Add support for Vensim TABBED ARRAYS

[rogersamso]

Description

Add support for TABBED ARRAYS

Type of change

• New feature (non-breaking change which adds functionality)

PR verification (to be filled by reviewers)

• [ ] The code follows the PEP 8 style
• [ ] The new code has been tested properly for all the possible cases
• [ ] The overall coverage has not dropped and other features have not been broken.
• [ ] If new features have been added, they have been properly documented
• [ ] docs/whats_new.rst has been updated
• [ ] PySD version has been updated

[rogersamso]

@enekomartinmartinez, please let me know if you are fine with this, and I will update the what's new and others.

[enekomartinmartinez]

@enekomartinmartinez, please let me know if you are fine with this, and I will update the what's new and others.

Hi @rogersamso, thank you for the changes. It is a good and clean implementation!

I just have some suggestions for the text model at https://github.com/SDXorg/test-models/pull/86

Could you please add TABBED ARRAYS of 1D and 2D there, and if possible a 3D var defined with several tabbed arrays. Also include some regular floating point numbers and with e (e.g. 0.075, 3e-05), they should work properly as the raw_number parsing works for other tests, but just in case is better to have a more general example test.

[enekomartinmartinez]

Hi @rogersamso

I noticed that there is a bug to be corrected. The current implementation will not accept tabbed arrays defined as a n unchangeable constants (==).

You should remove all the parsing done in the element_object and parse the tabbed_array() call at the component level (which will no be need to be visited). Please, let me know if I explained properly.

Also, please add an unchangeable constant defined tabbed array ( with two =) in the test model.

[rogersamso]

@enekomartinmartinez

Please check if this is better now.

However, there's still a fix to implement, as Unchangeable constants in PySD only work for scalars.

This code:

dim1:
    a, b, c
    ~   
    ~       |

unchangeable_vector[dim1]==
    1, 2, 3;
    ~   
    ~       |

translates to this:

@component.add(
    name="unchangeable_vector",
    subscripts=["dim1"],
    comp_type="Constant",
    comp_subtype="Normal",
)
def unchangeable_vector():
    return xr.DataArray([1.0, 2.0, 3.0], {"dim1": _subscript_dict["dim1"]}, ["dim1"])

Note that comp_subtype in the translated code is not Unchangeable.

This test should be added in tests/test-models/tests/unchangeable_constant/test_unchangeable_constant.mdl. If you want I can push it to the test repos too, while we prepare the fix in pysd.