12 |
use WebPAC::Input 0.03; |
use WebPAC::Input 0.03; |
13 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.03; |
14 |
use WebPAC::Normalize::XML; |
use WebPAC::Normalize::XML; |
15 |
|
use WebPAC::Normalize::Set; |
16 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
17 |
use YAML qw/LoadFile/; |
use YAML qw/LoadFile/; |
18 |
use Getopt::Long; |
use Getopt::Long; |
19 |
use File::Path; |
use File::Path; |
20 |
use Time::HiRes qw/time/; |
use Time::HiRes qw/time/; |
21 |
|
use File::Slurp; |
22 |
|
|
23 |
=head1 NAME |
=head1 NAME |
24 |
|
|
50 |
|
|
51 |
path to YAML configuration file |
path to YAML configuration file |
52 |
|
|
53 |
|
=item --force-set |
54 |
|
|
55 |
|
force conversion C<normalize->path> in C<config.yml> from |
56 |
|
C<.xml> to C<.pl> |
57 |
|
|
58 |
=back |
=back |
59 |
|
|
60 |
=cut |
=cut |
66 |
my $config = 'conf/config.yml'; |
my $config = 'conf/config.yml'; |
67 |
my $debug = 0; |
my $debug = 0; |
68 |
my $only_db_name; |
my $only_db_name; |
69 |
|
my $force_set = 0; |
70 |
|
|
71 |
GetOptions( |
GetOptions( |
72 |
"limit=i" => \$limit, |
"limit=i" => \$limit, |
76 |
"only=s" => \$only_db_name, |
"only=s" => \$only_db_name, |
77 |
"config" => \$config, |
"config" => \$config, |
78 |
"debug" => \$debug, |
"debug" => \$debug, |
79 |
|
"force-set" => \$force_set, |
80 |
); |
); |
81 |
|
|
82 |
$config = LoadFile($config); |
$config = LoadFile($config); |
200 |
prefix => $input->{name}, |
prefix => $input->{name}, |
201 |
); |
); |
202 |
|
|
203 |
|
my $rules; |
204 |
my $normalize_path = $input->{normalize}->{path}; |
my $normalize_path = $input->{normalize}->{path}; |
205 |
|
|
206 |
|
if ($force_set) { |
207 |
|
my $new_norm_path = $normalize_path; |
208 |
|
$new_norm_path =~ s/\.xml$/.pl/; |
209 |
|
if (-e $new_norm_path) { |
210 |
|
$log->info("--force-set replaced $normalize_path with $new_norm_path"); |
211 |
|
$normalize_path = $new_norm_path; |
212 |
|
} else { |
213 |
|
$log->warn("--force-set failed on $new_norm_path, fallback to $normalize_path"); |
214 |
|
} |
215 |
|
} |
216 |
|
|
217 |
if ($normalize_path =~ m/\.xml$/i) { |
if ($normalize_path =~ m/\.xml$/i) { |
218 |
$n->open( |
$n->open( |
219 |
tag => $input->{normalize}->{tag}, |
tag => $input->{normalize}->{tag}, |
220 |
xml_file => $input->{normalize}->{path}, |
xml_file => $normalize_path, |
221 |
); |
); |
222 |
} elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) { |
} elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) { |
223 |
$n->open_yaml( |
$n->open_yaml( |
224 |
path => $normalize_path, |
path => $normalize_path, |
225 |
tag => $input->{normalize}->{tag}, |
tag => $input->{normalize}->{tag}, |
226 |
); |
); |
227 |
|
} elsif ($normalize_path =~ m/\.(?:pl)$/i) { |
228 |
|
$n = undef; |
229 |
|
$log->info("using WebPAC::Normalize::Set to process $normalize_path"); |
230 |
|
$rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!"; |
231 |
} |
} |
232 |
|
|
233 |
foreach my $pos ( 0 ... $input_db->size ) { |
foreach my $pos ( 0 ... $input_db->size ) { |
242 |
push @{ $row->{'000'} }, $pos; |
push @{ $row->{'000'} }, $pos; |
243 |
} |
} |
244 |
|
|
245 |
my $ds = $n->data_structure($row); |
my $ds = $n ? $n->data_structure($row) : |
246 |
|
WebPAC::Normalize::Set::data_structure( |
247 |
|
row => $row, |
248 |
|
rules => $rules, |
249 |
|
lookup => $lookup->lookup_hash, |
250 |
|
); |
251 |
|
|
252 |
$indexer->add( |
$indexer->add( |
253 |
id => $input->{name} . "/" . $mfn, |
id => $input->{name} . "/" . $mfn, |