use string.translate in Python to transliterate Cyrillic?

Question

I'm getting UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-51: ordinal not in range(128) exception trying to use string.maketrans in Python. I'm kinda discouraged with this kind of error in following code (gist):

# -*- coding: utf-8 -*-

import string

def translit1(string):
    """ This function works just fine """
    capital_letters = {
        u'А': u'A',
        u'Б': u'B',
        u'В': u'V',
        u'Г': u'G',
        u'Д': u'D',
        u'Е': u'E',
        u'Ё': u'E',
        u'Ж': u'Zh',
        u'З': u'Z',
        u'И': u'I',
        u'Й': u'Y',
        u'К': u'K',
        u'Л': u'L',
        u'М': u'M',
        u'Н': u'N',
        u'О': u'O',
        u'П': u'P',
        u'Р': u'R',
        u'С': u'S',
        u'Т': u'T',
        u'У': u'U',
        u'Ф': u'F',
        u'Х': u'H',
        u'Ц': u'Ts',
        u'Ч': u'Ch',
        u'Ш': u'Sh',
        u'Щ': u'Sch',
        u'Ъ': u'',
        u'Ы': u'Y',
        u'Ь': u'',
        u'Э': u'E',
        u'Ю': u'Yu',
        u'Я': u'Ya'
    }

    lower_case_letters = {
        u'а': u'a',
        u'б': u'b',
        u'в': u'v',
        u'г': u'g',
        u'д': u'd',
        u'е': u'e',
        u'ё': u'e',
        u'ж': u'zh',
        u'з': u'z',
        u'и': u'i',
        u'й': u'y',
        u'к': u'k',
        u'л': u'l',
        u'м': u'm',
        u'н': u'n',
        u'о': u'o',
        u'п': u'p',
        u'р': u'r',
        u'с': u's',
        u'т': u't',
        u'у': u'u',
        u'ф': u'f',
        u'х': u'h',
        u'ц': u'ts',
        u'ч': u'ch',
        u'ш': u'sh',
        u'щ': u'sch',
        u'ъ': u'',
        u'ы': u'y',
        u'ь': u'',
        u'э': u'e',
        u'ю': u'yu',
        u'я': u'ya'
    }

    translit_string = ""

    for index, char in enumerate(string):
        if char in lower_case_letters.keys():
            char = lower_case_letters[char]
        elif char in capital_letters.keys():
            char = capital_letters[char]
            if len(string) > index+1:
                if string[index+1] not in lower_case_letters.keys():
                    char = char.upper()
            else:
                char = char.upper()
        translit_string += char

    return translit_string


def translit2(text):
    """ This method should be more easy to grasp, 
    but throws exception:
    UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-51: ordinal not in range(128)
    """

    symbols = string.maketrans(u"абвгдеёзийклмнопрстуфхъыьэАБВГДЕЁЗИЙКЛМНОПРСТУФХЪЫЬЭ",
                               u"abvgdeezijklmnoprstufh'y'eABVGDEEZIJKLMNOPRSTUFH'Y'E")
    sequence = {
        u'ж':'zh',
        u'ц':'ts',
        u'ч':'ch',
        u'ш':'sh',
        u'щ':'sch',
        u'ю':'ju',
        u'я':'ja',
        u'Ж':'Zh',
        u'Ц':'Ts',
        u'Ч':'Ch'
    }

    for char in sequence.keys():
        text = text.replace(char, sequence[char])

    return text.translate(symbols)

if __name__ == "__main__":
    print translit1(u"Привет") # prints Privet as expected
    print translit2(u"Привет") # throws exception: UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-51: ordinal not in range(128)

Original trace:

Traceback (most recent call last):
  File "translit_error.py", line 124, in <module>
    print translit2(u"Привет") # throws exception: UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-51: ordinal not in range(128)
  File "translit_error.py", line 103, in translit2
    u"abvgdeezijklmnoprstufh'y'eABVGDEEZIJKLMNOPRSTUFH'Y'E")
UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-51: ordinal not in range(128)

I mean, why Python string.maketrans trying to use ascii table anyway? And how comes English alphabet letters are out of 0-128 range?

$ python -c "print ord(u'A')"
65
$ python -c "print ord(u'z')"
122
$ python -c "print ord(u\"'\")"
39

After several hours I feel like absolutely exhausted to solve this issue.

Can someone say what is happening and how to fix it?

Answer 1

translate behaves differently when used with unicode strings. Instead of a maketrans table, you have to provide a dictionary ord(search)->ord(replace):

symbols = (u"абвгдеёжзийклмнопрстуфхцчшщъыьэюяАБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ",
           u"abvgdeejzijklmnoprstufhzcss_y_euaABVGDEEJZIJKLMNOPRSTUFHZCSS_Y_EUA")

tr = {ord(a):ord(b) for a, b in zip(*symbols)}

# for Python 2.*:
# tr = dict( [ (ord(a), ord(b)) for (a, b) in zip(*symbols) ] )

text = u'Добрый Ден'
print text.translate(tr)  # looks good

That said, I'd second the suggestion not to reinvent the wheel and to use an established library: http://pypi.python.org/pypi/Unidecode

Answer 2

You can use transliterate package (https://pypi.python.org/pypi/transliterate)

Example #1:

from transliterate import translit
print translit("Lorem ipsum dolor sit amet", "ru")
# Лорем ипсум долор сит амет

Example #2:

print translit(u"Лорем ипсум долор сит амет", "ru", reversed=True)
# Lorem ipsum dolor sit amet

Answer 3

Check out the CyrTranslit package, it's specifically made to transliterate from and to Cyrillic script text. It currently supports Serbian, Montenegrin, Macedonian, and Russian.

Example usage:

>>> import cyrtranslit
>>> cyrtranslit.supported()
['me', 'sr', 'mk', 'ru']

>>> cyrtranslit.to_latin('Моё судно на воздушной подушке полно угрей', 'ru')
'Moyo sudno na vozdushnoj podushke polno ugrej'

>>> cyrtranslit.to_cyrillic('Moyo sudno na vozdushnoj podushke polno ugrej')
'Моё судно на воздушной подушке полно угрей'

Answer 4

Here is another short solution with more accurate transliteration:

symbols = (u"абвгдеёжзийклмнопрстуфхцчшщъыьэюяАБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ ",
    (*list(u'abvgdee'), 'zh', *list(u'zijklmnoprstuf'), 'kh', 'z', 'ch', 'sh', 'sh', '',
    'y', '', 'e', 'yu','ya', *list(u'ABVGDEE'), 'ZH', 
    *list(u'ZIJKLMNOPRSTUF'), 'KH', 'Z', 'CH', 'SH', 'SH', *list(u'_Y_E'), 'YU', 'YA', ' '))

coding_dict = {source: dest for source, dest in zip(*symbols)}
translate = lambda x: ''.join([coding_dict[i] for i in x])

text = u'Добро пожаловать'
translate(text)
# 'Dobro pozhalovat'