#138 the system instruction inserted in the begining of user prompt

[hariprasathys22]

#138
feat: Add [merge_system_prompt] option for models that don't support system instructions (e.g., Gemma 3)

Description
Problem
When using certain AI models like Gemma 3 via Google AI API, the application fails with error:

Added a new boolean field merge_system_prompt to the OpenAIAgent model that allows administrators to configure how system instructions are sent to the API.

Changes Made

merge_system_prompt = models.BooleanField( _("Merge System Prompt to User Message"), default=False, help_text=_("Enable for models that don't support system instructions (e.g., Gemma 3)") )

How It Works
Before (Standard Behavior):
messages = [ {"role": "system", "content": "You are a translator..."}, {"role": "user", "content": "Hello, how are you?"} ]

After (With merge_system_prompt=True):
messages = [ {"role": "user", "content": "You are a translator...\n\nHello, how are you?"} ]

[versun]

Thank you @hariprasathys22 for your contribution! The merge_system_prompt feature is a great addition that enables support for models that don't support system instructions (like Gemma 3).

I've merged your PR and created a follow-up fix PR (#141) to address some token calculation issues. The feature is now working correctly.

Thanks again for your valuable contribution to RSSBox! 🎉

[hariprasathys22]

Thank you @versun

[hariprasathys22]

There is one more issue i have faced, when i integrated the gemma ai, that in the advanced settings, reasoning_factor is given value as "minimal", but for google ai model has reasoning_factor as "low", "high" etc.,

Can I fix it in the next PR.

[versun]

No need, as you can directly modify the value of this parameter in the advanced settings.