postcode token should have transliterated term in word_token

[nominatim.git] / src / nominatim_api / search / icu_tokenizer.py

diff --git a/src/nominatim_api/search/icu_tokenizer.py b/src/nominatim_api/search/icu_tokenizer.py
index b3e14f6a114a6ffb649e761953c6ae81950a7880..b6cd34646f4029a619d1cfa1ef262f015fab04b4 100644 (file)
--- a/src/nominatim_api/search/icu_tokenizer.py
+++ b/src/nominatim_api/search/icu_tokenizer.py
@@ -193,10 +193,12 @@ class ICUQueryAnalyzer(AbstractQueryAnalyzer):
 
         self.add_extra_tokens(query)
         for start, end, pc in self.postcode_parser.parse(query):
 
         self.add_extra_tokens(query)
         for start, end, pc in self.postcode_parser.parse(query):

+            term = ' '.join(n.term_lookup for n in query.nodes[start + 1:end + 1])

             query.add_token(qmod.TokenRange(start, end),
                             qmod.TOKEN_POSTCODE,
                             ICUToken(penalty=0.1, token=0, count=1, addr_count=1,
             query.add_token(qmod.TokenRange(start, end),
                             qmod.TOKEN_POSTCODE,
                             ICUToken(penalty=0.1, token=0, count=1, addr_count=1,

-                                     lookup_word=pc, word_token=pc, info=None))
+                                     lookup_word=pc, word_token=term,
+                                     info=None))

         self.rerank_tokens(query)
 
         log().table_dump('Word tokens', _dump_word_tokens(query))
         self.rerank_tokens(query)
 
         log().table_dump('Word tokens', _dump_word_tokens(query))