only consider partials in multi-words for initial count

[nominatim.git] / nominatim / tokenizer / legacy_icu_tokenizer.py

diff --git a/nominatim/tokenizer/legacy_icu_tokenizer.py b/nominatim/tokenizer/legacy_icu_tokenizer.py
index 5f83b73dc2c7c7fb6e7c444404549abf11c863cb..6bf409cca3ab3674b41605b06e8dfe49eda40e41 100644 (file)
--- a/nominatim/tokenizer/legacy_icu_tokenizer.py
+++ b/nominatim/tokenizer/legacy_icu_tokenizer.py
@@ -168,7 +168,8 @@ class LegacyICUTokenizer:
                 for name, cnt in cur:
                     terms = set()
                     for word in name_proc.get_variants_ascii(name_proc.get_normalized(name)):
-                        terms.update(word.split())
+                        if ' ' in word:
+                            terms.update(word.split())
                     for term in terms:
                         words[term] += cnt