latin/text2pensum_cloze.py

import sys
import re

for filename in sys.argv[1:]:
	with open(filename, 'r') as infile:
		content = infile.read()

	# Replace _string_ with html
	html_span_head = '<span>'
	html_input_head = '<input type="text" data-expected="'
	html_input_tail = '"/>'
	html_span_tail = '</span>'
	content = re.sub(r'(\s|\"|)([a-zA-Z\-\ÿ]+|)_([āēīōūa-zA-Z\-]+)_(\.|\,|)', r'\1' + html_span_head + r'\2' + html_input_head + r'\3' + html_input_tail + r'\4' + html_span_tail, content)

	# Replace vowel-dash-vowel with vowels with macrons
	content = content.replace('A-A', 'Ā')
	content = content.replace('a-a', 'ā')
	content = content.replace('E-E', 'Ē')
	content = content.replace('e-e', 'ē')
	content = content.replace('I-I', 'Ī')
	content = content.replace('i-i', 'ī')
	content = content.replace('O-O', 'Ō')
	content = content.replace('o-o', 'ō')
	content = content.replace('U-U', 'Ū')
	content = content.replace('u-u', 'ū')

	with open(filename + '.html', 'w') as outfile:
		outfile.write(content)