[
https://issues.apache.org/jira/browse/SOLR-8089?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Varun Thacker updated SOLR-8089:
--------------------------------
Attachment: SOLR-8089.patch
Patch with a test case of EDismax . The tests are failing for queries with
enablePositionIncrements=false
For the following two failing queries here is how the positions are getting put
q="the sky is blue"
sky at pos=0
blue at pos=1
q="sky blue"
sky at pos=0
blue at pos=1
The relevant code from QueryBuilder#analyzePhrase
{code}
while (stream.incrementToken()) {
if (enablePositionIncrements) {
position += posIncrAtt.getPositionIncrement();
} else {
position += 1;
}
builder.add(new Term(field, termAtt.getBytesRef()), position);
}
{code}
So I can see why they are not matching the document.
What am I missing here ?
> Support query parsers being able to set enablePositionIncrements
> ----------------------------------------------------------------
>
> Key: SOLR-8089
> URL: https://issues.apache.org/jira/browse/SOLR-8089
> Project: Solr
> Issue Type: Bug
> Reporter: Varun Thacker
> Attachments: SOLR-8089.patch
>
>
> In 5.0 LUCENE-4963 disallowed setting enablePositionIncrements=false on any
> token filter. The idea being no filter should change the token stream in a
> way that could mess up preceding or following analysis components
> So if a user wants to be able to have PhraseQueries that match across
> stopwords they cannot unless the parser is configured to not take position
> increments into account when generating phrase queries .
> This is documented in the "Token Position Increments" section here :
> https://lucene.apache.org/core/5_3_0/core/org/apache/lucene/analysis/package-summary.html
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
