X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/c325e1b0398ad786ab46f98707a641fbcd67739a..b4c87a09a194a7d966724ce48ba2d8ef0d044ca0:/sql/functions.sql diff --git a/sql/functions.sql b/sql/functions.sql index c025f221..e0d0897f 100644 --- a/sql/functions.sql +++ b/sql/functions.sql @@ -5,7 +5,6 @@ -- score integer --); - CREATE OR REPLACE FUNCTION getclasstypekey(c text, t text) RETURNS TEXT AS $$ DECLARE @@ -83,18 +82,24 @@ END; $$ LANGUAGE 'plpgsql' IMMUTABLE; +-- returns NULL if the word is too common CREATE OR REPLACE FUNCTION getorcreate_word_id(lookup_word TEXT) RETURNS INTEGER AS $$ DECLARE lookup_token TEXT; return_word_id INTEGER; + count INTEGER; BEGIN lookup_token := trim(lookup_word); - SELECT min(word_id) FROM word WHERE word_token = lookup_token and class is null and type is null into return_word_id; + SELECT min(word_id), max(search_name_count) FROM word WHERE word_token = lookup_token and class is null and type is null into return_word_id, count; IF return_word_id IS NULL THEN return_word_id := nextval('seq_word'); - INSERT INTO word VALUES (return_word_id, lookup_token, regexp_replace(lookup_token,E'([^0-9])\\1+',E'\\1','g'), null, null, null, null, 0, null); + INSERT INTO word VALUES (return_word_id, lookup_token, null, null, null, null, 0); + ELSE + IF count > get_maxwordfreq() THEN + return_word_id := NULL; + END IF; END IF; RETURN return_word_id; END; @@ -112,7 +117,7 @@ BEGIN SELECT min(word_id) FROM word WHERE word_token = lookup_token and class='place' and type='house' into return_word_id; IF return_word_id IS NULL THEN return_word_id := nextval('seq_word'); - INSERT INTO word VALUES (return_word_id, lookup_token, null, null, 'place', 'house', null, 0, null); + INSERT INTO word VALUES (return_word_id, lookup_token, null, 'place', 'house', null, 0); END IF; RETURN return_word_id; END; @@ -130,7 +135,7 @@ BEGIN SELECT min(word_id) FROM word WHERE word_token = lookup_token and country_code=lookup_country_code into return_word_id; IF return_word_id IS NULL THEN return_word_id := nextval('seq_word'); - INSERT INTO word VALUES (return_word_id, lookup_token, null, null, null, null, lookup_country_code, 0, null); + INSERT INTO word VALUES (return_word_id, lookup_token, null, null, null, lookup_country_code, 0); END IF; RETURN return_word_id; END; @@ -148,7 +153,7 @@ BEGIN SELECT min(word_id) FROM word WHERE word_token = lookup_token and class=lookup_class and type = lookup_type into return_word_id; IF return_word_id IS NULL THEN return_word_id := nextval('seq_word'); - INSERT INTO word VALUES (return_word_id, lookup_token, null, null, lookup_class, lookup_type, null, 0, null); + INSERT INTO word VALUES (return_word_id, lookup_token, null, lookup_class, lookup_type, null, 0); END IF; RETURN return_word_id; END; @@ -166,7 +171,7 @@ BEGIN SELECT min(word_id) FROM word WHERE word_token = lookup_token into return_word_id; IF return_word_id IS NULL THEN return_word_id := nextval('seq_word'); - INSERT INTO word VALUES (return_word_id, lookup_token, null, null, null, null, null, 0, null); + INSERT INTO word VALUES (return_word_id, lookup_token, null, null, null, null, 0); END IF; RETURN return_word_id; END; @@ -198,7 +203,7 @@ BEGIN SELECT min(word_id) FROM word WHERE word_token = lookup_token and class=lookup_class and type = lookup_type and operator = op into return_word_id; IF return_word_id IS NULL THEN return_word_id := nextval('seq_word'); - INSERT INTO word VALUES (return_word_id, lookup_token, null, null, lookup_class, lookup_type, null, 0, op, null); + INSERT INTO word VALUES (return_word_id, lookup_token, null, lookup_class, lookup_type, null, 0, op); END IF; RETURN return_word_id; END; @@ -217,7 +222,7 @@ BEGIN SELECT min(word_id) FROM word WHERE word_token = lookup_token and class is null and type is null into return_word_id; IF return_word_id IS NULL THEN return_word_id := nextval('seq_word'); - INSERT INTO word VALUES (return_word_id, lookup_token, regexp_replace(lookup_token,E'([^0-9])\\1+',E'\\1','g'), src_word, null, null, null, 0, null); + INSERT INTO word VALUES (return_word_id, lookup_token, src_word, null, null, null, 0); -- nospace_lookup_token := replace(replace(lookup_token, '-',''), ' ',''); -- IF ' '||nospace_lookup_token != lookup_token THEN -- INSERT INTO word VALUES (return_word_id, '-'||nospace_lookup_token, null, src_word, null, null, null, 0, null); @@ -290,6 +295,34 @@ END; $$ LANGUAGE plpgsql IMMUTABLE; +CREATE OR REPLACE FUNCTION create_country(src HSTORE, lookup_country_code varchar(2)) RETURNS VOID + AS $$ +DECLARE + s TEXT; + w INTEGER; + words TEXT[]; + item RECORD; + j INTEGER; +BEGIN + FOR item IN SELECT (each(src)).* LOOP + + s := make_standard_name(item.value); + w := getorcreate_country(s, lookup_country_code); + + words := regexp_split_to_array(item.value, E'[,;()]'); + IF array_upper(words, 1) != 1 THEN + FOR j IN 1..array_upper(words, 1) LOOP + s := make_standard_name(words[j]); + IF s != '' THEN + w := getorcreate_country(s, lookup_country_code); + END IF; + END LOOP; + END IF; + END LOOP; +END; +$$ +LANGUAGE plpgsql; + CREATE OR REPLACE FUNCTION make_keywords(src HSTORE) RETURNS INTEGER[] AS $$ DECLARE @@ -312,12 +345,18 @@ BEGIN result := result || w; END IF; + w := getorcreate_word_id(s); + + IF NOT (ARRAY[w] <@ result) THEN + result := result || w; + END IF; + words := string_to_array(s, ' '); IF array_upper(words, 1) IS NOT NULL THEN FOR j IN 1..array_upper(words, 1) LOOP IF (words[j] != '') THEN w = getorcreate_word_id(words[j]); - IF NOT (ARRAY[w] <@ result) THEN + IF w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN result := result || w; END IF; END IF; @@ -330,7 +369,7 @@ BEGIN s := make_standard_name(words[j]); IF s != '' THEN w := getorcreate_word_id(s); - IF NOT (ARRAY[w] <@ result) THEN + IF w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN result := result || w; END IF; END IF; @@ -374,12 +413,18 @@ BEGIN result := result || w; END IF; + w := getorcreate_word_id(s); + + IF NOT (ARRAY[w] <@ result) THEN + result := result || w; + END IF; + words := string_to_array(s, ' '); IF array_upper(words, 1) IS NOT NULL THEN FOR j IN 1..array_upper(words, 1) LOOP IF (words[j] != '') THEN w = getorcreate_word_id(words[j]); - IF NOT (ARRAY[w] <@ result) THEN + IF w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN result := result || w; END IF; END IF; @@ -392,7 +437,7 @@ BEGIN s := make_standard_name(words[j]); IF s != '' THEN w := getorcreate_word_id(s); - IF NOT (ARRAY[w] <@ result) THEN + IF w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN result := result || w; END IF; END IF; @@ -537,6 +582,20 @@ END; $$ LANGUAGE plpgsql IMMUTABLE; +CREATE OR REPLACE FUNCTION get_country_language_codes(search_country_code VARCHAR(2)) RETURNS TEXT[] + AS $$ +DECLARE + nearcountry RECORD; +BEGIN + FOR nearcountry IN select country_default_language_codes from country_name where country_code = search_country_code limit 1 + LOOP + RETURN lower(nearcountry.country_default_language_codes); + END LOOP; + RETURN NULL; +END; +$$ +LANGUAGE plpgsql IMMUTABLE; + CREATE OR REPLACE FUNCTION get_partition(place geometry, in_country_code VARCHAR(10)) RETURNS INTEGER AS $$ DECLARE @@ -590,7 +649,7 @@ BEGIN -- RAISE WARNING 'Adding location with rank > 25 (% rank %)', place_id, rank_search; - x := deleteLocationArea(partition, place_id); + x := deleteLocationArea(partition, place_id, rank_search); isarea := false; IF (ST_GeometryType(geometry) in ('ST_Polygon','ST_MultiPolygon') AND ST_IsValid(geometry)) THEN @@ -657,7 +716,7 @@ CREATE OR REPLACE FUNCTION update_location( DECLARE b BOOLEAN; BEGIN - b := deleteLocationArea(partition, place_id); + b := deleteLocationArea(partition, place_id, rank_search); -- result := add_location(NEW.place_id, NEW.country_code, NEW.partition, name_vector, NEW.rank_search, NEW.rank_address, NEW.geometry); RETURN add_location(place_id, place_country_code, name, rank_search, rank_address, geometry); END; @@ -735,8 +794,8 @@ CREATE OR REPLACE FUNCTION create_interpolation(wayid BIGINT, interpolationtype DECLARE newpoints INTEGER; - waynodes integer[]; - nodeid INTEGER; + waynodes BIGINT[]; + nodeid BIGINT; prevnode RECORD; nextnode RECORD; startnumber INTEGER; @@ -764,11 +823,18 @@ BEGIN FOR nodeidpos in 1..array_upper(waynodes, 1) LOOP - select min(place_id) from placex where osm_type = 'N' and osm_id = waynodes[nodeidpos]::INTEGER and type = 'house' INTO search_place_id; + select min(place_id) from placex where osm_type = 'N' and osm_id = waynodes[nodeidpos]::BIGINT and type = 'house' INTO search_place_id; IF search_place_id IS NULL THEN -- 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 * from placex where osm_type = 'N' and osm_id = waynodes[nodeidpos]::BIGINT 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; @@ -909,9 +975,9 @@ BEGIN default_language := get_country_language_code(NEW.calculated_country_code); IF default_language IS NOT NULL THEN IF NEW.name ? 'name' AND NOT NEW.name ? ('name:'||default_language) THEN - NEW.name := NEW.name || (('name:'||default_language) => (NEW.name -> 'name')); + NEW.name := NEW.name || hstore(('name:'||default_language), (NEW.name -> 'name')); ELSEIF NEW.name ? ('name:'||default_language) AND NOT NEW.name ? 'name' THEN - NEW.name := NEW.name || ('name' => (NEW.name -> ('name:'||default_language))); + NEW.name := NEW.name || hstore('name', (NEW.name -> ('name:'||default_language))); END IF; END IF; END IF; @@ -938,7 +1004,7 @@ BEGIN RETURN NULL; END IF; - NEW.name := 'ref'=>NEW.postcode; + NEW.name := hstore('ref', NEW.postcode); IF NEW.calculated_country_code = 'gb' THEN @@ -1061,6 +1127,9 @@ BEGIN ELSEIF NEW.class = 'landuse' AND ST_GeometryType(NEW.geometry) in ('ST_Polygon','ST_MultiPolygon') THEN NEW.rank_search := 22; NEW.rank_address := NEW.rank_search; + ELSEIF NEW.class = 'natural' and NEW.type in ('peak','volcano','mountain_range') THEN + NEW.rank_search := 18; + NEW.rank_address := 0; -- any feature more than 5 square miles is probably worth indexing ELSEIF ST_GeometryType(NEW.geometry) in ('ST_Polygon','ST_MultiPolygon') AND ST_Area(NEW.geometry) > 0.1 THEN NEW.rank_search := 22; @@ -1086,9 +1155,9 @@ BEGIN NEW.rank_address := NEW.rank_search; ELSEIF NEW.class = 'natural' and NEW.type in ('coastline') THEN RETURN NULL; - ELSEIF NEW.class = 'natural' and NEW.type in ('peak','volcano') THEN - NEW.rank_search := 18; - NEW.rank_address := 0; + ELSEIF NEW.class = 'mountain_pass' THEN + NEW.rank_search := 20; + NEW.rank_address := 0; END IF; END IF; @@ -1154,7 +1223,13 @@ BEGIN END IF; IF diameter > 0 THEN -- RAISE WARNING 'placex point insert: % % % % %',NEW.osm_type,NEW.osm_id,NEW.class,NEW.type,diameter; - 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); + IF NEW.rank_search >= 26 THEN + -- 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); + ELSE + -- for all other places the search terms may change as well + 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); + END IF; END IF; END IF; @@ -1213,6 +1288,8 @@ DECLARE location_rank_search INTEGER; location_distance FLOAT; + location_parent GEOMETRY; + location_isaddress BOOLEAN; tagpairid INTEGER; @@ -1225,7 +1302,14 @@ DECLARE result BOOLEAN; BEGIN - IF NEW.indexed_status != 0 OR OLD.indexed_status = 0 OR NEW.linked_place_id is not null THEN + -- 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 THEN RETURN NEW; END IF; @@ -1239,30 +1323,25 @@ BEGIN 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; - + -- TODO: this test is now redundant? IF OLD.indexed_status != 0 THEN - --DEBUG: RAISE WARNING 'placex_update_0 % %',NEW.osm_type,NEW.osm_id; NEW.indexed_date = now(); - IF NEW.class = 'place' AND NEW.type = 'houses' THEN - i := create_interpolation(NEW.osm_id, NEW.housenumber); + result := deleteSearchName(NEW.partition, NEW.place_id); + DELETE FROM place_addressline WHERE place_id = NEW.place_id; + DELETE FROM place_boundingbox where place_id = NEW.place_id; + result := deleteRoad(NEW.partition, NEW.place_id); + result := deleteLocationArea(NEW.partition, NEW.place_id, NEW.rank_search); + UPDATE placex set linked_place_id = null where linked_place_id = NEW.place_id; + + IF NEW.linked_place_id is not null THEN RETURN NEW; END IF; - IF OLD.indexed_status > 0 THEN - result := deleteSearchName(NEW.partition, NEW.place_id); - DELETE FROM place_addressline WHERE place_id = NEW.place_id; - DELETE FROM place_boundingbox where place_id = NEW.place_id; - result := deleteRoad(NEW.partition, NEW.place_id); - result := deleteLocationArea(NEW.partition, NEW.place_id); - UPDATE placex set linked_place_id = null where linked_place_id = NEW.place_id; + IF NEW.class = 'place' AND NEW.type = 'houses' THEN + i := create_interpolation(NEW.osm_id, NEW.housenumber); + RETURN NEW; END IF; -- Speed up searches - just use the centroid of the feature @@ -1292,9 +1371,9 @@ BEGIN default_language := get_country_language_code(NEW.calculated_country_code); IF default_language IS NOT NULL THEN IF NEW.name ? 'name' AND NOT NEW.name ? ('name:'||default_language) THEN - NEW.name := NEW.name || (('name:'||default_language) => (NEW.name -> 'name')); + NEW.name := NEW.name || hstore(('name:'||default_language), (NEW.name -> 'name')); ELSEIF NEW.name ? ('name:'||default_language) AND NOT NEW.name ? 'name' THEN - NEW.name := NEW.name || ('name' => (NEW.name -> ('name:'||default_language))); + NEW.name := NEW.name || hstore('name', (NEW.name -> ('name:'||default_language))); END IF; END IF; END IF; @@ -1326,6 +1405,9 @@ BEGIN --RAISE WARNING 'finding street for %', NEW; + -- We won't get a better centroid, besides these places are too small to care + NEW.centroid := place_centroid; + NEW.parent_place_id := null; -- to do that we have to find our parent road @@ -1341,7 +1423,7 @@ BEGIN FOR i IN 1..array_upper(relation.members, 1) BY 2 LOOP IF NEW.parent_place_id IS NULL AND relation.members[i+1] = 'street' THEN --RAISE WARNING 'node in relation %',relation; - SELECT place_id from placex where osm_type='W' and osm_id = substring(relation.members[i],2,200)::integer + SELECT place_id from placex where osm_type='W' and osm_id = substring(relation.members[i],2,200)::bigint and rank_search = 26 INTO NEW.parent_place_id; END IF; END LOOP; @@ -1370,7 +1452,7 @@ BEGIN FOR i IN 1..array_upper(relation.members, 1) BY 2 LOOP IF NEW.parent_place_id IS NULL AND relation.members[i+1] = 'street' THEN --RAISE WARNING 'node in way that is in a relation %',relation; - SELECT place_id from placex where osm_type='W' and osm_id = substring(relation.members[i],2,200)::integer + SELECT place_id from placex where osm_type='W' and osm_id = substring(relation.members[i],2,200)::bigint and rank_search = 26 INTO NEW.parent_place_id; END IF; END LOOP; @@ -1415,7 +1497,7 @@ BEGIN FOR i IN 1..array_upper(relation.members, 1) BY 2 LOOP IF NEW.parent_place_id IS NULL AND relation.members[i+1] = 'street' THEN --RAISE WARNING 'way that is in a relation %',relation; - SELECT place_id from placex where osm_type='W' and osm_id = substring(relation.members[i],2,200)::integer + SELECT place_id from placex where osm_type='W' and osm_id = substring(relation.members[i],2,200)::bigint and rank_search = 26 INTO NEW.parent_place_id; END IF; END LOOP; @@ -1469,6 +1551,7 @@ BEGIN -- Merge address from parent nameaddress_vector := array_merge(nameaddress_vector, location.nameaddress_vector); + nameaddress_vector := array_merge(nameaddress_vector, location.name_vector); --return NEW; -- Performance, it would be more acurate to do all the rest of the import process but it takes too long -- Just be happy with inheriting from parent road only @@ -1477,7 +1560,7 @@ BEGIN result := add_location(NEW.place_id, NEW.calculated_country_code, NEW.partition, name_vector, NEW.rank_search, NEW.rank_address, NEW.geometry); END IF; - result := insertSearchName(NEW.partition, NEW.place_id, NEW.calculated_country_code, name_vector, nameaddress_vector, NEW.rank_search, NEW.rank_address, NEW.importance, place_centroid); + result := insertSearchName(NEW.partition, NEW.place_id, NEW.calculated_country_code, name_vector, nameaddress_vector, NEW.rank_search, NEW.rank_address, NEW.importance, place_centroid, NEW.geometry); return NEW; END IF; @@ -1493,62 +1576,83 @@ BEGIN select members from planet_osm_rels where id = NEW.osm_id INTO relation_members; -- RAISE WARNING 'get_osm_rel_members, label'; - FOR relMember IN select get_osm_rel_members(relation_members,ARRAY['label']) as member LOOP + IF relation_members IS NOT NULL THEN + 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 := array_merge(name_vector, make_keywords(linkedPlacex.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; + IF NOT NEW.extratags ? linkedPlacex.class THEN + NEW.extratags := NEW.extratags || hstore(linkedPlacex.class, linkedPlacex.type); + END IF; - END LOOP; + -- mark the linked place (excludes from search results) + UPDATE placex set linked_place_id = NEW.place_id where place_id = linkedPlacex.place_id; - IF NEW.centroid IS NULL THEN + END LOOP; - FOR relMember IN select get_osm_rel_members(relation_members,ARRAY['admin_center','admin_centre']) as member LOOP + END 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; + IF NEW.centroid IS NULL 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 + FOR relMember IN select get_osm_rel_members(relation_members,ARRAY['admin_center','admin_centre']) as member LOOP + 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; + -- 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_address = linkedPlacex.rank_address THEN - -- merge in the name, re-init word vector - NEW.name := linkedPlacex.name || NEW.name; - name_vector := make_keywords(NEW.name); + -- 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 extra tags - NEW.extratags := linkedPlacex.extratags || NEW.extratags; + -- 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; - -- mark the linked place (excludes from search results) - UPDATE placex set linked_place_id = NEW.place_id where place_id = linkedPlacex.place_id; + -- merge in extra tags + IF NOT linkedPlacex.extratags IS NULL THEN + NEW.extratags := linkedPlacex.extratags || NEW.extratags; + 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; + IF NOT NEW.extratags ? linkedPlacex.class THEN + NEW.extratags := NEW.extratags || hstore(linkedPlacex.class, linkedPlacex.type); + END IF; - END LOOP; + -- 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; + + END LOOP; + + END LOOP; + + END IF; END IF; -- not found one yet? how about doing a name search @@ -1556,7 +1660,7 @@ BEGIN FOR linkedPlacex IN select placex.* from placex WHERE make_standard_name(name->'name') = make_standard_name(NEW.name->'name') - AND placex.rank_search = NEW.rank_search + AND placex.rank_address = NEW.rank_address AND placex.place_id != NEW.place_id AND placex.osm_type = 'N'::char(1) AND placex.rank_search < 26 AND st_covers(NEW.geometry, placex.geometry) @@ -1574,6 +1678,10 @@ BEGIN -- merge in extra tags NEW.extratags := linkedPlacex.extratags || NEW.extratags; + IF NOT NEW.extratags ? linkedPlacex.class THEN + NEW.extratags := NEW.extratags || hstore(linkedPlacex.class, linkedPlacex.type); + 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; @@ -1584,6 +1692,19 @@ BEGIN IF NEW.centroid IS NOT NULL THEN place_centroid := NEW.centroid; + -- Place might have had only a name tag before but has now received translations + -- from the linked place. Make sure a name tag for the default language exists in + -- this case. + IF NEW.name is not null AND array_upper(akeys(NEW.name),1) > 1 THEN + default_language := get_country_language_code(NEW.calculated_country_code); + IF default_language IS NOT NULL THEN + IF NEW.name ? 'name' AND NOT NEW.name ? ('name:'||default_language) THEN + NEW.name := NEW.name || hstore(('name:'||default_language), (NEW.name -> 'name')); + ELSEIF NEW.name ? ('name:'||default_language) AND NOT NEW.name ? 'name' THEN + NEW.name := NEW.name || hstore('name', (NEW.name -> ('name:'||default_language))); + END IF; + END IF; + END IF; END IF; -- Did we gain a wikipedia tag in the process? then we need to recalculate our importance @@ -1597,6 +1718,11 @@ BEGIN END IF; + -- make sure all names are in the word table + IF NEW.admin_level = 2 AND NEW.class = 'boundary' AND NEW.type = 'administrative' AND NEW.country_code IS NOT NULL THEN + perform create_country(NEW.name, lower(NEW.country_code)); + END IF; + NEW.parent_place_id = 0; parent_place_id_rank = 0; @@ -1608,8 +1734,15 @@ BEGIN FOR i IN 1..array_upper(isin, 1) LOOP address_street_word_id := get_name_id(make_standard_name(isin[i])); IF address_street_word_id IS NOT NULL AND NOT(ARRAY[address_street_word_id] <@ isin_tokens) THEN + nameaddress_vector := array_merge(nameaddress_vector, ARRAY[address_street_word_id]); isin_tokens := isin_tokens || address_street_word_id; END IF; + + -- merge word into address vector + address_street_word_id := get_word_id(make_standard_name(isin[i])); + IF address_street_word_id IS NOT NULL THEN + nameaddress_vector := array_merge(nameaddress_vector, ARRAY[address_street_word_id]); + END IF; END LOOP; END IF; END IF; @@ -1619,32 +1752,70 @@ BEGIN FOR i IN 1..array_upper(isin, 1) LOOP address_street_word_id := get_name_id(make_standard_name(isin[i])); IF address_street_word_id IS NOT NULL AND NOT(ARRAY[address_street_word_id] <@ isin_tokens) THEN + nameaddress_vector := array_merge(nameaddress_vector, ARRAY[address_street_word_id]); isin_tokens := isin_tokens || address_street_word_id; END IF; + + -- merge into address vector + address_street_word_id := get_word_id(make_standard_name(isin[i])); + IF address_street_word_id IS NOT NULL THEN + nameaddress_vector := array_merge(nameaddress_vector, ARRAY[address_street_word_id]); + END IF; END LOOP; END IF; END IF; + + -- for the USA we have an additional address table. Merge in zip codes from there too + IF NEW.rank_search = 26 AND NEW.calculated_country_code = 'us' THEN + FOR location IN SELECT distinct postcode from location_property_tiger where parent_place_id = NEW.place_id LOOP + address_street_word_id := get_name_id(make_standard_name(location.postcode)); + nameaddress_vector := array_merge(nameaddress_vector, ARRAY[address_street_word_id]); + isin_tokens := isin_tokens || address_street_word_id; + + -- also merge in the single word version + address_street_word_id := get_word_id(make_standard_name(location.postcode)); + nameaddress_vector := array_merge(nameaddress_vector, ARRAY[address_street_word_id]); + END LOOP; + END IF; + -- RAISE WARNING 'ISIN: %', isin_tokens; -- Process area matches - location_rank_search := 100; + location_rank_search := 0; location_distance := 0; --- RAISE WARNING ' getNearFeatures(%,''%'',%,''%'')',NEW.partition, place_centroid, search_maxrank, isin_tokens; - FOR location IN SELECT distinct * from getNearFeatures(NEW.partition, place_centroid, search_maxrank, isin_tokens) LOOP + 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; - 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_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[]); - 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; + IF location.rank_search > 4 THEN + nameaddress_vector := array_merge(nameaddress_vector, location.keywords::integer[]); + END IF; + 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; @@ -1661,21 +1832,25 @@ BEGIN IF array_upper(isin_tokens, 1) IS NOT NULL THEN FOR i IN 1..array_upper(isin_tokens, 1) LOOP --RAISE WARNING ' getNearestNamedFeature: % % % %',NEW.partition, place_centroid, search_maxrank, isin_tokens[i]; + IF NOT ARRAY[isin_tokens[i]] <@ nameaddress_vector THEN - FOR location IN SELECT distinct * from getNearestNamedFeature(NEW.partition, place_centroid, search_maxrank, isin_tokens[i]) LOOP + FOR location IN SELECT * from getNearestNamedFeature(NEW.partition, place_centroid, search_maxrank, isin_tokens[i]) LOOP ---RAISE WARNING ' ISIN: %',location; + --RAISE WARNING ' ISIN: %',location; - nameaddress_vector := array_merge(nameaddress_vector, location.keywords::integer[]); - INSERT INTO place_addressline VALUES (NEW.place_id, location.place_id, false, NOT address_havelevel[location.rank_address], location.distance, location.rank_address); - address_havelevel[location.rank_address] := true; + IF location.rank_search > 4 THEN + nameaddress_vector := array_merge(nameaddress_vector, location.keywords::integer[]); + INSERT INTO place_addressline VALUES (NEW.place_id, location.place_id, false, NOT address_havelevel[location.rank_address], location.distance, location.rank_address); + address_havelevel[location.rank_address] := true; - IF location.rank_address > parent_place_id_rank THEN - NEW.parent_place_id = location.place_id; - parent_place_id_rank = location.rank_address; - END IF; + IF location.rank_address > parent_place_id_rank THEN + NEW.parent_place_id = location.place_id; + parent_place_id_rank = location.rank_address; + END IF; + END IF; + END LOOP; - END LOOP; + END IF; END LOOP; END IF; @@ -1683,17 +1858,17 @@ BEGIN -- 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 - 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 THEN + IF location.rank_search > 4 AND location.distance < location_distance THEN -- Add it to the list of search terms nameaddress_vector := array_merge(nameaddress_vector, location.keywords::integer[]); @@ -1716,9 +1891,8 @@ BEGIN result := insertLocationRoad(NEW.partition, NEW.place_id, NEW.calculated_country_code, NEW.geometry); END IF; - result := insertSearchName(NEW.partition, NEW.place_id, NEW.calculated_country_code, name_vector, nameaddress_vector, NEW.rank_search, NEW.rank_address, NEW.importance, place_centroid); + result := insertSearchName(NEW.partition, NEW.place_id, NEW.calculated_country_code, name_vector, nameaddress_vector, NEW.rank_search, NEW.rank_address, NEW.importance, place_centroid, NEW.geometry); --- INSERT INTO search_name values (NEW.place_id, NEW.rank_search, NEW.rank_search, 0, NEW.calculated_country_code, name_vector, nameaddress_vector, place_centroid); END IF; -- If we've not managed to pick up a better one - default centroid @@ -1739,11 +1913,11 @@ DECLARE b BOOLEAN; classtable TEXT; BEGIN - RAISE WARNING 'placex_delete % %',OLD.osm_type,OLD.osm_id; + -- 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; - 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 @@ -1768,7 +1942,7 @@ BEGIN --DEBUG: RAISE WARNING 'placex_delete:08 % %',OLD.osm_type,OLD.osm_id; IF OLD.rank_address < 26 THEN - b := deleteLocationArea(OLD.partition, OLD.place_id); + b := deleteLocationArea(OLD.partition, OLD.place_id, OLD.rank_search); END IF; --DEBUG: RAISE WARNING 'placex_delete:09 % %',OLD.osm_type,OLD.osm_id; @@ -1806,8 +1980,8 @@ BEGIN --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; @@ -2074,7 +2248,9 @@ BEGIN END IF; END LOOP; - RETURN null; + -- anything will do as a fallback - just take the first name type thing there is + search := avals(name); + RETURN search[1]; END; $$ LANGUAGE plpgsql IMMUTABLE; @@ -2156,7 +2332,7 @@ 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)); - 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; @@ -2167,11 +2343,11 @@ END; $$ LANGUAGE plpgsql; -DROP TYPE addressline CASCADE; +DROP TYPE IF EXISTS addressline CASCADE; create type addressline as ( place_id BIGINT, osm_type CHAR(1), - osm_id INTEGER, + osm_id BIGINT, name HSTORE, class TEXT, type TEXT, @@ -2213,14 +2389,14 @@ BEGIN 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; - 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; @@ -2230,7 +2406,7 @@ BEGIN hadcountry := false; FOR location IN select placex.place_id, osm_type, osm_id, - CASE WHEN class = 'place' and type = 'postcode' THEN 'name' => postcode ELSE name END as name, + CASE WHEN class = 'place' and type = 'postcode' THEN hstore('name', postcode) ELSE name END as name, class, type, admin_level, true as fromarea, true as isaddress, CASE WHEN rank_address = 0 THEN 100 WHEN rank_address = 11 THEN 5 ELSE rank_address END as rank_address, 0 as distance, calculated_country_code @@ -2263,7 +2439,7 @@ BEGIN FOR location IN select placex.place_id, osm_type, osm_id, - CASE WHEN class = 'place' and type = 'postcode' THEN 'name' => postcode ELSE name END as name, + CASE WHEN class = 'place' and type = 'postcode' THEN hstore('name', postcode) ELSE name END as name, class, type, admin_level, fromarea, isaddress, CASE WHEN address_place_id = for_place_id AND rank_address = 0 THEN 100 WHEN rank_address = 11 THEN 5 ELSE rank_address END as rank_address, distance,calculated_country_code @@ -2308,7 +2484,7 @@ BEGIN END IF; IF searchcountrycode IS NOT NULL THEN - location := ROW(null, null, null, 'ref'=>searchcountrycode, 'place', 'country_code', null, true, false, 4, 0)::addressline; + location := ROW(null, null, null, hstore('ref', searchcountrycode), 'place', 'country_code', null, true, false, 4, 0)::addressline; RETURN NEXT location; END IF; @@ -2319,12 +2495,12 @@ BEGIN END IF; IF searchhousenumber IS NOT NULL THEN - location := ROW(in_place_id, null, null, 'ref'=>searchhousenumber, 'place', 'house_number', null, true, true, 28, 0)::addressline; + location := ROW(in_place_id, null, null, hstore('ref', searchhousenumber), 'place', 'house_number', null, true, true, 28, 0)::addressline; RETURN NEXT location; END IF; IF searchpostcode IS NOT NULL THEN - location := ROW(null, null, null, 'ref'=>searchpostcode, 'place', 'postcode', null, true, true, 5, 0)::addressline; + location := ROW(null, null, null, hstore('ref', searchpostcode), 'place', 'postcode', null, true, true, 5, 0)::addressline; RETURN NEXT location; END IF; @@ -2560,93 +2736,6 @@ END; $$ 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 $$ @@ -2907,3 +2996,70 @@ BEGIN END; $$ LANGUAGE plpgsql; + + +CREATE OR REPLACE FUNCTION place_force_delete(placeid BIGINT) RETURNS BOOLEAN + AS $$ +DECLARE + osmid BIGINT; + osmtype character(1); + pclass text; + ptype text; +BEGIN + SELECT osm_type, osm_id, class, type FROM placex WHERE place_id = placeid INTO osmtype, osmid, pclass, ptype; + DELETE FROM import_polygon_delete where osm_type = osmtype and osm_id = osmid and class = pclass and type = ptype; + DELETE FROM import_polygon_error where osm_type = osmtype and osm_id = osmid and class = pclass and type = ptype; + -- force delete from place/placex by making it a very small geometry + UPDATE place set geometry = ST_SetSRID(ST_Point(0,0), 4326) where osm_type = osmtype and osm_id = osmid and class = pclass and type = ptype; + DELETE FROM place where osm_type = osmtype and osm_id = osmid and class = pclass and type = ptype; + + RETURN TRUE; +END; +$$ +LANGUAGE plpgsql; + +CREATE OR REPLACE FUNCTION place_force_update(placeid BIGINT) RETURNS BOOLEAN + AS $$ +DECLARE + placegeom GEOMETRY; + geom GEOMETRY; + diameter FLOAT; + rank INTEGER; +BEGIN + SELECT geometry, rank_search FROM placex WHERE place_id = placeid INTO placegeom, rank; + IF placegeom IS NOT NULL AND ST_IsValid(placegeom) THEN + IF ST_GeometryType(placegeom) in ('ST_Polygon','ST_MultiPolygon') THEN + FOR geom IN select split_geometry(placegeom) FROM placex WHERE place_id = placeid LOOP + update placex set indexed_status = 2 where (st_covers(geom, placex.geometry) OR ST_Intersects(geom, placex.geometry)) + AND rank_search > rank and indexed_status = 0 and ST_geometrytype(placex.geometry) = 'ST_Point' and (rank_search < 28 or name is not null); + update placex set indexed_status = 2 where (st_covers(geom, placex.geometry) OR ST_Intersects(geom, placex.geometry)) + AND rank_search > rank and indexed_status = 0 and ST_geometrytype(placex.geometry) != 'ST_Point' and (rank_search < 28 or name is not null); + END LOOP; + ELSE + diameter := 0; + IF rank = 11 THEN + diameter := 0.05; + ELSEIF rank < 18 THEN + diameter := 0.1; + ELSEIF rank < 20 THEN + diameter := 0.05; + ELSEIF rank = 21 THEN + diameter := 0.001; + ELSEIF rank < 24 THEN + diameter := 0.02; + ELSEIF rank < 26 THEN + diameter := 0.002; -- 100 to 200 meters + ELSEIF rank < 28 THEN + diameter := 0.001; -- 50 to 100 meters + END IF; + IF diameter > 0 THEN + update placex set indexed_status = 2 where indexed_status = 0 and rank_search > rank and ST_DWithin(placex.geometry, placegeom, diameter) and (rank_search < 28 or name is not null); + END IF; + END IF; + RETURN TRUE; + END IF; + + RETURN FALSE; +END; +$$ +LANGUAGE plpgsql;