Retour au classement

gruns/icecream

Python

🍦 Never use print() to debug again.

pythonpython3libraryinspectsdebugdebuggingdebugging-toolprint
Croissance des étoiles
Étoiles
10.1k
Forks
224
Croissance hebdomadaire
Issues
50
5k10k
janv. 2023mars 2024mai 2025juil. 2026
ArtefactsPyPIpip install icecream
README

IceCream

IceCream — Never use print() to debug again

Do you ever use print() or log() to debug your code? Of course you do. IceCream, or ic for short, makes print debugging a little sweeter.

ic() is like print(), but better:

  1. It prints both variables and expressions along with their values.
  2. It's 60% faster to type.
  3. Data structures are formatted and pretty printed.
  4. Output is syntax highlighted.
  5. It optionally includes program context: filename, line number, and parent function.

IceCream is well tested, permissively licensed, and supports Python 3 and PyPy3.

IceCream is maintained by Jakeroid (Ivan Karabadzhak), with support from the confidential computing folks at 🌖 Lunal.

Inspect Variables

Have you ever printed variables or expressions to debug your program? If you've ever typed something like

print(foo('123'))

or the more thorough

print("foo('123')", foo('123'))

then ic() will put a smile on your face. With arguments, ic() inspects itself and prints both its own arguments and the values of those arguments.

from icecream import ic

def foo(i):
    return i + 333

ic(foo(123))

Prints

ic| foo(123): 456

Similarly,

d = {'key': {1: 'one'}}
ic(d['key'][1])

class klass():
    attr = 'yep'
ic(klass.attr)

Prints

ic| d['key'][1]: 'one'
ic| klass.attr: 'yep'

Just give ic() a variable or expression and you're done. Easy.

Inspect Execution

Have you ever used print() to determine which parts of your program are executed, and in which order they're executed? For example, if you've ever added print statements to debug code like

def foo():
    print(0)
    first()

    if expression:
        print(1)
        second()
    else:
        print(2)
        third()

then ic() helps here, too. Without arguments, ic() inspects itself and prints the calling filename, line number, and parent function.

from icecream import ic

def foo():
    ic()
    first()

    if expression:
        ic()
        second()
    else:
        ic()
        third()

Prints

ic| example.py:4 in foo()
ic| example.py:11 in foo()

Just call ic() and you're done. Simple.

Return Value

ic() returns its argument(s), so ic() can easily be inserted into pre-existing code.

>>> a = 6
>>> def half(i):
>>>     return i / 2
>>> b = half(ic(a))
ic| a: 6
>>> ic(b)
ic| b: 3

Miscellaneous

ic.format(*args) is like ic() but the output is returned as a string instead of written to stderr.

>>> from icecream import ic
>>> s = 'sup'
>>> out = ic.format(s)
>>> print(out)
ic| s: 'sup'

Additionally, ic()'s output can be entirely disabled, and later re-enabled, with ic.disable() and ic.enable() respectively.

from icecream import ic

ic(1)

ic.disable()
ic(2)

ic.enable()
ic(3)

Prints

ic| 1: 1
ic| 3: 3

ic() continues to return its arguments when disabled, of course; no existing code with ic() breaks.

Import Tricks

To make ic() available in every file without needing to be imported in every file, you can install() it. For example, in a root A.py:

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

from icecream import install
install()

from B import foo
foo()

and then in B.py, which is imported by A.py, just call ic():

# -*- coding: utf-8 -*-

def foo():
    x = 3
    ic(x)

install() adds ic() to the builtins module, which is shared amongst all files imported by the interpreter. Similarly, ic() can later be uninstall()ed, too.

ic() can also be imported in a manner that fails gracefully if IceCream isn't installed, like in production environments (i.e. not development). To that end, this fallback import snippet may prove useful:

try:
    from icecream import ic
except ImportError:  # Graceful fallback if IceCream isn't installed.
    ic = lambda *a: None if not a else (a[0] if len(a) == 1 else a)  # noqa

Configuration

ic.configureOutput(prefix, outputFunction, argToStringFunction, includeContext, contextAbsPath) controls ic()'s output.

prefix, if provided, adopts a custom output prefix. prefix can be a string, like

>>> from icecream import ic
>>> ic.configureOutput(prefix='hello -> ')
>>> ic('world')
hello -> 'world'

or a function.

>>> import time
>>> from icecream import ic
>>>  
>>> def unixTimestamp():
>>>     return '%i |> ' % int(time.time())
>>>
>>> ic.configureOutput(prefix=unixTimestamp)
>>> ic('world')
1519185860 |> 'world': 'world'

prefix's default value is ic| .

outputFunction, if provided, is called once for every ic() call with ic()'s output, as a string, instead of that string being written to stderr (the default).

>>> import logging
>>> from icecream import ic
>>>
>>> def warn(s):
>>>     logging.warning("%s", s)
>>>
>>> ic.configureOutput(outputFunction=warn)
>>> ic('eep')
WARNING:root:ic| 'eep': 'eep'

argToStringFunction, if provided, is called with argument values to be serialized to displayable strings. The default is PrettyPrint's pprint.pformat(), but this can be changed to, for example, handle non-standard datatypes in a custom fashion.

>>> from icecream import ic
>>>
>>> def toString(obj):
>>>    if isinstance(obj, str):
>>>        return '[!string %r with length %i!]' % (obj, len(obj))
>>>    return repr(obj)
>>>
>>> ic.configureOutput(argToStringFunction=toString)
>>> ic(7, 'hello')
ic| 7: 7, 'hello': [!string 'hello' with length 5!]

The default argToStringFunction is icecream.argumentToString, and has methods to register and unregister functions to be dispatched for specific classes using functools.singledispatch. It also has a registry property to view registered functions.

>>> from icecream import ic, argumentToString
>>> import numpy as np
>>>
>>> # Register a function to summarize numpy array
>>> @argumentToString.register(np.ndarray)
>>> def _(obj):
>>>     return f"ndarray, shape={obj.shape}, dtype={obj.dtype}"
>>>
>>> x = np.zeros((1, 2))
>>> ic(x)
ic| x: ndarray, shape=(1, 2), dtype=float64
>>>
>>> # View registered functions
>>> argumentToString.registry
mappingproxy({object: <function icecream.icecream.argumentToString(obj)>,
              numpy.ndarray: <function __main__._(obj)>})
>>>
>>> # Unregister a function and fallback to the default behavior
>>> argumentToString.unregister(np.ndarray)
>>> ic(x)
ic| x: array([[0., 0.]])

includeContext, if provided and True, adds the ic() call's filename, line number, and parent function to ic()'s output.

>>> from icecream import ic
>>> ic.configureOutput(includeContext=True)
>>>
>>> def foo():
>>>   i = 3
>>>   ic(i)
>>> foo()
ic| example.py:12 in foo()- i: 3

includeContext is False by default.

contextAbsPath, if provided and True, outputs absolute filepaths, like /path/to/foo.py, over just filenames, like foo.py, when ic() is called with includeContext == True. This is useful when debugging multiple files that share the same filename(s). Moreover, some editors, like VSCode, turn absolute filepaths into clickable links that open the file where ic() was called.

>>> from icecream import ic
>>> ic.configureOutput(includeContext=True, contextAbsPath=True)
>>>
>>> i = 3
>>>
>>> def foo():
>>>   ic(i)
>>> foo()
ic| /absolute/path/to/example.py:12 in foo()- i: 3
>>>
>>> ic.configureOutput(includeContext=True, contextAbsPath=False)
>>>
>>> def foo():
>>>   ic(i)
>>> foo()
ic| example.py:18 in foo()- i: 3

contextAbsPath is False by default.

If you want to use icecream with multiple log levels, like with Python’s logging module, you can use ic.format() to integrate icecream’s debugging with your logger:

import logging
from icecream import ic

foo = 'bar'
logging.debug(ic.format(foo))

❕ This is a bit clunky. Would you prefer built-in log level support in icecream? If so, please share your thoughts in issue.

Installation

Installing IceCream with pip is easy.

$ pip install icecream

ic() uses executing by @alexmojaki to reliably locate ic() calls in Python source. It's magic.

IceCream in Other Languages

Delicious IceCream should be enjoyed in every language.

If you'd like a similar ic() function in your favorite language, please open a pull request! IceCream's goal is to sweeten print debugging with a handy-dandy ic() function in every language.

Dépôts similaires
donnemartin/system-design-primer

Learn how to design large-scale systems. Prep for the system design interview. Includes Anki flashcards.

PythonPyPIOtherprogrammingdevelopment
358.7k57.3k
vinta/awesome-python

An opinionated list of Python frameworks, libraries, tools, and resources

PythonPyPIOtherawesomepython
awesome-python.com
309.6k28.4k
practical-tutorials/project-based-learning

Curated list of project-based tutorials

PythonPyPIMIT Licensetutorialproject
274.6k35.4k
TheAlgorithms/Python

All Algorithms implemented in Python

PythonPyPIMIT Licensepythonalgorithm
thealgorithms.github.io/Python/
223k50.9k
tensorflow/tensorflow

An Open Source Machine Learning Framework for Everyone

C++Apache License 2.0tensorflowmachine-learning
tensorflow.org
196.5k75.7k
Significant-Gravitas/AutoGPT

AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.

PythonPyPIOtheraiopenai
agpt.co
185.6k46.1k
CyC2018/CS-Notes

:books: 技术面试必备基础知识、Leetcode、计算机操作系统、计算机网络、系统设计

algorithmleetcode
cyc2018.xyz
184.8k50.8k
yt-dlp/yt-dlp

A feature-rich command-line audio/video downloader

PythonPyPIThe Unlicenseyoutube-dlpython
discord.gg/H5MNcFW63r
179.4k15.3k
521xueweihan/HelloGitHub

:octocat: 分享 GitHub 上有趣、入门级的开源项目。Share interesting, entry-level open source projects on GitHub.

PythonPyPIgithubhellogithub
hellogithub.com
166.5k12.4k
huggingface/transformers

🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

PythonPyPIApache License 2.0nlpnatural-language-processing
huggingface.co/transformers
162.8k34k
langgenius/dify

Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.

TypeScriptnpmOtheraigpt
dify.ai
149.7k23.6k
langchain-ai/langchain

The agent engineering platform.

PythonPyPIMIT Licenseaianthropic
docs.langchain.com/langchain/
142.3k23.7k