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;
18 use PublicInbox::Hval;
19 use URI::Escape qw(uri_escape_utf8 uri_unescape);
20 use constant SSOMA_URL => '//ssoma.public-inbox.org/';
21 use constant PI_URL => '//public-inbox.org/';
22 require PublicInbox::Git;
23 use PublicInbox::GitHTTPBackend;
24 our $INBOX_RE = qr!\A/([\w\.\-]+)!;
25 our $MID_RE = qr!([^/]+)!;
26 our $END_RE = qr!(T/|t/|t\.mbox(?:\.gz)?|t\.atom|raw|)!;
27 our $ATTACH_RE = qr!(\d[\.\d]*)-([[:alnum:]][\w\.-]+[[:alnum:]])!i;
30 my ($class, $pi_config) = @_;
31 $pi_config ||= PublicInbox::Config->new;
32 bless { pi_config => $pi_config }, $class;
35 # backwards compatibility, do not use
37 my ($req, $method) = @_;
38 PublicInbox::WWW->new->call($req->env);
42 my ($self, $env) = @_;
43 my $cgi = Plack::Request->new($env);
44 my $ctx = { cgi => $cgi, env => $env, www => $self,
45 pi_config => $self->{pi_config} };
47 # we don't care about multi-value
49 my ($k, $v) = split('=', $_, 2);
50 $v = '' unless defined $v;
53 } split(/[&;]/, uri_unescape($env->{QUERY_STRING}));
56 my $path_info = $env->{PATH_INFO};
57 my $method = $env->{REQUEST_METHOD};
59 if ($method eq 'POST' &&
60 $path_info =~ m!$INBOX_RE/(git-upload-pack)\z!) {
62 return (invalid_inbox($self, $ctx, $1) ||
63 serve_git($env, $ctx->{git}, $path));
65 elsif ($method !~ /\AGET|HEAD\z/) {
66 return r(405, 'Method Not Allowed');
69 # top-level indices and feeds
70 if ($path_info eq '/') {
72 } elsif ($path_info =~ m!$INBOX_RE\z!o) {
73 invalid_inbox($self, $ctx, $1) || r301($ctx, $1);
74 } elsif ($path_info =~ m!$INBOX_RE(?:/|/index\.html)?\z!o) {
75 invalid_inbox($self, $ctx, $1) || get_index($ctx);
76 } elsif ($path_info =~ m!$INBOX_RE/(?:atom\.xml|new\.atom)\z!o) {
77 invalid_inbox($self, $ctx, $1) || get_atom($ctx);
78 } elsif ($path_info =~ m!$INBOX_RE/new\.html\z!o) {
79 invalid_inbox($self, $ctx, $1) || get_new($ctx);
80 } elsif ($path_info =~ m!$INBOX_RE/
81 ($PublicInbox::GitHTTPBackend::ANY)\z!ox) {
83 invalid_inbox($self, $ctx, $1) ||
84 serve_git($env, $ctx->{git}, $path);
85 } elsif ($path_info =~ m!$INBOX_RE/([\w-]+).mbox\.gz\z!o) {
86 serve_mbox_range($self, $ctx, $1, $2);
87 } elsif ($path_info =~ m!$INBOX_RE/$MID_RE/$END_RE\z!o) {
88 msg_page($self, $ctx, $1, $2, $3);
90 } elsif ($path_info =~ m!$INBOX_RE/$MID_RE/$ATTACH_RE\z!o) {
91 my ($idx, $fn) = ($3, $4);
92 invalid_inbox_mid($self, $ctx, $1, $2) ||
93 get_attach($ctx, $idx, $fn);
94 # in case people leave off the trailing slash:
95 } elsif ($path_info =~ m!$INBOX_RE/$MID_RE/(T|t)\z!o) {
96 my ($inbox, $mid, $suffix) = ($1, $2, $3);
97 $suffix .= $suffix =~ /\A[tT]\z/ ? '/#u' : '/';
98 r301($ctx, $inbox, $mid, $suffix);
100 } elsif ($path_info =~ m!$INBOX_RE/$MID_RE/R/?\z!o) {
101 my ($inbox, $mid) = ($1, $2);
102 r301($ctx, $inbox, $mid, '#R');
104 } elsif ($path_info =~ m!$INBOX_RE/$MID_RE/f/?\z!o) {
107 # convenience redirects order matters
108 } elsif ($path_info =~ m!$INBOX_RE/([^/]{2,})\z!o) {
112 legacy_redirects($self, $ctx, $path_info);
116 # for CoW-friendliness, MOOOOO!
118 require PublicInbox::Feed;
119 require PublicInbox::View;
120 require PublicInbox::Thread;
125 foreach (qw(PublicInbox::Search PublicInbox::SearchView
126 PublicInbox::Mbox IO::Compress::Gzip
127 PublicInbox::NewsWWW)) {
132 # private functions below
136 if ($ctx && $ctx->{mid}) {
137 require PublicInbox::ExtMsg;
139 return PublicInbox::ExtMsg::ext_msg($ctx);
144 # simple response for errors
145 sub r { [ $_[0], ['Content-Type' => 'text/plain'], [ join(' ', @_, "\n") ] ] }
147 # returns undef if valid, array ref response if invalid
149 my ($self, $ctx, $inbox) = @_;
150 my $obj = $ctx->{pi_config}->lookup_name($inbox);
152 $ctx->{git_dir} = $obj->{mainrepo};
153 $ctx->{git} = $obj->git;
154 # for PublicInbox::HTTP::weaken_task:
155 $ctx->{cgi}->{env}->{'pi-httpd.inbox'} = $obj;
156 $ctx->{-inbox} = $obj;
157 $ctx->{inbox} = $inbox;
161 # sometimes linkifiers (not ours!) screw up automatic link
162 # generation and link things intended for nntp:// to https?://,
163 # so try to infer links and redirect them to the appropriate
165 $self->news_www->call($ctx->{cgi}->{env});
168 # returns undef if valid, array ref response if invalid
169 sub invalid_inbox_mid {
170 my ($self, $ctx, $inbox, $mid) = @_;
171 my $ret = invalid_inbox($self, $ctx, $inbox);
174 $ctx->{mid} = $mid = uri_unescape($mid);
175 if ($mid =~ /\A[a-f0-9]{40}\z/) {
176 # this is horiffically wasteful for legacy URLs:
177 if ($mid = mid2blob($ctx)) {
178 require Email::Simple;
179 use PublicInbox::MID qw/mid_clean/;
180 my $s = Email::Simple->new($mid);
181 $ctx->{mid} = mid_clean($s->header('Message-ID'));
187 # /$INBOX/new.atom -> Atom feed, includes replies
190 require PublicInbox::Feed;
191 PublicInbox::Feed::generate($ctx);
194 # /$INBOX/new.html -> HTML only
197 require PublicInbox::Feed;
198 PublicInbox::Feed::new_html($ctx);
201 # /$INBOX/?r=$GIT_COMMIT -> HTML only
204 require PublicInbox::Feed;
205 my $srch = searcher($ctx);
207 if ($ctx->{env}->{QUERY_STRING} =~ /(?:\A|[&;])q=/) {
208 require PublicInbox::SearchView;
209 PublicInbox::SearchView::sres_top_html($ctx);
211 PublicInbox::Feed::generate_html_index($ctx);
215 # just returns a string ref for the blob in the current ctx
218 $ctx->{-inbox}->msg_by_mid($ctx->{mid});
221 # /$INBOX/$MESSAGE_ID/raw -> raw mbox
224 my $x = mid2blob($ctx) or return r404($ctx);
225 require PublicInbox::Mbox;
226 PublicInbox::Mbox::emit1($ctx, $x);
229 # /$INBOX/$MESSAGE_ID/ -> HTML content (short quotes)
232 my $x = mid2blob($ctx) or return r404($ctx);
234 require PublicInbox::View;
235 my $foot = footer($ctx);
237 my $mime = Email::MIME->new($x);
239 PublicInbox::View::msg_html($ctx, $mime, $foot);
242 # /$INBOX/$MESSAGE_ID/t/
244 my ($ctx, $flat) = @_;
245 searcher($ctx) or return need_search($ctx);
246 $ctx->{flat} = $flat;
247 require PublicInbox::View;
248 PublicInbox::View::thread_html($ctx);
252 my ($ctx, $key) = @_;
253 my $val = $ctx->{$key};
254 (defined $val && $val ne '') or die "BUG: bad ctx, $key unusable";
260 return '' unless $ctx;
261 my $obj = $ctx->{-inbox} or return '';
263 # auto-generate a footer
264 chomp(my $desc = $obj->description);
265 $desc = PublicInbox::Hval::ascii_html($desc);
268 my @urls = @{$obj->cloneurl};
269 my %seen = map { $_ => 1 } @urls;
270 my $cgi = $ctx->{cgi};
271 my $http = $cgi->base->as_string . $obj->{name};
272 $seen{$http} or unshift @urls, $http;
273 my $ssoma_url = PublicInbox::Hval::prurl($ctx->{env}, SSOMA_URL);
274 if (scalar(@urls) == 1) {
275 $urls = "URL for <a\nhref=\"" . $ssoma_url .
276 qq(">ssoma</a> or <b>git clone --mirror $urls[0]</b>);
278 $urls = "URLs for <a\nhref=\"" . $ssoma_url .
279 qq(">ssoma</a> or <b>git clone --mirror</b>\n) .
280 join("\n", map { "\tgit clone --mirror $_" } @urls);
283 my $addr = $obj->{-primary_address};
284 $ctx->{footer} = join("\n",
287 PublicInbox::Hval::prurl($ctx->{cgi}->{env}, PI_URL) .
288 '">public-inbox</a>, ' .
289 'anybody may post in plain-text (not HTML):',
295 # search support is optional, returns undef if Xapian is not installed
296 # or not configured for the given GIT_DIR
300 require PublicInbox::Search;
301 $ctx->{srch} = $ctx->{-inbox}->search;
308 <html><head><title>Search not available for this
309 public-inbox</title><body><pre>Search is not available for this public-inbox
310 <a href="../">Return to index</a></pre></body></html>
312 [ 501, [ 'Content-Type' => 'text/html; charset=UTF-8' ], [ $msg ] ];
315 # /$INBOX/$MESSAGE_ID/t.mbox -> thread as mbox
316 # /$INBOX/$MESSAGE_ID/t.mbox.gz -> thread as gzipped mbox
317 # note: I'm not a big fan of other compression formats since they're
318 # significantly more expensive on CPU than gzip and less-widely available,
319 # especially on older systems. Stick to zlib since that's what git uses.
320 sub get_thread_mbox {
321 my ($ctx, $sfx) = @_;
322 my $srch = searcher($ctx) or return need_search($ctx);
323 require PublicInbox::Mbox;
324 PublicInbox::Mbox::thread_mbox($ctx, $srch, $sfx);
328 # /$INBOX/$MESSAGE_ID/t.atom -> thread as Atom feed
329 sub get_thread_atom {
331 searcher($ctx) or return need_search($ctx);
332 $ctx->{self_url} = $ctx->{cgi}->uri->as_string;
333 require PublicInbox::Feed;
334 PublicInbox::Feed::generate_thread_atom($ctx);
337 sub legacy_redirects {
338 my ($self, $ctx, $path_info) = @_;
340 # single-message pages
341 if ($path_info =~ m!$INBOX_RE/m/(\S+)/\z!o) {
343 } elsif ($path_info =~ m!$INBOX_RE/m/(\S+)/raw\z!o) {
344 r301($ctx, $1, $2, 'raw');
346 } elsif ($path_info =~ m!$INBOX_RE/f/(\S+)/\z!o) {
350 } elsif ($path_info =~ m!$INBOX_RE/t/(\S+)/\z!o) {
351 r301($ctx, $1, $2, 't/#u');
353 } elsif ($path_info =~ m!$INBOX_RE/t/(\S+)/mbox(\.gz)?\z!o) {
354 r301($ctx, $1, $2, "t.mbox$3");
356 # even older legacy redirects
357 } elsif ($path_info =~ m!$INBOX_RE/m/(\S+)\.html\z!o) {
360 } elsif ($path_info =~ m!$INBOX_RE/t/(\S+)\.html\z!o) {
361 r301($ctx, $1, $2, 't/#u');
363 } elsif ($path_info =~ m!$INBOX_RE/f/(\S+)\.html\z!o) {
366 } elsif ($path_info =~ m!$INBOX_RE/(?:m|f)/(\S+)\.txt\z!o) {
367 r301($ctx, $1, $2, 'raw');
369 } elsif ($path_info =~ m!$INBOX_RE/t/(\S+)(\.mbox(?:\.gz)?)\z!o) {
370 r301($ctx, $1, $2, "t$3");
372 # legacy convenience redirects, order still matters
373 } elsif ($path_info =~ m!$INBOX_RE/m/(\S+)\z!o) {
375 } elsif ($path_info =~ m!$INBOX_RE/t/(\S+)\z!o) {
376 r301($ctx, $1, $2, 't/#u');
377 } elsif ($path_info =~ m!$INBOX_RE/f/(\S+)\z!o) {
380 # some Message-IDs have slashes in them and the HTTP server
381 # may try to be clever and unescape them :<
382 } elsif ($path_info =~ m!$INBOX_RE/(\S+/\S+)/$END_RE\z!o) {
383 msg_page($self, $ctx, $1, $2, $3);
385 # in case people leave off the trailing slash:
386 } elsif ($path_info =~ m!$INBOX_RE/(\S+/\S+)/(T|t)\z!o) {
387 r301($ctx, $1, $2, $3 eq 't' ? 't/#u' : $3);
388 } elsif ($path_info =~ m!$INBOX_RE/(\S+/\S+)/f\z!o) {
391 $self->news_www->call($ctx->{cgi}->{env});
396 my ($ctx, $inbox, $mid, $suffix) = @_;
397 my $cgi = $ctx->{cgi};
398 my $obj = $ctx->{-inbox};
400 my $r404 = invalid_inbox($ctx->{www}, $ctx, $inbox);
401 return $r404 if $r404;
402 $obj = $ctx->{-inbox};
404 my $url = $obj->base_url($cgi);
405 my $qs = $ctx->{env}->{QUERY_STRING};
406 $url .= (uri_escape_utf8($mid) . '/') if (defined $mid);
407 $url .= $suffix if (defined $suffix);
408 $url .= "?$qs" if $qs ne '';
411 [ Location => $url, 'Content-Type' => 'text/plain' ],
412 [ "Redirecting to $url\n" ] ]
416 my ($self, $ctx, $inbox, $mid, $e) = @_;
418 $ret = invalid_inbox_mid($self, $ctx, $inbox, $mid) and return $ret;
419 '' eq $e and return get_mid_html($ctx);
420 'T/' eq $e and return get_thread($ctx, 1);
421 't/' eq $e and return get_thread($ctx);
422 't.atom' eq $e and return get_thread_atom($ctx);
423 't.mbox' eq $e and return get_thread_mbox($ctx);
424 't.mbox.gz' eq $e and return get_thread_mbox($ctx, '.gz');
425 'raw' eq $e and return get_mid_txt($ctx);
427 # legacy, but no redirect for compatibility:
428 'f/' eq $e and return get_mid_html($ctx);
433 my ($env, $git, $path) = @_;
434 PublicInbox::GitHTTPBackend::serve($env, $git, $path);
437 sub serve_mbox_range {
438 my ($self, $ctx, $inbox, $range) = @_;
439 invalid_inbox($self, $ctx, $inbox) || eval {
440 require PublicInbox::Mbox;
442 PublicInbox::Mbox::emit_range($ctx, $range);
448 my $nw = $self->{news_www};
450 require PublicInbox::NewsWWW;
451 $self->{news_www} = PublicInbox::NewsWWW->new($self->{pi_config});
455 my ($ctx, $idx, $fn) = @_;
456 require PublicInbox::WwwAttach;
457 PublicInbox::WwwAttach::get_attach($ctx, $idx, $fn);