[FR] Kreuzberg as an alternative content exractor to Apache Tika

[cpollmann]

Is your feature request related to a problem? Please describe.

According to the .env.example from the opencloud-compose repository, Apache Tika is disabled as a search extractor by default due to performance reasons.

Describe the solution you'd like

I would like to propose adding support for an additional content extractor based on the Kreuzberg project.

From Kreuzberg's README:

Extract text and metadata from a wide range of file formats (91+), generate embeddings and post-process at native speeds without needing a GPU.

Flexible deployment – Use as library, CLI tool, REST API server, or MCP server

Describe alternatives you've considered

N/A

Additional context

I am not a developer so I cannot estimate the effort required to develop such a content extractor for OpenCloud, nor can I validate the quality or production-readiness of the mentioned project. However, I would happily receive feedback on the idea in general.

My research showed that the co-founder of Kreuzberg is also located in Berlin! ;)

[No1234DDgg]

That would be very nice <3

[pascalwengerter]

After taking a look at https://docs.kreuzberg.dev/comparisons/kreuzberg-vs-tika and https://kreuzberg.dev/benchmarks I don't think that's a reasonable ask for the foreseable future - and that's coming from someone that has used kreuzberg and is a fan 😉