def get_suitable_codepages_for_file(
translation_file: Path): # FIXME: make async
codepages = get_supported_codepages().keys()
with open(translation_file, "r", encoding="utf-8") as fn:
po_file = parse_po.PoReader(fn)
translation_file_language = po_file.meta["Language"]
strings = [cleanup_string(entry.translation) for entry in po_file]
return filter_codepages(codepages, strings), translation_file_language
def prepare_dictionary(
dictionary: Iterable[Tuple[str, str]],
exclusions_leading: Set[str],
exclusions_trailing: Set[str],
) -> Iterable[Tuple[str, str]]:
for original_string, translation in dictionary:
if original_string and translation and translation != original_string:
translation = fix_spaces(original_string, translation,
exclusions_leading, exclusions_trailing)
yield original_string, cleanup_string(translation)
def get_suitable_codepages_for_directory(directory: Path,
language: str): # FIXME: make async
files = filter_files_by_language(directory, language)
codepages = get_supported_codepages().keys()
for file in files:
with open(directory / file, "r", encoding="utf-8") as fn:
po_file = parse_po.PoReader(fn)
strings = [cleanup_string(entry.translation) for entry in po_file]
codepages = filter_codepages(codepages, strings)
return codepages
def translate_single_raw_file(
source_file_path: Path,
destination_file_path: Path,
dictionary: Mapping[Tuple[str, Optional[str]], str],
encoding: str,
) -> Iterator[str]:
with open(source_file_path, encoding="cp437") as src:
with open(destination_file_path, "w", encoding=encoding) as dest:
yield destination_file_path.name
for line in translate_raw_file(src, dictionary):
line = cleanup_string(line)
try:
print(line, file=dest)
except UnicodeEncodeError:
line = line.encode(
encoding, errors="backslashreplace").decode(encoding)
print(
f"Some characters of this line: {line!r} "
f"cannot be represented in {encoding} encoding. Using backslashreplace mode.",
file=sys.stderr,
)
print(line, file=dest)
def test_cleanup_string(original, fixed):
assert cleanup_string(original) == fixed
def cleanup_translations_string(
original: str, translation: str,
exclusions_leading: Optional[Set[str]],
exclusions_trailing: Optional[Set[str]]) -> str:
return fix_spaces(original, cleanup_string(translation),
exclusions_leading, exclusions_trailing)