Merge pull request #2038 from lonvia/addresses-for-large-areas

Improve addresses for large areas

settings/address-levels.json

@@ -64,7 +64,15 @@
           "peak" : [18, 0],
           "volcano" : [18, 0],
           "mountain_range" : [18, 0],
-          "sea" : [4, 0]
+          "sea" : [4, 0],
+          "tree" : 30,
+          "spring"  : 30,
+          "cave_entrance" : 30,
+          "geyser" : 30,
+          "hot_spring" : 30,
+          "rock" : 30,
+          "stone" : 30,
+          "" : [22, 0]
       },
       "waterway" : {
           "river" : [19, 0],

sql/functions/placex_triggers.sql

@@ -509,6 +509,8 @@ DECLARE
   addr_street TEXT;
   addr_place TEXT;
 
+  max_rank SMALLINT;
+
   name_vector INTEGER[];
   nameaddress_vector INTEGER[];
   addr_nameaddress_vector INTEGER[];
@@ -906,14 +908,19 @@ BEGIN
     --DEBUG: RAISE WARNING 'Country names updated';
   END IF;
 
-  SELECT * FROM insert_addresslines(NEW.place_id, NEW.partition,
+  IF NEW.rank_address = 0 THEN
-                                    CASE WHEN NEW.rank_address = 0 THEN NEW.rank_search
+    max_rank := geometry_to_rank(NEW.rank_search, NEW.geometry, NEW.country_code);
-                                         WHEN NEW.rank_address > 25 THEN 25::smallint
+  ELSEIF NEW.rank_address > 25 THEN
-                                         ELSE NEW.rank_address END,
+    max_rank := 25;
+  ELSE
+    max_rank = NEW.rank_address;
+  END IF;
+
+  SELECT * FROM insert_addresslines(NEW.place_id, NEW.partition, max_rank,
                                     NEW.address,
-                                    CASE WHEN NEW.rank_search >= 26
+                                    CASE WHEN (NEW.rank_address = 0 or
-                                             AND NEW.rank_search < 30
+                                               NEW.rank_search between 26 and 29)
-                                      THEN NEW.geometry ELSE NEW.centroid END)
+                                         THEN NEW.geometry ELSE NEW.centroid END)
     INTO NEW.parent_place_id, NEW.postcode, nameaddress_vector;
 
   --DEBUG: RAISE WARNING 'RETURN insert_addresslines: %, %, %', NEW.parent_place_id, NEW.postcode, nameaddress_vector;

sql/functions/ranking.sql

@@ -55,6 +55,53 @@ END;
 $$
 LANGUAGE plpgsql IMMUTABLE;
 
+-- Compute a base address rank from the extent of the given geometry.
+--
+-- This is all simple guess work. We don't need particularly good estimates
+-- here. This just avoids to have very high ranked address parts in features
+-- that span very large areas (or vice versa).
+CREATE OR REPLACE FUNCTION geometry_to_rank(search_rank SMALLINT, geometry GEOMETRY, country_code TEXT)
+  RETURNS SMALLINT
+  AS $$
+DECLARE
+  area FLOAT;
+BEGIN
+  IF ST_GeometryType(geometry) in ('ST_Polygon','ST_MultiPolygon') THEN
+      area := ST_Area(geometry);
+  ELSIF ST_GeometryType(geometry) in ('ST_LineString','ST_MultiLineString') THEN
+      area := (ST_Length(geometry)^2) * 0.1;
+  ELSE
+    RETURN search_rank;
+  END IF;
+
+  -- adjust for the fact that countries come in different sizes
+  IF country_code IN ('ca', 'au', 'ru') THEN
+    area := area / 5;
+  ELSIF country_code IN ('br', 'kz', 'cn', 'us', 'ne', 'gb', 'za', 'sa', 'id', 'eh', 'ml', 'tm') THEN
+    area := area / 3;
+  ELSIF country_code IN ('bo', 'ar', 'sd', 'mn', 'in', 'et', 'cd', 'mz', 'ly', 'cl', 'zm') THEN
+    area := area / 2;
+  END IF;
+
+  IF area > 1 THEN
+    RETURN 7;
+  ELSIF area > 0.1 THEN
+    RETURN 9;
+  ELSIF area > 0.01 THEN
+    RETURN 13;
+  ELSIF area > 0.001 THEN
+    RETURN 17;
+  ELSIF area > 0.0001 THEN
+    RETURN 19;
+  ELSIF area > 0.000005 THEN
+    RETURN 21;
+  END IF;
+
+   RETURN 23;
+END;
+$$
+LANGUAGE plpgsql IMMUTABLE;
+
 
 -- Guess a ranking for postcodes from country and postcode format.
 CREATE OR REPLACE FUNCTION get_postcode_rank(country_code VARCHAR(2), postcode TEXT,

test/bdd/db/import/placex.feature

@@ -216,13 +216,13 @@ Feature: Import into placex
           | object | rank_search | rank_address |
           | N2     | 18          | 0 |
           | N4     | 18          | 0 |
-          | N5     | 30          | 30 |
+          | N5     | 22          | 0 |
           | W2     | 18          | 0 |
           | R3     | 18          | 0 |
-          | R4     | 30          | 30 |
+          | R4     | 22          | 0 |
           | R5     | 4           | 0 |
           | R6     | 4           | 0 |
-          | W3     | 30          | 30 |
+          | W3     | 22          | 0 |
 
     Scenario: boundary ways for countries and states are ignored
         Given the named places

test/bdd/db/update/simple.feature

@@ -35,7 +35,7 @@ Feature: Update of simple objects
         When importing
         Then placex contains
           | object | rank_address |
-          | R1     | 30 |
+          | R1     | 0 |
           | R2     | 26 |
           | W1     | 30 |
         When marking for delete R1,R2,W1