X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FFeed.pm;h=258ef03b7f716e1e2a70cfbdd4d1eef051b6eacc;hb=f76f265a851944b5dedcc3be5f3b5224b6ebda89;hp=da877f3f536d50209ee5d3965ad5cefcc4788826;hpb=cf0a2370a57fe49d0fca149409f98d2907efeb15;p=public-inbox.git

diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm
index da877f3f..258ef03b 100644
--- a/lib/PublicInbox/Feed.pm
+++ b/lib/PublicInbox/Feed.pm
@@ -1,186 +1,385 @@
-# Copyright (C) 2013, Eric Wong <normalperson@yhbt.net> and all contributors
+# Copyright (C) 2013-2015 all contributors <meta@public-inbox.org>
 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
 package PublicInbox::Feed;
 use strict;
 use warnings;
-use XML::Atom::SimpleFeed;
-use Email::MIME;
 use Email::Address;
-use URI::Escape qw/uri_escape/;
-use Encode qw/encode decode/;
-use Encode::MIME::Header;
-use DateTime::Format::Mail;
-use CGI qw(escapeHTML);
-use POSIX qw(strftime);
-use constant DATEFMT => '%Y-%m-%dT%H:%M:%SZ';
-our $dt_parser = DateTime::Format::Mail->new(loose => 1);
+use Email::MIME;
+use Date::Parse qw(strptime);
+use PublicInbox::Hval;
+use PublicInbox::GitCatFile;
+use PublicInbox::View;
+use PublicInbox::MID qw/mid_clean mid2path/;
+use POSIX qw/strftime/;
+use constant {
+	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 ($class, $git_dir, $max) = @_;
-	$max ||= 25;
-
-	local $ENV{GIT_DIR} = $git_dir;
-	my $feed_opts = get_feedopts();
-
-	my $feed = XML::Atom::SimpleFeed->new(
-		title => $feed_opts->{title},
-		link => $feed_opts->{url} || "http://example.com/",
-		link => {
-			rel => 'self',
-			href => $feed_opts->{atomUrl} ||
-				"http://example.com/atom",
-		},
-		id => $feed_opts->{email} || 'public-inbox@example.com',
-		updated => strftime(DATEFMT, gmtime),
-	);
-
-	my @entries;
+	my ($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], $ctx) };
+}
+
+# private subs
+
+sub title_tag {
+	my ($title) = @_;
+	# try to avoid the type attribute in title:
+	$title = PublicInbox::Hval->new_oneline($title)->as_html;
+	my $type = index($title, '&') >= 0 ? "\ntype=\"html\"" : '';
+	"<title$type>$title</title>";
+}
+
+sub atom_header {
+	my ($feed_opts, $title) = @_;
+
+	$title = title_tag($feed_opts->{description}) unless (defined $title);
+
+	qq(<?xml version="1.0" encoding="us-ascii"?>\n) .
+	qq{<feed\nxmlns="http://www.w3.org/2005/Atom">} .
+	qq{$title} .
+	qq(<link\nhref="$feed_opts->{url}"/>) .
+	qq(<link\nrel="self"\nhref="$feed_opts->{atomurl}"/>) .
+	qq(<id>mailto:$feed_opts->{id_addr}</id>);
+}
+
+sub emit_atom {
+	my ($cb, $ctx) = @_;
+	my $fh = $cb->([ 200, ['Content-Type' => 'application/atom+xml']]);
+	my $max = $ctx->{max} || MAX_PER_PAGE;
+	my $feed_opts = get_feedopts($ctx);
+	my $x = atom_header($feed_opts);
+	my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
+	each_recent_blob($ctx, sub {
+		my ($path, undef, $ts) = @_;
+		if (defined $x) {
+			$fh->write($x . '<updated>' .
+				   strftime(DATEFMT, gmtime($ts)) .
+				   '</updated>');
+			$x = undef;
+		}
+		add_to_feed($feed_opts, $fh, $path, $git);
+	});
+	$git = undef; # destroy pipes
+	_end_feed($fh);
+}
+
+sub _no_thread {
+	my ($cb) = @_;
+	my $fh = $cb->([404, ['Content-Type' => 'text/plain']]);
+	$fh->write("No feed found for thread\n");
+	$fh->close;
+}
+
+sub _end_feed {
+	my ($fh) = @_;
+	Email::Address->purge_cache;
+	$fh->write('</feed>');
+	$fh->close;
+}
+
+sub emit_atom_thread {
+	my ($cb, $ctx) = @_;
+	my $res = $ctx->{srch}->get_thread($ctx->{mid});
+	return _no_thread($cb) unless $res->{total};
+	my $fh = $cb->([200, ['Content-Type' => 'application/atom+xml']]);
+	my $feed_opts = get_feedopts($ctx);
+
+	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;
+
+	my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
+	foreach my $msg (@{$res->{msgs}}) {
+		add_to_feed($feed_opts, $fh, mid2path($msg->mid), $git);
+	}
+	$git = undef; # destroy pipes
+	_end_feed($fh);
+}
+
+sub emit_html_index {
+	my ($cb, $ctx) = @_;
+	my $fh = $cb->([200,['Content-Type'=>'text/html; charset=UTF-8']]);
+
+	my $max = $ctx->{max} || MAX_PER_PAGE;
+	my $feed_opts = get_feedopts($ctx);
+
+	my $title = $feed_opts->{description} || '';
+	$title = PublicInbox::Hval->new_oneline($title)->as_html;
+	my $atom_url = $feed_opts->{atomurl};
+	my ($footer, $param, $last);
+	my $state = { ctx => $ctx, seen => {}, anchor_idx => 0 };
+	my $srch = $ctx->{srch};
+
+	my $top = "<b>$title</b> (<a\nhref=\"$atom_url\">Atom feed</a>)";
+
+	if ($srch) {
+		$top = qq{<form\naction=""><tt>$top} .
+			  qq{ <input\nname=q\ntype=text />} .
+			  qq{<input\ntype=submit\nvalue=search />} .
+			  qq{</tt></form>} .
+			  PublicInbox::View::PRE_WRAP;
+	} else {
+		$top = PublicInbox::View::PRE_WRAP . $top . "\n";
+	}
+
+	$fh->write("<html><head><title>$title</title>" .
+		   "<link\nrel=alternate\ntitle=\"Atom feed\"\n".
+		   "href=\"$atom_url\"\ntype=\"application/atom+xml\"/>" .
+		   "</head><body>$top");
+
+	# if the 'r' query parameter is given, it is a legacy permalink
+	# which we must continue supporting:
+	my $cgi = $ctx->{cgi};
+	if ($cgi && !$cgi->param('r') && $srch) {
+		$state->{srch} = $srch;
+		$last = PublicInbox::View::emit_index_topics($state, $fh);
+		$param = 'o';
+	} else {
+		$last = emit_index_nosrch($ctx, $state, $fh);
+		$param = 'r';
+	}
+	$footer = nav_footer($cgi, $last, $feed_opts, $state, $param);
+	if ($footer) {
+		my $list_footer = $ctx->{footer};
+		$footer .= "\n\n" . $list_footer if $list_footer;
+		$footer = "<hr /><pre>$footer</pre>";
+	}
+	$fh->write("$footer</body></html>");
+	$fh->close;
+}
+
+sub emit_index_nosrch {
+	my ($ctx, $state, $fh) = @_;
+	my $git = PublicInbox::GitCatFile->new($ctx->{git_dir});
+	my (undef, $last) = each_recent_blob($ctx, sub {
+		my ($path, $commit, $ts, $u, $subj) = @_;
+		$state->{first} ||= $commit;
+
+		my $mime = do_cat_mail($git, $path) or return 0;
+		PublicInbox::View::index_entry($fh, $mime, 0, $state);
+		1;
+	});
+	Email::Address->purge_cache;
+	$last;
+}
+
+sub nav_footer {
+	my ($cgi, $last, $feed_opts, $state, $param) = @_;
+	$cgi or return '';
+	my $old_r = $cgi->param($param);
+	my $head = '    ';
+	my $next = '    ';
+	my $first = $state->{first};
+	my $anchor = $state->{anchor_idx};
+
+	if ($last) {
+		$next = qq!<a\nhref="?$param=$last">next</a>!;
+	}
+	if ($old_r) {
+		$head = $cgi->path_info;
+		$head = qq!<a\nhref="$head">head</a>!;
+	}
+	my $atom = "<a\nhref=\"$feed_opts->{atomurl}\">atom</a>";
+	"<a\nname=\"s$anchor\">page:</a> $next $head $atom";
+}
+
+sub each_recent_blob {
+	my ($ctx, $cb) = @_;
+	my $max = $ctx->{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 $refhex = qr/(?:HEAD|${hex}{4,40})(?:~\d+)?/;
+	my $cgi = $ctx->{cgi};
+
+	# revision ranges may be specified
+	my $range = 'HEAD';
+	my $r = $cgi->param('r') if $cgi;
+	if ($r && ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o)) {
+		$range = $r;
+	}
 
 	# 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 log --no-color --raw -r --no-abbrev HEAD |";
-	my $pid = open my $log, $cmd or die "open `$cmd' pipe failed: $!\n";
-	my %deleted;
+	my @cmd = ('git', "--git-dir=$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 %deleted; # only an optimization at this point
+	my $last;
 	my $nr = 0;
-	foreach my $line (<$log>) {
-		if ($line =~ /^:000000 100644 0{40} ([a-f0-9]{40})/) {
+	my ($cur_commit, $first_commit, $last_commit);
+	my ($ts, $subj, $u);
+	while (defined(my $line = <$log>)) {
+		if ($line =~ /$addmsg/o) {
 			my $add = $1;
-			next if $deleted{$add};
-			$nr += add_to_feed($feed_opts, $feed, $add);
-			last if $nr >= $max;
-		} elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) {
+			next if $deleted{$add}; # optimization-only
+			$nr += $cb->($add, $cur_commit, $ts, $u, $subj);
+			if ($nr >= $max) {
+				$last = 1;
+				last;
+			}
+		} elsif ($line =~ /$delmsg/o) {
 			$deleted{$1} = 1;
+		} elsif ($line =~ /^${hex}{7,40}/o) {
+			($cur_commit, $ts, $u, $subj) = split("\0", $line);
+			unless (defined $first_commit) {
+				$first_commit = $cur_commit;
+			}
+		}
+	}
+
+	if ($last) {
+		while (my $line = <$log>) {
+			if ($line =~ /^(${hex}{7,40})/o) {
+				$last_commit = $1;
+				last;
+			}
 		}
 	}
-	close $log;
 
-	$feed->as_string;
+	close $log; # we may EPIPE here
+	# for pagination
+	($first_commit, $last_commit);
 }
 
 # private functions below
 sub get_feedopts {
+	my ($ctx) = @_;
+	my $pi_config = $ctx->{pi_config};
+	my $listname = $ctx->{listname};
+	my $cgi = $ctx->{cgi};
 	my %rv;
-	foreach my $key (qw(title url atomUrl email)) {
-		my $tmp = `git config publicInboxFeed.$key`;
-		chomp $tmp;
-		$rv{$key} = $tmp;
+	if (open my $fh, '<', "$ctx->{git_dir}/description") {
+		chomp($rv{description} = <$fh>);
+		close $fh;
+	} else {
+		$rv{description} = '($GIT_DIR/description missing)';
+	}
+
+	if ($pi_config && defined $listname && $listname ne '') {
+		my $addr = $pi_config->get($listname, 'address') || "";
+		$rv{address} = $addr;
+		$addr = $addr->[0] if ref($addr);
+		$rv{id_addr} = $addr;
+	}
+	$rv{id_addr} ||= 'public-inbox@example.com';
+
+	my $url_base;
+	if ($cgi) {
+		my $base;
+		if (ref($cgi) eq 'CGI') {
+			$base = $cgi->url(-base);
+		} else {
+			$base = $cgi->base->as_string;
+			$base =~ s!/\z!!;
+		}
+		$url_base = "$base/$listname";
+		if (my $mid = $ctx->{mid}) { # per-thread feed:
+			$rv{atomurl} = "$url_base/$mid/t.atom";
+		} else {
+			$rv{atomurl} = "$url_base/new.atom";
+		}
+	} else {
+		$url_base = "http://example.com";
+		$rv{atomurl} = "$url_base/new.atom";
 	}
+	$rv{url} ||= "$url_base/";
+	$rv{midurl} = "$url_base/";
+
 	\%rv;
 }
 
-sub utf8_header {
+sub mime_header {
 	my ($mime, $name) = @_;
-	encode('utf8', decode('MIME-Header', $mime->header($name)));
+	PublicInbox::Hval->new_oneline($mime->header($name))->raw;
 }
 
 sub feed_date {
 	my ($date) = @_;
-	my $dt = $dt_parser->parse_datetime($date);
-        $dt ? $dt->strftime(DATEFMT) : 0;
+	my @t = eval { strptime($date) };
+
+	scalar(@t) ? strftime(DATEFMT, @t) : 0;
 }
 
 # returns 0 (skipped) or 1 (added)
 sub add_to_feed {
-	my ($feed_opts, $feed, $add) = @_;
+	my ($feed_opts, $fh, $add, $git) = @_;
 
-	# we can use git cat-file --batch if performance becomes a
-	# problem, but I doubt it...
-	my $str = `git cat-file blob $add`;
-	return 0 if $? != 0;
-	my $mime = Email::MIME->new($str);
+	my $mime = do_cat_mail($git, $add) or return 0;
+	my $url = $feed_opts->{url};
+	my $midurl = $feed_opts->{midurl};
 
-	my $content = msg_content($mime);
+	my $header_obj = $mime->header_obj;
+	my $mid = $header_obj->header('Message-ID');
+	defined $mid or return 0;
+	$mid = PublicInbox::Hval->new_msgid($mid);
+	my $href = $mid->as_href;
+	my $content = PublicInbox::View->feed_entry($mime, "$midurl$href/f/");
 	defined($content) or return 0;
+	$mime = undef;
 
-	my $mid_url = $feed_opts->{mid_url} || "http://example.com/mid/%s";
-	my $mid = utf8_header($mime, "Message-ID") or return 0;
-	$mid =~ s/\A<//;
-	$mid =~ s/>\z//;
-
-	my $subject = utf8_header($mime, "Subject") || "";
-	defined($subject) && length($subject) or return 0;
+	my $date = $header_obj->header('Date');
+	$date = PublicInbox::Hval->new_oneline($date);
+	$date = feed_date($date->raw) or return 0;
+	$date = "<updated>$date</updated>";
 
-	my $from = utf8_header($mime, "From") or return 0;
+	my $title = mime_header($header_obj, 'Subject') or return 0;
+	$title = title_tag($title);
 
-	my @from = Email::Address->parse($from);
-	my $name = $from[0]->name;
-	defined $name or $name = "";
+	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;
-	defined $email or $email = "";
-
-	my $url = sprintf($mid_url, uri_escape($mid));
-	my $date = utf8_header($mime, "Date");
-	$date or return 0;
-	$date = feed_date($date) or return 0;
-	$feed->add_entry(
-		author => { name => $name, email => $email },
-		title => $subject,
-		updated => $date,
-		content => { type => "html", content => $content },
-		link => $url,
-		id => $add,
-	);
+	$email = PublicInbox::Hval->new_oneline($email)->as_html;
+
+	if (delete $feed_opts->{emit_header}) {
+		$fh->write(atom_header($feed_opts, $title) . $date);
+	}
+	$fh->write("<entry><author><name>$name</name><email>$email</email>" .
+		   "</author>$title$date" .
+		   qq{<content\ntype="xhtml">} .
+		   qq{<div\nxmlns="http://www.w3.org/1999/xhtml">});
+	$fh->write($content);
+
+	$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);
+	$fh->write(qq!</div></content><link\nhref="$midurl$href/"/>!.
+		   "<id>$id</id></entry>");
 	1;
 }
 
-# returns a plain-text message body without quoted text
-# returns undef if there was nothing
-sub msg_content {
-	my ($mime) = @_;
-	my $rv;
-
-	# scan through all parts, looking for displayable text
-	$mime->walk_parts(sub {
-		return if $rv;
-		my ($part) = @_;
-		return if $part->subparts; # walk_parts already recurses
-		my $ct = $part->content_type || 'text/plain';
-		return if $ct !~ m!\btext/[a-z0-9\+\._-]+\b!i;
-		my @body;
-		my $killed_wrote; # omit "So-and-so wrote:" line
-
-		# no quoted text in Atom feed summary
-		# $part->body should already be decoded for us (no QP)
-
-		my $state = 0; # 0: beginning, 1: keep, 2: quoted
-		foreach my $l (split(/\r?\n/, $part->body)) {
-			if ($state == 0) {
-				# drop leading blank lines
-				next if $l =~ /\A\s*\z/;
-
-				$state = ($l =~ /\A>/) ? 2 : 1; # fall-through
-			}
-			if ($state == 2) { # quoted text, drop it
-				if ($l !~ /\A>/) {
-					push @body, "<quoted text snipped>";
-					if ($l =~ /\S/) {
-						push @body, $l;
-					}
-					$state = 1;
-				}
-			}
-			if ($state == 1) { # stuff we may keep
-				if ($l =~ /\A>/) {
-					# drop "So-and-so wrote:" line
-					if (@body && !$killed_wrote &&
-					    $body[-1] =~ /:\z/) {
-						$killed_wrote = 1;
-						pop @body;
-					}
-					$state = 2;
-				} else {
-					push @body, $l;
-				}
-			}
-		}
-		$rv = "<pre>" .
-			join("\n", map { escapeHTML($_) } @body) .
-			"</pre>";
-	});
-
-	$rv;
+sub do_cat_mail {
+	my ($git, $path) = @_;
+	my $mime = eval {
+		my $str = $git->cat_file("HEAD:$path");
+		Email::MIME->new($str);
+	};
+	$@ ? undef : $mime;
 }
 
 1;