1 # Copyright (C) 2015, all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
6 eval { require PublicInbox::Search; };
7 plan skip_all => "Xapian missing for search" if $@;
8 use File::Temp qw/tempdir/;
11 my $tmpdir = tempdir(CLEANUP => 1);
12 my $git_dir = "$tmpdir/a.git";
13 my ($root_id, $last_id);
15 is(0, system(qw(git init -q --bare), $git_dir), "git init (main)");
16 eval { PublicInbox::Search->new($git_dir) };
17 ok($@, "exception raised on non-existent DB");
19 my $rw = PublicInbox::Search->new($git_dir, 1);
20 my $ro = PublicInbox::Search->new($git_dir);
23 my $root = Email::MIME->create(
25 Date => 'Fri, 02 Oct 1993 00:00:00 +0000',
26 Subject => 'Hello world',
27 'Message-ID' => '<root@s>',
28 From => 'John Smith <js@example.com>',
29 To => 'list@example.com',
32 my $last = Email::MIME->create(
34 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
35 Subject => 'Re: Hello world',
36 'In-Reply-To' => '<root@s>',
37 'Message-ID' => '<last@s>',
38 From => 'John Smith <js@example.com>',
39 To => 'list@example.com',
41 body => "goodbye forever :<\n");
44 $root_id = $rw->add_message($root);
45 is($root_id, int($root_id), "root_id is an integer: $root_id");
46 $last_id = $rw->add_message($last);
47 is($last_id, int($last_id), "last_id is an integer: $last_id");
52 sort(map { $_->mid } @{$res->{msgs}});
57 my $found = $ro->lookup_message('<root@s>');
58 ok($found, "message found");
59 is($root_id, $found->{doc_id}, 'doc_id set correctly');
60 $found->ensure_metadata;
61 is($found->mid, 'root@s', 'mid set correctly');
62 ok(int($found->thread_id) > 0, 'thread_id is an integer');
64 my @exp = sort qw(root@s last@s);
65 my $res = $ro->query("path:hello_world");
66 my @res = filter_mids($res);
67 is_deeply(\@res, \@exp, 'got expected results for path: match');
69 foreach my $p (qw(hello hello_ hello_world2 hello_world_)) {
70 $res = $ro->query("path:$p");
71 is($res->{count}, 0, "path variant `$p' does not match");
74 $res = $ro->query('subject:(Hello world)');
75 @res = filter_mids($res);
76 is_deeply(\@res, \@exp, 'got expected results for subject:() match');
78 $res = $ro->query('subject:"Hello world"');
79 @res = filter_mids($res);
80 is_deeply(\@res, \@exp, 'got expected results for subject:"" match');
82 $res = $ro->query('subject:"Hello world"', {limit => 1});
83 is(scalar @{$res->{msgs}}, 1, "limit works");
84 my $first = $res->{msgs}->[0];
86 $res = $ro->query('subject:"Hello world"', {offset => 1});
87 is(scalar @{$res->{msgs}}, 1, "offset works");
88 my $second = $res->{msgs}->[0];
90 isnt($first, $second, "offset returned different result from limit");
92 foreach my $f (qw(inreplyto references)) {
93 $res = $ro->query($f . ':root@s');
94 @res = filter_mids($res);
95 is_deeply(\@res, [ 'last@s' ],
96 "got expected results for $f: match");
97 $res = $ro->query($f . ':root');
98 is($res->{count}, 0, "no partial mid match");
105 my $rmid = '<ghost-message@s>';
106 my $reply_to_ghost = Email::MIME->create(
108 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
109 Subject => 'Re: ghosts',
110 'Message-ID' => '<ghost-reply@s>',
111 'In-Reply-To' => $rmid,
112 From => 'Time Traveler <tt@example.com>',
113 To => 'list@example.com',
118 my $reply_id = $rw->add_message($reply_to_ghost);
119 is($reply_id, int($reply_id), "reply_id is an integer: $reply_id");
121 my $was_ghost = Email::MIME->create(
123 Date => 'Sat, 02 Oct 2010 00:00:01 +0000',
125 'Message-ID' => $rmid,
126 From => 'Laggy Sender <lag@example.com>',
127 To => 'list@example.com',
129 body => "are real\n");
131 my $ghost_id = $rw->add_message($was_ghost);
132 is($ghost_id, int($ghost_id), "ghost_id is an integer: $ghost_id");
133 ok($ghost_id < $reply_id, "ghost vivified from earlier message");
136 # search thread on ghost
141 my $res = $ro->query('ghost');
142 my @exp = sort qw(ghost-message@s ghost-reply@s);
143 my @res = filter_mids($res);
144 is_deeply(\@res, \@exp, 'got expected results for Subject match');
147 $res = $ro->query('goodbye');
148 is($res->{msgs}->[0]->mid, 'last@s', 'got goodbye message body');
155 my $long_mid = 'last' . ('x' x 60). '@s';
156 my $long_midc = Digest::SHA::sha1_hex($long_mid);
158 my $long = Email::MIME->create(
160 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
161 Subject => 'long message ID',
162 'References' => '<root@s> <last@s>',
163 'In-Reply-To' => '<last@s>',
164 'Message-ID' => "<$long_mid>",
165 From => '"Long I.D." <long-id@example.com>',
166 To => 'list@example.com',
169 my $long_id = $rw->add_message($long);
170 is($long_id, int($long_id), "long_id is an integer: $long_id");
173 my $res = $ro->query('references:root@s');
174 my @res = filter_mids($res);
175 is_deeply(\@res, [ sort('last@s', $long_midc) ],
176 "got expected results for references: match");
178 my $followups = $ro->get_followups('root@s');
179 $followups = [ filter_mids($followups) ];
180 is_deeply($followups, [ filter_mids($res) ], "get_followups matches");
182 my $long_reply_mid = 'reply-to-long@1';
183 my $long_reply = Email::MIME->create(
185 Subject => 'I break references',
186 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
187 'Message-ID' => "<$long_reply_mid>",
189 # 'References' => '<root@s> <last@s> <'.$long_mid.'>',
190 'In-Reply-To' => "<$long_mid>",
191 From => '"no1 <no1@example.com>',
192 To => 'list@example.com',
194 body => "no References\n");
195 ok($rw->add_message($long_reply) > $long_id, "inserted long reply");
198 my $t = $ro->get_thread('root@s');
199 is($t->{count}, 4, "got all 4 mesages in thread");
200 my @exp = sort($long_reply_mid, 'root@s', 'last@s', $long_midc);
201 @res = filter_mids($t);
202 is_deeply(\@res, \@exp, "get_thread works");
205 # quote prioritization
208 $rw->add_message(Email::MIME->create(
210 Date => 'Sat, 02 Oct 2010 00:00:01 +0000',
212 'Message-ID' => '<quote@a>',
213 From => 'Quoter <quoter@example.com>',
214 To => 'list@example.com',
216 body => "> theatre illusions\nfade\n"));
218 $rw->add_message(Email::MIME->create(
220 Date => 'Sat, 02 Oct 2010 00:00:02 +0000',
222 'Message-ID' => '<nquote@a>',
223 From => 'Non-Quoter<non-quoter@example.com>',
224 To => 'list@example.com',
226 body => "theatre\nfade\n"));
227 my $res = $rw->query("theatre");
228 is($res->{count}, 2, "got both matches");
229 is($res->{msgs}->[0]->mid, 'nquote@a', "non-quoted scores higher");
230 is($res->{msgs}->[1]->mid, 'quote@a', "quoted result still returned");
232 $res = $rw->query("illusions");
233 is($res->{count}, 1, "got a match for quoted text");
234 is($res->{msgs}->[0]->mid, 'quote@a',
235 "quoted result returned if nothing else");