1 |
dpavlin |
1 |
package CouchDB::Estraier; |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
|
|
use warnings; |
5 |
|
|
|
6 |
dpavlin |
3 |
our $VERSION = '0.01'; |
7 |
|
|
|
8 |
dpavlin |
1 |
use Search::Estraier; |
9 |
|
|
use Data::Dump qw/dump/; |
10 |
dpavlin |
3 |
use Getopt::Long; |
11 |
|
|
use JSON; |
12 |
dpavlin |
12 |
#use IO::Handle; |
13 |
|
|
use IO::File; |
14 |
dpavlin |
1 |
|
15 |
|
|
=head1 NAME |
16 |
|
|
|
17 |
|
|
CouchDB::Estraier - Hyper Estraier full-text search for CouchDB |
18 |
|
|
|
19 |
|
|
=head1 METHODS |
20 |
|
|
|
21 |
|
|
=cut |
22 |
|
|
|
23 |
|
|
our $c = { |
24 |
|
|
node_url => 'http://localhost:1978/node/', |
25 |
|
|
estuser => 'admin', |
26 |
|
|
estpasswd => 'admin', |
27 |
dpavlin |
12 |
debug => 0, |
28 |
dpavlin |
1 |
}; |
29 |
|
|
|
30 |
dpavlin |
12 |
my $log = $0; |
31 |
|
|
$log =~ s{^.*/([^/]+)(\.\w+)$}{/tmp/$1.log} or die "can't generate log name from $0"; |
32 |
dpavlin |
1 |
|
33 |
dpavlin |
12 |
warn "$0 log $log\n"; |
34 |
|
|
|
35 |
|
|
GetOptions($c, qw/node_url=s debug+ estuser=s estpasswd=s dbuser=s dbpasswd=s/) or die $!; |
36 |
|
|
|
37 |
|
|
#my $log = IO::Handle->new; |
38 |
|
|
#my $log = IO::Handle->new_from_fd(\*STDERR, 'w'); |
39 |
|
|
|
40 |
|
|
$log = IO::File->new( "> $log" ) || die "can't open $log: $!"; |
41 |
|
|
|
42 |
|
|
$log->autoflush( 1 ); |
43 |
|
|
$log->print("c: ", dump($c), "\n" ); # if $c->{debug}; |
44 |
|
|
|
45 |
|
|
my $in = IO::Handle->new_from_fd(\*STDIN, 'r'); |
46 |
|
|
my $out = IO::Handle->new_from_fd(\*STDOUT, 'w'); |
47 |
|
|
$out->autoflush( 1 ); |
48 |
|
|
|
49 |
dpavlin |
3 |
=head2 run_search |
50 |
dpavlin |
1 |
|
51 |
|
|
Process command line options and start helper |
52 |
|
|
|
53 |
|
|
CouchDB::Estraier->run; |
54 |
|
|
|
55 |
|
|
=cut |
56 |
|
|
|
57 |
dpavlin |
3 |
sub run_search { |
58 |
|
|
while ( 1 ) { |
59 |
dpavlin |
12 |
$log->print("search ready\n"); |
60 |
|
|
my $json = $in->getline; |
61 |
|
|
die unless defined $json; |
62 |
|
|
$log->print( "run_search $json\n" ); |
63 |
|
|
$out->print( |
64 |
|
|
encode_json( |
65 |
|
|
search( decode_json( $json ) ) |
66 |
|
|
) ,"\n" |
67 |
|
|
); |
68 |
dpavlin |
1 |
} |
69 |
|
|
} |
70 |
|
|
|
71 |
dpavlin |
3 |
sub run_update { |
72 |
|
|
while ( 1 ) { |
73 |
dpavlin |
12 |
$log->print("update ready\n"); |
74 |
|
|
my $json = $in->getline; |
75 |
|
|
die unless defined $json; |
76 |
|
|
$log->print( "run_update $json\n" ); |
77 |
|
|
update( decode_json( $json ) ) |
78 |
dpavlin |
3 |
} |
79 |
|
|
} |
80 |
dpavlin |
1 |
|
81 |
dpavlin |
3 |
|
82 |
dpavlin |
12 |
=head2 update |
83 |
dpavlin |
1 |
|
84 |
dpavlin |
12 |
CouchDB::Estraier::update( { db => $database, type => $type } ); |
85 |
dpavlin |
1 |
|
86 |
|
|
=cut |
87 |
|
|
|
88 |
dpavlin |
12 |
sub update { |
89 |
|
|
my $args = shift or die "no args"; |
90 |
|
|
$log->print( "add ",dump( $args ),"\n" ); |
91 |
dpavlin |
1 |
|
92 |
dpavlin |
12 |
my $ret = { |
93 |
|
|
code => 200, |
94 |
|
|
json => { |
95 |
|
|
args => $args, |
96 |
|
|
}, |
97 |
|
|
}; |
98 |
dpavlin |
3 |
|
99 |
dpavlin |
12 |
return $ret; |
100 |
|
|
|
101 |
|
|
my $database = $args->{db} or die "no db in ",dump( $args ); |
102 |
|
|
my $data = $args->{data} or die "no data in ",dump( $args ); |
103 |
|
|
|
104 |
dpavlin |
1 |
# create and configure node |
105 |
|
|
my $node = new Search::Estraier::Node( |
106 |
|
|
url => $c->{node_url} . $database, |
107 |
|
|
user => $c->{estuser}, |
108 |
|
|
passwd => $c->{estpasswd}, |
109 |
|
|
croak_on_error => 1, |
110 |
|
|
create => 1, |
111 |
|
|
debug => $c->{debug} >= 4 ? 1 : 0, |
112 |
|
|
); |
113 |
|
|
|
114 |
|
|
# create document |
115 |
|
|
my $doc = new Search::Estraier::Document; |
116 |
|
|
|
117 |
|
|
if (my $id = $data->{_id}) { |
118 |
|
|
$doc->add_attr('@uri', $id); |
119 |
|
|
} else { |
120 |
|
|
die "can't find pk_col column _id in results\n"; |
121 |
|
|
} |
122 |
|
|
|
123 |
|
|
while (my ($col,$val) = each %{$data}) { |
124 |
|
|
|
125 |
dpavlin |
12 |
if ( defined $val ) { |
126 |
dpavlin |
1 |
# add attributes (make column usable from attribute search) |
127 |
|
|
$doc->add_attr($col, $val); |
128 |
|
|
|
129 |
|
|
# add body text to document (make it searchable using full-text index) |
130 |
|
|
$doc->add_text($val); |
131 |
|
|
} |
132 |
|
|
|
133 |
|
|
} |
134 |
|
|
|
135 |
dpavlin |
12 |
$log->print("doc draft: ",$doc->dump_draft ) if ($c->{debug} >= 2); |
136 |
dpavlin |
1 |
|
137 |
|
|
die "error: ", $node->status,"\n" unless (eval { $node->put_doc($doc) }); |
138 |
|
|
|
139 |
dpavlin |
12 |
return $ret; |
140 |
dpavlin |
1 |
} |
141 |
|
|
|
142 |
|
|
=head2 search |
143 |
|
|
|
144 |
|
|
Implementation specification: L<http://wiki.apache.org/couchdb/FullTextSearch> |
145 |
|
|
|
146 |
dpavlin |
3 |
CouchDB::Estraier::search( $database, $query ); |
147 |
dpavlin |
1 |
|
148 |
|
|
=cut |
149 |
|
|
|
150 |
|
|
sub search { |
151 |
dpavlin |
12 |
my $args = shift or die "no args"; |
152 |
|
|
$log->print( "search ",dump( $args ),"\n" ); |
153 |
|
|
my $database = $args->{db} or die "no db in ",dump( $args ); |
154 |
|
|
my $query = $args->{query} or die "no query in ",dump( $args ); |
155 |
dpavlin |
1 |
|
156 |
|
|
# create and configure node |
157 |
|
|
my $node = new Search::Estraier::Node( |
158 |
|
|
url => $c->{node_url} . $database, |
159 |
|
|
user => $c->{estuser}, |
160 |
|
|
passwd => $c->{estpasswd}, |
161 |
|
|
croak_on_error => 1, |
162 |
dpavlin |
12 |
# create => 1, |
163 |
dpavlin |
1 |
debug => $c->{debug} >= 4 ? 1 : 0, |
164 |
|
|
); |
165 |
|
|
|
166 |
|
|
my $cond = new Search::Estraier::Condition; |
167 |
|
|
$cond->set_phrase( $query ); |
168 |
|
|
my $nres = $node->search($cond, 0); |
169 |
|
|
|
170 |
|
|
if ( defined($nres) ) { |
171 |
|
|
|
172 |
dpavlin |
12 |
$out->print( "ok\n" ); |
173 |
dpavlin |
1 |
for my $i ( 0 ... $nres->doc_num - 1 ) { |
174 |
|
|
my $rdoc = $nres->get_doc($i); |
175 |
|
|
print $rdoc->attr('@uri'),"\n",$i,"\n"; |
176 |
|
|
} |
177 |
dpavlin |
12 |
$out->print( "\n\n" ); |
178 |
dpavlin |
1 |
|
179 |
|
|
} else { |
180 |
dpavlin |
12 |
$out->print( "error\n", $node->status, "\n" ); |
181 |
dpavlin |
1 |
} |
182 |
|
|
|
183 |
|
|
} |
184 |
|
|
|
185 |
|
|
|
186 |
|
|
1; |