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;
6 use XML::Atom::SimpleFeed;
9 use URI::Escape qw/uri_escape/;
10 use Encode qw/find_encoding/;
11 use Encode::MIME::Header;
12 use CGI qw(escapeHTML);
13 use POSIX qw(strftime);
14 use Date::Parse qw(strptime str2time);
16 DATEFMT => '%Y-%m-%dT%H:%M:%SZ',
19 use PublicInbox::View;
21 my $enc_utf8 = find_encoding('utf8');
22 my $enc_ascii = find_encoding('us-ascii');
23 my $enc_mime = find_encoding('MIME-Header');
25 # FIXME: workaround https://rt.cpan.org/Public/Bug/Display.html?id=22817
29 my ($class, $args) = @_;
30 my $max = $args->{max} || MAX_PER_PAGE;
31 my $top = $args->{top}; # bool
33 local $ENV{GIT_DIR} = $args->{git_dir};
34 my $feed_opts = get_feedopts($args);
36 my $feed = XML::Atom::SimpleFeed->new(
37 title => $feed_opts->{description} || "unnamed feed",
38 link => $feed_opts->{url} || "http://example.com/",
41 href => $feed_opts->{atomurl} ||
42 "http://example.com/atom",
44 id => $feed_opts->{address} || 'public-inbox@example.com',
45 updated => strftime(DATEFMT, gmtime),
48 each_recent_blob($args, sub {
50 add_to_feed($feed_opts, $feed, $add, $top);
55 sub generate_html_index {
56 my ($class, $args) = @_;
57 my $max = $args->{max} || MAX_PER_PAGE;
58 my $top = $args->{top}; # bool
59 local $ENV{GIT_DIR} = $args->{git_dir};
60 my $feed_opts = get_feedopts($args);
61 my $title = xs_html($feed_opts->{description} || "");
63 my ($first, $last) = each_recent_blob($args, sub {
64 my $str = `git cat-file blob $_[0]`;
66 my $simple = Email::Simple->new($str);
67 if ($top && ($simple->header("In-Reply-To") ||
68 $simple->header("References"))) {
71 $simple->body_set(""); # save some memory
73 my $t = eval { str2time($simple->header('Date')) };
74 defined($t) or $t = 0;
75 $simple->header_set('X-PI-Date', $t);
76 push @messages, $simple;
80 my $th = Mail::Thread->new(@messages);
83 "<html><head><title>$title</title>" .
84 '<link rel=alternate title=Atom.feed href="' .
85 $feed_opts->{atomurl} . '" type="application/atom+xml"/>' .
86 '</head><body><pre>');
87 push @out, $feed_opts->{midurl};
89 # sort by date, most recent at top
92 $b->topmost->message->header('X-PI-Date') <=>
93 $a->topmost->message->header('X-PI-Date')
96 dump_html_line($_, 0, \@out) for $th->rootset;
98 my $footer = nav_footer($args->{cgi}, $first, $last);
99 $footer = "<hr /><pre>$footer</pre>" if $footer;
100 $out[0] . "</pre>$footer</html>";
106 my ($cgi, $first, $last) = @_;
108 my $old_r = $cgi->param('r');
111 my %opts = (-path => 1, -query => 1, -relative => 1);
114 $cgi->param('r', $last);
115 $next = $cgi->url(%opts);
116 $next = qq!<a href="$next">next</a>!;
118 if ($first && $old_r) {
119 $cgi->param('r', "$first..");
120 $prev = $cgi->url(%opts);
121 $prev = qq!<a href="$prev">prev</a>!;
126 sub each_recent_blob {
127 my ($args, $cb) = @_;
128 my $max = $args->{max} || MAX_PER_PAGE;
129 my $refhex = qr/[a-f0-9]{4,40}(?:~\d+)?/;
130 my $cgi = $args->{cgi};
132 # revision ranges may be specified
135 my $r = $cgi->param('r') if $cgi;
137 if ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o) {
139 } elsif ($r =~ /\A(?:$refhex\.\.)\z/o) {
145 # get recent messages
146 # we could use git log -z, but, we already know ssoma will not
147 # leave us with filenames with spaces in them..
148 my @cmd = qw/git log --no-notes --no-color --raw -r --no-abbrev/;
149 push @cmd, '--reverse' if $reverse;
153 my $pid = open(my $log, '-|', @cmd) or
154 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
159 while (my $line = <$log>) {
160 if ($line =~ /^:000000 100644 0{40} ([a-f0-9]{40})/) {
162 next if $deleted{$add};
168 } elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) {
170 } elsif ($line =~ /^commit ([a-f0-9]{40})/) {
176 while (my $line = <$log>) {
177 if ($line =~ /^commit ([a-f0-9]{40})/) {
184 close $log; # we may EPIPE here
186 $reverse ? ($commits[-1],$commits[0]) : ($commits[0],$commits[-1]);
189 # private functions below
192 my $pi_config = $args->{pi_config};
193 my $listname = $args->{listname};
194 my $cgi = $args->{cgi};
197 if ($pi_config && defined $listname && length $listname) {
198 foreach my $key (qw(description address)) {
199 $rv{$key} = $pi_config->get($listname, $key) || "";
204 my $cgi_url = $cgi->url(-path=>1, -relative=>1);
205 my $base = $cgi->url(-base);
206 $url_base = $cgi_url;
207 if ($url_base =~ s!/(?:|index\.html)?\z!!) {
208 $rv{atomurl} = "$base$url_base/index.atom.xml";
210 $url_base =~ s!/?(?:index|all)\.atom\.xml\z!!;
211 $rv{atomurl} = $base . $cgi_url;
212 $url_base = $base . $url_base; # XXX is this needed?
215 $url_base = "http://example.com";
216 $rv{atomurl} = "$url_base/index.atom.xml";
218 $rv{url} ||= "$url_base/";
219 $rv{midurl} = "$url_base/m/";
220 $rv{fullurl} = "$url_base/f/";
226 my ($simple, $name) = @_;
227 my $val = $simple->header($name);
228 return "" unless defined $val;
229 $val =~ tr/\t\n / /s;
231 $enc_utf8->encode($enc_mime->decode($val));
236 my @t = eval { strptime($date) };
238 scalar(@t) ? strftime(DATEFMT, @t) : 0;
241 # returns 0 (skipped) or 1 (added)
243 my ($feed_opts, $feed, $add, $top) = @_;
245 # we can use git cat-file --batch if performance becomes a
246 # problem, but I doubt it...
247 my $str = `git cat-file blob $add`;
249 my $mime = Email::MIME->new($str);
251 if ($top && $mime->header("In-Reply-To")) {
255 my $midurl = $feed_opts->{midurl} || 'http://example.com/m/';
256 my $fullurl = $feed_opts->{fullurl} || 'http://example.com/f/';
258 my $content = PublicInbox::View->as_feed_entry($mime, $fullurl);
259 defined($content) or return 0;
261 my $mid = utf8_header($mime, "Message-ID") or return 0;
262 $mid =~ s/\A<//; $mid =~ s/>\z//;
264 my $subject = utf8_header($mime, "Subject") || "";
265 length($subject) or return 0;
267 my $from = utf8_header($mime, "From") or return 0;
269 my @from = Email::Address->parse($from);
270 my $name = $from[0]->name;
271 defined $name or $name = "";
272 my $email = $from[0]->address;
273 defined $email or $email = "";
275 my $url = $midurl . uri_escape($mid);
276 my $date = utf8_header($mime, "Date");
278 $date = feed_date($date) or return 0;
280 author => { name => $name, email => $email },
283 content => { type => "html", content => $content },
291 my ($self, $level, $args) = @_; # args => [ $html, $midurl ]
292 $args->[0] .= (' ' x $level);
293 if ($self->message) {
294 my $simple = $self->message;
295 my $subj = utf8_header($simple, "Subject");
296 my $mid = utf8_header($simple, "Message-ID");
299 my $url = $args->[1] . uri_escape($mid);
300 my $from = utf8_header($simple, "From");
301 my @from = Email::Address->parse($from);
302 $from = $from[0]->name;
303 (defined($from) && length($from)) or $from = $from[0]->address;
304 $from = xs_html($from);
305 $subj = xs_html($subj);
306 $args->[0] .= "<a href=\"$url.html\">$subj</a> $from\n";
308 $args->[0] .= "[ Message not available ]\n";
310 dump_html_line($self->child, $level+1, $args) if $self->child;
311 dump_html_line($self->next, $level, $args) if $self->next;
315 $enc_ascii->encode(escapeHTML($enc_utf8->decode($_[0])),