repetition.py

import sys

if len(sys.argv) != 2:
	exit(1)

text = open(sys.argv[1], 'r').read()

words = text.replace("\n", " ").split(" ")

for i in range(1, len(words)):
	prev_word = words[i-1]
	curr_word = words[i]

	if prev_word == curr_word:
		if curr_word in ['', ' ']: # skip some leftovers
			continue

		# print word and context
		print('repeated word (', curr_word, ')', 'line: ', end='\n\t')
		for j in range(5):
			if (i-3+j) > 0:
				print(words[i-3+j], end=' ')
		print('\n')