In Python, objects can declare their textual representation using the __repr__ method. IPython expands on this idea and allows objects to declare other, richer representations including:
This Notebook shows how you can add custom display logic to your own classes, so that they can be displayed using these rich representations. There are two ways of accomplishing this:
_repr_html_.In this Notebook we show how both approaches work.
Before we get started, we will import the various display functions for displaying the different formats we will create.
In [1]:
from IPython.display import display
from IPython.display import (
display_html, display_jpeg, display_png,
display_javascript, display_svg, display_latex
)
The main idea of the first approach is that you have to implement special display methods, one for each representation you want to use. Here is a list of the names of the special methods and the values they must return:
_repr_html_: return raw HTML as a string_repr_json_: return raw JSON as a string_repr_jpeg_: return raw JPEG data_repr_png_: return raw PNG data_repr_svg_: return raw SVG data as a string_repr_latex_: return LaTeX commands in a string surrounded by "$".A prominent example of a package that has IPython-aware rich representations of its objects is pandas.
A pandas DataFrame has a rich HTML table representation,
using _repr_html_.
In [2]:
import io
import pandas
In [3]:
%%writefile data.csv
Date,Open,High,Low,Close,Volume,Adj Close
2012-06-01,569.16,590.00,548.50,584.00,14077000,581.50
2012-05-01,584.90,596.76,522.18,577.73,18827900,575.26
2012-04-02,601.83,644.00,555.00,583.98,28759100,581.48
2012-03-01,548.17,621.45,516.22,599.55,26486000,596.99
2012-02-01,458.41,547.61,453.98,542.44,22001000,540.12
2012-01-03,409.40,458.24,409.00,456.48,12949100,454.53
In [4]:
df = pandas.read_csv("data.csv")
pandas.set_option('display.notebook_repr_html', False)
df
Out[4]:
rich HTML can be activated via pandas.set_option.
In [5]:
pandas.set_option('display.notebook_repr_html', True)
df
Out[5]:
In [6]:
lines = df._repr_html_().splitlines()
print "\n".join(lines[:20])
Write a simple Circle Python class. Don't even worry about properties such as radius, position, colors, etc. To help you out use the following representations (remember to wrap them in Python strings):
For HTML:
○
For SVG:
<svg width="100px" height="100px">
<circle cx="50" cy="50" r="20" stroke="black" stroke-width="1" fill="white"/>
</svg>
For LaTeX (wrap with $ and use a raw Python string):
\bigcirc
After you write the class, create an instance and then use display_html, display_svg and display_latex to display those representations.
Tips : you can slightly tweek the representation to know from which _repr_*_ method it came from.
For example in my solution the svg representation is blue, and the HTML one show "HTML" between brackets.
Here is my simple MyCircle class:
In [8]:
%load soln/mycircle.py
Now create an instance and use the display methods:
In [11]:
c = MyCircle()
In [12]:
display_html(c)
In [13]:
display_svg(c)
In [14]:
display_latex(c)
In [15]:
display_javascript(c)
When you are directly writing your own classes, you can adapt them for display in IPython by following the above example. But in practice, we often need to work with existing code we can't modify. We now illustrate how to add these kinds of extended display capabilities to existing objects. To continue with our example above, we will add a PNG representation to our Circle class using Matplotlib.
SymPy is another model citizen that defines rich representations of its object.
Unlike pandas above, sympy registers display formatters via IPython's display formatter API, rather than declaring _repr_mime_ methods.
In [16]:
from sympy import Rational, pi, exp, I, symbols
x, y, z = symbols("x y z")
In [17]:
r = Rational(3,2)*pi + exp(I*x) / (x**2 + y)
r
Out[17]:
SymPy provides an init_printing function that sets up advanced $\LaTeX$
representations of its objects.
In [18]:
from sympy.interactive.printing import init_printing
init_printing()
r
Out[18]:
To add a display method to an existing class, we must use IPython's display formatter API. Here we show all of the available formatters:
In [6]:
ip = get_ipython()
for mime, formatter in ip.display_formatter.formatters.items():
print '%24s : %s' % (mime, formatter.__class__.__name__)
Let's grab the PNG formatter:
In [20]:
png_f = ip.display_formatter.formatters['image/png']
We will use the for_type method to register our display function.
In [21]:
png_f.for_type?
As the docstring describes, we need to define a function the takes the object as a parameter and returns the raw PNG data.
In [22]:
%matplotlib inline
import matplotlib.pyplot as plt
In [23]:
class AnotherCircle(object):
def __init__(self, radius=1, center=(0,0), color='r'):
self.radius = radius
self.center = center
self.color = color
def __repr__(self):
return "<%s Circle with r=%s at %s>" % (
self.color,
self.radius,
self.center,
)
c = AnotherCircle()
c
Out[23]:
In [24]:
from IPython.core.pylabtools import print_figure
def png_circle(circle):
"""Render AnotherCircle to png data using matplotlib"""
fig, ax = plt.subplots()
patch = plt.Circle(circle.center,
radius=circle.radius,
fc=circle.color,
)
ax.add_patch(patch)
plt.axis('scaled')
data = print_figure(fig, 'png')
# We MUST close the figure, otherwise IPython's display machinery
# will pick it up and send it as output, resulting in a double display
plt.close(fig)
return data
In [27]:
c = AnotherCircle()
print repr(png_circle(c)[:10])
Now we register the display function for the type:
In [28]:
png_f.for_type(AnotherCircle, png_circle)
Now all Circle instances have PNG representations!
In [29]:
c2 = AnotherCircle(radius=2, center=(1,0), color='g')
c2
Out[29]:
In [30]:
display_png(c2)
In [ ]:
# for demonstration purpose, I do the same with a circle that has no _repr_javascript method
class MyNoJSCircle(MyCircle):
def _repr_javascript_(self):
return
cNoJS = MyNoJSCircle()
Of course you can now still return the object, and this will use compute all the representations, store them in the notebook and show you the appropriate one.
In [ ]:
cNoJS
Or just use display(object) if you are in a middle of a loop
In [ ]:
for i in range(3):
display(cNoJS)
Advantage of using display() versus display_*() is that all representation will be stored in the notebook document and notebook file, they are then availlable for other frontends or post-processing tool like nbconvert.
Let's compare display() vs display_html() for our circle in the Notebook Web-app and we'll see later the difference in nbconvert.
In [ ]:
print "I should see a nice html circle in web-app, but"
print "nothing if the format I'm viewing the notebook in"
print "does not support html"
display_html(cNoJS)
In [ ]:
print "Whatever the format I will see a representation"
print "of my circle"
display(cNoJS)
In [ ]:
print "Same if I return the object"
cNoJS
In [ ]:
print "But not if I print it"
print cNoJS