From: Sarah Hoffmann <lonvia@denofr.de>
Date: Mon, 2 Nov 2020 10:41:53 +0000 (+0100)
Subject: use different area estimates for large countries
X-Git-Tag: v3.6.0~36^2
X-Git-Url: https://git.openstreetmap.org/nominatim.git/commitdiff_plain/fa574ae9fd1303c31759da21f116ebdd44e22885

use different area estimates for large countries
---

diff --git a/sql/functions/placex_triggers.sql b/sql/functions/placex_triggers.sql
index 433ceb92..6848140a 100644
--- a/sql/functions/placex_triggers.sql
+++ b/sql/functions/placex_triggers.sql
@@ -909,7 +909,7 @@ BEGIN
   END IF;
 
   IF NEW.rank_address = 0 THEN
-    max_rank := geometry_to_rank(NEW.rank_search, NEW.geometry);
+    max_rank := geometry_to_rank(NEW.rank_search, NEW.geometry, NEW.country_code);
   ELSEIF NEW.rank_address > 25 THEN
     max_rank := 25;
   ELSE
diff --git a/sql/functions/ranking.sql b/sql/functions/ranking.sql
index e918f924..a84269fe 100644
--- a/sql/functions/ranking.sql
+++ b/sql/functions/ranking.sql
@@ -60,7 +60,7 @@ LANGUAGE plpgsql IMMUTABLE;
 -- 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)
+CREATE OR REPLACE FUNCTION geometry_to_rank(search_rank SMALLINT, geometry GEOMETRY, country_code TEXT)
   RETURNS SMALLINT
   AS $$
 DECLARE
@@ -74,6 +74,15 @@ BEGIN
     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