1 # Copyright (C) 2013, Eric Wong <normalperson@yhbt.net> and all contributors
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
3 package PublicInbox::Feed;
8 use Date::Parse qw(strptime str2time);
10 use PublicInbox::GitCatFile;
11 use PublicInbox::View;
13 DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # atom standard
14 MAX_PER_PAGE => 25, # this needs to be tunable
15 PRE_WRAP => "<pre\nstyle=\"white-space:pre-wrap\">",
20 my ($class, $ctx) = @_;
21 require XML::Atom::SimpleFeed;
23 my $max = $ctx->{max} || MAX_PER_PAGE;
25 my $feed_opts = get_feedopts($ctx);
26 my $addr = $feed_opts->{address};
27 $addr = $addr->[0] if ref($addr);
28 my $feed = XML::Atom::SimpleFeed->new(
29 title => $feed_opts->{description} || "unnamed feed",
30 link => $feed_opts->{url} || "http://example.com/",
33 href => $feed_opts->{atomurl} ||
34 "http://example.com/atom.xml",
36 id => 'mailto:' . ($addr || 'public-inbox@example.com'),
37 updated => POSIX::strftime(DATEFMT, gmtime),
41 my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
42 each_recent_blob($ctx, sub {
44 add_to_feed($feed_opts, $feed, $add, $git);
46 $git = undef; # destroy pipes
47 Email::Address->purge_cache;
51 sub generate_html_index {
52 my ($class, $ctx) = @_;
53 require PublicInbox::Thread;
55 my $max = $ctx->{max} || MAX_PER_PAGE;
56 my $feed_opts = get_feedopts($ctx);
58 my $title = $feed_opts->{description} || '';
59 $title = PublicInbox::Hval->new_oneline($title)->as_html;
62 my $git_dir = $ctx->{git_dir};
63 my $git = PublicInbox::GitCatFile->new($git_dir);
64 my ($first, $last) = each_recent_blob($ctx, sub {
65 mime_load_for_sort($git, $_[0], \@messages);
67 $git = undef; # destroy pipes.
69 my $th = PublicInbox::Thread->new(@messages);
71 my $html = "<html><head><title>$title</title>" .
72 '<link rel="alternate" title="Atom feed"' . "\nhref=\"" .
73 $feed_opts->{atomurl} . "\"\ntype=\"application/atom+xml\"/>" .
74 '</head><body>' . PRE_WRAP;
76 # sort child messages in chronological order
77 $th->order(*PublicInbox::Thread::sort_ts);
79 # except we sort top-level messages reverse chronologically
80 my $state = [ $ctx->{srch}, {}, $first, 0 ];
81 for (PublicInbox::Thread::rsort_ts($th->rootset)) {
82 dump_msg($_, 0, \$html, $state)
84 Email::Address->purge_cache;
86 my $footer = nav_footer($ctx->{cgi}, $last, $feed_opts, $state);
88 my $list_footer = $ctx->{footer};
89 $footer .= "\n" . $list_footer if $list_footer;
90 $footer = "<hr />" . PRE_WRAP . "$footer</pre>";
92 $html . "</pre>$footer</body></html>";
98 my ($cgi, $last, $feed_opts, $state) = @_;
100 my $old_r = $cgi->param('r');
103 my $first = $state->[2];
104 my $anchor = $state->[3];
107 $next = qq!<a\nhref="?r=$last">next</a>!;
110 $head = $cgi->path_info;
111 $head = qq!<a\nhref="$head">head</a>!;
113 my $atom = "<a\nhref=\"$feed_opts->{atomurl}\">atom</a>";
114 my $permalink = "<a\nhref=\"?r=$first\">permalink</a>";
115 "<a\nname=\"s$anchor\">page:</a> $next $head $atom $permalink";
118 sub each_recent_blob {
120 my $max = $ctx->{max} || MAX_PER_PAGE;
121 my $hex = '[a-f0-9]';
122 my $addmsg = qr!^:000000 100644 \S+ \S+ A\t(${hex}{2}/${hex}{38})$!;
123 my $delmsg = qr!^:100644 000000 \S+ \S+ D\t(${hex}{2}/${hex}{38})$!;
124 my $refhex = qr/${hex}{4,40}(?:~\d+)?/;
125 my $cgi = $ctx->{cgi};
127 # revision ranges may be specified
129 my $r = $cgi->param('r') if $cgi;
130 if ($r && ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o)) {
134 # get recent messages
135 # we could use git log -z, but, we already know ssoma will not
136 # leave us with filenames with spaces in them..
137 my @cmd = ('git', "--git-dir=$ctx->{git_dir}",
138 qw/log --no-notes --no-color --raw -r
139 --abbrev=16 --abbrev-commit/);
142 my $pid = open(my $log, '-|', @cmd) or
143 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
144 my %deleted; # only an optimization at this point
148 while (my $line = <$log>) {
149 if ($line =~ /$addmsg/o) {
151 next if $deleted{$add};
157 } elsif ($line =~ /$delmsg/o) {
159 } elsif ($line =~ /^commit (${hex}{7,40})/o) {
165 while (my $line = <$log>) {
166 if ($line =~ /^commit (${hex}{7,40})/o) {
172 push @commits, undef;
175 close $log; # we may EPIPE here
177 ($commits[0], $commits[-1]);
180 # private functions below
183 my $pi_config = $ctx->{pi_config};
184 my $listname = $ctx->{listname};
185 my $cgi = $ctx->{cgi};
187 if (open my $fh, '<', "$ctx->{git_dir}/description") {
188 chomp($rv{description} = <$fh>);
192 if ($pi_config && defined $listname && length $listname) {
193 foreach my $key (qw(address)) {
194 $rv{$key} = $pi_config->get($listname, $key) || "";
199 my $path_info = $cgi->path_info;
201 if (ref($cgi) eq 'CGI') {
202 $base = $cgi->url(-base);
204 $base = $cgi->base->as_string;
207 $url_base = $path_info;
208 if ($url_base =~ s!/(?:|index\.html)?\z!!) {
209 $rv{atomurl} = "$base$url_base/atom.xml";
211 $url_base =~ s!/atom\.xml\z!!;
212 $rv{atomurl} = $base . $path_info;
213 $url_base = $base . $url_base; # XXX is this needed?
216 $url_base = "http://example.com";
217 $rv{atomurl} = "$url_base/atom.xml";
219 $rv{url} ||= "$url_base/";
220 $rv{midurl} = "$url_base/m/";
221 $rv{fullurl} = "$url_base/f/";
227 my ($mime, $name) = @_;
228 PublicInbox::Hval->new_oneline($mime->header($name))->raw;
233 my @t = eval { strptime($date) };
235 scalar(@t) ? POSIX::strftime(DATEFMT, @t) : 0;
238 # returns 0 (skipped) or 1 (added)
240 my ($feed_opts, $feed, $add, $git) = @_;
242 my $mime = do_cat_mail($git, $add) or return 0;
243 my $midurl = $feed_opts->{midurl} || 'http://example.com/m/';
244 my $fullurl = $feed_opts->{fullurl} || 'http://example.com/f/';
246 my $header_obj = $mime->header_obj;
247 my $mid = $header_obj->header_raw('Message-ID');
248 defined $mid or return 0;
249 $mid = PublicInbox::Hval->new_msgid($mid);
250 my $href = $mid->as_href . '.html';
251 my $content = PublicInbox::View->feed_entry($mime, $fullurl . $href);
252 defined($content) or return 0;
255 my $subject = mime_header($header_obj, 'Subject') or return 0;
257 my $from = mime_header($header_obj, 'From') or return 0;
258 my @from = Email::Address->parse($from);
259 my $name = $from[0]->name;
260 defined $name or $name = "";
261 my $email = $from[0]->address;
262 defined $email or $email = "";
264 my $date = $header_obj->header('Date');
265 $date = PublicInbox::Hval->new_oneline($date);
266 $date = feed_date($date->raw) or return 0;
269 my (@uuid5) = ($add =~ m!\A($h{8})($h{4})($h{4})($h{4})($h{12})!o);
272 author => { name => $name, email => $email },
275 content => { type => 'xhtml', content => $content },
276 link => $midurl . $href,
277 id => 'urn:uuid:' . join('-', @uuid5),
283 my ($self, $level, $html, $state) = @_;
284 my $mime = $self->message;
286 $$html .= PublicInbox::View->index_entry($mime, $level, $state);
288 dump_msg($self->child, $level+1, $html, $state) if $self->child;
289 dump_msg($self->next, $level, $html, $state) if $self->next;
293 my ($git, $path) = @_;
295 my $str = $git->cat_file("HEAD:$path");
296 Email::MIME->new($str);
301 sub mime_load_for_sort {
302 my ($git, $path, $messages) = @_;
303 my $mime = do_cat_mail($git, $path) or return 0;
305 my $t = eval { str2time($mime->header('Date')) };
306 defined($t) or $t = 0;
307 $mime->header_set('X-PI-TS', $t);
308 push @$messages, $mime;