X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/6a7e0d652b1d40a397e1c1386d500101796676c4..b4fe7d7c7df335ce62eb8ec626428b45b38235a6:/lib-sql/functions/normalization.sql?ds=sidebyside

diff --git a/lib-sql/functions/normalization.sql b/lib-sql/functions/normalization.sql
index 8bb4915b..c7bd2fc5 100644
--- a/lib-sql/functions/normalization.sql
+++ b/lib-sql/functions/normalization.sql
@@ -1,12 +1,12 @@
 -- Functions for term normalisation and access to the 'word' table.
 
 CREATE OR REPLACE FUNCTION transliteration(text) RETURNS text
-  AS '{modulepath}/nominatim.so', 'transliteration'
+  AS '{{ modulepath }}/nominatim.so', 'transliteration'
 LANGUAGE c IMMUTABLE STRICT;
 
 
 CREATE OR REPLACE FUNCTION gettokenstring(text) RETURNS text
-  AS '{modulepath}/nominatim.so', 'gettokenstring'
+  AS '{{ modulepath }}/nominatim.so', 'gettokenstring'
 LANGUAGE c IMMUTABLE STRICT;
 
 
@@ -47,6 +47,25 @@ END;
 $$
 LANGUAGE plpgsql;
 
+-- Create housenumber tokens from an OSM addr:housenumber.
+-- The housnumber is split at comma and semicolon as necessary.
+-- The function returns the normalized form of the housenumber suitable
+-- for comparison.
+CREATE OR REPLACE FUNCTION create_housenumber_id(housenumber TEXT)
+  RETURNS TEXT
+  AS $$
+DECLARE
+  normtext TEXT;
+BEGIN
+  SELECT array_to_string(array_agg(trans), ';')
+    INTO normtext
+    FROM (SELECT lookup_word as trans, getorcreate_housenumber_id(lookup_word)
+          FROM (SELECT make_standard_name(h) as lookup_word
+                FROM regexp_split_to_table(housenumber, '[,;]') h) x) y;
+
+  return normtext;
+END;
+$$ LANGUAGE plpgsql STABLE STRICT;
 
 CREATE OR REPLACE FUNCTION getorcreate_housenumber_id(lookup_word TEXT)
   RETURNS INTEGER
@@ -358,40 +377,26 @@ $$
 LANGUAGE plpgsql;
 
 
-CREATE OR REPLACE FUNCTION make_keywords(src TEXT)
-  RETURNS INTEGER[]
+CREATE OR REPLACE FUNCTION precompute_words(src TEXT)
+  RETURNS INTEGER
   AS $$
 DECLARE
-  result INTEGER[];
   s TEXT;
   w INTEGER;
   words TEXT[];
   i INTEGER;
   j INTEGER;
 BEGIN
-  result := '{}'::INTEGER[];
-
   s := make_standard_name(src);
   w := getorcreate_name_id(s, src);
 
-  IF NOT (ARRAY[w] <@ result) THEN
-    result := result || w;
-  END IF;
-
   w := getorcreate_word_id(s);
 
-  IF w IS NOT NULL AND 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 w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN
-          result := result || w;
-        END IF;
+        w := getorcreate_word_id(words[j]);
       END IF;
     END LOOP;
   END IF;
@@ -402,9 +407,6 @@ BEGIN
       s := make_standard_name(words[j]);
       IF s != '' THEN
         w := getorcreate_word_id(s);
-        IF w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN
-          result := result || w;
-        END IF;
       END IF;
     END LOOP;
   END IF;
@@ -414,13 +416,10 @@ BEGIN
     s := make_standard_name(s);
     IF s != '' THEN
       w := getorcreate_name_id(s, src);
-      IF NOT (ARRAY[w] <@ result) THEN
-        result := result || w;
-      END IF;
     END IF;
   END IF;
 
-  RETURN result;
+  RETURN 1;
 END;
 $$
 LANGUAGE plpgsql;