Welcome to PrettyPrinter’s documentation!

PrettyPrinter

Documentation

Syntax-highlighting, declarative and composable pretty printer for Python 3.5+

pip install prettyprinter

• Drop in replacement for the standard library pprint: just rename pprint to prettyprinter in your imports.
• Uses a modified Wadler-Leijen layout algorithm for optimal formatting
• Write pretty printers for your own types with a dead simple, declarative interface

Pretty print common Python values:

>>> from datetime import datetime
>>> from prettyprinter import pprint
>>> pprint({'beautiful output': datetime.now()})
{
    'beautiful output': datetime.datetime(
        year=2017,
        month=12,
        day=12,
        hour=0,
        minute=43,
        second=4,
        microsecond=752094
    )
}

As well as your own, without any manual string formatting:

>>> class MyClass:
...     def __init__(self, one, two):
...         self.one = one
...         self.two = two

>>> from prettyprinter import register_pretty, pretty_call

>>> @register_pretty(MyClass)
... def pretty_myclass(value, ctx):
...     return pretty_call(ctx, MyClass, one=value.one, two=value.two)

>>> pprint(MyClass((1, 2, 3), {'a': 1, 'b': 2}))
MyClass(one=(1, 2, 3), two={'a': 1, 'b': 2})

>>> pprint({'beautiful output': datetime.now(), 'beautiful MyClass instance': MyClass((1, 2, 3), {'a': 1, 'b': 2})})
{
    'beautiful MyClass instance': MyClass(
        one=(1, 2, 3),
        two={'a': 1, 'b': 2}
    ),
    'beautiful output': datetime.datetime(
        year=2017,
        month=12,
        day=12,
        hour=0,
        minute=44,
        second=18,
        microsecond=384219
    )
}

Comes packaged with the following pretty printer definitions, which you can enable by calling prettyprinter.install_extras():

• datetime - (installed by default)
• enum - (installed by default)
• pytz - (installed by default)
• dataclasses - any new class you create will be pretty printed automatically
• attrs - pretty prints any new class you create with attrs
• django - pretty prints your Models and QuerySets
• numpy - pretty prints numpy scalars with explicit types
• requests - pretty prints Requests, Responses, Sessions, and more from the requests library

• Free software: MIT license
• Documentation: Documentation.

User Guide

Usage

Install the package with pip:

pip install prettyprinter

Then, instead of

from pprint import pprint

do

from prettyprinter import cpprint

for colored output. For colorless output, remove the c prefix from the function name:

from prettyprinter import pprint

Usage in Python code

Call cpprint() for colored output or pprint() for uncolored output, just like pprint.pprint:

>>> from prettyprinter import cpprint
>>> cpprint({'a': 1, 'b': 2})
{'a': 1, 'b': 2}

The default style is meant for a dark background. If you’re on a light background, or want to set your own theme, you may do so with set_default_style()

>>> from prettyprinter import set_default_style
>>> set_default_style('light')

Possible values are 'light', 'dark', and any subclass of pygments.styles.Style.

Adding a pretty printer function to the global namespace

If you’re so inclined, you could add cpprint() to the global namespace in your application so you can use it in a similar way as the built in print function:

import builtins
import prettyprinter
builtins.pretty = prettyprinter.cpprint

pretty([1, 2, 3])

You’ll want to add this to a file that is executed during application initialization.

Usage with IPython

You can use prettyprinter with IPython so that values in the REPL will be printed with prettyprinter using syntax highlighting. You need to call prettyprinter initialization functions at the start of an IPython session, which IPython facilitates with profile startup files. To initialize prettyprinter in your default profile, add and edit a new startup file with the following commands:

touch "`ipython locate profile default`/startup/init_prettyprinter.py"
nano "`ipython locate profile default`/startup/init_prettyprinter.py"

The code in this file will be run upon entering the shell. Add these lines and comment out any extra packages you don’t need:

# Specify syntax higlighting theme in IPython;
# will be picked up by prettyprinter.
from pygments import styles

# For light terminal backgrounds.
from prettyprinter.color import GitHubLightStyle
ipy = get_ipython()
ipy.colors = 'LightBG'
ipy.highlighting_style = GitHubLightStyle

# For dark terminal background.
ipy = get_ipython()
ipy.colors = 'linux'
ipy.highlighting_style = styles.get_style_by_name('monokai')

import prettyprinter

prettyprinter.install_extras(
    # Comment out any packages you are not using.
    include=[
        'ipython',
        'ipython_repr_pretty',
        'attrs',
        'django',
        'requests',
        'dataclasses',
    ],
    warn_on_error=True
)

Usage in the default Python shell

PrettyPrinter integrates with the default shell by overriding sys.displayhook, so that values evaluated in the prompt will be printed using PrettyPrinter. The integration is set up as follows:

>>> from prettyprinter import install_extras
>>> install_extras(['python'])
>>> {'a': 1, 'b': 2}
{'a': 1, 'b': 2}  # <- this will be colored when run in a terminal.

If you don’t want to run this every time you open a shell, create a Python startup file that executes the above statements and point the environment variable PYTHONSTARTUP to that file in your shell initialization file (such as ~/.bashrc), and rerun ~/.bashrc to assign the correct PYTHONSTARTUP value in your current shell session. Here’s a bash script to do that for you:

echo 'import prettyprinter; prettyprinter.install_extras(["python"])\n' >> ~/python_startup.py
echo "\nexport PYTHONSTARTUP=~/python_startup.py" >> ~/.bashrc
source ~/.bashrc

If you’re using a light background in your terminal, run this to add a line to the Python startup file to change the color theme PrettyPrinter uses:

echo '\nprettyprinter.set_default_style("light")' >> ~/python_startup.py

Then, after starting the python shell,

python

values evaluated in the shell should be printed with PrettyPrinter without any other setup.

>>> {'a': 1, 'b': 2}
{'a': 1, 'b': 2} # <- the output should be colored when run in a terminal.

Pretty printing your own types

Given a custom class:

class MyClass(object):
    def __init__(self, one, two):
        self.one = one
        self.two = two

You can register a pretty printer:

from prettyprinter import register_pretty, pretty_call

@register_pretty(MyClass)
def pretty_myclass(value, ctx):
    return pretty_call(
        ctx,
        MyClass,
        one=value.one,
        two=value.two
    )

To get an output like this with simple data:

>>> prettyprinter.pprint(MyClass(1, 2))
MyClass(one=1, two=2)

The real utility is in how nested data pretty printing is handled for you, and how the function call is broken to multiple lines for easier legibility:

>>> prettyprinter.pprint(MyClass({'abc': 1, 'defg': 2, 'hijk': 3}, [1, 2]))
MyClass(
    one={
        'abc': 1,
        'defg': 2,
        'hijk': 3
    },
    two=[1, 2]
)

@register_pretty is a decorator that takes the type to register. Internally, functools.singledispatch is used to handle dispatch to the correct pretty printer. This means that any subclasses will also use the same printer.

The decorated function must accept exactly two positional arguments:

• value to pretty print, and
• ctx, a context value.

In most cases, you don’t need need to do anything with the context, except pass it along in nested calls. It can be used to affect rendering of nested data.

The function must return a Doc, which is either an instance of Doc or a str. pretty_call() returns a Doc that represents a function call. Given an arbitrary context ctx

pretty_call(ctx, round, 1.5)

Will be printed out as

round(1.5)

with syntax highlighting.

API Reference

API Reference

Top-level package for prettyprinter.

prettyprinter.cpprint(object, stream=UNSET, indent=UNSET, width=UNSET, depth=UNSET, *, compact=False, ribbon_width=UNSET, max_seq_len=UNSET, sort_dict_keys=UNSET, style=None, end='\n')

Pretty print a Python value object to stream, which defaults to sys.stdout. The output will be colored and syntax highlighted.

Parameters:

• indent – number of spaces to add for each level of nesting.
• stream – the output stream, defaults to sys.stdout
• width – a soft maximum allowed number of columns in the output, which the layout algorithm attempts to stay under.
• depth – maximum depth to print nested structures
• ribbon_width – a soft maximum allowed number of columns in the output, after indenting the line
• max_seq_len – a maximum sequence length that applies to subclasses of lists, sets, frozensets, tuples and dicts. A trailing comment that indicates the number of truncated elements. Setting max_seq_len to None disables truncation.
• sort_dict_keys – a bool value indicating if dict keys should be sorted in the output. Defaults to False, in which case the default order is used, which is the insertion order in CPython 3.6+.
• style – one of 'light', 'dark' or a subclass of pygments.styles.Style. If omitted, will use the default style. If the default style is not changed by the user with set_default_style(), the default is 'dark'.

prettyprinter.pprint(object, stream=UNSET, indent=UNSET, width=UNSET, depth=UNSET, *, compact=False, ribbon_width=UNSET, max_seq_len=UNSET, sort_dict_keys=UNSET, end='\n')

Pretty print a Python value object to stream, which defaults to sys.stdout. The output will not be colored.

Parameters:

• indent – number of spaces to add for each level of nesting.
• stream – the output stream, defaults to sys.stdout
• width – a soft maximum allowed number of columns in the output, which the layout algorithm attempts to stay under.
• depth – maximum depth to print nested structures
• ribbon_width – a soft maximum allowed number of columns in the output, after indenting the line
• max_seq_len – a maximum sequence length that applies to subclasses of lists, sets, frozensets, tuples and dicts. A trailing comment that indicates the number of truncated elements. Setting max_seq_len to None disables truncation.
• sort_dict_keys – a bool value indicating if dict keys should be sorted in the output. Defaults to False, in which case the default order is used, which is the insertion order in CPython 3.6+.

prettyprinter.pformat(object, indent=UNSET, width=UNSET, depth=UNSET, *, ribbon_width=UNSET, max_seq_len=UNSET, compact=UNSET, sort_dict_keys=UNSET)

Returns a pretty printed representation of the object as a str. Accepts the same parameters as pprint(). The output is not colored.

prettyprinter.pretty_repr(instance)

A function assignable to the __repr__ dunder method, so that the prettyprinter definition for the type is used to provide repr output. Usage:

from prettyprinter import pretty_repr

class MyClass:
    __repr__ = pretty_repr

prettyprinter.install_extras(include=frozenset({'attrs', 'python', 'dataclasses', 'ipython', 'requests', 'ipython_repr_pretty', 'django', 'numpy'}), *, exclude=frozenset(), raise_on_error=False, warn_on_error=True)

Installs extras.

Installing an extra means registering pretty printers for objects from third party libraries and/or enabling integrations with other python programs.

• 'attrs' - automatically pretty prints classes created using the attrs package.
• 'dataclasses' - automatically pretty prints classes created using the dataclasses module.

• 'django' - automatically pretty prints Model and QuerySet subclasses defined in your

  Django apps.

• numpy - automatically pretty prints numpy scalars with explicit types, and, for numpy>=1.14, numpy arrays.
• 'requests' - automatically pretty prints Requests, Responses, Sessions, etc.
• 'ipython' - makes prettyprinter the default printer in the IPython shell.
• 'python' - makes prettyprinter the default printer in the default Python shell.
• 'ipython_repr_pretty' - automatically prints objects that define a _repr_pretty_ method to integrate with IPython.lib.pretty.

Parameters:

• include – an iterable of strs representing the extras to include. All extras are included by default.
• exclude – an iterable of strs representing the extras to exclude.

prettyprinter.set_default_style(style)

Sets default global style to be used by prettyprinter.cpprint.

Parameters:style – the style to set, either subclass of pygments.styles.Style or one of 'dark', 'light'

prettyprinter.set_default_config(*, style=UNSET, max_seq_len=UNSET, width=UNSET, ribbon_width=UNSET, depth=UNSET, sort_dict_keys=UNSET)

Sets the default configuration values used when calling pprint, cpprint, or pformat, if those values weren’t explicitly provided. Only overrides the values provided in the keyword arguments.

prettyprinter.get_default_config()

Returns a read-only view of the current configuration

prettyprinter.register_pretty(type=None, predicate=None)

Returns a decorator that registers the decorated function as the pretty printer for instances of type.

Parameters:

• type – the type to register the pretty printer for, or a str to indicate the module and name, e.g.: 'collections.Counter'.
• predicate – a predicate function that takes one argument and returns a boolean indicating if the value should be handled by the registered pretty printer.

Only one of type and predicate may be supplied. That means that predicate will be run on unregistered types only.

The decorated function must accept exactly two positional arguments:

• value to pretty print, and
• ctx, a context value.

Here’s an example of the pretty printer for OrderedDict:

from collections import OrderedDict
from prettyprinter import register_pretty, pretty_call

@register_pretty(OrderedDict)
def pretty_orderreddict(value, ctx):
    return pretty_call(ctx, OrderedDict, list(value.items()))

prettyprinter.pretty_call(ctx, fn, *args, **kwargs)

Returns a Doc that represents a function call to fn with the remaining positional and keyword arguments.

You can only use this function on Python 3.6+. On Python 3.5, the order of keyword arguments is not maintained, and you have to use pretty_call_alt().

Given an arbitrary context ctx,:

pretty_call(ctx, sorted, [7, 4, 5], reverse=True)

Will result in output:

sorted([7, 4, 5], reverse=True)

The layout algorithm will automatically break the call to multiple lines if needed:

sorted(
    [7, 4, 5],
    reverse=True
)

pretty_call automatically handles syntax highlighting.

Parameters:

• ctx (prettyprinter.prettyprinter.PrettyContext) – a context value
• fn – a callable
• args – positional arguments to render to the call
• kwargs – keyword arguments to render to the call

Returns:

Doc

prettyprinter.pretty_call_alt(ctx, fn, args=(), kwargs=())

Returns a Doc that represents a function call to fn with the args and kwargs.

Given an arbitrary context ctx,:

pretty_call_alt(ctx, sorted, args=([7, 4, 5], ), kwargs=[('reverse', True)])

Will result in output:

sorted([7, 4, 5], reverse=True)

The layout algorithm will automatically break the call to multiple lines if needed:

sorted(
    [7, 4, 5],
    reverse=True
)

pretty_call_alt automatically handles syntax highlighting.

Parameters:

• ctx (prettyprinter.prettyprinter.PrettyContext) – a context value
• fn – a callable
• args – a tuple of positional arguments to render to the call
• kwargs – keyword arguments to render to the call. Either an instance of OrderedDict, or an iterable of two-tuples, where the first element is a str (key), and the second is the Python value for that keyword argument.

Returns:

Doc

prettyprinter.trailing_comment(value, comment_text)

Annotates a value with a comment text, so that the comment will be rendered “trailing”, e.g. in place of the last element in a list, set or tuple, or after the last argument in a function.

This will force the rendering of value to be broken to multiple lines as Python does not have inline comments.

>>> trailing_comment(['value'], '...and more')
[
    'value',
    # ...and more
]

prettyprinter.comment(value, comment_text)

Annotates a value or a Doc with a comment.

When printed by prettyprinter, the comment will be rendered next to the value or Doc.

prettyprinter.python_to_sdocs(value, indent, width, depth, ribbon_width, max_seq_len, sort_dict_keys)

prettyprinter.default_render_to_stream(stream, sdocs, newline='\n', separator=' ')

class prettyprinter.PrettyPrinter(*args, **kwargs)

Bases: object

format(object)

isreadable(object)

isrecursive(object)

pformat(object)

pprint(object)

prettyprinter.saferepr(object)

Version of repr() which can handle recursive data structures.

prettyprinter.isreadable(object)

Determine if saferepr(object) is readable by eval().

prettyprinter.isrecursive(object)

Determine if object requires a recursive representation.

Meta

Credits

• Tommi Kaikkonen
• GitHub user Cologler
• GitHub user johnnoone
• GitHub user dangirsh
• GitHub user RazerM
• GitHub user jdanbrown
• GitHub user @anntzer
• GitHub user @crowsonkb

History

0.18.0 (2019-06-21)

• Improve prettyprinting of builtin bound methods. by @anntzer
• Fix test suite compatibility with hypothesis4. by @anntzer
• Use $COLORFGBG to help decide whether to use a dark or light style by @crowsonkb
• Truncate numpy arrays with ndim >= 2 so that the total printed does not exceed max_seq_len by @crowsonkb
• Fixed ipython_repr_pretty extra raising an exception when printing Mock instances (GH #61)
• Added support for pretty printing types.SimpleNamespace (GH #60)
• Fixed dictionary pretty printing indentation when the value had a comment on the line above (GH #59)

0.17.0 (2019-03-14)

• Add prettyprinter for numpy ndarrays. by @anntzer
• Add helper to apply default config. by @anntzer
• A number of docs and CI improvements: #43, #44, #45 .Thanks @anntzer
• Add support for functools.partialmethod. by @anntzer
• Fix typo in changelog. Thanks @Vlad-Shcherbina

0.16.0 (2019-02-27)

• Adds a new extra for numpy. The extra currently registers pretty printers for numpy scalar types. Enable it with import prettyprinter; prettyprinter.install_extras(['numpy']). Thanks @anntzer
• C-API named tuples are now automatically prettyprinted. C-API named tuples are returned from expressions such as sys.flags, time.strptime(...), and os.stat(...). The fieldname of each tuple element is annotated using a comment in the output.

0.15.0 (2019-02-25)

This release brings bugfixes, an enhancement to pathlib prettyprinting (thanks @anntzer ) and a nice performance boost. There was an redundant subtree call in a tree normalization procedure that caused exponential runtime, worsening quickly if data was highly nested. That extra call is now removed.

• Fix exponential runtime in highly nested data
• Fix infinite loop when rendering strings in highly nested data
• Only split Path prettyprints on “/”, not on “-” or other nonword chars. , thanks @anntzer
• Add vim swapfiles to gitignore , thanks @anntzer
• Fix typo , thanks @anntzer

0.14.0 (2018-07-25)

Most likely no breaking changes.

• Added definitions for pathlib standard library module thanks to GitHub user RazerM
• Fixed unexpected error output inside Jupyter notebooks thanks to GitHub user jdanbrown
• Fixed missing commas in setup.py requirements list

0.13.2 (2018-05-29)

No breaking changes.

• Fixed the dataclasses pretty printer that had regressed after changes to the dataclasses API. Fix was contributed by GitHub user dangirsh.

0.13.1 (2018-02-03)

No breaking changes.

• Fixed GH issue #17 where Django models showed an incorrect display name for fields with choices.

0.13.0 (2018-02-03)

No breaking changes.

• Added definitions for the ast standard library module thanks to GitHub user johnnoone.

0.12.0 (2018-01-22)

No breaking changes.

• Added a definition for classes that look like they were built with collections.namedtuple
• If a pretty printer raises an exception, it is caught and emitted as a warning, and the default repr implementation will be used instead.
• Added definitions for collections.ChainMap, collections.defaultdict, collections.deque, functools.partial, and for exception objects.
• Made pretty printers for primitive types (dict, list, set, etc.) render a subclass constructor around them

0.11.0 (2018-01-20)

No breaking changes.

• Added Python 3.5 support
• Added pretty_call_alt function that doesn’t depend on dict maintaining insertion order
• Fixed bug in set_default_config where most configuration values were not updated
• Added get_default_config

0.10.1 (2018-01-10)

No breaking changes.

• Fixed regression with types.MappingProxyType not being properly registered.

0.10.0 (2018-01-09)

No breaking changes.

• Added support for deferred printer registration, where instead of a concrete type value, you can pass a qualified path to a type as a str to register_pretty. For an example, see the deferred printer registration for uuid.UUID

0.9.0 (2018-01-03)

No breaking changes.

• Added pretty printer definition for types.MappingProxyType thanks to GitHub user Cologler
• Added support for _repr_pretty_ in the extra ipython_repr_pretty.

0.8.1 (2018-01-01)

• Fixed issue #7 where having a str value for IPython’s highlighting_style setting was not properly handled in prettyprinter’s IPython integration, and raised an exception when trying to print data.

0.8.0 (2017-12-31)

Breaking changes:

• by default, dict keys are printed in the default order (insertion order in CPython 3.6+). Previously they were sorted like in the pprint standard library module. To let the user control this, an additional keyword argument sort_dict_keys was added to cpprint, pprint, and pformat. Pretty printer definitions can control dict key sorting with the PrettyContext instance passed to each pretty printer function.

Non-breaking changes:

• Improved performance of rendering colorized output by caching colors.
• Added prettyprinter.pretty_repr that is assignable to __repr__ dunder methods, so you don’t need to write it separately from the pretty printer definition.
• Deprecated use of PrettyContext.set in favor of less misleading PrettyContext.assoc
• Defined pretty printing for instances of type, i.e. classes.
• Defined pretty printing for functions

0.7.0 (2017-12-23)

Breaking change: instances of lists, sets, frozensets, tuples and dicts will be truncated to 1000 elements by default when printing.

• Added pretty printing definitions for dataclasses
• Improved performance of splitting strings to multiple lines by ~15%
• Added a maximum sequence length that applies to subclasses of lists, sets, frozensets, tuples and dicts. The default is 1000. There is a trailing comment that indicates the number of truncated elements. To remove truncation, you can set max_seq_len to None using set_default_config explained below.
• Added ability to change the default global configuration using set_default_config. The functions accepts zero to many keyword arguments and replaces those values in the global configuration with the ones provided.

from prettyprinter import set_default_config

set_default_config(
    style='dark',
    max_seq_len=1000,
    width=79,
    ribbon_width=71,
    depth=None,
)

0.6.0 (2017-12-21)

No backwards incompatible changes.

• Added pretty printer definitions for the requests library. To use it, include 'requests' in your install_extras call: prettyprinter.install_extras(include=['requests']).

0.5.0 (2017-12-21)

No backwards incompatible changes.

• Added integration for the default Python shell
• Wrote docs to explain integration with the default Python shell
• Check install_extras arguments for unknown extras

0.4.0 (2017-12-14)

• Revised comment to accept both normal Python values and Docs, and reversed the argument order to be more Pythonic

0.3.0 (2017-12-12)

• Add set_default_style function, improve docs on working with a light background

0.2.0 (2017-12-12)

• Numerous API changes and improvements.

0.1.0 (2017-12-07)

• First release on PyPI.