config/config.py

# RAG Configuration File
# Update these settings as needed

import os
from dotenv import load_dotenv

# Load environment variables from .env file
load_dotenv()

# Common LLM Settings
MAX_TOKENS = 1200 
TEMPERATURE = 0.4

# OCR Settings
OCR_SPACE_API_KEY = os.getenv("OCR_SPACE_API_KEY", "")

# OpenAI Settings
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY", "") 
OPENAI_MODEL = "gpt-3.5-turbo"

# Gemini Settings
GEMINI_API_KEY = os.getenv("GEMINI_API_KEY", "") 
GEMINI_MODEL = "gemini-1.5-flash"

# Groq Settings
GROQ_API_KEY = os.getenv("GROQ_API_KEY") 
GROQ_MODEL = os.getenv("GROQ_MODEL", "llama3-70b-8192")

GROQ_API_KEY_LITE = os.getenv("GROQ_API_KEY_LITE") 
GROQ_MODEL_LITE = "llama3-8b-8192"

# API Authentication
BEARER_TOKEN = os.getenv("BEARER_TOKEN", "c6cee5b5046310e401632a7effe9c684d071a9ef5ce09b96c9ec5c3ebd13085e")

# Chunking (TOKEN-BASED)
CHUNK_SIZE = 1600  #
CHUNK_OVERLAP = 200  

# Retrieval Settings
TOP_K = 12
SCORE_THRESHOLD = 0.3 
RERANK_TOP_K = 9  
BM25_WEIGHT = 0.3 
SEMANTIC_WEIGHT = 0.7

# Advanced RAG Settings
ENABLE_RERANKING = True
ENABLE_HYBRID_SEARCH = True
ENABLE_QUERY_EXPANSION = True
RERANKER_MODEL = "cross-encoder/ms-marco-MiniLM-L-6-v2"
QUERY_EXPANSION_COUNT = 3 
MAX_CONTEXT_LENGTH = 15000

USE_TOTAL_BUDGET_APPROACH = True  

# Embedding Settings
EMBEDDING_MODEL = "BAAI/bge-large-en"  
BATCH_SIZE = 16  

# Paths
OUTPUT_DIR = os.getenv("RAG_EMBEDDINGS_PATH", "./RAG/rag_embeddings")

# API Settings
API_HOST = "0.0.0.0"
API_PORT = 8000
API_RELOAD = True

# Multi-LLM failover system
sequence = ["primary", "secondary", "ternary", "quaternary", "quinary", "senary", "septenary", "octonary", "nonary", "denary"]

def get_provider_configs():
    """
    Get configurations for all provider instances with failover support.
    Supports multiple instances of each provider type for reliability.
    """
    configs = {
        "groq": [],
        "gemini": [],
        "openai": []
    }
    
    # Groq configurations with multiple API keys for failover
    DEFAULT_GROQ_MODEL = "qwen/qwen3-32b"
    configs["groq"] = [{
        "name": sequence[i-1],
        "api_key": os.getenv(f"GROQ_API_KEY_{i}"),
        "model": os.getenv(f"GROQ_MODEL_{i}", DEFAULT_GROQ_MODEL)} for i in range(1, 10) if os.getenv(f"GROQ_API_KEY_{i}", "")
    ]
    
    # Add main GROQ key as primary
    if os.getenv("GROQ_API_KEY"):
        configs["groq"].insert(0, {
            "name": "main",
            "api_key": os.getenv("GROQ_API_KEY"),
            "model": DEFAULT_GROQ_MODEL
        })
    
    # Gemini configurations with multiple API keys for failover
    DEFAULT_GEMINI_MODEL = "gemini-1.5-flash"
    configs["gemini"] = [{
            "name": sequence[i-1],
            "api_key": os.getenv(f"GEMINI_API_KEY_{i}"),
            "model": os.getenv(f"GEMINI_MODEL_{i}", DEFAULT_GEMINI_MODEL)
        } for i in range(1, 10) if os.getenv(f"GEMINI_API_KEY_{i}", "")
    ]
    
    # Add main GEMINI key as primary
    if os.getenv("GEMINI_API_KEY"):
        configs["gemini"].insert(0, {
            "name": "main",
            "api_key": os.getenv("GEMINI_API_KEY"),
            "model": DEFAULT_GEMINI_MODEL
        })
    
    # OpenAI configurations with multiple API keys for failover
    DEFAULT_OPENAI_MODEL = "gpt-4o-mini"
    configs["openai"] = [{
            "name": sequence[i-1],
            "api_key": os.getenv(f"OPENAI_API_KEY_{i}"),
            "model": os.getenv(f"OPENAI_MODEL_{i}", DEFAULT_OPENAI_MODEL)
        } for i in range(1, 10) if os.getenv(f"OPENAI_API_KEY_{i}", "")
    ]
    
    # Add main OPENAI key as primary
    if os.getenv("OPENAI_API_KEY"):
        configs["openai"].insert(0, {
            "name": "main",
            "api_key": os.getenv("OPENAI_API_KEY"),
            "model": DEFAULT_OPENAI_MODEL
        })
    
    return configs

# Specialized API keys for different tasks
GROQ_API_KEY_TABULAR = os.getenv("GROQ_API_KEY_TABULAR", GROQ_API_KEY)
GEMINI_API_KEY_IMAGE = os.getenv("GEMINI_API_KEY_IMAGE", GEMINI_API_KEY)
GEMINI_API_KEY_MULTILINGUAL = os.getenv("GEMINI_API_KEY_MULTILINGUAL", GEMINI_API_KEY)

# Validation (optional - comment out for production)
# assert OPENAI_API_KEY, "OPENAI KEY NOT SET"
# assert GEMINI_API_KEY, "GEMINI KEY NOT SET" 
# assert GROQ_API_KEY, "GROQ KEY NOT SET"