release 5.1.0.post1

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

diff --git a/lib-sql/tokenizer/icu_tokenizer.sql b/lib-sql/tokenizer/icu_tokenizer.sql
index ffe6648c38e959c6279efb2d1898d835514f32a7..8cf13120a11515db4fbe432c131332fdea054a45 100644 (file)
--- a/lib-sql/tokenizer/icu_tokenizer.sql
+++ b/lib-sql/tokenizer/icu_tokenizer.sql
@@ -1,3 +1,10 @@
+-- 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.
+

 -- Get tokens used for searching the given place.
 --
 -- These are the tokens that will be saved in the search_name table.
 -- Get tokens used for searching the given place.
 --
 -- These are the tokens that will be saved in the search_name table.


@@ -5,7 +12,7 @@ CREATE OR REPLACE FUNCTION token_get_name_search_tokens(info JSONB)
   RETURNS INTEGER[]
 AS $$
   SELECT (info->>'names')::INTEGER[]
   RETURNS INTEGER[]
 AS $$
   SELECT (info->>'names')::INTEGER[]

-$$ LANGUAGE SQL IMMUTABLE STRICT;
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 
 -- Get tokens for matching the place name against others.
 
 
 -- Get tokens for matching the place name against others.


@@ -15,7 +22,7 @@ CREATE OR REPLACE FUNCTION token_get_name_match_tokens(info JSONB)
   RETURNS INTEGER[]
 AS $$
   SELECT (info->>'names')::INTEGER[]
   RETURNS INTEGER[]
 AS $$
   SELECT (info->>'names')::INTEGER[]

-$$ LANGUAGE SQL IMMUTABLE STRICT;
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 
 -- Return the housenumber tokens applicable for the place.
 
 
 -- Return the housenumber tokens applicable for the place.


@@ -23,7 +30,7 @@ CREATE OR REPLACE FUNCTION token_get_housenumber_search_tokens(info JSONB)
   RETURNS INTEGER[]
 AS $$
   SELECT (info->>'hnr_tokens')::INTEGER[]
   RETURNS INTEGER[]
 AS $$
   SELECT (info->>'hnr_tokens')::INTEGER[]

-$$ LANGUAGE SQL IMMUTABLE STRICT;
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 
 -- Return the housenumber in the form that it can be matched during search.
 
 
 -- Return the housenumber in the form that it can be matched during search.


@@ -31,51 +38,77 @@ CREATE OR REPLACE FUNCTION token_normalized_housenumber(info JSONB)
   RETURNS TEXT
 AS $$
   SELECT info->>'hnr';
   RETURNS TEXT
 AS $$
   SELECT info->>'hnr';

-$$ LANGUAGE SQL IMMUTABLE STRICT;
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 
 
 

-CREATE OR REPLACE FUNCTION token_addr_street_match_tokens(info JSONB)
-  RETURNS INTEGER[]
+CREATE OR REPLACE FUNCTION token_is_street_address(info JSONB)
+  RETURNS BOOLEAN

 AS $$
 AS $$

-  SELECT (info->>'street')::INTEGER[]
-$$ LANGUAGE SQL IMMUTABLE STRICT;
+  SELECT info->>'street' is not null or info->>'place' is null;
+$$ LANGUAGE SQL IMMUTABLE PARALLEL SAFE;

 
 
 
 

-CREATE OR REPLACE FUNCTION token_addr_place_match_tokens(info JSONB)
-  RETURNS INTEGER[]
+CREATE OR REPLACE FUNCTION token_has_addr_street(info JSONB)
+  RETURNS BOOLEAN

 AS $$
 AS $$

-  SELECT (info->>'place_match')::INTEGER[]
-$$ LANGUAGE SQL IMMUTABLE STRICT;
+  SELECT info->>'street' is not null and info->>'street' != '{}';
+$$ LANGUAGE SQL IMMUTABLE PARALLEL SAFE;
+
+
+CREATE OR REPLACE FUNCTION token_has_addr_place(info JSONB)
+  RETURNS BOOLEAN
+AS $$
+  SELECT info->>'place' is not null;
+$$ LANGUAGE SQL IMMUTABLE PARALLEL SAFE;
+
+
+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 PARALLEL SAFE;
+
+
+CREATE OR REPLACE FUNCTION token_matches_place(info JSONB, place_tokens INTEGER[])
+  RETURNS BOOLEAN
+AS $$
+  SELECT (info->>'place')::INTEGER[] <@ place_tokens
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 
 CREATE OR REPLACE FUNCTION token_addr_place_search_tokens(info JSONB)
   RETURNS INTEGER[]
 AS $$
 
 
 CREATE OR REPLACE FUNCTION token_addr_place_search_tokens(info JSONB)
   RETURNS INTEGER[]
 AS $$

-  SELECT (info->>'place_search')::INTEGER[]
-$$ LANGUAGE SQL IMMUTABLE STRICT;
+  SELECT (info->>'place')::INTEGER[]
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;
+
+
+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 PARALLEL SAFE;
+

 
 

+CREATE OR REPLACE FUNCTION token_get_address_search_tokens(info JSONB, key TEXT)
+  RETURNS INTEGER[]
+AS $$
+  SELECT (info->'addr'->>key)::INTEGER[];
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 

-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_tokens(info JSONB)
-  RETURNS SETOF token_addresstoken
+CREATE OR REPLACE FUNCTION token_matches_address(info JSONB, key TEXT, tokens INTEGER[])
+  RETURNS BOOLEAN

 AS $$
 AS $$

-  SELECT key, (value->>1)::int[] as match_tokens,
-         (value->>0)::int[] as search_tokens
-  FROM jsonb_each(info->'addr');
-$$ LANGUAGE SQL IMMUTABLE STRICT;
+  SELECT (info->'addr'->>key)::INTEGER[] <@ tokens;
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 
 
 

-CREATE OR REPLACE FUNCTION token_normalized_postcode(postcode TEXT)
+CREATE OR REPLACE FUNCTION token_get_postcode(info JSONB)

   RETURNS TEXT
 AS $$
   RETURNS TEXT
 AS $$

-  SELECT CASE WHEN postcode SIMILAR TO '%(,|;)%' THEN NULL ELSE upper(trim(postcode))END;
-$$ LANGUAGE SQL IMMUTABLE STRICT;
+  SELECT info->>'postcode';
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 
 -- Return token info that should be saved permanently in the database.
 
 
 -- Return token info that should be saved permanently in the database.


@@ -83,7 +116,7 @@ CREATE OR REPLACE FUNCTION token_strip_info(info JSONB)
   RETURNS JSONB
 AS $$
   SELECT NULL::JSONB;
   RETURNS JSONB
 AS $$
   SELECT NULL::JSONB;

-$$ LANGUAGE SQL IMMUTABLE STRICT;
+$$ LANGUAGE SQL IMMUTABLE STRICT PARALLEL SAFE;

 
 --------------- private functions ----------------------------------------------
 
 
 --------------- private functions ----------------------------------------------
 


@@ -95,16 +128,14 @@ DECLARE
   partial_terms TEXT[] = '{}'::TEXT[];
   term TEXT;
   term_id INTEGER;
   partial_terms TEXT[] = '{}'::TEXT[];
   term TEXT;
   term_id INTEGER;

-  term_count INTEGER;

 BEGIN
   SELECT min(word_id) INTO full_token
     FROM word WHERE word = norm_term and type = 'W';
 
   IF full_token IS NULL THEN
     full_token := nextval('seq_word');
 BEGIN
   SELECT min(word_id) INTO full_token
     FROM word WHERE word = norm_term and type = 'W';
 
   IF full_token IS NULL THEN
     full_token := nextval('seq_word');

-    INSERT INTO word (word_id, word_token, type, word, info)
-      SELECT full_token, lookup_term, 'W', norm_term,
-             json_build_object('count', 0)
+    INSERT INTO word (word_id, word_token, type, word)
+      SELECT full_token, lookup_term, 'W', norm_term

         FROM unnest(lookup_terms) as lookup_term;
   END IF;
 
         FROM unnest(lookup_terms) as lookup_term;
   END IF;
 


@@ -117,20 +148,92 @@ BEGIN
 
   partial_tokens := '{}'::INT[];
   FOR term IN SELECT unnest(partial_terms) LOOP
 
   partial_tokens := '{}'::INT[];
   FOR term IN SELECT unnest(partial_terms) LOOP

-    SELECT min(word_id), max(info->>'count') INTO term_id, term_count
+    SELECT min(word_id) INTO term_id

       FROM word WHERE word_token = term and type = 'w';
 
     IF term_id IS NULL THEN
       term_id := nextval('seq_word');
       FROM word WHERE word_token = term and type = 'w';
 
     IF term_id IS NULL THEN
       term_id := nextval('seq_word');

-      term_count := 0;
-      INSERT INTO word (word_id, word_token, type, info)
-        VALUES (term_id, term, 'w', json_build_object('count', term_count));
+      INSERT INTO word (word_id, word_token, type)
+        VALUES (term_id, term, 'w');
+    END IF;
+
+    partial_tokens := array_merge(partial_tokens, ARRAY[term_id]);
+  END LOOP;
+END;
+$$
+LANGUAGE plpgsql;
+
+
+CREATE OR REPLACE FUNCTION getorcreate_full_word(norm_term TEXT,
+                                                 lookup_terms TEXT[],
+                                                 lookup_norm_terms TEXT[],
+                                                 OUT full_token INT,
+                                                 OUT partial_tokens INT[])
+  AS $$
+DECLARE
+  partial_terms TEXT[] = '{}'::TEXT[];
+  term TEXT;
+  term_id INTEGER;
+BEGIN
+  SELECT min(word_id) INTO full_token
+    FROM word WHERE word = norm_term and type = 'W';
+
+  IF full_token IS NULL THEN
+    full_token := nextval('seq_word');
+    IF lookup_norm_terms IS NULL THEN
+      INSERT INTO word (word_id, word_token, type, word)
+        SELECT full_token, lookup_term, 'W', norm_term
+          FROM unnest(lookup_terms) as lookup_term;
+    ELSE
+      INSERT INTO word (word_id, word_token, type, word, info)
+        SELECT full_token, t.lookup, 'W', norm_term,
+               CASE WHEN norm_term = t.norm THEN null
+               ELSE json_build_object('lookup', t.norm) END
+          FROM unnest(lookup_terms, lookup_norm_terms) as t(lookup, norm);

     END IF;
     END IF;

+  END IF;

 
 

-    IF term_count < {{ max_word_freq }} THEN
-      partial_tokens := array_merge(partial_tokens, ARRAY[term_id]);
+  FOR term IN SELECT unnest(string_to_array(unnest(lookup_terms), ' ')) LOOP
+    term := trim(term);
+    IF NOT (ARRAY[term] <@ partial_terms) THEN
+      partial_terms := partial_terms || term;

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

+
+  partial_tokens := '{}'::INT[];
+  FOR term IN SELECT unnest(partial_terms) LOOP
+    SELECT min(word_id) INTO term_id
+      FROM word WHERE word_token = term and type = 'w';
+
+    IF term_id IS NULL THEN
+      term_id := nextval('seq_word');
+      INSERT INTO word (word_id, word_token, type)
+        VALUES (term_id, term, 'w');
+    END IF;
+
+    partial_tokens := array_merge(partial_tokens, ARRAY[term_id]);
+  END LOOP;
+END;
+$$
+LANGUAGE plpgsql;
+
+
+CREATE OR REPLACE FUNCTION getorcreate_partial_word(partial TEXT)
+  RETURNS INTEGER
+  AS $$
+DECLARE
+  token INTEGER;
+BEGIN
+  SELECT min(word_id) INTO token
+    FROM word WHERE word_token = partial and type = 'w';
+
+  IF token IS NULL THEN
+    token := nextval('seq_word');
+    INSERT INTO word (word_id, word_token, type, info)
+        VALUES (token, partial, 'w', json_build_object('count', 0));
+  END IF;
+
+  RETURN token;

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


@@ -155,3 +258,49 @@ BEGIN
 END;
 $$
 LANGUAGE plpgsql;
 END;
 $$
 LANGUAGE plpgsql;

+
+
+CREATE OR REPLACE FUNCTION create_analyzed_hnr_id(norm_term TEXT, lookup_terms TEXT[])
+  RETURNS INTEGER
+  AS $$
+DECLARE
+  return_id INTEGER;
+BEGIN
+  SELECT min(word_id) INTO return_id
+    FROM word WHERE word = norm_term and type = 'H';
+
+  IF return_id IS NULL THEN
+    return_id := nextval('seq_word');
+    INSERT INTO word (word_id, word_token, type, word, info)
+      SELECT return_id, lookup_term, 'H', norm_term,
+             json_build_object('lookup', lookup_terms[1])
+        FROM unnest(lookup_terms) as lookup_term;
+  END IF;
+
+  RETURN return_id;
+END;
+$$
+LANGUAGE plpgsql;
+
+CREATE OR REPLACE FUNCTION create_postcode_word(postcode TEXT, lookup_terms TEXT[])
+  RETURNS BOOLEAN
+  AS $$
+DECLARE
+  existing INTEGER;
+BEGIN
+  SELECT count(*) INTO existing
+    FROM word WHERE word = postcode and type = 'P';
+
+  IF existing > 0 THEN
+    RETURN TRUE;
+  END IF;
+
+  -- postcodes don't need word ids
+  INSERT INTO word (word_token, type, word)
+    SELECT lookup_term, 'P', postcode FROM unnest(lookup_terms) as lookup_term;
+
+  RETURN FALSE;
+END;
+$$
+LANGUAGE plpgsql;
+