🐛 Bug Report: `TypeError` in `create_rdf_store()` when given `None` as input

[cedricdcc]

When calling create_rdf_store() with None as one of the inputs, a TypeError is raised. This happens because the function does not properly handle the case where None is passed in combination with other arguments, resulting in a NoneType being unpacked.

Stack Trace

TypeError: sema.commons.store.build.create_rdf_store() argument after * must be an iterable, not NoneType

Code Example

create_rdf_store(None)

Expected Behavior:
None should be safely ignored, and the function should return a MemoryRDFStore in this case.

Actual Behavior:
A TypeError is raised due to unpacking None.

---

🛠️ Proposed Solution

Modify the code to ensure that None values are properly filtered and handled before unpacking store_info. Specifically, ensure that store_info is always a valid iterable.

Current Code:

def create_rdf_store(*store_info) -> RDFStore:
    store_info = [el for el in store_info if el is not None]
    assert len(store_info) <= 2, "Too many arguments to create store {store_info=}"
    
    if len(store_info) == 0:
        return MemoryRDFStore()
    return URIRDFStore(*store_info)

---

✅ Proposed Fix:

Update the function to avoid unpacking store_info when it is empty or contains only None. Use a fallback to ensure a valid iterable is passed to URIRDFStore.

Suggested Code Update:

def create_rdf_store(*store_info) -> RDFStore:
    """Creates an rdf_store based on the passed non-None arguments.
    0 of those arguments yields a MemoryRDFStore,
    1-2 will be passed as read_uri and write_uri to URIRDFStore.
    Anything beyond is unacceptable.
    """
    store_info = tuple(el for el in store_info if el is not None)  # Ensure a valid tuple

    assert len(store_info) <= 2, f"Too many arguments to create store {store_info=}"

    if len(store_info) == 0:
        return MemoryRDFStore()

    return URIRDFStore(*store_info)  # Properly unpack only valid elements

---

🔬 Test Code

Here’s a test to confirm the bug and validate the fix:

def test_create_rdf_store():
    assert isinstance(create_rdf_store(), MemoryRDFStore)  # No input → MemoryRDFStore
    assert isinstance(create_rdf_store(None), MemoryRDFStore)  # None input → MemoryRDFStore
    assert isinstance(create_rdf_store("read_uri"), URIRDFStore)  # Single valid input → URIRDFStore
    assert isinstance(create_rdf_store("read_uri", "write_uri"), URIRDFStore)  # Two valid inputs → URIRDFStore

    # Test for failure if more than two arguments are passed
    try:
        create_rdf_store("read_uri", "write_uri", "extra_uri")
    except AssertionError as e:
        assert "Too many arguments" in str(e)

[cedricdcc]

The comment above is wrong.
The problems is not in the "None" itself but in the unpacking of the *None.

The function itself must not be called with create_rdf_store(*target_store_info) if in the class above that calls the function allows for a variable List[str] with default None. In Harvest the last argument called target_info_store should be changed to *target_store_info: List[str]

This would allow for unpacking of the files.
Check this for all other module sin sema that use target_store_info