define_generator.py updates needed

[dostiep]

@swhume I ran define_generator.py on the latest define.json I created and had some issues listed below:

1. def:ValueListRef are not created under ItemDef elements. This is most likely due to a recent update on define.json model where VLMs are now represented in "slices".

2. isReferenceData in define.json is a boolean value and needs to be converted to "Yes" or "No" in define.xml.

3. Class and SubClass elements in define.xml have no corresponding elements in define.json model. Jeremy is not in favor of adding those either. Those will need to be either extracted from CDISC Library for existing domains and a new method will likely be needed for sponsor defined domains.

I started looking at gaps and what's missing and realized it is probably best to add "placeholders" directly in the json file being generated rather than creating a new file to address the gaps with OIDs to merge back the updated information. I did that for 3 items: KeySequence at ItemGroup level where I added the attributes with a list value that contains "<>". I did the same for length and origin for variables where this information is missing. Length is set to null and the the origin type and source are set to "<>". That being said, having type with a value "<>" triggered an error as it is not a valid value (item.Origin.append(DEFINE.Origin(**attr))). I simply commented out this line of code in my branch for now.

Should I try on my end to default some Length value from variable's name? For example, --FL variables always 1, --DTC 19, etc?

[swhume]

1. I created an issue for this one that would be good to resolve. Maybe we can discuss this during an upcoming call DDE Issue #4

2. isReferenceData in define.json is a boolean value, and the define_generator converts that to Yes or No. Let me know if there's a bug to fix or a change that's needed.

3. There is a limited Class implementation in define_generator (assumes a single class if any). We'll need to tackle this during ADaM define.xml implementation.

looking at gaps and what's missing, and realized it is probably best to add "placeholders" directly in the json file

That seems sensible to me.

rather than creating a new file to address the gaps with OIDs to merge back the updated information.

I assume that merging in missing metadata from alternative sources might be a requirement. Let's chat about that in our next meeting, as this is a big design decision. I think this discussion can also cover whether or not it makes sense to create some default lengths.