convert-inno-setup

#!/usr/bin/env python3
import subprocess
import argparse
import tempfile
import os
import io
import re
import shutil

parser = argparse.ArgumentParser(description="""Convert Inno Setup localization
between UTF-8, which Transifex supports, to the various Windows code pages
Inno Setup requires.

TODO: implement Windows to UTF-8.""")
parser.add_argument("wininstaller_translations_dir")
args = parser.parse_args()

# TODO: pycountry exposes Debian's ISO 639 language database, but it lacks
# questionbly-standard language codes that Transifex uses like zh_CN and id_ID.

# TODO: Is there an existing Windows encoding list?
# Windows code pages by language - see the Inno Setup LanguageCodePage value or
# https://en.wikipedia.org/wiki/Windows_code_page#List
# 0 is ASCII:
# See http://www.jrsoftware.org/ishelp/index.php?topic=langoptionssection
code_pages = {
    "bg": 1251,  # Bulgarian
    "cs": 1250,  # Czech
    "de": 1252,  # German
    "el": 1253,  # Greek
    "en": 0,  # English
    "es": 1252,  # Spanish
    "fi": 1252,  # Finnish
    "fr": 1252,  # French
    "hr": 1250,  # Croatian
    "hu": 1250,  # Hungarian
    "id_ID": 0,  # Indonesian
    "it": 1252,  # Italian
    "ja": 932,   # Japanese
    "nl": 1252,  # Dutch
    "pl": 1250,  # Polish
    "pt_BR": 1252,  # Brazilian Portuguese
    "pt_PT": 1252,  # Portuguese (Portugal)
    "ru": 1251,  # Russian
    "sr": 1251,  # Serbian
    "zh_CN": 936,  # Simplified Chinese
    "zh_TW": 950,  # Traditional Chinese
}

utf8_encoded = re.compile("Messages_([a-zA-Z_]+)_utf8\.isl")
windows_template = "Messages_{}.isl"

os.chdir(os.path.abspath(args.wininstaller_translations_dir))

converted = []

for filename in os.listdir("."):
    match = utf8_encoded.match(filename)
    if match:
        # Unexpected to have a directory with such a name; could just ignore
        # if need be.
        assert os.path.isfile(filename)

        language_code = match.group(1)
        if language_code not in code_pages:
            print("Language code '{}' encoding not found; add it to "
                  "code_pages.".format(language_code))
            exit(1)
        code_page = code_pages[language_code]
        encoding = "CP{}".format(code_page) if code_page else "ASCII"

        # Remove a UTF-8 byte-order mark if present; iconv does not support it.
        temp_file = None
        with open(filename, 'rb') as file:
            file_stream = io.BufferedReader(file)
            has_bom = file_stream.read(3) == bytearray.fromhex("EFBBBF")
            if has_bom:
                temp_file = tempfile.NamedTemporaryFile()
                temp_file.file.write(file_stream.read())
                temp_file.file.flush()

        source = temp_file.name if temp_file else filename
        target = windows_template.format(language_code)
        subprocess.check_call(["iconv", "--from-code", "UTF-8",
                               "--to-code", encoding, "--output", target,
                               source])

        if temp_file:
            temp_file.close()

        converted.append(language_code)

print("Converted {} files: {}".format(len(converted), sorted(converted)))