X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=sql%2Fbase%2Fbase.sql;h=f5c120afc4101142df033b43e4688a4fa960064e;hp=af6340a3618dc1a86bd760ba1e8b04f7fae75146;hb=1f7d2d142333cfb4edfa33c374ae2841116add6d;hpb=d03d9de1d4c79626712b790b8f4f6a2ca086c93b diff --git a/sql/base/base.sql b/sql/base/base.sql index af6340a..f5c120a 100644 --- a/sql/base/base.sql +++ b/sql/base/base.sql @@ -117,7 +117,7 @@ CREATE OR REPLACE FUNCTION migration_tools.init (TEXT) RETURNS VOID AS $$ END; PERFORM migration_tools.exec( $1, 'DROP TABLE IF EXISTS ' || migration_schema || '.config;' ); PERFORM migration_tools.exec( $1, 'CREATE TABLE ' || migration_schema || '.config ( key TEXT UNIQUE, value TEXT);' ); - PERFORM migration_tools.exec( $1, 'INSERT INTO ' || migration_schema || '.config (key,value) VALUES ( ''production_tables'', ''asset.call_number,asset.call_number_prefix,asset.call_number_suffix,asset.copy_location,asset.copy,asset.stat_cat,asset.stat_cat_entry,asset.stat_cat_entry_copy_map,asset.copy_note,actor.usr,actor.card,actor.usr_address,actor.stat_cat,actor.stat_cat_entry,actor.stat_cat_entry_usr_map,actor.usr_note,actor.usr_standing_penalty,actor.usr_setting,action.circulation,action.hold_request,action.hold_notification,action.hold_request_note,action.hold_transit_copy,action.transit_copy,money.grocery,money.billing,money.cash_payment,money.forgive_payment,acq.provider,acq.provider_address,acq.provider_note,acq.provider_contact,acq.provider_contact_address,acq.fund,acq.fund_allocation,acq.fund_tag,acq.fund_tag_map,acq.funding_source,acq.funding_source_credit,acq.lineitem,acq.purchase_order,acq.po_item,acq.invoice,acq.invoice_item,acq.invoice_entry,acq.lineitem_detail,acq.fund_debit,acq.fund_transfer,acq.po_note,config.circ_matrix_matchpoint,config.circ_matrix_limit_set_map,config.hold_matrix_matchpoint,asset.copy_tag,asset.copy_tag_copy_map,config.copy_tag_type,serial.item,serial.item_note,serial.record_entry,biblio.record_entry'' );' ); + PERFORM migration_tools.exec( $1, 'INSERT INTO ' || migration_schema || '.config (key,value) VALUES ( ''production_tables'', ''asset.call_number,asset.call_number_prefix,asset.call_number_suffix,asset.copy_location,asset.copy,asset.copy_alert,asset.stat_cat,asset.stat_cat_entry,asset.stat_cat_entry_copy_map,asset.copy_note,actor.usr,actor.card,actor.usr_address,actor.stat_cat,actor.stat_cat_entry,actor.stat_cat_entry_usr_map,actor.usr_note,actor.usr_standing_penalty,actor.usr_setting,action.circulation,action.hold_request,action.hold_notification,action.hold_request_note,action.hold_transit_copy,action.transit_copy,money.grocery,money.billing,money.cash_payment,money.forgive_payment,acq.provider,acq.provider_address,acq.provider_note,acq.provider_contact,acq.provider_contact_address,acq.fund,acq.fund_allocation,acq.fund_tag,acq.fund_tag_map,acq.funding_source,acq.funding_source_credit,acq.lineitem,acq.purchase_order,acq.po_item,acq.invoice,acq.invoice_item,acq.invoice_entry,acq.lineitem_detail,acq.fund_debit,acq.fund_transfer,acq.po_note,config.circ_matrix_matchpoint,config.circ_matrix_limit_set_map,config.hold_matrix_matchpoint,asset.copy_tag,asset.copy_tag_copy_map,config.copy_tag_type,serial.item,serial.item_note,serial.record_entry,biblio.record_entry'' );' ); PERFORM migration_tools.exec( $1, 'INSERT INTO ' || migration_schema || '.config (key,value) VALUES ( ''country_code'', ''USA'' );' ); PERFORM migration_tools.exec( $1, 'DROP TABLE IF EXISTS ' || migration_schema || '.fields_requiring_mapping;' ); PERFORM migration_tools.exec( $1, 'CREATE TABLE ' || migration_schema || '.fields_requiring_mapping( table_schema TEXT, table_name TEXT, column_name TEXT, data_type TEXT);' ); @@ -244,7 +244,7 @@ CREATE OR REPLACE FUNCTION migration_tools.build_specific_base_staging_table (TE BEGIN base_staging_table = REPLACE( production_table, '.', '_' ); --RAISE INFO 'In migration_tools.build_specific_base_staging_table(%,%) -> %', migration_schema, production_table, base_staging_table; - PERFORM migration_tools.exec( $1, 'CREATE UNLOGGED TABLE ' || migration_schema || '.' || base_staging_table || ' ( LIKE ' || production_table || ' INCLUDING DEFAULTS EXCLUDING CONSTRAINTS );' ); + PERFORM migration_tools.exec( $1, 'CREATE TABLE ' || migration_schema || '.' || base_staging_table || ' ( LIKE ' || production_table || ' INCLUDING DEFAULTS EXCLUDING CONSTRAINTS );' ); PERFORM migration_tools.exec( $1, ' INSERT INTO ' || migration_schema || '.fields_requiring_mapping SELECT table_schema, table_name, column_name, data_type @@ -261,6 +261,45 @@ CREATE OR REPLACE FUNCTION migration_tools.build_specific_base_staging_table (TE END; $$ LANGUAGE PLPGSQL STRICT VOLATILE; +CREATE OR REPLACE FUNCTION migration_tools.create_linked_legacy_table_from (TEXT,TEXT,TEXT) RETURNS VOID AS $$ + DECLARE + migration_schema ALIAS FOR $1; + parent_table ALIAS FOR $2; + source_table ALIAS FOR $3; + columns RECORD; + create_sql TEXT; + insert_sql TEXT; + column_list TEXT := ''; + column_count INTEGER := 0; + BEGIN + create_sql := 'CREATE TABLE ' || migration_schema || '.' || parent_table || '_legacy ( '; + FOR columns IN + SELECT table_schema, table_name, column_name, data_type + FROM information_schema.columns + WHERE table_schema = migration_schema AND table_name = source_table + LOOP + column_count := column_count + 1; + if column_count > 1 then + create_sql := create_sql || ', '; + column_list := column_list || ', '; + end if; + create_sql := create_sql || columns.column_name || ' '; + if columns.data_type = 'ARRAY' then + create_sql := create_sql || 'TEXT[]'; + else + create_sql := create_sql || columns.data_type; + end if; + column_list := column_list || columns.column_name; + END LOOP; + create_sql := create_sql || ' ) INHERITS ( ' || migration_schema || '.' || parent_table || ' );'; + --RAISE INFO 'create_sql = %', create_sql; + EXECUTE create_sql; + insert_sql := 'INSERT INTO ' || migration_schema || '.' || parent_table || '_legacy (' || column_list || ') SELECT ' || column_list || ' FROM ' || migration_schema || '.' || source_table || ';'; + --RAISE INFO 'insert_sql = %', insert_sql; + EXECUTE insert_sql; + END; +$$ LANGUAGE PLPGSQL STRICT VOLATILE; + CREATE OR REPLACE FUNCTION migration_tools.insert_base_into_production (TEXT) RETURNS VOID AS $$ DECLARE migration_schema ALIAS FOR $1; @@ -534,6 +573,98 @@ CREATE OR REPLACE FUNCTION migration_tools.name_parse_out_fuller_last_first_midd END; $$ LANGUAGE PLPGSQL STRICT IMMUTABLE; +CREATE OR REPLACE FUNCTION migration_tools.name_parse_out_fuller_last_first_middle_and_random_affix2 (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 := BTRIM(full_name); + -- Use values, not structure, for prefix/suffix, unless we come up with a better idea + --IF temp ~ '^\S{2,}\.' THEN + -- prefix := REGEXP_REPLACE(temp, '^(\S{2,}\.).*$','\1'); + -- temp := BTRIM(REGEXP_REPLACE(temp, '^\S{2,}\.(.*)$','\1')); + --END IF; + --IF temp ~ '\S{2,}\.$' THEN + -- suffix := REGEXP_REPLACE(temp, '^.*(\S{2,}\.)$','\1'); + -- temp := REGEXP_REPLACE(temp, '^(.*)\S{2,}\.$','\1'); + --END IF; + IF temp ilike '%MR.%' THEN + prefix := 'Mr.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'MR\.\\s*', '', 'i' )); + END IF; + IF temp ilike '%MRS.%' THEN + prefix := 'Mrs.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'MRS\.\\s*', '', 'i' )); + END IF; + IF temp ilike '%MS.%' THEN + prefix := 'Ms.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'MS\.\\s*', '', 'i' )); + END IF; + IF temp ilike '%DR.%' THEN + prefix := 'Dr.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'DR\.\\s*', '', 'i' )); + END IF; + IF temp ilike '%JR.%' THEN + suffix := 'Jr.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'JR\.\\s*', '', 'i' )); + END IF; + IF temp ilike '%JR,%' THEN + suffix := 'Jr.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'JR,\\s*', ',', 'i' )); + END IF; + IF temp ilike '%SR.%' THEN + suffix := 'Sr.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'SR\.\\s*', '', 'i' )); + END IF; + IF temp ilike '%SR,%' THEN + suffix := 'Sr.'; + temp := BTRIM(REGEXP_REPLACE( temp, E'SR,\\s*', ',', 'i' )); + END IF; + IF temp like '%III%' THEN + suffix := 'III'; + temp := BTRIM(REGEXP_REPLACE( temp, E'III', '' )); + END IF; + IF temp like '%II%' THEN + suffix := 'II'; + temp := BTRIM(REGEXP_REPLACE( temp, E'II', '' )); + END IF; + + IF temp ~ ',' THEN + family_name = BTRIM(REGEXP_REPLACE(temp,'^(.*?,).*$','\1')); + temp := BTRIM(REPLACE( temp, family_name, '' )); + family_name := REPLACE( family_name, ',', '' ); + IF temp ~ ' ' THEN + first_given_name := BTRIM( REGEXP_REPLACE(temp,'^(.+)\s(.+)$','\1') ); + second_given_name := BTRIM( REGEXP_REPLACE(temp,'^(.+)\s(.+)$','\2') ); + ELSE + first_given_name := temp; + second_given_name := ''; + END IF; + ELSE + IF temp ~ '^\S+\s+\S+\s+\S+$' THEN + first_given_name := BTRIM( REGEXP_REPLACE(temp,'^(\S+)\s*(\S+)\s*(\S+)$','\1') ); + second_given_name := BTRIM( REGEXP_REPLACE(temp,'^(\S+)\s*(\S+)\s*(\S+)$','\2') ); + family_name := BTRIM( REGEXP_REPLACE(temp,'^(\S+)\s*(\S+)\s*(\S+)$','\3') ); + ELSE + first_given_name := BTRIM( REGEXP_REPLACE(temp,'^(\S+)\s*(\S+)$','\1') ); + second_given_name := temp; + family_name := BTRIM( REGEXP_REPLACE(temp,'^(\S+)\s*(\S+)$','\2') ); + END IF; + END IF; + + family_name := BTRIM(REPLACE(REPLACE(family_name,',',''),'"','')); + first_given_name := BTRIM(REPLACE(REPLACE(first_given_name,',',''),'"','')); + second_given_name := BTRIM(REPLACE(REPLACE(second_given_name,',',''),'"','')); + + RETURN ARRAY[ family_name, prefix, first_given_name, second_given_name, suffix ]; + END; +$$ LANGUAGE PLPGSQL STRICT IMMUTABLE; + CREATE OR REPLACE FUNCTION migration_tools.address_parse_out_citystatezip (TEXT) RETURNS TEXT[] AS $$ DECLARE city_state_zip TEXT := $1; @@ -648,552 +779,563 @@ CREATE OR REPLACE FUNCTION migration_tools.parse_out_address2 (TEXT) RETURNS TEX ]; $$ LANGUAGE PLPERLU STABLE; --- FIXME: there's probably a more efficient way to implement this -CREATE OR REPLACE FUNCTION migration_tools.normalize_address_suffix (TEXT) RETURNS TEXT AS $$ +DROP TABLE IF EXISTS migration_tools.usps_suffixes; +CREATE TABLE migration_tools.usps_suffixes ( suffix_from TEXT, suffix_to TEXT ); +INSERT INTO migration_tools.usps_suffixes VALUES + ('ALLEE','ALY'), + ('ALLEY','ALY'), + ('ALLY','ALY'), + ('ALY','ALY'), + ('ANEX','ANX'), + ('ANNEX','ANX'), + ('ANNX','ANX'), + ('ANX','ANX'), + ('ARCADE','ARC'), + ('ARC','ARC'), + ('AV','AVE'), + ('AVE','AVE'), + ('AVEN','AVE'), + ('AVENU','AVE'), + ('AVENUE','AVE'), + ('AVN','AVE'), + ('AVNUE','AVE'), + ('BAYOO','BYU'), + ('BAYOU','BYU'), + ('BCH','BCH'), + ('BEACH','BCH'), + ('BEND','BND'), + ('BLF','BLF'), + ('BLUF','BLF'), + ('BLUFF','BLF'), + ('BLUFFS','BLFS'), + ('BLVD','BLVD'), + ('BND','BND'), + ('BOT','BTM'), + ('BOTTM','BTM'), + ('BOTTOM','BTM'), + ('BOUL','BLVD'), + ('BOULEVARD','BLVD'), + ('BOULV','BLVD'), + ('BRANCH','BR'), + ('BR','BR'), + ('BRDGE','BRG'), + ('BRG','BRG'), + ('BRIDGE','BRG'), + ('BRK','BRK'), + ('BRNCH','BR'), + ('BROOK','BRK'), + ('BROOKS','BRKS'), + ('BTM','BTM'), + ('BURG','BG'), + ('BURGS','BGS'), + ('BYPA','BYP'), + ('BYPAS','BYP'), + ('BYPASS','BYP'), + ('BYP','BYP'), + ('BYPS','BYP'), + ('CAMP','CP'), + ('CANYN','CYN'), + ('CANYON','CYN'), + ('CAPE','CPE'), + ('CAUSEWAY','CSWY'), + ('CAUSWAY','CSWY'), + ('CEN','CTR'), + ('CENT','CTR'), + ('CENTER','CTR'), + ('CENTERS','CTRS'), + ('CENTR','CTR'), + ('CENTRE','CTR'), + ('CIRC','CIR'), + ('CIR','CIR'), + ('CIRCL','CIR'), + ('CIRCLE','CIR'), + ('CIRCLES','CIRS'), + ('CK','CRK'), + ('CLB','CLB'), + ('CLF','CLF'), + ('CLFS','CLFS'), + ('CLIFF','CLF'), + ('CLIFFS','CLFS'), + ('CLUB','CLB'), + ('CMP','CP'), + ('CNTER','CTR'), + ('CNTR','CTR'), + ('CNYN','CYN'), + ('COMMON','CMN'), + ('COR','COR'), + ('CORNER','COR'), + ('CORNERS','CORS'), + ('CORS','CORS'), + ('COURSE','CRSE'), + ('COURT','CT'), + ('COURTS','CTS'), + ('COVE','CV'), + ('COVES','CVS'), + ('CP','CP'), + ('CPE','CPE'), + ('CRCL','CIR'), + ('CRCLE','CIR'), + ('CR','CRK'), + ('CRECENT','CRES'), + ('CREEK','CRK'), + ('CRESCENT','CRES'), + ('CRES','CRES'), + ('CRESENT','CRES'), + ('CREST','CRST'), + ('CRK','CRK'), + ('CROSSING','XING'), + ('CROSSROAD','XRD'), + ('CRSCNT','CRES'), + ('CRSE','CRSE'), + ('CRSENT','CRES'), + ('CRSNT','CRES'), + ('CRSSING','XING'), + ('CRSSNG','XING'), + ('CRT','CT'), + ('CSWY','CSWY'), + ('CT','CT'), + ('CTR','CTR'), + ('CTS','CTS'), + ('CURVE','CURV'), + ('CV','CV'), + ('CYN','CYN'), + ('DALE','DL'), + ('DAM','DM'), + ('DIV','DV'), + ('DIVIDE','DV'), + ('DL','DL'), + ('DM','DM'), + ('DR','DR'), + ('DRIV','DR'), + ('DRIVE','DR'), + ('DRIVES','DRS'), + ('DRV','DR'), + ('DVD','DV'), + ('DV','DV'), + ('ESTATE','EST'), + ('ESTATES','ESTS'), + ('EST','EST'), + ('ESTS','ESTS'), + ('EXP','EXPY'), + ('EXPRESS','EXPY'), + ('EXPRESSWAY','EXPY'), + ('EXPR','EXPY'), + ('EXPW','EXPY'), + ('EXPY','EXPY'), + ('EXTENSION','EXT'), + ('EXTENSIONS','EXTS'), + ('EXT','EXT'), + ('EXTN','EXT'), + ('EXTNSN','EXT'), + ('EXTS','EXTS'), + ('FALL','FALL'), + ('FALLS','FLS'), + ('FERRY','FRY'), + ('FIELD','FLD'), + ('FIELDS','FLDS'), + ('FLAT','FLT'), + ('FLATS','FLTS'), + ('FLD','FLD'), + ('FLDS','FLDS'), + ('FLS','FLS'), + ('FLT','FLT'), + ('FLTS','FLTS'), + ('FORD','FRD'), + ('FORDS','FRDS'), + ('FOREST','FRST'), + ('FORESTS','FRST'), + ('FORGE','FRG'), + ('FORGES','FRGS'), + ('FORG','FRG'), + ('FORK','FRK'), + ('FORKS','FRKS'), + ('FORT','FT'), + ('FRD','FRD'), + ('FREEWAY','FWY'), + ('FREEWY','FWY'), + ('FRG','FRG'), + ('FRK','FRK'), + ('FRKS','FRKS'), + ('FRRY','FRY'), + ('FRST','FRST'), + ('FRT','FT'), + ('FRWAY','FWY'), + ('FRWY','FWY'), + ('FRY','FRY'), + ('FT','FT'), + ('FWY','FWY'), + ('GARDEN','GDN'), + ('GARDENS','GDNS'), + ('GARDN','GDN'), + ('GATEWAY','GTWY'), + ('GATEWY','GTWY'), + ('GATWAY','GTWY'), + ('GDN','GDN'), + ('GDNS','GDNS'), + ('GLEN','GLN'), + ('GLENS','GLNS'), + ('GLN','GLN'), + ('GRDEN','GDN'), + ('GRDN','GDN'), + ('GRDNS','GDNS'), + ('GREEN','GRN'), + ('GREENS','GRNS'), + ('GRN','GRN'), + ('GROVE','GRV'), + ('GROVES','GRVS'), + ('GROV','GRV'), + ('GRV','GRV'), + ('GTWAY','GTWY'), + ('GTWY','GTWY'), + ('HARB','HBR'), + ('HARBOR','HBR'), + ('HARBORS','HBRS'), + ('HARBR','HBR'), + ('HAVEN','HVN'), + ('HAVN','HVN'), + ('HBR','HBR'), + ('HEIGHT','HTS'), + ('HEIGHTS','HTS'), + ('HGTS','HTS'), + ('HIGHWAY','HWY'), + ('HIGHWY','HWY'), + ('HILL','HL'), + ('HILLS','HLS'), + ('HIWAY','HWY'), + ('HIWY','HWY'), + ('HL','HL'), + ('HLLW','HOLW'), + ('HLS','HLS'), + ('HOLLOW','HOLW'), + ('HOLLOWS','HOLW'), + ('HOLW','HOLW'), + ('HOLWS','HOLW'), + ('HRBOR','HBR'), + ('HT','HTS'), + ('HTS','HTS'), + ('HVN','HVN'), + ('HWAY','HWY'), + ('HWY','HWY'), + ('INLET','INLT'), + ('INLT','INLT'), + ('IS','IS'), + ('ISLAND','IS'), + ('ISLANDS','ISS'), + ('ISLANDS','SLNDS'), + ('ISLANDS','SS'), + ('ISLE','ISLE'), + ('ISLES','ISLE'), + ('ISLND','IS'), + ('I','SLNDS'), + ('ISS','ISS'), + ('JCTION','JCT'), + ('JCT','JCT'), + ('JCTN','JCT'), + ('JCTNS','JCTS'), + ('JCTS','JCTS'), + ('JUNCTION','JCT'), + ('JUNCTIONS','JCTS'), + ('JUNCTN','JCT'), + ('JUNCTON','JCT'), + ('KEY','KY'), + ('KEYS','KYS'), + ('KNL','KNL'), + ('KNLS','KNLS'), + ('KNOL','KNL'), + ('KNOLL','KNL'), + ('KNOLLS','KNLS'), + ('KY','KY'), + ('KYS','KYS'), + ('LAKE','LK'), + ('LAKES','LKS'), + ('LA','LN'), + ('LANDING','LNDG'), + ('LAND','LAND'), + ('LANE','LN'), + ('LANES','LN'), + ('LCK','LCK'), + ('LCKS','LCKS'), + ('LDGE','LDG'), + ('LDG','LDG'), + ('LF','LF'), + ('LGT','LGT'), + ('LIGHT','LGT'), + ('LIGHTS','LGTS'), + ('LK','LK'), + ('LKS','LKS'), + ('LNDG','LNDG'), + ('LNDNG','LNDG'), + ('LN','LN'), + ('LOAF','LF'), + ('LOCK','LCK'), + ('LOCKS','LCKS'), + ('LODGE','LDG'), + ('LODG','LDG'), + ('LOOP','LOOP'), + ('LOOPS','LOOP'), + ('MALL','MALL'), + ('MANOR','MNR'), + ('MANORS','MNRS'), + ('MDW','MDW'), + ('MDWS','MDWS'), + ('MEADOW','MDW'), + ('MEADOWS','MDWS'), + ('MEDOWS','MDWS'), + ('MEWS','MEWS'), + ('MILL','ML'), + ('MILLS','MLS'), + ('MISSION','MSN'), + ('MISSN','MSN'), + ('ML','ML'), + ('MLS','MLS'), + ('MNR','MNR'), + ('MNRS','MNRS'), + ('MNTAIN','MTN'), + ('MNT','MT'), + ('MNTN','MTN'), + ('MNTNS','MTNS'), + ('MOTORWAY','MTWY'), + ('MOUNTAIN','MTN'), + ('MOUNTAINS','MTNS'), + ('MOUNTIN','MTN'), + ('MOUNT','MT'), + ('MSN','MSN'), + ('MSSN','MSN'), + ('MTIN','MTN'), + ('MT','MT'), + ('MTN','MTN'), + ('NCK','NCK'), + ('NECK','NCK'), + ('ORCHARD','ORCH'), + ('ORCH','ORCH'), + ('ORCHRD','ORCH'), + ('OVAL','OVAL'), + ('OVERPASS','OPAS'), + ('OVL','OVAL'), + ('PARK','PARK'), + ('PARKS','PARK'), + ('PARKWAY','PKWY'), + ('PARKWAYS','PKWY'), + ('PARKWY','PKWY'), + ('PASSAGE','PSGE'), + ('PASS','PASS'), + ('PATH','PATH'), + ('PATHS','PATH'), + ('PIKE','PIKE'), + ('PIKES','PIKE'), + ('PINE','PNE'), + ('PINES','PNES'), + ('PK','PARK'), + ('PKWAY','PKWY'), + ('PKWY','PKWY'), + ('PKWYS','PKWY'), + ('PKY','PKWY'), + ('PLACE','PL'), + ('PLAINES','PLNS'), + ('PLAIN','PLN'), + ('PLAINS','PLNS'), + ('PLAZA','PLZ'), + ('PLN','PLN'), + ('PLNS','PLNS'), + ('PL','PL'), + ('PLZA','PLZ'), + ('PLZ','PLZ'), + ('PNES','PNES'), + ('POINT','PT'), + ('POINTS','PTS'), + ('PORT','PRT'), + ('PORTS','PRTS'), + ('PRAIRIE','PR'), + ('PRARIE','PR'), + ('PRK','PARK'), + ('PR','PR'), + ('PRR','PR'), + ('PRT','PRT'), + ('PRTS','PRTS'), + ('PT','PT'), + ('PTS','PTS'), + ('RADIAL','RADL'), + ('RADIEL','RADL'), + ('RADL','RADL'), + ('RAD','RADL'), + ('RAMP','RAMP'), + ('RANCHES','RNCH'), + ('RANCH','RNCH'), + ('RAPID','RPD'), + ('RAPIDS','RPDS'), + ('RDGE','RDG'), + ('RDG','RDG'), + ('RDGS','RDGS'), + ('RD','RD'), + ('RDS','RDS'), + ('REST','RST'), + ('RIDGE','RDG'), + ('RIDGES','RDGS'), + ('RIVER','RIV'), + ('RIV','RIV'), + ('RIVR','RIV'), + ('RNCH','RNCH'), + ('RNCHS','RNCH'), + ('ROAD','RD'), + ('ROADS','RDS'), + ('ROUTE','RTE'), + ('ROW','ROW'), + ('RPD','RPD'), + ('RPDS','RPDS'), + ('RST','RST'), + ('RUE','RUE'), + ('RUN','RUN'), + ('RVR','RIV'), + ('SHL','SHL'), + ('SHLS','SHLS'), + ('SHOAL','SHL'), + ('SHOALS','SHLS'), + ('SHOAR','SHR'), + ('SHOARS','SHRS'), + ('SHORE','SHR'), + ('SHORES','SHRS'), + ('SHR','SHR'), + ('SHRS','SHRS'), + ('SKYWAY','SKWY'), + ('SMT','SMT'), + ('SPG','SPG'), + ('SPGS','SPGS'), + ('SPNG','SPG'), + ('SPNGS','SPGS'), + ('SPRING','SPG'), + ('SPRINGS','SPGS'), + ('SPRNG','SPG'), + ('SPRNGS','SPGS'), + ('SPUR','SPUR'), + ('SPURS','SPUR'), + ('SQRE','SQ'), + ('SQR','SQ'), + ('SQRS','SQS'), + ('SQ','SQ'), + ('SQUARE','SQ'), + ('SQUARES','SQS'), + ('SQU','SQ'), + ('STA','STA'), + ('STATION','STA'), + ('STATN','STA'), + ('STN','STA'), + ('STRA','STRA'), + ('STRAVEN','STRA'), + ('STRAVENUE','STRA'), + ('STRAVE','STRA'), + ('STRAVN','STRA'), + ('STRAV','STRA'), + ('STREAM','STRM'), + ('STREETS','STS'), + ('STREET','ST'), + ('STREME','STRM'), + ('STRM','STRM'), + ('STR','ST'), + ('STRT','ST'), + ('STRVN','STRA'), + ('STRVNUE','STRA'), + ('ST','ST'), + ('SUMIT','SMT'), + ('SUMITT','SMT'), + ('SUMMIT','SMT'), + ('TERRACE','TER'), + ('TERR','TER'), + ('TER','TER'), + ('THROUGHWAY','TRWY'), + ('TPKE','TPKE'), + ('TPK','TPKE'), + ('TRACES','TRCE'), + ('TRACE','TRCE'), + ('TRACKS','TRAK'), + ('TRACK','TRAK'), + ('TRAFFICWAY','TRFY'), + ('TRAILS','TRL'), + ('TRAIL','TRL'), + ('TRAK','TRAK'), + ('TRCE','TRCE'), + ('TRFY','TRFY'), + ('TRKS','TRAK'), + ('TRK','TRAK'), + ('TRLS','TRL'), + ('TRL','TRL'), + ('TRNPK','TPKE'), + ('TRPK','TPKE'), + ('TR','TRL'), + ('TUNEL','TUNL'), + ('TUNLS','TUNL'), + ('TUNL','TUNL'), + ('TUNNELS','TUNL'), + ('TUNNEL','TUNL'), + ('TUNNL','TUNL'), + ('TURNPIKE','TPKE'), + ('TURNPK','TPKE'), + ('UNDERPASS','UPAS'), + ('UNIONS','UNS'), + ('UNION','UN'), + ('UN','UN'), + ('VALLEYS','VLYS'), + ('VALLEY','VLY'), + ('VALLY','VLY'), + ('VDCT','IA'), + ('VIADCT','VIA'), + ('VIADUCT','IA'), + ('VIADUCT','VIA'), + ('VIA','VIA'), + ('VIEWS','VWS'), + ('VIEW','VW'), + ('VILLAGES','VLGS'), + ('VILLAGE','VLG'), + ('VILLAG','VLG'), + ('VILLE','VL'), + ('VILLG','VLG'), + ('VILLIAGE','VLG'), + ('VILL','VLG'), + ('VISTA','VIS'), + ('VIST','VIS'), + ('VIS','VIS'), + ('VLGS','VLGS'), + ('VLG','VLG'), + ('VLLY','VLY'), + ('VL','VL'), + ('VLYS','VLYS'), + ('VLY','VLY'), + ('VSTA','VIS'), + ('VST','VIS'), + ('VWS','VWS'), + ('VW','VW'), + ('WALKS','WALK'), + ('WALK','WALK'), + ('WALL','WALL'), + ('WAYS','WAYS'), + ('WAY','WAY'), + ('WELLS','WLS'), + ('WELL','WL'), + ('WLS','WLS'), + ('WY','WAY'), + ('XING','XING'); + +-- this function should get a smaller range of inputs and benefit more from STABLE, hopefully speeding things up +CREATE OR REPLACE FUNCTION migration_tools._normalize_address_suffix (TEXT) RETURNS TEXT AS $$ DECLARE - street1 TEXT := $1; + suffix TEXT := $1; _r RECORD; BEGIN - DROP TABLE IF EXISTS mig_usps_suffixes; - CREATE TEMP TABLE mig_usps_suffixes ( suffix_from TEXT, suffix_to TEXT ); - INSERT INTO mig_usps_suffixes VALUES - ('ALLEE','ALY'), - ('ALLEY','ALY'), - ('ALLY','ALY'), - ('ALY','ALY'), - ('ANEX','ANX'), - ('ANNEX','ANX'), - ('ANNX','ANX'), - ('ANX','ANX'), - ('ARCADE','ARC'), - ('ARC','ARC'), - ('AV','AVE'), - ('AVE','AVE'), - ('AVEN','AVE'), - ('AVENU','AVE'), - ('AVENUE','AVE'), - ('AVN','AVE'), - ('AVNUE','AVE'), - ('BAYOO','BYU'), - ('BAYOU','BYU'), - ('BCH','BCH'), - ('BEACH','BCH'), - ('BEND','BND'), - ('BLF','BLF'), - ('BLUF','BLF'), - ('BLUFF','BLF'), - ('BLUFFS','BLFS'), - ('BLVD','BLVD'), - ('BND','BND'), - ('BOT','BTM'), - ('BOTTM','BTM'), - ('BOTTOM','BTM'), - ('BOUL','BLVD'), - ('BOULEVARD','BLVD'), - ('BOULV','BLVD'), - ('BRANCH','BR'), - ('BR','BR'), - ('BRDGE','BRG'), - ('BRG','BRG'), - ('BRIDGE','BRG'), - ('BRK','BRK'), - ('BRNCH','BR'), - ('BROOK','BRK'), - ('BROOKS','BRKS'), - ('BTM','BTM'), - ('BURG','BG'), - ('BURGS','BGS'), - ('BYPA','BYP'), - ('BYPAS','BYP'), - ('BYPASS','BYP'), - ('BYP','BYP'), - ('BYPS','BYP'), - ('CAMP','CP'), - ('CANYN','CYN'), - ('CANYON','CYN'), - ('CAPE','CPE'), - ('CAUSEWAY','CSWY'), - ('CAUSWAY','CSWY'), - ('CEN','CTR'), - ('CENT','CTR'), - ('CENTER','CTR'), - ('CENTERS','CTRS'), - ('CENTR','CTR'), - ('CENTRE','CTR'), - ('CIRC','CIR'), - ('CIR','CIR'), - ('CIRCL','CIR'), - ('CIRCLE','CIR'), - ('CIRCLES','CIRS'), - ('CK','CRK'), - ('CLB','CLB'), - ('CLF','CLF'), - ('CLFS','CLFS'), - ('CLIFF','CLF'), - ('CLIFFS','CLFS'), - ('CLUB','CLB'), - ('CMP','CP'), - ('CNTER','CTR'), - ('CNTR','CTR'), - ('CNYN','CYN'), - ('COMMON','CMN'), - ('COR','COR'), - ('CORNER','COR'), - ('CORNERS','CORS'), - ('CORS','CORS'), - ('COURSE','CRSE'), - ('COURT','CT'), - ('COURTS','CTS'), - ('COVE','CV'), - ('COVES','CVS'), - ('CP','CP'), - ('CPE','CPE'), - ('CRCL','CIR'), - ('CRCLE','CIR'), - ('CR','CRK'), - ('CRECENT','CRES'), - ('CREEK','CRK'), - ('CRESCENT','CRES'), - ('CRES','CRES'), - ('CRESENT','CRES'), - ('CREST','CRST'), - ('CRK','CRK'), - ('CROSSING','XING'), - ('CROSSROAD','XRD'), - ('CRSCNT','CRES'), - ('CRSE','CRSE'), - ('CRSENT','CRES'), - ('CRSNT','CRES'), - ('CRSSING','XING'), - ('CRSSNG','XING'), - ('CRT','CT'), - ('CSWY','CSWY'), - ('CT','CT'), - ('CTR','CTR'), - ('CTS','CTS'), - ('CURVE','CURV'), - ('CV','CV'), - ('CYN','CYN'), - ('DALE','DL'), - ('DAM','DM'), - ('DIV','DV'), - ('DIVIDE','DV'), - ('DL','DL'), - ('DM','DM'), - ('DR','DR'), - ('DRIV','DR'), - ('DRIVE','DR'), - ('DRIVES','DRS'), - ('DRV','DR'), - ('DVD','DV'), - ('DV','DV'), - ('ESTATE','EST'), - ('ESTATES','ESTS'), - ('EST','EST'), - ('ESTS','ESTS'), - ('EXP','EXPY'), - ('EXPRESS','EXPY'), - ('EXPRESSWAY','EXPY'), - ('EXPR','EXPY'), - ('EXPW','EXPY'), - ('EXPY','EXPY'), - ('EXTENSION','EXT'), - ('EXTENSIONS','EXTS'), - ('EXT','EXT'), - ('EXTN','EXT'), - ('EXTNSN','EXT'), - ('EXTS','EXTS'), - ('FALL','FALL'), - ('FALLS','FLS'), - ('FERRY','FRY'), - ('FIELD','FLD'), - ('FIELDS','FLDS'), - ('FLAT','FLT'), - ('FLATS','FLTS'), - ('FLD','FLD'), - ('FLDS','FLDS'), - ('FLS','FLS'), - ('FLT','FLT'), - ('FLTS','FLTS'), - ('FORD','FRD'), - ('FORDS','FRDS'), - ('FOREST','FRST'), - ('FORESTS','FRST'), - ('FORGE','FRG'), - ('FORGES','FRGS'), - ('FORG','FRG'), - ('FORK','FRK'), - ('FORKS','FRKS'), - ('FORT','FT'), - ('FRD','FRD'), - ('FREEWAY','FWY'), - ('FREEWY','FWY'), - ('FRG','FRG'), - ('FRK','FRK'), - ('FRKS','FRKS'), - ('FRRY','FRY'), - ('FRST','FRST'), - ('FRT','FT'), - ('FRWAY','FWY'), - ('FRWY','FWY'), - ('FRY','FRY'), - ('FT','FT'), - ('FWY','FWY'), - ('GARDEN','GDN'), - ('GARDENS','GDNS'), - ('GARDN','GDN'), - ('GATEWAY','GTWY'), - ('GATEWY','GTWY'), - ('GATWAY','GTWY'), - ('GDN','GDN'), - ('GDNS','GDNS'), - ('GLEN','GLN'), - ('GLENS','GLNS'), - ('GLN','GLN'), - ('GRDEN','GDN'), - ('GRDN','GDN'), - ('GRDNS','GDNS'), - ('GREEN','GRN'), - ('GREENS','GRNS'), - ('GRN','GRN'), - ('GROVE','GRV'), - ('GROVES','GRVS'), - ('GROV','GRV'), - ('GRV','GRV'), - ('GTWAY','GTWY'), - ('GTWY','GTWY'), - ('HARB','HBR'), - ('HARBOR','HBR'), - ('HARBORS','HBRS'), - ('HARBR','HBR'), - ('HAVEN','HVN'), - ('HAVN','HVN'), - ('HBR','HBR'), - ('HEIGHT','HTS'), - ('HEIGHTS','HTS'), - ('HGTS','HTS'), - ('HIGHWAY','HWY'), - ('HIGHWY','HWY'), - ('HILL','HL'), - ('HILLS','HLS'), - ('HIWAY','HWY'), - ('HIWY','HWY'), - ('HL','HL'), - ('HLLW','HOLW'), - ('HLS','HLS'), - ('HOLLOW','HOLW'), - ('HOLLOWS','HOLW'), - ('HOLW','HOLW'), - ('HOLWS','HOLW'), - ('HRBOR','HBR'), - ('HT','HTS'), - ('HTS','HTS'), - ('HVN','HVN'), - ('HWAY','HWY'), - ('HWY','HWY'), - ('INLET','INLT'), - ('INLT','INLT'), - ('IS','IS'), - ('ISLAND','IS'), - ('ISLANDS','ISS'), - ('ISLANDS','SLNDS'), - ('ISLANDS','SS'), - ('ISLE','ISLE'), - ('ISLES','ISLE'), - ('ISLND','IS'), - ('I','SLNDS'), - ('ISS','ISS'), - ('JCTION','JCT'), - ('JCT','JCT'), - ('JCTN','JCT'), - ('JCTNS','JCTS'), - ('JCTS','JCTS'), - ('JUNCTION','JCT'), - ('JUNCTIONS','JCTS'), - ('JUNCTN','JCT'), - ('JUNCTON','JCT'), - ('KEY','KY'), - ('KEYS','KYS'), - ('KNL','KNL'), - ('KNLS','KNLS'), - ('KNOL','KNL'), - ('KNOLL','KNL'), - ('KNOLLS','KNLS'), - ('KY','KY'), - ('KYS','KYS'), - ('LAKE','LK'), - ('LAKES','LKS'), - ('LA','LN'), - ('LANDING','LNDG'), - ('LAND','LAND'), - ('LANE','LN'), - ('LANES','LN'), - ('LCK','LCK'), - ('LCKS','LCKS'), - ('LDGE','LDG'), - ('LDG','LDG'), - ('LF','LF'), - ('LGT','LGT'), - ('LIGHT','LGT'), - ('LIGHTS','LGTS'), - ('LK','LK'), - ('LKS','LKS'), - ('LNDG','LNDG'), - ('LNDNG','LNDG'), - ('LN','LN'), - ('LOAF','LF'), - ('LOCK','LCK'), - ('LOCKS','LCKS'), - ('LODGE','LDG'), - ('LODG','LDG'), - ('LOOP','LOOP'), - ('LOOPS','LOOP'), - ('MALL','MALL'), - ('MANOR','MNR'), - ('MANORS','MNRS'), - ('MDW','MDW'), - ('MDWS','MDWS'), - ('MEADOW','MDW'), - ('MEADOWS','MDWS'), - ('MEDOWS','MDWS'), - ('MEWS','MEWS'), - ('MILL','ML'), - ('MILLS','MLS'), - ('MISSION','MSN'), - ('MISSN','MSN'), - ('ML','ML'), - ('MLS','MLS'), - ('MNR','MNR'), - ('MNRS','MNRS'), - ('MNTAIN','MTN'), - ('MNT','MT'), - ('MNTN','MTN'), - ('MNTNS','MTNS'), - ('MOTORWAY','MTWY'), - ('MOUNTAIN','MTN'), - ('MOUNTAINS','MTNS'), - ('MOUNTIN','MTN'), - ('MOUNT','MT'), - ('MSN','MSN'), - ('MSSN','MSN'), - ('MTIN','MTN'), - ('MT','MT'), - ('MTN','MTN'), - ('NCK','NCK'), - ('NECK','NCK'), - ('ORCHARD','ORCH'), - ('ORCH','ORCH'), - ('ORCHRD','ORCH'), - ('OVAL','OVAL'), - ('OVERPASS','OPAS'), - ('OVL','OVAL'), - ('PARK','PARK'), - ('PARKS','PARK'), - ('PARKWAY','PKWY'), - ('PARKWAYS','PKWY'), - ('PARKWY','PKWY'), - ('PASSAGE','PSGE'), - ('PASS','PASS'), - ('PATH','PATH'), - ('PATHS','PATH'), - ('PIKE','PIKE'), - ('PIKES','PIKE'), - ('PINE','PNE'), - ('PINES','PNES'), - ('PK','PARK'), - ('PKWAY','PKWY'), - ('PKWY','PKWY'), - ('PKWYS','PKWY'), - ('PKY','PKWY'), - ('PLACE','PL'), - ('PLAINES','PLNS'), - ('PLAIN','PLN'), - ('PLAINS','PLNS'), - ('PLAZA','PLZ'), - ('PLN','PLN'), - ('PLNS','PLNS'), - ('PL','PL'), - ('PLZA','PLZ'), - ('PLZ','PLZ'), - ('PNES','PNES'), - ('POINT','PT'), - ('POINTS','PTS'), - ('PORT','PRT'), - ('PORTS','PRTS'), - ('PRAIRIE','PR'), - ('PRARIE','PR'), - ('PRK','PARK'), - ('PR','PR'), - ('PRR','PR'), - ('PRT','PRT'), - ('PRTS','PRTS'), - ('PT','PT'), - ('PTS','PTS'), - ('RADIAL','RADL'), - ('RADIEL','RADL'), - ('RADL','RADL'), - ('RAD','RADL'), - ('RAMP','RAMP'), - ('RANCHES','RNCH'), - ('RANCH','RNCH'), - ('RAPID','RPD'), - ('RAPIDS','RPDS'), - ('RDGE','RDG'), - ('RDG','RDG'), - ('RDGS','RDGS'), - ('RD','RD'), - ('RDS','RDS'), - ('REST','RST'), - ('RIDGE','RDG'), - ('RIDGES','RDGS'), - ('RIVER','RIV'), - ('RIV','RIV'), - ('RIVR','RIV'), - ('RNCH','RNCH'), - ('RNCHS','RNCH'), - ('ROAD','RD'), - ('ROADS','RDS'), - ('ROUTE','RTE'), - ('ROW','ROW'), - ('RPD','RPD'), - ('RPDS','RPDS'), - ('RST','RST'), - ('RUE','RUE'), - ('RUN','RUN'), - ('RVR','RIV'), - ('SHL','SHL'), - ('SHLS','SHLS'), - ('SHOAL','SHL'), - ('SHOALS','SHLS'), - ('SHOAR','SHR'), - ('SHOARS','SHRS'), - ('SHORE','SHR'), - ('SHORES','SHRS'), - ('SHR','SHR'), - ('SHRS','SHRS'), - ('SKYWAY','SKWY'), - ('SMT','SMT'), - ('SPG','SPG'), - ('SPGS','SPGS'), - ('SPNG','SPG'), - ('SPNGS','SPGS'), - ('SPRING','SPG'), - ('SPRINGS','SPGS'), - ('SPRNG','SPG'), - ('SPRNGS','SPGS'), - ('SPUR','SPUR'), - ('SPURS','SPUR'), - ('SQRE','SQ'), - ('SQR','SQ'), - ('SQRS','SQS'), - ('SQ','SQ'), - ('SQUARE','SQ'), - ('SQUARES','SQS'), - ('SQU','SQ'), - ('STA','STA'), - ('STATION','STA'), - ('STATN','STA'), - ('STN','STA'), - ('STRA','STRA'), - ('STRAVEN','STRA'), - ('STRAVENUE','STRA'), - ('STRAVE','STRA'), - ('STRAVN','STRA'), - ('STRAV','STRA'), - ('STREAM','STRM'), - ('STREETS','STS'), - ('STREET','ST'), - ('STREME','STRM'), - ('STRM','STRM'), - ('STR','ST'), - ('STRT','ST'), - ('STRVN','STRA'), - ('STRVNUE','STRA'), - ('ST','ST'), - ('SUMIT','SMT'), - ('SUMITT','SMT'), - ('SUMMIT','SMT'), - ('TERRACE','TER'), - ('TERR','TER'), - ('TER','TER'), - ('THROUGHWAY','TRWY'), - ('TPKE','TPKE'), - ('TPK','TPKE'), - ('TRACES','TRCE'), - ('TRACE','TRCE'), - ('TRACKS','TRAK'), - ('TRACK','TRAK'), - ('TRAFFICWAY','TRFY'), - ('TRAILS','TRL'), - ('TRAIL','TRL'), - ('TRAK','TRAK'), - ('TRCE','TRCE'), - ('TRFY','TRFY'), - ('TRKS','TRAK'), - ('TRK','TRAK'), - ('TRLS','TRL'), - ('TRL','TRL'), - ('TRNPK','TPKE'), - ('TRPK','TPKE'), - ('TR','TRL'), - ('TUNEL','TUNL'), - ('TUNLS','TUNL'), - ('TUNL','TUNL'), - ('TUNNELS','TUNL'), - ('TUNNEL','TUNL'), - ('TUNNL','TUNL'), - ('TURNPIKE','TPKE'), - ('TURNPK','TPKE'), - ('UNDERPASS','UPAS'), - ('UNIONS','UNS'), - ('UNION','UN'), - ('UN','UN'), - ('VALLEYS','VLYS'), - ('VALLEY','VLY'), - ('VALLY','VLY'), - ('VDCT','IA'), - ('VIADCT','VIA'), - ('VIADUCT','IA'), - ('VIADUCT','VIA'), - ('VIA','VIA'), - ('VIEWS','VWS'), - ('VIEW','VW'), - ('VILLAGES','VLGS'), - ('VILLAGE','VLG'), - ('VILLAG','VLG'), - ('VILLE','VL'), - ('VILLG','VLG'), - ('VILLIAGE','VLG'), - ('VILL','VLG'), - ('VISTA','VIS'), - ('VIST','VIS'), - ('VIS','VIS'), - ('VLGS','VLGS'), - ('VLG','VLG'), - ('VLLY','VLY'), - ('VL','VL'), - ('VLYS','VLYS'), - ('VLY','VLY'), - ('VSTA','VIS'), - ('VST','VIS'), - ('VWS','VWS'), - ('VW','VW'), - ('WALKS','WALK'), - ('WALK','WALK'), - ('WALL','WALL'), - ('WAYS','WAYS'), - ('WAY','WAY'), - ('WELLS','WLS'), - ('WELL','WL'), - ('WLS','WLS'), - ('WY','WAY'), - ('XING','XING'); - FOR _r IN (SELECT * FROM mig_usps_suffixes) LOOP - street1 := REGEXP_REPLACE( street1, ' ' || _r.suffix_from || '$', ' ' || _r.suffix_to, 'i'); + --RAISE INFO 'suffix = %', suffix; + FOR _r IN (SELECT * FROM migration_tools.usps_suffixes) LOOP + suffix := REGEXP_REPLACE( suffix, _r.suffix_from, _r.suffix_to, 'i'); END LOOP; - RETURN street1; + RETURN suffix; END; -$$ LANGUAGE PLPGSQL STRICT VOLATILE; +$$ LANGUAGE PLPGSQL STRICT STABLE; + +CREATE OR REPLACE FUNCTION migration_tools.normalize_address_suffix (TEXT) RETURNS TEXT AS $$ + BEGIN + RETURN CASE + WHEN $1 ~ '\s\S+$' THEN REGEXP_REPLACE( $1, '^(.*\s)(\S+)$', '\1' ) || migration_tools._normalize_address_suffix( REGEXP_REPLACE( $1, '^(.*\s)(\S+)$', '\2' ) ) + ELSE $1 + END; + END; +$$ LANGUAGE PLPGSQL STRICT STABLE; CREATE OR REPLACE FUNCTION migration_tools.rebarcode (o TEXT, t BIGINT) RETURNS TEXT AS $$ DECLARE @@ -3401,6 +3543,27 @@ CREATE OR REPLACE FUNCTION migration_tools.get_marc_tags (TEXT, TEXT, TEXT, TEXT return \@texts; $$ LANGUAGE PLPERLU STABLE; +CREATE OR REPLACE FUNCTION migration_tools.get_marc_tags_filtered (TEXT, TEXT, TEXT, TEXT, TEXT) RETURNS TEXT[] AS $$ + my ($marcxml, $tag, $subfield, $delimiter, $match) = @_; + + use MARC::Record; + use MARC::File::XML; + use MARC::Field; + + my @fields; + eval { + my $marc = MARC::Record->new_from_xml($marcxml, 'UTF-8'); + @fields = $marc->field($tag); + }; + my @texts; + foreach my $field (@fields) { + if ($field->as_string() =~ qr/$match/) { + push @texts, $field->as_string($subfield,$delimiter); + } + } + return \@texts; +$$ LANGUAGE PLPERLU STABLE; + CREATE OR REPLACE FUNCTION migration_tools.find_hold_matrix_matchpoint (INTEGER) RETURNS INTEGER AS $$ SELECT action.find_hold_matrix_matchpoint( (SELECT pickup_lib FROM action.hold_request WHERE id = $1), @@ -3559,25 +3722,25 @@ CREATE OR REPLACE FUNCTION migration_tools.handle_shelf (TEXT,TEXT,TEXT,INTEGER) IF x_org_found THEN EXECUTE 'UPDATE ' || quote_ident(table_name) || ' a' - || ' SET x_shelf = id FROM asset_copy_location b' + || ' SET x_shelf = b.id FROM asset_copy_location b' || ' WHERE BTRIM(UPPER(a.desired_shelf)) = BTRIM(UPPER(b.name))' || ' AND b.owning_lib = x_org' || ' AND NOT b.deleted'; EXECUTE 'UPDATE ' || quote_ident(table_name) || ' a' - || ' SET x_shelf = id FROM asset.copy_location b' + || ' SET x_shelf = b.id FROM asset.copy_location b' || ' WHERE BTRIM(UPPER(a.desired_shelf)) = BTRIM(UPPER(b.name))' || ' AND b.owning_lib = x_org' || ' AND x_shelf IS NULL' || ' AND NOT b.deleted'; ELSE EXECUTE 'UPDATE ' || quote_ident(table_name) || ' a' - || ' SET x_shelf = id FROM asset_copy_location b' + || ' SET x_shelf = b.id FROM asset_copy_location b' || ' WHERE BTRIM(UPPER(a.desired_shelf)) = BTRIM(UPPER(b.name))' || ' AND b.owning_lib = $1' || ' AND NOT b.deleted' USING org; EXECUTE 'UPDATE ' || quote_ident(table_name) || ' a' - || ' SET x_shelf = id FROM asset_copy_location b' + || ' SET x_shelf = b.id FROM asset_copy_location b' || ' WHERE BTRIM(UPPER(a.desired_shelf)) = BTRIM(UPPER(b.name))' || ' AND b.owning_lib = $1' || ' AND x_shelf IS NULL' @@ -3587,7 +3750,7 @@ CREATE OR REPLACE FUNCTION migration_tools.handle_shelf (TEXT,TEXT,TEXT,INTEGER) FOREACH o IN ARRAY org_list LOOP EXECUTE 'UPDATE ' || quote_ident(table_name) || ' a' - || ' SET x_shelf = id FROM asset.copy_location b' + || ' SET x_shelf = b.id FROM asset.copy_location b' || ' WHERE BTRIM(UPPER(a.desired_shelf)) = BTRIM(UPPER(b.name))' || ' AND b.owning_lib = $1 AND x_shelf IS NULL' || ' AND NOT b.deleted' @@ -3718,7 +3881,7 @@ CREATE OR REPLACE FUNCTION migration_tools.handle_org (TEXT,TEXT) RETURNS VOID A || ' ADD COLUMN x_org INTEGER'; EXECUTE 'UPDATE ' || quote_ident(table_name) || ' a' - || ' SET x_org = id FROM actor.org_unit b' + || ' SET x_org = b.id FROM actor.org_unit b' || ' WHERE BTRIM(a.desired_org) = BTRIM(b.shortname)'; EXECUTE 'SELECT migration_tools.assert( @@ -4132,9 +4295,25 @@ BEGIN END $$ LANGUAGE plpgsql; --- yet another subfield 9 function, this one only adds the $9 if the ind1 = 1 or 4 and ind2 = 0 or 1 -DROP FUNCTION IF EXISTS migration_tools.strict_add_sf9(TEXT,TEXT); -CREATE OR REPLACE FUNCTION migration_tools.strict_add_sf9(marc TEXT, new_9 TEXT) +DROP FUNCTION IF EXISTS migration_tools.munge_sf9(INTEGER,TEXT,TEXT); +CREATE OR REPLACE FUNCTION migration_tools.merge_group(bib_id INTEGER,new_sf9 TEXT,force TEXT DEFAULT 'false') + RETURNS BOOLEAN AS +$BODY$ +DECLARE + marc_xml TEXT; + new_marc TEXT; +BEGIN + SELECT marc FROM biblio.record_entry WHERE id = bib_id INTO marc_xml; + + SELECT munge_sf9(marc_xml,new_sf9,force) INTO new_marc; + UPDATE biblio.record_entry SET marc = new_marc WHERE id = bib_id; + + RETURN true; +END; +$BODY$ LANGUAGE plpgsql; + +DROP FUNCTION IF EXISTS migration_tools.munge_sf9(TEXT,TEXT,TEXT); +CREATE OR REPLACE FUNCTION migration_tools.munge_sf9(marc_xml TEXT, new_9_to_set TEXT, force TEXT) RETURNS TEXT LANGUAGE plperlu AS $function$ @@ -4150,6 +4329,7 @@ binmode(STDERR, ':utf8'); my $marc_xml = shift; my $new_9_to_set = shift; +my $force = shift; $marc_xml =~ s/(.........)./${1}a/; @@ -4168,10 +4348,12 @@ return $marc_xml->as_xml_record() unless @uris; foreach my $field (@uris) { my $ind1 = $field->indicator('1'); if (!defined $ind1) { next; } - if ($ind1 ne '1' && $ind1 ne '4') { next; } + if ($ind1 ne '1' && $ind1 ne '4' && $force eq 'false') { next; } + if ($ind1 ne '1' && $ind1 ne '4' && $force eq 'true') { $field->set_indicator(1,'4'); } my $ind2 = $field->indicator('2'); if (!defined $ind2) { next; } - if ($ind2 ne '0' && $ind2 ne '1') { next; } + if ($ind2 ne '0' && $ind2 ne '1' && $force eq 'false') { next; } + if ($ind2 ne '0' && $ind2 ne '1' && $force eq 'true') { $field->set_indicator(2,'0'); } $field->add_subfields( '9' => $new_9_to_set ); } @@ -4179,10 +4361,8 @@ return $marc_xml->as_xml_record(); $function$; --- yet another subfield 9 function, this one only adds the $9 and forces --- ind1 = 4 if not already ind1 = 1 or 4 and ind2 = 0 if not already ind2 = 0 or 1 -DROP FUNCTION IF EXISTS migration_tools.force_add_sf9(TEXT,TEXT); -CREATE OR REPLACE FUNCTION migration_tools.force_add_sf9(marc TEXT, new_9 TEXT) +DROP FUNCTION IF EXISTS migration_tools.owner_change_sf9_substring_match(TEXT,TEXT,TEXT); +CREATE OR REPLACE FUNCTION migration_tools.owner_change_sf9_substring_match (marc_xml TEXT, substring_old_value TEXT, new_value TEXT, fix_indicators TEXT) RETURNS TEXT LANGUAGE plperlu AS $function$ @@ -4197,7 +4377,9 @@ binmode(STDOUT, ':utf8'); binmode(STDERR, ':utf8'); my $marc_xml = shift; -my $new_9_to_set = shift; +my $substring_old_value = shift; +my $new_value = shift; +my $fix_indicators = shift; $marc_xml =~ s/(.........)./${1}a/; @@ -4215,95 +4397,30 @@ return $marc_xml->as_xml_record() unless @uris; foreach my $field (@uris) { my $ind1 = $field->indicator('1'); - if (!defined $ind1) { next; } - if ($ind1 ne '1' && $ind1 ne '4') { $field->set_indicator(1,'4'); } - my $ind2 = $field->indicator('2'); - if (!defined $ind2) { next; } - if ($ind2 ne '0' && $ind2 ne '1') { $field->set_indicator(2,'0'); } - $field->add_subfields( '9' => $new_9_to_set ); -} - -return $marc_xml->as_xml_record(); - -$function$; - --- alternate adding subfield 9 function in that it adds them to existing tags where the 856$u matches a correct value only -DROP FUNCTION IF EXISTS migration_tools.add_sf9(TEXT,TEXT,TEXT); -CREATE OR REPLACE FUNCTION migration_tools.add_sf9(marc TEXT, partial_u TEXT, new_9 TEXT) - RETURNS TEXT - LANGUAGE plperlu -AS $function$ -use strict; -use warnings; - -use MARC::Record; -use MARC::File::XML (BinaryEncoding => 'utf8'); - -binmode(STDERR, ':bytes'); -binmode(STDOUT, ':utf8'); -binmode(STDERR, ':utf8'); - -my $marc_xml = shift; -my $matching_u_text = shift; -my $new_9_to_set = shift; - -$marc_xml =~ s/(.........)./${1}a/; - -eval { - $marc_xml = MARC::Record->new_from_xml($marc_xml); -}; -if ($@) { - #elog("could not parse $bibid: $@\n"); - import MARC::File::XML (BinaryEncoding => 'utf8'); - return; -} - -my @uris = $marc_xml->field('856'); -return unless @uris; - -foreach my $field (@uris) { - my $sfu = $field->subfield('u'); + if (defined $ind1) { + if ($ind1 ne '1' && $ind1 ne '4' && $fix_indicators eq 'true') { + $field->set_indicator(1,'4'); + } + } my $ind2 = $field->indicator('2'); - if (!defined $ind2) { next; } - if ($ind2 ne '0') { next; } - if (!defined $sfu) { next; } - if ($sfu =~ m/$matching_u_text/ or $matching_u_text eq 'pineapple') { - $field->add_subfields( '9' => $new_9_to_set ); - last; + if (defined $ind2) { + if ($ind2 ne '0' && $ind2 ne '1' && $fix_indicators eq 'true') { + $field->set_indicator(2,'0'); + } } + if ($field->as_string('9') =~ qr/$substring_old_value/) { + $field->delete_subfield('9'); + $field->add_subfields( '9' => $new_value ); + } + $marc_xml->delete_field($field); # -- we're going to dedup and add them back } +my %hash = (map { ($_->as_usmarc => $_) } @uris); # -- courtesy of an old Mike Rylander post :-) +$marc_xml->insert_fields_ordered( values( %hash ) ); + return $marc_xml->as_xml_record(); $function$; - -DROP FUNCTION IF EXISTS migration_tools.add_sf9(BIGINT, TEXT, TEXT, REGCLASS); -CREATE OR REPLACE FUNCTION migration_tools.add_sf9(bib_id BIGINT, target_u_text TEXT, sf9_text TEXT, bib_table REGCLASS) - RETURNS BOOLEAN AS -$BODY$ -DECLARE - source_xml TEXT; - new_xml TEXT; - r BOOLEAN; -BEGIN - - EXECUTE 'SELECT marc FROM ' || bib_table || ' WHERE id = ' || bib_id INTO source_xml; - - SELECT migration_tools.add_sf9(source_xml, target_u_text, sf9_text) INTO new_xml; - - r = FALSE; - new_xml = '$_$' || new_xml || '$_$'; - - IF new_xml != source_xml THEN - EXECUTE 'UPDATE ' || bib_table || ' SET marc = ' || new_xml || ' WHERE id = ' || bib_id; - r = TRUE; - END IF; - - RETURN r; - -END; -$BODY$ LANGUAGE plpgsql; - -- strip marc tag DROP FUNCTION IF EXISTS migration_tools.strip_tag(TEXT,TEXT); CREATE OR REPLACE FUNCTION migration_tools.strip_tag(marc TEXT, tag TEXT) @@ -5124,3 +5241,86 @@ BEGIN RETURN TRUE; END $function$; + + +-- convenience function for handling item barcode collisions in asset_copy_legacy + +CREATE OR REPLACE FUNCTION migration_tools.handle_asset_barcode_collisions(migration_schema TEXT) RETURNS VOID AS $function$ +DECLARE + x_barcode TEXT; + x_id BIGINT; + row_count NUMERIC; + internal_collision_count NUMERIC := 0; + incumbent_collision_count NUMERIC := 0; +BEGIN + FOR x_barcode IN SELECT barcode FROM asset_copy_legacy WHERE x_migrate GROUP BY 1 HAVING COUNT(*) > 1 + LOOP + FOR x_id IN SELECT id FROM asset_copy WHERE barcode = x_barcode + LOOP + UPDATE asset_copy SET barcode = migration_schema || '_internal_collision_' || id || '_' || barcode WHERE id = x_id; + GET DIAGNOSTICS row_count = ROW_COUNT; + internal_collision_count := internal_collision_count + row_count; + END LOOP; + END LOOP; + RAISE INFO '% internal collisions', internal_collision_count; + FOR x_barcode IN SELECT a.barcode FROM asset.copy a, asset_copy_legacy b WHERE x_migrate AND a.deleted IS FALSE AND a.barcode = b.barcode + LOOP + FOR x_id IN SELECT id FROM asset_copy_legacy WHERE barcode = x_barcode + LOOP + UPDATE asset_copy_legacy SET barcode = migration_schema || '_incumbent_collision_' || id || '_' || barcode WHERE id = x_id; + GET DIAGNOSTICS row_count = ROW_COUNT; + incumbent_collision_count := incumbent_collision_count + row_count; + END LOOP; + END LOOP; + RAISE INFO '% incumbent collisions', incumbent_collision_count; +END +$function$ LANGUAGE plpgsql; + +-- convenience function for handling patron barcode/usrname collisions in actor_usr_legacy +-- this should be ran prior to populating actor_card + +CREATE OR REPLACE FUNCTION migration_tools.handle_actor_barcode_collisions(migration_schema TEXT) RETURNS VOID AS $function$ +DECLARE + x_barcode TEXT; + x_id BIGINT; + row_count NUMERIC; + internal_collision_count NUMERIC := 0; + incumbent_barcode_collision_count NUMERIC := 0; + incumbent_usrname_collision_count NUMERIC := 0; +BEGIN + FOR x_barcode IN SELECT usrname FROM actor_usr_legacy WHERE x_migrate GROUP BY 1 HAVING COUNT(*) > 1 + LOOP + FOR x_id IN SELECT id FROM actor_usr_legacy WHERE x_migrate AND usrname = x_barcode + LOOP + UPDATE actor_usr_legacy SET usrname = migration_schema || '_internal_collision_' || id || '_' || usrname WHERE id = x_id; + GET DIAGNOSTICS row_count = ROW_COUNT; + internal_collision_count := internal_collision_count + row_count; + END LOOP; + END LOOP; + RAISE INFO '% internal usrname/barcode collisions', internal_collision_count; + + FOR x_barcode IN + SELECT a.barcode FROM actor.card a, actor_usr_legacy b WHERE x_migrate AND a.barcode = b.usrname + LOOP + FOR x_id IN SELECT DISTINCT id FROM actor_usr_legacy WHERE x_migrate AND usrname = x_barcode + LOOP + UPDATE actor_usr_legacy SET usrname = migration_schema || '_incumbent_barcode_collision_' || id || '_' || usrname WHERE id = x_id; + GET DIAGNOSTICS row_count = ROW_COUNT; + incumbent_barcode_collision_count := incumbent_barcode_collision_count + row_count; + END LOOP; + END LOOP; + RAISE INFO '% incumbent barcode collisions', incumbent_barcode_collision_count; + + FOR x_barcode IN + SELECT a.usrname FROM actor.usr a, actor_usr_legacy b WHERE x_migrate AND a.deleted IS FALSE AND a.usrname = b.usrname + LOOP + FOR x_id IN SELECT DISTINCT id FROM actor_usr_legacy WHERE x_migrate AND usrname = x_barcode + LOOP + UPDATE actor_usr_legacy SET usrname = migration_schema || '_incumbent_usrname_collision_' || id || '_' || usrname WHERE id = x_id; + GET DIAGNOSTICS row_count = ROW_COUNT; + incumbent_usrname_collision_count := incumbent_usrname_collision_count + row_count; + END LOOP; + END LOOP; + RAISE INFO '% incumbent usrname collisions (post barcode collision munging)', incumbent_usrname_collision_count; +END +$function$ LANGUAGE plpgsql;