1 # Copyright (C) 2015-2018 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <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(MID_ESC);
14 use PublicInbox::MIME;
15 require PublicInbox::Git;
16 require PublicInbox::SearchThread;
23 my $q = PublicInbox::SearchQuery->new($ctx->{qp});
25 require PublicInbox::Mbox;
26 return PublicInbox::Mbox::mbox_all($ctx, $q->{'q'}) if $x eq 'm';
32 my $srch = $ctx->{-inbox}->search or
33 return PublicInbox::WWW::need($ctx, 'Search');
34 my $q = PublicInbox::SearchQuery->new($ctx->{qp});
36 my $query = $q->{'q'};
39 # double the limit for expanded views:
46 my ($mset, $total, $err, $cb);
49 $mset = $srch->query($query, $opts);
50 $total = $mset->get_matches_estimated;
53 ctx_prepare($q, $ctx);
56 $ctx->{-html_tip} = '<pre>'.err_txt($ctx, $err).'</pre><hr>';
58 } elsif ($total == 0) {
59 if (defined($ctx->{-uxs_retried})) {
61 $q->{'q'} = $ctx->{-uxs_retried};
62 } elsif (index($q->{'q'}, '%') >= 0) {
63 $ctx->{-uxs_retried} = $q->{'q'};
64 $q->{'q'} = uri_unescape($q->{'q'});
68 $ctx->{-html_tip} = "<pre>\n[No results found]</pre><hr>";
71 return adump($_[0], $mset, $q, $ctx) if $x eq 'A';
73 $ctx->{-html_tip} = search_nav_top($mset, $q, $ctx);
75 $cb = mset_thread($ctx, $mset, $q);
77 $cb = mset_summary($ctx, $mset, $q);
80 PublicInbox::WwwStream->response($ctx, $code, $cb);
83 # allow undef for individual doc loads...
85 my ($srch, $mitem) = @_;
88 $srch->retry_reopen(sub {
89 PublicInbox::SearchMsg->load_doc($mitem->get_document)
94 # display non-nested search results similar to what users expect from
95 # regular WWW search engines:
97 my ($ctx, $mset, $q) = @_;
99 my $total = $mset->get_matches_estimated;
100 my $pad = length("$total");
101 my $pfx = ' ' x $pad;
102 my $res = \($ctx->{-html_tip});
103 my $ibx = $ctx->{-inbox};
104 my $srch = $ibx->search;
105 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
106 foreach my $m ($mset->items) {
107 my $rank = sprintf("%${pad}d", $m->get_rank + 1);
108 my $pct = $m->get_percent;
109 my $smsg = load_doc_retry($srch, $m);
112 $m = "$m ".$m->get_docid . " expired\n";
113 $ctx->{env}->{'psgi.errors'}->print($m);
117 my $s = ascii_html($smsg->subject);
118 my $f = ascii_html($smsg->from_name);
120 obfuscate_addrs($obfs_ibx, $s);
121 obfuscate_addrs($obfs_ibx, $f);
123 my $date = PublicInbox::View::fmt_ts($smsg->ds);
124 my $mid = PublicInbox::Hval->new_msgid($smsg->mid)->{href};
125 $s = '(no subject)' if $s eq '';
126 $$res .= qq{$rank. <b><a\nhref="$mid/">}.
128 $$res .= "$pfx - by $f @ $date UTC [$pct%]\n\n";
130 $$res .= search_nav_bot($mset, $q);
135 my ($ctx, $err) = @_;
136 my $u = $ctx->{-inbox}->base_url($ctx->{env}) . '_/text/help/';
137 $err =~ s/^\s*Exception:\s*//; # bad word to show users :P
138 $err = ascii_html($err);
139 "\nBad query: <b>$err</b>\n" .
140 qq{See <a\nhref="$u">$u</a> for help on using search};
144 my ($mset, $q, $ctx) = @_;
145 my $m = $q->qs_html(x => 'm', r => undef);
146 my $rv = qq{<form\naction="?$m"\nmethod="post"><pre>};
147 my $initial_q = $ctx->{-uxs_retried};
148 if (defined $initial_q) {
149 my $rewritten = $q->{'q'};
150 utf8::decode($initial_q);
151 utf8::decode($rewritten);
152 $initial_q = ascii_html($initial_q);
153 $rewritten = ascii_html($rewritten);
154 $rv .= " Warning: Initial query:\n <b>$initial_q</b>\n";
155 $rv .= " returned no results, used:\n";
156 $rv .= " <b>$rewritten</b>\n instead\n\n";
159 $rv .= 'Search results ordered by [';
161 my $d = $q->qs_html(r => 0);
162 $rv .= qq{<a\nhref="?$d">date</a>|<b>relevance</b>};
164 my $d = $q->qs_html(r => 1);
165 $rv .= qq{<b>date</b>|<a\nhref="?$d">relevance</a>};
172 my $t = $q->qs_html(x => 't');
173 $rv .= qq{<b>summary</b>|<a\nhref="?$t">nested</a>}
174 } elsif ($q->{x} eq 't') {
175 my $s = $q->qs_html(x => '');
176 $rv .= qq{<a\nhref="?$s">summary</a>|<b>nested</b>};
178 my $A = $q->qs_html(x => 'A', r => undef);
179 $rv .= qq{|<a\nhref="?$A">Atom feed</a>]};
180 $rv .= qq{\n\t\t\t\t\t\tdownload: };
181 $rv .= qq{<input\ntype=submit\nvalue="mbox.gz"/></pre></form><pre>};
186 my $total = $mset->get_matches_estimated;
189 my $end = $o + $mset->size;
191 my $rv = '</pre><hr><pre id=t>';
193 $rv .= "Results $beg-$end of $total";
194 $rv .= ' (estimated)' if $end != $total;
196 $rv .= "No more results, only $total";
201 my $qs = $q->qs_html(o => $n, l => $l);
202 $rv .= qq{ <a\nhref="?$qs"\nrel=next>next</a>}
205 $rv .= $n < $total ? '/' : ' ';
207 my $qs = $q->qs_html(o => ($p > 0 ? $p : 0));
208 $rv .= qq{<a\nhref="?$qs"\nrel=prev>prev</a>};
216 [ sort { (eval { $pct->{$b->topmost->{id}} } || 0)
218 (eval { $pct->{$a->topmost->{id}} } || 0)
223 my ($ctx, $mset, $q) = @_;
225 my $ibx = $ctx->{-inbox};
226 my $msgs = $ibx->search->retry_reopen(sub { [ map {
228 my $smsg = PublicInbox::SearchMsg->load_doc($i->get_document);
229 $pct{$smsg->mid} = $i->get_percent;
231 } ($mset->items) ]});
233 my $rootset = PublicInbox::SearchThread::thread($msgs,
234 $r ? sort_relevance(\%pct) : *PublicInbox::View::sort_ds,
236 my $skel = search_nav_bot($mset, $q). "<pre>";
238 $ctx->{anchor_idx} = 1;
239 $ctx->{cur_level} = 0;
240 $ctx->{dst} = \$skel;
241 $ctx->{mapping} = {};
243 $ctx->{prev_attr} = '';
244 $ctx->{prev_level} = 0;
245 $ctx->{s_nr} = scalar(@$msgs).'+ results';
247 # reduce hash lookups in skel_dump
248 $ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate};
249 PublicInbox::View::walk_thread($rootset, $ctx,
250 *PublicInbox::View::pre_thread);
252 @$msgs = reverse @$msgs if $r;
256 while (my $m = pop @$msgs) {
257 $smsg = $ibx->smsg_mime($m) and last;
260 return PublicInbox::View::index_entry($smsg, $ctx,
272 $qh = ascii_html($qh);
273 $ctx->{-q_value_html} = $qh;
274 $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef);
275 $ctx->{-title_html} = "$qh - search results";
277 $extra .= qq{<input\ntype=hidden\nname=r />} if $q->{r};
278 if (my $x = $q->{x}) {
280 $extra .= qq{<input\ntype=hidden\nname=x\nvalue="$x" />};
282 $ctx->{-extra_form_html} = $extra;
286 my ($cb, $mset, $q, $ctx) = @_;
287 my $ibx = $ctx->{-inbox};
288 my @items = $mset->items;
289 $ctx->{search_query} = $q;
290 my $srch = $ibx->search;
291 PublicInbox::WwwAtomStream->response($ctx, 200, sub {
292 while (my $x = shift @items) {
293 $x = load_doc_retry($srch, $x);
294 $x = $ibx->smsg_mime($x) and return $x;
300 package PublicInbox::SearchQuery;
303 use URI::Escape qw(uri_escape);
304 use PublicInbox::Hval;
305 use PublicInbox::MID qw(MID_ESC);
308 my ($class, $qp) = @_;
311 my ($l) = (($qp->{l} || '') =~ /(\d+)/);
312 $l = $LIM if !$l || $l > $LIM;
316 o => (($qp->{o} || '0') =~ /(\d+)/),
318 r => (defined $r && $r ne '0'),
323 my ($self, %over) = @_;
326 my $tmp = bless { %$self }, ref($self);
327 foreach my $k (keys %over) {
328 $tmp->{$k} = $over{$k};
333 my $q = uri_escape($self->{'q'}, MID_ESC);
334 $q =~ s/%20/+/g; # improve URL readability
337 if (my $o = $self->{o}) { # ignore o == 0
340 if (my $l = $self->{l}) {
341 $qs .= "&l=$l" unless $l == $LIM;
343 if (my $r = $self->{r}) {
346 if (my $x = $self->{x}) {
347 $qs .= "&x=$x" if ($x eq 't' || $x eq 'A' || $x eq 'm');