1 |
dpavlin |
131 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
dpavlin |
144 |
use Search::Estraier 0.06; |
5 |
dpavlin |
131 |
use DBI; |
6 |
|
|
use Data::Dumper; |
7 |
|
|
use Encode qw/from_to/; |
8 |
dpavlin |
146 |
use Time::HiRes qw/time/; |
9 |
dpavlin |
159 |
use Getopt::Long; |
10 |
dpavlin |
131 |
|
11 |
|
|
=head1 NAME |
12 |
|
|
|
13 |
|
|
dbi-indexer.pl - example indexer of DBI sources for Search::Estraier |
14 |
|
|
|
15 |
|
|
=cut |
16 |
|
|
|
17 |
dpavlin |
144 |
my $c = { |
18 |
dpavlin |
195 |
node_url => 'http://localhost:1978/node/dbi-template1', |
19 |
|
|
dbi => 'Pg:dbname=template1', |
20 |
|
|
dbuser => 'postgres', |
21 |
dpavlin |
144 |
sql => qq{ |
22 |
dpavlin |
195 |
select * from pg_database |
23 |
dpavlin |
144 |
}, |
24 |
dpavlin |
195 |
pk_col => 'datname', |
25 |
dpavlin |
145 |
db_encoding => 'iso-8859-2', |
26 |
dpavlin |
144 |
debug => 0, |
27 |
dpavlin |
195 |
estuser => 'admin', |
28 |
|
|
estpasswd => 'admin', |
29 |
|
|
quiet => 0, |
30 |
dpavlin |
131 |
}; |
31 |
|
|
|
32 |
dpavlin |
195 |
GetOptions($c, qw/node_url=s dbi=s sql=s pk_col=s eb_encoding=s debug+ quiet+ estuser=s estpasswd=s dbuser=s dbpasswd=s/); |
33 |
dpavlin |
159 |
|
34 |
|
|
warn "# c: ", Dumper($c) if ($c->{debug}); |
35 |
|
|
|
36 |
dpavlin |
131 |
# create and configure node |
37 |
|
|
my $node = new Search::Estraier::Node( |
38 |
dpavlin |
144 |
url => $c->{node_url}, |
39 |
dpavlin |
183 |
user => $c->{estuser}, |
40 |
|
|
passwd => $c->{estpasswd}, |
41 |
dpavlin |
131 |
croak_on_error => 1, |
42 |
dpavlin |
144 |
create => 1, |
43 |
dpavlin |
159 |
debug => $c->{debug} >= 4 ? 1 : 0, |
44 |
dpavlin |
131 |
); |
45 |
|
|
|
46 |
|
|
# create DBI connection |
47 |
dpavlin |
183 |
my $dbh = DBI->connect("DBI:$c->{dbi}", $c->{dbuser}, $c->{dbpasswd}) || die $DBI::errstr; |
48 |
dpavlin |
131 |
|
49 |
dpavlin |
144 |
my $sth = $dbh->prepare($c->{sql}) || die $dbh->errstr(); |
50 |
dpavlin |
131 |
$sth->execute() || die $sth->errstr(); |
51 |
|
|
|
52 |
dpavlin |
144 |
warn "# columns: ",join(",",@{ $sth->{NAME} }),"\n" if ($c->{debug}); |
53 |
dpavlin |
131 |
|
54 |
|
|
my $total = $sth->rows; |
55 |
|
|
my $i = 1; |
56 |
|
|
|
57 |
dpavlin |
146 |
my $t = time(); |
58 |
dpavlin |
147 |
my $pk_col = $c->{pk_col} || 'id'; |
59 |
dpavlin |
146 |
|
60 |
dpavlin |
131 |
while (my $row = $sth->fetchrow_hashref() ) { |
61 |
|
|
|
62 |
dpavlin |
159 |
warn "# row: ",Dumper($row) if ($c->{debug} >= 3); |
63 |
dpavlin |
131 |
|
64 |
|
|
# create document |
65 |
|
|
my $doc = new Search::Estraier::Document; |
66 |
|
|
|
67 |
dpavlin |
147 |
if (my $id = $row->{$pk_col}) { |
68 |
|
|
$doc->add_attr('@uri', $id); |
69 |
|
|
} else { |
70 |
|
|
die "can't find pk_col column '$pk_col' in results\n"; |
71 |
|
|
} |
72 |
dpavlin |
131 |
|
73 |
dpavlin |
195 |
my $out = ''; |
74 |
|
|
$out .= sprintf("%4d ",$i); |
75 |
dpavlin |
131 |
|
76 |
|
|
while (my ($col,$val) = each %{$row}) { |
77 |
|
|
|
78 |
|
|
if ($val) { |
79 |
|
|
# change encoding? |
80 |
dpavlin |
145 |
from_to($val, ($c->{db_encoding} || 'ISO-8859-1'), 'UTF-8'); |
81 |
dpavlin |
131 |
|
82 |
|
|
# add attributes (make column usable from attribute search) |
83 |
|
|
$doc->add_attr($col, $val); |
84 |
|
|
|
85 |
|
|
# add body text to document (make it searchable using full-text index) |
86 |
|
|
$doc->add_text($val); |
87 |
|
|
|
88 |
dpavlin |
195 |
$out .= "R"; |
89 |
dpavlin |
131 |
} else { |
90 |
dpavlin |
195 |
$out .= "."; |
91 |
dpavlin |
131 |
} |
92 |
|
|
|
93 |
|
|
} |
94 |
|
|
|
95 |
dpavlin |
159 |
warn "# doc draft: ",$doc->dump_draft, "\n" if ($c->{debug} >= 2); |
96 |
dpavlin |
131 |
|
97 |
|
|
die "error: ", $node->status,"\n" unless (eval { $node->put_doc($doc) }); |
98 |
dpavlin |
146 |
|
99 |
dpavlin |
195 |
printf ("%s %d%% %.1f/s\n", $out, int(( $i++ / $total) * 100), ( $i / (time() - $t) ) ) unless ($c->{quiet}); |
100 |
dpavlin |
146 |
|
101 |
dpavlin |
131 |
} |