7 |
use Data::Dumper; |
use Data::Dumper; |
8 |
use lib './lib'; |
use lib './lib'; |
9 |
|
|
10 |
|
use WebPAC::Common 0.02; |
11 |
use WebPAC::Lookup; |
use WebPAC::Lookup; |
12 |
use WebPAC::Input::ISIS; |
use WebPAC::Input::ISIS; |
13 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.03; |
14 |
use WebPAC::Normalize::XML; |
use WebPAC::Normalize::XML; |
15 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
16 |
use WebPAC::Output::Estraier 0.02; |
use WebPAC::Output::Estraier 0.05; |
17 |
use YAML qw/LoadFile/; |
use YAML qw/LoadFile/; |
18 |
use LWP::Simple; |
use LWP::Simple; |
19 |
|
|
29 |
|
|
30 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
31 |
|
|
32 |
|
my $log = _new WebPAC::Common()->_get_logger(); |
33 |
|
|
34 |
|
# |
35 |
|
# open Hyper Estraier database |
36 |
|
# |
37 |
|
|
38 |
|
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
39 |
|
$est_config->{database} = $database; |
40 |
|
|
41 |
|
$log->info("using HyperEstraier URL $est_config->{masterurl}"); |
42 |
|
|
43 |
|
my $est = new WebPAC::Output::Estraier( |
44 |
|
%{ $est_config }, |
45 |
|
); |
46 |
|
|
47 |
|
# |
48 |
|
# now WebPAC::Store |
49 |
|
# |
50 |
my $abs_path = abs_path($0); |
my $abs_path = abs_path($0); |
51 |
$abs_path =~ s#/[^/]*$#/#; |
$abs_path =~ s#/[^/]*$#/#; |
52 |
|
|
53 |
my $db_path = $config->{webpac}->{db_path} . '/' . $database; |
my $db_path = $config->{webpac}->{db_path} . '/' . $database; |
54 |
|
|
55 |
|
$log->info("working on $database in $db_path"); |
56 |
|
|
57 |
my $db = new WebPAC::Store( |
my $db = new WebPAC::Store( |
58 |
path => $db_path, |
path => $db_path, |
59 |
database => $database, |
database => $database, |
60 |
debug => 1, |
debug => 1, |
61 |
); |
); |
62 |
|
|
|
my $log = $db->_get_logger; |
|
|
$log->info("working on $database in $db_path"); |
|
|
|
|
|
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
|
|
$est_config->{database} = $database; |
|
|
|
|
|
$log->info("using HyperEstraier URL $est_config->{masterurl}"); |
|
|
|
|
|
my $est = new WebPAC::Output::Estraier( |
|
|
%{ $est_config }, |
|
|
); |
|
63 |
|
|
64 |
# |
# |
65 |
# now, iterate through input formats |
# now, iterate through input formats |
68 |
my @inputs; |
my @inputs; |
69 |
if (ref($db_config->{input}) eq 'ARRAY') { |
if (ref($db_config->{input}) eq 'ARRAY') { |
70 |
@inputs = @{ $db_config->{input} }; |
@inputs = @{ $db_config->{input} }; |
71 |
} else { |
} elsif ($db_config->{input}) { |
72 |
push @inputs, $db_config->{input}; |
push @inputs, $db_config->{input}; |
73 |
|
} else { |
74 |
|
$log->info("database $database doesn't have inputs defined"); |
75 |
} |
} |
76 |
|
|
77 |
foreach my $input (@inputs) { |
foreach my $input (@inputs) { |
97 |
code_page => $input->{encoding}, # database encoding |
code_page => $input->{encoding}, # database encoding |
98 |
); |
); |
99 |
|
|
|
$log->info( Dumper($lookup->{_lookup_data}) ); |
|
|
|
|
100 |
my $n = new WebPAC::Normalize::XML( |
my $n = new WebPAC::Normalize::XML( |
101 |
# filter => { 'foo' => sub { shift } }, |
# filter => { 'foo' => sub { shift } }, |
102 |
db => $db, |
db => $db, |
105 |
prefix => $input->{name}, |
prefix => $input->{name}, |
106 |
); |
); |
107 |
|
|
108 |
$n->open( |
my $normalize_path = $input->{normalize}->{path}; |
109 |
tag => $input->{normalize}->{tag}, |
|
110 |
xml_file => $input->{normalize}->{path}, |
if ($normalize_path =~ m/\.xml$/i) { |
111 |
); |
$n->open( |
112 |
|
tag => $input->{normalize}->{tag}, |
113 |
|
xml_file => $input->{normalize}->{path}, |
114 |
|
); |
115 |
|
} elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) { |
116 |
|
$n->open_yaml( |
117 |
|
path => $normalize_path, |
118 |
|
tag => $input->{normalize}->{tag}, |
119 |
|
); |
120 |
|
} |
121 |
|
|
122 |
for ( 0 ... $isis->size ) { |
for ( 0 ... $isis->size ) { |
123 |
|
|
139 |
}; |
}; |
140 |
|
|
141 |
$log->info("$total_rows records indexed"); |
$log->info("$total_rows records indexed"); |
142 |
|
|
143 |
|
# |
144 |
|
# add Hyper Estraier links to other databases |
145 |
|
# |
146 |
|
if (ref($db_config->{links}) eq 'ARRAY') { |
147 |
|
foreach my $link (@{ $db_config->{links} }) { |
148 |
|
$log->info("adding link $database -> $link->{to} [$link->{credit}]"); |
149 |
|
$est->add_link( |
150 |
|
from => $database, |
151 |
|
to => $link->{to}, |
152 |
|
credit => $link->{credit}, |
153 |
|
); |
154 |
|
} |
155 |
|
} |
156 |
|
|
157 |
} |
} |
158 |
|
|