src/config.py

# """
# Configuration settings for the insurance product classifier
# """

import os
from pathlib import Path

# Base directory (project root)
BASE_DIR = Path(__file__).resolve().parent.parent

# Model directory
MODEL_DIR = BASE_DIR / "models"

# Model files
EMBEDDINGS_FILE = MODEL_DIR / "category_embeddings_mpnet.npy"
METADATA_FILE = MODEL_DIR / "category_metadata.pkl"
CONFIG_FILE = MODEL_DIR / "config.json"

# Model configuration
MODEL_NAME = "sentence-transformers/all-mpnet-base-v2"
EMBEDDING_DIMENSION = 768

# Classification thresholds
AUTO_APPROVE_THRESHOLD = 0.75  # 75% confidence
QUICK_REVIEW_THRESHOLD = 0.60  # 60% confidence

# Keyword boosting
BOOST_FACTOR = 0.15  # 15% boost for keyword matches
MAX_BOOST = 0.30  # Maximum 30% total boost

# API settings
API_TITLE = "Insurance Product Classification API"
API_VERSION = "1.0.0"
API_DESCRIPTION = "ML-powered product categorization for insurance underwriting"

# Processing settings
DEFAULT_TOP_K = 5  # Return top 5 matches
BATCH_SIZE = 32  # For batch processing

# Keywords for boosting
PRODUCT_KEYWORDS = {
    # Electronics
    "iphone",
    "ipad",
    "macbook",
    "smartphone",
    "laptop",
    "tablet",
    "computer",
    "electronics",
    "phone",
    "mobile",
    "samsung",
    "apple",
    "dell",
    "hp",
    # Appliances
    "refrigerator",
    "dishwasher",
    "washing machine",
    "dryer",
    "oven",
    "microwave",
    "coffee maker",
    "blender",
    "toaster",
    "appliance",
    # Clothing
    "shoes",
    "shirt",
    "pants",
    "dress",
    "jacket",
    "sneakers",
    "boots",
    "clothing",
    "apparel",
    "footwear",
    # Books
    "book",
    "novel",
    "textbook",
    "ebook",
    "reading",
    "literature",
    # Sports
    "sports",
    "fitness",
    "exercise",
    "gym",
    "athletic",
    "running",
    "yoga",
    # Home
    "furniture",
    "decor",
    "bedding",
    "kitchen",
    "home",
    "garden",
}


def validate_files():
    """Validate that all required model files exist"""
    required_files = [EMBEDDINGS_FILE, METADATA_FILE, CONFIG_FILE]

    missing_files = []
    for file_path in required_files:
        if not file_path.exists():
            missing_files.append(str(file_path))

    if missing_files:
        raise FileNotFoundError(
            f"Missing required files:\n" + "\n".join(f"  - {f}" for f in missing_files)
        )

    return True


if __name__ == "__main__":
    print("Configuration Settings:")
    print(f"  Model Directory: {MODEL_DIR}")
    print(f"  Embeddings File: {EMBEDDINGS_FILE.name}")
    print(f"  Metadata File: {METADATA_FILE.name}")
    print(f"  Auto-Approve Threshold: {AUTO_APPROVE_THRESHOLD * 100}%")
    print(f"  Quick Review Threshold: {QUICK_REVIEW_THRESHOLD * 100}%")

    try:
        validate_files()
        print("\n✅ All required files found!")
    except FileNotFoundError as e:
        print(f"\n❌ Error: {e}")