cartodb4/data-services
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge branch 'master' of https://github.com/CartoDB/data-services

Browse Source 
merge
This commit is contained in:
Chris Henrick
2014-08-31 19:36:17 -04:00
parent 27b1678cfa a2d3addcef
commit 342fc8b4c1
6 changed files with 159 additions and 117 deletions
Download Patch File Download Diff File Expand all files Collapse all files

177
geocoder/admin1/sql/geocoder.sql
View File

@@ -1,97 +1,74 @@
--- Usage
--SELECT (geocode_admin1_polygons(Array['TX','Cuidad Real', 'sevilla'])).*
--- Function
CREATE OR REPLACE FUNCTION test_geocode_admin1_polygons(name text[])
RETURNS SETOF geocode_admin_v1 AS $$
DECLARE
ret geocode_admin_v1%rowtype;
BEGIN
FOR ret IN
SELECT
q, geom, CASE WHEN geom IS NULL THEN FALSE ELSE TRUE END AS success
FROM (
SELECT
q, (
SELECT the_geom
FROM global_province_polygons
WHERE d.c = ANY (synonyms)
-- To calculate frequency, I simply counted the number of users
-- we had signed up in each country. Countries with more users,
-- we favor higher in the geocoder :)
ORDER BY frequency DESC LIMIT 1
) geom
FROM (SELECT trim(replace(lower(unnest(name)),'.',' ')) c, unnest(name) q) d
) v
LOOP
RETURN NEXT ret;
END LOOP;
RETURN;
END
$$ LANGUAGE 'plpgsql' SECURITY DEFINER;
Text array, country name
-- CREATE OR REPLACE FUNCTION test_geocode_admin1_polygons(name text[])
-- RETURNS SETOF geocode_admin_v1 AS $$
-- DECLARE
-- ret geocode_admin_v1%rowtype;
-- BEGIN
-- -- FOR ret IN
-- RETURN QUERY
-- SELECT
-- d.q, n.the_geom as geom,
-- CASE WHEN s.adm1_code IS NULL then FALSE ELSE TRUE END AS success
-- FROM (
-- SELECT
-- q, lower(regexp_replace(q, '[^a-zA-Z\u00C0-\u00ff]+', '', 'g'))::text x
-- FROM (SELECT unnest(name) q) g
-- ) d
-- LEFT OUTER JOIN
-- admin1_synonyms s ON name_ = d.x
-- LEFT OUTER JOIN
-- ne_admin1_v3 n ON s.adm1_code = n.adm1_code;
-- END
-- $$ LANGUAGE 'plpgsql' SECURITY DEFINER;
--- Usage
--- SELECT (geocode_admin1_polygons(Array['az', 'Texas'], 'Ecuador')).*
--- Function
CREATE OR REPLACE FUNCTION test_geocode_admin1_polygons(name text[], inputcountry text)
RETURNS SETOF geocode_admin_v1 AS $$
RETURNS SETOF geocode_admin_country_v1 AS $$
DECLARE
ret geocode_admin_v1%rowtype;
ret geocode_admin_country_v1%rowtype;
adm0 TEXT;
adm0_check BOOLEAN := TRUE;
BEGIN
FOR ret IN WITH
p AS (SELECT r.c, r.q, (SELECT iso3 FROM country_decoder WHERE lower(inputcountry) = ANY (synonyms)) i FROM (SELECT trim(replace(lower(unnest(name)),'.',' ')) c, unnest(name) q) r)
SELECT
q, geom, CASE WHEN geom IS NULL THEN FALSE ELSE TRUE END AS success
FROM (
SELECT
q, (
SELECT the_geom
FROM global_province_polygons
WHERE p.c = ANY (synonyms)
AND iso3 = p.i
-- To calculate frequency, I simply counted the number of users
-- we had signed up in each country. Countries with more users,
-- we favor higher in the geocoder :)
ORDER BY frequency DESC LIMIT 1
) geom
FROM p) n
LOOP
RETURN NEXT ret;
END LOOP;
IF inputcountry IS NULL THEN
adm0_check = FALSE;
END IF;
IF trim(inputcountry)='' THEN
adm0_check = FALSE;
END IF;
IF adm0_check IS TRUE THEN
SELECT INTO adm0 adm0_a3 FROM admin0_synonyms WHERE name_ = lower(regexp_replace(inputcountry, '[^a-zA-Z\u00C0-\u00ff]+', '', 'g'))::text LIMIT 1;
FOR ret IN
SELECT
q, inputcountry, geom, CASE WHEN geom IS NULL THEN FALSE ELSE TRUE END AS success
FROM (
SELECT
q, (
SELECT the_geom FROM qs_adm1 WHERE global_id = (
SELECT global_id
FROM admin1_synonyms
WHERE name_ = lower(regexp_replace(d.q, '[^a-zA-Z\u00C0-\u00ff]+', '', 'g'))::text
AND adm0_a3 = adm0
LIMIT 1
)
) geom
FROM (SELECT unnest(name) q) d
) v
LOOP
RETURN NEXT ret;
END LOOP;
--Handle cases where country couldn't be found
ELSE
FOR ret IN
SELECT
q, inputcountry, geom, CASE WHEN geom IS NULL THEN FALSE ELSE TRUE END AS success
FROM (
SELECT
q, (
SELECT the_geom FROM qs_adm1 WHERE global_id = (
SELECT global_id
FROM admin1_synonyms
WHERE name_ = lower(regexp_replace(d.q, '[^a-zA-Z\u00C0-\u00ff]+', '', 'g'))::text
LIMIT 1
)
) geom
FROM (SELECT unnest(name) q) d
) v
LOOP
RETURN NEXT ret;
END LOOP;
END IF;
RETURN;
END
$$ LANGUAGE 'plpgsql' SECURITY DEFINER;
Text array, country array
$$ LANGUAGE 'plpgsql';
--Text array, country array
--- Usage
@@ -103,40 +80,12 @@ CREATE OR REPLACE FUNCTION test_geocode_admin1_polygons(names text[], country te
RETURNS SETOF geocode_admin_country_v1 AS $$
DECLARE
ret geocode_admin_country_v1%rowtype;
nans TEXT[];
BEGIN
SELECT array_agg(p) INTO nans FROM (SELECT unnest(names) p, unnest(country) c) g WHERE c IS NULL;
IF 0 < array_length(nans, 1) THEN
SELECT array_agg(p), array_agg(c) INTO names, country FROM (SELECT unnest(names) p, unnest(country) c) g WHERE c IS NOT NULL;
FOR ret IN SELECT g.q, NULL as c, g.geom, g.success FROM (SELECT (geocode_admin1_polygons(nans)).*) g LOOP
RETURN NEXT ret;
END LOOP;
END IF;
FOR ret IN WITH
p AS (SELECT r.p, r.q, c, (SELECT iso3 FROM country_decoder WHERE lower(r.c) = ANY (synonyms)) i FROM (SELECT trim(replace(lower(unnest(names)),'.',' ')) p, unnest(names) q, unnest(country) c) r)
SELECT
q, c, geom, CASE WHEN geom IS NULL THEN FALSE ELSE TRUE END AS success
FROM (
SELECT
q, c, (
SELECT the_geom
FROM global_province_polygons
WHERE p.p = ANY (synonyms)
AND iso3 = p.i
-- To calculate frequency, I simply counted the number of users
-- we had signed up in each country. Countries with more users,
-- we favor higher in the geocoder :)
ORDER BY frequency DESC LIMIT 1
) geom
FROM p) n
LOOP
FOR ret IN SELECT (test_geocode_admin1_polygons(array_agg(n), c)).* FROM (SELECT unnest(names) n, unnest(country) c) a GROUP BY c LOOP
RETURN NEXT ret;
END LOOP;
RETURN;
END
$$ LANGUAGE 'plpgsql' SECURITY DEFINER;
$$ LANGUAGE 'plpgsql';