Diacritical marks may exist in the patron record -- as they should, but this makes patron name search difficult
for librarians who may be unfamiliar with all the special characters used and also how to elicit them from a
keyboard. To ease this, accented characters will be converted into their 'plain' analogs for comparison purposes.
So, for example, if the patron's last name is Chávez, typing "Chavez" in the Last Name box in Patron Search will
match it.
Spaces in a name (like "De la Croix") will be squashed out so that name would be matched by "Delacroix" or "De la Croix" or variants.
The librarian can enter the accented characters or not.
Signed-off-by: Dan Pearl <dpearl@cwmars.org>
Signed-off-by:
Signed-off-by:
'Text::Aspell' => '0',
'Text::CSV' => '0',
'Text::Glob' => '0',
+ 'Text::Unaccent::PurePerl' => '0',
'Time::HiRes' => '0',
'Time::Local' => '0',
'Unicode::Normalize' => '0',
use OpenSRF::Utils qw/:datetime/;
use OpenILS::Utils::Fieldmapper;
use OpenSRF::Utils::SettingsClient;
+use OpenILS::Application::AppUtils;
use DateTime;
use DateTime::Format::ISO8601;
use DateTime::Set;
use DateTime::SpanSet;
+use Text::Unaccent::PurePerl qw(unac_string);
+
+my $apputils = "OpenILS::Application::AppUtils";
+my $U = $apputils;
+
my $_dt_parser = DateTime::Format::ISO8601->new;
my $log = 'OpenSRF::Utils::Logger';
NOTE
);
+sub _prepare_name_argument {
+ # Get rid of extra spaces, accents, and regex characters
+ my ($search) = _clean_regex_chars(@_);
+ $search =~ s/\s//g;
+ $search = Text::Unaccent::PurePerl::unac_string($search);
+
+ return $search;
+};
+
sub _clean_regex_chars {
my ($search) = @_;
# group 2 = phone, ident
# group 3 = barcode
- my $usr = join ' AND ', map { "evergreen.lowercase(CAST($_ AS text)) ~ ?" } grep { ''.$$search{$_}{group} eq '0' } keys %$search;
- my @usrv = map { "^" . _clean_regex_chars($$search{$_}{value}) } grep { ''.$$search{$_}{group} eq '0' } keys %$search;
+ # Treatment of name fields depends on whether the org has
+ # diacritic_insensitivity turned on or off.
+
+ my $diacritic_insensitive = $U->is_true($U->ou_ancestor_setting_value($ws_ou, 'circ.patron_search.diacritic_insensitive'));
+
+ my $usr;
+ my @usrv;
+
+ if ($diacritic_insensitive) {
+ $usr = join ' AND ', map { "evergreen.unaccent_and_squash(CAST($_ AS text)) ~ ?" } grep { ''.$$search{$_}{group} eq '0' } keys %$search;
+ @usrv = map { "^" . _prepare_name_argument($$search{$_}{value}) } grep { ''.$$search{$_}{group} eq '0' } keys %$search;
+
+ } else {
+ $usr = join ' AND ', map { "evergreen.lowercase(CAST($_ AS text)) ~ ?" } grep { ''.$$search{$_}{group} eq '0' } keys %$search;
+ @usrv = map { "^" . _clean_regex_chars($$search{$_}{value}) } grep { ''.$$search{$_}{group} eq '0' } keys %$search;
+ }
my $addr = join ' AND ', map { "evergreen.lowercase(CAST($_ AS text)) ~ ?" } grep { ''.$$search{$_}{group} eq '1' } keys %$search;
my @addrv = map { "^" . _clean_regex_chars($$search{$_}{value}) } grep { ''.$$search{$_}{group} eq '1' } keys %$search;
my @namev;
if (0 && $nv) {
for my $n ( qw/first_given_name second_given_name family_name/ ) {
- push @ns, "evergreen.lowercase($n) ~ ?";
+ if ($diacritic_insensitive) {
+ push @ns, "evergreen.unaccent_and_squash($n) ~ ?";
+ } else {
+ push @ns, "evergreen.lowercase($n) ~ ?";
+ }
push @namev, "^$nv";
}
$name = '(' . join(' OR ', @ns) . ')';
CREATE INDEX actor_usr_first_given_name_idx ON actor.usr (evergreen.lowercase(first_given_name));
CREATE INDEX actor_usr_second_given_name_idx ON actor.usr (evergreen.lowercase(second_given_name));
CREATE INDEX actor_usr_family_name_idx ON actor.usr (evergreen.lowercase(family_name));
+CREATE INDEX actor_usr_first_given_name_unaccent_idx ON actor.usr (evergreen.unaccent_and_squash(first_given_name));
+CREATE INDEX actor_usr_second_given_name_unaccent_idx ON actor.usr (evergreen.unaccent_and_squash(second_given_name));
+CREATE INDEX actor_usr_family_name_unaccent_idx ON actor.usr (evergreen.unaccent_and_squash(family_name));
CREATE INDEX actor_usr_usrname_idx ON actor.usr (evergreen.lowercase(usrname));
CREATE INDEX actor_usr_email_idx ON actor.usr (evergreen.lowercase(email));
'coust', 'description'),
'bool');
+INSERT INTO config.org_unit_setting_type
+( name, grp, label, description, datatype )
+VALUES
+('circ.patron_search.diacritic_insensitive',
+ 'circ',
+ oils_i18n_gettext('circ.patron_search.diacritic_insensitive',
+ 'Patron search diacritic insensitive',
+ 'coust', 'label'),
+ oils_i18n_gettext('circ.patron_search.diacritic_insensitive',
+ 'Match patron last, first, and middle names irrespective of usage of diacritical marks or spaces. (e.g., Ines will match Inés; de la Cruz will match Delacruz)',
+ 'coust', 'description'),
+ 'bool');
+
+INSERT INTO actor.org_unit_setting (
+ org_unit, name, value
+) VALUES (
+ (SELECT id FROM actor.org_unit WHERE parent_ou IS NULL),
+ 'circ.patron_search.diacritic_insensitive',
+ 'true'
+);
CREATE EXTENSION xml2;
CREATE EXTENSION hstore;
CREATE EXTENSION intarray;
+CREATE EXTENSION unaccent;
--- /dev/null
+BEGIN;
+
+SELECT evergreen.upgrade_deps_block_check('XXXX', :eg_version);
+
+CREATE EXTENSION unaccent;
+
+CREATE OR REPLACE FUNCTION evergreen.unaccent_and_squash ( IN arg text) RETURNS text
+ IMMUTABLE STRICT AS $$
+ BEGIN
+ RETURN evergreen.lowercase(evergreen.unaccent(regexp_replace(arg, '\s','','g')));
+ END;
+$$ LANGUAGE PLPGSQL;
+
+-- The unaccented indices for patron name fields
+CREATE INDEX actor_usr_first_given_name_unaccent_idx ON actor.usr (evergreen.unaccent_and_squash(first_given_name));
+CREATE INDEX actor_usr_second_given_name_unaccent_idx ON actor.usr (evergreen.unaccent_and_squash(second_given_name));
+CREATE INDEX actor_usr_family_name_unaccent_idx ON actor.usr (evergreen.unaccent_and_squash(family_name));
+
+-- DB setting to control behavior; true by default
+INSERT INTO config.org_unit_setting_type
+( name, grp, label, description, datatype )
+VALUES
+('circ.patron_search.diacritic_insensitive',
+ 'circ',
+ oils_i18n_gettext('circ.patron_search.diacritic_insensitive',
+ 'Patron search diacritic insensitive',
+ 'coust', 'label'),
+ oils_i18n_gettext('circ.patron_search.diacritic_insensitive',
+ 'Match patron last, first, and middle names irrespective of usage of diacritical marks or spaces. (e.g., Ines will match Inés; de la Cruz will match Delacruz)',
+ 'coust', 'description'),
+ 'bool');
+
+INSERT INTO actor.org_unit_setting (
+ org_unit, name, value
+) VALUES (
+ (SELECT id FROM actor.org_unit WHERE parent_ou IS NULL),
+ 'circ.patron_search.diacritic_insensitive',
+ 'true'
+);
+
+
+COMMIT;
+
--- /dev/null
+Accent Insensitive Patron Search
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+When performing a patron search, in addition to the (existing)
+case-insensitivity, these additional characteristics will govern the
+search:
+
+Accent (diacritic) insensitivity::
+Diacritics will be transformed into a plain character equivalent for comparison purposes. So if the patron name is Eugène Delacroix, for example, you could enter
+euge for the First Name, and it would match. Ligatures such as Πare expanded into the constituent characters "OE".
+
+Space insensitivity::
+Spaces will be squashed out for comparison purposes. If the patron is, again,
+Eugène Delacroix, you could enter "de la croix" in the Last Name field and it would match.
+
+This behavior affects the Last Name, First Name, and Middle Name fields
+of the search.
projects / working / Evergreen.git / commitdiff