MCP-RAG

import os from pathlib import Path from dotenv import load_dotenv # Load environment variables load_dotenv() class Config: # OpenAI API Key OPENAI_API_KEY = "xxxxxx" BASE_URL = "" MODEL_NAME = "claude-3-5-sonnet" # EMBEDDING_MODEL # File processing settings MAX_FILE_SIZE_MB = 200 # Maximum file size to process CHUNK_SIZE_BYTES = 8192 # For file reading MEMORY_THRESHOLD_MB = 30 # Threshold for chunking strategy PROCESSING_TIMEOUT_SECONDS = 600 # 10 minutes # Milvus settings MILVUS_HOST = os.getenv("MILVUS_HOST", "localhost") MILVUS_PORT = os.getenv("MILVUS_PORT", "19530") MILVUS_USER = os.getenv("MILVUS_USER", "") MILVUS_PASSWORD = os.getenv("MILVUS_PASSWORD", "") MILVUS_DB_NAME = os.getenv("MILVUS_DB_NAME", "rag_documents") MILVUS_COLLECTION_NAME = "document_embeddings" # RAG settings # EMBEDDING_MODEL = "all-MiniLM-L6-v2" # SentenceTransformer model EMBEDDING_MODEL = "text-embedding-bge-m3" USE_OPENAI_EMBEDDINGS = bool(OPENAI_API_KEY) # Use OpenAI if key available VECTOR_DB_PATH = "./data/chroma_db" T_SYSTEMS_MAX_BATCH_SIZE = 128 # Text chunking settings CHUNK_SIZE = 500 CHUNK_OVERLAP = 100 # Search settings DEFAULT_SEARCH_RESULTS = 10 MAX_CONTEXT_LENGTH = 4000 # Streamlit settings UPLOAD_FOLDER = "./data/uploads" MAX_UPLOAD_SIZE_MB = 200 PDF_BATCH_SIZE = 5 # Process 5 pages at a time # PowerPoint extraction settings EXTRACT_SLIDE_NOTES = True # Include slide notes in extraction EXTRACT_PRESENTATION_METADATA = True # Include presentation properties POWERPOINT_BATCH_SIZE = 5 # Slides per batch for large presentations # Image OCR settings IMAGE_OCR_ENHANCEMENT_LEVEL = "standard" # light, standard, aggressive IMAGE_OCR_LANGUAGES = "eng" # Tesseract language codes IMAGE_MIN_CONFIDENCE = 30 # Minimum OCR confidence threshold IMAGE_BATCH_SIZE = 5 # Images to process in batch # OCR preprocessing settings ENABLE_IMAGE_PREPROCESSING = True OCR_DPI_THRESHOLD = 300 # Minimum DPI for good OCR ENABLE_DESKEW = True # Auto-correct skewed images ENABLE_NOISE_REMOVAL = True # Create directories @classmethod def setup_directories(cls): Path(cls.VECTOR_DB_PATH).mkdir(parents=True, exist_ok=True) Path(cls.UPLOAD_FOLDER).mkdir(parents=True, exist_ok=True) # Initialize directories Config.setup_directories()