Python小练习：argparse的用法

作者：凯鲁嘎吉 - 博客园 http://www.cnblogs.com/kailugaji/

1. test.py

#!/usr/bin/env python3
# -*- coding=utf-8 -*-
# 参考：argparse --- 命令行选项、参数和子命令解析器
# https://docs.python.org/zh-cn/3/library/argparse.html#argumentparser-objects
# --------------------------------------------------------------------------------
# https://www.cnblogs.com/kailugaji/
import argparse
 
# I. 创建 ArgumentParser() 对象
parser = argparse.ArgumentParser(
    prog='kailugaji', usage='%(prog)s [options]',
    description = 'argparse --- 命令行选项、参数和子命令解析器',
    epilog = '凯鲁嘎吉 - 博客园: https://www.cnblogs.com/kailugaji/')
 
# II. 调用 add_argument() 方法添加参数
parser.add_argument('--dataset', '-ds', default='cifar10',
                    choices=['cifar10', 'cifar100', 'tin', 'IN'], help='which dataset to use')
parser.add_argument('--optimizer', '-opt', default='sgd',
                    choices=['sgd', 'adam', 'adagrad'], help='which optimizer to use')
parser.add_argument('--num_workers', '-cpus', default=16, type=int)
parser.add_argument('--batch-size', '-bs', type=int, default=256, help='batch size for training')
# 在参数解析时，参数中并不区分字符‘-’和‘_’
# 如上所述，在添加是使用的是’–batch-size’，但是在解析时使用的是args.batch_size
 
# III. 使用 parse_args() 解析添加的参数
args = parser.parse_args()
 
# 打印
print("文件test.py的帮助信息：")
parser.print_help()
print("\n-------------------------------------------------------------------------------------------------------")
print("方式一：")
print("1. args: ", args)
print("2. Dataset: ",args.dataset)
print("3. Num_workers: ", args.num_workers)
print("4. Batch Size: ", args.batch_size)
print("5. Optimizer: ", args.optimizer)
 
args_2 = parser.parse_args(
    ['-bs', '128', '-ds', 'tin', '-opt', 'adagrad', '-cpus', '8'])
# 打印
print("\n-------------------------------------------------------------------------------------------------------")
print("方式二：")
print("1. args_2: ", args_2)
print("2. Dataset: ",args_2.dataset)
print("3. Num_workers: ", args_2.num_workers)
print("4. Batch Size: ", args_2.batch_size)
print("5. Optimizer: ", args_2.optimizer)

2. 结果

文件test.py的帮助信息：
usage: kailugaji [options]
 
argparse --- 命令行选项、参数和子命令解析器
 
optional arguments:
  -h, --help            show this help message and exit
  --dataset {cifar10,cifar100,tin,IN}, -ds {cifar10,cifar100,tin,IN}
                        which dataset to use
  --optimizer {sgd,adam,adagrad}, -opt {sgd,adam,adagrad}
                        which optimizer to use
  --num_workers NUM_WORKERS, -cpus NUM_WORKERS
  --batch-size BATCH_SIZE, -bs BATCH_SIZE
                        batch size for training
 
凯鲁嘎吉 - 博客园: https://www.cnblogs.com/kailugaji/
 
-------------------------------------------------------------------------------------------------------
方式一：
1. args:  Namespace(batch_size=256, dataset='cifar10', num_workers=16, optimizer='sgd')
2. Dataset:  cifar10
3. Num_workers:  16
4. Batch Size:  256
5. Optimizer:  sgd
 
-------------------------------------------------------------------------------------------------------
方式二：
1. args_2:  Namespace(batch_size=128, dataset='tin', num_workers=8, optimizer='adagrad')
2. Dataset:  tin
3. Num_workers:  8
4. Batch Size:  128
5. Optimizer:  adagrad