Merge pull request #192 from CartoDB/182-removing_tests

Removing namedplace tests and adding granularity on them

common-data/README.md

@@ -1247,6 +1247,84 @@ Data library
 
 **Source URL**: http://www.mass.gov/anf/research-and-tech/it-serv-and-support/application-serv/office-of-geographic-information-massgis/datalayers/structures.html
 
+#### Municipalities of Spain
+**Description**: Geometries for the municipalities of Spain. Official dataset from Instituto Geográfico Nacional de España.
+
+**Source**: [Instituto Geográfico Nacional](http://www.ign.es/)
+
+**License**: Public data - Attribution required  
+
+**Table Name**: ign_spanish_adm3_municipalities
+
+**Sync Table**: `false`
+
+**Source URL**: http://centrodedescargas.cnig.es/CentroDescargas/equipamiento/lineas_limite.zip
+
+#### Municipalities of Spain with displaced Canary Islands
+**Description**: Geometries for the municipalities of Spain with the Canary Islands displaced. Official dataset from Instituto Geográfico Nacional de España.
+
+**Source**: [Instituto Geográfico Nacional](http://www.ign.es/)
+
+**License**: Public data - Attribution required  
+
+**Table Name**: ign_spanish_adm3_municipalities_displaced_canary
+
+**Sync Table**: `false`
+
+**Source URL**: http://centrodedescargas.cnig.es/CentroDescargas/equipamiento/lineas_limite.zip
+
+#### Provinces of Spain
+**Description**: Geometries for the provinces of Spain. Official dataset from Instituto Geográfico Nacional de España.
+
+**Source**: [Instituto Geográfico Nacional](http://www.ign.es/)
+
+**License**: Public data - Attribution required 
+
+**Table Name**: ign_spanish_adm2_provinces
+
+**Sync Table**: `false`
+
+**Source URL**: http://centrodedescargas.cnig.es/CentroDescargas/equipamiento/lineas_limite.zip
+
+#### Provinces of Spain with displaced Canary Islands
+**Description**: Geometries for the provinces of Spain with the Canary Islands displaced. Official dataset from Instituto Geográfico Nacional de España.
+
+**Source**: [Instituto Geográfico Nacional](http://www.ign.es/)
+
+**License**: Public data - Attribution required
+
+**Table Name**: ign_spanish_adm2_provinces_displaced_canary
+
+**Sync Table**: `false`
+
+**Source URL**: http://centrodedescargas.cnig.es/CentroDescargas/equipamiento/lineas_limite.zip
+
+#### Autonomous Communities of Spain.
+**Description**: Geometries for  the autonomous communities of Spain. Official dataset from Instituto Geográfico Nacional de España.
+
+**Source**: [Instituto Geográfico Nacional](http://www.ign.es/)
+
+**License**: Public data - Attribution required
+
+**Table Name**: ign_spanish_adm1_ccaa
+
+**Sync Table**: `false`
+
+**Source URL**: http://centrodedescargas.cnig.es/CentroDescargas/equipamiento/lineas_limite.zip
+
+#### Autonomous communities of Spain with displaced Canary Islands
+**Description**: Geometries for  the autonomous communities of Spain with  the Canary Islands displaced. Official dataset from Instituto Geográfico Nacional de España.
+
+**Source**: [Instituto Geográfico Nacional](http://www.ign.es/)
+
+**License**: Public data - Attribution required
+
+**Table Name**: ign_spanish_adm1_ccaa_displaced_canary
+
+**Sync Table**: `false`
+
+**Source URL**: http://centrodedescargas.cnig.es/CentroDescargas/equipamiento/lineas_limite.zip
+
 ***
 
 ### Historic

geocoder/extension/expected/40_namedplaces_test.out

@@ -46,13 +46,25 @@ SELECT (geocode_namedplace(Array['Portland', 'Portland', 'New York City'], Array
  Portland      | Maine  | USA |      | f
 (3 rows)
 
+SELECT namedplace_guess_country(Array['granada', 'jaen', 'cordoba', 'madrid', 'valladolid']);
+ namedplace_guess_country 
+--------------------------
+ 
+(1 row)
+
 -- Add a named place source
 COPY global_cities_alternates_limited (geoname_id, name, the_geom, created_at, updated_at, the_geom_webmercator, preferred, lowername, cartodb_id, admin1_geonameid, iso2, admin1) FROM stdin;
 COPY global_cities_points_limited (geoname_id, name, asciiname, altnames, featclass, featcode, iso2, admin1, admin2, population, the_geom, created_at, updated_at, the_geom_webmercator, cartodb_id, lowername) FROM stdin;
 -- Check that the geocoding function is callable, should return success = true
-SELECT (geocode_namedplace(Array['Barcelona'])).*
+SELECT (geocode_namedplace(Array['Barcelona'])).*;
      q     |                        geom                        | success 
 -----------+----------------------------------------------------+---------
  Barcelona | 0101000020E6100000CA15DEE522E653C0A4C2D842902B4540 | t
 (1 row)
 
+SELECT namedplace_guess_country(Array['Barcelona']);
+ namedplace_guess_country 
+--------------------------
+ ES
+(1 row)
+

geocoder/extension/sql/0.0.1/40_namedplaces.sql

@@ -288,6 +288,40 @@ CREATE OR REPLACE FUNCTION geocode_namedplace(places text[]) RETURNS SETOF geoco
 END
 $$;
 
+CREATE OR REPLACE FUNCTION namedplace_guess_country(places text[])
+RETURNS text AS $$
+DECLARE
+  country_code text;
+  threshold CONSTANT float := 0.8;
+  input_length integer := array_length(places, 1);
+BEGIN
+  BEGIN
+    WITH hist AS (
+        SELECT count(DISTINCT(lower(p.s), gp.iso2)) AS c, iso2
+        FROM global_cities_points_limited gp
+        inner join (SELECT unnest(places) AS s) p
+        ON (gp.lowername = lower(s))
+        GROUP BY iso2
+      ),
+      best_two AS (
+        SELECT iso2, c
+          FROM hist
+          WHERE c > input_length * threshold
+          ORDER BY c DESC
+          LIMIT 2
+      )
+    SELECT iso2 INTO STRICT country_code
+      FROM (SELECT iso2, c, max(c) over() AS maxcount FROM best_two) bt
+      WHERE bt.c = bt.maxcount;
+    EXCEPTION
+      WHEN NO_DATA_FOUND OR too_many_rows THEN
+        RETURN NULL;
+  END;
+  RETURN country_code;
+END;
+$$ LANGUAGE 'plpgsql' SECURITY DEFINER IMMUTABLE;
+
+
 --------------------------------------------------------------------------------
 
 -- Support tables

geocoder/extension/sql/40_namedplaces_test.sql

@@ -5,6 +5,7 @@ SELECT (geocode_namedplace(Array['sunapee', 'sunapeeee', 'New York City', 'Madri
 SELECT (geocode_namedplace(Array['Portland', 'Portland', 'New York City'], Array['Maine',    'Oregon',    NULL], 'USA')).*;
 SELECT (geocode_namedplace(Array['Portland'], 'Oregon', 'USA')).*;
 SELECT (geocode_namedplace(Array['Portland', 'Portland', 'New York City'], Array['Maine',    'Oregon',    NULL], Array['USA'])).*;
+SELECT namedplace_guess_country(Array['granada', 'jaen', 'cordoba', 'madrid', 'valladolid']);
 
 -- Add a named place source
 COPY global_cities_alternates_limited (geoname_id, name, the_geom, created_at, updated_at, the_geom_webmercator, preferred, lowername, cartodb_id, admin1_geonameid, iso2, admin1) FROM stdin;
@@ -16,5 +17,6 @@ COPY global_cities_points_limited (geoname_id, name, asciiname, altnames, featcl
 \.
 
 -- Check that the geocoding function is callable, should return success = true
-SELECT (geocode_namedplace(Array['Barcelona'])).*
+SELECT (geocode_namedplace(Array['Barcelona'])).*;
+SELECT namedplace_guess_country(Array['Barcelona']);
 

geocoder/namedplace/sql/namedplace_country_guessing.sql

@@ -1,35 +0,0 @@
--- Return a guess about the country where the places are located, if possible.
--- E.g: SELECT namedplace_guess_country(Array['granada', 'jaen', 'cordoba', 'madrid', 'valladolid']); => NULL
--- E.g: SELECT namedplace_guess_country(Array['granada', 'jaén', 'córdoba', 'madrid', 'valladolid', 'peligros'])); => 'ES'
-CREATE OR REPLACE FUNCTION namedplace_guess_country(places text[])
-RETURNS text AS $$
-DECLARE
-  country_code text;
-  threshold CONSTANT float := 0.8;
-  input_length integer := array_length(places, 1);
-BEGIN
-  BEGIN
-    WITH hist AS (
-        SELECT count(DISTINCT(lower(p.s), gp.iso2)) AS c, iso2
-        FROM global_cities_points_limited gp
-        inner join (SELECT unnest(places) AS s) p
-        ON (gp.lowername = lower(s))
-        GROUP BY iso2
-      ),
-      best_two AS (
-        SELECT iso2, c
-          FROM hist
-          WHERE c > input_length * threshold
-          ORDER BY c DESC
-          LIMIT 2
-      )
-    SELECT iso2 INTO STRICT country_code
-      FROM (SELECT iso2, c, max(c) over() AS maxcount FROM best_two) bt
-      WHERE bt.c = bt.maxcount;
-    EXCEPTION
-      WHEN NO_DATA_FOUND OR too_many_rows THEN
-        RETURN NULL;
-  END;
-  RETURN country_code;
-END;
-$$ LANGUAGE 'plpgsql' SECURITY DEFINER IMMUTABLE;