created test_ version of new admin1 geocoder function

geocoder/admin1/sql/geocoder.sql

@@ -2,37 +2,73 @@
 
 --- 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 
+  IF inputcountry IS NULL THEN
-    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)
+    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, geom, CASE WHEN geom IS NULL THEN FALSE ELSE TRUE END AS success
+        q, inputcountry, geom, CASE WHEN geom IS NULL THEN FALSE ELSE TRUE END AS success
       FROM (
         SELECT 
           q, (
-          SELECT the_geom 
+            SELECT the_geom FROM qs_adm1 WHERE global_id = (
-          FROM global_province_polygons
+              SELECT global_id
-          WHERE p.c = ANY (synonyms) 
+              FROM admin1_synonyms
-          AND iso3 = p.i
+              WHERE name_ =  lower(regexp_replace(d.q, '[^a-zA-Z\u00C0-\u00ff]+', '', 'g'))::text
-          -- To calculate frequency, I simply counted the number of users
+                AND adm0_a3 = adm0
-          -- we had signed up in each country. Countries with more users, 
+              LIMIT 1
-          -- we favor higher in the geocoder :)
+              )
-          ORDER BY frequency DESC LIMIT 1
             ) geom
-      FROM p) n
+        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
-$$ LANGUAGE 'plpgsql' SECURITY DEFINER;
+$$ LANGUAGE 'plpgsql';
-Text array, country array
+
+
+--Text array, country array
 
 --- Usage
 
@@ -44,40 +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
 
-
+  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 
-  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
     RETURN NEXT ret;
   END LOOP;
   RETURN;
 END
-$$ LANGUAGE 'plpgsql' SECURITY DEFINER;
+$$ LANGUAGE 'plpgsql';
+

geocoder/setup/indexes.sql

@@ -4,6 +4,10 @@ CREATE INDEX idx_admin0_synonyms_name_ ON admin0_synonyms (name_);
 CREATE INDEX idx_admin0_synonyms_rank ON admin0_synonyms (rank);
 -- CREATE INDEX idx_admin0_synonyms_name_rank ON admin0_synonyms (name_, rank);
 
+-- Index on admin1 id
+CREATE UNIQUE INDEX idx_qs_adm1_global_id ON qs_adm1 (global_id)
+CREATE INDEX idx_admin1_synonyms_name_adm0 ON admin1_synonyms (name_, adm0_a3)
+
 -- create indexes on polygon table
 CREATE UNIQUE INDEX idx_ne_admin0_v3_adm0_a3 ON ne_admin0_v3 (adm0_a3);