Script to aid on a "reverse translation" FROM [a portuguese codebase with no i18n] TO [an english one with i18n]

i18n_from_portuguese.py

# Script to aid on a "reverse translation":
#  * FROM a portuguese codebase with no i18n
#  * TO an english one with i18n
#
# Process:
#
# In the cenario wher your have all your code with portuguese strings and no i18n
# (Naturally that applies to other languages as well)
#
# 1. Mark all your translatable portuguese strings with gettext marks _(...)
# 2. generate a temporary .po file and save it as version "pt"
# 3. translate to english inplace (that means in the files itself)
# 4. generate another temporary .po file and save it as version "en"
# 5. call translate(...) to generate the final .po file from the previous two


def read_lines_from_file(filename):
    with open(filename, "r") as f:
        return f.readlines()


def write_lines_to_file(filename, content):
    with open(filename, "w") as text_file:
        text_file.writelines(content)


def translated_sequence(en, pt):
    translations = dict((a, b.replace('msgid', 'msgstr'))
        for a, b in zip(en, pt)
        if a != b and a.startswith('msgid'))
    en = iter(en)
    line = next(en, None)
    while line:
        match = translations.get(line, None)
        if match:
            yield line
            assert next(en) == 'msgstr ""\n'  # we skip this one
            yield match
        else:
            yield line
        line = next(en, None)


def translate(en_filename, pt_filename, final_filename):
    en, pt = read_lines_from_file(en_filename), read_lines_from_file(pt_filename)
    write_lines_to_file(final_filename, translated_sequence(en, pt))