ICU: add an index over word_ids

[nominatim.git] / lib-sql / tokenizer / legacy_tokenizer.sql

diff --git a/lib-sql/tokenizer/legacy_tokenizer.sql b/lib-sql/tokenizer/legacy_tokenizer.sql
index fe82762e47d53867281e0ec68d5a4e9b74c6a046..2b734e6f2a95a5cfe97243e81e3b4c47485a3d92 100644 (file)
--- a/lib-sql/tokenizer/legacy_tokenizer.sql
+++ b/lib-sql/tokenizer/legacy_tokenizer.sql
@@ -34,17 +34,31 @@ AS $$
 $$ LANGUAGE SQL IMMUTABLE STRICT;
 
 
-CREATE OR REPLACE FUNCTION token_addr_street_match_tokens(info JSONB)
-  RETURNS INTEGER[]
+CREATE OR REPLACE FUNCTION token_has_addr_street(info JSONB)
+  RETURNS BOOLEAN
 AS $$
-  SELECT (info->>'street')::INTEGER[]
+  SELECT info->>'street' is not null;
+$$ LANGUAGE SQL IMMUTABLE;
+
+
+CREATE OR REPLACE FUNCTION token_has_addr_place(info JSONB)
+  RETURNS BOOLEAN
+AS $$
+  SELECT info->>'place_match' is not null;
+$$ LANGUAGE SQL IMMUTABLE;
+
+
+CREATE OR REPLACE FUNCTION token_matches_street(info JSONB, street_tokens INTEGER[])
+  RETURNS BOOLEAN
+AS $$
+  SELECT (info->>'street')::INTEGER[] && street_tokens
 $$ LANGUAGE SQL IMMUTABLE STRICT;
 
 
-CREATE OR REPLACE FUNCTION token_addr_place_match_tokens(info JSONB)
-  RETURNS INTEGER[]
+CREATE OR REPLACE FUNCTION token_matches_place(info JSONB, place_tokens INTEGER[])
+  RETURNS BOOLEAN
 AS $$
-  SELECT (info->>'place_match')::INTEGER[]
+  SELECT (info->>'place_match')::INTEGER[] && place_tokens
 $$ LANGUAGE SQL IMMUTABLE STRICT;
 
 
@@ -55,19 +69,24 @@ AS $$
 $$ LANGUAGE SQL IMMUTABLE STRICT;
 
 
-DROP TYPE IF EXISTS token_addresstoken CASCADE;
-CREATE TYPE token_addresstoken AS (
-  key TEXT,
-  match_tokens INT[],
-  search_tokens INT[]
-);
+CREATE OR REPLACE FUNCTION token_get_address_keys(info JSONB)
+  RETURNS SETOF TEXT
+AS $$
+  SELECT * FROM jsonb_object_keys(info->'addr');
+$$ LANGUAGE SQL IMMUTABLE STRICT;
+
+
+CREATE OR REPLACE FUNCTION token_get_address_search_tokens(info JSONB, key TEXT)
+  RETURNS INTEGER[]
+AS $$
+  SELECT (info->'addr'->key->>0)::INTEGER[];
+$$ LANGUAGE SQL IMMUTABLE STRICT;
+
 
-CREATE OR REPLACE FUNCTION token_get_address_tokens(info JSONB)
-  RETURNS SETOF token_addresstoken
+CREATE OR REPLACE FUNCTION token_matches_address(info JSONB, key TEXT, tokens INTEGER[])
+  RETURNS BOOLEAN
 AS $$
-  SELECT key, (value->>1)::int[] as match_tokens,
-         (value->>0)::int[] as search_tokens
-  FROM jsonb_each(info->'addr');
+  SELECT (info->'addr'->key->>1)::INTEGER[] && tokens;
 $$ LANGUAGE SQL IMMUTABLE STRICT;
 
 
@@ -287,26 +306,21 @@ DECLARE
   s TEXT;
   w INTEGER;
   words TEXT[];
-  item RECORD;
+  value TEXT;
   j INTEGER;
 BEGIN
   result := '{}'::INTEGER[];
 
-  FOR item IN SELECT (each(src)).* LOOP
-
-    s := make_standard_name(item.value);
-    w := getorcreate_name_id(s, item.value);
+  FOR value IN SELECT unnest(regexp_split_to_array(svals(src), E'[,;]')) LOOP
+    -- full name
+    s := make_standard_name(value);
+    w := getorcreate_name_id(s, value);
 
     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;
-
+    -- partial single-word terms
     words := string_to_array(s, ' ');
     IF array_upper(words, 1) IS NOT NULL THEN
       FOR j IN 1..array_upper(words, 1) LOOP
@@ -319,24 +333,23 @@ BEGIN
       END LOOP;
     END IF;
 
-    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_word_id(s);
-          IF w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN
-            result := result || w;
-          END IF;
+    -- consider parts before an opening braket a full word as well
+    words := regexp_split_to_array(value, E'[(]');
+    IF array_upper(words, 1) > 1 THEN
+      s := make_standard_name(words[1]);
+      IF s != '' THEN
+        w := getorcreate_name_id(s, words[1]);
+        IF w IS NOT NULL AND NOT (ARRAY[w] <@ result) THEN
+          result := result || w;
         END IF;
-      END LOOP;
+      END IF;
     END IF;
 
-    s := regexp_replace(item.value, '市$', '');
-    IF s != item.value THEN
+    s := regexp_replace(value, '市$', '');
+    IF s != value THEN
       s := make_standard_name(s);
       IF s != '' THEN
-        w := getorcreate_name_id(s, item.value);
+        w := getorcreate_name_id(s, value);
         IF NOT (ARRAY[w] <@ result) THEN
           result := result || w;
         END IF;