Add a mapping function in transforms.io

[staydelight]

Add a function to create a JSON file that maps input and output paths.

Fixes #7557 .

Description

A few sentences describing the changes proposed in this pull request.

Types of changes

• Non-breaking change (fix or new feature that would not break existing functionality).
• Breaking change (fix or new feature that would cause existing functionality to change).
• New tests added to cover the changes.
• Integration tests passed locally by running ./runtests.sh -f -u --net --coverage.
• Quick tests passed locally by running ./runtests.sh --quick --unittests --disttests.
• In-line docstrings updated.
• Documentation updated, tested make html command in the docs/ folder.

[staydelight]

I had a few minor added comments but otherwise it looks good. If you could please address what I and @KumoLiu have mentioned we can get this through.

@ericspod @KumoLiu All comments have been addressed.If there are any additioanal changes,please let me know.

[ericspod]

@ericspod @KumoLiu All comments have been addressed.If there are any additioanal changes,please let me know.

Do we not also need a dictionary version of this transform? If not we can approve this as it is now.

[KumoLiu]

Do we not also need a dictionary version of this transform? If not we can approve this as it is now.

Adding a dictionary version would be better since dictionary version pipeline may be more commonly used.

[ericspod]

Do we not also need a dictionary version of this transform? If not we can approve this as it is now.

Adding a dictionary version would be better since dictionary version pipeline may be more commonly used.

The dictionary version can be something very simple I think, like this:

class WriteFileMappingd(MapTransform):
    def __init__(self, keys: KeysCollection, mapping_file_path: Path | str = "mapping.json", allow_missing_keys: bool = False) -> None:
        super().__init__(keys, allow_missing_keys)
        self.mapping = WriteFileMapping(mapping_file_path)

    def __call__(self, data: Mapping[Hashable, NdarrayOrTensor]) -> dict[Hashable, NdarrayOrTensor]:
        d = dict(data)
        for key in self.key_iterator(d):
            d[key] = self.mapping(d[key])
        return d

This would needed added tests as well.

[KumoLiu]

Hi @staydelight, just a quick note—if you want this new feature included in version 1.4, we only have one or two weeks left to add new features.

[staydelight]

Do we not also need a dictionary version of this transform? If not we can approve this as it is now.

Adding a dictionary version would be better since dictionary version pipeline may be more commonly used.

The dictionary version can be something very simple I think, like this:

class WriteFileMappingd(MapTransform):
    def __init__(self, keys: KeysCollection, mapping_file_path: Path | str = "mapping.json", allow_missing_keys: bool = False) -> None:
        super().__init__(keys, allow_missing_keys)
        self.mapping = WriteFileMapping(mapping_file_path)

    def __call__(self, data: Mapping[Hashable, NdarrayOrTensor]) -> dict[Hashable, NdarrayOrTensor]:
        d = dict(data)
        for key in self.key_iterator(d):
            d[key] = self.mapping(d[key])
        return d

This would needed added tests as well.

I would like to ask if data: Mapping[Hashable, NdarrayOrTensor]) -> dict[Hashable, NdarrayOrTensor] must include Ndarray. Since both the input and output paths require metadata, the input parameter for WriteFileMapping is MetaTensor. In this case, is it possible to use data: Mapping[Hashable, torch.Tensor]) -> dict[Hashable, torch.Tensor] or data: Mapping[Hashable, MetaTensor]) -> dict[Hashable, MetaTensor]?

Also, I've added the test for the dictionary. Please let me know if it's okay. Thank you!

[KumoLiu]

Thanks for the quick update, overall looks good to me.
@ericspod @binliunls, do you have any further comments on this one?

[KumoLiu]

/build