argument-parsing 1.0.1

A zero dependency argument parser written in python.

Introduction

A zero dependency, single file string argument parser written in python.
You can define commands by creating simple python dictionaries, parse strings for the defined fields, and receive a structured result. The argument parser handles some basic python types, typecasting, missing values, default values, nested lists and tuples, error reporting, and any combination of the above.

Installation

This package is available on PyPI:

pip install argument-parsing

Another way to install this module is to do a local pip install.

git clone [e-mail]:flpeters/argument_parsing.git
cd argument_parsing
pip install -e .

Documentation

This entire module provides exactly two functions.
A single function is provided for processing your definition of a "command" as well as the string "arguments" you want to parse, into structured data.

def parse_arguments(command:dict, arguments:str) -> (bool, dict, dict):

A second function is provided for changing the settings of how errors and warnings are reported.

def set_report_options(report_error:bool=True, report_warning:bool=True,
                       raise_error:bool=False, raise_warning:bool=False,
                       silent:bool=False) -> None:

The command

is a simple dictionary. The keys are the keywords which will be parsed out from the arguments string. Each keyword maps to either a datatype, or a default value which will be used to infere the datatype.
Example command:

{
    'name'   : str,
    'weather': 'sunny',
    'celsius': float,
    'age'    : int,
    'thirsty': bool,
    'tired'  : bool
}

The arguments

are passed as a whitespacespace separated string. Keywords start with a hyphen-minus ('-'), and depending on the datatype of the keyword, are followed by zero, one, multiple, or optional values.
Example arguments:

'-name bob -age 99 -celsius 30.5 -thirsty'

The return value

is a three-tuple of "success", "result", and "is_set".

• success is a single bool, which tells you whether or not parsing was successful. If this is False, the other two arguments are not guaranteed to be valid. There will be an error message with details on why parsing failed.
• result is a dict with exactly the same keys as the input command. The values will be set to what was parsed from the arguments string. In cases where success is False, this might only be partially filled out, so success should always be checked.
• is_set is a dict, which also contains exactly the same keys as command. The values are all either True or False depending on if the keyword was present in the arguments string. In cases where a default value is given in command, only if the default was overwritten will the value be True. This holds even for bool arguments.

return values given the example inputs above:

success = True
result  = {'name': 'bob',
           'weather': 'sunny',
           'celsius': 30.5,
           'age': 99,
           'thirsty': True,
           'tired': False}
is_set  = {'name': True,
           'weather': False,
           'celsius': True,
           'age': True,
           'thirsty': True,
           'tired': False}

Supported datatypes

A primitive type is a simple python datatype. Depending on which type is used when defining the keywords of a command, the parsing rules will change.

---

The following primitive types are supported:

string

A keyword of type str always requires exactly one value.

command   = {'weather': str}
arguments = '-weather sunny'
success, result, is_set = parse_arguments(command, arguments)
result    = {'weather': 'sunny'}
is_set    = {'weather': True}

As long as this value doesn't contain any whitespace, the characters it is made up of don't matter. Any punctuation or numbers will just be treated as part of the string.

arguments = '-weather 1234'
...
result    = {'weather': '1234'}
is_set    = {'weather': True}

arguments = '-weather -cloudy'
...
result    = {'weather': '-cloudy'}
is_set    = {'weather': True}

When a default value is set, the keyword becomes optional. Should the keyword not be part of the arguments, then the default value will be placed in the results instead.

command   = {'weather': 'cloudy'}
arguments = ''
...
result    = {'weather': 'cloudy'}
is_set    = {'weather': False}

boolean

A keyword of type bool requires no value. If the keyword appears in the arguments, the result is automatically set to True. Setting the datatype of a keyword to bool in command is the same as specifying the default value False. This also means that the result of a bool keyword with a default value of True can never be set to False.

command   = {'rainy': bool}
arguments = '-rainy'
...
result    = {'rainy': True}
is_set    = {'rainy': True}

command   = {'rainy': bool}
arguments = ''
...
result    = {'rainy': False}
is_set    = {'rainy': False}

command   = {'rainy': True}
arguments = ''
...
result    = {'rainy': True}
is_set    = {'rainy': False}

integer

A keyword of type int always requires exactly one value.

command   = {'meters': int}
arguments = '-meters 26'
...
result    = {'meters': 26}
is_set    = {'meters': True}

The value which comes after the keyword will first be cast to float, and then to int. This is partly due to how python works, but also to check for a remainder in case the provided value was actually a float. This also means that a sign in front of the number will be respected.

arguments = '-meters -15'
...
result    = {'meters': -15}

arguments = '-meters +15.0'
...
result    = {'meters': 15}

arguments = '-meters 15.4'
...
result    = {'meters': 15}
# This will result in a warning that the remainder of 0.4 has been cut off.

The default value behaviour is the same as for strings.

command   = {'meters': 33}
arguments = ''
...
result    = {'meters': 33}
is_set    = {'meters': False}

float

A keyword of type float always requires exactly one value. The value which comes after the keyword will simply be cast to a python float. What is and what isn't a valid python float can be suprising, so you should check the casting rules beforehand.

command   = {'celsius': float}
arguments = '-celsius 23.6'
...
result    = {'celsius': 23.6}
is_set    = {'celsius': True}

command   = {'celsius': float}
arguments = '-celsius -10.0'
...
result    = {'celsius': -10.0}

command   = {'celsius': float}
arguments = '-celsius inf'
...
result    = {'celsius': inf}

The default value behaviour is the same as for strings.

command   = {'celsius': float('nan')}
arguments = ''
...
result    = {'celsius': nan}
is_set    = {'celsius': False}

---

A composite type is a list or tuple containing one or more values of potentially multiple datatypes. Since list or tuple have the same semantics and can be used interchangeably, they will be referred to as 'array' for the purpose of this documentation.

The following composite types are supported:

un-bounded array

Specifying only the type list or tuple, will result in an 'unbounded array' of that type, meaning that all values following the keyword will be added to the array, until either the end of the arguments string is reached, or a value starts with a hyphen-minus ('-'), which denotes the start of the next argument. All values of this unbounded array will be of type str. This kind of argument should be used with caution because, for instance, negative numbers will be treated as the start of a new argument due to them starting with a hyphen-minus sign.

command = {
    'unbounded_list' : list,
    'unbounded_tuple': tuple,
}
arguments = '-unbounded_list 1 2 3 4 5 a b c -unbounded_tuple 1.0 + 1.0 = 2.0'
...
result = {'unbounded_list' : ['1', '2', '3', '4', '5', 'a', 'b', 'c'],
          'unbounded_tuple': ('1.0', '+', '1.0', '=', '2.0')}
is_set = {'unbounded_list': True, 'unbounded_tuple': True}

fixed array

By specifying an array containing the types, default values, and ordering you want the values to have, you can define a fixed array. Their construction can get arbitrarily complex, mixing and matching any supported primitive type and fixed arrays you want. The only thing not allowed, is using an unbounded array inside a fixed array. All values will be cast to the corresponding type using all the same semantics as if they were single values (see above).
Please note that while default values can be used to define the array, you currently need to specify all of the values, even if the array has default values at the end and they could technically be omitted. Default values in the middle of the array also currently don't have a mechanism for not setting them. This also means that since the value of a bool type can't be decided based on presence or absence, a value has to be given. We recognize a value of either 'True', 'False', or one interpretable as a float, which will then be cast to a bool. This means that e.g. '0.0' will result in False, and '123' will result in 'True' (careful, check the casting rules first).

command = {
    'arg1': [int]*5,
    'arg2': (3.14, 'pi', bool),
    'arg3': (bool, str, 123)*2,
    'arg4': [[0]*3, [1]*3, [str]*3],
    'arg5': [str, int, bool, True, [1, '2', 3, bool], (2.1, float)]
}
arguments = '-arg1 4 5 6 7 8 \
             -arg2 6.28 tau True \
             -arg3 False helloworld 123 True HelloWorld 321 \
             -arg4 0 0 0 1 1 1 2 2 2 \
             -arg5 1 1 1 1 1 1 1 1 1 1'
...
result = {
    'arg1': [4, 5, 6, 7, 8],
    'arg2': (6.28, 'tau', True),
    'arg3': (False, 'helloworld', 123, True, 'HelloWorld', 321),
    'arg4': [[0, 0, 0], [1, 1, 1], ['2', '2', '2']],
    'arg5': ['1', 1, True, True, [1, '1', 1, True], (1.0, 1.0)]}
is_set = {'arg1': True, 'arg2': True, 'arg3': True, 'arg4': True, 'arg5': True}

Attribution

This argument parser is largely inspired by and based upon, but in no way affiliated with, the work by Jonathan Blow as seen in these two videos on his youtube channel.

Jonathan Blow - Livestream: Metaprogramming Use Case: Command-Line Argument Parsing - Part 1
Jonathan Blow - Livestream: Metaprogramming Use Case: Command-Line Argument Parsing - Part 2