From ebb7ad9f114865db720271deacb5d2bec173aba7 Mon Sep 17 00:00:00 2001 From: Jason Etheridge Date: Tue, 11 Aug 2009 10:37:09 +0000 Subject: [PATCH] name parser --- sql/base/base.sql | 53 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 53 insertions(+), 0 deletions(-) diff --git a/sql/base/base.sql b/sql/base/base.sql index 834231c..f8c68e8 100644 --- a/sql/base/base.sql +++ b/sql/base/base.sql @@ -207,6 +207,59 @@ CREATE OR REPLACE FUNCTION migration_tools.insert_into_production (TEXT,TEXT) RE END; $$ LANGUAGE PLPGSQL STRICT VOLATILE; +CREATE OR REPLACE FUNCTION migration_tools.name_parse_out_last_comma_prefix_first_middle_suffix (TEXT) RETURNS TEXT[] AS $$ + DECLARE + full_name TEXT := $1; + temp TEXT; + family_name TEXT := ''; + first_given_name TEXT := ''; + second_given_name TEXT := ''; + suffix TEXT := ''; + prefix TEXT := ''; + BEGIN + temp := full_name; + -- Use values, not structure, for prefix/suffix, unless we come up with a better idea + IF temp ilike '%MR.%' THEN + prefix := 'Mr.'; + temp := REGEXP_REPLACE( temp, E'MR\.\\s*', '', 'i' ); + END IF; + IF temp ilike '%MRS.%' THEN + prefix := 'Mrs.'; + temp := REGEXP_REPLACE( temp, E'MRS\.\\s*', '', 'i' ); + END IF; + IF temp ilike '%MS.%' THEN + prefix := 'Ms.'; + temp := REGEXP_REPLACE( temp, E'MS\.\\s*', '', 'i' ); + END IF; + IF temp ilike '%DR.%' THEN + prefix := 'Dr.'; + temp := REGEXP_REPLACE( temp, E'DR\.\\s*', '', 'i' ); + END IF; + IF temp ilike '%JR%' THEN + suffix := 'Jr.'; + temp := REGEXP_REPLACE( temp, E'JR\.?\\s*', '', 'i' ); + END IF; + IF temp ilike '%SR%' THEN + suffix := 'Sr.'; + temp := REGEXP_REPLACE( temp, E'SR\.?\\s*', '', 'i' ); + END IF; + IF temp ~ E'\\sII$' THEN + suffix := 'II'; + temp := REGEXP_REPLACE( temp, E'II$', '', 'i' ); + END IF; + IF temp ~ E'\\sIII$' THEN + suffix := 'III'; + temp := REGEXP_REPLACE( temp, E'III$', '', 'i' ); + END IF; + + family_name := BTRIM( REGEXP_REPLACE(temp,E'^([^,]*)\\s*,.*$',E'\\1') ); + first_given_name := BTRIM( CASE WHEN temp ~ ',' THEN REGEXP_REPLACE(temp,E'^[^,]*\\s*,\\s*([^,\\s]*)\\s*.*$',E'\\1') ELSE 'N/A' END ); + second_given_name := BTRIM( CASE WHEN temp ~ ',' THEN REGEXP_REPLACE(temp,E'^[^,]*\\s*,\\s*[^,\\s]*\\s*(.*)$',E'\\1') ELSE '' END ); + + RETURN ARRAY[ family_name, prefix, first_given_name, second_given_name, suffix ]; + END; +$$ LANGUAGE PLPGSQL STRICT VOLATILE; + CREATE OR REPLACE FUNCTION migration_tools.address_parse_out_citystatezip (TEXT) RETURNS TEXT[] AS $$ DECLARE city_state_zip TEXT := $1; -- 1.7.2.5