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-threaded 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 $obfs = $ctx->{-inbox}->{obfuscate};
93 foreach my $m ($mset->items) {
94 my $rank = sprintf("%${pad}d", $m->get_rank + 1);
95 my $pct = $m->get_percent;
96 my $smsg = load_doc_retry($srch, $m);
99 $m = "$m ".$m->get_docid . " expired\n";
100 $ctx->{env}->{'psgi.errors'}->print($m);
104 my $s = ascii_html($smsg->subject);
105 my $f = ascii_html($smsg->from_name);
110 my $ts = PublicInbox::View::fmt_ts($smsg->ts);
111 my $mid = PublicInbox::Hval->new_msgid($smsg->mid)->{href};
112 $$res .= qq{$rank. <b><a\nhref="$mid/">}.
114 $$res .= "$pfx - by $f @ $ts UTC [$pct%]\n\n";
116 $$res .= search_nav_bot($mset, $q);
121 my ($ctx, $err) = @_;
122 my $u = $ctx->{-inbox}->base_url($ctx->{env}) . '_/text/help/';
123 $err =~ s/^\s*Exception:\s*//; # bad word to show users :P
124 $err = ascii_html($err);
125 "\nBad query: <b>$err</b>\n" .
126 qq{See <a\nhref="$u">$u</a> for help on using search};
130 my ($mset, $q, $ctx) = @_;
133 my $initial_q = $ctx->{-uxs_retried};
134 if (defined $initial_q) {
135 my $rewritten = $q->{'q'};
136 utf8::decode($initial_q);
137 utf8::decode($rewritten);
138 $initial_q = ascii_html($initial_q);
139 $rewritten = ascii_html($rewritten);
140 $rv .= " Warning: Initial query:\n <b>$initial_q</b>\n";
141 $rv .= " returned no results, used:\n";
142 $rv .= " <b>$rewritten</b>\n instead\n\n";
145 $rv .= 'Search results ordered by [';
147 my $d = $q->qs_html(r => 0);
148 $rv .= qq{<a\nhref="?$d">date</a>|<b>relevance</b>};
150 my $d = $q->qs_html(r => 1);
151 $rv .= qq{<b>date</b>|<a\nhref="?$d">relevance</a>};
158 my $t = $q->qs_html(x => 't');
159 $rv .= qq{<b>summary</b>|<a\nhref="?$t">threaded</a>}
160 } elsif ($q->{x} eq 't') {
161 my $s = $q->qs_html(x => '');
162 $rv .= qq{<a\nhref="?$s">summary</a>|<b>threaded</b>};
164 my $A = $q->qs_html(x => 'A', r => undef);
165 $rv .= qq{|<a\nhref="?$A">Atom feed</a>]};
170 my $total = $mset->get_matches_estimated;
171 my $nr = scalar $mset->items;
175 my $rv = '</pre><hr><pre id=t>';
177 $rv .= "Results $beg-$end of $total";
178 $rv .= ' (estimated)' if $end != $total;
180 $rv .= "No more results, only $total";
185 my $qs = $q->qs_html(o => $n);
186 $rv .= qq{ <a\nhref="?$qs"\nrel=next>next</a>}
189 $rv .= $n < $total ? '/' : ' ';
191 my $qs = $q->qs_html(o => ($p > 0 ? $p : 0));
192 $rv .= qq{<a\nhref="?$qs"\nrel=prev>prev</a>};
200 [ sort { (eval { $pct->{$b->topmost->{id}} } || 0)
202 (eval { $pct->{$a->topmost->{id}} } || 0)
207 my ($ctx, $mset, $q) = @_;
209 my $msgs = $ctx->{srch}->retry_reopen(sub { [ map {
211 my $smsg = PublicInbox::SearchMsg->load_doc($i->get_document);
212 $pct{$smsg->mid} = $i->get_percent;
214 } ($mset->items) ]});
216 my $rootset = PublicInbox::SearchThread::thread($msgs,
217 $r ? sort_relevance(\%pct) : *PublicInbox::View::sort_ts);
218 my $skel = search_nav_bot($mset, $q). "<pre>";
219 my $inbox = $ctx->{-inbox};
221 $ctx->{anchor_idx} = 1;
222 $ctx->{cur_level} = 0;
223 $ctx->{dst} = \$skel;
224 $ctx->{mapping} = {};
226 $ctx->{prev_attr} = '';
227 $ctx->{prev_level} = 0;
229 $ctx->{s_nr} = scalar(@$msgs).'+ results';
231 # reduce hash lookups in skel_dump
232 $ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate};
233 PublicInbox::View::walk_thread($rootset, $ctx,
234 *PublicInbox::View::pre_thread);
236 @$msgs = reverse @$msgs if $r;
240 while ($mime = pop @$msgs) {
241 $mime = $inbox->msg_by_smsg($mime) and last;
244 $mime = PublicInbox::MIME->new($mime);
245 return PublicInbox::View::index_entry($mime, $ctx,
257 $qh = ascii_html($qh);
258 $ctx->{-q_value_html} = $qh;
259 $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef);
260 $ctx->{-title_html} = "$qh - search results";
262 $extra .= qq{<input\ntype=hidden\nname=r />} if $q->{r};
263 if (my $x = $q->{x}) {
265 $extra .= qq{<input\ntype=hidden\nname=x\nvalue="$x" />};
267 $ctx->{-extra_form_html} = $extra;
271 my ($cb, $mset, $q, $ctx) = @_;
272 my $ibx = $ctx->{-inbox};
273 my @items = $mset->items;
274 $ctx->{search_query} = $q;
275 my $srch = $ctx->{srch};
276 PublicInbox::WwwAtomStream->response($ctx, 200, sub {
277 while (my $x = shift @items) {
278 $x = load_doc_retry($srch, $x);
279 $x = $ibx->msg_by_smsg($x) and
280 return PublicInbox::MIME->new($x);
286 package PublicInbox::SearchQuery;
289 use URI::Escape qw(uri_escape);
290 use PublicInbox::Hval;
291 use PublicInbox::MID qw(MID_ESC);
294 my ($class, $qp) = @_;
300 o => (($qp->{o} || '0') =~ /(\d+)/),
301 r => (defined $r && $r ne '0'),
306 my ($self, %over) = @_;
309 my $tmp = bless { %$self }, ref($self);
310 foreach my $k (keys %over) {
311 $tmp->{$k} = $over{$k};
316 my $q = uri_escape($self->{'q'}, MID_ESC);
317 $q =~ s/%20/+/g; # improve URL readability
320 if (my $o = $self->{o}) { # ignore o == 0
323 if (my $r = $self->{r}) {
326 if (my $x = $self->{x}) {
327 $qs .= "&x=$x" if ($x eq 't' || $x eq 'A');