New stcructure v1.2

config.py

@@ -1,24 +1,20 @@
 """
 🇬🇧 Module: config.py
-Purpose: Central configuration for environment variables and constants
-used across the entire TradeLink AI Analyzer project.
+Purpose: Central configuration for environment variables and constants.
 
 🇷🇺 Модуль: config.py
-Назначение: централизованная конфигурация переменных окружения и констант
-для всего проекта TradeLink AI Analyzer.
+Назначение: централизованная конфигурация переменных окружения и констант проекта.
 """
 
 import os
 
-# === Hugging Face LLM Configuration ===
-HF_LLM_API_KEY = os.getenv("HF_LLM_integrate")  # token name in Space settings
-HF_LLM_API_URL = "https://api-inference.huggingface.co/models"
+# === Featherless.ai Configuration ===
+FEATHERLESS_API_KEY = os.getenv("featherless")
+FEATHERLESS_MODEL = "meta-llama/Meta-Llama-3.1-8B-Instruct"
 
 # === API Base URLs ===
 TRADELINK_API_URL = "https://api.tradelink.pro"
 
-# === Request/Connection Settings ===
-REQUEST_TIMEOUT = 15  # seconds
-
-# === Debug Mode ===
+# === Other Settings ===
+REQUEST_TIMEOUT = 15
 DEBUG = os.getenv("DEBUG", "false").lower() == "true"

requirements.txt

@@ -1,5 +1,6 @@
 gradio>=4.29.0
-requests>=2.31.0
-httpx>=0.27.0
-pandas>=2.2.0
-matplotlib>=3.8.0
+openai>=1.30.1
+requests
+httpx
+pandas
+matplotlib

services/llm_client.py

@@ -1,56 +1,39 @@
 """
 🇬🇧 Module: llm_client.py
-Purpose: Adapter for Hugging Face Inference API (LLM).
-Provides a unified interface for text generation and future model extensions.
+Purpose: Adapter for Featherless.ai (OpenAI-compatible API).
 
 🇷🇺 Модуль: llm_client.py
-Назначение: адаптер для LLM через Hugging Face Inference API.
-Обеспечивает единый интерфейс для генерации текста и расширения под другие модели в будущем.
+Назначение: адаптер для LLM-инференса через Featherless.ai (совместимо с OpenAI API).
 """
 
-import requests
+import os
 from typing import List, Dict, Generator
-from config import HF_LLM_API_KEY, HF_LLM_API_URL, DEBUG
+from openai import OpenAI
+from config import FEATHERLESS_API_KEY, FEATHERLESS_MODEL
 
 
-class HuggingFaceLLMClient:
-    """Lightweight client for Hugging Face text generation models."""
+class FeatherlessLLM:
+    """Wrapper for Featherless.ai LLM inference."""
 
-    def __init__(self, api_key: str):
+    def __init__(self, api_key: str = FEATHERLESS_API_KEY, model: str = FEATHERLESS_MODEL):
         if not api_key:
-            raise ValueError("Missing Hugging Face API key (HF_LLM_integrate).")
-        self.api_key = api_key
-        self.headers = {"Authorization": f"Bearer {self.api_key}"}
-
-    def complete(self, model: str, prompt: str) -> str:
-        """Non-streaming completion request."""
-        url = f"{HF_LLM_API_URL}/{model}"
-        payload = {"inputs": prompt}
-
-        if DEBUG:
-            print(f"[DEBUG] Sending request to {url}")
-
-        resp = requests.post(url, headers=self.headers, json=payload)
-        resp.raise_for_status()
-        result = resp.json()
-
-        # Handle HF response structures
-        if isinstance(result, list) and "generated_text" in result[0]:
-            return result[0]["generated_text"]
-        if isinstance(result, dict) and "generated_text" in result:
-            return result["generated_text"]
-
-        return str(result)
-
-    def stream_chat(self, *, messages: List[Dict], model: str) -> Generator[str, None, None]:
-        """
-        Placeholder for streaming logic (not all Hugging Face endpoints support streaming).
-        For now, concatenates messages and sends as one prompt.
-        """
-        prompt = "\n".join(m["content"] for m in messages)
-        text = self.complete(model, prompt)
-        yield text
-
-
-# Global instance for reuse
-llm_service = HuggingFaceLLMClient(api_key=HF_LLM_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()