Lines 34-39
use Search::Elasticsearch;
Link Here
|
34 |
use Try::Tiny; |
34 |
use Try::Tiny; |
35 |
use YAML::Syck; |
35 |
use YAML::Syck; |
36 |
|
36 |
|
|
|
37 |
use List::Util qw( sum0 ); |
38 |
use Search::Elasticsearch; |
39 |
use MARC::File::XML; |
40 |
|
37 |
use Data::Dumper; # TODO remove |
41 |
use Data::Dumper; # TODO remove |
38 |
|
42 |
|
39 |
__PACKAGE__->mk_ro_accessors(qw( index )); |
43 |
__PACKAGE__->mk_ro_accessors(qw( index )); |
Lines 69-74
sub new {
Link Here
|
69 |
return $self; |
73 |
return $self; |
70 |
} |
74 |
} |
71 |
|
75 |
|
|
|
76 |
sub get_elasticsearch { |
77 |
my $self = shift @_; |
78 |
unless (defined $self->{elasticsearch}) { |
79 |
my $conf = $self->get_elasticsearch_params(); |
80 |
$self->{elasticsearch} = Search::Elasticsearch->new( |
81 |
client => "5_0::Direct", |
82 |
nodes => $conf->{nodes}, |
83 |
cxn_pool => 'Sniff' |
84 |
); |
85 |
} |
86 |
return $self->{elasticsearch}; |
87 |
} |
88 |
|
72 |
=head2 get_elasticsearch_params |
89 |
=head2 get_elasticsearch_params |
73 |
|
90 |
|
74 |
my $params = $self->get_elasticsearch_params(); |
91 |
my $params = $self->get_elasticsearch_params(); |
Lines 180-185
sub get_elasticsearch_mappings {
Link Here
|
180 |
include_in_all => JSON::false, |
197 |
include_in_all => JSON::false, |
181 |
type => "text", |
198 |
type => "text", |
182 |
}, |
199 |
}, |
|
|
200 |
marc_xml => { |
201 |
store => JSON::true, |
202 |
analyzer => "keyword", |
203 |
index => JSON::false, |
204 |
include_in_all => JSON::false, |
205 |
type => "text", |
206 |
}, |
183 |
} |
207 |
} |
184 |
} |
208 |
} |
185 |
}; |
209 |
}; |
Lines 312-317
sub sort_fields {
Link Here
|
312 |
return $self->_sort_fields_accessor(); |
336 |
return $self->_sort_fields_accessor(); |
313 |
} |
337 |
} |
314 |
|
338 |
|
|
|
339 |
sub marc_records_to_documents { |
340 |
my ($self, $records) = @_; |
341 |
my $rules = $self->get_marc_mapping_rules(); |
342 |
my $control_fields_rules = $rules->{control_fields}; |
343 |
my $data_fields_rules = $rules->{data_fields}; |
344 |
my $marcflavour = lc C4::Context->preference('marcflavour'); |
345 |
|
346 |
my @record_documents; |
347 |
|
348 |
sub _process_mappings { |
349 |
my ($mappings, $data, $record_document) = @_; |
350 |
foreach my $mapping (@{$mappings}) { |
351 |
my ($target, $options) = @{$mapping}; |
352 |
my $_data = $data; |
353 |
$record_document->{$target} //= []; |
354 |
if ($options->{substr}) { |
355 |
my ($offset, $length) = @{$options->{substr}}; |
356 |
$_data = substr $data, $offset, $length; |
357 |
} |
358 |
if ($options->{property}) { |
359 |
$_data = { |
360 |
$options->{property} => $_data |
361 |
} |
362 |
} |
363 |
push @{$record_document->{$target}}, $_data; |
364 |
} |
365 |
} |
366 |
foreach my $record (@{$records}) { |
367 |
my $record_document = {}; |
368 |
my $mappings = $rules->{leader}; |
369 |
if ($mappings) { |
370 |
_process_mappings($mappings, $record->leader(), $record_document); |
371 |
} |
372 |
foreach my $field ($record->fields()) { |
373 |
if($field->is_control_field()) { |
374 |
my $mappings = $control_fields_rules->{$field->tag()}; |
375 |
if ($mappings) { |
376 |
_process_mappings($mappings, $field->data(), $record_document); |
377 |
} |
378 |
} |
379 |
else { |
380 |
my $subfields_mappings = $data_fields_rules->{$field->tag()}; |
381 |
if ($subfields_mappings) { |
382 |
my $wildcard_mappings = $subfields_mappings->{'*'}; |
383 |
foreach my $subfield ($field->subfields()) { |
384 |
my ($code, $data) = @{$subfield}; |
385 |
my $mappings = $subfields_mappings->{$code} // []; |
386 |
if ($wildcard_mappings) { |
387 |
$mappings = [@{$mappings}, @{$wildcard_mappings}]; |
388 |
} |
389 |
if (@{$mappings}) { |
390 |
_process_mappings($mappings, $data, $record_document); |
391 |
} |
392 |
} |
393 |
} |
394 |
} |
395 |
} |
396 |
foreach my $field (keys %{$rules->{defaults}}) { |
397 |
unless (defined $record_document->{$field}) { |
398 |
$record_document->{$field} = $rules->{defaults}->{$field}; |
399 |
} |
400 |
} |
401 |
foreach my $field (@{$rules->{sum}}) { |
402 |
if (defined $record_document->{$field}) { |
403 |
# TODO: validate numeric? filter? |
404 |
# TODO: Or should only accept fields without nested values? |
405 |
# TODO: Quick and dirty, improve if needed |
406 |
$record_document->{$field} = sum0(grep { ref($_) eq 'SCALAR' && m/\d+([.,]\d+)?/} @{$record_document->{$field}}); |
407 |
} |
408 |
} |
409 |
# TODO: Perhaps should check if $records_document non empty, but really should never be the case |
410 |
$record->encoding('UTF-8'); |
411 |
$record_document->{'marc_xml'} = $record->as_xml_record($marcflavour); |
412 |
my $id = $record->subfield('999', 'c'); |
413 |
push @record_documents, [$id, $record_document]; |
414 |
} |
415 |
return \@record_documents; |
416 |
} |
417 |
|
418 |
# Provides the rules for marc to Elasticsearch JSON document conversion. |
419 |
sub get_marc_mapping_rules { |
420 |
my ($self) = @_; |
421 |
|
422 |
my $marcflavour = lc C4::Context->preference('marcflavour'); |
423 |
my @rules; |
424 |
|
425 |
sub _field_mappings { |
426 |
my ($facet, $suggestible, $sort, $target_name, $target_type, $range) = @_; |
427 |
my %mapping_defaults = (); |
428 |
my @mappings; |
429 |
|
430 |
my $substr_args = undef; |
431 |
if ($range) { |
432 |
my ($offset, $end) = map(int, split /-/, $range, 2); |
433 |
$substr_args = [$offset]; |
434 |
push @{$substr_args}, (defined $end ? $end - $offset : 1); |
435 |
} |
436 |
my $default_options = {}; |
437 |
if ($substr_args) { |
438 |
$default_options->{substr} = $substr_args; |
439 |
} |
440 |
|
441 |
my $mapping = [$target_name, $default_options]; |
442 |
push @mappings, $mapping; |
443 |
|
444 |
my @suffixes = (); |
445 |
push @suffixes, 'facet' if $facet; |
446 |
push @suffixes, 'suggestion' if $suggestible; # Check condition, also if undef? |
447 |
push @suffixes, 'sort' if $sort; |
448 |
foreach my $suffix (@suffixes) { |
449 |
my $mapping = ["${target_name}__$suffix"]; |
450 |
# Hack, fix later in less hideous manner |
451 |
if ($suffix eq 'suggestion') { |
452 |
push @{$mapping}, {%{$default_options}, property => 'input'}; |
453 |
} |
454 |
else { |
455 |
push @{$mapping}, $default_options; |
456 |
} |
457 |
push @mappings, $mapping; |
458 |
} |
459 |
return @mappings; |
460 |
}; |
461 |
my $field_spec_regexp = qr/^([0-9]{3})([0-9a-z]+)?(?:_\/(\d+(?:-\d+)?))?$/; |
462 |
my $leader_regexp = qr/^leader(?:_\/(\d+(?:-\d+)?))?$/; |
463 |
my $rules = { |
464 |
'leader' => [], |
465 |
'control_fields' => {}, |
466 |
'data_fields' => {}, |
467 |
'sum' => [], |
468 |
'defaults' => {} |
469 |
}; |
470 |
|
471 |
$self->_foreach_mapping(sub { |
472 |
my ( $name, $type, $facet, $suggestible, $sort, $marc_type, $marc_field ) = @_; |
473 |
return if $marc_type ne $marcflavour; |
474 |
|
475 |
if ($type eq 'sum') { |
476 |
push @{$rules->{sum}}, $name; |
477 |
} |
478 |
elsif($type eq 'boolean') { |
479 |
# boolean gets special handling, if value doesn't exist for a field, |
480 |
# it is set to false |
481 |
$rules->{defaults}->{$name} = 0; |
482 |
} |
483 |
|
484 |
if ($marc_field =~ $field_spec_regexp) { |
485 |
my $field_tag = $1; |
486 |
my $subfields = defined $2 ? $2 : '*'; |
487 |
my $range = defined $3 ? $3 : undef; |
488 |
if ($field_tag < 10) { |
489 |
$rules->{control_fields}->{$field_tag} //= []; |
490 |
my @mappings = _field_mappings($facet, $suggestible, $sort, $name, $type, $range); |
491 |
push @{$rules->{control_fields}->{$field_tag}}, @mappings; |
492 |
} |
493 |
else { |
494 |
$rules->{data_fields}->{$field_tag} //= {}; |
495 |
foreach my $subfield (split //, $subfields) { |
496 |
$rules->{data_fields}->{$field_tag}->{$subfield} //= []; |
497 |
my @mappings = _field_mappings($facet, $suggestible, $sort, $name, $type, $range); |
498 |
push @{$rules->{data_fields}->{$field_tag}->{$subfield}}, @mappings; |
499 |
} |
500 |
} |
501 |
} |
502 |
elsif ($marc_field =~ $leader_regexp) { |
503 |
my $range = defined $1 ? $1 : undef; |
504 |
my @mappings = _field_mappings($facet, $suggestible, $sort, $name, $type, $range); |
505 |
push @{$rules->{leader}}, @mappings; |
506 |
} |
507 |
else { |
508 |
die("Invalid marc field: $marc_field"); |
509 |
} |
510 |
}); |
511 |
return $rules; |
512 |
} |
513 |
|
315 |
# Provides the rules for data conversion. |
514 |
# Provides the rules for data conversion. |
316 |
sub get_fixer_rules { |
515 |
sub get_fixer_rules { |
317 |
my ($self) = @_; |
516 |
my ($self) = @_; |