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 uri_escape);
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;
20 my $q = PublicInbox::SearchQuery->new($ctx->{qp});
22 require PublicInbox::Mbox;
23 return PublicInbox::Mbox::mbox_all($ctx, $q->{'q'}) if $x eq 'm';
29 my $srch = $ctx->{-inbox}->search or
30 return PublicInbox::WWW::need($ctx, 'Search');
31 my $q = PublicInbox::SearchQuery->new($ctx->{qp});
33 my $query = $q->{'q'};
38 $o = -($o + 1); # so [-1] is the last element, like Perl lists
42 # double the limit for expanded views:
50 my ($mset, $total, $err, $html);
53 $mset = $srch->query($query, $opts);
54 $total = $mset->get_matches_estimated;
57 ctx_prepare($q, $ctx);
60 $html = '<pre>'.err_txt($ctx, $err).'</pre><hr>';
61 } elsif ($total == 0) {
62 if (defined($ctx->{-uxs_retried})) {
64 $q->{'q'} = $ctx->{-uxs_retried};
65 } elsif (index($q->{'q'}, '%') >= 0) {
66 $ctx->{-uxs_retried} = $q->{'q'};
67 $q->{'q'} = uri_unescape($q->{'q'});
71 $html = "<pre>\n[No results found]</pre><hr>";
73 return adump($_[0], $mset, $q, $ctx) if $x eq 'A';
75 $ctx->{-html_tip} = search_nav_top($mset, $q, $ctx);
76 return mset_thread($ctx, $mset, $q) if $x eq 't';
77 mset_summary($ctx, $mset, $q); # appends to {-html_tip}
80 html_oneshot($ctx, $code);
83 # display non-nested search results similar to what users expect from
84 # regular WWW search engines:
86 my ($ctx, $mset, $q) = @_;
88 my $total = $mset->get_matches_estimated;
89 my $pad = length("$total");
91 my $res = \($ctx->{-html_tip});
92 my $ibx = $ctx->{-inbox};
93 my $srch = $ibx->search;
94 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
95 foreach my $m ($mset->items) {
96 my $rank = sprintf("%${pad}d", $m->get_rank + 1);
97 my $pct = get_pct($m);
98 my $smsg = PublicInbox::Smsg::from_mitem($m, $srch);
101 $m = "$m ".$m->get_docid . " expired\n";
102 $ctx->{env}->{'psgi.errors'}->print($m);
106 my $s = ascii_html($smsg->{subject});
107 my $f = ascii_html($smsg->{from_name});
109 obfuscate_addrs($obfs_ibx, $s);
110 obfuscate_addrs($obfs_ibx, $f);
112 my $date = PublicInbox::View::fmt_ts($smsg->{ds});
113 my $mid = mid_href($smsg->{mid});
114 $s = '(no subject)' if $s eq '';
115 $$res .= qq{$rank. <b><a\nhref="$mid/">}.
117 $$res .= "$pfx - by $f @ $date UTC [$pct%]\n\n";
119 $$res .= search_nav_bot($mset, $q);
123 # shorten "/full/path/to/Foo/Bar.pm" to "Foo/Bar.pm" so error
124 # messages don't reveal FS layout info in case people use non-standard
128 if (my $short = $rmap_inc{$full}) {
130 } elsif (!scalar(keys %rmap_inc) && -e $full) {
131 %rmap_inc = map {; "$INC{$_}" => $_ } keys %INC;
132 # fall back to basename as last resort
133 $rmap_inc{$full} // (split('/', $full))[-1];
140 my ($ctx, $err) = @_;
141 my $u = $ctx->{-inbox}->base_url($ctx->{env}) . '_/text/help/';
142 $err =~ s/^\s*Exception:\s*//; # bad word to show users :P
143 $err =~ s!(\S+)!path2inc($1)!sge;
144 $err = ascii_html($err);
145 "\nBad query: <b>$err</b>\n" .
146 qq{See <a\nhref="$u">$u</a> for help on using search};
150 my ($mset, $q, $ctx) = @_;
151 my $m = $q->qs_html(x => 'm', r => undef);
152 my $rv = qq{<form\naction="?$m"\nmethod="post"><pre>};
153 my $initial_q = $ctx->{-uxs_retried};
154 if (defined $initial_q) {
155 my $rewritten = $q->{'q'};
156 utf8::decode($initial_q);
157 utf8::decode($rewritten);
158 $initial_q = ascii_html($initial_q);
159 $rewritten = ascii_html($rewritten);
160 $rv .= " Warning: Initial query:\n <b>$initial_q</b>\n";
161 $rv .= " returned no results, used:\n";
162 $rv .= " <b>$rewritten</b>\n instead\n\n";
165 $rv .= 'Search results ordered by [';
167 my $d = $q->qs_html(r => 0);
168 $rv .= qq{<a\nhref="?$d">date</a>|<b>relevance</b>};
170 my $d = $q->qs_html(r => 1);
171 $rv .= qq{<b>date</b>|<a\nhref="?$d">relevance</a>};
178 my $t = $q->qs_html(x => 't');
179 $rv .= qq{<b>summary</b>|<a\nhref="?$t">nested</a>}
180 } elsif ($q->{x} eq 't') {
181 my $s = $q->qs_html(x => '');
182 $rv .= qq{<a\nhref="?$s">summary</a>|<b>nested</b>};
184 my $A = $q->qs_html(x => 'A', r => undef);
185 $rv .= qq{|<a\nhref="?$A">Atom feed</a>]};
186 $rv .= qq{\n\t\t\t\t\t\tdownload: };
187 $rv .= qq{<input\ntype=submit\nvalue="mbox.gz"/></pre></form><pre>};
192 my $total = $mset->get_matches_estimated;
194 my $rv = '</pre><hr><pre id=t>';
196 my $off = $o < 0 ? -($o + 1) : $o;
197 my $end = $off + $mset->size;
201 $rv .= "Results $beg-$end of $total";
202 $rv .= ' (estimated)' if $end != $total;
204 $rv .= "No more results, only $total";
206 my ($next, $join, $prev);
208 if ($o >= 0) { # sort descending
211 $next = $q->qs_html(o => $n, l => $l);
214 $join = $n < $total ? '/' : ' ';
216 $prev = $q->qs_html(o => ($p > 0 ? $p : 0));
218 } else { # o < 0, sort ascending
222 $next = $q->qs_html(o => $n, l => $l);
225 $join = -$n < $total ? '/' : ' ';
227 $prev = $q->qs_html(o => ($p < 0 ? $p : 0));
231 $rv .= qq{ <a\nhref="?$next"\nrel=next>next</a>} if $next;
232 $rv .= $join if $join;
233 $rv .= qq{<a\nhref="?$prev"\nrel=prev>prev</a>} if $prev;
235 my $rev = $q->qs_html(o => $o < 0 ? 0 : -1);
236 $rv .= qq{ | <a\nhref="?$rev">reverse results</a></pre>};
241 (eval { $b->topmost->{pct} } // 0) <=>
242 (eval { $a->topmost->{pct} } // 0)
247 # Capped at "99%" since "100%" takes an extra column in the
248 # thread skeleton view. <xapian/mset.h> says the value isn't
249 # very meaningful, anyways.
250 my $n = $_[0]->get_percent;
258 my $smsg = PublicInbox::Smsg::from_mitem($mi);
259 $smsg->{pct} = get_pct($mi);
265 my ($ctx, $mset, $q) = @_;
266 my $ibx = $ctx->{-inbox};
267 my $msgs = $ibx->search->retry_reopen(\&load_msgs, $mset);
269 my $rootset = PublicInbox::SearchThread::thread($msgs,
270 $r ? \&sort_relevance : \&PublicInbox::View::sort_ds,
272 my $skel = search_nav_bot($mset, $q). "<pre>";
274 $ctx->{anchor_idx} = 1;
275 $ctx->{cur_level} = 0;
276 $ctx->{skel} = \$skel;
277 $ctx->{mapping} = {};
278 $ctx->{searchview} = 1;
279 $ctx->{prev_attr} = '';
280 $ctx->{prev_level} = 0;
281 $ctx->{s_nr} = scalar(@$msgs).'+ results';
283 # reduce hash lookups in skel_dump
284 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
285 PublicInbox::View::walk_thread($rootset, $ctx,
286 \&PublicInbox::View::pre_thread);
288 @$msgs = reverse @$msgs if $r;
289 $ctx->{msgs} = $msgs;
290 PublicInbox::WwwStream::aresponse($ctx, 200, \&mset_thread_i);
293 # callback for PublicInbox::WwwStream::getline
295 my ($ctx, $eml) = @_;
296 $ctx->zmore($ctx->html_top) if exists $ctx->{-html_tip};
297 $eml and return PublicInbox::View::eml_entry($ctx, $eml);
298 my $smsg = shift @{$ctx->{msgs}} or
299 $ctx->zmore(${delete($ctx->{skel})});
307 $qh = ascii_html($qh);
308 $ctx->{-q_value_html} = $qh;
309 $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef);
310 $ctx->{-title_html} = "$qh - search results";
312 $extra .= qq{<input\ntype=hidden\nname=r />} if $q->{r};
313 if (my $x = $q->{x}) {
315 $extra .= qq{<input\ntype=hidden\nname=x\nvalue="$x" />};
317 $ctx->{-extra_form_html} = $extra;
321 my ($cb, $mset, $q, $ctx) = @_;
322 $ctx->{items} = [ $mset->items ];
323 $ctx->{search_query} = $q; # used by WwwAtomStream::atom_header
324 $ctx->{srch} = $ctx->{-inbox}->search;
325 PublicInbox::WwwAtomStream->response($ctx, 200, \&adump_i);
328 # callback for PublicInbox::WwwAtomStream::getline
331 while (my $mi = shift @{$ctx->{items}}) {
333 PublicInbox::Smsg::from_mitem($mi, $ctx->{srch});
339 package PublicInbox::SearchQuery;
342 use URI::Escape qw(uri_escape);
343 use PublicInbox::MID qw(MID_ESC);
346 my ($class, $qp) = @_;
349 my ($l) = (($qp->{l} || '') =~ /([0-9]+)/);
350 $l = $LIM if !$l || $l > $LIM;
354 o => (($qp->{o} || '0') =~ /(-?[0-9]+)/),
356 r => (defined $r && $r ne '0'),
361 my ($self, %override) = @_;
363 if (scalar(keys(%override))) {
364 $self = bless { (%$self, %override) }, ref($self);
367 my $q = uri_escape($self->{'q'}, MID_ESC);
368 $q =~ s/%20/+/g; # improve URL readability
371 if (my $o = $self->{o}) { # ignore o == 0
374 if (my $l = $self->{l}) {
375 $qs .= "&l=$l" unless $l == $LIM;
377 if (my $r = $self->{r}) {
380 if (my $x = $self->{x}) {
381 $qs .= "&x=$x" if ($x eq 't' || $x eq 'A' || $x eq 'm');