Dict- Arranging a elements in a order

[psbasha]

Hi,

In case of the 'dict',the stored elements will not be in a order.It will be randomly arranged ( not as list,in list the elements will be arranged in order as we store it).
For Example :
---------------------
Input : Elements stored in order
>>> d = {1:[1,1],100:[1,2],2:[1]}

output ( Not as expected): Order of the dict Elements as we stored has been changed
>>> d
{1: [1, 1], 2: [1], 100: [1, 2]}
>>>

How to arrange the elements in a order inside a dict?

Thanks
PSB

[bartonc]

Hi,

In case of the 'dict',the stored elements will not be in a order.It will be randomly arranged ( not as list,in list the elements will be arranged in order as we store it).
For Example :
---------------------
Input : Elements stored in order
>>> d = {1:[1,1],100:[1,2],2:[1]}

output ( Not as expected): Order of the dict Elements as we stored has been changed
>>> d
{1: [1, 1], 2: [1], 100: [1, 2]}
>>>

How to arrange the elements in a order inside a dict?

Thanks
PSB

Use a tuple instead if a ditionary...

>>> t = ({1:(1,1)},{100:(1,2)},{2:(1,)})
>>> t
({1: (1, 1)}, {100: (1, 2)}, {2: (1,)})
>>> t[1]
{100: (1, 2)}
>>> 

I have used tuples as the items of the dictionaries as an efficiency boost. Lists take much longer to create in memory that do tuples.

[bartonc]

Use a tuple instead if a ditionary...

>>> t = ({1:(1,1)},{100:(1,2)},{2:(1,)})
>>> t
({1: (1, 1)}, {100: (1, 2)}, {2: (1,)})
>>> t[1]
{100: (1, 2)}
>>> 

I have used tuples as the items of the dictionaries as an efficiency boost. Lists take much longer to create in memory that do tuples.

For example, this implementation is almost four times faster than yours:

>>> import timeit
 
>>> # a million dictionaries of int:tuple pairs
>>> t = timeit.Timer("d = {1:(1,1),100:(1,2),2:(1,)}")
>>> t.timeit()
1.1584151058304997
 
>>> # a million tuples with 3 dictionaries of int:tuple pairs
>>> t = timeit.Timer("t = ({1:(1,1)},{100:(1,2)},{2:(1,)})")
>>> t.timeit()
1.942572589533027
 
>>> # a million dictionaries of int:list pairs
>>> t = timeit.Timer("d = {1:[1,1],100:[1,2],2:[1]}")
>>> t.timeit()
4.0557003753270351
>>> 

And creating three dictionaries with lists as the items took agonizingly long:

>>> # a million tuples with 3 dictionaries of int:list pairs
>>> t = timeit.Timer("t = ({1:[1,1]},{100:[1,2]},{2:[1]})")
>>> t.timeit()
4.8632546873974434
>>> 

[bvdet]

Hi,

In case of the 'dict',the stored elements will not be in a order.It will be randomly arranged ( not as list,in list the elements will be arranged in order as we store it).
For Example :
---------------------
Input : Elements stored in order
>>> d = {1:[1,1],100:[1,2],2:[1]}

output ( Not as expected): Order of the dict Elements as we stored has been changed
>>> d
{1: [1, 1], 2: [1], 100: [1, 2]}
>>>

How to arrange the elements in a order inside a dict?

Thanks
PSB

There are several implementations of ordered dictionaries <url remove> on activestate

[bartonc]

There are several implementations of ordered dictionaries <url remove> on activestate

Sorry BV. The powers that be would have my head if I let a link to a competitive site stay. Besides, I'm sure our friend, here, wants to maintain the order of creation; not the key value order (which of course can't be done in a dictionary - for good reason).

[bvdet]

Sorry BV. The powers that be would have my head if I let a link to a competitive site stay. Besides, I'm sure our friend, here, wants to maintain the order of creation; not the key value order (which of course can't be done in a dictionary - for good reason).

Oops! I did not consider that site as competition. A Google search on "ordered dictionary python" provides plenty of links. The implementations I looked at preserved the order of creation.

[bartonc]

Oops! I did not consider that site as competition. A Google search on "ordered dictionary python" provides plenty of links. The implementations I looked at preserved the order of creation.

I've held back linking there due to their forum structure.

Our friend here has complained that his python implementation is 4 to 5 times slower that another implementation. Simple is probably better than anything else (with an eye toward efficiency). <Just my impressions on the situation.>

[bartonc]

Oops! I did not consider that site as competition. A Google search on "ordered dictionary python" provides plenty of links. The implementations I looked at preserved the order of creation.

Here is an example of what is allowed from such sites:

extract = lambda keys, dict: reduce(lambda x, y: x.update({y[0]:y[1]}) or x,
                                    map(None, keys, map(dict.get, keys)), {})
 

That way they get the credit that is due, the member gets the info with a bonus, an we have not provided a link.