3.4. Star Arguments

• Unpack and Arbitrary Number of Parameters and Arguments

• * is used for positional arguments

• ** is used for keyword arguments

• echo(*data) - unpacks from sequence (tuple, list, set, etc)

• echo(**data) - unpacks from mapping (dict, etc)

• echo(*data1, **data2) - unpacks from sequence and mappings

3.4.1. Recap

• Argument - value passed to the function

• Argument can be: positional or keyword

• Positional arguments - resolved by position, order is important, must be at the left side

• Keyword arguments - resolved by name, order is not important, must be on the right side

• After first keyword argument, all following arguments must also be keyword

>>> def echo(a, b):
...     ...

Positional arguments (order is important):

>>> echo(1, 2)
>>> echo(2, 1)

Keyword arguments (order is not important):

>>> echo(a=1, b=2)
>>> echo(b=2, a=1)

Positional and keyword arguments:

>>> echo(1, b=2)

Positional arguments must be at the left side:

>>> echo(a=1, 2)
Traceback (most recent call last):
SyntaxError: positional argument follows keyword argument

3.4.2. Positional Arguments

• echo(*data) - unpacks from sequence (tuple, list, set, etc)

• * is used for positional arguments

• There is no convention, so you can use any name, for example *data

>>> def echo(a, b, c, d):
...     print(f'{a=}, {b=}, {c=}, {d=}')
...
>>> data = (1, 2, 3, 4)

Without star unpacking:

>>> echo(data[0], data[1], data[2], data[3])
a=1, b=2, c=3, d=4

With start unpacking:

>>> echo(*data)
a=1, b=2, c=3, d=4

3.4.3. Keyword Arguments

• echo(**data) - unpacks from mapping (dict, etc)

• ** is used for keyword arguments

• There is no convention, so you can use any name, for example **data

Keyword arguments passed directly:

>>> def echo(a, b, c, d):
...     print(f'{a=}, {b=}, {c=}, {d=}')
...
>>> data = {'a':1, 'b':2, 'c':3, 'd':4}

Without star unpacking:

>>> echo(a=data['a'], b=data['b'], c=data['c'], d=data['d'])
a=1, b=2, c=3, d=4

With start unpacking:

>>> echo(**data)
a=1, b=2, c=3, d=4

3.4.4. Positional and Keyword Arguments

• echo(*data1, **data2) - unpacks from sequence and mappings

• * is used for positional arguments

• ** is used for keyword arguments

• There is no convention, so you can use any name, for example *data1

• There is no convention, so you can use any name, for example **data2

>>> def echo(a, b, c, d):
...     print(f'{a=}, {b=}, {c=}, {d=}')
...
>>> data1 = (1, 2)
>>> data2 = {'c':3, 'd':4}

Without star unpacking:

>>> echo(data1[0], data1[1], c=data2['c'], d=data2['d'])
a=1, b=2, c=3, d=4

With star unpacking:

>>> echo(*data1, **data2)
a=1, b=2, c=3, d=4

3.4.5. Merge Kwargs

• echo(**data1, **data2)

>>> def echo(a, b, c, d):
...     return locals()
>>>
>>> data1 = {'a':1, 'b':2}
>>> data2 = {'c':3, 'd':4}

With star unpacking:

>>> echo(**data1, **data2)
{'a': 1, 'b': 2, 'c': 3, 'd': 4}

3.4.6. Merge Dicts

• dict(**data1, **data2) - old way

• {**data1, **data2} - old way

• data1 | data2 - since Python 3.9 preferred way

>>> data1 = {'a':1, 'b':2}
>>> data2 = {'c':3, 'd':4}

Before Python 3.9 merging dicts was done with dict or {}:

>>> dict(**data1, **data2)
{'a': 1, 'b': 2, 'c': 3, 'd': 4}

>>> {**data1, **data2}
{'a': 1, 'b': 2, 'c': 3, 'd': 4}

Since Python 3.9 there is a dedicated operator for merging dicts:

>>> data1 | data2
{'a': 1, 'b': 2, 'c': 3, 'd': 4}

3.4.7. Create Objects

• One object from sequence

• One object from mapping

• Many objects from sequence of sequences

• Many objects from sequence of mappings

>>> class User:
...     def __init__(self, firstname, lastname):
...         self.firstname = firstname
...         self.lastname = lastname
...
...     def __repr__(self):
...         return f"User(firstname='{self.firstname}', lastname='{self.lastname}')"

One object from sequence:

>>> data = ('Mark', 'Watney')
>>>
>>> result = User(*data)
>>> result
User(firstname='Mark', lastname='Watney')

One object from mapping:

>>> data = {'firstname': 'Mark', 'lastname': 'Watney'}
>>>
>>> result = User(**data)
>>> result
User(firstname='Mark', lastname='Watney')

Many objects from sequence of sequences:

>>> data = [
...     ('Mark', 'Watney'),
...     ('Melissa', 'Lewis'),
...     ('Rick', 'Martinez'),
... ]
>>>
>>> result = [User(*row) for row in data]
>>> result  
[User(firstname='Mark', lastname='Watney'),
 User(firstname='Melissa', lastname='Lewis'),
 User(firstname='Rick', lastname='Martinez')]

Many objects from sequence of mappings:

>>> data = [
...     {'firstname': 'Mark', 'lastname': 'Watney'},
...     {'firstname': 'Melissa', 'lastname': 'Lewis'},
...     {'firstname': 'Rick', 'lastname': 'Martinez'},
... ]
>>>
>>> result = [User(**row) for row in data]
>>> result  
[User(firstname='Mark', lastname='Watney'),
 User(firstname='Melissa', lastname='Lewis'),
 User(firstname='Rick', lastname='Martinez')]

3.4.8. Recap

• * is used for positional arguments

• ** is used for keyword arguments

• echo(*data) - unpacks from sequence (tuple, list, set, etc)

• echo(**data) - unpacks from mapping (dict, etc)

• echo(*data1, **data2) - unpacks from sequence and mappings

• Create one object from sequence

• Create one object from mapping

• Create many objects from sequence of sequences

• Create many objects from sequence of mappings

• Old way to merge dicts dict(**data1, **data2) or {**data1, **data2}

• Since Python 3.9 merge dicts with: data1 | data2

3.4.9. Use Case - 1

Calling a function which has similar parameters. Passing configuration to the function, which sets parameters from the config:

>>> def draw_line(x, y, color, type, width, markers):
...     ...

>>> draw_line(x=1, y=2, color='red', type='dashed', width='2px', markers='disc')
>>> draw_line(x=3, y=4, color='red', type='dashed', width='2px', markers='disc')
>>> draw_line(x=5, y=6, color='red', type='dashed', width='2px', markers='disc')

>>> style = {'color': 'red',
...          'type': 'dashed',
...          'width': '2px',
...          'markers': 'disc'}
>>>
>>> draw_line(x=1, y=2, **style)
>>> draw_line(x=3, y=4, **style)
>>> draw_line(x=5, y=6, **style)

3.4.10. Use Case - 2

>>> def print_coordinates(x, y, z):
...     print(f'{x=}, {y=}, {z=}')

Passing sequence to the function:

>>> point = [1, 2, 3]
>>>
>>> print_coordinates(point[0], point[1], point[2])
x=1, y=2, z=3
>>>
>>> print_coordinates(*point)
x=1, y=2, z=3

Passing mapping to the function:

>>> point = {'x': 1, 'y': 2, 'z': 3}
>>>
>>> print_coordinates(x=point['x'], y=point['y'], z=point['z'])
x=1, y=2, z=3
>>>
>>> print_coordinates(**point)
x=1, y=2, z=3
>>>
>>> print_coordinates(*point.values())
x=1, y=2, z=3

Passing sequence and mapping to the function:

>>> point2d = (1, 2)
>>> point3d = {'z': 3}
>>>
>>> print_coordinates(*point2d, **point3d)
x=1, y=2, z=3

3.4.11. Use Case - 3

>>> def database_connect(host, port, username, password, database):
...     ...

After reading config from file we have a dict:

>>> CONFIG = {
...     'host': 'example.com',
...     'port': 5432,
...     'username': 'myusername',
...     'password': 'mypassword',
...     'database': 'mydatabase'}

Database connection configuration read from config file:

>>> connection = database_connect(
...     host=CONFIG['host'],
...     port=CONFIG['port'],
...     username=CONFIG['username'],
...     password=CONFIG['password'],
...     database=CONFIG['database'])

Or:

>>> connection = database_connect(**CONFIG)

3.4.12. Use Case - 4

>>> from dataclasses import dataclass
>>>
>>>
>>> @dataclass
... class Point:
...     x: int
...     y: int
...     z: int = 0
>>>
>>>
>>> MOVEMENT = [(0, 0),
...             (1, 0),
...             (2, 1, 1),
...             (3, 2),
...             (3, 3, -1),
...             (2, 3),
... ]

>>> movement = [Point(x,y) for x,y in MOVEMENT]
Traceback (most recent call last):
ValueError: too many values to unpack (expected 2)

>>> movement = [Point(*coordinates) for coordinates in MOVEMENT]
>>>
>>> movement  
[Point(x=0, y=0, z=0),
 Point(x=1, y=0, z=0),
 Point(x=2, y=1, z=1),
 Point(x=3, y=2, z=0),
 Point(x=3, y=3, z=-1),
 Point(x=2, y=3, z=0)]

3.4.13. Use Case - 4

>>> from dataclasses import dataclass
>>>
>>>
>>> @dataclass
... class Iris:
...     sepal_length: float
...     sepal_width: float
...     petal_length: float
...     petal_width: float
...     species: str
>>>
>>>
>>> DATA = [
...     (5.8, 2.7, 5.1, 1.9, 'virginica'),
...     (5.1, 3.5, 1.4, 0.2, 'setosa'),
...     (5.7, 2.8, 4.1, 1.3, 'versicolor'),
...     (6.3, 2.9, 5.6, 1.8, 'virginica'),
...     (6.4, 3.2, 4.5, 1.5, 'versicolor'),
...     (4.7, 3.2, 1.3, 0.2, 'setosa'),
... ]
>>>
>>>
>>> result = [Iris(*row) for row in DATA]
>>>
>>> print(result)  
[Iris(sepal_length=5.8, sepal_width=2.7, petal_length=5.1, petal_width=1.9, species='virginica'),
 Iris(sepal_length=5.1, sepal_width=3.5, petal_length=1.4, petal_width=0.2, species='setosa'),
 Iris(sepal_length=5.7, sepal_width=2.8, petal_length=4.1, petal_width=1.3, species='versicolor'),
 Iris(sepal_length=6.3, sepal_width=2.9, petal_length=5.6, petal_width=1.8, species='virginica'),
 Iris(sepal_length=6.4, sepal_width=3.2, petal_length=4.5, petal_width=1.5, species='versicolor'),
 Iris(sepal_length=4.7, sepal_width=3.2, petal_length=1.3, petal_width=0.2, species='setosa')]

3.4.14. Use Case - 5

>>> from dataclasses import dataclass
>>>
>>>
>>> @dataclass
... class Iris:
...     sepal_length: float
...     sepal_width: float
...     petal_length: float
...     petal_width: float
...     species: str
>>>
>>>
>>> DATA = [
...     {"sepal_length":5.8,"sepal_width":2.7,"petal_length":5.1,"petal_width":1.9,"species":"virginica"},
...     {"sepal_length":5.1,"sepal_width":3.5,"petal_length":1.4,"petal_width":0.2,"species":"setosa"},
...     {"sepal_length":5.7,"sepal_width":2.8,"petal_length":4.1,"petal_width":1.3,"species":"versicolor"},
...     {"sepal_length":6.3,"sepal_width":2.9,"petal_length":5.6,"petal_width":1.8,"species":"virginica"},
...     {"sepal_length":6.4,"sepal_width":3.2,"petal_length":4.5,"petal_width":1.5,"species":"versicolor"},
...     {"sepal_length":4.7,"sepal_width":3.2,"petal_length":1.3,"petal_width":0.2,"species":"setosa"},
... ]
>>>
>>>
>>> result = [Iris(**row) for row in DATA]
>>>
>>> print(result)  
[Iris(sepal_length=5.8, sepal_width=2.7, petal_length=5.1, petal_width=1.9, species='virginica'),
 Iris(sepal_length=5.1, sepal_width=3.5, petal_length=1.4, petal_width=0.2, species='setosa'),
 Iris(sepal_length=5.7, sepal_width=2.8, petal_length=4.1, petal_width=1.3, species='versicolor'),
 Iris(sepal_length=6.3, sepal_width=2.9, petal_length=5.6, petal_width=1.8, species='virginica'),
 Iris(sepal_length=6.4, sepal_width=3.2, petal_length=4.5, petal_width=1.5, species='versicolor'),
 Iris(sepal_length=4.7, sepal_width=3.2, petal_length=1.3, petal_width=0.2, species='setosa')]

3.4.15. Use Case - 6

Calling function with all variables from higher order function. locals() will return a dict with all the variables in local scope of the function:

>>> def template(template, **user_data):
...     print('Template:', template)
...     print('Data:', user_data)
>>>
>>>
>>> def controller(firstname, lastname, uid=0):
...     groups = ['admins', 'astronauts']
...     permission = ['all', 'everywhere']
...     return template('user_details.html', **locals())
>>>
>>>     # template('user_details.html',
>>>     #    firstname='Mark',
>>>     #    lastname='Watney',
>>>     #    uid=0,
>>>     #    groups=['admins', 'astronauts'],
>>>     #    permission=['all', 'everywhere'])
>>>
>>>
>>> controller('Mark', 'Watney')  
Template: user_details.html
Data: {'firstname': 'Mark',
       'lastname': 'Watney',
       'uid': 0,
       'groups': ['admins', 'astronauts'],
       'permission': ['all', 'everywhere']}

3.4.16. Use Case - 7

• Definition of pandas.read_csv() function [1]

• Proxy functions. One of the most common use of *args, **kwargs:

>>> def read_csv(filepath_or_buffer, /, *, sep=', ', delimiter=None,
...              header='infer', names=None, index_col=None, usecols=None,
...              squeeze=False, prefix=None, mangle_dupe_cols=True,
...              dtype=None, engine=None, converters=None, true_values=None,
...              false_values=None, skipinitialspace=False, skiprows=None,
...              nrows=None, na_values=None, keep_default_na=True,
...              na_filter=True, verbose=False, skip_blank_lines=True,
...              parse_dates=False, infer_datetime_format=False,
...              keep_date_col=False, date_parser=None, dayfirst=False,
...              iterator=False, chunksize=None, compression='infer',
...              thousands=None, decimal=b'.', lineterminator=None,
...              quotechar='"', quoting=0, escapechar=None, comment=None,
...              encoding=None, dialect=None, tupleize_cols=None,
...              error_bad_lines=True, warn_bad_lines=True, skipfooter=0,
...              doublequote=True, delim_whitespace=False, low_memory=True,
...              memory_map=False, float_precision=None): ...

Calling function with positional only arguments is insane. In Python we don't do that, because we have keyword arguments.

>>> read_csv('/tmp/myfile.csv', ';', None, 'infer', None, None, None, False,
...          True, None, None, None, None, None, False, None, None, None,
...          None, True, True, False, True, False, False, False, None, False,
...          False, None, 'infer', None, b',', None, '"', 0, None, None,
...          None, None, None, True, True, 0, True, False, True, False, None)
Traceback (most recent call last):
TypeError: read_csv() takes 1 positional argument but 49 were given

Keyword arguments with sensible defaults are your best friends. The number of function parameters suddenly is not a problem:

>>> read_csv('myfile1.csv', delimiter=';', decimal=b',')
>>> read_csv('myfile2.csv', delimiter=';', decimal=b',')
>>> read_csv('myfile3.csv', delimiter=';', decimal=b',')
>>> read_csv('myfile4.csv', delimiter=';', decimal=b',')
>>> read_csv('myfile5.csv', delimiter=';', decimal=b',')

Proxy functions allows for changing defaults to the original function. One simply define a function which has sensible defaults and call the original function setting default values automatically:

>>> def mycsv(file, delimiter=';', decimal=b',', **kwargs):
...     return read_csv(file, delimiter=delimiter, decimal=decimal, **kwargs)

Thanks to using **kwargs there is no need to specify all the values from the original function. The uncovered arguments will simply be put in kwargs dictionary and passed to the original function:

>>> mycsv('/tmp/myfile1.csv')
>>> mycsv('/tmp/myfile2.csv')
>>> mycsv('/tmp/myfile3.csv')
>>> mycsv('/tmp/myfile4.csv')
>>> mycsv('/tmp/myfile5.csv')

This allows for cleaner code. Each parameter will be passed to mycsv function. Then it will be checked if there is a different default value already defined. If not, then parameter will be stored in kwargs and passed to the original function:

>>> mycsv('/tmp/myfile.csv', encoding='utf-8')
>>> mycsv('/tmp/myfile.csv', encoding='utf-8', verbose=True)
>>> mycsv('/tmp/myfile.csv', verbose=True, usecols=['sepal_length', 'species'])

3.4.17. Use Case - 8

Decorators are functions, which get reference to the decorated function as it's argument, and has closure which gets original function arguments as positional and keyword arguments:

>>> def mydecorator(func):
...     def wrapper(*args, **kwargs):
...         return func(*args, **kwargs)
...     return wrapper

Decorators could be used on any function, therefore we could not predict what would be the name of the parameter passed to it:

>>> @mydecorator
... def add(a, b):
...     return a + b

>>> @mydecorator
... def echo(text):
...     return text

Moreover it depends on a user whether he/she chooses to run function positionally, using keyword arguments or even both at the same time:

>>> add(1, 2)
3

>>> add(a=1, b=2)
3

>>> add(1, b=2)
3

>>> echo('hello')
'hello'

3.4.18. References

[1]

The pandas development team. API Reference Pandas.read_csv(). Year: 2021. Retrieved: 2021-12-05. URL: https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.read_csv.html

3.4.19. Assignments

# %% License
# - Copyright 2025, Matt Harasymczuk <matt@python3.info>
# - This code can be used only for learning by humans
# - This code cannot be used for teaching others
# - This code cannot be used for teaching LLMs and AI algorithms
# - This code cannot be used in commercial or proprietary products
# - This code cannot be distributed in any form
# - This code cannot be changed in any form outside of training course
# - This code cannot have its license changed
# - If you use this code in your product, you must open-source it under GPLv2
# - Exception can be granted only by the author

# %% Run
# - PyCharm: right-click in the editor and `Run Doctest in ...`
# - PyCharm: keyboard shortcut `Control + Shift + F10`
# - Terminal: `python -m doctest -v myfile.py`

# %% About
# - Name: Star Arguments Define
# - Difficulty: easy
# - Lines: 3
# - Minutes: 8

# %% English
# 1. Define `result: list[dict]`
# 2. Iterate over `DATA` separating `values` from `species`
# 3. To `result` append dict with:
#    - key: `species`, value: species name
#    - key: `mean`, value: arithmetic mean of `values`
# 4. Run doctests - all must succeed

# %% Polish
# 1. Zdefiniuj `result: list[dict]`
# 2. Iteruj po `DATA` separując `values` od `species`
# 3. Do `result` dodawaj dict z:
#    - klucz: `species`, wartość: nazwa gatunku
#    - klucz: `mean`, wartość: wynik średniej arytmetycznej `values`
# 4. Uruchom doctesty - wszystkie muszą się powieść

# %% Tests
"""
>>> import sys; sys.tracebacklimit = 0
>>> assert sys.version_info >= (3, 9), \
'Python 3.9+ required'

>>> assert type(result) is list, \
'Result must be a list'

>>> assert all(type(row) is dict for row in result), \
'All elements in result must be a dict'

>>> result  # doctest: +NORMALIZE_WHITESPACE
[{'species': 'virginica', 'mean': 3.875},
 {'species': 'setosa', 'mean': 2.65},
 {'species': 'versicolor', 'mean': 3.475},
 {'species': 'virginica', 'mean': 6.0},
 {'species': 'versicolor', 'mean': 3.95},
 {'species': 'setosa', 'mean': 4.7}]
"""

DATA = [
    ('sepal_length', 'sepal_width', 'petal_length', 'petal_width', 'species'),
    (5.8, 2.7, 5.1, 1.9, 'virginica'),
    (5.1, 0.2, 'setosa'),
    (5.7, 2.8, 4.1, 1.3, 'versicolor'),
    (6.3, 5.7, 'virginica'),
    (6.4, 1.5, 'versicolor'),
    (4.7, 'setosa'),
]


def mean(*args):
    return sum(args) / len(args)


# calculate mean and append dict with {'species': ..., 'mean': ...}
# type: list[dict]
result = ...

# %% License
# - Copyright 2025, Matt Harasymczuk <matt@python3.info>
# - This code can be used only for learning by humans
# - This code cannot be used for teaching others
# - This code cannot be used for teaching LLMs and AI algorithms
# - This code cannot be used in commercial or proprietary products
# - This code cannot be distributed in any form
# - This code cannot be changed in any form outside of training course
# - This code cannot have its license changed
# - If you use this code in your product, you must open-source it under GPLv2
# - Exception can be granted only by the author

# %% Run
# - PyCharm: right-click in the editor and `Run Doctest in ...`
# - PyCharm: keyboard shortcut `Control + Shift + F10`
# - Terminal: `python -m doctest -v myfile.py`

# %% About
# - Name: Star Arguments Range
# - Difficulty: medium
# - Lines: 25
# - Minutes: 13

# %% English
# 1. Write own implementation of a built-in function `range()`,
#    example usage: `myrange(0, 10)` or `myrange(0, 10, 2)`
# 2. Note, that function does not take any keyword arguments
# 3. How to implement passing only stop argument, i.e. `myrange(10)`?
# 4. Use lenght check of `*args` and `**kwargs`
# 5. Run doctests - all must succeed

# %% Polish
# 1. Zaimplementuj własne rozwiązanie wbudowanej funkcji `range()`,
#    przykład użycia: `myrange(0, 10)` lub `myrange(0, 10, 2)`
# 2. Zauważ, że funkcja nie przyjmuje żanych argumentów nazwanych (keyword)
# 3. Jak zaimplementować możliwość podawania tylko końca, tj. `myrange(10)`?
# 4. Użyj sprawdzania długości `*args` i `**kwargs`
# 5. Uruchom doctesty - wszystkie muszą się powieść

# %% Hints
# - https://github.com/python/cpython/blob/main/Objects/rangeobject.c#LC75
# - `raise TypeError('error message')`
# - `if len(args) == ...`

# %% Tests
"""
>>> import sys; sys.tracebacklimit = 0
>>> assert sys.version_info >= (3, 9), \
'Python 3.9+ required'

>>> from inspect import isfunction
>>> assert isfunction(myrange)

>>> myrange(0, 10, 2)
[0, 2, 4, 6, 8]

>>> myrange(0, 5)
[0, 1, 2, 3, 4]

>>> myrange(5)
[0, 1, 2, 3, 4]

>>> myrange()
Traceback (most recent call last):
TypeError: myrange expected at least 1 argument, got 0

>>> myrange(1,2,3,4)
Traceback (most recent call last):
TypeError: myrange expected at most 3 arguments, got 4

>>> myrange(stop=2)
Traceback (most recent call last):
TypeError: myrange() takes no keyword arguments

>>> myrange(start=1, stop=2)
Traceback (most recent call last):
TypeError: myrange() takes no keyword arguments

>>> myrange(start=1, stop=2, step=2)
Traceback (most recent call last):
TypeError: myrange() takes no keyword arguments
"""


# Write own implementation of a built-in function `range()`
# example: myrange(0, 10, 2), myrange(0, 10)
# note: function does not take keyword arguments
# type: Callable[[int,int,int], list[int]]
def myrange(*args, **kwargs):
    if kwargs:
        raise TypeError('myrange() takes no keyword arguments')

    current = start
    result = []

    while current < stop:
        result.append(current)
        current += step

    return result