|
Lines 135-179
sub build_query {
Link Here
|
| 135 |
return $res; |
135 |
return $res; |
| 136 |
} |
136 |
} |
| 137 |
|
137 |
|
| 138 |
=head2 build_browse_query |
|
|
| 139 |
|
| 140 |
my $browse_query = $builder->build_browse_query($field, $query); |
| 141 |
|
| 142 |
This performs a "starts with" style query on a particular field. The field |
| 143 |
to be searched must have been indexed with an appropriate mapping as a |
| 144 |
"phrase" subfield, which pretty much everything has. |
| 145 |
|
| 146 |
=cut |
| 147 |
|
| 148 |
# XXX this isn't really a browse query like we want in the end |
| 149 |
sub build_browse_query { |
| 150 |
my ( $self, $field, $query ) = @_; |
| 151 |
|
| 152 |
my $fuzzy_enabled = C4::Context->preference("QueryFuzzy") || 0; |
| 153 |
|
| 154 |
return { query => '*' } if !defined $query; |
| 155 |
|
| 156 |
# TODO this should come from Koha::SearchEngine::Elasticsearch |
| 157 |
my %field_whitelist = ( |
| 158 |
title => 1, |
| 159 |
author => 1, |
| 160 |
); |
| 161 |
$field = 'title' if !exists $field_whitelist{$field}; |
| 162 |
my $sort = $self->_sort_field($field); |
| 163 |
my $res = { |
| 164 |
query => { |
| 165 |
match_phrase_prefix => { |
| 166 |
"$field.phrase" => { |
| 167 |
query => $query, |
| 168 |
operator => 'or', |
| 169 |
fuzziness => $fuzzy_enabled ? 'auto' : '0', |
| 170 |
} |
| 171 |
} |
| 172 |
}, |
| 173 |
sort => [ { $sort => { order => "asc" } } ], |
| 174 |
}; |
| 175 |
} |
| 176 |
|
| 177 |
=head2 build_query_compat |
138 |
=head2 build_query_compat |
| 178 |
|
139 |
|
| 179 |
my ( |
140 |
my ( |
|
Lines 199-252
sub build_query_compat {
Link Here
|
| 199 |
$lang, $params ) |
160 |
$lang, $params ) |
| 200 |
= @_; |
161 |
= @_; |
| 201 |
|
162 |
|
| 202 |
#die Dumper ( $self, $operators, $operands, $indexes, $orig_limits, $sort_by, $scan, $lang ); |
163 |
my $query; |
| 203 |
my @sort_params = $self->_convert_sort_fields(@$sort_by); |
164 |
my $query_str = ''; |
| 204 |
my @index_params = $self->_convert_index_fields(@$indexes); |
165 |
my $search_param_query_str = ''; |
| 205 |
my $limits = $self->_fix_limit_special_cases($orig_limits); |
166 |
my $limits = (); |
| 206 |
if ( $params->{suppress} ) { push @$limits, "suppress:0"; } |
167 |
if ( $scan ) { |
| 207 |
|
168 |
($query, $query_str) = $self->_build_scan_query( $operands, $indexes ); |
| 208 |
# Merge the indexes in with the search terms and the operands so that |
169 |
$search_param_query_str = $query_str; |
| 209 |
# each search thing is a handy unit. |
170 |
} else { |
| 210 |
unshift @$operators, undef; # The first one can't have an op |
171 |
my @sort_params = $self->_convert_sort_fields(@$sort_by); |
| 211 |
my @search_params; |
172 |
my @index_params = $self->_convert_index_fields(@$indexes); |
| 212 |
my $truncate = C4::Context->preference("QueryAutoTruncate") || 0; |
173 |
my $limits = $self->_fix_limit_special_cases($orig_limits); |
| 213 |
my $ea = each_array( @$operands, @$operators, @index_params ); |
174 |
if ( $params->{suppress} ) { push @$limits, "suppress:0"; } |
| 214 |
while ( my ( $oand, $otor, $index ) = $ea->() ) { |
175 |
|
| 215 |
next if ( !defined($oand) || $oand eq '' ); |
176 |
# Merge the indexes in with the search terms and the operands so that |
| 216 |
$oand = $self->_clean_search_term($oand); |
177 |
# each search thing is a handy unit. |
| 217 |
$oand = $self->_truncate_terms($oand) if ($truncate); |
178 |
unshift @$operators, undef; # The first one can't have an op |
| 218 |
push @search_params, { |
179 |
my @search_params; |
| 219 |
operand => $oand, # the search terms |
180 |
my $truncate = C4::Context->preference("QueryAutoTruncate") || 0; |
| 220 |
operator => defined($otor) ? uc $otor : undef, # AND and so on |
181 |
my $ea = each_array( @$operands, @$operators, @index_params ); |
| 221 |
$index ? %$index : (), |
182 |
while ( my ( $oand, $otor, $index ) = $ea->() ) { |
| 222 |
}; |
183 |
next if ( !defined($oand) || $oand eq '' ); |
| 223 |
} |
184 |
$oand = $self->_clean_search_term($oand); |
|
|
185 |
$oand = $self->_truncate_terms($oand) if ($truncate); |
| 186 |
push @search_params, { |
| 187 |
operand => $oand, # the search terms |
| 188 |
operator => defined($otor) ? uc $otor : undef, # AND and so on |
| 189 |
$index ? %$index : (), |
| 190 |
}; |
| 191 |
} |
| 224 |
|
192 |
|
| 225 |
# We build a string query from limits and the queries. An alternative |
193 |
# We build a string query from limits and the queries. An alternative |
| 226 |
# would be to pass them separately into build_query and let it build |
194 |
# would be to pass them separately into build_query and let it build |
| 227 |
# them into a structured ES query itself. Maybe later, though that'd be |
195 |
# them into a structured ES query itself. Maybe later, though that'd be |
| 228 |
# more robust. |
196 |
# more robust. |
| 229 |
my $search_param_query_str = join( ' ', $self->_create_query_string(@search_params) ); |
197 |
$search_param_query_str = join( ' ', $self->_create_query_string(@search_params) ); |
| 230 |
my $query_str = join( ' AND ', |
198 |
my $query_str = join( ' AND ', |
| 231 |
$search_param_query_str || (), |
199 |
$search_param_query_str || (), |
| 232 |
$self->_join_queries( $self->_convert_index_strings(@$limits) ) || () ); |
200 |
$self->_join_queries( $self->_convert_index_strings(@$limits) ) || () ); |
| 233 |
|
201 |
|
| 234 |
my @fields = '_all'; |
202 |
my @fields = '_all'; |
| 235 |
if ( defined($params->{weighted_fields}) && $params->{weighted_fields} ) { |
203 |
if ( defined($params->{weighted_fields}) && $params->{weighted_fields} ) { |
| 236 |
push @fields, sprintf("%s^%s", $_->name, $_->weight) for Koha::SearchFields->weighted_fields; |
204 |
push @fields, sprintf("%s^%s", $_->name, $_->weight) for Koha::SearchFields->weighted_fields; |
| 237 |
} |
205 |
} |
| 238 |
|
206 |
|
| 239 |
# If there's no query on the left, let's remove the junk left behind |
207 |
# If there's no query on the left, let's remove the junk left behind |
| 240 |
$query_str =~ s/^ AND //; |
208 |
$query_str =~ s/^ AND //; |
| 241 |
my %options; |
209 |
my %options; |
| 242 |
$options{fields} = \@fields; |
210 |
$options{fields} = \@fields; |
| 243 |
$options{sort} = \@sort_params; |
211 |
$options{sort} = \@sort_params; |
| 244 |
my $query = $self->build_query( $query_str, %options ); |
212 |
$options{expanded_facet} = $params->{expanded_facet}; |
|
|
213 |
$query = $self->build_query( $query_str, %options ); |
| 214 |
} |
| 245 |
|
215 |
|
| 246 |
# We roughly emulate the CGI parameters of the zebra query builder |
216 |
# We roughly emulate the CGI parameters of the zebra query builder |
| 247 |
my $query_cgi = ''; |
217 |
my $query_cgi = ''; |
| 248 |
shift @$operators; # Shift out the one we unshifted before |
218 |
shift @$operators; # Shift out the one we unshifted before |
| 249 |
$ea = each_array( @$operands, @$operators, @$indexes ); |
219 |
my $ea = each_array( @$operands, @$operators, @$indexes ); |
| 250 |
while ( my ( $oand, $otor, $index ) = $ea->() ) { |
220 |
while ( my ( $oand, $otor, $index ) = $ea->() ) { |
| 251 |
$query_cgi .= '&' if $query_cgi; |
221 |
$query_cgi .= '&' if $query_cgi; |
| 252 |
$query_cgi .= 'idx=' . uri_escape_utf8( $index // '') . '&q=' . uri_escape_utf8( $oand ); |
222 |
$query_cgi .= 'idx=' . uri_escape_utf8( $index // '') . '&q=' . uri_escape_utf8( $oand ); |
|
Lines 480-485
sub build_authorities_query_compat {
Link Here
|
| 480 |
return $query; |
450 |
return $query; |
| 481 |
} |
451 |
} |
| 482 |
|
452 |
|
|
|
453 |
=head2 _build_scan_query |
| 454 |
|
| 455 |
my ($query, $query_str) = $builder->build_scan_query(\@operands, \@indexes) |
| 456 |
|
| 457 |
This will build an aggregation scan query that can be issued to elasticsearch from |
| 458 |
the provided string input. |
| 459 |
|
| 460 |
=cut |
| 461 |
|
| 462 |
our %scan_field_convert = ( |
| 463 |
'ti' => 'title', |
| 464 |
'au' => 'author', |
| 465 |
'su' => 'subject', |
| 466 |
'se' => 'title-series', |
| 467 |
'pb' => 'publisher', |
| 468 |
); |
| 469 |
|
| 470 |
sub _build_scan_query { |
| 471 |
my ( $self, $operands, $indexes ) = @_; |
| 472 |
|
| 473 |
my $term = scalar( @$operands ) == 0 ? '' : $operands->[0]; |
| 474 |
my $index = scalar( @$indexes ) == 0 ? 'title' : $indexes->[0]; |
| 475 |
|
| 476 |
my ( $f, $d ) = split( /,/, $index); |
| 477 |
$index = $scan_field_convert{$f} || $f; |
| 478 |
|
| 479 |
my $res; |
| 480 |
$res->{query} = { |
| 481 |
query_string => { |
| 482 |
query => '*' |
| 483 |
} |
| 484 |
}; |
| 485 |
$res->{aggregations} = { |
| 486 |
$index => { |
| 487 |
terms => { |
| 488 |
field => $index . '__facet', |
| 489 |
order => { '_term' => 'asc' }, |
| 490 |
include => $self->_create_regex_filter($self->_clean_search_term($term)) . '.*' |
| 491 |
} |
| 492 |
} |
| 493 |
}; |
| 494 |
return ($res, $term); |
| 495 |
} |
| 496 |
|
| 497 |
=head2 _create_regex_filter |
| 498 |
|
| 499 |
my $filter = $builder->create_regex_filter('term') |
| 500 |
|
| 501 |
This will create a regex filter that can be used with an aggregation query. |
| 502 |
|
| 503 |
=cut |
| 504 |
|
| 505 |
sub _create_regex_filter { |
| 506 |
my ($self, $term) = @_; |
| 507 |
|
| 508 |
my $result = ''; |
| 509 |
foreach my $c (split(//, quotemeta($term))) { |
| 510 |
my $lc = lc($c); |
| 511 |
my $uc = uc($c); |
| 512 |
$result .= $lc ne $uc ? '[' . $lc . $uc . ']' : $c; |
| 513 |
} |
| 514 |
return $result; |
| 515 |
} |
| 516 |
|
| 483 |
=head2 _convert_sort_fields |
517 |
=head2 _convert_sort_fields |
| 484 |
|
518 |
|
| 485 |
my @sort_params = _convert_sort_fields(@sort_by) |
519 |
my @sort_params = _convert_sort_fields(@sort_by) |
|
Lines 726-732
sub _modify_string_by_type {
Link Here
|
| 726 |
return $str unless $str; # Empty or undef, we can't use it. |
760 |
return $str unless $str; # Empty or undef, we can't use it. |
| 727 |
|
761 |
|
| 728 |
$str .= '*' if $type eq 'right-truncate'; |
762 |
$str .= '*' if $type eq 'right-truncate'; |
| 729 |
$str = '"' . $str . '"' if $type eq 'phrase'; |
763 |
$str = '"' . $str . '"' if $type eq 'phrase' && $str !~ /^".*"$/; |
| 730 |
if ($type eq 'st-year') { |
764 |
if ($type eq 'st-year') { |
| 731 |
if ($str =~ /^(.*)-(.*)$/) { |
765 |
if ($str =~ /^(.*)-(.*)$/) { |
| 732 |
my $from = $1 || '*'; |
766 |
my $from = $1 || '*'; |