X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/b9fbfeff67b420905a4176f4f5e9312746d0c42e..bffbe68ec3ad93aa6188083441677ab0fdd54659:/nominatim/tokenizer/legacy_icu_tokenizer.py

diff --git a/nominatim/tokenizer/legacy_icu_tokenizer.py b/nominatim/tokenizer/legacy_icu_tokenizer.py
index 6bf409cc..c585c5af 100644
--- a/nominatim/tokenizer/legacy_icu_tokenizer.py
+++ b/nominatim/tokenizer/legacy_icu_tokenizer.py
@@ -163,7 +163,9 @@ class LegacyICUTokenizer:
             words = Counter()
             name_proc = ICUNameProcessor(self.naming_rules)
             with conn.cursor(name="words") as cur:
-                cur.execute("SELECT svals(name) as v, count(*) FROM place GROUP BY v")
+                cur.execute(""" SELECT v, count(*) FROM
+                                  (SELECT svals(name) as v FROM place)x
+                                WHERE length(v) < 75 GROUP BY v""")
 
                 for name, cnt in cur:
                     terms = set()