mirror of https://github.com/Textualize/rich.git
241 lines
9.3 KiB
ReStructuredText
241 lines
9.3 KiB
ReStructuredText
Pretty Printing
|
|
===============
|
|
|
|
In addition to syntax highlighting, Rich will format (i.e. *pretty print*) containers such as lists, dicts, and sets.
|
|
|
|
Run the following command to see an example of pretty printed output::
|
|
|
|
python -m rich.pretty
|
|
|
|
Note how the output will change to fit within the terminal width.
|
|
|
|
pprint method
|
|
-------------
|
|
|
|
The :func:`~rich.pretty.pprint` method offers a few more arguments you can use to tweak how objects are pretty printed. Here's how you would import it::
|
|
|
|
>>> from rich.pretty import pprint
|
|
>>> pprint(locals())
|
|
|
|
Indent guides
|
|
~~~~~~~~~~~~~
|
|
|
|
Rich can draw *indent guides* to highlight the indent level of a data structure. These can make it easier to read more deeply nested output. The pprint method enables indent guides by default. You can set ``indent_guides=False`` to disable this feature.
|
|
|
|
Expand all
|
|
~~~~~~~~~~
|
|
|
|
Rich is quite conservative about expanding data structures and will try to fit as much in each line as it can. If you prefer, you can tell Rich to fully expand all data structures by setting ``expand_all=True``. Here's an example::
|
|
|
|
>>> pprint(["eggs", "ham"], expand_all=True)
|
|
|
|
Truncating pretty output
|
|
~~~~~~~~~~~~~~~~~~~~~~~~
|
|
|
|
Very long data structures can be difficult to read and you may find yourself scrolling through multiple pages in the terminal to find the data you are interested in. Rich can truncate containers and long strings to give you an overview without swamping your terminal.
|
|
|
|
|
|
If you set the ``max_length`` argument to an integer, Rich will truncate containers with more than the given number of elements. If data is truncated, Rich will display an ellipsis ``...`` and the number of elements not shown.
|
|
|
|
Here's an example::
|
|
|
|
>>> pprint(locals(), max_length=2)
|
|
|
|
If you set the ``max_string`` argument to an integer, Rich will truncate strings over that length. Truncated string will be appended with the number of characters that have not been shown. Here's an example::
|
|
|
|
>>> pprint("Where there is a Will, there is a Way", max_string=21)
|
|
|
|
Pretty renderable
|
|
-----------------
|
|
|
|
Rich offers a :class:`~rich.pretty.Pretty` class which you can use to insert pretty printed data in to another renderable.
|
|
|
|
The following example displays pretty printed data within a simple panel::
|
|
|
|
from rich import print
|
|
from rich.pretty import Pretty
|
|
from rich.panel import Panel
|
|
|
|
pretty = Pretty(locals())
|
|
panel = Panel(pretty)
|
|
print(panel)
|
|
|
|
There are a large number of options to tweak the pretty formatting, See the :class:`~rich.pretty.Pretty` reference for details.
|
|
|
|
Rich Repr Protocol
|
|
------------------
|
|
|
|
Rich is able to syntax highlight any output, but the formatting is restricted to builtin containers, dataclasses, and other objects Rich knows about, such as objects generated by the `attrs <https://www.attrs.org/en/stable/>`_ library. To add Rich formatting capabilities to custom objects, you can implement the *rich repr protocol*.
|
|
|
|
Run the following command to see an example of what the Rich repr protocol can generate::
|
|
|
|
python -m rich.repr
|
|
|
|
First, let's look at a class that might benefit from a Rich repr::
|
|
|
|
class Bird:
|
|
def __init__(self, name, eats=None, fly=True, extinct=False):
|
|
self.name = name
|
|
self.eats = list(eats) if eats else []
|
|
self.fly = fly
|
|
self.extinct = extinct
|
|
|
|
def __repr__(self):
|
|
return f"Bird({self.name!r}, eats={self.eats!r}, fly={self.fly!r}, extinct={self.extinct!r})"
|
|
|
|
BIRDS = {
|
|
"gull": Bird("gull", eats=["fish", "chips", "ice cream", "sausage rolls"]),
|
|
"penguin": Bird("penguin", eats=["fish"], fly=False),
|
|
"dodo": Bird("dodo", eats=["fruit"], fly=False, extinct=True)
|
|
}
|
|
print(BIRDS)
|
|
|
|
The result of this script would be::
|
|
|
|
{'gull': Bird('gull', eats=['fish', 'chips', 'ice cream', 'sausage rolls'], fly=True, extinct=False), 'penguin': Bird('penguin', eats=['fish'], fly=False, extinct=False), 'dodo': Bird('dodo', eats=['fruit'], fly=False, extinct=True)}
|
|
|
|
The output is long enough to wrap on to the next line, which can make it hard to read. The repr strings are informative but a little verbose since they include default arguments. If we print this with Rich, things are improved somewhat::
|
|
|
|
{
|
|
'gull': Bird('gull', eats=['fish', 'chips', 'ice cream', 'sausage rolls'],
|
|
fly=True, extinct=False),
|
|
'penguin': Bird('penguin', eats=['fish'], fly=False, extinct=False),
|
|
'dodo': Bird('dodo', eats=['fruit'], fly=False, extinct=True)
|
|
}
|
|
|
|
Rich knows how to format the container dict, but the repr strings are still verbose, and there is some wrapping of the output (assumes an 80 character terminal).
|
|
|
|
We can solve both these issues by adding the following ``__rich_repr__`` method::
|
|
|
|
def __rich_repr__(self):
|
|
yield self.name
|
|
yield "eats", self.eats
|
|
yield "fly", self.fly, True
|
|
yield "extinct", self.extinct, False
|
|
|
|
Now if we print the same object with Rich we would see the following::
|
|
|
|
{
|
|
'gull': Bird(
|
|
'gull',
|
|
eats=['fish', 'chips', 'ice cream', 'sausage rolls']
|
|
),
|
|
'penguin': Bird('penguin', eats=['fish'], fly=False),
|
|
'dodo': Bird('dodo', eats=['fruit'], fly=False, extinct=True)
|
|
}
|
|
|
|
The default arguments have been omitted, and the output has been formatted nicely. The output remains readable even if we have less room in the terminal, or our objects are part of a deeply nested data structure::
|
|
|
|
{
|
|
'gull': Bird(
|
|
'gull',
|
|
eats=[
|
|
'fish',
|
|
'chips',
|
|
'ice cream',
|
|
'sausage rolls'
|
|
]
|
|
),
|
|
'penguin': Bird(
|
|
'penguin',
|
|
eats=['fish'],
|
|
fly=False
|
|
),
|
|
'dodo': Bird(
|
|
'dodo',
|
|
eats=['fruit'],
|
|
fly=False,
|
|
extinct=True
|
|
)
|
|
}
|
|
|
|
You can add a ``__rich_repr__`` method to any class to enable the Rich formatting. This method should return an iterable of tuples. You could return a list of tuples, but it's easier to express with the ``yield`` keywords, making it a *generator*.
|
|
|
|
Each tuple specifies an element in the output.
|
|
|
|
- ``yield value`` will generate a positional argument.
|
|
- ``yield name, value`` will generate a keyword argument.
|
|
- ``yield name, value, default`` will generate a keyword argument *if* ``value`` is not equal to ``default``.
|
|
|
|
You can also tell Rich to generate the *angular bracket* style of repr, which tend to be used where there is no easy way to recreate the object's constructor. To do this set the function attribute ``"angular"`` to ``True`` immediately after your ``__rich_repr__`` method. For example::
|
|
|
|
__rich_repr__.angular = True
|
|
|
|
This will change the output of the Rich repr example to the following::
|
|
|
|
{
|
|
'gull': <Bird 'gull' eats=['fish', 'chips', 'ice cream', 'sausage rolls']>,
|
|
'penguin': <Bird 'penguin' eats=['fish'] fly=False>,
|
|
'dodo': <Bird 'dodo' eats=['fruit'] fly=False extinct=True>
|
|
}
|
|
|
|
|
|
Note that you can add ``__rich_repr__`` methods to third-party libraries *without* including Rich as a dependency. If Rich is not installed, then nothing will break. Hopefully more third-party libraries will adopt Rich repr methods in the future.
|
|
|
|
Typing
|
|
~~~~~~
|
|
|
|
If you want to type the Rich repr method you can import and return ``rich.repr.Result``, which will help catch logical errors::
|
|
|
|
|
|
import rich.repr
|
|
|
|
|
|
class Bird:
|
|
def __init__(self, name, eats=None, fly=True, extinct=False):
|
|
self.name = name
|
|
self.eats = list(eats) if eats else []
|
|
self.fly = fly
|
|
self.extinct = extinct
|
|
|
|
def __rich_repr__(self) -> rich.repr.Result:
|
|
yield self.name
|
|
yield "eats", self.eats
|
|
yield "fly", self.fly, True
|
|
yield "extinct", self.extinct, False
|
|
|
|
|
|
Automatic Rich Repr
|
|
~~~~~~~~~~~~~~~~~~~
|
|
|
|
Rich can generate a rich repr automatically if the parameters are named the same as your attributes.
|
|
|
|
To automatically build a rich repr, use the :meth:`~rich.repr.auto` class decorator. The Bird example above follows the above rule, so we don't strictly need to implement our own ``__rich_repr__``. The following code would generate the same repr::
|
|
|
|
import rich.repr
|
|
|
|
@rich.repr.auto
|
|
class Bird:
|
|
def __init__(self, name, eats=None, fly=True, extinct=False):
|
|
self.name = name
|
|
self.eats = list(eats) if eats else []
|
|
self.fly = fly
|
|
self.extinct = extinct
|
|
|
|
|
|
BIRDS = {
|
|
"gull": Bird("gull", eats=["fish", "chips", "ice cream", "sausage rolls"]),
|
|
"penguin": Bird("penguin", eats=["fish"], fly=False),
|
|
"dodo": Bird("dodo", eats=["fruit"], fly=False, extinct=True)
|
|
}
|
|
from rich import print
|
|
print(BIRDS)
|
|
|
|
Note that the decorator will also create a `__repr__`, so you you will get an auto-generated repr even if you don't print with Rich.
|
|
|
|
If you want to auto-generate the angular type of repr, then set ``angular=True`` on the decorator::
|
|
|
|
@rich.repr.auto(angular=True)
|
|
class Bird:
|
|
def __init__(self, name, eats=None, fly=True, extinct=False):
|
|
self.name = name
|
|
self.eats = list(eats) if eats else []
|
|
self.fly = fly
|
|
self.extinct = extinct
|
|
|
|
|
|
Example
|
|
-------
|
|
|
|
See `repr.py <https://github.com/willmcgugan/rich/blob/master/examples/repr.py>`_ for the example code used in this page.
|