sankalp
7016cff3a7
Add support for the Ollama provider in the AI generation pipeline and expose a configurable host address. New Features: - Introduce `ollama` as a selectable provider. - Add `ollama_host` configuration option with default `http://localhost:11434`. Enhancements: - Extend provider list in error message to include Ollama.
38 lines
1.8 KiB
Python
38 lines
1.8 KiB
Python
from importlib import import_module
|
|
|
|
|
|
def generate_with_ai(
|
|
prompt, provider="gemini", api_key=None, model="gemini-2.5-flash", **kwargs
|
|
):
|
|
"""Generates AI-based content such as commit messages using the specified provider and model.
|
|
This function dynamically loads and initializes the provider class, sets the API key, and invokes the provider's generate method with the provided prompt and additional parameters.
|
|
|
|
Args:
|
|
prompt (str): The input prompt used to generate content.
|
|
provider (str): The name of the provider module and class to use; defaults to "gemini".
|
|
api_key (str, optional): The API key for authenticating with the provider. Should be provided via config file or kwargs.
|
|
model (str): The name of the model to use with the provider; defaults to "gemini".
|
|
**kwargs: Additional keyword arguments to pass to the provider's generate method.
|
|
|
|
Returns:
|
|
str: The generated content produced by the AI provider.
|
|
"""
|
|
try:
|
|
provider_module = import_module(f"devgen.providers.{provider}")
|
|
class_name = "".join([x.capitalize() for x in provider.split("_")]) + "Provider"
|
|
provider_class = getattr(provider_module, class_name)
|
|
except ModuleNotFoundError as e:
|
|
raise ImportError(
|
|
f"Provider `{provider}` is not a built-in module: {e}. "
|
|
f"Available providers: gemini, openai, anthropic, huggingface, "
|
|
f"openrouter, ollama."
|
|
) from e
|
|
except AttributeError as e:
|
|
raise ImportError(
|
|
f"Provider `{provider}` exists but does not expose a "
|
|
f"`{class_name}` class. This is a packaging bug."
|
|
) from e
|
|
|
|
provider_instance = provider_class()
|
|
return provider_instance.generate(prompt, api_key=api_key, model=model, **kwargs)
|