Automatically creates a new .po file with msgstr strings built from msgid strings prefixed by [`lang`]. Goal is ability to check easily where you forgot to add a gettext call.

mock_po_file.py

"""
Based on a .po file `source`, creates a .po file `destination` with msgstr
strings built from msgid strings prefixed by [`lang`]

$ python3 utils/mock_po_file.py --lang FR --source en.po --destination fr.po

Example :

    # en.po
    msgid "Hello"
    msgstr ""

    # fr.po
    msgid "Hello"
    msgstr "[FR] Hello"
"""
import sys
import optparse
from os import path

BLANK_LINE = '\n'
MSGID_START = 'msgid "'
MSGID_MULTI = 'msgid ""\n"'
MSGID_MULTI_START_BY_NL = 'msgid ""\n"\\n'
MSGID_PLURAL_START = 'msgid_plural "'
MSGID_PLURAL_MULTI = 'msgid_plural ""\n"'
MSGID_PLURAL_MULTI_START_BY_NL = 'msgid_plural ""\n"\\n'


def _parse_args():
    """
    Parse the command line for options
    """
    parser = optparse.OptionParser()
    parser.add_option(
        '--lang', dest='lang',
        help='destination language (ex: EN)')
    parser.add_option(
        '--source', dest='source',
        help='Path to .po source file')
    parser.add_option(
        '--destination', dest='destination',
        help='Path to .po destination file')
    options, args = parser.parse_args()

    if not options.lang:
        parser.error('Lang not given')

    if not options.source:
        parser.error('Source file not given')
    elif not path.isfile(options.source):
        parser.error('Source file does not exist')

    if not options.destination:
        parser.error('Destination file not given')
    elif not path.isdir(path.abspath(path.dirname(options.destination))):
        parser.error('Destination directory does not exist')

    return options


def _msgid2msgstr(msgid, is_plural, has_plural, lang):
    """
        msgid can be :
        - classical form (no singular/plural)
        - singular form
        - plural form
        msgid can also :
        - be mono-line
        - be multi-lines and start directly with text
        - be multi-lines and start with a \n
    """
    if not is_plural and not has_plural:
        msgstr_start = 'msgstr'
        start = MSGID_START
        start_multi = MSGID_MULTI
        start_by_nl_multi = MSGID_MULTI_START_BY_NL
    elif not is_plural and has_plural:
        msgstr_start = 'msgstr[0]'
        start = MSGID_START
        start_multi = MSGID_MULTI
        start_by_nl_multi = MSGID_MULTI_START_BY_NL
    elif is_plural and has_plural:
        msgstr_start = 'msgstr[1]'
        start = MSGID_PLURAL_START
        start_multi = MSGID_PLURAL_MULTI
        start_by_nl_multi = MSGID_PLURAL_MULTI_START_BY_NL

    if msgid.startswith(start_by_nl_multi):
        msgstr = msgid.replace(
            start_by_nl_multi,
            '%s ""\n"\\n[%s]' % (msgstr_start, lang))
    elif msgid.startswith(start_multi):
        msgstr = msgid.replace(
            start_multi,
            '%s ""\n"[%s] ' % (msgstr_start, lang))
    else:
        msgstr = msgid.replace(
            start,
            '%s "[%s] ' % (msgstr_start, lang))

    return msgstr


def main():
    """ Fill a po file with msgid preceded by [<lang>] """
    options = _parse_args()

    if path.isfile(options.destination):
        print("/!\ Warning /!\ ")
        print("Do really want to override %s?" % options.destination)
        conf = input('(Y/N) > ')
        if conf != 'Y':
            exit()

    content = ''
    msgid = None
    msgid_plural = None
    in_header = True
    with open(options.source, 'r') as f:
        for i, line in enumerate(f):

            if in_header and line == BLANK_LINE:
                in_header = False
            elif line.startswith('"Language: '):
                content += '"Language: {}\\n"\n'.format(options.lang)
                continue
            elif in_header:
                # keep any line intact in file headers (until first blank line)
                content += '{}'.format(line)
                continue

            if not line.startswith("msgstr"):
                # keep any line intact but msgstr
                content += '{}'.format(line)

            if line.startswith(MSGID_START):
                # We are reading the begining of a msgid string
                # Start collect it
                msgid = '{}'.format(line)
                continue
            elif line.startswith(MSGID_PLURAL_START):
                # We are reading the begining of a msgid plural string
                # Start collect it
                msgid_plural = '{}'.format(line)
                continue
            elif msgid_plural is not None and line.startswith('"'):
                # Continue collecting msgid_plural string
                msgid_plural += '{}'.format(line)
            elif msgid is not None and line.startswith('"'):
                # Continue collecting msgid string
                msgid += '{}'.format(line)
            elif msgid is not None and line.startswith('msgstr'):

                has_plural = msgid_plural is not None

                # Transform msgid into msgstr prefixed by language code
                msgstr = _msgid2msgstr(
                    msgid,
                    False,
                    has_plural,
                    options.lang)
                content += '{}'.format(msgstr)

                if msgid_plural is not None:
                    # If any, transform msgid plural form into msgstr prefixed
                    # by language code
                    msgstr = _msgid2msgstr(
                        msgid_plural,
                        True,
                        has_plural,
                        options.lang)
                    content += '{}'.format(msgstr)

                msgid = None
                msgid_plural = None

            else:
                pass

    # Write destination po file
    with open(options.destination, 'w') as f:
        f.write(content)


if __name__ == '__main__':
    sys.exit(main())