infrastructure/llm_client.py

"""
🇬🇧 Module: llm_client.py
Purpose: Adapter for Featherless.ai (OpenAI-compatible API).

🇷🇺 Модуль: llm_client.py
Назначение: адаптер для LLM-инференса через Featherless.ai (совместимо с OpenAI API).
"""

import os
from typing import List, Dict, Generator
from openai import OpenAI
from config import FEATHERLESS_API_KEY, FEATHERLESS_MODEL


class FeatherlessLLM:
    """Wrapper for Featherless.ai LLM inference."""

    def __init__(self, api_key: str = FEATHERLESS_API_KEY, model: str = FEATHERLESS_MODEL):
        if not api_key:
            raise RuntimeError("❌ Environment variable 'featherless' (API key) is missing.")
        self.client = OpenAI(base_url="https://api.featherless.ai/v1", api_key=api_key)
        self.model = model

    def stream_chat(self, *, messages: List[Dict], model: str = None) -> Generator[str, None, None]:
        """Stream chat completion using Featherless.ai."""
        used_model = model or self.model
        response = self.client.chat.completions.create(
            model=used_model,
            messages=messages,
            stream=True,
        )
        for chunk in response:
            delta = chunk.choices[0].delta.content
            if delta:
                yield delta


# === Global singleton instance ===
llm_service = FeatherlessLLM()