The inheritance_explorer
is an open source python package for inspecting
other packages. In particular, it focuses on class inheritance structures, allowing
you to produce graphs that recursively map out child classes from a given
starting class. There are a number of useful applications: from learning structures
of unfamiliar code bases to identifying duplicate code.
Useful links: github repository, full documentation
To install the latest release into your current active environment:
$ pip install inheritance_explorer
If you want to produce static images of any graphs, you'll also need to
install graphviz
following the instructions at
https://graphviz.org/download/. graphviz
is not needed for the interactive plotting and similarity calculations (see
Quick Start below).
The primary function of the inheritance_explorer
is to trace out the class
inheritance starting from a single parent class and recursively traversing all
child classes. To get started, initialize a ClassGraphTree with the starting
class:
from matplotlib.axes import Axes
from inheritance_explorer import ClassGraphTree
base_class = Axes # the starting class to map from
cgt = ClassGraphTree(base_class) # traces an inheritance tree
From here, there are a number of ways to visualize the inheritance graph.
- From a jupyter notebook, you can construct a
pyvis
interactive network graph - with:
graph = cgt.build_interactive_graph(width="1200px",
height="1200px",
bgcolor='#222222',
font_color='white')
graph.show('_tmp.html') # render the pyvis interactive graph
- If you have
graphviz
installed, you can also render a static graph either to - display in a jupyter notebook or save to file. To display in a notebook:
cgt.show_graph()
To save a file, you can access the underlying graph object directly:
cgt.graph.write('matplotlib_Axes.png')
The ClassGraphTree
can also track a selected function of a class during traversal.
When a child class overrides the function, the source code is stored. To use this
functionality, use the funcname
keyword argument to provide the function name
as a string:
from matplotlib.axes import Axes
from inheritance_explorer import ClassGraphTree
base_class = Axes # the starting class to map from
cgt = ClassGraphTree(base_class, funcname='clear')
By default, after the traversal completes, the different versions of the function that is tracked will be run through a code-similarity calculation. Cases where the similarity fraction is above a cutoff value (default of .75), the two nodes will be connected on the resulting graph:
Additionally, you can inspect the source code itself for any of the classes that
define it using cgt.get_source_code(class_name)
. If in a jupyter notebook,
you can view it with syntax highlighting with
from IPython.display import Code
Code(cgt.get_source_code('PolarAxes'), language="python")
For a complete description of the code similarity calculation and how to modify it, check out the full documentation.
The following additional features are described in more detail in the full documentation: