1 # Copyright (C) 2015-2020 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);
10 use PublicInbox::Hval qw(ascii_html obfuscate_addrs mid_href);
11 use PublicInbox::View;
12 use PublicInbox::WwwAtomStream;
13 use PublicInbox::WwwStream qw(html_oneshot);
14 use PublicInbox::SearchThread;
15 use PublicInbox::SearchQuery;
16 use PublicInbox::Search qw(mdocid);
21 my $q = PublicInbox::SearchQuery->new($ctx->{qp});
23 require PublicInbox::Mbox;
24 return PublicInbox::Mbox::mbox_all($ctx, $q->{'q'}) if $x eq 'm';
30 my $srch = $ctx->{-inbox}->search or
31 return PublicInbox::WWW::need($ctx, 'Search');
32 my $q = PublicInbox::SearchQuery->new($ctx->{qp});
34 my $query = $q->{'q'};
39 $o = -($o + 1); # so [-1] is the last element, like Perl lists
43 # double the limit for expanded views:
51 my ($mset, $total, $err, $html);
54 $mset = $srch->query($query, $opts);
55 $total = $mset->get_matches_estimated;
58 ctx_prepare($q, $ctx);
61 $html = '<pre>'.err_txt($ctx, $err).'</pre><hr>';
62 } elsif ($total == 0) {
63 if (defined($ctx->{-uxs_retried})) {
65 $q->{'q'} = $ctx->{-uxs_retried};
66 } elsif (index($q->{'q'}, '%') >= 0) {
67 $ctx->{-uxs_retried} = $q->{'q'};
68 $q->{'q'} = uri_unescape($q->{'q'});
72 $html = "<pre>\n[No results found]</pre><hr>";
74 return adump($_[0], $mset, $q, $ctx) if $x eq 'A';
76 $ctx->{-html_tip} = search_nav_top($mset, $q, $ctx);
77 return mset_thread($ctx, $mset, $q) if $x eq 't';
78 mset_summary($ctx, $mset, $q); # appends to {-html_tip}
81 html_oneshot($ctx, $code);
84 # display non-nested search results similar to what users expect from
85 # regular WWW search engines:
87 my ($ctx, $mset, $q) = @_;
89 my $total = $mset->get_matches_estimated;
90 my $pad = length("$total");
92 my $res = \($ctx->{-html_tip});
93 my $ibx = $ctx->{-inbox};
94 my $nshard = $ibx->search->{nshard} // 1;
95 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
96 my @nums = map { mdocid($nshard, $_) } $mset->items;
97 my %num2msg = map { $_->{num} => $_ } @{$ibx->over->get_all(@nums)};
99 foreach my $m ($mset->items) {
100 my $rank = sprintf("%${pad}d", $m->get_rank + 1);
101 my $pct = get_pct($m);
102 my $num = shift @nums;
103 my $smsg = delete($num2msg{$num}) or do {
105 $m = "$m $num expired\n";
106 $ctx->{env}->{'psgi.errors'}->print($m);
110 my $s = ascii_html($smsg->{subject});
111 my $f = ascii_html($smsg->{from_name});
113 obfuscate_addrs($obfs_ibx, $s);
114 obfuscate_addrs($obfs_ibx, $f);
116 my $date = PublicInbox::View::fmt_ts($smsg->{ds});
117 my $mid = mid_href($smsg->{mid});
118 $s = '(no subject)' if $s eq '';
119 $$res .= qq{$rank. <b><a\nhref="$mid/">}.
121 $$res .= "$pfx - by $f @ $date UTC [$pct%]\n\n";
123 $$res .= search_nav_bot($mset, $q);
127 # shorten "/full/path/to/Foo/Bar.pm" to "Foo/Bar.pm" so error
128 # messages don't reveal FS layout info in case people use non-standard
132 if (my $short = $rmap_inc{$full}) {
134 } elsif (!scalar(keys %rmap_inc) && -e $full) {
135 %rmap_inc = map {; "$INC{$_}" => $_ } keys %INC;
136 # fall back to basename as last resort
137 $rmap_inc{$full} // (split('/', $full))[-1];
144 my ($ctx, $err) = @_;
145 my $u = $ctx->{-inbox}->base_url($ctx->{env}) . '_/text/help/';
146 $err =~ s/^\s*Exception:\s*//; # bad word to show users :P
147 $err =~ s!(\S+)!path2inc($1)!sge;
148 $err = ascii_html($err);
149 "\nBad query: <b>$err</b>\n" .
150 qq{See <a\nhref="$u">$u</a> for help on using search};
154 my ($mset, $q, $ctx) = @_;
155 my $m = $q->qs_html(x => 'm', r => undef);
156 my $rv = qq{<form\naction="?$m"\nmethod="post"><pre>};
157 my $initial_q = $ctx->{-uxs_retried};
158 if (defined $initial_q) {
159 my $rewritten = $q->{'q'};
160 utf8::decode($initial_q);
161 utf8::decode($rewritten);
162 $initial_q = ascii_html($initial_q);
163 $rewritten = ascii_html($rewritten);
164 $rv .= " Warning: Initial query:\n <b>$initial_q</b>\n";
165 $rv .= " returned no results, used:\n";
166 $rv .= " <b>$rewritten</b>\n instead\n\n";
169 $rv .= 'Search results ordered by [';
171 my $d = $q->qs_html(r => 0);
172 $rv .= qq{<a\nhref="?$d">date</a>|<b>relevance</b>};
174 my $d = $q->qs_html(r => 1);
175 $rv .= qq{<b>date</b>|<a\nhref="?$d">relevance</a>};
182 my $t = $q->qs_html(x => 't');
183 $rv .= qq{<b>summary</b>|<a\nhref="?$t">nested</a>}
184 } elsif ($q->{x} eq 't') {
185 my $s = $q->qs_html(x => '');
186 $rv .= qq{<a\nhref="?$s">summary</a>|<b>nested</b>};
188 my $A = $q->qs_html(x => 'A', r => undef);
189 $rv .= qq{|<a\nhref="?$A">Atom feed</a>]};
190 $rv .= qq{\n\t\t\t\t\t\tdownload: };
191 $rv .= qq{<input\ntype=submit\nvalue="mbox.gz"/></pre></form><pre>};
196 my $total = $mset->get_matches_estimated;
198 my $rv = '</pre><hr><pre id=t>';
200 my $off = $o < 0 ? -($o + 1) : $o;
201 my $end = $off + $mset->size;
205 $rv .= "Results $beg-$end of $total";
206 $rv .= ' (estimated)' if $end != $total;
208 $rv .= "No more results, only $total";
210 my ($next, $join, $prev);
212 if ($o >= 0) { # sort descending
215 $next = $q->qs_html(o => $n, l => $l);
218 $join = $n < $total ? '/' : ' ';
220 $prev = $q->qs_html(o => ($p > 0 ? $p : 0));
222 } else { # o < 0, sort ascending
226 $next = $q->qs_html(o => $n, l => $l);
229 $join = -$n < $total ? '/' : ' ';
231 $prev = $q->qs_html(o => ($p < 0 ? $p : 0));
235 $rv .= qq{ <a\nhref="?$next"\nrel=next>next</a>} if $next;
236 $rv .= $join if $join;
237 $rv .= qq{<a\nhref="?$prev"\nrel=prev>prev</a>} if $prev;
239 my $rev = $q->qs_html(o => $o < 0 ? 0 : -1);
240 $rv .= qq{ | <a\nhref="?$rev">reverse results</a></pre>};
245 (eval { $b->topmost->{pct} } // 0) <=>
246 (eval { $a->topmost->{pct} } // 0)
251 # Capped at "99%" since "100%" takes an extra column in the
252 # thread skeleton view. <xapian/mset.h> says the value isn't
253 # very meaningful, anyways.
254 my $n = $_[0]->get_percent;
259 my ($ctx, $mset, $q) = @_;
260 my $ibx = $ctx->{-inbox};
261 my $nshard = $ibx->search->{nshard} // 1;
262 my %pct = map { mdocid($nshard, $_) => get_pct($_) } $mset->items;
263 my $msgs = $ibx->over->get_all(keys %pct);
264 $_->{pct} = $pct{$_->{num}} for @$msgs;
266 my $rootset = PublicInbox::SearchThread::thread($msgs,
267 $r ? \&sort_relevance : \&PublicInbox::View::sort_ds,
269 my $skel = search_nav_bot($mset, $q). "<pre>";
271 $ctx->{anchor_idx} = 1;
272 $ctx->{cur_level} = 0;
273 $ctx->{skel} = \$skel;
274 $ctx->{mapping} = {};
275 $ctx->{searchview} = 1;
276 $ctx->{prev_attr} = '';
277 $ctx->{prev_level} = 0;
278 $ctx->{s_nr} = scalar(@$msgs).'+ results';
280 # reduce hash lookups in skel_dump
281 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
282 PublicInbox::View::walk_thread($rootset, $ctx,
283 \&PublicInbox::View::pre_thread);
285 @$msgs = reverse @$msgs if $r;
286 $ctx->{msgs} = $msgs;
287 PublicInbox::WwwStream::aresponse($ctx, 200, \&mset_thread_i);
290 # callback for PublicInbox::WwwStream::getline
292 my ($ctx, $eml) = @_;
293 $ctx->zmore($ctx->html_top) if exists $ctx->{-html_tip};
294 $eml and return PublicInbox::View::eml_entry($ctx, $eml);
295 my $smsg = shift @{$ctx->{msgs}} or
296 $ctx->zmore(${delete($ctx->{skel})});
304 $qh = ascii_html($qh);
305 $ctx->{-q_value_html} = $qh;
306 $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef);
307 $ctx->{-title_html} = "$qh - search results";
309 $extra .= qq{<input\ntype=hidden\nname=r />} if $q->{r};
310 if (my $x = $q->{x}) {
312 $extra .= qq{<input\ntype=hidden\nname=x\nvalue="$x" />};
314 $ctx->{-extra_form_html} = $extra;
318 my ($cb, $mset, $q, $ctx) = @_;
319 my $nshard = $ctx->{-inbox}->search->{nshard} // 1;
320 $ctx->{ids} = [ map { mdocid($nshard, $_) } $mset->items ];
321 $ctx->{search_query} = $q; # used by WwwAtomStream::atom_header
322 PublicInbox::WwwAtomStream->response($ctx, 200, \&adump_i);
325 # callback for PublicInbox::WwwAtomStream::getline
328 while (my $num = shift @{$ctx->{ids}}) {
329 my $smsg = eval { $ctx->{-inbox}->over->get_art($num) } or next;