-
Type: Improvement
-
Status: Resolved
-
Priority: Minor
-
Resolution: Fixed
-
Affects Version/s: 6.0, 7.4
-
Fix Version/s: 10.3
-
Component/s: Elasticsearch
-
Release Notes Description:
-
Epic Link:
-
Sprint:nxcore 10.3.6
-
Story Points:3
The quick search (DEFAULT_DOCUMENT_SUGGESTION) page provider use a fulltext search which is based on simple_query_string operator.
There is an * appended to the input to have a suggestion.
The problem is that the search query with a wildcard pattern is not analyzed according to an analyzer, it is just lowercased.
This lead to some missing match depending on the indexing analyzer used.
NXP-17692 is a work around for accented character.
This does not solve the problem if the pattern has been stemmed by the fulltext analyzer. For instance searching for 'foos*' will not match an indexed text 'foos'.
Because the indexed term is "foo", the minimal_english stemmer has removed the ending 's'.
Also for ideographic characters like in Chinese, the fulltext analyzer creates a token per idograph, exact search is working but adding a wildcard does not work.
A better suggestion operator is to use match_phrase_prefix.