A small program to look through a Python source directory, pull out the class hierarchy, and generate a .dot file. It also produces a .pdf file using the dot program.

dot.py

#!/usr/bin/python

import re
import networkx as nx
import os

prog = re.compile("class (?P<subclass>\w+)\((?P<superclasses>[\s\w,]+)\):")

def grep_stream(location):
    return os.popen(
    'find %s -name "*.py" | xargs grep -h "^class \w*\([\w\s][\s\w,]*\)"'%location)

def nx_from_stream(f):
    assert isinstance(f, file)
    G = nx.DiGraph()
    for line in f:
        match = prog.match(line)
        if not match:
            continue
        subclass = match.groupdict()['subclass']
        for superclass in match.groupdict()['superclasses'].split(','):
            G.add_edge(superclass.strip(), subclass)
    return G

if __name__ == "__main__":
    from sys import argv, exit
    if len(argv) == 3:
        location, name = argv[1:]
    else:
        print "Usage:\n%s sympy/core/ core\nevince core.pdf"%argv[0]
        exit()

    stream = grep_stream(location)
    G = nx_from_stream(stream)
    Gpd = nx.to_pydot(G)
    Gpd.set_rankdir("LR")
    Gpd.set_overlap("false")
    Gpd.write("%s.dot"%name)
    os.system("dot -Tpdf %s.dot -o %s.pdf"%(name, name))

To produce a dot file this requires:

• Standard Unix utilities find, xargs, grep
• python libraries networkx, pydot

To produce a pdf this requires

• dot (the graphvis executable)

Here are some examples of the sympy code repository

Here are some examples of the sympy code repository

(the previous link was relative)