From: Sarah Hoffmann <lonvia@denofr.de>
Date: Wed, 19 Feb 2025 19:51:20 +0000 (+0100)
Subject: Merge remote-tracking branch 'upstream/master'
X-Git-Url: https://git.openstreetmap.org./nominatim.git/commitdiff_plain/19dfe0fdd45e0541b7171a96c16ecc9a84dd4c36?ds=inline;hp=-c

Merge remote-tracking branch 'upstream/master'
---

19dfe0fdd45e0541b7171a96c16ecc9a84dd4c36
diff --combined src/nominatim_api/search/icu_tokenizer.py
index 8f2069c1,487dd171..44afce9f
--- a/src/nominatim_api/search/icu_tokenizer.py
+++ b/src/nominatim_api/search/icu_tokenizer.py
@@@ -208,12 -208,6 +208,12 @@@ class ICUQueryAnalyzer(AbstractQueryAna
          log().section('Analyze query (using ICU tokenizer)')
          for func in self.preprocessors:
              phrases = func(phrases)
 +
 +        if len(phrases) == 1 \
 +                and phrases[0].text.count(' ') > 3 \
 +                and max(len(s) for s in phrases[0].text.split()) < 3:
 +            normalized = []
 +
          query = qmod.QueryStruct(phrases)
  
          log().var_dump('Normalized query', query.source)
@@@ -250,7 -244,7 +250,7 @@@
              standardized form search will work with. All information removed
              at this stage is inevitably lost.
          """
-         return cast(str, self.normalizer.transliterate(text))
+         return cast(str, self.normalizer.transliterate(text)).strip('-: ')
  
      def split_query(self, query: qmod.QueryStruct) -> Tuple[QueryParts, WordDict]:
          """ Transliterate the phrases and split them into tokens.