->{-no_obfuscate} = {};
38 $self->{-limiters} = {};
39 $self->{-code_repos} = {}; # nick => PublicInbox::Git object
40 $self->{-cgitrc_unparsed} = $self->{'publicinbox.cgitrc'};
42 if (my $no = delete $self->{'publicinbox.noobfuscate'}) {
45 foreach my $n (@$no) {
46 my @n = split(/\s+/, $n);
48 if (/\S+@\S+/) { # full address
49 $self->{-no_obfuscate}->{lc $_} = 1;
51 # allow "example.com" or "@example.com"
53 push @domains, quotemeta($_);
57 my $nod = join('|', @domains);
58 $self->{-no_obfuscate_re} = qr/(?:$nod)\z/i;
60 if (my $css = delete $self->{'publicinbox.css'}) {
61 $self->{css} = _array($css);
68 sub fill_all ($) { each_inbox($_[0], \&noop) }
70 sub _lookup_fill ($$$) {
71 my ($self, $cache, $key) = @_;
72 $self->{$cache}->{$key} // do {
74 $self->{$cache}->{$key};
79 my ($self, $recipient) = @_;
80 _lookup_fill($self, '-by_addr', lc($recipient));
84 my ($self, $list_id) = @_;
85 _lookup_fill($self, '-by_list_id', lc($list_id));
88 sub lookup_name ($$) {
89 my ($self, $name) = @_;
90 $self->{-by_name}->{$name} // _fill($self, "publicinbox.$name");
94 my ($self, $name) = @_;
95 $self->{-ei_by_name}->{$name} //= _fill_ei($self, "extindex.$name");
98 # special case for [extindex "all"]
99 sub ALL { lookup_ei($_[0], 'all') }
102 my ($self, $cb, @arg) = @_;
103 # may auto-vivify if config file is non-existent:
104 foreach my $section (@{$self->{-section_order}}) {
105 next if $section !~ m!\Apublicinbox\.([^/]+)\z!;
106 my $ibx = lookup_name($self, $1) or next;
111 sub lookup_newsgroup {
112 my ($self, $ng) = @_;
113 _lookup_fill($self, '-by_newsgroup', lc($ng));
117 my ($self, $name) = @_;
118 $self->{-limiters}->{$name} //= do {
119 require PublicInbox::Qspawn;
120 my $max = $self->{"publicinboxlimiter.$name.max"} || 1;
121 my $limiter = PublicInbox::Qspawn::Limiter->new($max);
122 $limiter->setup_rlimit($name, $self);
127 sub config_dir { $ENV{PI_DIR} // "$ENV{HOME}/.public-inbox" }
130 $ENV{PI_CONFIG} // (config_dir() . '/config');
133 sub config_fh_parse ($$$) {
134 my ($fh, $rs, $fs) = @_;
135 my (%rv, %seen, @section_order, $line, $k, $v, $section, $cur, $i);
137 while (defined($line = <$fh>)) { # perf critical with giant configs
138 $i = index($line, $fs);
139 $k = substr($line, 0, $i);
140 $v = substr($line, $i + 1, -1); # chop off $fs
141 $section = substr($k, 0, rindex($k, '.'));
142 $seen{$section} //= push(@section_order, $section);
144 if (defined($cur = $rv{$k})) {
145 if (ref($cur) eq "ARRAY") {
148 $rv{$k} = [ $cur, $v ];
154 $rv{-section_order} = \@section_order;
159 sub git_config_dump {
161 return {} unless -e $file;
162 my $cmd = [ qw(git config -z -l --includes), "--file=$file" ];
163 my $fh = popen_rd($cmd);
164 my $rv = config_fh_parse($fh, "\0", "\n");
165 close $fh or die "failed to close (@$cmd) pipe: $?";
169 sub valid_inbox_name ($) {
172 # Similar rules found in git.git/remote.c::valid_remote_nick
173 # and git.git/refs.c::check_refname_component
174 # We don't reject /\.lock\z/, however, since we don't lock refs
175 if ($name eq '' || $name =~ /\@\{/ ||
176 $name =~ /\.\./ || $name =~ m![/:\?\[\]\^~\s\f[:cntrl:]\*]! ||
177 $name =~ /\A\./ || $name =~ /\.\z/) {
181 # Note: we allow URL-unfriendly characters; users may configure
182 # non-HTTP-accessible inboxes
186 # XXX needs testing for cgit compatibility
187 # cf. cgit/scan-tree.c::add_repo
188 sub cgit_repo_merge ($$$) {
189 my ($self, $base, $repo) = @_;
190 my $path = $repo->{dir};
191 if (defined(my $se = $self->{-cgit_strict_export})) {
192 return unless -e "$path/$se";
194 return if -e "$path/noweb";
195 # this comes from the cgit config, and AFAIK cgit only allows
196 # repos to have one URL, but that's just the PATH_INFO component,
197 # not the Host: portion
198 # $repo = { url => 'foo.git', dir => '/path/to/foo.git' }
199 my $rel = $repo->{url};
200 unless (defined $rel) {
201 my $off = index($path, $base, 0);
205 $rel = substr($path, length($base) + 1);
208 $rel =~ s!/\.git\z!! or
211 $self->{-cgit_remove_suffix} and
212 $rel =~ s!/?\.git\z!!;
214 $self->{"coderepo.$rel.dir"} //= $path;
215 $self->{"coderepo.$rel.cgiturl"} //= _array($rel);
220 -d "$git_dir/objects" && -f "$git_dir/HEAD";
223 # XXX needs testing for cgit compatibility
224 sub scan_path_coderepo {
225 my ($self, $base, $path) = @_;
226 opendir(my $dh, $path) or do {
227 warn "error opening directory: $path\n";
231 if (is_git_dir($git_dir) || is_git_dir($git_dir .= '/.git')) {
232 my $repo = { dir => $git_dir };
233 cgit_repo_merge($self, $base, $repo);
236 while (defined(my $dn = readdir $dh)) {
237 next if $dn eq '.' || $dn eq '..';
238 if (index($dn, '.') == 0 && !$self->{-cgit_scan_hidden_path}) {
241 my $dir = "$path/$dn";
242 scan_path_coderepo($self, $base, $dir) if -d $dir;
246 sub scan_tree_coderepo ($$) {
247 my ($self, $path) = @_;
248 scan_path_coderepo($self, $path, $path);
251 sub scan_projects_coderepo ($$$) {
252 my ($self, $list, $path) = @_;
253 open my $fh, '<', $list or do {
254 warn "failed to open cgit projectlist=$list: $!\n";
259 scan_path_coderepo($self, $path, "$path/$_");
264 my ($self, $cgitrc, $nesting) = @_;
267 my %s = map { $_ => 1 } qw(/cgit.css /cgit.png
268 /favicon.ico /robots.txt);
269 $self->{-cgit_static} = \%s;
272 # same limit as cgit/configfile.c::parse_configfile
273 return if $nesting > 8;
275 open my $fh, '<', $cgitrc or do {
276 warn "failed to open cgitrc=$cgitrc: $!\n";
280 # FIXME: this doesn't support macro expansion via $VARS, yet
284 if (m!\Arepo\.url=(.+?)/*\z!) {
286 cgit_repo_merge($self, $repo->{dir}, $repo) if $repo;
287 $repo = { url => $nick };
288 } elsif (m!\Arepo\.path=(.+)\z!) {
292 warn "$_ without repo.url\n";
294 } elsif (m!\Ainclude=(.+)\z!) {
295 parse_cgitrc($self, $1, $nesting + 1);
296 } elsif (m!\A(scan-hidden-path|remove-suffix)=([0-9]+)\z!) {
297 my ($k, $v) = ($1, $2);
299 $self->{"-cgit_$k"} = $v;
300 } elsif (m!\A(project-list|strict-export)=(.+)\z!) {
301 my ($k, $v) = ($1, $2);
303 $self->{"-cgit_$k"} = $v;
304 } elsif (m!\Ascan-path=(.+)\z!) {
305 if (defined(my $list = $self->{-cgit_project_list})) {
306 scan_projects_coderepo($self, $list, $1);
308 scan_tree_coderepo($self, $1);
310 } elsif (m!\A(?:css|favicon|logo|repo\.logo)=(/.+)\z!) {
311 # absolute paths for static files via PublicInbox::Cgit
312 $self->{-cgit_static}->{$1} = 1;
315 cgit_repo_merge($self, $repo->{dir}, $repo) if $repo;
319 # Only git is supported at the moment, but SVN and Hg are possibilities
320 sub _fill_code_repo {
321 my ($self, $nick) = @_;
322 my $pfx = "coderepo.$nick";
324 # TODO: support gitweb and other repository viewers?
325 if (defined(my $cgitrc = delete $self->{-cgitrc_unparsed})) {
326 parse_cgitrc($self, $cgitrc, 0);
328 my $dir = $self->{"$pfx.dir"}; # aka "GIT_DIR"
329 unless (defined $dir) {
330 warn "$pfx.dir unset\n";
334 my $git = PublicInbox::Git->new($dir);
335 foreach my $t (qw(blob commit tree tag)) {
336 $git->{$t.'_url_format'} =
337 _array($self->{lc("$pfx.${t}UrlFormat")});
340 if (defined(my $cgits = $self->{"$pfx.cgiturl"})) {
341 $git->{cgit_url} = $cgits = _array($cgits);
342 $self->{"$pfx.cgiturl"} = $cgits;
344 # cgit supports "/blob/?id=%s", but it's only a plain-text
345 # display and requires an unabbreviated id=
346 foreach my $t (qw(blob commit tag)) {
347 $git->{$t.'_url_format'} //= map {
357 my ($val) = $_[-1]; # $_[0] may be $self, or $val
358 if ($val =~ /\A(?:false|no|off|[\-\+]?(?:0x)?0+)\z/i) {
360 } elsif ($val =~ /\A(?:true|yes|on|[\-\+]?(?:0x)?[0-9]+)\z/i) {
367 # abs_path resolves symlinks, so we want to avoid it if rel2abs
368 # is sufficient and doesn't leave "/.." or "/../"
369 sub rel2abs_collapsed {
371 my $p = File::Spec->rel2abs($_[-1]);
372 return $p if substr($p, -3, 3) ne '/..' && index($p, '/../') < 0;
378 my ($self, $pfx) = @_;
381 for my $k (qw(watch nntpserver)) {
382 my $v = $self->{"$pfx.$k"};
383 $ibx->{$k} = $v if defined $v;
385 for my $k (qw(filter inboxdir newsgroup replyto httpbackendmax feedmax
386 indexlevel indexsequentialshard)) {
387 if (defined(my $v = $self->{"$pfx.$k"})) {
388 if (ref($v) eq 'ARRAY') {
390 W: $pfx.$k has multiple values, only using `$v->[-1]'
392 $ibx->{$k} = $v->[-1];
399 # "mainrepo" is backwards compatibility:
400 my $dir = $ibx->{inboxdir} //= $self->{"$pfx.mainrepo"} // return;
401 if (index($dir, "\n") >= 0) {
402 warn "E: `$dir' must not contain `\\n'\n";
405 foreach my $k (qw(obfuscate)) {
406 my $v = $self->{"$pfx.$k"};
408 if (defined(my $bval = git_bool($v))) {
411 warn "Ignoring $pfx.$k=$v in config, not boolean\n";
414 # TODO: more arrays, we should support multi-value for
415 # more things to encourage decentralization
416 foreach my $k (qw(address altid nntpmirror coderepo hide listid url
417 infourl watchheader)) {
418 if (defined(my $v = $self->{"$pfx.$k"})) {
419 $ibx->{$k} = _array($v);
423 my $name = substr($pfx, length('publicinbox.'));
424 if (!valid_inbox_name($name)) {
425 warn "invalid inbox name: '$name'\n";
429 $ibx->{name} = $name;
430 $ibx->{-pi_cfg} = $self;
431 $ibx = PublicInbox::Inbox->new($ibx);
432 foreach (@{$ibx->{address}}) {
433 my $lc_addr = lc($_);
434 $self->{-by_addr}->{$lc_addr} = $ibx;
435 $self->{-no_obfuscate}->{$lc_addr} = 1;
437 if (my $listids = $ibx->{listid}) {
438 # RFC2919 section 6 stipulates "case insensitive equality"
439 foreach my $list_id (@$listids) {
440 $self->{-by_list_id}->{lc($list_id)} = $ibx;
443 if (defined(my $ngname = $ibx->{newsgroup})) {
445 delete $ibx->{newsgroup};
446 warn 'multiple newsgroups not supported: '.
447 join(', ', @$ngname). "\n";
448 # Newsgroup name needs to be compatible with RFC 3977
449 # wildmat-exact and RFC 3501 (IMAP) ATOM-CHAR.
450 # Leave out a few chars likely to cause problems or conflicts:
451 # '|', '<', '>', ';', '#', '$', '&',
452 } elsif ($ngname =~ m![^A-Za-z0-9/_\.\-\~\@\+\=:]! ||
454 delete $ibx->{newsgroup};
455 warn "newsgroup name invalid: `$ngname'\n";
457 # PublicInbox::NNTPD does stricter ->nntp_usable
458 # checks, keep this lean for startup speed
459 $self->{-by_newsgroup}->{$ngname} = $ibx;
462 unless (defined $ibx->{newsgroup}) { # for ->eidx_key
463 my $abs = rel2abs_collapsed($dir);
465 warn "W: `$dir' canonicalized to `$abs'\n";
466 $ibx->{inboxdir} = $abs;
469 $self->{-by_name}->{$name} = $ibx;
470 if ($ibx->{obfuscate}) {
471 $ibx->{-no_obfuscate} = $self->{-no_obfuscate};
472 $ibx->{-no_obfuscate_re} = $self->{-no_obfuscate_re};
473 fill_all($self); # noop to populate -no_obfuscate
476 if (my $ibx_code_repos = $ibx->{coderepo}) {
477 my $code_repos = $self->{-code_repos};
478 my $repo_objs = $ibx->{-repo_objs} = [];
479 foreach my $nick (@$ibx_code_repos) {
480 my @parts = split(m!/!, $nick);
482 $valid += valid_inbox_name($_) foreach (@parts);
483 $valid == scalar(@parts) or next;
485 my $repo = $code_repos->{$nick} //=
486 _fill_code_repo($self, $nick);
487 push @$repo_objs, $repo if $repo;
490 if (my $es = ALL($self)) {
491 require PublicInbox::Isearch;
492 $ibx->{isrch} = PublicInbox::Isearch->new($ibx, $es);
494 $self->{-by_eidx_key}->{$ibx->eidx_key} = $ibx;
498 my ($self, $pfx) = @_;
499 eval { require PublicInbox::ExtSearch } or return;
500 my $d = $self->{"$pfx.topdir"};
501 defined($d) && -d $d ? PublicInbox::ExtSearch->new($d) : undef;
505 my ($self, $key, $url) = @_;
506 state $urlmatch_broken; # requires git 1.8.5
507 return if $urlmatch_broken;
508 my $file = default_file();
509 my $cmd = [qw/git config -z --includes --get-urlmatch/,
510 "--file=$file", $key, $url ];
511 my $fh = popen_rd($cmd);
518 $urlmatch_broken = 1 if (($? >> 8) != 1);
526 for my $mod (qw(Cpanel::JSON::XS JSON::MaybeXS JSON JSON::PP)) {
527 eval "require $mod" or next;
528 # ->ascii encodes non-ASCII to "\uXXXX"
529 $json = $mod->new->ascii(1) and last;