fix: check_invalid_binary_vector should not raise on non-sized values

[lokeshramchand-ctrl]

Problem

check_invalid_binary_vector raises an unhandled TypeError instead of returning False when a BINARY_VECTOR entity's values list contains an element with no __len__ (e.g. None, an int):

>>> check_invalid_binary_vector([{"type": DataType.BINARY_VECTOR, "values": [b"\x00\x01", None]}])
TypeError: object of type 'NoneType' has no len()

The per-row loop calls len(values) before checking isinstance(values, bytes):

for values in entity["values"]:
    if len(values) * 8 != dim:        # len() evaluated first
        return False
    if not isinstance(values, bytes):  # too late — never reached on non-sized input
        return False

Impact

This function's whole job is to return False so the caller can raise a clean error. grpc_handler.py:1094 (and :1228, plus the async handler) does:

if not check_invalid_binary_vector(entities):
    raise ParamError(message="Invalid binary vector data exists")

So instead of ParamError("Invalid binary vector data exists"), the user gets a raw TypeError out of the validation helper. Reachable from a malformed binary-vector insert, e.g. Collection.insert(data=[[b"\x00\x01", None]]).

This is a sibling gap to the one fixed in #3694 (and → or for the empty-list case), in the same function, that survived that fix. The existing test_non_bytes_values test only exercises nested lists as the invalid case, which still have __len__, so it never caught a genuinely non-sized element.

Filed as #3768.

Fix

Swap the two checks: verify isinstance(values, bytes) before calling len(values).

for values in entity["values"]:
    if not isinstance(values, bytes):
        return False
    if len(values) * 8 != dim:
        return False

Testing

Added test_non_sized_value to TestCheckInvalidBinaryVector in tests/unit/test_utils.py, alongside the existing test_inconsistent_dimensions / test_non_bytes_values / test_empty_values cases.

• uv run pytest tests/unit/test_utils.py -q → 71 passed
• uv run pytest tests/unit/ -q → 4719 passed, 3 skipped, 1 failed (tests/unit/test_check.py::TestGetCommit::test_get_commit, a pre-existing unrelated failure — same one called out in Reject an empty binary vector list instead of raising IndexError #3694's own PR description — verified to fail identically on master before this change)
• Manually verified the repro above raises TypeError before the fix and returns False after

---

This change was AI-assisted (Claude). I found the bug while doing a broader audit for unclaimed defects, reproduced the TypeError directly, traced the call sites in grpc_handler.py / async_grpc_handler.py to confirm the intended contract, and ran the full unit suite plus the baseline comparison above myself.

[sre-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: lokeshramchand-ctrl
To complete the pull request process, please assign xuanyang-cn after the PR has been reviewed.
You can assign the PR to them by writing /assign @xuanyang-cn in a comment when ready.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[sre-ci-robot]

Welcome @lokeshramchand-ctrl! It looks like this is your first PR to milvus-io/pymilvus 🎉

[lokeshramchand-ctrl]

/assign @XuanYang-cn -- Please look into this and let me know if any changes

[mergify]

Tick the box to add this pull request to the merge queue (same as @mergifyio queue).

• Queue this pull request