src/utils/model_versioning.py

import hashlib
import os
import logging

logger = logging.getLogger("ModelVersioning")

# Global cache for model hashes
_HASH_CACHE = {}

def get_file_hash(filepath: str, hash_type: str = "sha256") -> str:
    """
    Calculate MD5 or SHA256 hash of a file with caching to avoid redundant disk I/O.
    """
    if not filepath:
        return "unknown_no_path"
        
    # Standardize path
    abs_path = os.path.abspath(filepath)
    if not os.path.exists(abs_path):
        return "unknown_not_found"
        
    if abs_path in _HASH_CACHE:
        return _HASH_CACHE[abs_path]
        
    try:
        logger.info(f"Computing {hash_type} for {abs_path}...")
        hash_func = hashlib.sha256() if hash_type == "sha256" else hashlib.md5()
        
        with open(abs_path, "rb") as f:
            for chunk in iter(lambda: f.read(65536), b""):
                hash_func.update(chunk)
                
        file_hash = hash_func.hexdigest()
        _HASH_CACHE[abs_path] = file_hash
        logger.info(f"Cached hash for {abs_path}: {file_hash}")
        return file_hash
    except Exception as e:
        logger.error(f"Error computing hash for {abs_path}: {e}")
        return "error_computing_hash"

def get_model_metadata(model_name: str, filepath: str, version: str = "1.0.0") -> dict:
    """
    Return structured model metadata dictionary.
    """
    return {
        "name": model_name,
        "version": version,
        "hash": get_file_hash(filepath)
    }