[
https://issues.apache.org/jira/browse/LUCENE-7429?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Adrien Grand updated LUCENE-7429:
---------------------------------
Attachment: LUCENE-7429.patch
Agreed on that point. Since all we need for normalization is to use the same
attribute factory as the regular analysis chain, would it work better for you
if we tried to borrow the attribute source from a token stream created by
{{Analyzer.tokenStream}} like in the attached patch (the relevant bits are the
removal of {{Analyzer.attributeFactory()}} and how {{Analyzer.normalize}}
assigns the {{attributeFactory}} variable?
> DelegatingAnalyzerWrapper should delegate normalization too
> -----------------------------------------------------------
>
> Key: LUCENE-7429
> URL: https://issues.apache.org/jira/browse/LUCENE-7429
> Project: Lucene - Core
> Issue Type: Bug
> Affects Versions: 6.2
> Reporter: Adrien Grand
> Priority: Minor
> Attachments: LUCENE-7355.patch, LUCENE-7429.patch, LUCENE-7429.patch
>
>
> This is something that I overlooked in LUCENE-7355:
> (Delegating)AnalyzerWrapper uses the default implementation of
> initReaderForNormalization and normalize, meaning that by default the
> normalization is a no-op. It should delegate to the wrapped analyzer.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
