4 |
|
|
5 |
use Cwd qw/abs_path/; |
use Cwd qw/abs_path/; |
6 |
use File::Temp qw/tempdir/; |
use File::Temp qw/tempdir/; |
|
use Data::Dumper; |
|
7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Lookup; |
use WebPAC::Lookup 0.03; |
11 |
use WebPAC::Input 0.03; |
use WebPAC::Input 0.07; |
12 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.03; |
13 |
use WebPAC::Normalize; |
use WebPAC::Normalize 0.11; |
14 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
15 |
use WebPAC::Validate; |
use WebPAC::Validate; |
16 |
|
use WebPAC::Output::MARC; |
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; |
use File::Slurp; |
22 |
use MARC::Record 2.0; # need 2.0 for utf-8 encoding see marcpm.sf.net |
use Data::Dump qw/dump/; |
23 |
|
use Storable qw/dclone/; |
24 |
|
|
25 |
=head1 NAME |
=head1 NAME |
26 |
|
|
64 |
|
|
65 |
turn on extra validation of imput records, see L<WebPAC::Validation> |
turn on extra validation of imput records, see L<WebPAC::Validation> |
66 |
|
|
67 |
|
=item --marc-normalize conf/normalize/mapping.pl |
68 |
|
|
69 |
|
This option specifies normalisation file for MARC creation |
70 |
|
|
71 |
|
=item --marc-output out/marc/test.marc |
72 |
|
|
73 |
|
Optional path to output file |
74 |
|
|
75 |
|
=item --marc-lint |
76 |
|
|
77 |
|
By default turned on if C<--marc-normalize> is used. You can disable lint |
78 |
|
messages with C<--no-marc-lint>. |
79 |
|
|
80 |
|
=item --marc-dump |
81 |
|
|
82 |
|
Force dump or input and marc record for debugging. |
83 |
|
|
84 |
=back |
=back |
85 |
|
|
86 |
=cut |
=cut |
94 |
my $only_filter; |
my $only_filter; |
95 |
my $stats = 0; |
my $stats = 0; |
96 |
my $validate_path; |
my $validate_path; |
97 |
|
my ($marc_normalize, $marc_output); |
98 |
|
my $marc_lint = 1; |
99 |
|
my $marc_dump = 0; |
100 |
|
|
101 |
GetOptions( |
GetOptions( |
102 |
"limit=i" => \$limit, |
"limit=i" => \$limit, |
105 |
"one=s" => \$only_filter, |
"one=s" => \$only_filter, |
106 |
"only=s" => \$only_filter, |
"only=s" => \$only_filter, |
107 |
"config" => \$config, |
"config" => \$config, |
108 |
"debug" => \$debug, |
"debug+" => \$debug, |
109 |
"stats" => \$stats, |
"stats" => \$stats, |
110 |
"validate=s" => \$validate_path, |
"validate=s" => \$validate_path, |
111 |
|
"marc-normalize=s" => \$marc_normalize, |
112 |
|
"marc-output=s" => \$marc_output, |
113 |
|
"marc-lint!" => \$marc_lint, |
114 |
|
"marc-dump!" => \$marc_dump, |
115 |
); |
); |
116 |
|
|
117 |
$config = LoadFile($config); |
$config = LoadFile($config); |
118 |
|
|
119 |
print "config = ",Dumper($config) if ($debug); |
print "config = ",dump($config) if ($debug); |
120 |
|
|
121 |
die "no databases in config file!\n" unless ($config->{databases}); |
die "no databases in config file!\n" unless ($config->{databases}); |
122 |
|
|
136 |
$log->info("using $use_indexer indexing engine..."); |
$log->info("using $use_indexer indexing engine..."); |
137 |
} |
} |
138 |
|
|
139 |
|
# disable indexing when creating marc |
140 |
|
$use_indexer = undef if ($marc_normalize); |
141 |
|
|
142 |
my $total_rows = 0; |
my $total_rows = 0; |
143 |
my $start_t = time(); |
my $start_t = time(); |
144 |
|
|
222 |
|
|
223 |
die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs)); |
die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs)); |
224 |
|
|
225 |
my $lookup = new WebPAC::Lookup( |
my $lookup; |
226 |
lookup_file => $input->{lookup}, |
if ($input->{lookup}) { |
227 |
) if ($input->{lookup}); |
$lookup = new WebPAC::Lookup( |
228 |
|
lookup_file => $input->{lookup}, |
229 |
|
); |
230 |
|
delete( $input->{lookup} ); |
231 |
|
} |
232 |
|
|
233 |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
234 |
|
|
238 |
|
|
239 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
240 |
module => $input_module, |
module => $input_module, |
241 |
code_page => $config->{webpac}->{webpac_encoding}, |
encoding => $config->{webpac}->{webpac_encoding}, |
242 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
243 |
offset => $offset, |
offset => $offset, |
244 |
lookup => $lookup, |
lookup_coderef => sub { |
245 |
|
my $rec = shift || return; |
246 |
|
$lookup->add( $rec ); |
247 |
|
}, |
248 |
recode => $input->{recode}, |
recode => $input->{recode}, |
249 |
stats => $stats, |
stats => $stats, |
250 |
|
modify_records => $input->{modify_records}, |
251 |
); |
); |
252 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
253 |
|
|
260 |
my @norm_array = ref($input->{normalize}) eq 'ARRAY' ? |
my @norm_array = ref($input->{normalize}) eq 'ARRAY' ? |
261 |
@{ $input->{normalize} } : ( $input->{normalize} ); |
@{ $input->{normalize} } : ( $input->{normalize} ); |
262 |
|
|
263 |
|
if ($marc_normalize) { |
264 |
|
@norm_array = ( { |
265 |
|
path => $marc_normalize, |
266 |
|
output => $marc_output || 'out/marc/' . $database . '-' . $input->{name} . '.marc', |
267 |
|
} ); |
268 |
|
} |
269 |
|
|
270 |
foreach my $normalize (@norm_array) { |
foreach my $normalize (@norm_array) { |
271 |
|
|
|
my $rules; |
|
272 |
my $normalize_path = $normalize->{path} || $log->logdie("can't find normalize path in config"); |
my $normalize_path = $normalize->{path} || $log->logdie("can't find normalize path in config"); |
273 |
|
|
274 |
$log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i ); |
$log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i ); |
277 |
|
|
278 |
$log->info("Using $normalize_path for normalization..."); |
$log->info("Using $normalize_path for normalization..."); |
279 |
|
|
280 |
my $marc_fh; |
my $marc = new WebPAC::Output::MARC( |
281 |
if (my $path = $normalize->{marc21}) { |
path => $normalize->{output}, |
282 |
open($marc_fh, '>', $path) || |
lint => $marc_lint, |
283 |
$log->logdie("can't open MARC output $path: $!"); |
dump => $marc_dump, |
284 |
|
) if ($normalize->{output}); |
|
$log->info("Creating MARC export file $path\n"); |
|
|
} |
|
285 |
|
|
286 |
# reset position in database |
# reset position in database |
287 |
$input_db->seek(1); |
$input_db->seek(1); |
304 |
$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors); |
$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors); |
305 |
} |
} |
306 |
|
|
307 |
|
my $ds_config = dclone($db_config); |
308 |
|
|
309 |
|
# default values -> database key |
310 |
|
$ds_config->{_} = $database; |
311 |
|
|
312 |
|
# current mfn |
313 |
|
$ds_config->{_mfn} = $mfn; |
314 |
|
|
315 |
|
# attach current input |
316 |
|
$ds_config->{input} = $input; |
317 |
|
|
318 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
319 |
row => $row, |
row => $row, |
320 |
rules => $rules, |
rules => $rules, |
321 |
lookup => $lookup ? $lookup->lookup_hash : undef, |
lookup => $lookup ? $lookup->lookup_hash : undef, |
322 |
|
config => $ds_config, |
323 |
marc_encoding => 'utf-8', |
marc_encoding => 'utf-8', |
324 |
); |
); |
325 |
|
|
335 |
type => $config->{$use_indexer}->{type}, |
type => $config->{$use_indexer}->{type}, |
336 |
) if ($indexer && $ds); |
) if ($indexer && $ds); |
337 |
|
|
338 |
if ($marc_fh) { |
if ($marc) { |
339 |
my $marc = new MARC::Record; |
my $i = 0; |
340 |
$marc->encoding( 'utf-8' ); |
|
341 |
$marc->add_fields( WebPAC::Normalize::_get_marc21_fields() ); |
while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) { |
342 |
print $marc_fh $marc->as_usmarc; |
$marc->add( |
343 |
|
id => $mfn . ( $i ? "/$i" : '' ), |
344 |
|
fields => $fields, |
345 |
|
leader => WebPAC::Normalize::marc_leader(), |
346 |
|
row => $row, |
347 |
|
); |
348 |
|
$i++; |
349 |
|
} |
350 |
|
|
351 |
|
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
352 |
} |
} |
353 |
|
|
354 |
$total_rows++; |
$total_rows++; |
357 |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
358 |
|
|
359 |
# close MARC file |
# close MARC file |
360 |
close($marc_fh) if ($marc_fh); |
$marc->finish if ($marc); |
361 |
|
|
362 |
} |
} |
363 |
|
|
375 |
# |
# |
376 |
# add Hyper Estraier links to other databases |
# add Hyper Estraier links to other databases |
377 |
# |
# |
378 |
if (ref($db_config->{links}) eq 'ARRAY') { |
if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) { |
379 |
foreach my $link (@{ $db_config->{links} }) { |
foreach my $link (@{ $db_config->{links} }) { |
380 |
if ($use_indexer eq 'hyperestraier') { |
if ($use_indexer eq 'hyperestraier') { |
381 |
$log->info("saving link $database -> $link->{to} [$link->{credit}]"); |
$log->info("saving link $database -> $link->{to} [$link->{credit}]"); |