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