def _fix_encode_to_binary(tokens: List[Token], i: int) -> None:
parts = rfind_string_parts(tokens, i - 2)
if not parts:
return
# .encode()
if (
i + 2 < len(tokens) and
tokens[i + 1].src == '(' and
tokens[i + 2].src == ')'
):
victims = slice(i - 1, i + 3)
latin1_ok = False
# .encode('encoding')
elif (
i + 3 < len(tokens) and
tokens[i + 1].src == '(' and
tokens[i + 2].name == 'STRING' and
tokens[i + 3].src == ')'
):
victims = slice(i - 1, i + 4)
prefix, rest = parse_string_literal(tokens[i + 2].src)
if 'f' in prefix.lower():
return
encoding = ast.literal_eval(prefix + rest)
if is_codec(encoding, 'ascii') or is_codec(encoding, 'utf-8'):
latin1_ok = False
elif is_codec(encoding, 'iso8859-1'):
latin1_ok = True
else:
return
else:
return
for part in parts:
prefix, rest = parse_string_literal(tokens[part].src)
escapes = set(ESCAPE_RE.findall(rest))
if (
not is_ascii(rest) or
'\\u' in escapes or
'\\U' in escapes or
'\\N' in escapes or
('\\x' in escapes and not latin1_ok) or
'f' in prefix.lower()
):
return
for part in parts:
prefix, rest = parse_string_literal(tokens[part].src)
prefix = 'b' + prefix.replace('u', '').replace('U', '')
tokens[part] = tokens[part]._replace(src=prefix + rest)
del tokens[victims]
def visit_Call(
state: State,
node: ast.Call,
parent: ast.AST,
) -> Iterable[Tuple[Offset, TokenFunc]]:
if (state.settings.min_version >= (3, )
and isinstance(node.func, ast.Attribute)
and isinstance(node.func.value, ast.Str)
and node.func.attr == 'encode' and not has_starargs(node)
and len(node.args) == 1 and isinstance(node.args[0], ast.Str)
and is_codec(node.args[0].s, 'utf-8')):
yield ast_to_offset(node), _fix_default_encoding