1 # Copyright (C) 2013-2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
4 # Used for generating Atom feeds for web-accessible mailing list archives.
5 package PublicInbox::Feed;
10 use Date::Parse qw(strptime);
11 use PublicInbox::Hval;
12 use PublicInbox::GitCatFile;
13 use PublicInbox::View;
14 use PublicInbox::MID qw/mid_clean mid2path/;
15 use POSIX qw/strftime/;
17 DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # Atom standard
18 MAX_PER_PAGE => 25, # this needs to be tunable
21 use Encode qw/find_encoding/;
22 my $enc_utf8 = find_encoding('UTF-8');
27 sub { emit_atom($_[0], $ctx) };
30 sub generate_thread_atom {
32 sub { emit_atom_thread($_[0], $ctx) };
35 sub generate_html_index {
37 sub { emit_html_index($_[0], $ctx) };
44 # try to avoid the type attribute in title:
45 $title = PublicInbox::Hval->new_oneline($title)->as_html;
46 my $type = index($title, '&') >= 0 ? "\ntype=\"html\"" : '';
47 "<title$type>$title</title>";
51 my ($feed_opts, $title) = @_;
53 $title = title_tag($feed_opts->{description}) unless (defined $title);
55 qq(<?xml version="1.0" encoding="us-ascii"?>\n) .
56 qq{<feed\nxmlns="http://www.w3.org/2005/Atom">} .
58 qq(<link\nhref="$feed_opts->{url}"/>) .
59 qq(<link\nrel="self"\nhref="$feed_opts->{atomurl}"/>) .
60 qq(<id>mailto:$feed_opts->{id_addr}</id>);
65 my $fh = $cb->([ 200, ['Content-Type' => 'application/atom+xml']]);
66 my $max = $ctx->{max} || MAX_PER_PAGE;
67 my $feed_opts = get_feedopts($ctx);
68 my $x = atom_header($feed_opts);
69 my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
70 each_recent_blob($ctx, sub {
71 my ($path, undef, $ts) = @_;
73 $fh->write($x . feed_updated(undef, $ts));
76 add_to_feed($feed_opts, $fh, $path, $git);
78 $git = undef; # destroy pipes
84 my $fh = $cb->([404, ['Content-Type' => 'text/plain']]);
85 $fh->write("No feed found for thread\n");
91 Email::Address->purge_cache;
92 $fh->write('</feed>');
96 sub emit_atom_thread {
98 my $res = $ctx->{srch}->get_thread($ctx->{mid});
99 return _no_thread($cb) unless $res->{total};
100 my $fh = $cb->([200, ['Content-Type' => 'application/atom+xml']]);
101 my $feed_opts = get_feedopts($ctx);
103 my $html_url = $feed_opts->{atomurl} = $ctx->{self_url};
104 $html_url =~ s!/t\.atom\z!/!;
105 $feed_opts->{url} = $html_url;
106 $feed_opts->{emit_header} = 1;
108 my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
109 foreach my $msg (@{$res->{msgs}}) {
110 add_to_feed($feed_opts, $fh, mid2path($msg->mid), $git);
112 $git = undef; # destroy pipes
116 sub emit_html_index {
118 my $fh = $cb->([200,['Content-Type'=>'text/html; charset=UTF-8']]);
120 my $max = $ctx->{max} || MAX_PER_PAGE;
121 my $feed_opts = get_feedopts($ctx);
123 my $title = $feed_opts->{description} || '';
124 $title = PublicInbox::Hval->new_oneline($title)->as_html;
125 my $atom_url = $feed_opts->{atomurl};
126 my ($footer, $param, $last);
127 my $state = { ctx => $ctx, seen => {}, anchor_idx => 0 };
128 my $srch = $ctx->{srch};
130 my $top = "<b>$title</b> (<a\nhref=\"$atom_url\">Atom feed</a>)";
133 $top = qq{<form\naction=""><tt>$top} .
134 qq{ <input\nname=q\ntype=text />} .
135 qq{<input\ntype=submit\nvalue=search />} .
137 PublicInbox::View::PRE_WRAP;
139 $top = PublicInbox::View::PRE_WRAP . $top . "\n";
142 $fh->write("<html><head><title>$title</title>" .
143 "<link\nrel=alternate\ntitle=\"Atom feed\"\n".
144 "href=\"$atom_url\"\ntype=\"application/atom+xml\"/>" .
145 "</head><body>$top");
147 # if the 'r' query parameter is given, it is a legacy permalink
148 # which we must continue supporting:
149 my $cgi = $ctx->{cgi};
150 if ($cgi && !$cgi->param('r') && $srch) {
151 $state->{srch} = $srch;
152 $last = PublicInbox::View::emit_index_topics($state, $fh);
155 $last = emit_index_nosrch($ctx, $state, $fh);
158 $footer = nav_footer($cgi, $last, $feed_opts, $state, $param);
160 my $list_footer = $ctx->{footer};
161 $footer .= "\n\n" . $list_footer if $list_footer;
162 $footer = "<hr /><pre>$footer</pre>";
164 $fh->write("$footer</body></html>");
168 sub emit_index_nosrch {
169 my ($ctx, $state, $fh) = @_;
170 my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
171 my (undef, $last) = each_recent_blob($ctx, sub {
172 my ($path, $commit, $ts, $u, $subj) = @_;
173 $state->{first} ||= $commit;
175 my $mime = do_cat_mail($git, $path) or return 0;
176 PublicInbox::View::index_entry($fh, $mime, 0, $state);
179 Email::Address->purge_cache;
184 my ($cgi, $last, $feed_opts, $state, $param) = @_;
186 my $old_r = $cgi->param($param);
189 my $first = $state->{first};
190 my $anchor = $state->{anchor_idx};
193 $next = qq!<a\nhref="?$param=$last">next</a>!;
196 $head = $cgi->path_info;
197 $head = qq!<a\nhref="$head">head</a>!;
199 my $atom = "<a\nhref=\"$feed_opts->{atomurl}\">Atom</a>";
200 "<a\nname=\"s$anchor\">page:</a> $next $head $atom";
203 sub each_recent_blob {
205 my $max = $ctx->{max} || MAX_PER_PAGE;
206 my $hex = '[a-f0-9]';
207 my $addmsg = qr!^:000000 100644 \S+ \S+ A\t(${hex}{2}/${hex}{38})$!;
208 my $delmsg = qr!^:100644 000000 \S+ \S+ D\t(${hex}{2}/${hex}{38})$!;
209 my $refhex = qr/(?:HEAD|${hex}{4,40})(?:~\d+)?/;
210 my $cgi = $ctx->{cgi};
212 # revision ranges may be specified
214 my $r = $cgi->param('r') if $cgi;
215 if ($r && ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o)) {
219 # get recent messages
220 # we could use git log -z, but, we already know ssoma will not
221 # leave us with filenames with spaces in them..
222 my @cmd = ('git', "--git-dir=$ctx->{git_dir}",
223 qw/log --no-notes --no-color --raw -r
224 --abbrev=16 --abbrev-commit/,
225 "--format=%h%x00%ct%x00%an%x00%s%x00");
228 my $pid = open(my $log, '-|', @cmd) or
229 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
230 my %deleted; # only an optimization at this point
233 my ($cur_commit, $first_commit, $last_commit);
235 while (defined(my $line = <$log>)) {
236 if ($line =~ /$addmsg/o) {
238 next if $deleted{$add}; # optimization-only
239 $nr += $cb->($add, $cur_commit, $ts, $u, $subj);
244 } elsif ($line =~ /$delmsg/o) {
246 } elsif ($line =~ /^${hex}{7,40}/o) {
247 ($cur_commit, $ts, $u, $subj) = split("\0", $line);
248 unless (defined $first_commit) {
249 $first_commit = $cur_commit;
255 while (my $line = <$log>) {
256 if ($line =~ /^(${hex}{7,40})/o) {
263 close $log; # we may EPIPE here
265 ($first_commit, $last_commit);
268 # private functions below
271 my $pi_config = $ctx->{pi_config};
272 my $listname = $ctx->{listname};
273 my $cgi = $ctx->{cgi};
275 if (open my $fh, '<', "$ctx->{git_dir}/description") {
276 chomp($rv{description} = <$fh>);
279 $rv{description} = '($GIT_DIR/description missing)';
282 if ($pi_config && defined $listname && $listname ne '') {
283 my $addr = $pi_config->get($listname, 'address') || "";
284 $rv{address} = $addr;
285 $addr = $addr->[0] if ref($addr);
286 $rv{id_addr} = $addr;
288 $rv{id_addr} ||= 'public-inbox@example.com';
293 if (ref($cgi) eq 'CGI') {
294 $base = $cgi->url(-base);
296 $base = $cgi->base->as_string;
299 $url_base = "$base/$listname";
300 if (my $mid = $ctx->{mid}) { # per-thread feed:
301 $rv{atomurl} = "$url_base/$mid/t.atom";
303 $rv{atomurl} = "$url_base/new.atom";
306 $url_base = "http://example.com";
307 $rv{atomurl} = "$url_base/new.atom";
309 $rv{url} ||= "$url_base/";
310 $rv{midurl} = "$url_base/";
316 my ($mime, $name) = @_;
317 PublicInbox::Hval->new_oneline($mime->header($name))->raw;
321 my ($date, $ts) = @_;
322 my @t = eval { strptime($date) } if defined $date;
323 @t = gmtime($ts || time) unless scalar @t;
325 '<updated>' . strftime(DATEFMT, @t) . '</updated>';
328 # returns 0 (skipped) or 1 (added)
330 my ($feed_opts, $fh, $add, $git) = @_;
332 my $mime = do_cat_mail($git, $add) or return 0;
333 my $url = $feed_opts->{url};
334 my $midurl = $feed_opts->{midurl};
336 my $header_obj = $mime->header_obj;
337 my $mid = $header_obj->header('Message-ID');
338 defined $mid or return 0;
339 $mid = PublicInbox::Hval->new_msgid($mid);
340 my $href = $mid->as_href;
341 my $content = PublicInbox::View->feed_entry($mime, "$midurl$href/f/");
342 defined($content) or return 0;
345 my $date = $header_obj->header('Date');
346 my $updated = feed_updated($date);
348 my $title = mime_header($header_obj, 'Subject') or return 0;
349 $title = title_tag($title);
351 my $from = mime_header($header_obj, 'From') or return 0;
352 my @from = Email::Address->parse($from) or return 0;
353 my $name = PublicInbox::Hval->new_oneline($from[0]->name)->as_html;
354 my $email = $from[0]->address;
355 $email = PublicInbox::Hval->new_oneline($email)->as_html;
357 if (delete $feed_opts->{emit_header}) {
358 $fh->write(atom_header($feed_opts, $title) . $updated);
360 $fh->write("<entry><author><name>$name</name><email>$email</email>" .
361 "</author>$title$updated" .
362 qq{<content\ntype="xhtml">} .
363 qq{<div\nxmlns="http://www.w3.org/1999/xhtml">});
364 $fh->write($content);
368 my (@uuid5) = ($add =~ m!\A($h{8})($h{4})($h{4})($h{4})($h{12})!o);
369 my $id = 'urn:uuid:' . join('-', @uuid5);
370 $fh->write(qq!</div></content><link\nhref="$midurl$href/"/>!.
371 "<id>$id</id></entry>");
376 my ($git, $path) = @_;
378 my $str = $git->cat_file("HEAD:$path");
379 Email::MIME->new($str);