repair.py

# vim: fileencoding=utf-8 tw=100 expandtab ts=4 sw=4 :
#
# GoodMorningPlanet.com
# author : Vincent Loy <[email protected]>
# (c) 2015 ActivKonnect

"""
Repair bad formatted Markdown Files - replace **foo** by #foo
only when a line start and end by strong tag
"""

import re
import codecs

file_list = ['history', 'politician', 'politics', 'geography', 'money',
             'language', 'culture', 'transport', 'administrativ',
             'health', 'religion', 'festivals', 'food', 'dailylife',
             'luggage', 'mustsee', 'sport']

file_suffix = '_{locale}.md'.format(locale='en')
file_prefix = 'info_'
strong_regex = re.compile(ur'(^\*{2}([aA-zZ0-9_\s]+)\*{2}[\s]?$)', re.MULTILINE)

for file in file_list:
    file_name = '{pref}{name}{suf}'.format(pref=file_prefix, suf=file_suffix, name=file)

    with codecs.open(file_name, 'r', encoding='UTF-8') as chapt:
        original = chapt.read()

    with codecs.open(file_name, 'w+', encoding='UTF-8') as chapt:
        match = re.findall(strong_regex, original)

        if match:
            print("*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*")
            print(file_name)
            for m in match:
                n = '#' + m[1]
                original = original.replace(m[0], n)
                print(m)

                # print(original)
        chapt.write(u'' + original)