dividing tuple elements with an int or float

[royG]

hi
i am trying to resize some images.First i'd read the size as a 2
tuple and then i want to divide it by 2 or 4 or 2.5 etc..

suppose
origsz=(400,300 )
i want to divide the origsize by 2.5 so i can resize to (160,120)

scale=2.5
how can i get the newsz?
obviously origsz/2.5 won't work ..
thanks
RG

[Steven D'Aprano]

Re: dividing tuple elements with an int or float

On Wed, 19 Mar 2008 23:06:44 -0700, royG wrote:

suppose
origsz=(400,300 )
i want to divide the origsize by 2.5 so i can resize to (160,120)
>
scale=2.5
how can i get the newsz?
obviously origsz/2.5 won't work ..

newsz = (origsz[0]/scale, origsz[1]/scale)




--
Steven

[Jerry Hill]

Re: dividing tuple elements with an int or float

On Thu, Mar 20, 2008 at 3:42 AM, Steven D'Aprano
<steve@remove-this-cybersource.com .auwrote:

On Wed, 19 Mar 2008 23:06:44 -0700, royG wrote:
>
>

suppose
origsz=(400,300 )
i want to divide the origsize by 2.5 so i can resize to (160,120)
>
scale=2.5
how can i get the newsz?
obviously origsz/2.5 won't work ..

>
newsz = (origsz[0]/scale, origsz[1]/scale)

That works fine for a 2-tuple, but might get unwieldy for larger
tuples, or if you don't know the length until runtime. A more general
solution might use a generator expression, like this:

newsz = tuple(x/scale for x in origsz)

--
Jerry

[castironpi@gmail.com]

Re: dividing tuple elements with an int or float

On Mar 20, 9:28 am, "Jerry Hill" <malaclyp...@gm ail.comwrote:

On Thu, Mar 20, 2008 at 3:42 AM, Steven D'Aprano
>
<st...@remove-this-cybersource.com .auwrote:

On Wed, 19 Mar 2008 23:06:44 -0700, royG wrote:

>

 suppose
 origsz=(400,300 )
 i want to divide the origsize by 2.5 so i can resize to (160,120)

>

 scale=2.5
 how can i get the newsz?
 obviously origsz/2.5 won't work  ..

>

 newsz = (origsz[0]/scale, origsz[1]/scale)

>
That works fine for a 2-tuple, but might get unwieldy for larger
tuples, or if you don't know the length until runtime.  A more general
solution might use a generator expression, like this:
>
newsz = tuple(x/scale for x in origsz)

You want to perform a uniform call on the elements of a collection.
"Diagram A" appends 0 to every item of a list.

>>y= [ [] for k in range( 10 ) ]
>>def x( fun, *ar, **kw ):

... def post( *xr, **xw ):
... return fun( *(xr+ ar), **kw )
... return post
...

>>x( list.append, 0 )( y[0] )
>>y

[[0], [], [], [], [], [], [], [], [], []]

>>x( list.pop, 0 )( y[0] )

0

>>y

[[], [], [], [], [], [], [], [], [], []]

>>list( map( x( list.append, 0 ), y ) )

[None, None, None, None, None, None, None, None, None, None]

>>y

[[0], [0], [0], [0], [0], [0], [0], [0], [0], [0]]

If the elements are immutable,

>>y= [ () for k in range( 10 ) ]
>>list( map( x( tuple.__add__, ( 0, ) ), y ) )

[(0,), (0,), (0,), (0,), (0,), (0,), (0,), (0,), (0,), (0,)]

you get your new list from map. Here it is with integer elements:

>>y= ( 300, 400 )
>>tuple( map( x( int.__add__, 1, ), y ) )

(301, 401)

--You could spell the call like this:
latemap( y, int.__add__, 1 )

With mul:

>>tuple( map( x( int.__mul__, 2, ), y ) )

(600, 800)

It's like partial, but the argument addition is commuted. The key is
that 'y' is going in the first slot, as 'self'. It's not clear that
argument addition commution applies (matches, is correct, generalizes
with right meaning): you just want the first parameter to come from a
separate call. Other uses cases of the generalizer might not align.
+ a fraction on 'x'.

Floats get harder, since you're looking at composition.

>>tuple( map( compose( float, x( float.__mul__, 1/ 2.5 ) ), y ) )

(120.0, 160.0) # (160, 120)

+1 compose.

def compose( *f ):
def post( *ar, **kw ):
if len( f )1:
return f[0]( compose( *f[1:] )( *ar, **kw ) )
return f[0]( *ar, **kw )
return post

[castironpi@gmail.com]

Re: dividing tuple elements with an int or float

On Mar 21, 8:14 am, castiro...@gmai l.com wrote:

On Mar 20, 9:28 am, "Jerry Hill" <malaclyp...@gm ail.comwrote:

On Thu, Mar 20, 2008 at 3:42 AM, Steven D'Aprano
<st...@remove-this-cybersource.com .auwrote:

On Wed, 19 Mar 2008 23:06:44 -0700, royG wrote:

>

 suppose
 origsz=(400,300 )
 i want to divide the origsize by 2.5 so i can resize to (160,120)

>

 scale=2.5
 how can i get the newsz?
 obviously origsz/2.5 won't work  ..

>

 newsz = (origsz[0]/scale, origsz[1]/scale)

>

That works fine for a 2-tuple, but might get unwieldy for larger
tuples, or if you don't know the length until runtime.  A more general
solution might use a generator expression, like this:

>

newsz = tuple(x/scale for x in origsz)

>
You want to perform a uniform call on the elements of a collection.
+1 compose.

By the way. -And sorry for interrupting the OP's time- I feel that
control flow objects could be a really powerful addition to a
language. Compose a for-loop.

[Gabriel Genellina]

Re: dividing tuple elements with an int or float

On 21 mar, 10:14, castiro...@gmai l.com wrote:

On Mar 20, 9:28 am, "Jerry Hill" <malaclyp...@gm ail.comwrote:

A more general
solution might use a generator expression, like this:
newsz = tuple(x/scale for x in origsz)

>
You want to perform a uniform call on the elements of a collection.
"Diagram A" appends 0 to every item of a list.
>

>y= [ [] for k in range( 10 ) ]
>def x( fun, *ar, **kw ):

>
...     def post( *xr, **xw ):
...             return fun( *(xr+ ar), **kw )
...     return post
...>>x( list.append, 0 )( y[0] )

Sane programmers replace that crazyness with this code:

for x in collection:
x.append(0)

>y= ( 300, 400 )
>tuple( map( x( int.__add__, 1, ), y ) )

>
(301, 401)

Sane programmers replace that crazyness with this code:

tuple(x+1 for x in y)

Floats get harder, since you're looking at composition.
>

>tuple( map( compose( float, x( float.__mul__, 1/ 2.5 ) ), y ) )

>
(120.0, 160.0) # (160, 120)

Sane programmers -like D'Aprano, Jerry Hill and me- replace that
crazyness with this code:

tuple(x/2.5 for x in y)

def compose( *f ):
    def post( *ar, **kw ):
            if len( f )1:
                    return f[0]( compose( *f[1:] )( *ar, **kw ) )
            return f[0]( *ar, **kw )
    return post

Sane programmers don't write such semi-functional things (unless it
helps expressing the problem in certain domains).
I now think that deprecating map, lambda & Co. was a good thing after
all.

--
Gabriel Genellina

[castironpi@gmail.com]

Re: dividing tuple elements with an int or float

On Mar 21, 1:08 pm, Gabriel Genellina <gagsl-...@yahoo.com.a rwrote:

Sane programmers replace that crazyness with this code:
>
tuple(x+1 for x in y)
>
Sane programmers -like D'Aprano, Jerry Hill and me- replace that
crazyness with this code:
>
tuple(x/2.5 for x in y)
>
Sane programmers don't write such semi-functional things (unless it
helps expressing the problem in certain domains).
I now think that deprecating map, lambda & Co. was a good thing after
all.

If you write it that way the first time, you need therapy. Actually,
at this point, I (for one, personally) want to investigate 'certain
domains'. Tell me it's really bad at everything or what it's good
at. What can I respect about it?

[Lie]

Re: dividing tuple elements with an int or float

On Mar 22, 2:23 pm, castiro...@gmai l.com wrote:

Sane programmers don't write such semi-functional things (unless it
helps expressing the problem in certain domains).
I now think that deprecating map, lambda & Co. was a good thing after
all.

>
If you write it that way the first time, you need therapy.  Actually,
at this point, I (for one, personally) want to investigate 'certain
domains'.  Tell me it's really bad at everything or what it's good
at.  What can I respect about it?

If you (castiro..) write it your way, you'll surely win the Obfuscated
Python Code Contest.

[castironpi@gmail.com]

Re: dividing tuple elements with an int or float

On Mar 23, 12:22 pm, Lie <Lie.1...@gmail .comwrote:

On Mar 22, 2:23 pm, castiro...@gmai l.com wrote:
>

Sane programmers don't write such semi-functional things (unless it
helps expressing the problem in certain domains).
I now think that deprecating map, lambda & Co. was a good thing after
all.

>

If you write it that way the first time, you need therapy.  Actually,
at this point, I (for one, personally) want to investigate 'certain
domains'.  Tell me it's really bad at everything or what it's good
at.  What can I respect about it?

>
If you (castiro..) write it your way, you'll surely win the Obfuscated
Python Code Contest.

Sometimes you're on a roll and you don't want to back up a step. What
are good "on a roll" habits? (Or, what roll?)