[
https://issues.apache.org/jira/browse/LUCENE-3099?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Martijn van Groningen updated LUCENE-3099:
------------------------------------------
Attachment: LUCENE-3099.patch
Attached an initial idea of abstracting away the source of the group values.
In the patch there are four concepts:
* GroupValue. An abstraction of a value.
* GroupValueSource. An abstraction of the source of the group value.
* GroupHolder. An abstract holder to get the groups in a efficient manner.
* GroupSpecification. A factory class that glues all the above concepts
together.
I think with this infrastructure it is quite straight forward to add the
ability to group by a function. The patch contains implementations for all
fields (string, int, double etc.).
Some concepts are look like with what is already in Lucene / Solr. For example
Lucene's ValueSource and DocValues or Solr's DocValues and MutableValue. I just
started from scratch to see what grouping really needs. The Lucene's DocValues
for example didn't have all functionality grouping needs.
Furthermore I have added research group collectors. I ported the
MatchAllGroupsCollector and FirstPassGroupingCollector to use this new
infrastructure. Just to get a feeling of how it all fits together.
I also included a simple runner class that runs these research collectors and
compares them to the variants already added to the code base. You can then
easily check search times and the group results.
> Grouping module should allow subclasses to set the group key per document
> -------------------------------------------------------------------------
>
> Key: LUCENE-3099
> URL: https://issues.apache.org/jira/browse/LUCENE-3099
> Project: Lucene - Java
> Issue Type: Improvement
> Reporter: Michael McCandless
> Fix For: 3.2, 4.0
>
> Attachments: LUCENE-3099.patch
>
>
> The new grouping module can only group by a single-valued indexed field.
> But, if we make the 'getGroupKey' a method that a subclass could override,
> then I think we could refactor Solr over to the module, because it could do
> function queries and normal queries via subclass (I think).
> This also makes the impl more extensible to apps that might have their own
> interesting group values per document.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
