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