Lines 48-53
use URI::Escape;
Link Here
|
48 |
|
48 |
|
49 |
use C4::Context; |
49 |
use C4::Context; |
50 |
use Koha::Exceptions; |
50 |
use Koha::Exceptions; |
|
|
51 |
use Koha::Caches; |
51 |
|
52 |
|
52 |
=head2 build_query |
53 |
=head2 build_query |
53 |
|
54 |
|
Lines 90-98
sub build_query {
Link Here
|
90 |
query => $query, |
91 |
query => $query, |
91 |
fuzziness => $fuzzy_enabled ? 'auto' : '0', |
92 |
fuzziness => $fuzzy_enabled ? 'auto' : '0', |
92 |
default_operator => 'AND', |
93 |
default_operator => 'AND', |
93 |
default_field => '_all', |
94 |
fields => $self->_search_fields({ is_opac => $options{is_opac}, weighted_fields => $options{weighted_fields} }), |
94 |
lenient => JSON::true, |
95 |
lenient => JSON::true, |
95 |
fields => $options{fields} || [], |
|
|
96 |
} |
96 |
} |
97 |
}; |
97 |
}; |
98 |
|
98 |
|
Lines 231-247
sub build_query_compat {
Link Here
|
231 |
join( ' ', $self->_create_query_string(@search_params) ) || (), |
231 |
join( ' ', $self->_create_query_string(@search_params) ) || (), |
232 |
$self->_join_queries( $self->_convert_index_strings(@$limits) ) || () ); |
232 |
$self->_join_queries( $self->_convert_index_strings(@$limits) ) || () ); |
233 |
|
233 |
|
234 |
my @fields = '_all'; |
|
|
235 |
if ( defined($params->{weighted_fields}) && $params->{weighted_fields} ) { |
236 |
push @fields, sprintf("%s^%s", $_->name, $_->weight) for Koha::SearchFields->weighted_fields; |
237 |
} |
238 |
|
239 |
# If there's no query on the left, let's remove the junk left behind |
234 |
# If there's no query on the left, let's remove the junk left behind |
240 |
$query_str =~ s/^ AND //; |
235 |
$query_str =~ s/^ AND //; |
241 |
my %options; |
236 |
my %options; |
242 |
$options{fields} = \@fields; |
|
|
243 |
$options{sort} = \@sort_params; |
237 |
$options{sort} = \@sort_params; |
244 |
$options{expanded_facet} = $params->{expanded_facet}; |
238 |
$options{expanded_facet} = $params->{expanded_facet}; |
|
|
239 |
$options{is_opac} = $params->{is_opac}; |
240 |
$options{weighted_fields} = $params->{weighted_fields}; |
245 |
my $query = $self->build_query( $query_str, %options ); |
241 |
my $query = $self->build_query( $query_str, %options ); |
246 |
|
242 |
|
247 |
#die Dumper($query); |
243 |
#die Dumper($query); |
Lines 302-323
sub build_authorities_query {
Link Here
|
302 |
|
298 |
|
303 |
foreach my $s ( @{ $search->{searches} } ) { |
299 |
foreach my $s ( @{ $search->{searches} } ) { |
304 |
my ( $wh, $op, $val ) = @{$s}{qw(where operator value)}; |
300 |
my ( $wh, $op, $val ) = @{$s}{qw(where operator value)}; |
305 |
$wh = '_all' if $wh eq ''; |
301 |
if ( $op eq 'is' || $op eq '=' || $op eq 'exact') { |
306 |
if ( $op eq 'is' || $op eq '=' || $op eq 'exact' ) { |
302 |
if ($wh) { |
307 |
|
303 |
# Match the whole field, case insensitive, UTF normalized. |
308 |
# look for something that matches a term completely |
304 |
push @query_parts, { term => { "$wh.ci_raw" => $val } }; |
309 |
# note, '=' is about numerical vals. May need special handling. |
305 |
} |
310 |
# Also, we lowercase our search because the ES |
306 |
else { |
311 |
# index lowercases its values, and term searches don't get the |
307 |
# Match the whole field for all searchable fields, case insensitive, |
312 |
# search analyzer applied to them. |
308 |
# UTF normalized. |
313 |
push @query_parts, { match_phrase => {"$wh.phrase" => lc $val} }; |
309 |
# Given that field data is "The quick brown fox" |
|
|
310 |
# "The quick brown fox" and "the quick brown fox" will match |
311 |
# but not "quick brown fox". |
312 |
push @query_parts, { |
313 |
multi_match => { |
314 |
query => $val, |
315 |
fields => $self->_search_fields({ subfield => 'ci_raw' }), |
316 |
} |
317 |
}; |
318 |
} |
314 |
} |
319 |
} |
315 |
elsif ( $op eq 'start' ) { |
320 |
elsif ( $op eq 'start') { |
316 |
# startswith search, uses lowercase untokenized version of heading |
321 |
# Match the prefix within a field for all searchable fields. |
317 |
push @query_parts, { match_phrase_prefix => {"$wh.phrase" => lc $val} }; |
322 |
# Given that field data is "The quick brown fox" |
|
|
323 |
# "The quick bro" will match, but not "quick bro" |
324 |
|
325 |
# Does not seems to be a multi prefix query |
326 |
# so we need to create one |
327 |
if ($wh) { |
328 |
# Match prefix of the field. |
329 |
push @query_parts, { prefix => {"$wh.ci_raw" => $val} }; |
330 |
} |
331 |
else { |
332 |
my @prefix_queries; |
333 |
foreach my $field (@{$self->_search_fields()}) { |
334 |
push @prefix_queries, { |
335 |
prefix => { "$field.ci_raw" => $val } |
336 |
}; |
337 |
} |
338 |
push @query_parts, { |
339 |
'bool' => { |
340 |
'should' => \@prefix_queries, |
341 |
'minimum_should_match' => 1 |
342 |
} |
343 |
}; |
344 |
} |
318 |
} |
345 |
} |
319 |
else { |
346 |
else { |
320 |
# regular wordlist stuff |
347 |
# Query all searchable fields. |
|
|
348 |
# Given that field data is "The quick brown fox" |
349 |
# a search containing any of the words will match, regardless |
350 |
# of order. |
351 |
|
321 |
my @tokens = $self->_split_query( $val ); |
352 |
my @tokens = $self->_split_query( $val ); |
322 |
foreach my $token ( @tokens ) { |
353 |
foreach my $token ( @tokens ) { |
323 |
$token = $self->_truncate_terms( |
354 |
$token = $self->_truncate_terms( |
Lines 325-360
sub build_authorities_query {
Link Here
|
325 |
); |
356 |
); |
326 |
} |
357 |
} |
327 |
my $query = $self->_join_queries( @tokens ); |
358 |
my $query = $self->_join_queries( @tokens ); |
328 |
push @query_parts, { query_string => { default_field => $wh, query => $query } }; |
359 |
|
|
|
360 |
if ($wh) { |
361 |
push @query_parts, { query_string => { default_field => $wh, query => $query } }; |
362 |
} |
363 |
else { |
364 |
push @query_parts, { |
365 |
query_string => { |
366 |
query => $query, |
367 |
fields => $self->_search_fields(), |
368 |
} |
369 |
}; |
370 |
} |
329 |
} |
371 |
} |
330 |
} |
372 |
} |
331 |
|
373 |
|
332 |
# Merge the query parts appropriately |
374 |
# Merge the query parts appropriately |
333 |
# 'should' behaves like 'or' |
375 |
# 'should' behaves like 'or' |
334 |
# 'must' behaves like 'and' |
376 |
# 'must' behaves like 'and' |
335 |
# Zebra results seem to match must so using that here |
377 |
# Zebra behaviour seem to match must so using that here |
336 |
my $query = { query => |
378 |
my $elastic_query = {}; |
337 |
{ bool => |
379 |
$elastic_query->{bool}->{must} = \@query_parts; |
338 |
{ must => \@query_parts } |
380 |
|
339 |
} |
381 |
# Filter by authtypecode if set |
340 |
}; |
382 |
if ($search->{authtypecode}) { |
341 |
|
383 |
$elastic_query->{bool}->{filter} = { |
342 |
my %s; |
384 |
term => { |
343 |
if ( exists $search->{sort} ) { |
385 |
"authtype.raw" => $search->{authtypecode} |
344 |
foreach my $k ( keys %{ $search->{sort} } ) { |
386 |
} |
345 |
my $f = $self->_sort_field($k); |
387 |
}; |
346 |
$s{$f} = $search->{sort}{$k}; |
|
|
347 |
} |
348 |
$search->{sort} = \%s; |
349 |
} |
388 |
} |
350 |
|
389 |
|
351 |
# add the sort stuff |
390 |
my $query = { |
352 |
$query->{sort} = [ $search->{sort} ] if exists $search->{sort}; |
391 |
query => $elastic_query |
|
|
392 |
}; |
393 |
|
394 |
# Add the sort stuff |
395 |
$query->{sort} = [ $search->{sort} ] if exists $search->{sort}; |
353 |
|
396 |
|
354 |
return $query; |
397 |
return $query; |
355 |
} |
398 |
} |
356 |
|
399 |
|
357 |
|
|
|
358 |
=head2 build_authorities_query_compat |
400 |
=head2 build_authorities_query_compat |
359 |
|
401 |
|
360 |
my ($query) = |
402 |
my ($query) = |
Lines 452-459
sub build_authorities_query_compat {
Link Here
|
452 |
|
494 |
|
453 |
my %sort; |
495 |
my %sort; |
454 |
my $sort_field = |
496 |
my $sort_field = |
455 |
( $orderby =~ /^heading/ ) ? 'heading' |
497 |
( $orderby =~ /^heading/ ) ? 'heading__sort' |
456 |
: ( $orderby =~ /^auth/ ) ? 'local-number' |
498 |
: ( $orderby =~ /^auth/ ) ? 'local-number__sort' |
457 |
: undef; |
499 |
: undef; |
458 |
if ($sort_field) { |
500 |
if ($sort_field) { |
459 |
my $sort_order = ( $orderby =~ /asc$/ ) ? 'asc' : 'desc'; |
501 |
my $sort_order = ( $orderby =~ /asc$/ ) ? 'asc' : 'desc'; |
Lines 520-526
types.
Link Here
|
520 |
=cut |
562 |
=cut |
521 |
|
563 |
|
522 |
our %index_field_convert = ( |
564 |
our %index_field_convert = ( |
523 |
'kw' => '_all', |
565 |
'kw' => '', |
524 |
'ab' => 'abstract', |
566 |
'ab' => 'abstract', |
525 |
'au' => 'author', |
567 |
'au' => 'author', |
526 |
'lcn' => 'local-classification', |
568 |
'lcn' => 'local-classification', |
Lines 614-620
sub _convert_index_fields {
Link Here
|
614 |
# If a field starts with mc- we save it as it's used (and removed) later |
656 |
# If a field starts with mc- we save it as it's used (and removed) later |
615 |
# when joining things, to indicate we make it an 'OR' join. |
657 |
# when joining things, to indicate we make it an 'OR' join. |
616 |
# (Sorry, this got a bit ugly after special cases were found.) |
658 |
# (Sorry, this got a bit ugly after special cases were found.) |
617 |
grep { $_->{field} } map { |
659 |
map { |
618 |
# Lower case all field names |
660 |
# Lower case all field names |
619 |
my ( $f, $t ) = map(lc, split /,/); |
661 |
my ( $f, $t ) = map(lc, split /,/); |
620 |
my $mc = ''; |
662 |
my $mc = ''; |
Lines 627-633
sub _convert_index_fields {
Link Here
|
627 |
type => $index_type_convert{ $t // '__default' } |
669 |
type => $index_type_convert{ $t // '__default' } |
628 |
}; |
670 |
}; |
629 |
$r->{field} = ($mc . $r->{field}) if $mc && $r->{field}; |
671 |
$r->{field} = ($mc . $r->{field}) if $mc && $r->{field}; |
630 |
$r; |
672 |
$r->{field} ? $r : undef; |
631 |
} @indexes; |
673 |
} @indexes; |
632 |
} |
674 |
} |
633 |
|
675 |
|
Lines 656-663
sub _convert_index_strings {
Link Here
|
656 |
push @res, $s; |
698 |
push @res, $s; |
657 |
next; |
699 |
next; |
658 |
} |
700 |
} |
659 |
push @res, $conv->{field} . ":" |
701 |
push @res, ($conv->{field} ? $conv->{field} . ':' : '') |
660 |
. $self->_modify_string_by_type( %$conv, operand => $term ); |
702 |
. $self->_modify_string_by_type( %$conv, operand => $term ); |
661 |
} |
703 |
} |
662 |
return @res; |
704 |
return @res; |
663 |
} |
705 |
} |
Lines 945-948
sub _split_query {
Link Here
|
945 |
return @tokens; |
987 |
return @tokens; |
946 |
} |
988 |
} |
947 |
|
989 |
|
|
|
990 |
=head2 _search_fields |
991 |
my $weighted_fields = $self->_search_fields({ |
992 |
is_opac => 0, |
993 |
weighted_fields => 1, |
994 |
subfield => 'raw' |
995 |
}); |
996 |
|
997 |
Generate a list of searchable fields to be used for Elasticsearch queries |
998 |
applied to multiple fields. |
999 |
|
1000 |
Returns an arrayref of field names for either OPAC or Staff client, with |
1001 |
possible weights and subfield appended to each field name depending on the |
1002 |
options provided. |
1003 |
|
1004 |
=over 4 |
1005 |
|
1006 |
=item C<$params> |
1007 |
|
1008 |
Hashref with options. The parameter C<is_opac> indicates whether the searchable |
1009 |
fields for OPAC or Staff client should be retrieved. If C<weighted_fields> is set |
1010 |
fields weights will be applied on returned fields. C<subfield> can be used to |
1011 |
provide a subfield that will be appended to fields as "C<field_name>.C<subfield>". |
1012 |
|
1013 |
=back |
1014 |
|
1015 |
=cut |
1016 |
|
1017 |
sub _search_fields { |
1018 |
my ($self, $params) = @_; |
1019 |
$params //= { |
1020 |
is_opac => 0, |
1021 |
weighted_fields => 0, |
1022 |
# This is a hack for authorities build_authorities_query |
1023 |
# can hopefully be removed in the future |
1024 |
subfield => undef, |
1025 |
}; |
1026 |
|
1027 |
my $cache = Koha::Caches->get_instance(); |
1028 |
my $cache_key = 'elasticsearch_search_fields' . ($params->{is_opac} ? '_opac' : '_staff_client'); |
1029 |
my $search_fields = $cache->get_from_cache($cache_key, { unsafe => 1 }); |
1030 |
|
1031 |
if (!$search_fields) { |
1032 |
# The reason we don't use Koha::SearchFields->search here is we don't |
1033 |
# want or need resultset wrapped as Koha::SearchField object. |
1034 |
# It does not make any sense in this context and would cause |
1035 |
# unnecessary overhead sice we are only querying for data |
1036 |
# Also would not work, or produce strange results, with the "columns" |
1037 |
# option. |
1038 |
my $schema = Koha::Database->schema; |
1039 |
my $result = $schema->resultset('SearchField')->search( |
1040 |
{ |
1041 |
$params->{is_opac} ? ( |
1042 |
'opac' => 1, |
1043 |
) : ( |
1044 |
'staff_client' => 1 |
1045 |
), |
1046 |
'search_marc_map.index_name' => $self->index, |
1047 |
'search_marc_map.marc_type' => C4::Context->preference('marcflavour'), |
1048 |
'search_marc_to_fields.search' => 1, |
1049 |
}, |
1050 |
{ |
1051 |
columns => [qw/name weight/], |
1052 |
collapse => 1, |
1053 |
join => {search_marc_to_fields => 'search_marc_map'}, |
1054 |
} |
1055 |
); |
1056 |
my @search_fields; |
1057 |
while (my $search_field = $result->next) { |
1058 |
push @search_fields, [ |
1059 |
$search_field->name, |
1060 |
$search_field->weight ? $search_field->weight : () |
1061 |
]; |
1062 |
} |
1063 |
$search_fields = \@search_fields; |
1064 |
$cache->set_in_cache($cache_key, $search_fields); |
1065 |
} |
1066 |
if ($params->{subfield}) { |
1067 |
my $subfield = $params->{subfield}; |
1068 |
$search_fields = [ |
1069 |
map { |
1070 |
# Copy values to avoid mutating cached |
1071 |
# data (since unsafe is used) |
1072 |
my ($field, $weight) = @{$_}; |
1073 |
["${field}.${subfield}", $weight]; |
1074 |
} @{$search_fields} |
1075 |
]; |
1076 |
} |
1077 |
|
1078 |
if ($params->{weighted_fields}) { |
1079 |
return [map { join('^', @{$_}) } @{$search_fields}]; |
1080 |
} |
1081 |
else { |
1082 |
# Exclude weight from field |
1083 |
return [map { $_->[0] } @{$search_fields}]; |
1084 |
} |
1085 |
} |
1086 |
|
948 |
1; |
1087 |
1; |