fix edge cases when linking nodes

[nominatim.git] / sql / functions.sql

diff --git a/sql/functions.sql b/sql/functions.sql
index 20e97c1c87ec167fa3ce41da9595480ff964c3dd..4b51ed3c84579b93e7e741532dd82fe1ac2284e8 100644 (file)
--- a/sql/functions.sql
+++ b/sql/functions.sql
@@ -769,6 +769,13 @@ BEGIN
           -- null record of right type
           select * from placex where osm_type = 'N' and osm_id = waynodes[nodeidpos]::INTEGER and type = 'house' limit 1 INTO nextnode;
           select ST_SetSRID(ST_Point(lon::float/10000000,lat::float/10000000),4326) from planet_osm_nodes where id = waynodes[nodeidpos] INTO nextnode.geometry;
           -- null record of right type
           select * from placex where osm_type = 'N' and osm_id = waynodes[nodeidpos]::INTEGER and type = 'house' limit 1 INTO nextnode;
           select ST_SetSRID(ST_Point(lon::float/10000000,lat::float/10000000),4326) from planet_osm_nodes where id = waynodes[nodeidpos] INTO nextnode.geometry;

+          IF nextnode.geometry IS NULL THEN
+            -- we don't have any information about this point, most likely
+            -- because an excerpt was updated and the node never imported
+            -- because the interpolation is outside the region of the excerpt.
+            -- Give up.
+            RETURN newpoints;
+          END IF;

         ELSE
           select * from placex where place_id = search_place_id INTO nextnode;
         END IF;
         ELSE
           select * from placex where place_id = search_place_id INTO nextnode;
         END IF;


@@ -1213,6 +1220,8 @@ DECLARE
 
   location_rank_search INTEGER;
   location_distance FLOAT;
 
   location_rank_search INTEGER;
   location_distance FLOAT;

+  location_parent GEOMETRY;
+  location_isaddress BOOLEAN;

 
   tagpairid INTEGER;
 
 
   tagpairid INTEGER;
 


@@ -1225,6 +1234,13 @@ DECLARE
   result BOOLEAN;
 BEGIN
 
   result BOOLEAN;
 BEGIN
 

+  -- deferred delete
+  IF OLD.indexed_status = 100 THEN
+    --DEBUG: RAISE WARNING 'placex_update_delete % %',NEW.osm_type,NEW.osm_id;
+    delete from placex where place_id = OLD.place_id;
+    RETURN NULL;
+  END IF;
+

   IF NEW.indexed_status != 0 OR OLD.indexed_status = 0 OR NEW.linked_place_id is not null THEN
     RETURN NEW;
   END IF;
   IF NEW.indexed_status != 0 OR OLD.indexed_status = 0 OR NEW.linked_place_id is not null THEN
     RETURN NEW;
   END IF;


@@ -1239,13 +1255,6 @@ BEGIN
     RETURN NEW;
   END IF;
 
     RETURN NEW;
   END IF;
 

-  -- deferred delete
-  IF OLD.indexed_status = 100 THEN
-    --DEBUG: RAISE WARNING 'placex_update_delete % %',NEW.osm_type,NEW.osm_id;
-    delete from placex where place_id = OLD.place_id;
-    RETURN NULL;
-  END IF;
-

   IF OLD.indexed_status != 0 THEN
     --DEBUG: RAISE WARNING 'placex_update_0 % %',NEW.osm_type,NEW.osm_id;
 
   IF OLD.indexed_status != 0 THEN
     --DEBUG: RAISE WARNING 'placex_update_0 % %',NEW.osm_type,NEW.osm_id;
 


@@ -1495,23 +1504,29 @@ BEGIN
 -- RAISE WARNING 'get_osm_rel_members, label';
       FOR relMember IN select get_osm_rel_members(relation_members,ARRAY['label']) as member LOOP
 
 -- RAISE WARNING 'get_osm_rel_members, label';
       FOR relMember IN select get_osm_rel_members(relation_members,ARRAY['label']) as member LOOP
 

-        select * from placex where osm_type = upper(substring(relMember.member,1,1))::char(1) 
-          and osm_id = substring(relMember.member,2,10000)::bigint order by rank_search desc limit 1 into linkedPlacex;
+        FOR linkedPlacex IN select * from placex where osm_type = upper(substring(relMember.member,1,1))::char(1) 
+          and osm_id = substring(relMember.member,2,10000)::bigint order by rank_search desc limit 1 LOOP

 
 

-        -- If we don't already have one use this as the centre point of the geometry
-        IF NEW.centroid IS NULL THEN
-          NEW.centroid := coalesce(linkedPlacex.centroid,st_centroid(linkedPlacex.geometry));
-        END IF;
+          -- If we don't already have one use this as the centre point of the geometry
+          IF NEW.centroid IS NULL THEN
+            NEW.centroid := coalesce(linkedPlacex.centroid,st_centroid(linkedPlacex.geometry));
+          END IF;

 
 

-        -- merge in the label name, re-init word vector
-        NEW.name := linkedPlacex.name || NEW.name;
-        name_vector := make_keywords(NEW.name);
+          -- merge in the label name, re-init word vector
+          IF NOT linkedPlacex.name IS NULL THEN
+            NEW.name := linkedPlacex.name || NEW.name;
+            name_vector := make_keywords(NEW.name);
+          END IF;
+
+          -- merge in extra tags
+          IF NOT linkedPlacex.extratags IS NULL THEN
+            NEW.extratags := linkedPlacex.extratags || NEW.extratags;
+          END IF;

 
 

-        -- merge in extra tags
-        NEW.extratags := linkedPlacex.extratags || NEW.extratags;
+          -- mark the linked place (excludes from search results)
+          UPDATE placex set linked_place_id = NEW.place_id where place_id = linkedPlacex.place_id;

 
 

-        -- mark the linked place (excludes from search results)
-        UPDATE placex set linked_place_id = NEW.place_id where place_id = linkedPlacex.place_id;
+        END LOOP;

 
       END LOOP;
 
 
       END LOOP;
 


@@ -1519,33 +1534,39 @@ BEGIN
 
         FOR relMember IN select get_osm_rel_members(relation_members,ARRAY['admin_center','admin_centre']) as member LOOP
 
 
         FOR relMember IN select get_osm_rel_members(relation_members,ARRAY['admin_center','admin_centre']) as member LOOP
 

-          select * from placex where osm_type = upper(substring(relMember.member,1,1))::char(1) 
-            and osm_id = substring(relMember.member,2,10000)::bigint order by rank_search desc limit 1 into linkedPlacex;
+          FOR linkedPlacex IN select * from placex where osm_type = upper(substring(relMember.member,1,1))::char(1) 
+            and osm_id = substring(relMember.member,2,10000)::bigint order by rank_search desc limit 1 LOOP

 
 

-          -- For an admin centre we also want a name match - still not perfect, for example 'new york, new york'
-          -- But that can be fixed by explicitly setting the label in the data
-          IF make_standard_name(NEW.name->'name') = make_standard_name(linkedPlacex.name->'name') 
-            AND NEW.rank_search = linkedPlacex.rank_search THEN
+            -- For an admin centre we also want a name match - still not perfect, for example 'new york, new york'
+            -- But that can be fixed by explicitly setting the label in the data
+            IF make_standard_name(NEW.name->'name') = make_standard_name(linkedPlacex.name->'name') 
+              AND NEW.rank_search = linkedPlacex.rank_search THEN

 
 
 
 

-            -- If we don't already have one use this as the centre point of the geometry
-            IF NEW.centroid IS NULL THEN
-              NEW.centroid := coalesce(linkedPlacex.centroid,st_centroid(linkedPlacex.geometry));
-            END IF;
+              -- If we don't already have one use this as the centre point of the geometry
+              IF NEW.centroid IS NULL THEN
+                NEW.centroid := coalesce(linkedPlacex.centroid,st_centroid(linkedPlacex.geometry));
+              END IF;

 
 

-            -- merge in the name, re-init word vector
-            NEW.name := linkedPlacex.name || NEW.name;
-            name_vector := make_keywords(NEW.name);
+              -- merge in the name, re-init word vector
+              IF NOT linkedPlacex.name IS NULL THEN
+                NEW.name := linkedPlacex.name || NEW.name;
+                name_vector := make_keywords(NEW.name);
+              END IF;

 
 

-            -- merge in extra tags
-            NEW.extratags := linkedPlacex.extratags || NEW.extratags;
+              -- merge in extra tags
+              IF NOT linkedPlacex.extratags IS NULL THEN
+                NEW.extratags := linkedPlacex.extratags || NEW.extratags;
+              END IF;

 
 

-            -- mark the linked place (excludes from search results)
-            UPDATE placex set linked_place_id = NEW.place_id where place_id = linkedPlacex.place_id;
+              -- mark the linked place (excludes from search results)
+              UPDATE placex set linked_place_id = NEW.place_id where place_id = linkedPlacex.place_id;

 
 

-            -- keep a note of the node id in case we need it for wikipedia in a bit
-            linked_node_id := linkedPlacex.osm_id;
-          END IF;
+              -- keep a note of the node id in case we need it for wikipedia in a bit
+              linked_node_id := linkedPlacex.osm_id;
+            END IF;
+
+          END LOOP;

 
         END LOOP;
 
 
         END LOOP;
 


@@ -1627,24 +1648,39 @@ BEGIN
 -- RAISE WARNING 'ISIN: %', isin_tokens;
 
     -- Process area matches
 -- RAISE WARNING 'ISIN: %', isin_tokens;
 
     -- Process area matches

-    location_rank_search := 100;
+    location_rank_search := 0;

     location_distance := 0;
     location_distance := 0;

--- RAISE WARNING '  getNearFeatures(%,''%'',%,''%'')',NEW.partition, place_centroid, search_maxrank, isin_tokens;
+    location_parent := NULL;
+    -- added ourself as address already
+    address_havelevel[NEW.rank_address] := true;
+    -- RAISE WARNING '  getNearFeatures(%,''%'',%,''%'')',NEW.partition, place_centroid, search_maxrank, isin_tokens;

     FOR location IN SELECT * from getNearFeatures(NEW.partition, place_centroid, search_maxrank, isin_tokens) LOOP
 
 --RAISE WARNING '  AREA: %',location;
 
     FOR location IN SELECT * from getNearFeatures(NEW.partition, place_centroid, search_maxrank, isin_tokens) LOOP
 
 --RAISE WARNING '  AREA: %',location;
 

-      IF location.rank_search < location_rank_search THEN
-        location_rank_search := location.rank_search;
+      IF location.rank_address != location_rank_search THEN
+        location_rank_search := location.rank_address;

         location_distance := location.distance * 1.5;
       END IF;
 
       IF location.distance < location_distance OR NOT location.isguess THEN
 
         location_distance := location.distance * 1.5;
       END IF;
 
       IF location.distance < location_distance OR NOT location.isguess THEN
 

+        location_isaddress := NOT address_havelevel[location.rank_address];
+        IF location_isaddress AND location.isguess AND location_parent IS NOT NULL THEN
+            location_isaddress := ST_Contains(location_parent,location.centroid);
+        END IF;
+
+        -- RAISE WARNING '% isaddress: %', location.place_id, location_isaddress;

         -- Add it to the list of search terms
         nameaddress_vector := array_merge(nameaddress_vector, location.keywords::integer[]);
         -- Add it to the list of search terms
         nameaddress_vector := array_merge(nameaddress_vector, location.keywords::integer[]);

-        INSERT INTO place_addressline VALUES (NEW.place_id, location.place_id, true, NOT address_havelevel[location.rank_address], location.distance, location.rank_address); 
-        address_havelevel[location.rank_address] := true;
+        INSERT INTO place_addressline VALUES (NEW.place_id, location.place_id, true, location_isaddress, location.distance, location.rank_address);
+
+        IF location_isaddress THEN
+            address_havelevel[location.rank_address] := true;
+            IF NOT location.isguess THEN
+                SELECT geometry FROM placex WHERE place_id = location.place_id INTO location_parent;
+            END IF;
+        END IF;

 
 --RAISE WARNING '  Terms: (%) %',location, nameaddress_vector;
 
 
 --RAISE WARNING '  Terms: (%) %',location, nameaddress_vector;
 


@@ -1683,13 +1719,13 @@ BEGIN
     -- for long ways we should add search terms for the entire length
     IF st_length(NEW.geometry) > 0.05 THEN
 
     -- for long ways we should add search terms for the entire length
     IF st_length(NEW.geometry) > 0.05 THEN
 

-      location_rank_search := 100;
+      location_rank_search := 0;

       location_distance := 0;
 
       FOR location IN SELECT * from getNearFeatures(NEW.partition, NEW.geometry, search_maxrank, isin_tokens) LOOP
 
       location_distance := 0;
 
       FOR location IN SELECT * from getNearFeatures(NEW.partition, NEW.geometry, search_maxrank, isin_tokens) LOOP
 

-        IF location.rank_search < location_rank_search THEN
-          location_rank_search := location.rank_search;
+        IF location.rank_address != location_rank_search THEN
+          location_rank_search := location.rank_address;

           location_distance := location.distance * 1.5;
         END IF;
 
           location_distance := location.distance * 1.5;
         END IF;
 


@@ -1741,9 +1777,9 @@ DECLARE
 BEGIN
   -- RAISE WARNING 'placex_delete % %',OLD.osm_type,OLD.osm_id;
 
 BEGIN
   -- RAISE WARNING 'placex_delete % %',OLD.osm_type,OLD.osm_id;
 

-  update placex set linked_place_id = null where linked_place_id = OLD.place_id;
+  update placex set linked_place_id = null, indexed_status = 2 where linked_place_id = OLD.place_id and indexed_status = 0;

   --DEBUG: RAISE WARNING 'placex_delete:01 % %',OLD.osm_type,OLD.osm_id;
   --DEBUG: RAISE WARNING 'placex_delete:01 % %',OLD.osm_type,OLD.osm_id;

-  update placex set indexed_status = 2 where linked_place_id = OLD.place_id and indexed_status = 0;
+  update placex set linked_place_id = null where linked_place_id = OLD.place_id;

   --DEBUG: RAISE WARNING 'placex_delete:02 % %',OLD.osm_type,OLD.osm_id;
 
   IF OLD.rank_address < 30 THEN
   --DEBUG: RAISE WARNING 'placex_delete:02 % %',OLD.osm_type,OLD.osm_id;
 
   IF OLD.rank_address < 30 THEN


@@ -1806,8 +1842,8 @@ BEGIN
 
   --DEBUG: RAISE WARNING 'delete: % % % %',OLD.osm_type,OLD.osm_id,OLD.class,OLD.type;
 
 
   --DEBUG: RAISE WARNING 'delete: % % % %',OLD.osm_type,OLD.osm_id,OLD.class,OLD.type;
 

-  -- deleting large polygons can have a massive effect ont he system - require manual intervention to let them through
-  IF st_area(OLD.geometry) > 2 THEN
+  -- deleting large polygons can have a massive effect on the system - require manual intervention to let them through
+  IF st_area(OLD.geometry) > 2 and st_isvalid(OLD.geometry) THEN

     insert into import_polygon_delete values (OLD.osm_type,OLD.osm_id,OLD.class,OLD.type);
     RETURN NULL;
   END IF;
     insert into import_polygon_delete values (OLD.osm_type,OLD.osm_id,OLD.class,OLD.type);
     RETURN NULL;
   END IF;


@@ -2156,7 +2192,7 @@ BEGIN
 
   FOR location IN select * from get_addressdata(for_place_id) where isaddress order by rank_address desc LOOP
     currresult := trim(get_name_by_language(location.name, languagepref));
 
   FOR location IN select * from get_addressdata(for_place_id) where isaddress order by rank_address desc LOOP
     currresult := trim(get_name_by_language(location.name, languagepref));

-    IF currresult != prevresult AND currresult IS NOT NULL THEN
+    IF currresult != prevresult AND currresult IS NOT NULL AND result[(100 - location.rank_address)] IS NULL THEN

       result[(100 - location.rank_address)] := trim(get_name_by_language(location.name, languagepref));
       prevresult := currresult;
     END IF;
       result[(100 - location.rank_address)] := trim(get_name_by_language(location.name, languagepref));
       prevresult := currresult;
     END IF;


@@ -2213,14 +2249,14 @@ BEGIN
   END IF;
 
   IF for_place_id IS NULL THEN
   END IF;
 
   IF for_place_id IS NULL THEN

-    select parent_place_id, calculated_country_code, housenumber, rank_address, postcode, name, class, type from placex 
+    select parent_place_id, calculated_country_code, housenumber, rank_search, postcode, name, class, type from placex 

       WHERE place_id = in_place_id and rank_address = 30 
       INTO for_place_id, searchcountrycode, searchhousenumber, searchrankaddress, searchpostcode, searchhousename, searchclass, searchtype;
   END IF;
 
   IF for_place_id IS NULL THEN
     for_place_id := in_place_id;
       WHERE place_id = in_place_id and rank_address = 30 
       INTO for_place_id, searchcountrycode, searchhousenumber, searchrankaddress, searchpostcode, searchhousename, searchclass, searchtype;
   END IF;
 
   IF for_place_id IS NULL THEN
     for_place_id := in_place_id;

-    select calculated_country_code, housenumber, rank_address, postcode, null from placex where place_id = for_place_id 
+    select calculated_country_code, housenumber, rank_search, postcode, null from placex where place_id = for_place_id 

       INTO searchcountrycode, searchhousenumber, searchrankaddress, searchpostcode, searchhousename;
   END IF;
 
       INTO searchcountrycode, searchhousenumber, searchrankaddress, searchpostcode, searchhousename;
   END IF;
 


@@ -2560,93 +2596,6 @@ END;
 $$
 LANGUAGE plpgsql;
 
 $$
 LANGUAGE plpgsql;
 

-CREATE OR REPLACE FUNCTION tigger_create_interpolation(linegeo GEOMETRY, in_startnumber INTEGER, 
-  in_endnumber INTEGER, interpolationtype TEXT, 
-  in_street TEXT, in_isin TEXT, in_postcode TEXT) RETURNS INTEGER
-  AS $$
-DECLARE
-  
-  startnumber INTEGER;
-  endnumber INTEGER;
-  stepsize INTEGER;
-  housenum INTEGER;
-  newpoints INTEGER;
-  numberrange INTEGER;
-  rangestartnumber INTEGER;
-  place_centroid GEOMETRY;
-  out_partition INTEGER;
-  out_parent_place_id BIGINT;
-  location RECORD;
-  address_street_word_id INTEGER;  
-
-BEGIN
-
-  IF in_endnumber > in_startnumber THEN
-    startnumber = in_startnumber;
-    endnumber = in_endnumber;
-  ELSE
-    startnumber = in_endnumber;
-    endnumber = in_startnumber;
-  END IF;
-
-  numberrange := endnumber - startnumber;
-  rangestartnumber := startnumber;
-
-  IF (interpolationtype = 'odd' AND startnumber%2 = 0) OR (interpolationtype = 'even' AND startnumber%2 = 1) THEN
-    startnumber := startnumber + 1;
-    stepsize := 2;
-  ELSE
-    IF (interpolationtype = 'odd' OR interpolationtype = 'even') THEN
-      stepsize := 2;
-    ELSE -- everything else assumed to be 'all'
-      stepsize := 1;
-    END IF;
-  END IF;
-
-  -- Filter out really broken tiger data
-  IF numberrange > 0 AND (numberrange::float/stepsize::float > 500) 
-                    AND ST_length(linegeo)/(numberrange::float/stepsize::float) < 0.000001 THEN
-    RAISE WARNING 'Road too short for number range % to % on %, % (%)',startnumber,endnumber,in_street,in_isin,
-                  ST_length(linegeo)/(numberrange::float/stepsize::float);    
-    RETURN 0;
-  END IF;
-
-  place_centroid := ST_Centroid(linegeo);
-  out_partition := get_partition(place_centroid, 'us');
-  out_parent_place_id := null;
-
-  address_street_word_id := get_name_id(make_standard_name(in_street));
-  IF address_street_word_id IS NOT NULL THEN
-    FOR location IN SELECT * from getNearestNamedRoadFeature(out_partition, place_centroid, address_street_word_id) LOOP
-      out_parent_place_id := location.place_id;
-    END LOOP;
-  END IF;
-
-  IF out_parent_place_id IS NULL THEN
-    FOR location IN SELECT place_id FROM getNearestParellelRoadFeature(out_partition, linegeo) LOOP
-      out_parent_place_id := location.place_id;
-    END LOOP;    
-  END IF;
-
-  IF out_parent_place_id IS NULL THEN
-    FOR location IN SELECT place_id FROM getNearestRoadFeature(out_partition, place_centroid) LOOP
-      out_parent_place_id := location.place_id;
-    END LOOP;    
-  END IF;
-
-  newpoints := 0;
-  FOR housenum IN startnumber..endnumber BY stepsize LOOP
-    insert into location_property_tiger (place_id, partition, parent_place_id, housenumber, postcode, centroid)
-    values (nextval('seq_place'), out_partition, out_parent_place_id, housenum, in_postcode,
-      ST_Line_Interpolate_Point(linegeo, (housenum::float-rangestartnumber::float)/numberrange::float));
-    newpoints := newpoints + 1;
-  END LOOP;
-
-  RETURN newpoints;
-END;
-$$
-LANGUAGE plpgsql;
-

 CREATE OR REPLACE FUNCTION aux_create_property(pointgeo GEOMETRY, in_housenumber TEXT, 
   in_street TEXT, in_isin TEXT, in_postcode TEXT, in_countrycode char(2)) RETURNS INTEGER
   AS $$
 CREATE OR REPLACE FUNCTION aux_create_property(pointgeo GEOMETRY, in_housenumber TEXT, 
   in_street TEXT, in_isin TEXT, in_postcode TEXT, in_countrycode char(2)) RETURNS INTEGER
   AS $$