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;
17 use PublicInbox::Config qw(try_cat);
18 use URI::Escape qw(uri_escape_utf8 uri_unescape);
19 use constant SSOMA_URL => '//ssoma.public-inbox.org/';
20 use constant PI_URL => '//public-inbox.org/';
21 require PublicInbox::Git;
22 use PublicInbox::GitHTTPBackend;
23 our $LISTNAME_RE = qr!\A/([\w\.\-]+)!;
24 our $MID_RE = qr!([^/]+)!;
25 our $END_RE = qr!(f/|T/|t/|R/|t\.mbox(?:\.gz)?|t\.atom|raw|)!;
28 my ($class, $pi_config) = @_;
29 $pi_config ||= PublicInbox::Config->new;
30 bless { pi_config => $pi_config }, $class;
33 # backwards compatibility, do not use
35 my ($req, $method) = @_;
36 PublicInbox::WWW->new->call($req->env);
40 my ($self, $env) = @_;
41 my $cgi = Plack::Request->new($env);
42 my $ctx = { cgi => $cgi, pi_config => $self->{pi_config} };
43 my $path_info = $cgi->path_info;
45 my $method = $cgi->method;
46 if ($method eq 'POST' &&
47 $path_info =~ m!$LISTNAME_RE/(git-upload-pack)\z!) {
49 return (invalid_list($self, $ctx, $1) ||
50 serve_git($cgi, $ctx->{git}, $path));
52 elsif ($method !~ /\AGET|HEAD\z/) {
53 return r(405, 'Method Not Allowed');
56 # top-level indices and feeds
57 if ($path_info eq '/') {
59 } elsif ($path_info =~ m!$LISTNAME_RE\z!o) {
60 invalid_list($self, $ctx, $1) || r301($ctx, $1);
61 } elsif ($path_info =~ m!$LISTNAME_RE(?:/|/index\.html)?\z!o) {
62 invalid_list($self, $ctx, $1) || get_index($ctx);
63 } elsif ($path_info =~ m!$LISTNAME_RE/(?:atom\.xml|new\.atom)\z!o) {
64 invalid_list($self, $ctx, $1) || get_atom($ctx);
66 } elsif ($path_info =~ m!$LISTNAME_RE/
67 ($PublicInbox::GitHTTPBackend::ANY)\z!ox) {
69 invalid_list($self, $ctx, $1) ||
70 serve_git($cgi, $ctx->{git}, $path);
71 } elsif ($path_info =~ m!$LISTNAME_RE/$MID_RE/$END_RE\z!o) {
72 msg_page($self, $ctx, $1, $2, $3);
74 # in case people leave off the trailing slash:
75 } elsif ($path_info =~ m!$LISTNAME_RE/$MID_RE/(f|T|t|R)\z!o) {
76 my ($listname, $mid, $suffix) = ($1, $2, $3);
77 $suffix .= $suffix =~ /\A[tT]\z/ ? '/#u' : '/';
78 r301($ctx, $listname, $mid, $suffix);
80 # convenience redirects order matters
81 } elsif ($path_info =~ m!$LISTNAME_RE/([^/]{2,})\z!o) {
85 legacy_redirects($self, $ctx, $path_info);
89 # for CoW-friendliness, MOOOOO!
91 require PublicInbox::Feed;
92 require PublicInbox::View;
93 require PublicInbox::Thread;
98 foreach (qw(PublicInbox::Search PublicInbox::SearchView
99 PublicInbox::Mbox IO::Compress::Gzip
100 PublicInbox::NewsWWW PublicInbox::NewsGroup)) {
105 # private functions below
109 if ($ctx && $ctx->{mid}) {
110 require PublicInbox::ExtMsg;
112 return PublicInbox::ExtMsg::ext_msg($ctx);
117 # simple response for errors
118 sub r { [ $_[0], ['Content-Type' => 'text/plain'], [ join(' ', @_, "\n") ] ] }
120 # returns undef if valid, array ref response if invalid
122 my ($self, $ctx, $listname, $mid) = @_;
123 my $git_dir = $ctx->{pi_config}->get($listname, "mainrepo");
124 if (defined $git_dir) {
125 $ctx->{git_dir} = $git_dir;
126 $ctx->{git} = PublicInbox::Git->new($git_dir);
127 $ctx->{listname} = $listname;
131 # sometimes linkifiers (not ours!) screw up automatic link
132 # generation and link things intended for nntp:// to https?://,
133 # so try to infer links and redirect them to the appropriate
135 $self->news_www->call($ctx->{cgi}->{env});
138 # returns undef if valid, array ref response if invalid
139 sub invalid_list_mid {
140 my ($self, $ctx, $listname, $mid) = @_;
141 my $ret = invalid_list($self, $ctx, $listname, $mid);
144 $ctx->{mid} = $mid = uri_unescape($mid);
145 if ($mid =~ /\A[a-f0-9]{40}\z/) {
146 # this is horiffically wasteful for legacy URLs:
147 if ($mid = mid2blob($ctx)) {
148 require Email::Simple;
149 use PublicInbox::MID qw/mid_clean/;
150 my $s = Email::Simple->new($mid);
151 $ctx->{mid} = mid_clean($s->header('Message-ID'));
157 # /$LISTNAME/new.atom -> Atom feed, includes replies
160 require PublicInbox::Feed;
161 PublicInbox::Feed::generate($ctx);
164 # /$LISTNAME/?r=$GIT_COMMIT -> HTML only
167 require PublicInbox::Feed;
168 my $srch = searcher($ctx);
170 if (defined $ctx->{cgi}->param('q')) {
171 require PublicInbox::SearchView;
172 PublicInbox::SearchView::sres_top_html($ctx);
174 PublicInbox::Feed::generate_html_index($ctx);
178 # just returns a string ref for the blob in the current ctx
181 require PublicInbox::MID;
182 my $path = PublicInbox::MID::mid2path($ctx->{mid});
183 $ctx->{git}->cat_file("HEAD:$path");
186 # /$LISTNAME/$MESSAGE_ID/raw -> raw mbox
189 my $x = mid2blob($ctx) or return r404($ctx);
190 require PublicInbox::Mbox;
191 PublicInbox::Mbox::emit1($ctx, $x);
194 # /$LISTNAME/$MESSAGE_ID/ -> HTML content (short quotes)
197 my $x = mid2blob($ctx) or return r404($ctx);
199 require PublicInbox::View;
200 my $foot = footer($ctx);
202 my $mime = Email::MIME->new($x);
204 [ 200, [ 'Content-Type' => 'text/html; charset=UTF-8' ],
205 [ PublicInbox::View::msg_html($ctx, $mime, 'f/', $foot) ] ];
208 # /$LISTNAME/$MESSAGE_ID/f/ -> HTML content (fullquotes)
211 my $x = mid2blob($ctx) or return r404($ctx);
213 require PublicInbox::View;
214 my $foot = footer($ctx);
216 my $mime = Email::MIME->new($x);
218 [ 200, [ 'Content-Type' => 'text/html; charset=UTF-8' ],
219 [ PublicInbox::View::msg_html($ctx, $mime, undef, $foot)] ];
222 # /$LISTNAME/$MESSAGE_ID/R/ -> HTML content (fullquotes)
225 my $x = mid2blob($ctx) or return r404($ctx);
227 require PublicInbox::View;
228 my $foot = footer($ctx);
230 my $hdr = Email::MIME->new($x)->header_obj;
231 [ 200, [ 'Content-Type' => 'text/html; charset=UTF-8' ],
232 [ PublicInbox::View::msg_reply($ctx, $hdr, $foot)] ];
235 # /$LISTNAME/$MESSAGE_ID/t/
237 my ($ctx, $flat) = @_;
238 my $srch = searcher($ctx) or return need_search($ctx);
239 require PublicInbox::View;
240 my $foot = footer($ctx);
241 $ctx->{flat} = $flat;
242 PublicInbox::View::thread_html($ctx, $foot, $srch);
246 my ($ctx, $key) = @_;
247 my $val = $ctx->{$key};
248 (defined $val && $val ne '') or die "BUG: bad ctx, $key unusable";
254 return '' unless $ctx;
255 my $git_dir = ctx_get($ctx, 'git_dir');
257 # favor user-supplied footer
258 my $footer = try_cat("$git_dir/public-inbox/footer.html");
259 if (defined $footer) {
261 $ctx->{footer} = $footer;
265 # auto-generate a footer
266 my $listname = ctx_get($ctx, 'listname');
267 my $desc = try_cat("$git_dir/description");
268 $desc = '$GIT_DIR/description missing' unless defined $desc;
271 my $urls = try_cat("$git_dir/cloneurl");
272 my @urls = split(/\r?\n/, $urls || '');
273 my %seen = map { $_ => 1 } @urls;
274 my $cgi = $ctx->{cgi};
275 my $http = $cgi->base->as_string . $listname;
276 $seen{$http} or unshift @urls, $http;
277 my $ssoma_url = PublicInbox::Hval::prurl($cgi->{env}, SSOMA_URL);
278 if (scalar(@urls) == 1) {
279 $urls = "URL for <a\nhref=\"" . $ssoma_url .
280 qq(">ssoma</a> or <b>git clone --mirror $urls[0]</b>);
282 $urls = "URLs for <a\nhref=\"" . $ssoma_url .
283 qq(">ssoma</a> or <b>git clone --mirror</b>\n) .
284 join("\n", map { "\tgit clone --mirror $_" } @urls);
287 my $addr = $ctx->{pi_config}->get($listname, 'address');
288 if (ref($addr) eq 'ARRAY') {
289 $addr = $addr->[0]; # first address is primary
292 $addr = "<a\nhref=\"mailto:$addr\">$addr</a>";
294 $ctx->{footer} = join("\n",
297 PublicInbox::Hval::prurl($ctx->{cgi}->{env}, PI_URL) .
298 '">public-inbox</a>, ' .
299 'anybody may post in plain-text (not HTML):',
305 # search support is optional, returns undef if Xapian is not installed
306 # or not configured for the given GIT_DIR
310 require PublicInbox::Search;
311 $ctx->{srch} = PublicInbox::Search->new($ctx->{git_dir});
318 <html><head><title>Search not available for this
319 public-inbox</title><body><pre>Search is not available for this public-inbox
320 <a href="../">Return to index</a></pre></body></html>
322 [ 501, [ 'Content-Type' => 'text/html; charset=UTF-8' ], [ $msg ] ];
325 # /$LISTNAME/$MESSAGE_ID/t.mbox -> thread as mbox
326 # /$LISTNAME/$MESSAGE_ID/t.mbox.gz -> thread as gzipped mbox
327 # note: I'm not a big fan of other compression formats since they're
328 # significantly more expensive on CPU than gzip and less-widely available,
329 # especially on older systems. Stick to zlib since that's what git uses.
330 sub get_thread_mbox {
331 my ($ctx, $sfx) = @_;
332 my $srch = searcher($ctx) or return need_search($ctx);
333 require PublicInbox::Mbox;
334 PublicInbox::Mbox::thread_mbox($ctx, $srch, $sfx);
338 # /$LISTNAME/$MESSAGE_ID/t.atom -> thread as Atom feed
339 sub get_thread_atom {
341 searcher($ctx) or return need_search($ctx);
342 $ctx->{self_url} = $ctx->{cgi}->uri->as_string;
343 require PublicInbox::Feed;
344 PublicInbox::Feed::generate_thread_atom($ctx);
347 sub legacy_redirects {
348 my ($self, $ctx, $path_info) = @_;
350 # single-message pages
351 if ($path_info =~ m!$LISTNAME_RE/m/(\S+)/\z!o) {
353 } elsif ($path_info =~ m!$LISTNAME_RE/m/(\S+)/raw\z!o) {
354 r301($ctx, $1, $2, 'raw');
356 } elsif ($path_info =~ m!$LISTNAME_RE/f/(\S+)/\z!o) {
357 r301($ctx, $1, $2, 'f/');
360 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)/\z!o) {
361 r301($ctx, $1, $2, 't/#u');
363 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)/mbox(\.gz)?\z!o) {
364 r301($ctx, $1, $2, "t.mbox$3");
366 # even older legacy redirects
367 } elsif ($path_info =~ m!$LISTNAME_RE/m/(\S+)\.html\z!o) {
370 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)\.html\z!o) {
371 r301($ctx, $1, $2, 't/#u');
373 } elsif ($path_info =~ m!$LISTNAME_RE/f/(\S+)\.html\z!o) {
374 r301($ctx, $1, $2, 'f/');
376 } elsif ($path_info =~ m!$LISTNAME_RE/(?:m|f)/(\S+)\.txt\z!o) {
377 r301($ctx, $1, $2, 'raw');
379 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)(\.mbox(?:\.gz)?)\z!o) {
380 r301($ctx, $1, $2, "t$3");
382 # legacy convenience redirects, order still matters
383 } elsif ($path_info =~ m!$LISTNAME_RE/m/(\S+)\z!o) {
385 } elsif ($path_info =~ m!$LISTNAME_RE/t/(\S+)\z!o) {
386 r301($ctx, $1, $2, 't/#u');
387 } elsif ($path_info =~ m!$LISTNAME_RE/f/(\S+)\z!o) {
388 r301($ctx, $1, $2, 'f/');
390 # some Message-IDs have slashes in them and the HTTP server
391 # may try to be clever and unescape them :<
392 } elsif ($path_info =~ m!$LISTNAME_RE/(\S+/\S+)/$END_RE\z!o) {
393 msg_page($self, $ctx, $1, $2, $3);
395 # in case people leave off the trailing slash:
396 } elsif ($path_info =~ m!$LISTNAME_RE/(\S+/\S+)/(f|T|t)\z!o) {
397 r301($ctx, $1, $2, $3 eq 't' ? 't/#u' : $3);
399 $self->news_www->call($ctx->{cgi}->{env});
404 my ($ctx, $listname, $mid, $suffix) = @_;
405 my $cgi = $ctx->{cgi};
407 my $qs = $cgi->env->{QUERY_STRING};
408 $url = $cgi->base->as_string . $listname . '/';
409 $url .= (uri_escape_utf8($mid) . '/') if (defined $mid);
410 $url .= $suffix if (defined $suffix);
411 $url .= "?$qs" if $qs ne '';
414 [ Location => $url, 'Content-Type' => 'text/plain' ],
415 [ "Redirecting to $url\n" ] ]
419 my ($self, $ctx, $list, $mid, $e) = @_;
421 $ret = invalid_list_mid($self, $ctx, $list, $mid) and return $ret;
422 '' eq $e and return get_mid_html($ctx);
423 't/' eq $e and return get_thread($ctx);
424 't.atom' eq $e and return get_thread_atom($ctx);
425 't.mbox' eq $e and return get_thread_mbox($ctx);
426 't.mbox.gz' eq $e and return get_thread_mbox($ctx, '.gz');
427 'T/' eq $e and return get_thread($ctx, 1);
428 'raw' eq $e and return get_mid_txt($ctx);
429 'f/' eq $e and return get_full_html($ctx);
430 'R/' eq $e and return get_reply_html($ctx);
435 my ($cgi, $git, $path) = @_;
436 PublicInbox::GitHTTPBackend::serve($cgi, $git, $path);
441 my $nw = $self->{news_www};
443 require PublicInbox::NewsWWW;
444 $self->{news_www} = PublicInbox::NewsWWW->new($self->{pi_config});