X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FView.pm;h=81e83d97738256511a530bc4093bfd2793267eb8;hb=956ede734d7c2e8d0a3003c6e2d554114586643e;hp=388207c6b1125f736ea47c0746271fbdc8db6ef0;hpb=eb9db159012d8302695500fc2af675c68d5ed9e6;p=public-inbox.git

diff --git a/lib/PublicInbox/View.pm b/lib/PublicInbox/View.pm
index 388207c6..81e83d97 100644
--- a/lib/PublicInbox/View.pm
+++ b/lib/PublicInbox/View.pm
@@ -24,12 +24,13 @@ sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
 sub msg_html {
 	my ($ctx, $mime) = @_;
 	my $hdr = $mime->header_obj;
-	my $obfs = $ctx->{-inbox}->{obfuscate};
-	my $tip = _msg_html_prepare($hdr, $ctx, $obfs);
+	my $ibx = $ctx->{-inbox};
+	my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
+	my $tip = _msg_html_prepare($hdr, $ctx, $obfs_ibx);
 	PublicInbox::WwwStream->response($ctx, 200, sub {
 		my ($nr, undef) = @_;
 		if ($nr == 1) {
-			$tip . multipart_text_as_html($mime, '', $obfs) .
+			$tip . multipart_text_as_html($mime, '', $obfs_ibx) .
 				'</pre><hr>'
 		} elsif ($nr == 2) {
 			# fake an EOF if generating the footer fails;
@@ -103,9 +104,8 @@ EOF
 sub in_reply_to {
 	my ($hdr) = @_;
 	my %mid = map { $_ => 1 } $hdr->header_raw('Message-ID');
-	my @refs = ($hdr->header_raw('References'),
-			$hdr->header_raw('In-Reply-To'));
-	@refs = ((join(' ', @refs)) =~ /<([^>]+)>/g);
+	my @refs = (($hdr->header_raw('References') || '') =~ /<([^>]+)>/g);
+	push(@refs, (($hdr->header_raw('In-Reply-To') || '') =~ /<([^>]+)>/g));
 	while (defined(my $irt = pop @refs)) {
 		next if $mid{"<$irt>"};
 		return $irt;
@@ -138,11 +138,11 @@ sub index_entry {
 
 	my $root_anchor = $ctx->{root_anchor} || '';
 	my $irt;
-	my $obfs = $ctx->{-obfuscate};
+	my $obfs_ibx = $ctx->{-obfs_ibx};
 
 	my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
 	$subj = '<b>'.ascii_html($subj).'</b>';
-	obfuscate_addrs($subj) if $obfs;
+	obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
 	$subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
 	$rv .= $subj . "\n";
 	$rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
@@ -150,12 +150,12 @@ sub index_entry {
 	foreach my $f (qw(To Cc)) {
 		my $dst = _hdr_names_html($hdr, $f);
 		if ($dst ne '') {
-			obfuscate_addrs($dst) if $obfs;
+			obfuscate_addrs($obfs_ibx, $dst) if $obfs_ibx;
 			push @tocc, "$f: $dst";
 		}
 	}
 	my $from = _hdr_names_html($hdr, 'From');
-	obfuscate_addrs($from) if $obfs;
+	obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
 	$rv .= "From: $from @ "._msg_date($hdr)." UTC";
 	my $upfx = $ctx->{-upfx};
 	my $mhref = $upfx . mid_escape($mid_raw) . '/';
@@ -173,7 +173,7 @@ sub index_entry {
 	$rv .= "\n";
 
 	# scan through all parts, looking for displayable text
-	msg_iter($mime, sub { $rv .= add_text_body($mhref, $obfs, $_[0]) });
+	msg_iter($mime, sub { $rv .= add_text_body($mhref, $obfs_ibx, $_[0]) });
 
 	# add the footer
 	$rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
@@ -186,17 +186,17 @@ sub index_entry {
 		$rv .= "\t[relevance $pct->{$mid_raw}%]";
 		$hr = 1;
 	} elsif ($mapping) {
-		my $threaded = 'threaded';
+		my $nested = 'nested';
 		my $flat = 'flat';
 		my $end = '';
 		if ($ctx->{flat}) {
 			$hr = 1;
 			$flat = "<b>$flat</b>";
 		} else {
-			$threaded = "<b>$threaded</b>";
+			$nested = "<b>$nested</b>";
 		}
 		$rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
-		$rv .= "|<a\nhref=\"${mhref}t/#u\">$threaded</a>]";
+		$rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
 		$rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
 	} else {
 		$hr = $ctx->{-hr};
@@ -319,7 +319,7 @@ sub stream_thread ($$) {
 	}
 	return missing_thread($ctx) unless $mime;
 
-	$ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate};
+	$ctx->{-obfs_ibx} = $inbox->{obfuscate} ? $inbox : undef;
 	$mime = PublicInbox::MIME->new($mime);
 	$ctx->{-title_html} = ascii_html($mime->header('Subject'));
 	$ctx->{-html_tip} = thread_index_entry($ctx, $level, $mime);
@@ -367,21 +367,20 @@ sub thread_html {
 	$ctx->{prev_attr} = '';
 	$ctx->{prev_level} = 0;
 	$ctx->{root_anchor} = anchor_for($mid);
-	$ctx->{seen} = {};
 	$ctx->{mapping} = {};
 	$ctx->{s_nr} = "$nr+ messages in thread";
 
-	my $rootset = thread_results($msgs);
+	my $rootset = thread_results($msgs, $srch);
 
 	# reduce hash lookups in pre_thread->skel_dump
-	$ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate};
+	my $inbox = $ctx->{-inbox};
+	$ctx->{-obfs_ibx} = $inbox->{obfuscate} ? $inbox : undef;
 	walk_thread($rootset, $ctx, *pre_thread);
 
 	$skel .= '</pre>';
 	return stream_thread($rootset, $ctx) unless $ctx->{flat};
 
 	# flat display: lazy load the full message from smsg
-	my $inbox = $ctx->{-inbox};
 	my $mime;
 	while ($mime = shift @$msgs) {
 		$mime = $inbox->msg_by_smsg($mime) and last;
@@ -406,11 +405,11 @@ sub thread_html {
 }
 
 sub multipart_text_as_html {
-	my ($mime, $upfx, $obfs) = @_;
+	my ($mime, $upfx, $obfs_ibx) = @_;
 	my $rv = "";
 
 	# scan through all parts, looking for displayable text
-	msg_iter($mime, sub { $rv .= add_text_body($upfx, $obfs, $_[0]) });
+	msg_iter($mime, sub { $rv .= add_text_body($upfx, $obfs_ibx, $_[0]) });
 	$rv;
 }
 
@@ -463,7 +462,7 @@ sub attach_link ($$$$;$) {
 }
 
 sub add_text_body {
-	my ($upfx, $obfs, $p) = @_;
+	my ($upfx, $obfs_ibx, $p) = @_;
 	# $p - from msg_iter: [ Email::MIME, depth, @idx ]
 	my ($part, $depth) = @$p; # attachment @idx is unused
 	my $ct = $part->content_type || 'text/plain';
@@ -515,10 +514,10 @@ sub add_text_body {
 
 	if (@quot) { # ugh, top posted
 		flush_quote(\$s, $l, \@quot);
-		obfuscate_addrs($s) if $obfs;
+		obfuscate_addrs($obfs_ibx, $s) if $obfs_ibx;
 		$s;
 	} else {
-		obfuscate_addrs($s) if $obfs;
+		obfuscate_addrs($obfs_ibx, $s) if $obfs_ibx;
 		if ($s =~ /\n\z/s) { # common, last line ends with a newline
 			$s;
 		} else { # some editors don't do newlines...
@@ -528,7 +527,7 @@ sub add_text_body {
 }
 
 sub _msg_html_prepare {
-	my ($hdr, $ctx, $obfs) = @_;
+	my ($hdr, $ctx, $obfs_ibx) = @_;
 	my $srch = $ctx->{srch} if $ctx;
 	my $atom = '';
 	my $rv = "<pre\nid=b>"; # anchor for body start
@@ -547,7 +546,7 @@ sub _msg_html_prepare {
 		if ($h eq 'From') {
 			my @n = PublicInbox::Address::names($v->raw);
 			$title[1] = ascii_html(join(', ', @n));
-			obfuscate_addrs($title[1]) if $obfs;
+			obfuscate_addrs($obfs_ibx, $title[1]) if $obfs_ibx;
 		} elsif ($h eq 'Subject') {
 			$title[0] = $v->as_html;
 			if ($srch) {
@@ -557,7 +556,7 @@ sub _msg_html_prepare {
 			}
 		}
 		$v = $v->as_html;
-		obfuscate_addrs($v) if $obfs;
+		obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
 		$rv .= "$h: $v\n";
 
 	}
@@ -596,8 +595,9 @@ sub thread_skel {
 	$$dst .= "$nr+ messages in thread ($expand";
 	$$dst .= qq! / <a\nhref="#b">[top]</a>)\n!;
 
-	my $subj = $srch->subject_path($hdr->header('Subject'));
-	$ctx->{seen} = { $subj => 1 };
+	my $subj = $hdr->header('Subject');
+	defined $subj or $subj = '';
+	$ctx->{prev_subj} = [ split(/ /, $srch->subject_normalized($subj)) ];
 	$ctx->{cur} = $mid;
 	$ctx->{prev_attr} = '';
 	$ctx->{prev_level} = 0;
@@ -605,8 +605,9 @@ sub thread_skel {
 	$sres = load_results($srch, $sres);
 
 	# reduce hash lookups in skel_dump
-	$ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate};
-	walk_thread(thread_results($sres), $ctx, *skel_dump);
+	my $ibx = $ctx->{-inbox};
+	$ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
+	walk_thread(thread_results($sres, $srch), $ctx, *skel_dump);
 
 	$ctx->{parent_msg} = $parent;
 }
@@ -719,7 +720,7 @@ sub ghost_parent {
 
 sub indent_for {
 	my ($level) = @_;
-	INDENT x ($level - 1);
+	$level ? INDENT x ($level - 1) : '';
 }
 
 sub load_results {
@@ -735,9 +736,9 @@ sub msg_timestamp {
 }
 
 sub thread_results {
-	my ($msgs) = @_;
+	my ($msgs, $srch) = @_;
 	require PublicInbox::SearchThread;
-	PublicInbox::SearchThread::thread($msgs, *sort_ts);
+	PublicInbox::SearchThread::thread($msgs, *sort_ts, $srch);
 }
 
 sub missing_thread {
@@ -754,6 +755,25 @@ sub _msg_date {
 
 sub fmt_ts { POSIX::strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
 
+sub dedupe_subject {
+	my ($prev_subj, $subj, $val) = @_;
+
+	my $omit = ''; # '"' denotes identical text omitted
+	my (@prev_pop, @curr_pop);
+	while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
+		push(@prev_pop, pop(@$prev_subj));
+		push(@curr_pop, pop(@$subj));
+		$omit ||= $val;
+	}
+	pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
+	if (scalar(@curr_pop) == 1) {
+		$omit = '';
+		push @$prev_subj, @prev_pop;
+		push @$subj, @curr_pop;
+	}
+	$omit;
+}
+
 sub skel_dump {
 	my ($ctx, $level, $node) = @_;
 	my $smsg = $node->{smsg} or return _skel_ghost($ctx, $level, $node);
@@ -763,8 +783,8 @@ sub skel_dump {
 	my $mid = $smsg->{mid};
 
 	my $f = ascii_html($smsg->from_name);
-	my $obfs = $ctx->{-obfuscate};
-	obfuscate_addrs($f) if $obfs;
+	my $obfs_ibx = $ctx->{-obfs_ibx};
+	obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
 
 	my $d = fmt_ts($smsg->{ts}) . ' ' . indent_for($level) . th_pfx($level);
 	my $attr = $f;
@@ -791,20 +811,25 @@ sub skel_dump {
 	# Subject is never undef, this mail was loaded from
 	# our Xapian which would've resulted in '' if it were
 	# really missing (and Filter rejects empty subjects)
-	my $subj = $smsg->subject;
-	my $h = $ctx->{srch}->subject_path($subj);
-	if ($ctx->{seen}->{$h}) {
-		$subj = undef;
+	my @subj = split(/ /, $ctx->{srch}->subject_normalized($smsg->subject));
+
+	# remove common suffixes from the subject if it matches the previous,
+	# so we do not show redundant text at the end.
+	my $prev_subj = $ctx->{prev_subj} || [];
+	$ctx->{prev_subj} = [ @subj ];
+	my $omit = dedupe_subject($prev_subj, \@subj, '&#34; ');
+	my $end;
+	if (@subj) {
+		my $subj = join(' ', @subj);
+		$subj = ascii_html($subj);
+		obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
+		$end = "$subj</a> $omit$f\n"
 	} else {
-		$ctx->{seen}->{$h} = 1;
-		$subj = PublicInbox::Hval->new($subj);
-		$subj = $subj->as_html;
-		obfuscate_addrs($subj) if $obfs;
+		$end = "$f</a>\n";
 	}
 	my $m;
 	my $id = '';
 	my $mapping = $ctx->{mapping};
-	my $end = defined($subj) ? "$subj</a> $f\n" : "$f</a>\n";
 	if ($mapping) {
 		my $map = $mapping->{$mid};
 		$id = id_compress($mid, 1);
@@ -896,7 +921,9 @@ sub dump_topics {
 	}
 
 	my @out;
-	my $obfs = $ctx->{-inbox}->{obfuscate};
+	my $ibx = $ctx->{-inbox};
+	my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
+	my $srch = $ctx->{srch};
 
 	# sort by recency, this allows new posts to "bump" old topics...
 	foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
@@ -905,6 +932,7 @@ sub dump_topics {
 		next unless defined $top;  # ghost topic
 		my $mid = delete $seen->{$top};
 		my $href = mid_escape($mid);
+		my $prev_subj = [ split(/ /, $top) ];
 		$top = PublicInbox::Hval->new($top)->as_html;
 		$ts = fmt_ts($ts);
 
@@ -927,11 +955,15 @@ sub dump_topics {
 			my $level = $ex[$i];
 			my $subj = $ex[$i + 1];
 			$mid = delete $seen->{$subj};
-			$subj = ascii_html($subj);
-			obfuscate_addrs($subj) if $obfs;
+			my @subj = split(/ /, $srch->subject_normalized($subj));
+			my @next_prev = @subj; # full copy
+			my $omit = dedupe_subject($prev_subj, \@subj, ' &#34;');
+			$prev_subj = \@next_prev;
+			$subj = ascii_html(join(' ', @subj));
+			obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
 			$href = mid_escape($mid);
 			$s .= indent_for($level) . TCHILD;
-			$s .= "<a\nhref=\"$href/T/#u\">$subj</a>\n";
+			$s .= "<a\nhref=\"$href/T/#u\">$subj</a>$omit\n";
 		}
 		push @out, $s;
 	}
@@ -974,7 +1006,7 @@ sub index_topics {
 	my $nr = scalar @{$sres->{msgs}};
 	if ($nr) {
 		$sres = load_results($srch, $sres);
-		walk_thread(thread_results($sres), $ctx, *acc_topic);
+		walk_thread(thread_results($sres, $srch), $ctx, *acc_topic);
 	}
 	$ctx->{-next_o} = $off+ $nr;
 	$ctx->{-cur_o} = $off;