[DiadkoShmek]

Summary

Keep a single-component embedding as a list when get_vector() squeezes a supported array or tensor. np.array([0.5]) and pd.Series([0.5]) currently become the scalar 0.5, despite the helper's list-return contract. _Pack.single() then raises TypeError: object of type 'float' has no len(); the equivalent Python list works.

Wrap scalar conversion results in a list in both the direct squeeze().tolist() path and the numpy() fallback. Existing list results are retained, without flattening multi-vectors or changing request shape classification.

Validation

• Added regression coverage with actual NumPy float32/float64, pandas, polars and list inputs, including byte-packing parity. A mock tensor exercises the NumPy fallback with one, multiple and zero elements.
• On unchanged main (d1e185eaa5425cc0884c28689bbaa11d5eec7898), the vector test file reports 4 failed, 9 passed. With this change, vector, utility and byte-operations tests report 57 passed.
• A separate offline replay constructs near-vector protobuf requests with the real client modules for server version 1.29.0. NumPy/pandas single-component inputs fail before and succeed after; list, two-component and multi-vector controls retain their bytes. No server or RPC was used.
• Repository-pinned Ruff 0.14.7 checks/formatting, flake8 with development plugins, and git diff --check: passed on both changed files.

Real TensorFlow/PyTorch runtimes and server integration were not tested. The helper also normalizes a zero-dimensional array conversion to a list; zero-dimensional query input remains rejected by the existing shape classifier.

AI disclosure: Codex prepared and tested this change, with a separate AI review of the complete diff and focused cases. No personal human code review is claimed.

[orca-security-eu]

Orca Security Scan Summary

Status	Check	Issues by priority
Passed	Infrastructure as Code	0   0   0   0	View in Orca
Passed	SAST	0   0   0   0	View in Orca
Passed	Secrets	0   0   0   0	View in Orca
Passed	Vulnerabilities	0   0   0   0	View in Orca

[weaviate-git-bot]

To avoid any confusion in the future about your contribution to Weaviate, we work with a Contributor License Agreement. If you agree, you can simply add a comment to this PR that you agree with the CLA so that we can merge.

beep boop - the Weaviate bot 👋🤖

PS:
Are you already a member of the Weaviate Forum?

[DiadkoShmek]

I agree to the Weaviate Contributor License Agreement: https://weaviate.io/service/contributor-license-agreement