Units of measurement

[Paul Rubin]

I'm sure this has been done before, but it just struck my fancy, an
example of Python's "emulating numeric types", inspired by the old
Unix "units" utility, and the Frink language. The language reference
for these features is somewhat unclear and I'll enter some sourceforge
items. Maybe I'll write a calculator-with-units around it. If I were
really industrious, I'd attempt a Frink-like interpreter, but don't
hold your breath.

================================================== ==============

try:
set
except NameError:
from sets import Set as set # python 2.3 ...

class Unit:
def __init__(self, coef, dimensions):
self.dimensions = dimensions
self.coef = coef
def __add__(self, other):
if self.dimensions != other.dimensions:
raise TypeError, ('dimension mismatch', self, other)
return Unit(self.coef + other.coef, self.dimensions)
def __sub__(self,other):
return self + (-1.0) * other
def __rmul__(self, x):
return x*self
def __rdiv__(self, x):
return x/self
def __mul__(self, other):
pc = self.coef * other.coef
sd, od = self.dimensions, other.dimensions
basis = set(sd.keys()) | set(od.keys())
pa = [(d, sd.get(d,0)+od.get(d,0)) for d in basis]
pd = dict([(d,a) for d,a in pa if a != 0])
return Unit(pc, pd)
def __repr__(self):
a = [repr(self.coef)]
for d,c in self.dimensions.iteritems():
s = str(d)
if c != 1: s += '^'+str(c)
a.append(s)
return '*'.join(a)
def __div__(self, other):
od = other.dimensions
inv = Unit(1.0 / other.coef,
dict([(d, -od[d]) for d in od]))
return self * inv
def __pow__(self, n):
n = n.coef
if self.dimensions and type(n) not in (int,long):
raise TypeError, ('exponent must be integer', self,n)
cn = self.coef ** n
sd = self.dimensions
return Unit(self.coef ** n,
dict([(d,sd[d]*n) for d in sd]))

def __coerce__(self, other):
if isinstance(other, Unit): return self, other
return self, Unit(other, {})

def __float__(self):
if self.dimensions:
raise TypeError, ('unit must be dimensionless for float cast', self)
return float(self.coef)
def __int__(self):
return int(float(self))

def base_unit(name):
return Unit(1.0, {name : 1})

meter = base_unit('m')
second = base_unit('s')
kg = base_unit('kg')
coulomb = base_unit('coulomb')

centimeter = meter / 100
inch = 2.54 * centimeter
foot = ft = 12 * inch
mile = 5280*foot
minute=60*second
hour=60*minute
speed_limit = 55 * mile / hour
furlong = mile / 8
day = 24 * hour
fortnight = 14 * day
# could include more units but you get the idea

c = 186282*mile/second
print 'speed of light =', c/(furlong/fortnight), 'furlongs per fortnight'
# ...

[Robert Kern]

Paul Rubin wrote:

I'm sure this has been done before, but it just struck my fancy, an
example of Python's "emulating numeric types", inspired by the old
Unix "units" utility, and the Frink language.

Oh yeah, it's been done before. Several times over, in fact.

Unum
http://home.tiscali.be/be052320/Unum.html

Caltech's pyre.units
http://www.cacr.caltech.edu/projects/pyre/

Konrad Hinsen's Scientific.Physics.PhysicalQuantity
http://dirac.cnrs-orleans.fr/ScientificPython/

Will Ware posted one to the list.
http://groups.google.com/group/comp....05aea353c1563d

And there was another one announced here sometime in the past year or so, IIRC,
but I don't recall the name of it or that of the author. :-(

--
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
that is made terrible by our own mad attempt to interpret it as though it had
an underlying truth."
-- Umberto Eco

[Russ]

Robert Kern wrote:

Paul Rubin wrote:

I'm sure this has been done before, but it just struck my fancy, an
example of Python's "emulating numeric types", inspired by the old
Unix "units" utility, and the Frink language.

Oh yeah, it's been done before. Several times over, in fact.

<cut>

And there was another one announced here sometime in the past year or so, IIRC,
but I don't recall the name of it or that of the author. :-(

Perhaps you are referring to the scalar class at
http://RussP.us/scalar.htm

The nice thing about this one (which I wrote myself) is that after you
do your development and are ready for "production" runs, you can easily
disable the unit checks and get the efficiency of built-in numeric
types. That can be two orders of magnitude faster!

It comes with a complete user manual too (pdf and html). I'm using it
for my engineering work, and it's working great!

[Robert Kern]

Russ wrote:

Robert Kern wrote:

>And there was another one announced here sometime in the past year or so, IIRC,
but I don't recall the name of it or that of the author. :-(

Perhaps you are referring to the scalar class at
http://RussP.us/scalar.htm

Yup, you're the one.

--
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
that is made terrible by our own mad attempt to interpret it as though it had
an underlying truth."
-- Umberto Eco

[Tim Roberts]

Paul Rubin <http://ph****@NOSPAM.invalidwrote:

>
I'm sure this has been done before, but it just struck my fancy, an
example of Python's "emulating numeric types", inspired by the old
Unix "units" utility, and the Frink language.
...
# could include more units but you get the idea
...
c = 186282*mile/second
print 'speed of light =', c/(furlong/fortnight), 'furlongs per fortnight'
# ...

I could not skip to the next message without cutting, pasting, and
executing that script to find out what the speed of light actually was in
furlongs per fortnight.

Now I need to figure out how to work that into a cocktail party
conversation. "Hey, the deficit isn't the only thing that is approaching
1.8 trillion..."
--
Tim Roberts, ti**@probo.com
Providenza & Boekelheide, Inc.