Merge remote-tracking branch 'upstream/master'

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

diff --git a/lib-sql/functions/placex_triggers.sql b/lib-sql/functions/placex_triggers.sql
index fa7156ec904c396a8376b780d03c09a37045fb43..9463bb27a8ed4e50f04cfcc4850ba7b830936fe4 100644 (file)
--- a/lib-sql/functions/placex_triggers.sql
+++ b/lib-sql/functions/placex_triggers.sql
@@ -1,30 +1,41 @@
+-- SPDX-License-Identifier: GPL-2.0-only
+--
+-- This file is part of Nominatim. (https://nominatim.org)
+--
+-- Copyright (C) 2022 by the Nominatim developer community.
+-- For a full list of authors see the git log.
+

 -- Trigger functions for the placex table.
 
 -- Trigger functions for the placex table.
 

+-- Information returned by update preparation.
+DROP TYPE IF EXISTS prepare_update_info CASCADE;
+CREATE TYPE prepare_update_info AS (
+  name HSTORE,
+  address HSTORE,
+  rank_address SMALLINT,
+  country_code TEXT,
+  class TEXT,
+  type TEXT,
+  linked_place_id BIGINT
+);
+

 -- Retrieve the data needed by the indexer for updating the place.
 -- Retrieve the data needed by the indexer for updating the place.

---
--- Return parameters:
---  name            list of names
---  address         list of address tags, either from the object or a surrounding
---                  building
---  country_feature If the place is a country feature, this contains the
---                  country code, otherwise it is null.
-CREATE OR REPLACE FUNCTION placex_prepare_update(p placex,
-                                                 OUT name HSTORE,
-                                                 OUT address HSTORE,
-                                                 OUT country_feature VARCHAR,
-                                                 OUT linked_place_id BIGINT)
+CREATE OR REPLACE FUNCTION placex_indexing_prepare(p placex)
+  RETURNS prepare_update_info

   AS $$
 DECLARE
   location RECORD;
   AS $$
 DECLARE
   location RECORD;

+  result prepare_update_info;
+  extra_names HSTORE;

 BEGIN
   -- For POI nodes, check if the address should be derived from a surrounding
   -- building.
 BEGIN
   -- For POI nodes, check if the address should be derived from a surrounding
   -- building.

-  IF p.rank_search < 30 OR p.osm_type != 'N' OR p.address is not null THEN
-    address := p.address;
-  ELSE
+  IF p.rank_search < 30 OR p.osm_type != 'N' THEN
+    result.address := p.address;
+  ELSEIF p.address is null THEN

     -- The additional && condition works around the misguided query
     -- planner of postgis 3.0.
     -- The additional && condition works around the misguided query
     -- planner of postgis 3.0.

-    SELECT placex.address || hstore('_inherited', '') INTO address
+    SELECT placex.address || hstore('_inherited', '') INTO result.address

       FROM placex
      WHERE ST_Covers(geometry, p.centroid)
            and geometry && p.centroid
       FROM placex
      WHERE ST_Covers(geometry, p.centroid)
            and geometry && p.centroid


@@ -32,29 +43,57 @@ BEGIN
            and (placex.address ? 'housenumber' or placex.address ? 'street' or placex.address ? 'place')
            and rank_search = 30 AND ST_GeometryType(geometry) in ('ST_Polygon','ST_MultiPolygon')
      LIMIT 1;
            and (placex.address ? 'housenumber' or placex.address ? 'street' or placex.address ? 'place')
            and rank_search = 30 AND ST_GeometryType(geometry) in ('ST_Polygon','ST_MultiPolygon')
      LIMIT 1;

+  ELSE
+    result.address := p.address;
+    -- See if we can inherit addtional address tags from an interpolation.
+    -- These will become permanent.
+    FOR location IN
+      SELECT (address - 'interpolation'::text - 'housenumber'::text) as address
+        FROM place, planet_osm_ways w
+        WHERE place.osm_type = 'W' and place.address ? 'interpolation'
+              and place.geometry && p.geometry
+              and place.osm_id = w.id
+              and p.osm_id = any(w.nodes)
+    LOOP
+      result.address := location.address || result.address;
+    END LOOP;

   END IF;
 
   END IF;
 

-  address := address - '_unlisted_place'::TEXT;
-  name := p.name;
+  -- remove internal and derived names
+  result.address := result.address - '_unlisted_place'::TEXT;
+  SELECT hstore(array_agg(key), array_agg(value)) INTO result.name
+    FROM each(p.name) WHERE key not like '\_%';
+
+  result.class := p.class;
+  result.type := p.type;
+  result.country_code := p.country_code;
+  result.rank_address := p.rank_address;

 
   -- Names of linked places need to be merged in, so search for a linkable
   -- place already here.
   SELECT * INTO location FROM find_linked_place(p);
 
   IF location.place_id is not NULL THEN
 
   -- Names of linked places need to be merged in, so search for a linkable
   -- place already here.
   SELECT * INTO location FROM find_linked_place(p);
 
   IF location.place_id is not NULL THEN

-    linked_place_id := location.place_id;
+    result.linked_place_id := location.place_id;
+
+    IF location.name is not NULL THEN
+      {% if debug %}RAISE WARNING 'Names original: %, location: %', result.name, location.name;{% endif %}
+      -- Add all names from the place nodes that deviate from the name
+      -- in the relation with the prefix '_place_'. Deviation means that
+      -- either the value is different or a given key is missing completely
+      SELECT hstore(array_agg('_place_' || key), array_agg(value)) INTO extra_names
+        FROM each(location.name - result.name);
+      {% if debug %}RAISE WARNING 'Extra names: %', extra_names;{% endif %}
+
+      IF extra_names is not null THEN
+          result.name := result.name || extra_names;
+      END IF;

 
 

-    IF NOT location.name IS NULL THEN
-      name := location.name || name;
+      {% if debug %}RAISE WARNING 'Final names: %', result.name;{% endif %}

     END IF;
   END IF;
 
     END IF;
   END IF;
 

-  country_feature := CASE WHEN p.admin_level = 2
-                               and p.class = 'boundary' and p.type = 'administrative'
-                               and p.osm_type = 'R'
-                          THEN p.country_code
-                          ELSE null
-                     END;
+  RETURN result;

 END;
 $$
 LANGUAGE plpgsql STABLE;
 END;
 $$
 LANGUAGE plpgsql STABLE;


@@ -104,8 +143,7 @@ CREATE OR REPLACE FUNCTION find_parent_for_poi(poi_osm_type CHAR(1),
                                                poi_osm_id BIGINT,
                                                poi_partition SMALLINT,
                                                bbox GEOMETRY,
                                                poi_osm_id BIGINT,
                                                poi_partition SMALLINT,
                                                bbox GEOMETRY,

-                                               addr_street INTEGER[],
-                                               addr_place INTEGER[],
+                                               token_info JSONB,

                                                is_place_addr BOOLEAN)
   RETURNS BIGINT
   AS $$
                                                is_place_addr BOOLEAN)
   RETURNS BIGINT
   AS $$


@@ -119,23 +157,10 @@ BEGIN
   parent_place_id := find_associated_street(poi_osm_type, poi_osm_id);
 
   IF parent_place_id is null THEN
   parent_place_id := find_associated_street(poi_osm_type, poi_osm_id);
 
   IF parent_place_id is null THEN

-    parent_place_id := find_parent_for_address(addr_street, addr_place,
-                                               poi_partition, bbox);
+    parent_place_id := find_parent_for_address(token_info, poi_partition, bbox);

   END IF;
 
   IF parent_place_id is null and poi_osm_type = 'N' THEN
   END IF;
 
   IF parent_place_id is null and poi_osm_type = 'N' THEN

-    -- Is this node part of an interpolation?
-    FOR location IN
-      SELECT q.parent_place_id
-        FROM location_property_osmline q, planet_osm_ways x
-       WHERE q.linegeo && bbox and x.id = q.osm_id
-             and poi_osm_id = any(x.nodes)
-       LIMIT 1
-    LOOP
-      {% if debug %}RAISE WARNING 'Get parent from interpolation: %', location.parent_place_id;{% endif %}
-      RETURN location.parent_place_id;
-    END LOOP;
-

     FOR location IN
       SELECT p.place_id, p.osm_id, p.rank_search, p.address,
              coalesce(p.centroid, ST_Centroid(p.geometry)) as centroid
     FOR location IN
       SELECT p.place_id, p.osm_id, p.rank_search, p.address,
              coalesce(p.centroid, ST_Centroid(p.geometry)) as centroid


@@ -333,13 +358,15 @@ BEGIN
     WHERE s.place_id = parent_place_id;
 
   FOR addr_item IN
     WHERE s.place_id = parent_place_id;
 
   FOR addr_item IN

-    SELECT (get_addr_tag_rank(key, country)).*, match_tokens, search_tokens
-      FROM token_get_address_tokens(token_info)
-      WHERE not search_tokens <@ parent_address_vector
+    SELECT ranks.*, key,
+           token_get_address_search_tokens(token_info, key) as search_tokens
+      FROM token_get_address_keys(token_info) as key,
+           LATERAL get_addr_tag_rank(key, country) as ranks
+      WHERE not token_get_address_search_tokens(token_info, key) <@ parent_address_vector

   LOOP
     addr_place := get_address_place(in_partition, geometry,
                                     addr_item.from_rank, addr_item.to_rank,
   LOOP
     addr_place := get_address_place(in_partition, geometry,
                                     addr_item.from_rank, addr_item.to_rank,

-                                    addr_item.extent, addr_item.match_tokens);
+                                    addr_item.extent, token_info, addr_item.key);

 
     IF addr_place is null THEN
       -- No place found in OSM that matches. Make it at least searchable.
 
     IF addr_place is null THEN
       -- No place found in OSM that matches. Make it at least searchable.


@@ -446,15 +473,19 @@ BEGIN
   address_havelevel := array_fill(false, ARRAY[maxrank]);
 
   FOR location IN
   address_havelevel := array_fill(false, ARRAY[maxrank]);
 
   FOR location IN

-    SELECT (get_address_place(partition, geometry, from_rank, to_rank,
-                              extent, match_tokens)).*, search_tokens
-      FROM (SELECT (get_addr_tag_rank(key, country)).*, match_tokens, search_tokens
-              FROM token_get_address_tokens(token_info)) x
+    SELECT apl.*, key
+      FROM (SELECT extra.*, key
+              FROM token_get_address_keys(token_info) as key,
+                   LATERAL get_addr_tag_rank(key, country) as extra) x,
+           LATERAL get_address_place(partition, geometry, from_rank, to_rank,
+                              extent, token_info, key) as apl

       ORDER BY rank_address, distance, isguess desc
   LOOP
     IF location.place_id is null THEN
       {% if not db.reverse_only %}
       ORDER BY rank_address, distance, isguess desc
   LOOP
     IF location.place_id is null THEN
       {% if not db.reverse_only %}

-      nameaddress_vector := array_merge(nameaddress_vector, location.search_tokens);
+      nameaddress_vector := array_merge(nameaddress_vector,
+                                        token_get_address_search_tokens(token_info,
+                                                                        location.key));

       {% endif %}
     ELSE
       {% if not db.reverse_only %}
       {% endif %}
     ELSE
       {% if not db.reverse_only %}


@@ -616,10 +647,7 @@ BEGIN
 {% if not disable_diff_updates %}
   -- The following is not needed until doing diff updates, and slows the main index process down
 
 {% if not disable_diff_updates %}
   -- The following is not needed until doing diff updates, and slows the main index process down
 

-  IF NEW.osm_type = 'N' and NEW.rank_search > 28 THEN
-      -- might be part of an interpolation
-      result := osmline_reinsert(NEW.osm_id, NEW.geometry);
-  ELSEIF NEW.rank_address > 0 THEN
+  IF NEW.rank_address > 0 THEN

     IF (ST_GeometryType(NEW.geometry) in ('ST_Polygon','ST_MultiPolygon') AND ST_IsValid(NEW.geometry)) THEN
       -- Performance: We just can't handle re-indexing for country level changes
       IF st_area(NEW.geometry) < 1 THEN
     IF (ST_GeometryType(NEW.geometry) in ('ST_Polygon','ST_MultiPolygon') AND ST_IsValid(NEW.geometry)) THEN
       -- Performance: We just can't handle re-indexing for country level changes
       IF st_area(NEW.geometry) < 1 THEN


@@ -646,7 +674,7 @@ BEGIN
           -- roads may cause reparenting for >27 rank places
           update placex set indexed_status = 2 where indexed_status = 0 and rank_search > NEW.rank_search and ST_DWithin(placex.geometry, NEW.geometry, diameter);
           -- reparenting also for OSM Interpolation Lines (and for Tiger?)
           -- roads may cause reparenting for >27 rank places
           update placex set indexed_status = 2 where indexed_status = 0 and rank_search > NEW.rank_search and ST_DWithin(placex.geometry, NEW.geometry, diameter);
           -- reparenting also for OSM Interpolation Lines (and for Tiger?)

-          update location_property_osmline set indexed_status = 2 where indexed_status = 0 and ST_DWithin(location_property_osmline.linegeo, NEW.geometry, diameter);
+          update location_property_osmline set indexed_status = 2 where indexed_status = 0 and startnumber is not null and ST_DWithin(location_property_osmline.linegeo, NEW.geometry, diameter);

         ELSEIF NEW.rank_search >= 16 THEN
           -- up to rank 16, street-less addresses may need reparenting
           update placex set indexed_status = 2 where indexed_status = 0 and rank_search > NEW.rank_search and ST_DWithin(placex.geometry, NEW.geometry, diameter) and (rank_search < 28 or name is not null or address ? 'place');
         ELSEIF NEW.rank_search >= 16 THEN
           -- up to rank 16, street-less addresses may need reparenting
           update placex set indexed_status = 2 where indexed_status = 0 and rank_search > NEW.rank_search and ST_DWithin(placex.geometry, NEW.geometry, diameter) and (rank_search < 28 or name is not null or address ? 'place');


@@ -689,9 +717,6 @@ DECLARE
   parent_address_level SMALLINT;
   place_address_level SMALLINT;
 
   parent_address_level SMALLINT;
   place_address_level SMALLINT;
 

-  addr_street INTEGER[];
-  addr_place INTEGER[];
-

   max_rank SMALLINT;
 
   name_vector INTEGER[];
   max_rank SMALLINT;
 
   name_vector INTEGER[];


@@ -860,8 +885,6 @@ BEGIN
   END IF;
 
   NEW.housenumber := token_normalized_housenumber(NEW.token_info);
   END IF;
 
   NEW.housenumber := token_normalized_housenumber(NEW.token_info);

-  addr_street := token_addr_street_match_tokens(NEW.token_info);
-  addr_place := token_addr_place_match_tokens(NEW.token_info);

 
   NEW.postcode := null;
 
 
   NEW.postcode := null;
 


@@ -907,7 +930,7 @@ BEGIN
     NEW.parent_place_id := find_parent_for_poi(NEW.osm_type, NEW.osm_id,
                                                NEW.partition,
                                                ST_Envelope(NEW.geometry),
     NEW.parent_place_id := find_parent_for_poi(NEW.osm_type, NEW.osm_id,
                                                NEW.partition,
                                                ST_Envelope(NEW.geometry),

-                                               addr_street, addr_place,
+                                               NEW.token_info,

                                                is_place_address);
 
     -- If we found the road take a shortcut here.
                                                is_place_address);
 
     -- If we found the road take a shortcut here.


@@ -1040,16 +1063,22 @@ BEGIN
      AND NEW.class = 'boundary' AND NEW.type = 'administrative'
      AND NEW.country_code IS NOT NULL AND NEW.osm_type = 'R'
   THEN
      AND NEW.class = 'boundary' AND NEW.type = 'administrative'
      AND NEW.country_code IS NOT NULL AND NEW.osm_type = 'R'
   THEN

-    -- Update the list of country names. Adding an additional sanity
-    -- check here: make sure the country does overlap with the area where
-    -- we expect it to be as per static country grid.
+    -- Update the list of country names.
+    -- Only take the name from the largest area for the given country code
+    -- in the hope that this is the authoritive one.
+    -- Also replace any old names so that all mapping mistakes can
+    -- be fixed through regular OSM updates.

     FOR location IN
     FOR location IN

-      SELECT country_code FROM country_osm_grid
-       WHERE ST_Covers(geometry, NEW.centroid) and country_code = NEW.country_code
+      SELECT osm_id FROM placex
+       WHERE rank_search = 4 and osm_type = 'R'
+             and country_code = NEW.country_code
+       ORDER BY ST_Area(geometry) desc

        LIMIT 1
     LOOP
        LIMIT 1
     LOOP

-      {% if debug %}RAISE WARNING 'Updating names for country '%' with: %', NEW.country_code, NEW.name;{% endif %}
-      UPDATE country_name SET name = name || NEW.name WHERE country_code = NEW.country_code;
+      IF location.osm_id = NEW.osm_id THEN
+        {% if debug %}RAISE WARNING 'Updating names for country '%' with: %', NEW.country_code, NEW.name;{% endif %}
+        UPDATE country_name SET derived_name = NEW.name WHERE country_code = NEW.country_code;
+      END IF;

     END LOOP;
   END IF;
 
     END LOOP;
   END IF;