config-argument-parser
A package to help automatically create command-line interface from configuration or code.
It contains three modules CAP🧢(ConfigArgumentParser
), TAP🚰(TypeArgumentParser
), and GAP🕳️(GlobalArgumentParser
).
Read the documentation here.
Motivation
Configuration files are highly readable and useful for specifying options, but sometimes they are not convenient as command-line interface. However, it requires writing a lot of code to produce a CLI. This package automates the building process, by utilizing the Python standard libraries configparser
and argparse
.
The design is to minimize the changes to your original scripts, so as to facilitate maintenance.
Features
- Only few extra lines are needed to build a CLI from an existing script.
- The comments are parsed as help messages. (Most libraries do not preserve the comments.)
- Consistent format between configuration and script provides ease of use.
Usage
Case 1: create CLI from an object
If you use class to store arguments, such as the script example.py
below.
import configargparser
class Args:
a_string = "abc"
a_float = 1.23
a_boolean = False
an_integer = 0
args = Args()
parser = configargparser.ConfigArgumentParser()
parser.parse_obj(args, shorts="sfb")
print(args.a_string)
print(args.a_float)
print(args.a_boolean)
print(args.an_integer)
In fact, only the snippet below is added to the original script. Moreover, removing this minimal modification does not affect the original script. shorts
is optional. If provided, add short options for the first few arguments in order.
import configargparser
parser = configargparser.ConfigArgumentParser()
parser.parse_obj(args)
Default arguments are defined as class attributes, and parsed arguments are stored as instance attributes. The good is that auto-completion can be triggered in editors.
Show help, python example.py -h
:
$ python example.py -h
usage: example.py [-h] [-s A_STRING] [-f A_FLOAT] [-b] [--an-integer AN_INTEGER]
options:
-h, --help show this help message and exit
-s A_STRING, --a-string A_STRING
Help message of the first argument. Help is optional. (default: abc)
-f A_FLOAT, --a-float A_FLOAT
a_float (default: 1.23)
-b, --a-boolean Help can span multiple lines. This is another line. (default: False)
--an-integer AN_INTEGER
an_integer (default: 0)
Run with options, for example, python example.py -b -f 1
:
$ python example.py -b -f 1
abc
1.0
True
0
Note that the values are changed.
For the best practice, see Case 4.
Case 2: create CLI from configuration
If you use configuration file, create an example script example.py
:
import configargparser
parser = configargparser.ConfigArgumentParser()
parser.read("config.ini")
parser.parse_args(shorts="sfb")
print("Configs:", parser.defaults)
print("Args: ", parser.args)
Create a configuration file config.ini
in the same directory:
[DEFAULT]
a_string = 'abc'
a_float = 1.23
a_boolean = False
an_integer = 0
Regular run, python example.py
:
$ python example.py
Configs: {'a_string': 'abc', 'a_float': 1.23, 'a_boolean': False, 'an_integer': 0}
Args: {'a_string': 'abc', 'a_float': 1.23, 'a_boolean': False, 'an_integer': 0}
Run with options, such as python example.py -b -f 1
:
$ python example.py -b -f 1
Configs: {'a_string': 'abc', 'a_float': 1.23, 'a_boolean': False, 'an_integer': 0}
Args: {'a_string': 'abc', 'a_float': 1.0, 'a_boolean': True, 'an_integer': 0}
Case 3: create CLI from global variables
If you use global variables, define the variables at top of file as the script example.py
below:
a_string = "abc"
a_float = 1.23
a_boolean = False
an_integer = 0
import configargparser
parser = configargparser.ConfigArgumentParser()
parser.read_py("example.py")
parser.parse_args(shorts="sfb")
globals().update(parser.args)
print(a_string)
print(a_float)
print(a_boolean)
print(an_integer)
Use it as in case 1. For example, python example.py -b -f 1
:
$ python example.py -b -f 1
abc
1.0
True
0
Case 4: create CLI from a dataclass object (preferred)
Suppose you have a script example.py
below, which uses a dataclass
object to store arguments:
from dataclasses import dataclass
@dataclass
class Args:
a_string: str = "abc"
a_float: float = 1.23
a_boolean: bool = False
an_integer: int = 0
args = Args()
Add these lines to the script to create CLI:
import configargparser
parser = configargparser.TypeArgumentParser()
parser.parse_obj(args, shorts="sfb")
print(args)
Use it as in case 1. For example, python example.py -b -f 1
to change the values:
$ python example.py -b -f 1
Args(a_string='abc', a_float=1.0, a_boolean=True, an_integer=0)
This requires less code than case 3, but the comments are not parsed, as the script example.py
below:
a_string = "abc"
a_float = 1.23
a_boolean = False
an_integer = 0
import configargparser
parser = configargparser.GlobalArgumentParser()
parser.parse_globals(shorts="sfb")
print(a_string)
print(a_float)
print(a_boolean)
print(an_integer)
Use it as in case 1. For example, python example.py -b -f 1
:
$ python example.py -b -f 1
abc
1.0
True
0
Installation
Install from PyPI:
python -m pip install --upgrade pip
pip install config-argument-parser
Alternatively, install from source:
git clone https://github.com/yuanx749/config-argument-parser.git
cd config-argument-parser
then install in development mode:
git checkout main
python -m pip install --upgrade pip
pip install -e .
or:
git checkout dev
python -m pip install --upgrade pip
pip install -e .[dev]
pre-commit install
Uninstall:
pip uninstall config-argument-parser
Notes
This package uses Semantic Versioning.