tm1py-samples: dim_create.py with attribute values

[ubihw]

Hello,

https://github.com/cubewise-code/tm1py-samples/blob/master/Metadata/dim_create.py shows how to create a dimension, including attributes. What is missing, is, how the attribute values are written. I noticed, that the Element Object accepts an attributes list as additional parameter, and extended the example like follows:

from TM1py.Objects import Dimension, Element, ElementAttribute, Hierarchy
from TM1py.Services import TM1Service

from config import tm1params

name = 'TM1py Region'

# Connection to TM1. Needs IP, Port, Credentials, and SSL
with TM1Service(**tm1params) as tm1:
    # create elements objects
    # the attribute values must be in the same order as the attributes in element_attributes!
    elements = [Element(name='Europe', element_type='Consolidated', attributes=['Europe','EUR','mixed']),
                Element(name='CH', element_type='Numeric', attributes=['Switzerland','SWI','CHF']),
                Element(name='UK', element_type='Numeric', attributes=['United Kingdom','UnK','BP']),
                Element(name='BE', element_type='Numeric', attributes=['Belgium','BEL','EUR'])]

    # create edge object
    edges = {('Europe', 'CH'): 1,
             ('Europe', 'UK'): 1,
             ('Europe', 'BE'): 1}

    # create the element_attributes
    element_attributes = [ElementAttribute(name='Name Long', attribute_type='Alias'),
                          ElementAttribute(name='Name Short', attribute_type='Alias'),
                          ElementAttribute(name='Currency', attribute_type='String')]

    # create hierarchy object
    hierarchy = Hierarchy(name=name, dimension_name=name, elements=elements, element_attributes=element_attributes,
                          edges=edges)

    # create dimension object
    d = Dimension(name=name, hierarchies=[hierarchy])

    # create dimension in TM1 !
    tm1.dimensions.create(d)

    # populate attribute cube
    attrcube = '}ElementAttributes_' + name
    attrvals = dict()
    for i in range(len(element_attributes)):
        attrvals.update({(e.name,element_attributes[i].name) : e.element_attributes[i] for e in elements})
    tm1.cubes.cells.write_values(attrcube,attrvals)

It worked for me, but I dont know, if I used the attributes parameter in the way it was intended for. If there is a more TM1py-ish way to populate the attributes, please let me know. And, maybe dim_create.py could be extended with the right way to populate attributes?

[MariusWirtz]

Hi @ubihw,

TM1py treats attribute values as cell values. So to write attribute values you can write directly to the }ElementAttributes cube, just like you did.
In many ways, the write_values function is inferior to the standard write function. You can read about the different write functions here:
https://cubewise-code.github.io/tm1py-tales/2021/write-with-tm1py.html
I think the only real use case of the write_values function is to use data spread shortcuts on cells.

Regarding the attributes parameter in the Element constructor.
One could assume that the attribute values are updated in TM1 after this line is executed.

tm1.dimensions.create(d)

Unfortunately, TM1py can't send the attribute values in one operation together with the element updates to TM1.
So the attributes parameter in the Element constructor is only used when TM1py retrieves data from TM1 hierarchies. For write-back you need to write straight to the attributes cube, just like you did.

Yes. I think the way you use the attributes property to store the attribute values in the elements makes sense.

You can read more about dimension updates here:
https://cubewise-code.github.io/tm1py-tales/2021/sync-dimensions.html