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});
25 my $query = $q->{'q'};
26 return PublicInbox::Mbox::mbox_all($ctx, $query) if $x eq 'm';
29 # double the limit for expanded views:
36 my ($mset, $total, $err, $cb);
39 $mset = $ctx->{srch}->query($query, $opts);
40 $total = $mset->get_matches_estimated;
43 ctx_prepare($q, $ctx);
46 $ctx->{-html_tip} = '<pre>'.err_txt($ctx, $err).'</pre><hr>';
48 } elsif ($total == 0) {
49 if (defined($ctx->{-uxs_retried})) {
51 $q->{'q'} = $ctx->{-uxs_retried};
52 } elsif (index($q->{'q'}, '%') >= 0) {
53 $ctx->{-uxs_retried} = $q->{'q'};
54 $q->{'q'} = uri_unescape($q->{'q'});
58 $ctx->{-html_tip} = "<pre>\n[No results found]</pre><hr>";
61 return adump($_[0], $mset, $q, $ctx) if $x eq 'A';
63 $ctx->{-html_tip} = search_nav_top($mset, $q, $ctx) . "\n\n";
65 $cb = mset_thread($ctx, $mset, $q);
67 $cb = mset_summary($ctx, $mset, $q);
70 PublicInbox::WwwStream->response($ctx, $code, $cb);
73 # allow undef for individual doc loads...
75 my ($srch, $mitem) = @_;
78 $srch->retry_reopen(sub {
79 PublicInbox::SearchMsg->load_doc($mitem->get_document)
84 # display non-nested search results similar to what users expect from
85 # regular WWW search engines:
87 my ($ctx, $mset, $q) = @_;
89 my $total = $mset->get_matches_estimated;
90 my $pad = length("$total");
92 my $res = \($ctx->{-html_tip});
93 my $srch = $ctx->{srch};
94 my $ibx = $ctx->{-inbox};
95 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
96 foreach my $m ($mset->items) {
97 my $rank = sprintf("%${pad}d", $m->get_rank + 1);
98 my $pct = $m->get_percent;
99 my $smsg = load_doc_retry($srch, $m);
102 $m = "$m ".$m->get_docid . " expired\n";
103 $ctx->{env}->{'psgi.errors'}->print($m);
107 my $s = ascii_html($smsg->subject);
108 my $f = ascii_html($smsg->from_name);
110 obfuscate_addrs($obfs_ibx, $s);
111 obfuscate_addrs($obfs_ibx, $f);
113 my $ts = PublicInbox::View::fmt_ts($smsg->ts);
114 my $mid = PublicInbox::Hval->new_msgid($smsg->mid)->{href};
115 $$res .= qq{$rank. <b><a\nhref="$mid/">}.
117 $$res .= "$pfx - by $f @ $ts UTC [$pct%]\n\n";
119 $$res .= search_nav_bot($mset, $q);
124 my ($ctx, $err) = @_;
125 my $u = $ctx->{-inbox}->base_url($ctx->{env}) . '_/text/help/';
126 $err =~ s/^\s*Exception:\s*//; # bad word to show users :P
127 $err = ascii_html($err);
128 "\nBad query: <b>$err</b>\n" .
129 qq{See <a\nhref="$u">$u</a> for help on using search};
133 my ($mset, $q, $ctx) = @_;
136 my $initial_q = $ctx->{-uxs_retried};
137 if (defined $initial_q) {
138 my $rewritten = $q->{'q'};
139 utf8::decode($initial_q);
140 utf8::decode($rewritten);
141 $initial_q = ascii_html($initial_q);
142 $rewritten = ascii_html($rewritten);
143 $rv .= " Warning: Initial query:\n <b>$initial_q</b>\n";
144 $rv .= " returned no results, used:\n";
145 $rv .= " <b>$rewritten</b>\n instead\n\n";
148 $rv .= 'Search results ordered by [';
150 my $d = $q->qs_html(r => 0);
151 $rv .= qq{<a\nhref="?$d">date</a>|<b>relevance</b>};
153 my $d = $q->qs_html(r => 1);
154 $rv .= qq{<b>date</b>|<a\nhref="?$d">relevance</a>};
161 my $t = $q->qs_html(x => 't');
162 $rv .= qq{<b>summary</b>|<a\nhref="?$t">nested</a>}
163 } elsif ($q->{x} eq 't') {
164 my $s = $q->qs_html(x => '');
165 $rv .= qq{<a\nhref="?$s">summary</a>|<b>nested</b>};
167 my $A = $q->qs_html(x => 'A', r => undef);
168 $rv .= qq{|<a\nhref="?$A">Atom feed</a>]};
169 my $m = $q->qs_html(x => 'm', r => undef);
171 $rv .= qq{\n\t\t\t\t\t\tdownload: };
172 $rv .= qq{<a\nhref="?$m"\nrel="nofollow">mbox.gz</a>};
177 my $total = $mset->get_matches_estimated;
178 my $nr = scalar $mset->items;
182 my $rv = '</pre><hr><pre id=t>';
184 $rv .= "Results $beg-$end of $total";
185 $rv .= ' (estimated)' if $end != $total;
187 $rv .= "No more results, only $total";
192 my $qs = $q->qs_html(o => $n);
193 $rv .= qq{ <a\nhref="?$qs"\nrel=next>next</a>}
196 $rv .= $n < $total ? '/' : ' ';
198 my $qs = $q->qs_html(o => ($p > 0 ? $p : 0));
199 $rv .= qq{<a\nhref="?$qs"\nrel=prev>prev</a>};
207 [ sort { (eval { $pct->{$b->topmost->{id}} } || 0)
209 (eval { $pct->{$a->topmost->{id}} } || 0)
214 my ($ctx, $mset, $q) = @_;
216 my $srch = $ctx->{srch};
217 my $msgs = $srch->retry_reopen(sub { [ map {
219 my $smsg = PublicInbox::SearchMsg->load_doc($i->get_document);
220 $pct{$smsg->mid} = $i->get_percent;
222 } ($mset->items) ]});
224 my $rootset = PublicInbox::SearchThread::thread($msgs,
225 $r ? sort_relevance(\%pct) : *PublicInbox::View::sort_ts,
227 my $skel = search_nav_bot($mset, $q). "<pre>";
228 my $inbox = $ctx->{-inbox};
230 $ctx->{anchor_idx} = 1;
231 $ctx->{cur_level} = 0;
232 $ctx->{dst} = \$skel;
233 $ctx->{mapping} = {};
235 $ctx->{prev_attr} = '';
236 $ctx->{prev_level} = 0;
238 $ctx->{s_nr} = scalar(@$msgs).'+ results';
240 # reduce hash lookups in skel_dump
241 $ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate};
242 PublicInbox::View::walk_thread($rootset, $ctx,
243 *PublicInbox::View::pre_thread);
245 @$msgs = reverse @$msgs if $r;
249 while ($mime = pop @$msgs) {
250 $mime = $inbox->msg_by_smsg($mime) and last;
253 $mime = PublicInbox::MIME->new($mime);
254 return PublicInbox::View::index_entry($mime, $ctx,
266 $qh = ascii_html($qh);
267 $ctx->{-q_value_html} = $qh;
268 $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef);
269 $ctx->{-title_html} = "$qh - search results";
271 $extra .= qq{<input\ntype=hidden\nname=r />} if $q->{r};
272 if (my $x = $q->{x}) {
274 $extra .= qq{<input\ntype=hidden\nname=x\nvalue="$x" />};
276 $ctx->{-extra_form_html} = $extra;
280 my ($cb, $mset, $q, $ctx) = @_;
281 my $ibx = $ctx->{-inbox};
282 my @items = $mset->items;
283 $ctx->{search_query} = $q;
284 my $srch = $ctx->{srch};
285 PublicInbox::WwwAtomStream->response($ctx, 200, sub {
286 while (my $x = shift @items) {
287 $x = load_doc_retry($srch, $x);
288 $x = $ibx->msg_by_smsg($x) and
289 return PublicInbox::MIME->new($x);
295 package PublicInbox::SearchQuery;
298 use URI::Escape qw(uri_escape);
299 use PublicInbox::Hval;
300 use PublicInbox::MID qw(MID_ESC);
303 my ($class, $qp) = @_;
309 o => (($qp->{o} || '0') =~ /(\d+)/),
310 r => (defined $r && $r ne '0'),
315 my ($self, %over) = @_;
318 my $tmp = bless { %$self }, ref($self);
319 foreach my $k (keys %over) {
320 $tmp->{$k} = $over{$k};
325 my $q = uri_escape($self->{'q'}, MID_ESC);
326 $q =~ s/%20/+/g; # improve URL readability
329 if (my $o = $self->{o}) { # ignore o == 0
332 if (my $r = $self->{r}) {
335 if (my $x = $self->{x}) {
336 $qs .= "&x=$x" if ($x eq 't' || $x eq 'A' || $x eq 'm');