$title

X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FFeed.pm;h=d7df07ee9c4577a1d15fe5410f68e6bdd68ee3e2;hb=be4d92e74dd0b1c8de85bc08c3c522af2d8d8656;hp=b53255975a3f309be44059fcc6fc7ec4436d86c8;hpb=f79d00b03dfa4f3f3c13bee4654d243c1d2fcd97;p=public-inbox.git diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm index b5325597..d7df07ee 100644 --- a/lib/PublicInbox/Feed.pm +++ b/lib/PublicInbox/Feed.pm @@ -1,131 +1,111 @@ -# Copyright (C) 2013, Eric Wong and all contributors -# License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt) +# Copyright (C) 2013-2018 all contributors +# License: AGPL-3.0+ +# +# 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 str2time); -use PublicInbox::Hval; -use PublicInbox::GitCatFile; +use PublicInbox::MIME; use PublicInbox::View; -use constant { - DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # atom standard - MAX_PER_PAGE => 25, # this needs to be tunable - PRE_WRAP => "", -}; +use PublicInbox::WwwAtomStream; +use PublicInbox::SearchMsg; # this loads w/o Search::Xapian # main function sub generate { - my ($class, $args) = @_; - require XML::Atom::SimpleFeed; - require POSIX; - my $max = $args->{max} || MAX_PER_PAGE; - - my $feed_opts = get_feedopts($args); - my $addr = $feed_opts->{address}; - $addr = $addr->[0] if ref($addr); - my $feed = XML::Atom::SimpleFeed->new( - title => $feed_opts->{description} || "unnamed feed", - link => $feed_opts->{url} || "http://example.com/", - link => { - rel => 'self', - href => $feed_opts->{atomurl} || - "http://example.com/atom.xml", - }, - id => 'mailto:' . ($addr || 'public-inbox@example.com'), - updated => POSIX::strftime(DATEFMT, gmtime), - ); - - my $git = PublicInbox::GitCatFile->new($args->{git_dir}); - each_recent_blob($args, sub { - my ($add) = @_; - add_to_feed($feed_opts, $feed, $add, $git); + my ($ctx) = @_; + my $msgs = recent_msgs($ctx); + return _no_thread() unless @$msgs; + + my $ibx = $ctx->{-inbox}; + PublicInbox::WwwAtomStream->response($ctx, 200, sub { + while (my $smsg = shift @$msgs) { + $ibx->smsg_mime($smsg) and return $smsg; + } }); - $git = undef; # destroy pipes - Email::Address->purge_cache; - $feed->as_string; } -sub generate_html_index { - my ($class, $args) = @_; - require PublicInbox::Thread; - - my $max = $args->{max} || MAX_PER_PAGE; - my $feed_opts = get_feedopts($args); - - my $title = $feed_opts->{description} || ''; - $title = PublicInbox::Hval->new_oneline($title)->as_html; - - my @messages; - my $git_dir = $args->{git_dir}; - my $git = PublicInbox::GitCatFile->new($git_dir); - my ($first, $last) = each_recent_blob($args, sub { - mime_load_for_sort($git, $_[0], \@messages); +sub generate_thread_atom { + my ($ctx) = @_; + my $mid = $ctx->{mid}; + my $msgs = $ctx->{srch}->get_thread($mid); + return _no_thread() unless @$msgs; + + my $ibx = $ctx->{-inbox}; + my $html_url = $ibx->base_url($ctx->{env}); + $html_url .= PublicInbox::Hval->new_msgid($mid)->{href}; + $ctx->{-html_url} = $html_url; + PublicInbox::WwwAtomStream->response($ctx, 200, sub { + while (my $smsg = shift @$msgs) { + $ibx->smsg_mime($smsg) and return $smsg; + } }); - $git = undef; # destroy pipes. - - my $th = PublicInbox::Thread->new(@messages); - $th->thread; - my $html = "$title" . - '{atomurl} . "\"\ntype=\"application/atom+xml\"/>" . - '' . PRE_WRAP; - - # sort child messages in chronological order - $th->order(*PublicInbox::Thread::sort_ts); +} - # except we sort top-level messages reverse chronologically - my $state = [ time, {}, $first, 0 ]; - for (PublicInbox::Thread::rsort_ts($th->rootset)) { - dump_msg($_, 0, \$html, $state) +sub generate_html_index { + my ($ctx) = @_; + # 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} && $ctx->{srch}) { + return PublicInbox::View::index_topics($ctx); } - Email::Address->purge_cache; - my $footer = nav_footer($args->{cgi}, $last, $feed_opts, $state); - if ($footer) { - my $list_footer = $args->{footer}; - $footer .= "\n" . $list_footer if $list_footer; - $footer = "

" . PRE_WRAP . "$footer"; + my $env = $ctx->{env}; + my $url = $ctx->{-inbox}->base_url($env) . 'new.html'; + my $qs = $env->{QUERY_STRING}; + $url .= "?$qs" if $qs ne ''; + [302, [ 'Location', $url, 'Content-Type', 'text/plain'], + [ "Redirecting to $url\n" ] ]; +} + +sub new_html { + my ($ctx) = @_; + my $msgs = recent_msgs($ctx); + if (!@$msgs) { + return [404, ['Content-Type', 'text/plain'], + ["No messages, yet\n"] ]; } - $html . "$footer"; + $ctx->{-html_tip} = '

';
+	$ctx->{-upfx} = '';
+	$ctx->{-hr} = 1;
+	my $ibx = $ctx->{-inbox};
+	PublicInbox::WwwStream->response($ctx, 200, sub {
+		while (my $smsg = shift @$msgs) {
+			my $m = $ibx->smsg_mime($smsg) or next;
+			my $more = scalar @$msgs;
+			return PublicInbox::View::index_entry($m, $ctx, $more);
+		}
+		PublicInbox::View::pagination_footer($ctx, './new.html');
+	});
 }
 
 # private subs
 
-sub nav_footer {
-	my ($cgi, $last, $feed_opts, $state) = @_;
-	$cgi or return '';
-	my $old_r = $cgi->param('r');
-	my $head = '    ';
-	my $next = '    ';
-	my $first = $state->[2];
-	my $anchor = $state->[3];
+sub _no_thread () {
+	[404, ['Content-Type', 'text/plain'], ["No feed found for thread\n"]];
+}
 
-	if ($last) {
-		$next = qq!next!;
+sub recent_msgs {
+	my ($ctx) = @_;
+	my $ibx = $ctx->{-inbox};
+	my $max = $ibx->{feedmax};
+	my $qp = $ctx->{qp};
+	my $v = $ibx->{version} || 1;
+	if ($v > 2) {
+		die "BUG: unsupported inbox version: $v\n";
 	}
-	if ($old_r) {
-		$head = $cgi->path_info;
-		$head = qq!head!;
+	if (my $srch = $ibx->search) {
+		return PublicInbox::View::paginate_recent($ctx, $max);
 	}
-	my $atom = "{atomurl}\">atom";
-	my $permalink = "permalink";
-	"page: $next $head $atom $permalink";
-}
 
-sub each_recent_blob {
-	my ($args, $cb) = @_;
-	my $max = $args->{max} || MAX_PER_PAGE;
 	my $hex = '[a-f0-9]';
-	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 $refhex = qr/${hex}{4,40}(?:~\d+)?/;
-	my $cgi = $args->{cgi};
+	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+)?/;
 
 	# 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 = $r;
 	}
@@ -133,177 +113,41 @@ sub each_recent_blob {
 	# get recent messages
 	# we could use git log -z, but, we already know ssoma will not
 	# leave us with filenames with spaces in them..
-	my @cmd = ('git', "--git-dir=$args->{git_dir}",
-			qw/log --no-notes --no-color --raw -r
-			   --abbrev=16 --abbrev-commit/);
-	push @cmd, $range;
-
-	my $pid = open(my $log, '-|', @cmd) or
-		die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
+	my $log = $ibx->git->popen(qw/log
+				--no-notes --no-color --raw -r
+				--no-abbrev --abbrev-commit/,
+				"--format=%h", $range);
 	my %deleted; # only an optimization at this point
 	my $last;
-	my $nr = 0;
-	my @commits = ();
-	while (my $line = <$log>) {
+	my $last_commit;
+	local $/ = "\n";
+	my @oids;
+	while (defined(my $line = <$log>)) {
 		if ($line =~ /$addmsg/o) {
 			my $add = $1;
-			next if $deleted{$add};
-			$nr += $cb->($add);
-			if ($nr >= $max) {
+			next if $deleted{$add}; # optimization-only
+			push @oids, $add;
+			if (scalar(@oids) >= $max) {
 				$last = 1;
 				last;
 			}
 		} elsif ($line =~ /$delmsg/o) {
 			$deleted{$1} = 1;
-		} elsif ($line =~ /^commit (${hex}{7,40})/o) {
-			push @commits, $1;
 		}
 	}
 
 	if ($last) {
+		local $/ = "\n";
 		while (my $line = <$log>) {
-			if ($line =~ /^commit (${hex}{7,40})/o) {
-				push @commits, $1;
+			if ($line =~ /^(${hex}{7,40})/) {
+				$last_commit = $1;
 				last;
 			}
 		}
-	} else {
-		push @commits, undef;
 	}
 
-	close $log; # we may EPIPE here
-	# for pagination
-	($commits[0], $commits[-1]);
-}
-
-# private functions below
-sub get_feedopts {
-	my ($args) = @_;
-	my $pi_config = $args->{pi_config};
-	my $listname = $args->{listname};
-	my $cgi = $args->{cgi};
-	my %rv;
-	if (open my $fh, '<', "$args->{git_dir}/description") {
-		chomp($rv{description} = <$fh>);
-		close $fh;
-	}
-
-	if ($pi_config && defined $listname && length $listname) {
-		foreach my $key (qw(address)) {
-			$rv{$key} = $pi_config->get($listname, $key) || "";
-		}
-	}
-	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?
-		}
-	} else {
-		$url_base = "http://example.com";
-		$rv{atomurl} = "$url_base/atom.xml";
-	}
-	$rv{url} ||= "$url_base/";
-	$rv{midurl} = "$url_base/m/";
-	$rv{fullurl} = "$url_base/f/";
-
-	\%rv;
-}
-
-sub mime_header {
-	my ($mime, $name) = @_;
-	PublicInbox::Hval->new_oneline($mime->header($name))->raw;
-}
-
-sub feed_date {
-	my ($date) = @_;
-	my @t = eval { strptime($date) };
-
-	scalar(@t) ? POSIX::strftime(DATEFMT, @t) : 0;
-}
-
-# returns 0 (skipped) or 1 (added)
-sub add_to_feed {
-	my ($feed_opts, $feed, $add, $git) = @_;
-
-	my $mime = do_cat_mail($git, $add) or return 0;
-	my $midurl = $feed_opts->{midurl} || 'http://example.com/m/';
-	my $fullurl = $feed_opts->{fullurl} || 'http://example.com/f/';
-
-	my $mid = $mime->header_obj->header_raw('Message-ID');
-	defined $mid or return 0;
-	$mid = PublicInbox::Hval->new_msgid($mid);
-	my $href = $mid->as_href . '.html';
-	my $content = PublicInbox::View->feed_entry($mime, $fullurl . $href);
-	defined($content) or return 0;
-
-	my $subject = mime_header($mime, 'Subject') or return 0;
-
-	my $from = mime_header($mime, 'From') or return 0;
-	my @from = Email::Address->parse($from);
-	my $name = $from[0]->name;
-	defined $name or $name = "";
-	my $email = $from[0]->address;
-	defined $email or $email = "";
-
-	my $date = $mime->header('Date');
-	$date = PublicInbox::Hval->new_oneline($date);
-	$date = feed_date($date->raw) or return 0;
-	$add =~ tr!/!!d;
-	my $h = '[a-f0-9]';
-	my (@uuid5) = ($add =~ m!\A($h{8})($h{4})($h{4})($h{4})($h{12})!o);
-
-	$feed->add_entry(
-		author => { name => $name, email => $email },
-		title => $subject,
-		updated => $date,
-		content => { type => 'xhtml', content => $content },
-		link => $midurl . $href,
-		id => 'urn:uuid:' . join('-', @uuid5),
-	);
-	1;
-}
-
-sub dump_msg {
-	my ($self, $level, $html, $state) = @_;
-	my $mime = $self->message;
-	if ($mime) {
-		$$html .= PublicInbox::View->index_entry($mime, $level, $state);
-	}
-	dump_msg($self->child, $level+1, $html, $state) if $self->child;
-	dump_msg($self->next, $level, $html, $state) if $self->next;
-}
-
-sub do_cat_mail {
-	my ($git, $path) = @_;
-	my $mime = eval {
-		my $str = $git->cat_file("HEAD:$path");
-		Email::MIME->new($str);
-	};
-	$@ ? undef : $mime;
-}
-
-sub mime_load_for_sort {
-	my ($git, $path, $messages) = @_;
-	my $mime = do_cat_mail($git, $path) or return 0;
-
-	my $t = eval { str2time($mime->header('Date')) };
-	defined($t) or $t = 0;
-	$mime->header_set('X-PI-TS', $t);
-	push @$messages, $mime;
-	1;
+	$ctx->{next_page} = "r=$last_commit" if $last_commit;
+	[ map { bless {blob => $_ }, 'PublicInbox::SearchMsg' } @oids ];
 }
 
 1;