relabel_sequential() fails for every unsigned dtype: OverflowError: Python integer -1 out of bounds for uint16

[!NOTE]
This whole message is AI-generated. The issue was automatically discovered and reported by an AI agent (Claude) during an autonomous bug hunt on the spatialdata code base. It has not been verified or triaged by a human yet; the needs: triage label is set so that a maintainer can confirm it. The reproduction script below was executed by the agent in an isolated environment (see Environment) and its output is pasted verbatim.

Summary

With NumPy ≥ 2 the Python int -1 cannot be cast to an unsigned dtype (NEP 50), so relabel_sequential raises for uint8/16/32/64 inputs and works only for signed dtypes. Reproduced with numpy 2.4.4 and 2.5.3.

Severity (agent's assessment): high — segmentation masks are almost always unsigned, and map_raster(relabel=True) points users to this function

Where: src/spatialdata/_core/operations/map.py::relabel_sequential (da.full(max_label + 1, -1, dtype=arr.dtype))

Expected behaviour

Works for all integer dtypes.

Reproduction

Save as repro.py and run uv run repro.py (the PEP 723 header pins spatialdata to the commit the bug was found on; replace the URL fragment with @main to test the current main branch).

# /// script
# requires-python = ">=3.12"
# dependencies = [
#     "spatialdata @ git+https://github.com/scverse/spatialdata.git@ccf1ea048d054b6624214bf618008a9f9ae223e0",
# ]
# ///
"""relabel_sequential fails for every unsigned dtype (numpy 2: -1 cannot be cast to uintN)."""
import warnings
import numpy as np
import dask.array as da
from spatialdata import relabel_sequential

warnings.simplefilter("ignore")
print("numpy", np.__version__)
bug = False
for dt in [np.uint8, np.uint16, np.uint32, np.int32, np.int64]:
    arr = da.from_array(np.array([[0, 5, 9], [9, 5, 0]], dtype=dt), chunks=(2, 3))
    try:
        print(f"{dt.__name__:7s}: {relabel_sequential(arr).compute().tolist()}")
    except Exception as e:  # noqa: BLE001
        print(f"{dt.__name__:7s}: {type(e).__name__}: {e}")
        bug = True
print("VERDICT:", "BUG REPRODUCED" if bug else "NOT REPRODUCED")

Observed output

numpy 2.5.3
uint8  : OverflowError: Python integer -1 out of bounds for uint8
uint16 : OverflowError: Python integer -1 out of bounds for uint16
uint32 : OverflowError: Python integer -1 out of bounds for uint32
int32  : [[0, 1, 2], [2, 1, 0]]
int64  : [[0, 1, 2], [2, 1, 0]]
VERDICT: BUG REPRODUCED

Possible fix direction (unverified)

Use a fill value valid for the dtype (0, since ids not in unique_labels are background anyway, or np.iinfo(dtype).max) and add unsigned-dtype tests. Note also that the lookup array must be a single dask chunk (rechunk(-1)) for the following map_blocks(operator.getitem, ...) to be correct when max_label is very large.

Environment

uv run repro.py with the PEP 723 metadata in the script (fresh, isolated environment; spatialdata built from main @ ccf1ea0 (2026-08-28); Python 3.13, latest releases of the dependencies at run time: pandas 3.0, anndata 0.13, zarr 3.3, dask 2026.8, numpy 2.5, geopandas 1.1, shapely 2.1). macOS (arm64). Also reproduced in a second environment with pandas 2.3.3 / anndata 0.12.11 / numpy 2.4.4 / zarr 3.2.1.

Automatically generated; discovered by an AI agent (Claude) and not yet reviewed by a human.

Start in src/spatialdata/_core/operations/map.py at relabel_sequential and run the provided repro.py script with the listed unsigned dtypes. Add regression coverage for uint8, uint16, uint32, and uint64 inputs, then verify that relabel_sequential produces the expected labels for unsigned and signed integer arrays.