Compute histogram of method line count

compute_function_length_hist.py

import ast
import os
import matplotlib

matplotlib.use('Agg')

import matplotlib.pyplot as plt
from collections import defaultdict

def walk_dir(dirname, ext):
  for root, dirs, files in os.walk(dirname):
    for file in files:
      if file.endswith(ext):
        yield os.path.join(root, file)

def merge_dicts(dict1, dict2):
    merged_keys = list(set(dict1.keys()) | set(dict2.keys()))
    merged = {}
    for k in merged_keys:
        merged[k] = dict1.get(k, 0) + dict2.get(k, 0)
    return merged

def count_function_lengths(pth):
    # load file
    f = open(pth, 'r')

    # parse the AST of the file
    t = ast.parse(f.read())

    # count the length of the file
    f.seek(0)
    file_length = 0
    for line in f:
        file_length += 1

    classes = [e for e in t.body if type(e) == ast.ClassDef]
    funcs   = [f for f in t.body if type(f) == ast.FunctionDef]

    # get the starting line for each function and method
    for kls in classes:
        methods = [f for f in kls.body if type(f) == ast.FunctionDef]
        funcs.extend(methods)

    line_numbers = [f.lineno for f in funcs]
    line_numbers.append(file_length)

    # sort just to be sure
    line_numbers = sorted(line_numbers)

    # count function line lengths
    counts = defaultdict(int)

    # prepare histogram
    for ix, ln in enumerate(line_numbers[0:-1]):
        next_ln = line_numbers[ix+1]
        function_length = next_ln - ln
        counts[function_length] += 1

    return counts

hist = {}
for filename in walk_dir('YOUR_PROJECT_DIRECTORY_HERE', '.py'):
    counts = count_function_lengths(filename)
    hist = merge_dicts(hist, counts)

# print hist

# plot the histogram
plt.bar(hist.keys(), hist.values(), width=1, color='b')
plt.savefig('hist.png')%