www: do not mangle characters from search queries

[public-inbox.git] / lib / PublicInbox / SearchView.pm

diff --git a/lib/PublicInbox/SearchView.pm b/lib/PublicInbox/SearchView.pm
index 50a2c01c4749f05ab54dc73a5d05ee04747d4884..42bc648260f01f7b9259a8b16ca786660480db0e 100644 (file)
--- a/lib/PublicInbox/SearchView.pm
+++ b/lib/PublicInbox/SearchView.pm
@@ -10,10 +10,10 @@ use PublicInbox::Hval qw/ascii_html/;
 use PublicInbox::View;
 use PublicInbox::WwwAtomStream;
 use PublicInbox::MID qw(mid2path mid_mime mid_clean mid_escape);
-use Email::MIME;
+use PublicInbox::MIME;
 require PublicInbox::Git;
 require PublicInbox::SearchThread;
-our $LIM = 50;
+our $LIM = 200;
 
 sub noop {}
 
@@ -145,15 +145,21 @@ sub search_nav_bot {
        my $o = $q->{o};
        my $end = $o + $nr;
        my $beg = $o + 1;
-       my $rv = "</pre><hr><pre>Results $beg-$end of $total";
+       my $rv = '</pre><hr><pre id=t>';
+       if ($beg <= $end) {
+               $rv .= "Results $beg-$end of $total";
+               $rv .= ' (estimated)' if $end != $total;
+       } else {
+               $rv .= "No more results, only $total";
+       }
        my $n = $o + $LIM;
 
        if ($n < $total) {
                my $qs = $q->qs_html(o => $n);
-               $rv .= qq{, <a\nhref="?$qs"\nrel=next>next</a>}
+               $rv .= qq{  <a\nhref="?$qs"\nrel=next>next</a>}
        }
        if ($o > 0) {
-               $rv .= $n < $total ? '/' : ',      ';
+               $rv .= $n < $total ? '/' : '       ';
                my $p = $o - $LIM;
                my $qs = $q->qs_html(o => ($p > 0 ? $p : 0));
                $rv .= qq{<a\nhref="?$qs"\nrel=prev>prev</a>};
@@ -161,6 +167,15 @@ sub search_nav_bot {
        $rv .= '</pre>';
 }
 
+sub sort_relevance {
+       my ($pct) = @_;
+       sub {
+               [ sort { (eval { $pct->{$b->topmost->{id}} } || 0)
+                               <=>
+                       (eval { $pct->{$a->topmost->{id}} } || 0)
+       } @{$_[0]} ] };
+}
+
 sub mset_thread {
        my ($ctx, $mset, $q) = @_;
        my %pct;
@@ -170,19 +185,9 @@ sub mset_thread {
                $pct{$smsg->mid} = $i->get_percent;
                $smsg;
        } ($mset->items) ]});
-
-       my $th = PublicInbox::SearchThread->new($msgs);
-       $th->thread;
-       if ($q->{r}) { # order by relevance
-               $th->order(sub {
-                       [ sort { (eval { $pct{$b->topmost->{id}} } || 0)
-                                       <=>
-                               (eval { $pct{$a->topmost->{id}} } || 0)
-                       } @{$_[0]} ];
-               });
-       } else { # order by time (default for threaded view)
-               $th->order(*PublicInbox::View::sort_ts);
-       }
+       my $r = $q->{r};
+       my $rootset = PublicInbox::SearchThread::thread($msgs,
+               $r ? sort_relevance(\%pct) : *PublicInbox::View::sort_ts);
        my $skel = search_nav_bot($mset, $q). "<pre>";
        my $inbox = $ctx->{-inbox};
        $ctx->{-upfx} = '';
@@ -196,17 +201,17 @@ sub mset_thread {
        $ctx->{seen} = {};
        $ctx->{s_nr} = scalar(@$msgs).'+ results';
 
-       PublicInbox::View::walk_thread($th, $ctx,
+       PublicInbox::View::walk_thread($rootset, $ctx,
                *PublicInbox::View::pre_thread);
-
+       @$msgs = reverse @$msgs if $r;
        my $mime;
        sub {
                return unless $msgs;
-               while ($mime = shift @$msgs) {
+               while ($mime = pop @$msgs) {
                        $mime = $inbox->msg_by_smsg($mime) and last;
                }
                if ($mime) {
-                       $mime = Email::MIME->new($mime);
+                       $mime = PublicInbox::MIME->new($mime);
                        return PublicInbox::View::index_entry($mime, $ctx,
                                scalar @$msgs);
                }
@@ -217,7 +222,9 @@ sub mset_thread {
 
 sub ctx_prepare {
        my ($q, $ctx) = @_;
-       my $qh = ascii_html($q->{'q'});
+       my $qh = $q->{'q'};
+       utf8::decode($qh);
+       $qh = ascii_html($qh);
        $ctx->{-q_value_html} = $qh;
        $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef);
        $ctx->{-title_html} = "$qh - search results";
@@ -240,7 +247,7 @@ sub adump {
                while (my $x = shift @items) {
                        $x = load_doc_retry($srch, $x);
                        $x = $ibx->msg_by_smsg($x) and
-                                       return Email::MIME->new($x);
+                                       return PublicInbox::MIME->new($x);
                }
                return undef;
        });
@@ -249,8 +256,9 @@ sub adump {
 package PublicInbox::SearchQuery;
 use strict;
 use warnings;
+use URI::Escape qw(uri_escape);
 use PublicInbox::Hval;
-use PublicInbox::MID qw(mid_escape);
+use PublicInbox::MID qw(MID_ESC);
 
 sub new {
        my ($class, $qp) = @_;
@@ -275,7 +283,7 @@ sub qs_html {
                $self = $tmp;
        }
 
-       my $q = mid_escape($self->{'q'});
+       my $q = uri_escape($self->{'q'}, MID_ESC);
        $q =~ s/%20/+/g; # improve URL readability
        my $qs = "q=$q";