X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FView.pm;h=a78ce31d291ea0859edc5cf75a7e2d253579c3a2;hb=23a4e44bedabe5b8b651346cabc2a870c5377a30;hp=ac44d442b166ba1189772d363612a4c638450891;hpb=fdf9b900b8c35cb4259dbdf56400d44db61e7a9e;p=public-inbox.git

diff --git a/lib/PublicInbox/View.pm b/lib/PublicInbox/View.pm
index ac44d442..a78ce31d 100644
--- a/lib/PublicInbox/View.pm
+++ b/lib/PublicInbox/View.pm
@@ -10,15 +10,16 @@ use URI::Escape qw/uri_escape_utf8/;
 use Date::Parse qw/str2time/;
 use Encode qw/find_encoding/;
 use Encode::MIME::Header;
-use Email::MIME::ContentType qw/parse_content_type/;
 use PublicInbox::Hval qw/ascii_html/;
 use PublicInbox::Linkify;
 use PublicInbox::MID qw/mid_clean id_compress mid2path mid_mime/;
+use PublicInbox::MsgIter;
+use PublicInbox::Address;
 require POSIX;
 
 use constant INDENT => '  ';
-
-my $enc_utf8 = find_encoding('UTF-8');
+use constant TCHILD => '` ';
+sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
 
 # public functions:
 sub msg_html {
@@ -26,14 +27,14 @@ sub msg_html {
 	$footer = defined($footer) ? "\n$footer" : '';
 	my $hdr = $mime->header_obj;
 	headers_to_html_header($hdr, $ctx) .
-		multipart_text_as_html($mime) .
+		multipart_text_as_html($mime, '') .
 		'</pre><hr /><pre>' .
-		html_footer($hdr, 1, $ctx) .
+		html_footer($hdr, 1, $ctx, 'R/') .
 		$footer .
 		'</pre></body></html>';
 }
 
-# /$LISTNAME/$MESSAGE_ID/R/
+# /$INBOX/$MESSAGE_ID/R/
 sub msg_reply {
 	my ($ctx, $hdr, $footer) = @_;
 	my $s = $hdr->header('Subject');
@@ -71,13 +72,6 @@ sub msg_reply {
 	'</pre><hr /><pre>' . $footer .  '</pre></body></html>';
 }
 
-sub feed_entry {
-	my ($class, $mime) = @_;
-
-	# no <head> here for <style>...
-	PublicInbox::Hval::PRE . multipart_text_as_html($mime) . '</pre>';
-}
-
 sub in_reply_to {
 	my ($hdr) = @_;
 	my $irt = $hdr->header_raw('In-Reply-To');
@@ -97,9 +91,7 @@ sub index_entry {
 	my $midx = $state->{anchor_idx}++;
 	my $ctx = $state->{ctx};
 	my $srch = $ctx->{srch};
-	my $part_nr = 0;
 	my $hdr = $mime->header_obj;
-	my $enc = enc_for($hdr->header("Content-Type"));
 	my $subj = $hdr->header('Subject');
 
 	my $mid_raw = mid_clean(mid_mime($mime));
@@ -108,9 +100,7 @@ sub index_entry {
 	$seen->{$id} = "#$id"; # save the anchor for children, later
 
 	my $mid = PublicInbox::Hval->new_msgid($mid_raw);
-	my $from = $hdr->header('From');
-	my @from = Email::Address->parse($from);
-	$from = $from[0]->name;
+	my $from = PublicInbox::Address::from_name($hdr->header('From'));
 
 	my $root_anchor = $state->{root_anchor} || '';
 	my $path = $root_anchor ? '../../' : '';
@@ -133,11 +123,8 @@ sub index_entry {
 	my $mhref = "${path}$href/";
 
 	# scan through all parts, looking for displayable text
-	$mime->walk_parts(sub {
-		index_walk($fh, $_[0], $enc, \$part_nr);
-	});
-	$mime->body_set('');
-	$rv = "\n" . html_footer($hdr, 0, $ctx, $mhref);
+	msg_iter($mime, sub { index_walk($fh, $mhref, $_[0]) });
+	$rv = "\n" . html_footer($hdr, 0, $ctx, "$path$href/R/");
 
 	if (defined $irt) {
 		unless (defined $parent_anchor) {
@@ -152,13 +139,15 @@ sub index_entry {
 	} elsif ($srch) {
 		my $threaded = 'threaded';
 		my $flat = 'flat';
+		my $end = '';
 		if ($ctx->{flat}) {
 			$flat = "<b>$flat</b>";
+			$end = "\n"; # for lynx
 		} else {
 			$threaded = "<b>$threaded</b>";
 		}
 		$rv .= " [<a\nhref=\"${path}$href/t/#u\">$threaded</a>";
-		$rv .= "|<a\nhref=\"${path}$href/T/#u\">$flat</a>]";
+		$rv .= "|<a\nhref=\"${path}$href/T/#u\">$flat</a>]$end";
 	}
 	$fh->write($rv .= '</pre>');
 }
@@ -174,10 +163,10 @@ sub thread_html {
 sub emit_thread_html {
 	my ($res, $ctx, $foot, $srch) = @_;
 	my $mid = $ctx->{mid};
-	my $msgs = load_results($srch->get_thread($mid));
+	my $flat = $ctx->{flat};
+	my $msgs = load_results($srch->get_thread($mid, { asc => $flat }));
 	my $nr = scalar @$msgs;
 	return missing_thread($res, $ctx) if $nr == 0;
-	my $flat = $ctx->{flat};
 	my $seen = {};
 	my $state = {
 		res => $res,
@@ -201,7 +190,6 @@ sub emit_thread_html {
 				('</ul></li>' x ($max - 1)) . '</ul>');
 		}
 	}
-	Email::Address->purge_cache;
 
 	# there could be a race due to a message being deleted in git
 	# but still being in the Xapian index:
@@ -220,8 +208,8 @@ sub emit_thread_html {
 }
 
 sub index_walk {
-	my ($fh, $part, $enc, $part_nr) = @_;
-	my $s = add_text_body($enc, $part, $part_nr);
+	my ($fh, $upfx, $p) = @_;
+	my $s = add_text_body($upfx, $p);
 
 	return if $s eq '';
 
@@ -230,49 +218,22 @@ sub index_walk {
 	$fh->write($s);
 }
 
-sub enc_for {
-	my ($ct, $default) = @_;
-	$default ||= $enc_utf8;
-	defined $ct or return $default;
-	my $ct_parsed = parse_content_type($ct);
-	if ($ct_parsed) {
-		if (my $charset = $ct_parsed->{attributes}->{charset}) {
-			my $enc = find_encoding($charset);
-			return $enc if $enc;
-		}
-	}
-	$default;
-}
-
 sub multipart_text_as_html {
-	my ($mime) = @_;
+	my ($mime, $upfx) = @_;
 	my $rv = "";
-	my $part_nr = 0;
-	my $enc = enc_for($mime->header("Content-Type"));
 
 	# scan through all parts, looking for displayable text
-	$mime->walk_parts(sub {
-		my ($part) = @_;
-		$part = add_text_body($enc, $part, \$part_nr);
-		$rv .= $part;
-		$rv .= "\n" if $part ne '';
+	msg_iter($mime, sub {
+		my ($p) = @_;
+		$p = add_text_body($upfx, $p);
+		$rv .= $p;
+		$rv .= "\n" if $p ne '';
 	});
-	$mime->body_set('');
 	$rv;
 }
 
-sub add_filename_line {
-	my ($enc, $fn) = @_;
-	my $len = 72;
-	my $pad = "-";
-	$fn = $enc->decode($fn);
-	$len -= length($fn);
-	$pad x= ($len/2) if ($len > 0);
-	"$pad " . ascii_html($fn) . " $pad\n";
-}
-
 sub flush_quote {
-	my ($s, $l, $quot, $part_nr) = @_;
+	my ($s, $l, $quot) = @_;
 
 	# show everything in the full version with anchor from
 	# short version (see above)
@@ -285,35 +246,59 @@ sub flush_quote {
 	$$s .= qq(<span\nclass="q">) . $rv . '</span>'
 }
 
-sub add_text_body {
-	my ($enc_msg, $part, $part_nr) = @_;
-	return '' if $part->subparts;
+sub attach_link ($$$$) {
+	my ($upfx, $ct, $p, $fn) = @_;
+	my ($part, $depth, @idx) = @$p;
+	my $nl = $idx[-1] > 1 ? "\n" : '';
+	my $idx = join('.', @idx);
+	my $size = bytes::length($part->body);
+	$ct ||= 'text/plain';
+	$ct =~ s/;.*//; # no attributes
+	$ct = ascii_html($ct);
+	my $desc = $part->header('Content-Description');
+	$desc = $fn unless defined $desc;
+	$desc = '' unless defined $desc;
+	my $sfn;
+	if (defined $fn && $fn =~ /\A[[:alnum:]][\w\.-]+[[:alnum:]]\z/) {
+		$sfn = $fn;
+	} elsif ($ct eq 'text/plain') {
+		$sfn = 'a.txt';
+	} else {
+		$sfn = 'a.bin';
+	}
+	my @ret = qq($nl<a\nhref="$upfx$idx-$sfn">[-- Attachment #$idx: );
+	my $ts = "Type: $ct, Size: $size bytes";
+	push(@ret, ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]");
+	join('', @ret, '</a>');
+}
 
+sub add_text_body {
+	my ($upfx, $p) = @_; # from msg_iter: [ Email::MIME, depth, @idx ]
+	my ($part, $depth, @idx) = @$p;
 	my $ct = $part->content_type;
-	# account for filter bugs...
+	my $fn = $part->filename;
+
 	if (defined $ct && $ct =~ m!\btext/x?html\b!i) {
-		$part->body_set('');
-		return '';
+		return attach_link($upfx, $ct, $p, $fn);
 	}
-	my $enc = enc_for($ct, $enc_msg);
-	my $s = $part->body;
-	$part->body_set('');
-	$s = $enc->decode($s);
+
+	my $s = eval { $part->body_str };
+
+	# badly-encoded message? tell the world about it!
+	return attach_link($upfx, $ct, $p, $fn) if $@;
+
 	my @lines = split(/^/m, $s);
 	$s = '';
-
-	if ($$part_nr > 0) {
-		my $fn = $part->filename;
-		defined($fn) or $fn = "part #" . ($$part_nr + 1);
-		$s .= add_filename_line($enc, $fn);
+	if (defined($fn) || $depth > 0) {
+		$s .= attach_link($upfx, $ct, $p, $fn);
+		$s .= "\n\n";
 	}
-
 	my @quot;
 	my $l = PublicInbox::Linkify->new;
 	while (defined(my $cur = shift @lines)) {
 		if ($cur !~ /^>/) {
 			# show the previously buffered quote inline
-			flush_quote(\$s, $l, \@quot, $$part_nr) if @quot;
+			flush_quote(\$s, $l, \@quot) if @quot;
 
 			# regular line, OK
 			$cur = $l->linkify_1($cur);
@@ -324,9 +309,7 @@ sub add_text_body {
 		}
 	}
 
-	flush_quote(\$s, $l, \@quot, $$part_nr) if @quot;
-	++$$part_nr;
-
+	flush_quote(\$s, $l, \@quot) if @quot;
 	$s =~ s/[ \t]+$//sgm; # kill per-line trailing whitespace
 	$s =~ s/\A\n+//s; # kill leading blank lines
 	$s =~ s/\s+\z//s; # kill all trailing spaces (final "\n" added if ne '')
@@ -354,8 +337,7 @@ sub headers_to_html_header {
 		$v = PublicInbox::Hval->new($v);
 
 		if ($h eq 'From') {
-			my @from = Email::Address->parse($v->raw);
-			$title[1] = ascii_html($from[0]->name);
+			$title[1] = PublicInbox::Address::from_name($v->raw);
 		} elsif ($h eq 'Subject') {
 			$title[0] = $v->as_html;
 			if ($srch) {
@@ -379,19 +361,20 @@ sub headers_to_html_header {
 }
 
 sub thread_skel {
-	my ($dst, $ctx, $hdr, $upfx) = @_;
+	my ($dst, $ctx, $hdr, $tpfx) = @_;
 	my $srch = $ctx->{srch};
 	my $mid = mid_clean($hdr->header_raw('Message-ID'));
 	my $sres = $srch->get_thread($mid);
 	my $nr = $sres->{total};
-	my $expand = "<a\nhref=\"${upfx}t/#u\">expand</a> " .
-			"/ <a\nhref=\"${upfx}t.mbox.gz\">mbox.gz</a>";
+	my $expand = qq(<a\nhref="${tpfx}t/#u">expand</a> ) .
+			qq(/ <a\nhref="${tpfx}t.mbox.gz">mbox.gz</a> ) .
+			qq(/ <a\nhref="${tpfx}t.atom">Atom feed</a>);
 
 	my $parent = in_reply_to($hdr);
 	if ($nr <= 1) {
 		if (defined $parent) {
 			$$dst .= "($expand)\n ";
-			$$dst .= ghost_parent("$upfx../", $parent) . "\n";
+			$$dst .= ghost_parent("$tpfx../", $parent) . "\n";
 		} else {
 			$$dst .= "[no followups, yet] ($expand)\n";
 		}
@@ -412,7 +395,7 @@ sub thread_skel {
 		prev_level => 0,
 	};
 	for (thread_results(load_results($sres))->rootset) {
-		skel_dump($dst, $state, $upfx, $_, 0);
+		skel_dump($dst, $state, $tpfx, $_, 0);
 	}
 	$ctx->{next_msg} = $state->{next_msg};
 	$ctx->{parent_msg} = $parent;
@@ -463,15 +446,13 @@ sub mailto_arg_link {
 	foreach my $h (qw(From To Cc)) {
 		my $v = $hdr->header($h);
 		defined($v) && ($v ne '') or next;
-		my @addrs = Email::Address->parse($v);
-		foreach my $recip (@addrs) {
-			my $address = $recip->address;
+		my @addrs = PublicInbox::Address::emails($v);
+		foreach my $address (@addrs) {
 			my $dst = lc($address);
 			$cc{$dst} ||= $address;
 			$to ||= $dst;
 		}
 	}
-	Email::Address->purge_cache;
 	my @arg;
 
 	my $subj = $hdr->header('Subject') || '';
@@ -493,30 +474,28 @@ sub mailto_arg_link {
 }
 
 sub html_footer {
-	my ($hdr, $standalone, $ctx, $mhref) = @_;
+	my ($hdr, $standalone, $ctx, $rhref) = @_;
 
 	my $srch = $ctx->{srch} if $ctx;
 	my $upfx = '../';
 	my $tpfx = '';
 	my $idx = $standalone ? " <a\nhref=\"$upfx\">index</a>" : '';
 	my $irt = '';
-
-	if ($srch && $standalone) {
-		$idx .= qq{ / follow: <a\nhref="${tpfx}t.atom">Atom feed</a>\n};
-	}
 	if ($idx && $srch) {
+		$idx .= "\n";
 		thread_skel(\$idx, $ctx, $hdr, $tpfx);
 		my $p = $ctx->{parent_msg};
 		my $next = $ctx->{next_msg};
 		if ($p) {
 			$p = PublicInbox::Hval->new_msgid($p);
 			$p = $p->as_href;
-			$irt = "<a\nhref=\"$upfx$p/\">parent</a> ";
+			$irt = "<a\nhref=\"$upfx$p/\"rel=prev>parent</a> ";
 		} else {
 			$irt = ' ' x length('parent ');
 		}
 		if ($next) {
-			$irt .= "<a\nhref=\"$upfx$next/\">next</a> ";
+			my $n = PublicInbox::Hval->new_msgid($next)->as_href;
+			$irt .= "<a\nhref=\"$upfx$n/\"\nrel=next>next</a> ";
 		} else {
 			$irt .= ' ' x length('next ');
 		}
@@ -524,7 +503,7 @@ sub html_footer {
 		$irt = '';
 	}
 
-	$irt . qq(<a\nhref="${tpfx}R/">reply</a>) . $idx;
+	$irt . qq(<a\nhref="$rhref">reply</a>) . $idx;
 }
 
 sub linkify_ref_nosrch {
@@ -668,21 +647,11 @@ sub msg_timestamp {
 }
 
 sub thread_results {
-	my ($msgs, $nosubject, $nosort) = @_;
+	my ($msgs) = @_;
 	require PublicInbox::Thread;
 	my $th = PublicInbox::Thread->new(@$msgs);
-
-	# WARNING! both these Mail::Thread knobs were found by inspecting
-	# the Mail::Thread 2.55 source code, and we have some monkey patches
-	# in PublicInbox::Thread to fix memory leaks.  Since Mail::Thread
-	# appears unmaintained, I suppose it's safe to depend on these
-	# variables for now:
-	no warnings 'once';
-	$Mail::Thread::nosubject = $nosubject;
-	# Keep ghosts with only a single direct child:
-	$Mail::Thread::noprune = 1;
 	$th->thread;
-	$th->order(*sort_ts) unless $nosort;
+	$th->order(*sort_ts);
 	$th
 }
 
@@ -703,13 +672,12 @@ sub fmt_ts { POSIX::strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
 
 sub _skel_header {
 	my ($dst, $state, $upfx, $hdr, $level) = @_;
-	my $dot = $level == 0 ? '' : '` ';
 
 	my $cur = $state->{cur};
 	my $mid = mid_clean($hdr->header_raw('Message-ID'));
 	my $f = ascii_html($hdr->header('X-PI-From'));
 	my $d = _msg_date($hdr);
-	my $pfx = ' ' . $d . ' ' . indent_for($level);
+	my $pfx = "$d " . indent_for($level) . th_pfx($level);
 	my $attr = $f;
 	$state->{first_level} ||= $level;
 
@@ -723,7 +691,7 @@ sub _skel_header {
 	if ($cur) {
 		if ($cur eq $mid) {
 			delete $state->{cur};
-			$$dst .= "$pfx$dot<b><a\nid=r\nhref=\"#b\">".
+			$$dst .= "$pfx<b><a\nid=r\nhref=\"#t\">".
 				 "$attr [this message]</a></b>\n";
 
 			return;
@@ -746,11 +714,8 @@ sub _skel_header {
 	}
 	my $m = PublicInbox::Hval->new_msgid($mid);
 	$m = $upfx . '../' . $m->as_href . '/';
-	if (defined $s) {
-		$$dst .= "$pfx$dot<a\nhref=\"$m\">$s</a> $attr\n";
-	} else {
-		$$dst .= "$pfx$dot<a\nhref=\"$m\">$f</a>\n";
-	}
+	$$dst .= "$pfx<a\nhref=\"$m\">";
+	$$dst .= defined($s) ? "$s</a> $f\n" : "$f</a>\n";
 }
 
 sub skel_dump {
@@ -765,9 +730,8 @@ sub skel_dump {
 		if ($mid eq 'subject dummy') {
 			$$dst .= "\t[no common parent]\n";
 		} else {
-			$$dst .= '      [not found] ';
-			my $dot = $level == 0 ? '' : '` ';
-			$$dst .= indent_for($level) . $dot;
+			$$dst .= '     [not found] ';
+			$$dst .= indent_for($level) . th_pfx($level);
 			$mid = PublicInbox::Hval->new_msgid($mid);
 			my $href = "$upfx../" . $mid->as_href . '/';
 			my $html = $mid->as_html;
@@ -844,11 +808,10 @@ sub emit_topics {
 			$cur = undef;
 		}
 		$cur ||= [ $ts, '' ];
-		my $dot = $level == 0 ? '' : '` ';
 		$cur->[0] = $ts if $ts > $cur->[0];
-		$cur->[1] .= "$nl$pfx$dot<a\nhref=\"$mid/t/#u\"><b>";
-		$cur->[1] .= $subj;
-		$cur->[1] .= "</b></a>\n";
+		$cur->[1] .= $nl . $pfx . th_pfx($level) .
+				"<a\nhref=\"$mid/t/#u\"><b>" .
+				$subj . "</b></a>\n";
 
 		$ts = fmt_ts($ts);
 		my $attr = " $ts UTC";
@@ -872,18 +835,17 @@ sub emit_topics {
 
 sub emit_index_topics {
 	my ($state) = @_;
-	my $off = $state->{ctx}->{cgi}->param('o');
-	$off = 0 unless defined $off;
+	my ($off) = (($state->{ctx}->{cgi}->param('o') || '0') =~ /(\d+)/);
 	$state->{order} = [];
 	$state->{subjs} = {};
 	$state->{latest} = {};
 	my $max = 25;
-	my %opts = ( offset => int $off, limit => $max * 4 );
+	my %opts = ( offset => $off, limit => $max * 4 );
 	while (scalar @{$state->{order}} < $max) {
 		my $sres = $state->{srch}->query('', \%opts);
 		my $nr = scalar @{$sres->{msgs}} or last;
 
-		for (thread_results(load_results($sres), 1)->rootset) {
+		for (thread_results(load_results($sres))->rootset) {
 			add_topic($state, $_, 0);
 		}
 		$opts{offset} += $nr;
