search

danswer-ai / danswer

Gen-AI Chat for Teams - Think ChatGPT if it had access to your team's unique knowledge.
https://docs.danswer.dev/
Other
9.77k stars 1.09k forks source link

Confluence/Jira Data Center support? #508

Open voarsh2 opened 9 months ago

voarsh2 commented 9 months ago

Was excited to see this project.

A shame data center of Atlassian's apps have been left behind - I can't make use of Danswer without Data Center variant supported.

Any plan for this?

Weves commented 9 months ago

@voarsh2 are you saying that the self-hosted versions of Jira / Confluence don't work with the current connectors? Sorry, I'm not too familiar with the data center products. How do they differ (besides being self-hosted)? What would we need to do to support them (we'd love to have this support if possible)?

voarsh2 commented 9 months ago

@Weves - Yeah, the self-hosted (DC versions) would require an instance URL, not the default going to atlassian.com.

The https://github.com/GerevAI/gerev project (does similar to this project of indexing and searching of data on connecting sources) on GitHub supports Confluence Cloud/Jira Cloud and DC variants. See https://github.com/GerevAI/gerev/tree/main/app/data_source/sources/confluence - confluence.py/confluence_cloud.py

The Cloud version was "forked" (from Confluence/Jira) sometime in 2013-2015 and they diverged. I can't tell you exactly how they differ (I am not a programmer), but it might help if someone was to check the cloud/DC implementation in GerevAI project. The API's can't be too different (at least I hope).

As a side note, I am not sure how enterprises would feel sharing the data with OpenAI - the whole point of DC products is to be totally secure with data not going to third-parties. Even Atlassian Cloud is trying to make use of third-party AI services, but they would have a special agreement with OpenAI to make sure that data is not used for training and only kept for service provisioning. At least with this project you can setup local LLM's with it and avoid OpenAI (with a personal token, no special agreement around the data retention).

Weves commented 8 months ago

Hey @voarsh2 ! Just wanted to let you know that we just added support for Confluence Data Center (https://github.com/danswer-ai/danswer/pull/624)!

Let me know if you have any questions / if you run into any issues if you get the chance to test this out

voarsh2 commented 8 months ago

Cool - I'll be sure to test it out as soon as I can, not sure if you've tested it thoroughly though.

:)

voarsh2 commented 8 months ago

I gave it a quick test, and it seems personal spaces can't be indexed. E.G. my personal space would be [domain]/display/~voarsh

Can also access with [domain]/spaces/viewspace.action?key=~voarsh Returns error:

Error Message:
'str' object has no attribute 'get'
voarsh2 commented 8 months ago

Retested 0.2.55 tag and it's working for the most part. Any thoughts on Jira DC? 🤔 Any interest?

useruid commented 4 months ago

@voarsh2 , Jira DC is now also supported :)