sankalp
8069895bc7
Add token limit error detection and formatting across all provider modules to surface clear diagnostics when context windows are exceeded, improving error messages and preventing silent failures. New Features - Introduce `is_token_limit_error` checks in Anthropic, Gemini, Hugging Face, Ollama, OpenAI, and OpenRouter providers. - Use `format_token_limit_error` to generate user‑friendly error messages for token limit overs. Enhancements - Centralize token limit error handling for consistent diagnostics across providers. - Explicitly re‑raise `RuntimeError` in Hugging Face to preserve original behavior. Bug Fixes - None
55 lines
1.9 KiB
Python
55 lines
1.9 KiB
Python
from devgen.utils import format_token_limit_error, is_token_limit_error
|
|
|
|
|
|
class OllamaProvider:
|
|
"""Generates content using a local Ollama server.
|
|
|
|
Ollama exposes an OpenAI-compatible endpoint at /v1/chat/completions,
|
|
so we can reuse the openai client. The host defaults to
|
|
http://localhost:11434 but can be overridden via the
|
|
`ollama_host` kwarg (or `ollama_host` in ~/.devgen.yaml).
|
|
"""
|
|
|
|
DEFAULT_HOST = "http://localhost:11434"
|
|
DEFAULT_MODEL = "llama3.2"
|
|
|
|
def generate(
|
|
self,
|
|
prompt: str,
|
|
api_key: str | None = None,
|
|
model: str | None = None,
|
|
**kwargs,
|
|
) -> str:
|
|
from openai import OpenAI
|
|
|
|
host = kwargs.get("ollama_host") or self.DEFAULT_HOST
|
|
base_url = f"{host.rstrip('/')}/v1"
|
|
chosen_model = model or self.DEFAULT_MODEL
|
|
|
|
client = OpenAI(base_url=base_url, api_key=api_key or "ollama")
|
|
|
|
kwargs.pop("debug", None)
|
|
kwargs.pop("ollama_host", None)
|
|
|
|
try:
|
|
response = client.chat.completions.create(
|
|
model=chosen_model,
|
|
messages=[{"role": "user", "content": prompt}],
|
|
**kwargs,
|
|
)
|
|
except Exception as e:
|
|
if is_token_limit_error(e):
|
|
raise RuntimeError(format_token_limit_error("Ollama", e)) from e
|
|
raise RuntimeError(
|
|
f"Ollama request failed (host={host}, model={chosen_model}): {e}. "
|
|
"Is the Ollama server running? Start it with `ollama serve` and "
|
|
"pull a model with `ollama pull <model>`."
|
|
) from e
|
|
|
|
if not response.choices:
|
|
raise RuntimeError(
|
|
f"Ollama returned no choices for model {chosen_model!r}. "
|
|
"The model may not be installed locally."
|
|
)
|
|
return response.choices[0].message.content.strip()
|