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 ($config->{databases}); |
26 |
|
27 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
28 |
|
29 |
my $type = lc($db_config->{input}->{type}); |
30 |
|
31 |
die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis'); |
32 |
|
33 |
my $abs_path = abs_path($0); |
34 |
$abs_path =~ s#/[^/]*$#/#; |
35 |
|
36 |
my $lookup = new WebPAC::Lookup( |
37 |
lookup_file => $db_config->{input}->{lookup}, |
38 |
); |
39 |
|
40 |
my $db_path = $config->{webpac}->{db_path} . '/' . $database; |
41 |
|
42 |
|
43 |
my $log = $lookup->_get_logger; |
44 |
$log->info("working on $database in $db_path"); |
45 |
|
46 |
my $isis = new WebPAC::Input::ISIS( |
47 |
code_page => $config->{webpac}->{webpac_encoding}, |
48 |
limit_mfn => $db_config->{input}->{limit}, |
49 |
); |
50 |
|
51 |
my $maxmfn = $isis->open( |
52 |
filename => $db_config->{input}->{path}, |
53 |
code_page => $db_config->{input}->{encoding}, # database encoding |
54 |
); |
55 |
|
56 |
my $path = './db/'; |
57 |
|
58 |
my $db = new WebPAC::Store( |
59 |
path => $db_path, |
60 |
); |
61 |
|
62 |
my $n = new WebPAC::Normalize::XML( |
63 |
# filter => { 'foo' => sub { shift } }, |
64 |
db => $db, |
65 |
lookup_regex => $lookup->regex, |
66 |
lookup => $lookup, |
67 |
); |
68 |
|
69 |
$n->open( |
70 |
tag => $db_config->{normalize}->{tag}, |
71 |
xml_file => $db_config->{normalize}->{path}, |
72 |
); |
73 |
|
74 |
my $out = new WebPAC::Output::TT( |
75 |
include_path => $config->{webpac}->{template_path}, |
76 |
filters => { foo => sub { shift } }, |
77 |
); |
78 |
|
79 |
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
80 |
$est_config->{database} = $database; |
81 |
|
82 |
$log->info("using HyperEstraier URL $est_config->{url}"); |
83 |
|
84 |
my $est = new WebPAC::Output::Estraier( |
85 |
%{ $est_config }, |
86 |
); |
87 |
|
88 |
my $total_rows = 0; |
89 |
|
90 |
for ( 0 ... $isis->size ) { |
91 |
|
92 |
my $row = $isis->fetch || next; |
93 |
|
94 |
my $mfn = $row->{'000'}->[0] || die "can't find MFN"; |
95 |
|
96 |
my $ds = $n->data_structure($row); |
97 |
|
98 |
# print STDERR Dumper($row, $ds); |
99 |
|
100 |
# my $html = $out->apply( |
101 |
# template => 'html_ffzg.tt', |
102 |
# data => $ds, |
103 |
# ); |
104 |
# |
105 |
# # create test output |
106 |
# |
107 |
# my $file = sprintf('out/%02d.html', $mfn ); |
108 |
# open(my $fh, '>', $file) or die "can't open $file: $!"; |
109 |
# print $fh $html; |
110 |
# close($fh); |
111 |
# |
112 |
# $html =~ s#\s*[\n\r]+\s*##gs; |
113 |
# |
114 |
# print STDERR $html; |
115 |
|
116 |
$est->add( |
117 |
id => $mfn, |
118 |
ds => $ds, |
119 |
type => $config->{hyperestraier}->{type}, |
120 |
); |
121 |
|
122 |
$total_rows++; |
123 |
|
124 |
}; |
125 |
|
126 |
$log->info("$total_rows records indexed"); |
127 |
} |
128 |
|