1 # Copyright (C) 2014-2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
4 # Main web interface for mailing list archives
6 # We focus on the lowest common denominators here:
7 # - targeted at text-only console browsers (w3m, links, etc..)
8 # - Only basic HTML, CSS only for line-wrapping <pre> text content for GUIs
9 # - No JavaScript, graphics or icons allowed.
10 # - Must not rely on static content
11 # - UTF-8 is only for user-content, 7-bit US-ASCII for us
12 package PublicInbox::WWW;
16 use PublicInbox::Config qw(try_cat);
17 use URI::Escape qw(uri_escape_utf8 uri_unescape);
18 use constant SSOMA_URL => '//ssoma.public-inbox.org/';
19 use constant PI_URL => '//public-inbox.org/';
20 require PublicInbox::Git;
21 use PublicInbox::GitHTTPBackend;
22 our $LISTNAME_RE = qr!\A/([\w\.\-]+)!;
23 our $MID_RE = qr!([^/]+)!;
24 our $END_RE = qr!(f/|T/|t/|R/|t\.mbox(?:\.gz)?|t\.atom|raw|)!;
27 my ($class, $pi_config) = @_;
28 $pi_config ||= PublicInbox::Config->new;
29 bless { pi_config => $pi_config }, $class;
32 # backwards compatibility, do not use
34 my ($req, $method) = @_;
35 PublicInbox::WWW->new->call($req->env);
39 my ($self, $env) = @_;
40 my $cgi = Plack::Request->new($env);
41 my $ctx = { cgi => $cgi, pi_config => $self->{pi_config} };
42 my $path_info = $cgi->path_info;
44 my $method = $cgi->method;
45 if ($method eq 'POST' &&
46 $path_info =~ m!$LISTNAME_RE/(git-upload-pack)\z!) {
48 return (invalid_list($ctx, $1) ||
49 serve_git($cgi, $ctx->{git}, $path));
51 elsif ($method !~ /\AGET|HEAD\z/) {
52 return r(405, 'Method Not Allowed');
55 # top-level indices and feeds
56 if ($path_info eq '/') {
58 } elsif ($path_info =~ m!$LISTNAME_RE\z!o) {
59 invalid_list($ctx, $1) || r301($ctx, $1);
60 } elsif ($path_info =~ m!$LISTNAME_RE(?:/|/index\.html)?\z!o) {
61 invalid_list($ctx, $1) || get_index($ctx);
62 } elsif ($path_info =~ m!$LISTNAME_RE/(?:atom\.xml|new\.atom)\z!o) {
63 invalid_list($ctx, $1) || get_atom($ctx);
65 } elsif ($path_info =~ m!$LISTNAME_RE/
66 ($PublicInbox::GitHTTPBackend::ANY)\z!ox) {
68 invalid_list($ctx, $1) || serve_git($cgi, $ctx->{git}, $path);
69 } elsif ($path_info =~ m!$LISTNAME_RE/$MID_RE/$END_RE\z!o) {
70 msg_page($ctx, $1, $2, $3);
72 # in case people leave off the trailing slash:
73 } elsif ($path_info =~ m!$LISTNAME_RE/$MID_RE/(f|T|t|R)\z!o) {
74 my ($listname, $mid, $suffix) = ($1, $2, $3);
75 $suffix .= $suffix =~ /\A[tT]\z/ ? '/#u' : '/';
76 r301($ctx, $listname, $mid, $suffix);
78 # convenience redirects order matters
79 } elsif ($path_info =~ m!$LISTNAME_RE/([^/]{2,})\z!o) {
83 legacy_redirects($ctx, $path_info);
87 # for CoW-friendliness, MOOOOO!
89 require PublicInbox::Feed;
90 require PublicInbox::View;
91 require PublicInbox::Thread;
97 require PublicInbox::Search;
98 require PublicInbox::SearchView;
99 require PublicInbox::Mbox;
100 require IO::Compress::Gzip;
104 # private functions below
108 if ($ctx && $ctx->{mid}) {
109 require PublicInbox::ExtMsg;
111 return PublicInbox::ExtMsg::ext_msg($ctx);
116 # simple response for errors
117 sub r { [ $_[0], ['Content-Type' => 'text/plain'], [ join(' ', @_, "\n") ] ] }
119 # returns undef if valid, array ref response if invalid
121 my ($ctx, $listname) = @_;
122 my $git_dir = $ctx->{pi_config}->get($listname, "mainrepo");
123 if (defined $git_dir) {
124 $ctx->{git_dir} = $git_dir;
125 $ctx->{git} = PublicInbox::Git->new($git_dir);
126 $ctx->{listname} = $listname;
132 # returns undef if valid, array ref response if invalid
133 sub invalid_list_mid {
134 my ($ctx, $listname, $mid) = @_;
135 my $ret = invalid_list($ctx, $listname, $mid);
138 $ctx->{mid} = $mid = uri_unescape($mid);
139 if ($mid =~ /\A[a-f0-9]{40}\z/) {
140 if ($mid = mid2blob($ctx)) {
141 require Email::Simple;
142 use PublicInbox::MID qw/mid_clean/;
143 $mid = Email::Simple->new($mid);
144 $ctx->{mid} = mid_clean($mid->header('Message-ID'));
150 # /$LISTNAME/new.atom -> Atom feed, includes replies
153 require PublicInbox::Feed;
154 PublicInbox::Feed::generate($ctx);
157 # /$LISTNAME/?r=$GIT_COMMIT -> HTML only
160 require PublicInbox::Feed;
161 my $srch = searcher($ctx);
163 if (defined $ctx->{cgi}->param('q')) {
164 require PublicInbox::SearchView;
165 PublicInbox::SearchView::sres_top_html($ctx);
167 PublicInbox::Feed::generate_html_index($ctx);
171 # just returns a string ref for the blob in the current ctx
174 require PublicInbox::MID;
175 my $path = PublicInbox::MID::mid2path($ctx->{mid});
176 $ctx->{git}->cat_file("HEAD:$path");
179 # /$LISTNAME/$MESSAGE_ID/raw -> raw mbox
182 my $x = mid2blob($ctx) or return r404($ctx);
183 require PublicInbox::Mbox;
184 PublicInbox::Mbox::emit1($ctx, $x);
187 # /$LISTNAME/$MESSAGE_ID/ -> HTML content (short quotes)
190 my $x = mid2blob($ctx) or return r404($ctx);
192 require PublicInbox::View;
193 my $foot = footer($ctx);
195 my $mime = Email::MIME->new($x);
197 [ 200, [ 'Content-Type' => 'text/html; charset=UTF-8' ],
198 [ PublicInbox::View::msg_html($ctx, $mime, 'f/', $foot) ] ];
201 # /$LISTNAME/$MESSAGE_ID/f/ -> HTML content (fullquotes)
204 my $x = mid2blob($ctx) or return r404($ctx);
206 require PublicInbox::View;
207 my $foot = footer($ctx);
209 my $mime = Email::MIME->new($x);
211 [ 200, [ 'Content-Type' => 'text/html; charset=UTF-8' ],
212 [ PublicInbox::View::msg_html($ctx, $mime, undef, $foot)] ];
215 # /$LISTNAME/$MESSAGE_ID/R/ -> HTML content (fullquotes)
218 my $x = mid2blob($ctx) or return r404($ctx);
220 require PublicInbox::View;
221 my $foot = footer($ctx);
223 my $hdr = Email::MIME->new($x)->header_obj;
224 [ 200, [ 'Content-Type' => 'text/html; charset=UTF-8' ],
225 [ PublicInbox::View::msg_reply($ctx, $hdr, $foot)] ];
228 # /$LISTNAME/$MESSAGE_ID/t/
230 my ($ctx, $flat) = @_;
231 my $srch = searcher($ctx) or return need_search($ctx);
232 require PublicInbox::View;
233 my $foot = footer($ctx);
234 $ctx->{flat} = $flat;
235 PublicInbox::View::thread_html($ctx, $foot, $srch);
239 my ($ctx, $key) = @_;
240 my $val = $ctx->{$key};
241 (defined $val && $val ne '') or die "BUG: bad ctx, $key unusable\n";
247 return '' unless $ctx;
248 my $git_dir = ctx_get($ctx, 'git_dir');
250 # favor user-supplied footer
251 my $footer = try_cat("$git_dir/public-inbox/footer.html");
252 if (defined $footer) {
254 $ctx->{footer} = $footer;
258 # auto-generate a footer
259 my $listname = ctx_get($ctx, 'listname');
260 my $desc = try_cat("$git_dir/description");
261 $desc = '$GIT_DIR/description missing' unless defined $desc;
264 my $urls = try_cat("$git_dir/cloneurl");
265 my @urls = split(/\r?\n/, $urls || '');
266 my %seen = map { $_ => 1 } @urls;
267 my $cgi = $ctx->{cgi};
268 my $http = $cgi->base->as_string . $listname;
269 $seen{$http} or unshift @urls, $http;
270 my $ssoma_url = PublicInbox::Hval::prurl($cgi->{env}, SSOMA_URL);
271 if (scalar(@urls) == 1) {
272 $urls = "URL for <a\nhref=\"" . $ssoma_url .
273 qq(">ssoma</a> or <b>git clone --mirror \$URL</b> :) .
276 $urls = "URLs for <a\nhref=\"" . $ssoma_url .
277 qq(">ssoma</a> or <b>git clone --mirror \$URL</b>\n) .
278 join("\n", map { "\t$_" } @urls);
281 my $addr = $ctx->{pi_config}->get($listname, 'address');
282 if (ref($addr) eq 'ARRAY') {
283 $addr = $addr->[0]; # first address is primary
286 $addr = "<a\nhref=\"mailto:$addr\">$addr</a>";
288 $ctx->{footer} = join("\n",
291 PublicInbox::Hval::prurl($ctx->{cgi}->{env}, PI_URL) .
292 '">public-inbox</a>, ' .
293 'anybody may post in plain-text (not HTML):',
299 # search support is optional, returns undef if Xapian is not installed
300 # or not configured for the given GIT_DIR
304 require PublicInbox::Search;
305 $ctx->{srch} = PublicInbox::Search->new($ctx->{git_dir});
312 <html><head><title>Search not available for this
313 public-inbox</title><body><pre>Search is not available for this public-inbox
314 <a href="../">Return to index</a></pre></body></html>
316 [ 501, [ 'Content-Type' => 'text/html; charset=UTF-8' ], [ $msg ] ];
319 # /$LISTNAME/$MESSAGE_ID/t.mbox -> thread as mbox
320 # /$LISTNAME/$MESSAGE_ID/t.mbox.gz -> thread as gzipped mbox
321 # note: I'm not a big fan of other compression formats since they're
322 # significantly more expensive on CPU than gzip and less-widely available,
323 # especially on older systems. Stick to zlib since that's what git uses.
324 sub get_thread_mbox {
325 my ($ctx, $sfx) = @_;
326 my $srch = searcher($ctx) or return need_search($ctx);
327 require PublicInbox::Mbox;
328 PublicInbox::Mbox::thread_mbox($ctx, $srch, $sfx);
332 # /$LISTNAME/$MESSAGE_ID/t.atom -> thread as Atom feed
333 sub get_thread_atom {
335 searcher($ctx) or return need_search($ctx);
336 $ctx->{self_url} = $ctx->{cgi}->uri->as_string;
337 require PublicInbox::Feed;
338 PublicInbox::Feed::generate_thread_atom($ctx);
341 sub legacy_redirects {
342 my ($ctx, $path_info) = @_;
344 # single-message pages
345 if ($path_info =~ m!$LISTNAME_RE/m/(\S+)/\z!o) {
347 } elsif ($path_info =~ m!$LISTNAME_RE/m/(\S+)/raw\z!o) {
348 r301($ctx, $1, $2, 'raw');
350 } elsif ($path_info =~ m!$LISTNAME_RE/f/(\S+)/\z!o) {
351 r301($ctx, $1, $2, 'f/');
354 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)/\z!o) {
355 r301($ctx, $1, $2, 't/#u');
357 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)/mbox(\.gz)?\z!o) {
358 r301($ctx, $1, $2, "t.mbox$3");
360 # even older legacy redirects
361 } elsif ($path_info =~ m!$LISTNAME_RE/m/(\S+)\.html\z!o) {
364 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)\.html\z!o) {
365 r301($ctx, $1, $2, 't/#u');
367 } elsif ($path_info =~ m!$LISTNAME_RE/f/(\S+)\.html\z!o) {
368 r301($ctx, $1, $2, 'f/');
370 } elsif ($path_info =~ m!$LISTNAME_RE/(?:m|f)/(\S+)\.txt\z!o) {
371 r301($ctx, $1, $2, 'raw');
373 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)(\.mbox(?:\.gz)?)\z!o) {
374 r301($ctx, $1, $2, "t$3");
376 # legacy convenience redirects, order still matters
377 } elsif ($path_info =~ m!$LISTNAME_RE/m/(\S+)\z!o) {
379 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)\z!o) {
380 r301($ctx, $1, $2, 't/#u');
381 } elsif ($path_info =~ m!$LISTNAME_RE/f/(\S+)\z!o) {
382 r301($ctx, $1, $2, 'f/');
384 # some Message-IDs have slashes in them and the HTTP server
385 # may try to be clever and unescape them :<
386 } elsif ($path_info =~ m!$LISTNAME_RE/(\S+/\S+)/$END_RE\z!o) {
387 msg_page($ctx, $1, $2, $3);
389 # in case people leave off the trailing slash:
390 } elsif ($path_info =~ m!$LISTNAME_RE/(\S+/\S+)/(f|T|t)\z!o) {
391 r301($ctx, $1, $2, $3 eq 't' ? 't/#u' : $3);
398 my ($ctx, $listname, $mid, $suffix) = @_;
399 my $cgi = $ctx->{cgi};
401 my $qs = $cgi->env->{QUERY_STRING};
402 $url = $cgi->base->as_string . $listname . '/';
403 $url .= (uri_escape_utf8($mid) . '/') if (defined $mid);
404 $url .= $suffix if (defined $suffix);
405 $url .= "?$qs" if $qs ne '';
408 [ Location => $url, 'Content-Type' => 'text/plain' ],
409 [ "Redirecting to $url\n" ] ]
413 my ($ctx, $list, $mid, $e) = @_;
414 unless (invalid_list_mid($ctx, $list, $mid)) {
415 '' eq $e and return get_mid_html($ctx);
416 't/' eq $e and return get_thread($ctx);
417 't.atom' eq $e and return get_thread_atom($ctx);
418 't.mbox' eq $e and return get_thread_mbox($ctx);
419 't.mbox.gz' eq $e and return get_thread_mbox($ctx, '.gz');
420 'T/' eq $e and return get_thread($ctx, 1);
421 'raw' eq $e and return get_mid_txt($ctx);
422 'f/' eq $e and return get_full_html($ctx);
423 'R/' eq $e and return get_reply_html($ctx);
429 my ($cgi, $git, $path) = @_;
430 PublicInbox::GitHTTPBackend::serve($cgi, $git, $path);