Data from Solr

[adjouama]

Hello,

first, thank you a lot for this repo. I have a quick question.
my data is indexed in Solr. is it possible to use it instead of Elastic search ?

I can't find documentation in this repo regarding data preparation. Any guidance ?

thank you a lot in advance.

Best

[tholor]

Hey @adjouama,

We currently don't have Solr integrated. However, as Haystack is completely modular, it would definitely be possible to do that.

What you would need to do:

1. Implement a SolrDocumentStore
   You could take the ElasticsearchDocumentStore as an example and adjust it to Solr.
   This will be the major work, but you won't need all of the implemented functions there (e.g. query_by_embedding is not mandatory). The documentstore is needed for indexing docs, triggering queries and storing the mapping of fields from DB to haystack Documents

2. Implement a SolrRetriever (or use the in-memory TfidfRetriever for smaller prototyping)
   The retriever could be rather light-weight, e.g. consider the basic ElasticsearchRetriever which has only ~ 6 lines of code.

The other major components (Reader, Finder ...) are independent of this and will work out-of-the-box. We also plan to have some advanced Retrievers soon that are independent of the datastore.

Let us know if you want to tackle this in a PR. We are happy to support you along the way.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 14 days if no further activity occurs.

[Always-prog]

Hi @adjouama.
I also need to refactor the work haystack to it will work with Solr.
But, did you do it? And if you refactored haystack for Solr, can I use your code?
Thanks.

[adjouama]

Hi @adjouama.
I also need to refactor the work haystack to it will work with Solr.
But, did you do it? And if you refactored haystack for Solr, can I use your code?
Thanks.

Hi @Always-prog , unfortunately I did not spend time on the Solr implementation. I moved all my data on Elasticsearch and I am very much satisfied about the result. I'd suggest that you migrate :)