clig - CLI Generator

A single module, pure python, Command Line Interface Generator.

OBS: currently under development.

Installation

pip install clig

User guide

clig is a single module, written in pure python, that wraps around the stdlib module argparse to generate command line interfaces through simple functions.

If you know how to use argparse, you may want to use clig.

Basic usage

Create or import some function and call clig.run() with it:

# example01.py
import clig

def printperson(name, title="Mister"):
    print(f"{title} {name}")

clig.run(printperson)

In general, the function arguments that have a "default" value are turned into optional flagged (--) command line arguments, while the "non default" will be positional arguments.

> python example01.py -h

usage: printperson [-h] [--title TITLE] name

positional arguments:
  name

options:
  -h, --help     show this help message and exit
  --title TITLE

The script can then be used in the same way as used with argparse:

> python example01.py John

Mister John

> python example01.py Maria --title Miss

Miss Maria

You can also pass arguments in code (like with the original parse_args() method)

>>> import clig
>>> def printperson(name, title="Mister"):
...     print(f"{title} {name}")
...
>>> clig.run(printperson, ["Isaac", "--title", "Sir"])
Sir Isaac

The clig.run() function also accepts other arguments to customize the interface

Helps

Arguments and command Helps are taken from the docstring when possible:

# example02.py
import clig

def greetings(name, greet="Hello"):
    """Description of the command: A greeting prompt!

    Args:
        name: The name to greet
        greet: The greeting used. Defaults to "Hello".
    """
    print(f"Greetings: {greet} {name}!")

clig.run(greetings)

> python example02.py --help

usage: greetings [-h] [--greet GREET] name

Description of the command: A greeting prompt!

positional arguments:
  name           The name to greet

options:
  -h, --help     show this help message and exit
  --greet GREET  The greeting used. Defaults to "Hello".

There is an internal list of docstring templates from which you can choose if the inferred docstring is not correct. It is also possible to specify your own custom docstring template.

Argument inference

Based on type annotations, some arguments can be inferred from the function signature to pass data to the original add_argument() method:

# example03.py
import clig

def recordperson(name: str, age: int, height: float):
    print(locals())

clig.run(recordperson)

The types in the annotation may be used in the add_argument() method as type keyword argument, when possible:

> python example03.py John 37 1.73

{'name': 'John', 'age': 37, 'height': 1.73}

And the type conversions are performed as usual

> python example03.py Mr John Doe

usage: recordperson [-h] name age height
recordperson: error: argument age: invalid int value: 'John'

Booleans

Booleans are transformed into arguments with action of kind "store_true" or "store_false" (depending on the default value).

# example04.py
import clig

def recordperson(name: str, employee: bool = False):
    print(locals())

clig.run(recordperson)

> python example04.py -h

usage: recordperson [-h] [--employee] name

positional arguments:
  name

options:
  -h, --help  show this help message and exit
  --employee

> python example04.py --employee Leo

{'name': 'Leo', 'employee': True}

> python example04.py Ana

{'name': 'Ana', 'employee': False}

Required booleans

If no default is given to the boolean, a required=True keyword argument is used in the add_argument() method and a BooleanOptionalAction is used as action keyword argument, adding support for a boolean complement action in the form --no-option:

# example05.py
import clig

def recordperson(name: str, employee: bool):
    print(locals())

clig.run(recordperson)

> python example05.py -h

usage: recordperson [-h] --employee | --no-employee name

positional arguments:
  name

options:
  -h, --help            show this help message and exit
  --employee, --no-employee

> python example05.py Ana

usage: recordperson [-h] --employee | --no-employee name
recordperson: error: the following arguments are required: --employee/--no-employee

Tuples, Lists and Sequences: nargs

The original nargs keyword argument associates a different number of command-line arguments with a single action. This is inferrend in types using tuple, list and Sequence.

Tuples

If the type is a tuple of specified length N, the argument automatically uses nargs=N.

# example06.py
import clig

def main(name: tuple[str, str]):
    print(locals())

clig.run(main)

> python example06.py -h

usage: main [-h] name name

positional arguments:
  name

options:
  -h, --help  show this help message and exit

> python example06.py rocky yoco

{'name': ('rocky', 'yoco')}

> python example06.py rocky

usage: main [-h] name name
main: error: the following arguments are required: name

The argument can be positional (required, as above) or optional (with a default).

# example07.py
import clig

def main(name: tuple[str, str, str] = ("john", "mary", "jean")):
    print(locals())

clig.run(main)

> python example07.py

{'name': ('john', 'mary', 'jean')}

> python example07.py --name yoco

usage: main [-h] [--name NAME NAME NAME]
main: error: argument --name: expected 3 arguments

> python example07.py --name yoco rocky sand

{'name': ('yoco', 'rocky', 'sand')}

List, Sequences and Tuples of any length

If the type is a generic Sequence, a list or a tuple of any length (i.e., tuple[<type>, ...]), it uses nargs="+" if it is required (non default value) or nargs="*" if it is not required (has a default value).

# example08.py
import clig

def main(names: list[str]):
    print(locals())

clig.run(main)

In this example, we have names using nargs="+"

> python example08.py -h

usage: main [-h] names [names ...]

positional arguments:
  names

options:
  -h, --help  show this help message and exit

> python example08.py chester philip

{'names': ['chester', 'philip']}

> python example08.py

usage: main [-h] names [names ...]
main: error: the following arguments are required: names

In the next example, we have names as optional argument, using nargs="*"

# example09.py
import clig

def main(names: list[str] | None = None):
    print(locals())

clig.run(main)

> python example09.py -h

usage: main [-h] [--names [NAMES ...]]

options:
  -h, --help           show this help message and exit
  --names [NAMES ...]

> python example09.py --names katy buba

{'names': ['katy', 'buba']}

> python example09.py

{'names': None}

Literals and Enums: choices

If the type is a Literal or a Enum the argument automatically uses choices.

# example10.py
from typing import Literal
import clig

def main(name: str, move: Literal["rock", "paper", "scissors"]):
    print(locals())

clig.run(main)

> python example10.py -h

usage: main [-h] name {rock,paper,scissors}

positional arguments:
  name
  {rock,paper,scissors}

options:
  -h, --help            show this help message and exit

As is expected in argparse, an error message will be displayed if the argument was not one of the acceptable values:

> python example10.py John knife

usage: main [-h] name {rock,paper,scissors}
main: error: argument move: invalid choice: 'knife' (choose from rock, paper, scissors)

> python example10.py Mary paper

{'name': 'Mary', 'move': 'paper'}

Passing Enums

In the command line, Enum should be passed by name, regardless of if it is a number Enum or ar string Enum

# example11.py
from enum import Enum, StrEnum
import clig

class Color(Enum):
    red = 1
    blue = 2
    yellow = 3

class Statistic(StrEnum):
    minimun = "minimun"
    mean = "mean"
    maximum = "maximum"

def main(color: Color, statistic: Statistic):
    print(locals())

clig.run(main)

> python example11.py -h

usage: main [-h] {red,blue,yellow} {minimun,mean,maximum}

positional arguments:
  {red,blue,yellow}
  {minimun,mean,maximum}

options:
  -h, --help            show this help message and exit

It is correctly passed to the function

> python example11.py red mean

{'color': <Color.red: 1>, 'statistic': <Statistic.mean: 'mean'>}

> python example11.py green

usage: main [-h] {red,blue,yellow} {minimun,mean,maximum}
main: error: argument color: invalid choice: 'green' (choose from red, blue, yellow)

Literal with Enum

You can even mix Enum and Literal, following the Literal specification

# example12.py
from typing import Literal
from enum import Enum
import clig

class Color(Enum):
    red = 1
    blue = 2
    yellow = 3

def main(color: Literal[Color.red, "green", "black"]):
    print(locals())

clig.run(main)

> python example12.py red

{'color': <Color.red: 1>}

> python example12.py green

{'color': 'green'}

Variadic arguments (*args and **kwargs): Partial parsing

When the function has variadic arguments in the form *args or **kwargs, the parse_known_args() method will be used internally to gather unspecified arguments:

>>> import clig
>>> def variadics(foo: str, *args, **kwargs):
...     print(locals())
...
>>> clig.run(variadics, "bar badger BAR spam --name adam --title mister".split())
{'foo': 'bar', 'args': ('badger', 'BAR', 'spam'), 'kwargs': {'name': 'adam', 'title': 'mister'}}

*args

For arbitrary arguments in the form *args, the unspecified arguments will be wrapped up in a tuple of strings, by default. If there is a type annotation, the conversion is made in the whole tuple:

>>> import clig
>>> def variadicstyped(number: float, *integers: int):
...     print(locals())
...
>>> clig.run(variadicstyped, ["36.7", "1", "2", "3", "4", "5"])
{'number': 36.7, 'integers': (1, 2, 3, 4, 5)}

**kwargs

For arbitrary keyword arguments in the form **kwargs, the unspecified arguments will be wrapped up in a dictionary of strings by default. The keys of the dictionary are the names used with the option delimiter in the command line (usually - or --). If there are more than one value for each option, they are gathered in a list:

# example13.py
import clig

def foobar(name: str, **kwargs):
    print(locals())

clig.run(foobar)

> python example13.py joseph --nickname joe --uncles jack jean adam

{'name': 'joseph', 'kwargs': {'nickname': 'joe', 'uncles': ['jack', 'jean', 'adam']}}

If there is a type annotation, the conversion is made in all elements of the dictionary

# example14.py
import clig

def foobartyped(name: str, **intergers: int):
    print(locals())

clig.run(foobartyped)

> python example14.py joseph --age 23 --numbers 25 27 30

{'name': 'joseph', 'intergers': {'age': 23, 'numbers': [25, 27, 30]}}

> python example14.py joseph --age 23 --numbers jack jean adam

ValueError: invalid literal for int() with base 10: 'jack'

Error when passing flagged arguments to *args

The flag delimiters (usually - or --, which can be changed) are always interpreted as prefix for keyword arguments, raising the correct error when not allowed:

# example15.py
import clig

def bazham(name: str, *uncles: str):
    print(locals())

clig.run(bazham)

> python example15.py joseph jack john

{'name': 'joseph', 'uncles': ('jack', 'john')}

> python example15.py joseph --uncles jack john

TypeError: bazham() got an unexpected keyword argument 'uncles'

Argument specification

In some complex cases supported by argparse, the arguments may not be completely inferred by clig.run() on the function signature.

In theses cases, you can directly specificy the arguments parameters using the Annotated typing (or its clig's alias Arg) with its "metadata" created with the data() function.

The data() function accepts all possible arguments of the original add_argument() method:

name or flags

The name_or_flags parameter can be used to define additional flags for the arguments, like -f or --foo:

# example16.py
from clig import Arg, data, run

def main(foobar: Arg[str, data("-f", "--foo")] = "baz"):
    print(locals())

run(main)

> python example16.py -h

usage: main [-h] [-f FOOBAR]

options:
  -h, --help            show this help message and exit
  -f FOOBAR, --foo FOOBAR

name or flags can also be used to turn a positional argument (without default) into a required flagged argument (a required option):

# example17.py
from clig import Arg, data, run

def main(foo: Arg[str, data("-f")]):
    print(locals())

run(main)

> python example17.py -h

usage: main [-h] -f FOO

options:
  -h, --help         show this help message and exit
  -f FOO, --foo FOO

> python example17.py

usage: main [-h] -f FOO
main: error: the following arguments are required: -f/--foo

Note

As you can see above, clig tries to create a long flag (--) for the argument when only short flags (-) are defined, but not when long flags are already defined. However, this behavior can be disabled.

Some options for the name or flags parameter can also be set in the run() function.

nargs

Other cases of nargs can be specified in the data() function.

The next example uses an optional argument with nargs="?" and const, which brings 3 different behaviors for the optional argument:

• value passed
• value not passed (sets default value)
• option passed without value (sets const value):

>>> from clig import Arg, data, run
...
>>> def main(foo: Arg[str, data(nargs="?", const="c")] = "d"):
...     print(locals())
...
>>> run(main, ["--foo", "YY"])
{'foo': 'YY'}
>>> run(main, [])
{'foo': 'd'}
>>> run(main, ["--foo"])
{'foo': 'c'}

The next example makes optional a positional argument (not flagged), by using nargs="?" and default (which would default to None):

>>> from clig import Arg, data, run
>>> def main(foo: Arg[str, data(nargs="?", default="d")]):
...     print(locals())
...
>>> run(main, ["YY"])
{'foo': 'YY'}
>>> run(main, [])
{'foo': 'd'}

action

Other options for the action parameter can also be used in the data() function:

>>> from clig import Arg, data, run
>>> def append(foo: Arg[list[str], data(action="append")] = ["0"]):
...     print(locals())
...
>>> def append_const(bar: Arg[list[int], data(action="append_const", const=42)] = [42]):
...     print(locals())
...
>>> def extend(baz: Arg[list[float], data(action="extend")] = [0]):
...     print(locals())
...
>>> def count(ham: Arg[int, data(action="count")] = 0):
...     print(locals())
...
>>> run(append, "--foo 1 --foo 2".split())
{'foo': ['0', '1', '2']}
...
>>> run(append_const, "--bar --bar --bar --bar".split())
{'bar': [42, 42, 42, 42, 42]}
...
>>> run(extend, "--baz 25 --baz 50 65 75".split())
{'baz': [0, 25.0, 50.0, 65.0, 75.0]}
...
>>> run(count, "--ham --ham --ham".split())
{'ham': 3}

metavar

The parameter metavar is used to set alternative names in help messages to refer to arguments. By default, they would be referend as just the argument name, if positional, and the argument name uppercased, if optional.

# example18.py
from clig import Arg, data, run

def main(ham: Arg[str, data(metavar="YYY")], foo: Arg[str, data("-f", metavar="<foobar>")]):
    print(locals())

run(main)

> python example18.py -h

usage: main [-h] -f <foobar> YYY

positional arguments:
  YYY

options:
  -h, --help            show this help message and exit
  -f <foobar>, --foo <foobar>

Some options for the metavar argument can also be set in the run() function.

help

It is more convenient to specify helps for arguments in the docstring.

However, you can define helps using the data() function in the same way as in the original method add_argument(). Helps passed in the data() function takes precedence.

# example19.py
from clig import Arg, data, run

def mycommand(number: Arg[int, data(help="a different help for the number")]):
    """Description of the command

    Args:
        number: a number to compute
    """
    pass

run(mycommand)

> python example19.py -h

usage: mycommand [-h] number

Description of the command

positional arguments:
  number      a different help for the number

options:
  -h, --help  show this help message and exit

Some options for the help argument can also be set in the run() function.

Argument groups

The argparse module has the features of argument groups and mutually exclusive argument groups. These features can be used in clig with 2 additional classes: ArgumentGroup and MutuallyExclusiveGroup.

The object created with these classes can be used in the group parameter of the data() function.

Each class accepts all the parameters of the original methods add_argument_group() and add_mutually_exclusive_group().

# example20.py
from clig import Arg, data, run, ArgumentGroup

g = ArgumentGroup(title="Group of arguments", description="This is my group of arguments")

def main(foo: Arg[str, data(group=g)], bar: Arg[int, data(group=g)] = 42):
    print(locals())

run(main)

> python example20.py -h

usage: main [-h] [--bar BAR] foo

options:
  -h, --help  show this help message and exit

Group of arguments:
  This is my group of arguments

  foo
  --bar BAR

Remember that mutually exclusive arguments must be optional (either by using a flag in the data function, or by setting a deafult value):

# example21.py
from clig import Arg, data, run, MutuallyExclusiveGroup

g = MutuallyExclusiveGroup()

def main(foo: Arg[str, data("-f", group=g)], bar: Arg[int, data(group=g)] = 42):
    print(locals())

run(main)

> python example21.py --foo rocky --bar 23

usage: main [-h] [-f FOO | --bar BAR]
main: error: argument --bar: not allowed with argument -f/--foo

Required mutually exclusive group

A required argument is accepted by the MutuallyExclusiveGroup in the same way it is done with the original method add_mutually_exclusive_group() (to indicate that at least one of the mutually exclusive arguments is required):

# example22.py
from clig import Arg, data, run, MutuallyExclusiveGroup

g = MutuallyExclusiveGroup(required=True)

def main(foo: Arg[str, data(group=g)] = "baz", bar: Arg[int, data(group=g)] = 42):
    print(locals())

run(main)

> python example22.py -h

usage: main [-h] (--foo FOO | --bar BAR)

options:
  -h, --help  show this help message and exit
  --foo FOO
  --bar BAR

> python example22.py

usage: main [-h] (--foo FOO | --bar BAR)
main: error: one of the arguments --foo --bar is required

Mutually exclusive group added to an argument group

The MutuallyExclusiveGroup constructor class also accepts an additional argument_group parameter, because a mutually exclusive group can be added to an argument group.

# example23.py
from clig import Arg, data, run, ArgumentGroup, MutuallyExclusiveGroup

ag = ArgumentGroup(title="Group of arguments", description="This is my group")
meg = MutuallyExclusiveGroup(argument_group=ag)

def main(
    foo: Arg[str, data(group=meg)] = "baz",
    bar: Arg[int, data(group=meg)] = 42,
):
    print(locals())

run(main)

> python example23.py -h

usage: main [-h] [--foo FOO | --bar BAR]

options:
  -h, --help  show this help message and exit

Group of arguments:
  This is my group

  --foo FOO
  --bar BAR

However, you can define just the MutuallyExclusiveGroup object passing the parameters of ArgumentGroup to the constructor of the former class, which supports them:

# example24.py
from clig import Arg, data, run, MutuallyExclusiveGroup

g = MutuallyExclusiveGroup(
    title="Group of arguments",
    description="This is my exclusive group of arguments",
)

def main(
    foo: Arg[str, data("-f", group=g)],
    bar: Arg[int, data("-b", group=g)],
):
    print(locals())

run(main)

> python example24.py -h

usage: main [-h] [-f FOO | -b BAR]

options:
  -h, --help         show this help message and exit

Group of arguments:
  This is my exclusive group of arguments

  -f FOO, --foo FOO
  -b BAR, --bar BAR

Using the walrus operator (:=)

You can do argument group definition all in one single line (in the function declaration) by using the walrus operator (:=):

# example25.py
from clig import Arg, data, run, MutuallyExclusiveGroup

def main(
    foo: Arg[str, data(group=(g := MutuallyExclusiveGroup(title="My group")))] = "baz",
    bar: Arg[int, data(group=g)] = 42,
):
    print(locals())

run(main)

> python example25.py -h

usage: main [-h] [--foo FOO | --bar BAR]

options:
  -h, --help  show this help message and exit

My group:
  --foo FOO
  --bar BAR