如何从保存/转储文件加载数值，替代硬编码系数？附代码示例

Hey there! Let's tackle your two questions one by one—loading values from files to replace hardcoded coefficients, and loading numbers from saved/dumped files. I'll use Python examples since it's super common for these tasks, but you can adapt the logic to other languages if needed.

Hardcoding coefficients (like learning_rate = 0.01) in your code makes it a hassle to adjust values without editing the code itself. Here are a few practical, structured ways to fix this:

Option 1: Simple Text File (Key-Value Pairs) #

If you prefer a lightweight format without extra dependencies, a plain text file with key-value lines works great. Let's say you have a file named coefficients.txt:

learning_rate=0.01
threshold=0.8
weight=2.5

This is a comment, we'll skip it #

You can load it like this:

coefficients = {}
try:
    with open('coefficients.txt', 'r') as f:
        for line in f:
            # Skip empty lines and comments
            stripped_line = line.strip()
            if not stripped_line or stripped_line.startswith('#'):
                continue
            key, value = stripped_line.split('=')
            # Convert to the correct numeric type
            coefficients[key] = float(value) if '.' in value else int(value)
except FileNotFoundError:
    print("Error: coefficients.txt not found! Falling back to defaults.")
    # Set default values as a safety net
    coefficients = {"learning_rate": 0.01, "threshold": 0.8, "weight": 2.5}

# Now use the loaded values instead of hardcoding
learning_rate = coefficients['learning_rate']
threshold = coefficients['threshold']
print(f"Loaded learning rate: {learning_rate}, threshold: {threshold}")

Option 2: JSON File (Structured & Language-Agnostic) #

JSON is a universal format that's easy to read and works across different programming languages. First, you can save your coefficients to coeffs.json (run this once to create the file):

import json

initial_coeffs = {"learning_rate": 0.01, "threshold": 0.8, "weight": 2.5}
with open('coeffs.json', 'w') as f:
    json.dump(initial_coeffs, f, indent=4)  # Indent makes it human-readable

Then load it whenever you need:

import json

try:
    with open('coeffs.json', 'r') as f:
        coeffs = json.load(f)
except FileNotFoundError:
    coeffs = {"learning_rate": 0.01, "threshold": 0.8, "weight": 2.5}

learning_rate = coeffs['learning_rate']
print(f"Loaded learning rate from JSON: {learning_rate}")

Option 3: YAML File (Clean & Human-Friendly) #

YAML is even more readable than JSON, making it perfect for configuration files. You'll need to install the pyyaml package first (pip install pyyaml).

Your coeffs.yaml file would look like:

learning_rate: 0.01
threshold: 0.8
weight: 2.5

Load it with:

import yaml

try:
    with open('coeffs.yaml', 'r') as f:
        coeffs = yaml.safe_load(f)  # safe_load prevents arbitrary code execution
except FileNotFoundError:
    coeffs = {"learning_rate": 0.01, "threshold": 0.8, "weight": 2.5}

threshold = coeffs['threshold']
print(f"Loaded threshold from YAML: {threshold}")

If you've serialized (dumped) numbers or arrays into a file (like using Python's pickle or NumPy's formats), here's how to load them back:

Option 1: Pickle (Python-Specific Serialization) #

Pickle lets you save almost any Python object, including lists of numbers or single values. First, if you haven't dumped the file yet:

import pickle

# Save a list of numbers
numbers = [1.2, 3.4, 5.6, 7.8]
with open('numbers.pkl', 'wb') as f:
    pickle.dump(numbers, f)

To load it back:

import pickle

try:
    with open('numbers.pkl', 'rb') as f:
        loaded_numbers = pickle.load(f)
except FileNotFoundError:
    loaded_numbers = []

print(f"Loaded numbers from pickle: {loaded_numbers}")

Important note: Only load pickle files from trusted sources—they can execute malicious code if tampered with.

Option 2: NumPy Files (For Numerical Arrays) #

If you're working with numerical arrays (common in data science or machine learning), NumPy's .npy and .npz formats are optimized for this use case.

Loading a single array from .npy #

import numpy as np

# Save first (if needed)
# arr = np.array([10, 20, 30, 40])
# np.save('numbers.npy', arr)

loaded_arr = np.load('numbers.npy')
print(f"Loaded NumPy array: {loaded_arr}")

Loading multiple arrays from .npz #

import numpy as np

# Save multiple arrays with named keys
# np.savez('multi_numbers.npz', arr1=np.array([1,2]), arr2=np.array([3,4]))

loaded_data = np.load('multi_numbers.npz')
arr1 = loaded_data['arr1']
arr2 = loaded_data['arr2']
print(f"Loaded arr1: {arr1}, arr2: {arr2}")

Quick Pro Tip #

Always add error handling (like try-except blocks) to handle cases where the file doesn't exist or is corrupted—it makes your code way more robust and user-friendly!

内容的提问来源于stack exchange，提问作者Zhang Jinzhe