From: Sarah Hoffmann <lonvia@denofr.de>
Date: Thu, 20 Mar 2025 20:33:27 +0000 (+0100)
Subject: do not write any word counts on initial word insert
X-Git-Url: https://git.openstreetmap.org./nominatim.git/commitdiff_plain/58ef032a2b4d0fafe4c768a1ea88ebfdf0dcdd73?hp=--cc

do not write any word counts on initial word insert
---

58ef032a2b4d0fafe4c768a1ea88ebfdf0dcdd73
diff --git a/lib-sql/tokenizer/icu_tokenizer.sql b/lib-sql/tokenizer/icu_tokenizer.sql
index f0c30f1b..29039987 100644
--- a/lib-sql/tokenizer/icu_tokenizer.sql
+++ b/lib-sql/tokenizer/icu_tokenizer.sql
@@ -128,16 +128,14 @@ DECLARE
   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');
-    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;
 
@@ -150,14 +148,13 @@ BEGIN
 
   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');
-      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]);