I find that when doing bit-twiddling in pure Python,
fixed-width integer support is an extremely handy
capability to have in Python regardless of what
the apologists (for its absence) say.
I added some stuff to fixedint.py to make
x=SByte(80)
x+200 fixedint.Signed Type(8)(24)
the above work. Perhaps you can review the
code to see if there are any major flaws in it?
Right after I started using it, I realized
there has to be a lot more thought put into
the conventions for the coecion rules. What
should
x=SByte(80)
200+x
return for example... ?
----- START OF CODE -----
import operator
def _toFixedUnsigne d(n, b):
"Truncate n to a b-bit unsigned long."
return n & ((1L << b) - 1) # bitmask of b 1's
def _toFixedSigned( n, b):
"Truncate n to a b-bit signed long."
result = _toFixedUnsigne d(n, b)
if result >= (1L << b - 1): # More than "maxint"?
result -= 1L << b # Then wrap around.
return result
class _Integer(object ):
"Abstract base class for SignedType and UnsignedType."
# Numeric conversions
def __long__(self):
return self._value
def __int__(self):
return int(self._value )
def __float__(self) :
return float(self._val ue)
def __nonzero__(sel f):
return self._value != 0
# String conversions.
def __str__(self):
return str(self._value )
# Arithmetic
def __pos__(self):
return self
def __neg__(self):
return type(self)(-self._value)
def __add__(self, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value + other._value)
else:
return type(self)(self ._value + self.__class__( other)._value)
def __sub__(self, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value - other._value)
else:
return type(self)(self ._value - self.__class__( other)._value)
def __mul__(self, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value * other._value)
else:
return type(self)(self ._value * self.__class__( other)._value)
def __floordiv__(se lf, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value // other._value)
else:
return type(self)(self ._value // self.__class__( other)._value)
def __mod__(self, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value % other._value)
else:
return type(self)(self ._value % self.__class__( other)._value)
def __divmod__(self , other):
return self // other, self % other
# Relational
def __cmp__(self, other):
return cmp(long(self), other)
# Bit-bashing
def __lshift__(self , other):
return type(self)(self ._value << other)
def __rshift__(self , other):
return type(self)(self ._value >> other)
def __invert__(self ):
return type(self)(~sel f._value)
def __and__(self, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value & other._value)
else:
return type(self)(self ._value & self.__class__( other)._value)
def __or__(self, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value | other._value)
else:
return type(self)(self ._value | self.__class__( other)._value)
def __xor__(self, other):
if isinstance(othe r, type(self)):
return type(self)(self ._value ^ other._value)
else:
return type(self)(self ._value ^ self.__class__( other)._value)
_utypes = {}
def UnsignedType(bi ts):
"Returns a fixed-width unsigned int type with the given number of bits."
if bits in _utypes:
return _utypes[bits]
else:
class unsigned(_Integ er):
__doc__ = '%d-bit unsigned integer type' % bits
def __init__(self, value):
self._value = _toFixedUnsigne d(value, bits)
def __repr__(self):
return 'fixedint.Unsig nedType(%d)(%d) ' % (bits, self._value)
return unsigned
Byte = UnsignedType(8)
UShort = UnsignedType(16 )
UInt = UnsignedType(32 )
ULong = UnsignedType(64 )
_stypes = {}
def SignedType(bits ):
"Returns a fixed-width signed int type with the given number of bits."
if bits in _stypes:
return _stypes[bits]
else:
class signed(_Integer ):
__doc__ = '%d-bit signed integer type' % bits
def __init__(self, value):
self._value = _toFixedSigned( value, bits)
def __repr__(self):
return 'fixedint.Signe dType(%d)(%d)' % (bits, self._value)
return signed
SByte = SignedType(8)
Short = SignedType(16)
Int = SignedType(32)
Long = SignedType(64)
