Support for Ollama

[Mulugruntz]

Is your feature request related to a problem? Please describe.
新功能是否与解决某个问题相关, 请描述
When trying to have the extension target my local Mistral handled by Ollama, it fails (HTTP 403).

The settings:

I haven't digged into the code yet, but maybe it's because it uses a different API.

Can see here that Ollama WebUI works as expected

and that Ollama is running

Describe the solution you'd like
你期望的新功能实现方案
ChatGPTBox to work with Ollama.

Additional context
其他
Yes, see above :).

[josStorer]

Could you please capture the request through F12 Network in the Ollama WebUI, and provide the relevant request headers and body

[Donno191]

It would be nice to use ollama ! Maybe it works with the support of ollama openai api ? Example of settings ?

[josStorer]

@Mulugruntz @Donno191

After consulting the documentation of ollama, I found that ollama provides APIs compatible with OpenAI API.

And for the 403 error, it is actually caused by cross-origin requests. According to https://github.com/ollama/ollama/blob/main/docs/faq.md#how-can-i-allow-additional-web-origins-to-access-ollama, you can set environment variable OLLAMA_ORIGINS to *, and then restart ollama to allow requests from browser extensions.

The settings of ChatGPTBox are as follows:

[BarfingLemurs]

thanks for the tip, filling the model name eg: "gemma:2b" is a requirement for ollama to works, along with: export OLLAMA_ORIGINS=* on linux.

[Katzenwerfer]

Something to note.
Setting origins with a wildcard is generally not recommended.
It is better to be specific and ideally add moz-extension://[ID OF THE EXTENSION], or with a wildcard if you can't find the ID moz-extension://*