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);
15 eval { require Git }; # this is GPLv2+, so we are OK to use it
17 DATEFMT => '%Y-%m-%dT%H:%M:%SZ',
20 use PublicInbox::View;
22 my $enc_utf8 = find_encoding('utf8');
23 my $enc_ascii = find_encoding('us-ascii');
24 my $enc_mime = find_encoding('MIME-Header');
26 # FIXME: workaround https://rt.cpan.org/Public/Bug/Display.html?id=22817
30 my ($class, $args) = @_;
31 my $max = $args->{max} || MAX_PER_PAGE;
32 my $top = $args->{top}; # bool
34 local $ENV{GIT_DIR} = $args->{git_dir};
35 my $feed_opts = get_feedopts($args);
37 my $feed = XML::Atom::SimpleFeed->new(
38 title => $feed_opts->{description} || "unnamed feed",
39 link => $feed_opts->{url} || "http://example.com/",
42 href => $feed_opts->{atomurl} ||
43 "http://example.com/atom",
45 id => $feed_opts->{address} || 'public-inbox@example.com',
46 updated => strftime(DATEFMT, gmtime),
49 my $git = try_git_pm($args->{git_dir});
50 each_recent_blob($args, sub {
52 add_to_feed($feed_opts, $feed, $add, $top, $git);
57 sub generate_html_index {
58 my ($class, $args) = @_;
59 my $max = $args->{max} || MAX_PER_PAGE;
60 my $top = $args->{top}; # bool
61 local $ENV{GIT_DIR} = $args->{git_dir};
62 my $feed_opts = get_feedopts($args);
63 my $title = xs_html($feed_opts->{description} || "");
65 my $git = try_git_pm($args->{git_dir});
66 my ($first, $last) = each_recent_blob($args, sub {
67 my $simple = do_cat_mail($git, 'Email::Simple', $_[0])
69 if ($top && ($simple->header("In-Reply-To") ||
70 $simple->header("References"))) {
73 $simple->body_set(""); # save some memory
75 my $t = eval { str2time($simple->header('Date')) };
76 defined($t) or $t = 0;
77 $simple->header_set('X-PI-Date', $t);
78 push @messages, $simple;
82 my $th = Mail::Thread->new(@messages);
85 "<html><head><title>$title</title>" .
86 '<link rel=alternate title=Atom.feed href="' .
87 $feed_opts->{atomurl} . '" type="application/atom+xml"/>' .
88 '</head><body><pre>');
89 push @out, $feed_opts->{midurl};
91 # sort by date, most recent at top
94 $b->topmost->message->header('X-PI-Date') <=>
95 $a->topmost->message->header('X-PI-Date')
98 dump_html_line($_, 0, \@out) for $th->rootset;
100 my $footer = nav_footer($args->{cgi}, $first, $last);
101 $footer = "<hr /><pre>$footer</pre>" if $footer;
102 $out[0] . "</pre>$footer</html>";
108 my ($cgi, $first, $last) = @_;
110 my $old_r = $cgi->param('r');
113 my %opts = (-path => 1, -query => 1, -relative => 1);
116 $cgi->param('r', $last);
117 $next = $cgi->url(%opts);
118 $next = qq!<a href="$next">next</a>!;
120 if ($first && $old_r) {
121 $cgi->param('r', "$first..");
122 $prev = $cgi->url(%opts);
123 $prev = qq!<a href="$prev">prev</a>!;
128 sub each_recent_blob {
129 my ($args, $cb) = @_;
130 my $max = $args->{max} || MAX_PER_PAGE;
131 my $refhex = qr/[a-f0-9]{4,40}(?:~\d+)?/;
132 my $cgi = $args->{cgi};
134 # revision ranges may be specified
137 my $r = $cgi->param('r') if $cgi;
139 if ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o) {
141 } elsif ($r =~ /\A(?:$refhex\.\.)\z/o) {
147 # get recent messages
148 # we could use git log -z, but, we already know ssoma will not
149 # leave us with filenames with spaces in them..
150 my @cmd = qw/git log --no-notes --no-color --raw -r --no-abbrev/;
151 push @cmd, '--reverse' if $reverse;
155 my $pid = open(my $log, '-|', @cmd) or
156 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
161 while (my $line = <$log>) {
162 if ($line =~ /^:000000 100644 0{40} ([a-f0-9]{40})/) {
164 next if $deleted{$add};
170 } elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) {
172 } elsif ($line =~ /^commit ([a-f0-9]{40})/) {
178 while (my $line = <$log>) {
179 if ($line =~ /^commit ([a-f0-9]{40})/) {
186 close $log; # we may EPIPE here
188 $reverse ? ($commits[-1],$commits[0]) : ($commits[0],$commits[-1]);
191 # private functions below
194 my $pi_config = $args->{pi_config};
195 my $listname = $args->{listname};
196 my $cgi = $args->{cgi};
199 if ($pi_config && defined $listname && length $listname) {
200 foreach my $key (qw(description address)) {
201 $rv{$key} = $pi_config->get($listname, $key) || "";
206 my $cgi_url = $cgi->url(-path=>1, -relative=>1);
207 my $base = $cgi->url(-base);
208 $url_base = $cgi_url;
209 if ($url_base =~ s!/(?:|index\.html)?\z!!) {
210 $rv{atomurl} = "$base$url_base/index.atom.xml";
212 $url_base =~ s!/?(?:index|all)\.atom\.xml\z!!;
213 $rv{atomurl} = $base . $cgi_url;
214 $url_base = $base . $url_base; # XXX is this needed?
217 $url_base = "http://example.com";
218 $rv{atomurl} = "$url_base/index.atom.xml";
220 $rv{url} ||= "$url_base/";
221 $rv{midurl} = "$url_base/m/";
222 $rv{fullurl} = "$url_base/f/";
228 my ($simple, $name) = @_;
229 my $val = $simple->header($name);
230 return "" unless defined $val;
231 $val =~ tr/\t\n / /s;
233 $enc_utf8->encode($enc_mime->decode($val));
238 my @t = eval { strptime($date) };
240 scalar(@t) ? strftime(DATEFMT, @t) : 0;
243 # returns 0 (skipped) or 1 (added)
245 my ($feed_opts, $feed, $add, $top, $git) = @_;
247 my $mime = do_cat_mail($git, 'Email::MIME', $add) or return 0;
248 if ($top && $mime->header("In-Reply-To")) {
252 my $midurl = $feed_opts->{midurl} || 'http://example.com/m/';
253 my $fullurl = $feed_opts->{fullurl} || 'http://example.com/f/';
255 my $content = PublicInbox::View->as_feed_entry($mime, $fullurl);
256 defined($content) or return 0;
258 my $mid = utf8_header($mime, "Message-ID") or return 0;
259 $mid =~ s/\A<//; $mid =~ s/>\z//;
261 my $subject = utf8_header($mime, "Subject") || "";
262 length($subject) or return 0;
264 my $from = utf8_header($mime, "From") or return 0;
266 my @from = Email::Address->parse($from);
267 my $name = $from[0]->name;
268 defined $name or $name = "";
269 my $email = $from[0]->address;
270 defined $email or $email = "";
272 my $url = $midurl . uri_escape($mid);
273 my $date = utf8_header($mime, "Date");
275 $date = feed_date($date) or return 0;
277 author => { name => $name, email => $email },
280 content => { type => "html", content => $content },
288 my ($self, $level, $args) = @_; # args => [ $html, $midurl ]
289 $args->[0] .= (' ' x $level);
290 if ($self->message) {
291 my $simple = $self->message;
292 my $subj = utf8_header($simple, "Subject");
293 my $mid = utf8_header($simple, "Message-ID");
296 my $url = $args->[1] . uri_escape($mid);
297 my $from = utf8_header($simple, "From");
298 my @from = Email::Address->parse($from);
299 $from = $from[0]->name;
300 (defined($from) && length($from)) or $from = $from[0]->address;
301 $from = xs_html($from);
302 $subj = xs_html($subj);
303 $args->[0] .= "<a href=\"$url.html\">$subj</a> $from\n";
305 $args->[0] .= "[ Message not available ]\n";
307 dump_html_line($self->child, $level+1, $args) if $self->child;
308 dump_html_line($self->next, $level, $args) if $self->next;
312 $enc_ascii->encode(escapeHTML($enc_utf8->decode($_[0])),
318 eval { Git->repository(Directory => $dir) };
322 my ($git, $class, $sha1) = @_;
325 open my $fh, '>', \$str or
326 die "failed to setup string handle: $!\n";
328 my $bytes = $git->cat_blob($sha1, $fh);
329 return if $bytes <= 0;
330 close $fh or die "failed to close string handle: $!\n";
332 $str = `git cat-file blob $sha1`;
333 return if $? != 0 || length($str) == 0;