contributions: some additional rules for AI use

[nominatim.git] / lib-sql / functions / interpolation.sql

diff --git a/lib-sql/functions/interpolation.sql b/lib-sql/functions/interpolation.sql
index 9bb9102172b81169028ba302a52fa8d7b7676a0b..2fd21e8a86f907a3c99b9ec30814c0e059cf060c 100644 (file)
--- a/lib-sql/functions/interpolation.sql
+++ b/lib-sql/functions/interpolation.sql
@@ -67,10 +67,6 @@ BEGIN
     END LOOP;
   END IF;
 
-  IF parent_place_id is null THEN
-    RETURN 0;
-  END IF;
-
   RETURN parent_place_id;
 END;
 $$
@@ -125,6 +121,8 @@ LANGUAGE plpgsql;
 CREATE OR REPLACE FUNCTION osmline_insert()
   RETURNS TRIGGER
   AS $$
+DECLARE
+  centroid GEOMETRY;
 BEGIN
   NEW.place_id := nextval('seq_place');
   NEW.indexed_date := now();
@@ -139,10 +137,11 @@ BEGIN
       END IF;
 
       NEW.indexed_status := 1; --STATUS_NEW
-      NEW.country_code := lower(get_country_code(NEW.linegeo));
+      centroid := get_center_point(NEW.linegeo);
+      NEW.country_code := lower(get_country_code(centroid));
 
       NEW.partition := get_partition(NEW.country_code);
-      NEW.geometry_sector := geometry_sector(NEW.partition, NEW.linegeo);
+      NEW.geometry_sector := geometry_sector(NEW.partition, centroid);
   END IF;
 
   RETURN NEW;
@@ -164,7 +163,7 @@ DECLARE
   newend INTEGER;
   moddiff SMALLINT;
   linegeo GEOMETRY;
-  splitline GEOMETRY;
+  splitpoint FLOAT;
   sectiongeo GEOMETRY;
   postcode TEXT;
   stepmod SMALLINT;
@@ -180,8 +179,15 @@ BEGIN
   END IF;
 
   NEW.parent_place_id := get_interpolation_parent(NEW.token_info, NEW.partition,
-                                                 ST_PointOnSurface(NEW.linegeo),
-                                                 NEW.linegeo);
+                                                  get_center_point(NEW.linegeo),
+                                                  NEW.linegeo);
+
+  -- Cannot find a parent street. We will not be able to display a reliable
+  -- address, so drop entire interpolation.
+  IF NEW.parent_place_id is NULL THEN
+    DELETE FROM location_property_osmline where place_id = OLD.place_id;
+    RETURN NULL;
+  END IF;
 
   NEW.token_info := token_strip_info(NEW.token_info);
   IF NEW.address ? '_inherited' THEN
@@ -219,19 +225,32 @@ BEGIN
           -- formatted postcode and therefore 'postcode' contains a derived
           -- variant.
           CASE WHEN address ? 'postcode' THEN placex.postcode ELSE NULL::text END as postcode,
-          substring(address->'housenumber','[0-9]+')::integer as hnr
+          (address->'housenumber')::integer as hnr
         FROM placex, generate_series(1, array_upper(waynodes, 1)) nodeidpos
         WHERE osm_type = 'N' and osm_id = waynodes[nodeidpos]::BIGINT
               and address is not NULL and address ? 'housenumber'
+              and address->'housenumber' ~ '^[0-9]{1,6}$'
+              and ST_Distance(NEW.linegeo, geometry) < 0.0005
         ORDER BY nodeidpos
     LOOP
       {% if debug %}RAISE WARNING 'processing point % (%)', nextnode.hnr, ST_AsText(nextnode.geometry);{% endif %}
       IF linegeo is null THEN
         linegeo := NEW.linegeo;
       ELSE
-        splitline := ST_Split(ST_Snap(linegeo, nextnode.geometry, 0.0005), nextnode.geometry);
-        sectiongeo := ST_GeometryN(splitline, 1);
-        linegeo := ST_GeometryN(splitline, 2);
+        splitpoint := ST_LineLocatePoint(linegeo, nextnode.geometry);
+        IF splitpoint = 0 THEN
+          -- Corner case where the splitpoint falls on the first point
+          -- and thus would not return a geometry. Skip that section.
+          sectiongeo := NULL;
+        ELSEIF splitpoint = 1 THEN
+          -- Point is at the end of the line.
+          sectiongeo := linegeo;
+          linegeo := NULL;
+        ELSE
+          -- Split the line.
+          sectiongeo := ST_LineSubstring(linegeo, 0, splitpoint);
+          linegeo := ST_LineSubstring(linegeo, splitpoint, 1);
+        END IF;
       END IF;
 
       IF prevnode.hnr is not null
@@ -239,6 +258,9 @@ BEGIN
          -- regularly mapped housenumbers.
          -- (Conveniently also fails if one of the house numbers is not a number.)
          and abs(prevnode.hnr - nextnode.hnr) > NEW.step
+         -- If the interpolation geometry is broken or two nodes are at the
+         -- same place, then splitting might produce a point. Ignore that.
+         and ST_GeometryType(sectiongeo) = 'ST_LineString'
       THEN
         IF prevnode.hnr < nextnode.hnr THEN
           startnumber := prevnode.hnr;
@@ -300,12 +322,12 @@ BEGIN
                   NEW.address, postcode,
                   NEW.country_code, NEW.geometry_sector, 0);
         END IF;
+      END IF;
 
-        -- early break if we are out of line string,
-        -- might happen when a line string loops back on itself
-        IF ST_GeometryType(linegeo) != 'ST_LineString' THEN
-            RETURN NEW;
-        END IF;
+      -- early break if we are out of line string,
+      -- might happen when a line string loops back on itself
+      IF linegeo is null or ST_GeometryType(linegeo) != 'ST_LineString' THEN
+          RETURN NEW;
       END IF;
 
       prevnode := nextnode;