1 # Copyright (C) 2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
4 # Displays search results for the web interface
5 package PublicInbox::SearchView;
8 use URI::Escape qw(uri_unescape uri_escape);
9 use PublicInbox::SearchMsg;
10 use PublicInbox::Hval qw/ascii_html obfuscate_addrs/;
11 use PublicInbox::View;
12 use PublicInbox::WwwAtomStream;
13 use PublicInbox::MID qw(mid2path mid_mime mid_clean mid_escape MID_ESC);
14 use PublicInbox::MIME;
15 require PublicInbox::Git;
16 require PublicInbox::SearchThread;
23 my $q = PublicInbox::SearchQuery->new($ctx->{qp});
26 # double the limit for expanded views:
33 my ($mset, $total, $err, $cb);
36 $mset = $ctx->{srch}->query($q->{'q'}, $opts);
37 $total = $mset->get_matches_estimated;
40 ctx_prepare($q, $ctx);
43 $ctx->{-html_tip} = '<pre>'.err_txt($ctx, $err).'</pre><hr>';
45 } elsif ($total == 0) {
46 if (defined($ctx->{-uxs_retried})) {
48 $q->{'q'} = $ctx->{-uxs_retried};
49 } elsif (index($q->{'q'}, '%') >= 0) {
50 $ctx->{-uxs_retried} = $q->{'q'};
51 $q->{'q'} = uri_unescape($q->{'q'});
55 $ctx->{-html_tip} = "<pre>\n[No results found]</pre><hr>";
59 return adump($_[0], $mset, $q, $ctx) if $x eq 'A';
61 $ctx->{-html_tip} = search_nav_top($mset, $q, $ctx) . "\n\n";
63 $cb = mset_thread($ctx, $mset, $q);
65 $cb = mset_summary($ctx, $mset, $q);
68 PublicInbox::WwwStream->response($ctx, $code, $cb);
71 # allow undef for individual doc loads...
73 my ($srch, $mitem) = @_;
76 $srch->retry_reopen(sub {
77 PublicInbox::SearchMsg->load_doc($mitem->get_document)
82 # display non-nested search results similar to what users expect from
83 # regular WWW search engines:
85 my ($ctx, $mset, $q) = @_;
87 my $total = $mset->get_matches_estimated;
88 my $pad = length("$total");
90 my $res = \($ctx->{-html_tip});
91 my $srch = $ctx->{srch};
92 my $ibx = $ctx->{-inbox};
93 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
94 foreach my $m ($mset->items) {
95 my $rank = sprintf("%${pad}d", $m->get_rank + 1);
96 my $pct = $m->get_percent;
97 my $smsg = load_doc_retry($srch, $m);
100 $m = "$m ".$m->get_docid . " expired\n";
101 $ctx->{env}->{'psgi.errors'}->print($m);
105 my $s = ascii_html($smsg->subject);
106 my $f = ascii_html($smsg->from_name);
108 obfuscate_addrs($obfs_ibx, $s);
109 obfuscate_addrs($obfs_ibx, $f);
111 my $ts = PublicInbox::View::fmt_ts($smsg->ts);
112 my $mid = PublicInbox::Hval->new_msgid($smsg->mid)->{href};
113 $$res .= qq{$rank. <b><a\nhref="$mid/">}.
115 $$res .= "$pfx - by $f @ $ts UTC [$pct%]\n\n";
117 $$res .= search_nav_bot($mset, $q);
122 my ($ctx, $err) = @_;
123 my $u = $ctx->{-inbox}->base_url($ctx->{env}) . '_/text/help/';
124 $err =~ s/^\s*Exception:\s*//; # bad word to show users :P
125 $err = ascii_html($err);
126 "\nBad query: <b>$err</b>\n" .
127 qq{See <a\nhref="$u">$u</a> for help on using search};
131 my ($mset, $q, $ctx) = @_;
134 my $initial_q = $ctx->{-uxs_retried};
135 if (defined $initial_q) {
136 my $rewritten = $q->{'q'};
137 utf8::decode($initial_q);
138 utf8::decode($rewritten);
139 $initial_q = ascii_html($initial_q);
140 $rewritten = ascii_html($rewritten);
141 $rv .= " Warning: Initial query:\n <b>$initial_q</b>\n";
142 $rv .= " returned no results, used:\n";
143 $rv .= " <b>$rewritten</b>\n instead\n\n";
146 $rv .= 'Search results ordered by [';
148 my $d = $q->qs_html(r => 0);
149 $rv .= qq{<a\nhref="?$d">date</a>|<b>relevance</b>};
151 my $d = $q->qs_html(r => 1);
152 $rv .= qq{<b>date</b>|<a\nhref="?$d">relevance</a>};
159 my $t = $q->qs_html(x => 't');
160 $rv .= qq{<b>summary</b>|<a\nhref="?$t">nested</a>}
161 } elsif ($q->{x} eq 't') {
162 my $s = $q->qs_html(x => '');
163 $rv .= qq{<a\nhref="?$s">summary</a>|<b>nested</b>};
165 my $A = $q->qs_html(x => 'A', r => undef);
166 $rv .= qq{|<a\nhref="?$A">Atom feed</a>]};
171 my $total = $mset->get_matches_estimated;
172 my $nr = scalar $mset->items;
176 my $rv = '</pre><hr><pre id=t>';
178 $rv .= "Results $beg-$end of $total";
179 $rv .= ' (estimated)' if $end != $total;
181 $rv .= "No more results, only $total";
186 my $qs = $q->qs_html(o => $n);
187 $rv .= qq{ <a\nhref="?$qs"\nrel=next>next</a>}
190 $rv .= $n < $total ? '/' : ' ';
192 my $qs = $q->qs_html(o => ($p > 0 ? $p : 0));
193 $rv .= qq{<a\nhref="?$qs"\nrel=prev>prev</a>};
201 [ sort { (eval { $pct->{$b->topmost->{id}} } || 0)
203 (eval { $pct->{$a->topmost->{id}} } || 0)
208 my ($ctx, $mset, $q) = @_;
210 my $srch = $ctx->{srch};
211 my $msgs = $srch->retry_reopen(sub { [ map {
213 my $smsg = PublicInbox::SearchMsg->load_doc($i->get_document);
214 $pct{$smsg->mid} = $i->get_percent;
216 } ($mset->items) ]});
218 my $rootset = PublicInbox::SearchThread::thread($msgs,
219 $r ? sort_relevance(\%pct) : *PublicInbox::View::sort_ts,
221 my $skel = search_nav_bot($mset, $q). "<pre>";
222 my $inbox = $ctx->{-inbox};
224 $ctx->{anchor_idx} = 1;
225 $ctx->{cur_level} = 0;
226 $ctx->{dst} = \$skel;
227 $ctx->{mapping} = {};
229 $ctx->{prev_attr} = '';
230 $ctx->{prev_level} = 0;
232 $ctx->{s_nr} = scalar(@$msgs).'+ results';
234 # reduce hash lookups in skel_dump
235 $ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate};
236 PublicInbox::View::walk_thread($rootset, $ctx,
237 *PublicInbox::View::pre_thread);
239 @$msgs = reverse @$msgs if $r;
243 while ($mime = pop @$msgs) {
244 $mime = $inbox->msg_by_smsg($mime) and last;
247 $mime = PublicInbox::MIME->new($mime);
248 return PublicInbox::View::index_entry($mime, $ctx,
260 $qh = ascii_html($qh);
261 $ctx->{-q_value_html} = $qh;
262 $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef);
263 $ctx->{-title_html} = "$qh - search results";
265 $extra .= qq{<input\ntype=hidden\nname=r />} if $q->{r};
266 if (my $x = $q->{x}) {
268 $extra .= qq{<input\ntype=hidden\nname=x\nvalue="$x" />};
270 $ctx->{-extra_form_html} = $extra;
274 my ($cb, $mset, $q, $ctx) = @_;
275 my $ibx = $ctx->{-inbox};
276 my @items = $mset->items;
277 $ctx->{search_query} = $q;
278 my $srch = $ctx->{srch};
279 PublicInbox::WwwAtomStream->response($ctx, 200, sub {
280 while (my $x = shift @items) {
281 $x = load_doc_retry($srch, $x);
282 $x = $ibx->msg_by_smsg($x) and
283 return PublicInbox::MIME->new($x);
289 package PublicInbox::SearchQuery;
292 use URI::Escape qw(uri_escape);
293 use PublicInbox::Hval;
294 use PublicInbox::MID qw(MID_ESC);
297 my ($class, $qp) = @_;
303 o => (($qp->{o} || '0') =~ /(\d+)/),
304 r => (defined $r && $r ne '0'),
309 my ($self, %over) = @_;
312 my $tmp = bless { %$self }, ref($self);
313 foreach my $k (keys %over) {
314 $tmp->{$k} = $over{$k};
319 my $q = uri_escape($self->{'q'}, MID_ESC);
320 $q =~ s/%20/+/g; # improve URL readability
323 if (my $o = $self->{o}) { # ignore o == 0
326 if (my $r = $self->{r}) {
329 if (my $x = $self->{x}) {
330 $qs .= "&x=$x" if ($x eq 't' || $x eq 'A');