| 1 |
#!/usr/bin/perl -w |
| 2 |
|
| 3 |
use strict; |
| 4 |
|
| 5 |
use Cwd qw/abs_path/; |
| 6 |
use File::Temp qw/tempdir/; |
| 7 |
use Data::Dumper; |
| 8 |
use lib './lib'; |
| 9 |
|
| 10 |
use WebPAC::Lookup; |
| 11 |
use WebPAC::Input::ISIS; |
| 12 |
use WebPAC::Store 0.03; |
| 13 |
use WebPAC::Normalize::XML; |
| 14 |
use WebPAC::Output::TT; |
| 15 |
use WebPAC::Output::Estraier 0.02; |
| 16 |
use YAML qw/LoadFile/; |
| 17 |
use LWP::Simple; |
| 18 |
|
| 19 |
my $limit = shift @ARGV; |
| 20 |
|
| 21 |
my $config = LoadFile('conf/config.yml'); |
| 22 |
|
| 23 |
print "config = ",Dumper($config); |
| 24 |
|
| 25 |
die "no databases in config file!\n" unless ($config->{databases}); |
| 26 |
|
| 27 |
my $total_rows = 0; |
| 28 |
|
| 29 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
| 30 |
|
| 31 |
my $type = lc($db_config->{input}->{type}); |
| 32 |
|
| 33 |
die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis'); |
| 34 |
|
| 35 |
my $abs_path = abs_path($0); |
| 36 |
$abs_path =~ s#/[^/]*$#/#; |
| 37 |
|
| 38 |
my $lookup = new WebPAC::Lookup( |
| 39 |
lookup_file => $db_config->{input}->{lookup}, |
| 40 |
); |
| 41 |
|
| 42 |
my $db_path = $config->{webpac}->{db_path} . '/' . $database; |
| 43 |
|
| 44 |
|
| 45 |
my $log = $lookup->_get_logger; |
| 46 |
$log->info("working on $database in $db_path"); |
| 47 |
|
| 48 |
my $db = new WebPAC::Store( |
| 49 |
path => $db_path, |
| 50 |
database => $database, |
| 51 |
debug => 1, |
| 52 |
); |
| 53 |
|
| 54 |
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
| 55 |
$est_config->{database} = $database; |
| 56 |
|
| 57 |
$log->info("using HyperEstraier URL $est_config->{masterurl}"); |
| 58 |
|
| 59 |
my $est = new WebPAC::Output::Estraier( |
| 60 |
%{ $est_config }, |
| 61 |
); |
| 62 |
|
| 63 |
# |
| 64 |
# now, iterate through input formats |
| 65 |
# |
| 66 |
|
| 67 |
my @inputs; |
| 68 |
if (ref($db_config->{input}) eq 'ARRAY') { |
| 69 |
@inputs = @{ $db_config->{input} }; |
| 70 |
} else { |
| 71 |
push @inputs, $db_config->{input}; |
| 72 |
} |
| 73 |
|
| 74 |
foreach my $input (@inputs) { |
| 75 |
$log->info("working on input $input->{path} [$input->{type}]"); |
| 76 |
|
| 77 |
my $isis = new WebPAC::Input::ISIS( |
| 78 |
code_page => $config->{webpac}->{webpac_encoding}, |
| 79 |
limit_mfn => $input->{limit}, |
| 80 |
); |
| 81 |
|
| 82 |
my $maxmfn = $isis->open( |
| 83 |
filename => $input->{path}, |
| 84 |
code_page => $input->{encoding}, # database encoding |
| 85 |
); |
| 86 |
|
| 87 |
my $n = new WebPAC::Normalize::XML( |
| 88 |
# filter => { 'foo' => sub { shift } }, |
| 89 |
db => $db, |
| 90 |
lookup_regex => $lookup->regex, |
| 91 |
lookup => $lookup, |
| 92 |
prefix => $input->{name}, |
| 93 |
); |
| 94 |
|
| 95 |
$n->open( |
| 96 |
tag => $input->{normalize}->{tag}, |
| 97 |
xml_file => $input->{normalize}->{path}, |
| 98 |
); |
| 99 |
|
| 100 |
for ( 0 ... $isis->size ) { |
| 101 |
|
| 102 |
my $row = $isis->fetch || next; |
| 103 |
|
| 104 |
my $mfn = $row->{'000'}->[0] || die "can't find MFN"; |
| 105 |
|
| 106 |
my $ds = $n->data_structure($row); |
| 107 |
|
| 108 |
$est->add( |
| 109 |
id => $input->{name} . "#" . $mfn, |
| 110 |
ds => $ds, |
| 111 |
type => $config->{hyperestraier}->{type}, |
| 112 |
); |
| 113 |
|
| 114 |
$total_rows++; |
| 115 |
} |
| 116 |
|
| 117 |
}; |
| 118 |
|
| 119 |
$log->info("$total_rows records indexed"); |
| 120 |
} |
| 121 |
|
Parent Directory
| 
Revision Log