X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FFeed.pm;h=2f141c4413b9473e0f4ee3cdaaaa692633a95428;hb=e158d56a54d3c6d4890aa6ac4caa28a834279af0;hp=f4d19e0208d1355101e8e03cd521b99b6c568b8f;hpb=e9fda329d958fca206543b92bc8ebea8e91eb43a;p=public-inbox.git diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm index f4d19e02..2f141c44 100644 --- a/lib/PublicInbox/Feed.pm +++ b/lib/PublicInbox/Feed.pm @@ -1,138 +1,231 @@ -# Copyright (C) 2013-2015, all contributors +# Copyright (C) 2013-2015 all contributors # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt) +# +# Used for generating Atom feeds for web-accessible mailing list archives. package PublicInbox::Feed; use strict; use warnings; -use Email::Address; use Email::MIME; use Date::Parse qw(strptime); -use PublicInbox::Hval; -use PublicInbox::GitCatFile; +use PublicInbox::Hval qw/ascii_html/; +use PublicInbox::Git; use PublicInbox::View; -use PublicInbox::MID qw/mid_clean mid_compressed/; +use PublicInbox::MID qw/mid_clean mid2path/; +use PublicInbox::Address; +use POSIX qw/strftime/; use constant { - DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # atom standard + DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # Atom standard MAX_PER_PAGE => 25, # this needs to be tunable }; -use Encode qw/find_encoding/; -my $enc_utf8 = find_encoding('UTF-8'); - # main function sub generate { my ($ctx) = @_; sub { emit_atom($_[0], $ctx) }; } +sub generate_thread_atom { + my ($ctx) = @_; + sub { emit_atom_thread($_[0], $ctx) }; +} + sub generate_html_index { my ($ctx) = @_; sub { emit_html_index($_[0], $ctx) }; } +sub new_html { + my ($ctx) = @_; + my @paths; + my (undef, $last) = each_recent_blob($ctx, sub { + my ($path, $commit, $ts, $u, $subj) = @_; + $ctx->{first} ||= $commit; + push @paths, $path; + }); + if (!@paths) { + return [404, ['Content-Type', 'text/plain'], + ["No messages, yet\n"] ]; + } + $ctx->{-html_tip} = '
';
+	$ctx->{-upfx} = '';
+	PublicInbox::WwwStream->response($ctx, 200, sub {
+		while (my $path = shift @paths) {
+			my $m = do_cat_mail($ctx->{-inbox}, $path) or next;
+			my $more = scalar @paths;
+			my $s = PublicInbox::View::index_entry($m, $ctx, $more);
+			$s .= '
' unless $more; + return $s; + } + undef; + }); +} + # private subs +sub title_tag { + my ($title) = @_; + $title =~ tr/\t\n / /s; # squeeze spaces + # try to avoid the type attribute in title: + $title = ascii_html($title); + my $type = index($title, '&') >= 0 ? "\ntype=\"html\"" : ''; + "$title"; +} + +sub atom_header { + my ($feed_opts, $title) = @_; + + $title = title_tag($feed_opts->{description}) unless (defined $title); + + qq(\n) . + qq{} . + qq{$title} . + qq() . + qq({atomurl}"/>) . + qq(mailto:$feed_opts->{id_addr}); +} + sub emit_atom { my ($cb, $ctx) = @_; - require POSIX; - my $fh = $cb->([ 200, ['Content-Type' => 'application/xml']]); - my $max = $ctx->{max} || MAX_PER_PAGE; my $feed_opts = get_feedopts($ctx); - my $addr = $feed_opts->{address}; - $addr = $addr->[0] if ref($addr); - $addr ||= 'public-inbox@example.com'; - my $title = $feed_opts->{description} || "unnamed feed"; - $title = PublicInbox::Hval->new_oneline($title)->as_html; - my $type = index($title, '&') >= 0 ? "\ntype=\"html\"" : ''; - my $url = $feed_opts->{url} || "http://example.com/"; - my $atomurl = $feed_opts->{atomurl}; - $fh->write(qq(\n) . - qq{} . - qq{$title} . - qq{} . - qq{} . - qq{mailto:$addr} . - '' . POSIX::strftime(DATEFMT, gmtime) . ''); - - my $git = PublicInbox::GitCatFile->new($ctx->{git_dir}); + my $fh = $cb->([ 200, ['Content-Type' => 'application/atom+xml']]); + my $max = $ctx->{max} || MAX_PER_PAGE; + my $x = atom_header($feed_opts); + my $ibx = $ctx->{-inbox}; each_recent_blob($ctx, sub { - my ($add, undef) = @_; - add_to_feed($feed_opts, $fh, $add, $git); + my ($path, undef, $ts) = @_; + if (defined $x) { + $fh->write($x . feed_updated(undef, $ts)); + $x = undef; + } + my $s = feed_entry($feed_opts, $path, $ibx) or return 0; + $fh->write($s); + 1; }); - $git = undef; # destroy pipes - Email::Address->purge_cache; - $fh->write(""); - $fh->close; + end_feed($fh); } +sub _no_thread { + my ($cb) = @_; + $cb->([404, ['Content-Type', 'text/plain'], + ["No feed found for thread\n"]]); +} -sub emit_html_index { - my ($cb, $ctx) = @_; - my $fh = $cb->([200,['Content-Type'=>'text/html; charset=UTF-8']]); +sub end_feed { + my ($fh) = @_; + $fh->write(''); + $fh->close; +} - my $max = $ctx->{max} || MAX_PER_PAGE; +sub emit_atom_thread { + my ($cb, $ctx) = @_; + my $res = $ctx->{srch}->get_thread($ctx->{mid}); + return _no_thread($cb) unless $res->{total}; my $feed_opts = get_feedopts($ctx); + my $fh = $cb->([200, ['Content-Type' => 'application/atom+xml']]); - my $title = $feed_opts->{description} || ''; - $title = PublicInbox::Hval->new_oneline($title)->as_html; - my $atom_url = $feed_opts->{atomurl}; + my $html_url = $feed_opts->{atomurl} = $ctx->{self_url}; + $html_url =~ s!/t\.atom\z!/!; + $feed_opts->{url} = $html_url; + $feed_opts->{emit_header} = 1; - $fh->write("$title" . - "" . - '' . PublicInbox::View::PRE_WRAP . - "$title (Atom feed)\n"); + my $ibx = $ctx->{-inbox}; + foreach my $msg (@{$res->{msgs}}) { + my $s = feed_entry($feed_opts, mid2path($msg->mid), $ibx); + $fh->write($s) if defined $s; + } + end_feed($fh); +} - my $state; - my $git = PublicInbox::GitCatFile->new($ctx->{git_dir}); - my $topics; - my $srch = $ctx->{srch}; - $srch and $topics = [ [], {} ]; - my (undef, $last) = each_recent_blob($ctx, sub { - my ($path, $commit, $ts, $u, $subj) = @_; - $state ||= [ undef, {}, $commit, 0 ]; - - if ($srch) { - add_topic($git, $srch, $topics, $path, $ts, $u, $subj); - } else { - my $mime = do_cat_mail($git, $path) or return 0; - PublicInbox::View::index_entry($fh, $mime, 0, $state); - 1; - } - }); - Email::Address->purge_cache; - $git = undef; # destroy pipes. +sub _html_index_top { + my ($feed_opts, $srch) = @_; - my $footer = nav_footer($ctx->{cgi}, $last, $feed_opts, $state); + my $title = ascii_html($feed_opts->{description} || ''); + my $top = "$title (Atom feed)"; + if ($srch) { + $top = qq{
$top} .
+			  qq{ } .
+			  qq{} .
+			  q{
}
+	} else {
+		$top = '
' . $top . "\n";
+	}
+
+	"$title" .
+		"" .
+		PublicInbox::Hval::STYLE .
+		"$top";
+}
+
+sub emit_html_index {
+	my ($res, $ctx) = @_;
+	my $feed_opts = get_feedopts($ctx);
+	my $fh = $res->([200,['Content-Type'=>'text/html; charset=UTF-8']]);
+
+	my $max = $ctx->{max} || MAX_PER_PAGE;
+	$ctx->{-upfx} = '';
+
+	my ($footer, $param, $last);
+	$ctx->{seen} = {};
+	$ctx->{anchor_idx} = 0;
+	$ctx->{fh} = $fh;
+	my $srch = $ctx->{srch};
+	$fh->write(_html_index_top($feed_opts, $srch));
+
+	# if the 'r' query parameter is given, it is a legacy permalink
+	# which we must continue supporting:
+	my $qp = $ctx->{qp};
+	if ($qp && !$qp->{r} && $srch) {
+		$last = PublicInbox::View::emit_index_topics($ctx);
+		$param = 'o';
+	} else {
+		$last = emit_index_nosrch($ctx);
+		$param = 'r';
+	}
+	$footer = nav_footer($ctx, $last, $feed_opts, $param);
 	if ($footer) {
 		my $list_footer = $ctx->{footer};
-		$footer .= "\n" . $list_footer if $list_footer;
+		$footer .= "\n\n" . $list_footer if $list_footer;
 		$footer = "
$footer
"; } - $fh->write(dump_topics($topics)) if $topics; $fh->write("$footer"); $fh->close; } +sub emit_index_nosrch { + my ($ctx) = @_; + my $ibx = $ctx->{-inbox}; + my $fh = $ctx->{fh}; + my (undef, $last) = each_recent_blob($ctx, sub { + my ($path, $commit, $ts, $u, $subj) = @_; + $ctx->{first} ||= $commit; + + my $mime = do_cat_mail($ibx, $path) or return 0; + $fh->write(PublicInbox::View::index_entry($mime, $ctx, 1)); + 1; + }); + $last; +} + sub nav_footer { - my ($cgi, $last, $feed_opts, $state) = @_; - $cgi or return ''; - my $old_r = $cgi->param('r'); + my ($ctx, $last, $feed_opts, $param) = @_; + my $qp = $ctx->{qp} or return ''; + my $old_r = $qp->{$param}; my $head = ' '; my $next = ' '; - # $state = [ undef, {}, $first_commit, $last_anchor ]; - my $first = $state->[2]; - my $anchor = $state->[3]; + my $first = $ctx->{first}; + my $anchor = $ctx->{anchor_idx}; if ($last) { - $next = qq!next!; + $next = qq!next!; } if ($old_r) { - $head = $cgi->path_info; + $head = $ctx->{env}->{PATH_INFO}; $head = qq!head!; } - my $atom = "{atomurl}\">atom"; - my $permalink = "permalink"; - "page: $next $head $atom $permalink"; + my $atom = "{atomurl}\">Atom feed"; + "page: $next $head $atom"; } sub each_recent_blob { @@ -142,11 +235,11 @@ sub each_recent_blob { my $addmsg = qr!^:000000 100644 \S+ \S+ A\t(${hex}{2}/${hex}{38})$!; my $delmsg = qr!^:100644 000000 \S+ \S+ D\t(${hex}{2}/${hex}{38})$!; my $refhex = qr/(?:HEAD|${hex}{4,40})(?:~\d+)?/; - my $cgi = $ctx->{cgi}; + my $qp = $ctx->{qp}; # revision ranges may be specified my $range = 'HEAD'; - my $r = $cgi->param('r') if $cgi; + my $r = $qp->{r} if $qp; if ($r && ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o)) { $range = $r; } @@ -154,19 +247,17 @@ sub each_recent_blob { # get recent messages # we could use git log -z, but, we already know ssoma will not # leave us with filenames with spaces in them.. - my @cmd = ('git', "--git-dir=$ctx->{git_dir}", - qw/log --no-notes --no-color --raw -r - --abbrev=16 --abbrev-commit/, - "--format=%h%x00%ct%x00%an%x00%s%x00"); - push @cmd, $range; - - my $pid = open(my $log, '-|', @cmd) or - die('open `'.join(' ', @cmd) . " pipe failed: $!\n"); + my $log = $ctx->{-inbox}->git->popen(qw/log + --no-notes --no-color --raw -r + --abbrev=16 --abbrev-commit/, + "--format=%h%x00%ct%x00%an%x00%s%x00", + $range); my %deleted; # only an optimization at this point my $last; my $nr = 0; my ($cur_commit, $first_commit, $last_commit); my ($ts, $subj, $u); + local $/ = "\n"; while (defined(my $line = <$log>)) { if ($line =~ /$addmsg/o) { my $add = $1; @@ -187,6 +278,7 @@ sub each_recent_blob { } if ($last) { + local $/ = "\n"; while (my $line = <$log>) { if ($line =~ /^(${hex}{7,40})/o) { $last_commit = $1; @@ -195,7 +287,6 @@ sub each_recent_blob { } } - close $log; # we may EPIPE here # for pagination ($first_commit, $last_commit); } @@ -204,168 +295,85 @@ sub each_recent_blob { sub get_feedopts { my ($ctx) = @_; my $pi_config = $ctx->{pi_config}; - my $listname = $ctx->{listname}; + my $inbox = $ctx->{inbox}; + my $obj = $ctx->{-inbox}; my $cgi = $ctx->{cgi}; - my %rv; - if (open my $fh, '<', "$ctx->{git_dir}/description") { - chomp($rv{description} = <$fh>); - close $fh; - } + my %rv = ( description => $obj->description ); - if ($pi_config && defined $listname && length $listname) { - foreach my $key (qw(address)) { - $rv{$key} = $pi_config->get($listname, $key) || ""; - } - } + $rv{address} = $obj->{address}; + $rv{id_addr} = $obj->{-primary_address}; my $url_base; - if ($cgi) { - my $path_info = $cgi->path_info; - my $base; - if (ref($cgi) eq 'CGI') { - $base = $cgi->url(-base); - } else { - $base = $cgi->base->as_string; - $base =~ s!/\z!!; - } - $url_base = $path_info; - if ($url_base =~ s!/(?:|index\.html)?\z!!) { - $rv{atomurl} = "$base$url_base/atom.xml"; - } else { - $url_base =~ s!/atom\.xml\z!!; - $rv{atomurl} = $base . $path_info; - $url_base = $base . $url_base; # XXX is this needed? - } + $url_base = $obj->base_url($cgi); # CGI may be undef + if (my $mid = $ctx->{mid}) { # per-thread feed: + $rv{atomurl} = "$url_base$mid/t.atom"; } else { - $url_base = "http://example.com"; - $rv{atomurl} = "$url_base/atom.xml"; + $rv{atomurl} = $url_base."new.atom"; } - $rv{url} ||= "$url_base/"; - $rv{midurl} = "$url_base/m/"; - $rv{fullurl} = "$url_base/f/"; + $rv{url} ||= $url_base; + $rv{midurl} = $url_base; \%rv; } -sub mime_header { - my ($mime, $name) = @_; - PublicInbox::Hval->new_oneline($mime->header($name))->raw; -} - -sub feed_date { - my ($date) = @_; - my @t = eval { strptime($date) }; +sub feed_updated { + my ($date, $ts) = @_; + my @t = eval { strptime($date) } if defined $date; + @t = gmtime($ts || time) unless scalar @t; - scalar(@t) ? POSIX::strftime(DATEFMT, @t) : 0; + '' . strftime(DATEFMT, @t) . ''; } -# returns 0 (skipped) or 1 (added) -sub add_to_feed { - my ($feed_opts, $fh, $add, $git) = @_; +# returns undef or string +sub feed_entry { + my ($feed_opts, $add, $ibx) = @_; - my $mime = do_cat_mail($git, $add) or return 0; - my $fullurl = $feed_opts->{fullurl} || 'http://example.com/f/'; + my $mime = do_cat_mail($ibx, $add) or return; + my $url = $feed_opts->{url}; + my $midurl = $feed_opts->{midurl}; my $header_obj = $mime->header_obj; - my $mid = $header_obj->header('Message-ID'); - defined $mid or return 0; + my $mid = $header_obj->header_raw('Message-ID'); + defined $mid or return; $mid = PublicInbox::Hval->new_msgid($mid); - my $href = $mid->as_href . '.html'; - my $content = PublicInbox::View->feed_entry($mime, $fullurl . $href); - defined($content) or return 0; - $mime = undef; + my $href = $midurl.$mid->as_href; - my $title = mime_header($header_obj, 'Subject') or return 0; - $title = PublicInbox::Hval->new_oneline($title)->as_html; - my $type = index($title, '&') >= 0 ? "\ntype=\"html\"" : ''; + my $date = $header_obj->header('Date'); + my $updated = feed_updated($date); - my $from = mime_header($header_obj, 'From') or return 0; - my @from = Email::Address->parse($from) or return 0; - my $name = PublicInbox::Hval->new_oneline($from[0]->name)->as_html; - my $email = $from[0]->address; - $email = PublicInbox::Hval->new_oneline($email)->as_html; + my $title = $header_obj->header('Subject'); + defined $title or return; + $title = title_tag($title); - my $date = $header_obj->header('Date'); - $date = PublicInbox::Hval->new_oneline($date); - $date = feed_date($date->raw) or return 0; + my $from = $header_obj->header('From') or return; + my ($email) = PublicInbox::Address::emails($from); + my $name = join(', ',PublicInbox::Address::names($from)); + $name = ascii_html($name); + $email = ascii_html($email); - $fh->write("$name$email" . - "$title" . - "$date" . - qq{} . - qq{}); - $fh->write($content); + my $s = ''; + if (delete $feed_opts->{emit_header}) { + $s .= atom_header($feed_opts, $title) . $updated; + } + $s .= "$name$email" . + "$title$updated" . + qq{} . + qq{} . + qq() . + PublicInbox::View::multipart_text_as_html($mime, $href) . + '
'; $add =~ tr!/!!d; my $h = '[a-f0-9]'; my (@uuid5) = ($add =~ m!\A($h{8})($h{4})($h{4})($h{4})($h{12})!o); my $id = 'urn:uuid:' . join('-', @uuid5); - my $midurl = $feed_opts->{midurl} || 'http://example.com/m/'; - $fh->write(qq{}. - "$id"); - 1; + $s .= qq!!. + "$id"; } sub do_cat_mail { - my ($git, $path) = @_; - my $mime = eval { - my $str = $git->cat_file("HEAD:$path"); - Email::MIME->new($str); - }; - $@ ? undef : $mime; -} - -# accumulate recent topics if search is supported -sub add_topic { - my ($git, $srch, $topics, $path, $ts, $u, $subj) = @_; - my ($order, $subjs) = @$topics; - my $header_obj; - - # legacy ssoma did not set commit titles based on Subject - $subj = $enc_utf8->decode($subj); - if ($subj eq 'mda') { - my $mime = do_cat_mail($git, $path) or return 0; - $header_obj = $mime->header_obj; - $subj = mime_header($header_obj, 'Subject'); - } - - $subj = $srch->subject_normalized($subj); - if (++$subjs->{$subj} == 1) { - unless ($header_obj) { - my $mime = do_cat_mail($git, $path) or return 0; - $header_obj = $mime->header_obj; - } - my $mid = $header_obj->header('Message-ID'); - $mid = mid_compressed(mid_clean($mid)); - $u = $enc_utf8->decode($u); - push @$order, [ $mid, $ts, $u, $subj ]; - return 1; - } - 0; # old topic, continue going -} - -sub dump_topics { - my ($topics) = @_; - my ($order, $subjs) = @$topics; - my $dst = ''; - $dst .= "\n[No recent topics]" unless (scalar @$order); - while (defined(my $info = shift @$order)) { - my ($mid, $ts, $u, $subj) = @$info; - my $n = delete $subjs->{$subj}; - $mid = PublicInbox::Hval->new($mid)->as_href; - $subj = PublicInbox::Hval->new($subj)->as_html; - $u = PublicInbox::Hval->new($u)->as_html; - $dst .= "\n$subj\n- "; - $ts = POSIX::strftime('%Y-%m-%d %H:%M', gmtime($ts)); - if ($n == 1) { - $dst .= "created by $u @ $ts UTC\n" - } else { - # $n isn't the total number of posts on the topic, - # just the number of posts in the current "git log" - # window, so leave it unlabeled - $dst .= "updated by $u @ $ts UTC ($n)\n" - } - } - $dst .= '
' + my ($ibx, $path) = @_; + my $mime = eval { $ibx->msg_by_path($path) } or return; + Email::MIME->new($mime); } 1;