will/sonoma-import
1
0
Fork 0
mirror of https://github.com/zyphlar/sonoma-import.git synced 2024-03-08 15:07:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update necessary sql

Browse Source
This commit is contained in:
Will Bradley 2020-12-21 14:09:50 -08:00
parent f5bf0bb964
commit 24c15b41e1
2 changed files with 37 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
.gitignore vendored
View File

@ -6,5 +6,8 @@
*.shx *.shx
*.xml *.xml
# backups
original_data-backup.sql
# VIM stuff # VIM stuff
*.swp *.swp

62
conflation.sql
View File

@ -12,6 +12,7 @@ CREATE INDEX ON sonoma_county_building_outlines USING GIST (loc_geom);
-- added fields for the parcels table -- added fields for the parcels table
ALTER TABLE parcels__public_ ALTER TABLE parcels__public_
ADD COLUMN "addr:housenumber" text,
ADD COLUMN "addr:street" text, ADD COLUMN "addr:street" text,
ADD COLUMN loc_geom geometry(multipolygon,32616), ADD COLUMN loc_geom geometry(multipolygon,32616),
ADD COLUMN building_count integer, ADD COLUMN building_count integer,
@ -22,34 +23,38 @@ UPDATE parcels__public_ SET loc_geom = ST_MakeValid(ST_Transform(geom,32616));
CREATE INDEX ON parcels__public_ USING GIST (loc_geom); CREATE INDEX ON parcels__public_ USING GIST (loc_geom);
-- parse and expand parcel street addresses -- parse and expand parcel street addresses
UPDATE parcels__public_ SET "addr:street" = initcap(addrst)||' '|| update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+)$', '\1')),
CASE "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+)$', '\2')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+)'
WHEN upper(addrsf) = 'AV' THEN 'Avenue' AND situsfmt1 NOT LIKE '%NONE' AND situsfmt1 NOT SIMILAR TO '([0-9]+) ([A-Z ]+) (AVE|DR|RD|ST|LN|CT|PL|CIR|TER|BLVD|PKWY|HWY)' AND situsfmt1 NOT SIMILAR TO '% ([^A-Z]+)';
WHEN upper(addrsf) = 'DR' THEN 'Drive' -- select situsfmt1, "addr:housenumber", "addr:street" from parcels__public_ limit 100;
WHEN upper(addrsf) = 'RD' THEN 'Road' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z]+) AVE$', '\1')),
WHEN upper(addrsf) = 'ST' THEN 'Street' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) AVE$', '\2 Avenue')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) AVE';
WHEN upper(addrsf) = 'LN' THEN 'Lane' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) DR$', '\1')),
WHEN upper(addrsf) = 'CT' THEN 'Court' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) DR$', '\2 Drive')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) DR';
WHEN upper(addrsf) = 'PL' THEN 'Place' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) RD$', '\1')),
WHEN upper(addrsf) = 'CR' THEN 'Circle' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) RD$', '\2 Road')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) RD';
WHEN upper(addrsf) = 'TE' THEN 'Terrace' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) ST$', '\1')),
WHEN upper(addrsf) = 'PK' THEN 'Park' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) ST$', '\2 Street')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) ST';
WHEN upper(addrsf) = 'WY' THEN 'Way' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) LN$', '\1')),
WHEN upper(addrsf) = 'BV' THEN 'Boulevard' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) LN$', '\2 Lane')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) LN';
WHEN upper(addrsf) = 'PW' THEN 'Parkway' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) CT$', '\1')),
WHEN upper(addrsf) = 'TL' THEN 'Trail' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) CT$', '\2 Court')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) CT';
WHEN upper(addrsf) = 'HW' THEN 'Highway' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) PL$', '\1')),
WHEN upper(addrsf) = 'WA' THEN 'Way' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) PL$', '\2 Place')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) PL';
WHEN upper(addrsf) = 'TR' THEN 'Terrace' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) CIR$', '\1')),
WHEN upper(addrsf) = 'SQ' THEN 'Square' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) CIR$', '\2 Circle')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) CIR';
WHEN upper(addrsf) = 'AL' THEN 'Alley' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) TER$', '\1')),
WHEN upper(addrsf) = 'BL' THEN 'Boulevard' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) TER$', '\2 Terrace')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) TER';
WHEN upper(addrsf) = 'CI' THEN 'Circle' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) BLVD$', '\1')),
WHEN upper(addrsf) = 'PT' THEN 'Point' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) BLVD$', '\2 Boulevard')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) BLVD';
WHEN upper(addrsf) = 'PI' THEN 'Pike' update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) PKWY$', '\1')),
WHEN upper(addrsf) = 'LA' THEN 'Lane' "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) PKWY$', '\2 Parkway')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) PKWY';
ELSE '' -- NULL cases mostly have the suffix in the name field update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) HWY$', '\1')),
END; "addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) ([A-Z ]+) HWY$', '\2 Highway')) where situsfmt1 SIMILAR TO '([0-9]+) ([A-Z ]+) HWY';
-- select situsfmt1, "addr:housenumber", "addr:street" from parcels__public_ limit 100;
update parcels__public_ SET "addr:housenumber" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) HWY ([0-9]+)$', '\1')),
"addr:street" = initcap(REGEXP_REPLACE(situsfmt1, '^([0-9]+) HWY ([0-9]+)$', 'Highway \2')) where situsfmt1 SIMILAR TO '([0-9]+) HWY ([0-9]+)';
-- select situsfmt1, "addr:housenumber", "addr:street" from parcels__public_ where situsfmt1 SIMILAR TO '([0-9]+) HWY ([0-9]+)';
-- identify repeating parcels (indicates multiple addresses associated with buildings) -- identify repeating parcels (indicates multiple addresses associated with buildings)
@ -64,6 +69,7 @@ UPDATE parcels__public_ SET repeating = TRUE
FROM geom_counts2 FROM geom_counts2
WHERE ids @> ARRAY[gid]; WHERE ids @> ARRAY[gid];
-------------------------------------------------------TODO
-- identify parcels with multiple buildings -- identify parcels with multiple buildings
UPDATE parcels__public_ SET building_count = NULL WHERE building_count IS NOT NULL; UPDATE parcels__public_ SET building_count = NULL WHERE building_count IS NOT NULL;