[Feature]: TermSetWrapper with DataIO

[mavaylon1]

What would you like to see added to HDMF?

The TermSetWrapper class is meant to be used with HDF5 datasets and attributes. These datasets can also be used with DataIO. If a user wants to use both, we need to provide support for that functionality. The workflow would look something like

-->H5DataIO(data=TermSetWrapper(item=data), compression=True)

This way the data is validated before going into DataIO. Once in DataIO, the wrapper would need (most likely) to be removed to not interfere with IO operations.

Is your feature request related to a problem?

No response

What solution would you like?

Refer Above.

Do you have any interest in helping implement the feature?

Yes.

Code of Conduct

• I agree to follow this project's Code of Conduct
• Have you checked the Contributing document?
• Have you ensured this change was not already requested?

[oruebel]

An alternative would be to swap the two, that is have TermSetWrapper wrap H5DataIO

TermSetWrapper(item=H5DataIO(data=data), compression=True))

In principal, I think either way is fine. With TermSetWrapper on the outside it may just work because DataIO exposes the data. With the option where TermSetWrapper is on the inside, we would probably need to unwrap the data for I/O. One option may be to update the DataIO.data property to do the unwrapping :

hdmf/src/hdmf/data_utils.py

Lines 995 to 998 in ddc842b

	@property
	def data(self):
	"""Get the wrapped data object"""
	return self.__data

[mavaylon1]

An alternative would be to swap the two, that is have TermSetWrapper wrap H5DataIO

TermSetWrapper(item=H5DataIO(data=data), compression=True))

In principal, I think either way is fine. With TermSetWrapper on the outside it may just work because DataIO exposes the data. With the option where TermSetWrapper is on the inside, we would probably need to unwrap the data for I/O. One option may be to update the DataIO.data property to do the unwrapping :

hdmf/src/hdmf/data_utils.py

Lines 995 to 998 in ddc842b

	@property
	def data(self):
	"""Get the wrapped data object"""
	return self.__data

I agree either way in practice is fine, but I think implicitly it makes sense that the TermSetWrapper comes first and wraps around the data to show its validating the data and not the IO Wrapper. Though that's just a mental preference I have.

[mavaylon1]

Similar to supporting iterative write with the wrapper, this is a special case that we will push back on till requested by the community.

[oruebel]

Did you test with

TermSetWrapper(item=H5DataIO(data=data), compression=True))

I think this should likely work as is and may just need a couple of unit tests.