Annotate window / path / on_gpu_failure on public geotiff API (#1654) (#1658)

* Annotate window/path/on_gpu_failure on public geotiff API (#1654)

The api-consistency sweep on 2026-05-12 found the same parameter is
annotated on some siblings but not on others across the public
xrspatial.geotiff surface. Pin each annotation so type-checkers and
IDEs validate user code consistently.

- window: tuple | None on open_geotiff and read_vrt (read_geotiff_dask
  and read_geotiff_gpu already had it).
- path: str | BinaryIO on to_geotiff and write_geotiff_gpu. write_vrt
  stays str-only because VRT writes are path-only by design.
- on_gpu_failure: str on open_geotiff and read_geotiff_gpu. The
  deprecated gpu alias on read_geotiff_gpu carries the same str hint.

Annotation-only change; no runtime behaviour, defaults, or kwarg
renames. BinaryIO is imported under TYPE_CHECKING so the runtime
import cost stays at zero with from __future__ import annotations.

test_signature_annotations_1654.py pins each annotation to guard
against future signature drift. Also updates the api-consistency
sweep state CSV.

* Use tmp_path fixture for Windows-compatible cleanup (#1658)

Author: brendancol

.claude/sweep-api-consistency-state.csv

@@ -1,4 +1,3 @@
 module,last_inspected,issue,severity_max,categories_found,notes
-geotiff,2026-05-11,1644,MEDIUM,3,"Filed write_geotiff_gpu compression docstring drift vs to_geotiff (MEDIUM Cat 3, #1644). Fix on deep-sweep-api-consistency-geotiff-2026-05-11-1778545740: sync the full 9-codec list into the docstring and note GPU vs CPU encode paths; regression test test_compression_docstring_1644.py pins the codec list and exercises each CPU-fallback codec end-to-end. Other potential drifts surveyed: write_vrt returns str while to_geotiff/write_geotiff_gpu return None (LOW, intentional backward-compat); write_vrt nodata typed float|None vs int-accepting siblings (LOW, PEP 484 int->float compat); kwarg-only ordering drift across read functions (LOW, no user impact). Prior issues 1631/1637/1615/1560/1541/1562 all CLOSED."
-geotiff,2026-05-11,1652,MEDIUM,5,"Filed MEDIUM file-like cog=True drift #1652 (write_geotiff_gpu accepted BytesIO+cog=True; to_geotiff blocked it). Fixed in PR (TBD): mirror to_geotiff's gate on the explicit GPU writer; add regression tests in test_bytesio_source.py. Also filed #1651 (JPEG acceptance drift) but downgraded to LOW after #1647 confirmed write_geotiff_gpu(jpeg) is deliberate advanced-API; PR (TBD) carries the docstring clarification. Prior 1631/1644 noted in earlier rows (1644 open, fix in PR #1649). LOW: streaming_buffer_bytes default drift to_geotiff=256MB vs write_geotiff_gpu=None (no functional impact, explicit forwarding); to_geotiff data: annotation misses cupy.ndarray (accepted via auto-dispatch). cuda-validated."
+geotiff,2026-05-12,1654,MEDIUM,3,"Filed type-annotation drift #1654 (MEDIUM Cat 3): public window/path/on_gpu_failure annotations missing on some siblings while others carry them. Fixed on deep-sweep-api-consistency-geotiff-2026-05-12 with PR (TBD): added tuple|None on open_geotiff/read_vrt window, str|BinaryIO on to_geotiff/write_geotiff_gpu path, str on open_geotiff/read_geotiff_gpu on_gpu_failure plus the deprecated gpu alias. Regression test test_signature_annotations_1654.py pins each annotation. Prior drifts surveyed: chunks default 512 vs None (intentional, read_geotiff_dask is dask-only); crs vs crs_wkt on write_vrt (VRT is WKT-only); write_vrt returns str (intentional backward-compat); nodata float|None vs unannotated (LOW, prior sweep). cuda-validated."
 reproject,2026-05-10,1570,HIGH,2;5,"Filed cross-module attrs['vertical_crs'] type collision (string vs EPSG int) vs xrspatial.geotiff. Fixed in PR (TBD): reproject now writes EPSG int and preserves friendly token under vertical_datum. MEDIUM kwarg-order drift (transform_precision vs chunk_size) and missing type hints vs geotiff documented but not fixed (cosmetic, kwarg-only)."

xrspatial/geotiff/__init__.py

@@ -33,10 +33,14 @@
 
 import math
 import warnings
+from typing import TYPE_CHECKING
 
 import numpy as np
 import xarray as xr
 
+if TYPE_CHECKING:
+    from typing import BinaryIO
+
 from ._geotags import GeoTransform, RASTER_PIXEL_IS_AREA, RASTER_PIXEL_IS_POINT
 from ._reader import read_to_array
 from ._writer import write
@@ -446,14 +450,16 @@ def _populate_attrs_from_geo_info(attrs: dict, geo_info, *, window=None) -> None
                 break
 
 
-def open_geotiff(source, *, dtype=None, window=None,
+def open_geotiff(source, *, dtype=None,
+                 window: tuple | None = None,
                  overview_level: int | None = None,
                  band: int | None = None,
                  name: str | None = None,
                  chunks: int | tuple | None = None,
                  gpu: bool = False,
                  max_pixels: int | None = None,
-                 on_gpu_failure=_ON_GPU_FAILURE_SENTINEL) -> xr.DataArray:
+                 on_gpu_failure: str = _ON_GPU_FAILURE_SENTINEL,
+                 ) -> xr.DataArray:
     """Read a GeoTIFF, COG, or VRT file into an xarray.DataArray.
 
     Automatically dispatches to the best backend:
@@ -905,7 +911,8 @@ def _extract_rich_tags(attrs: dict) -> dict:
     }
 
 
-def to_geotiff(data: xr.DataArray | np.ndarray, path, *,
+def to_geotiff(data: xr.DataArray | np.ndarray,
+               path: str | BinaryIO, *,
                crs: int | str | None = None,
                nodata=None,
                compression: str = 'zstd',
@@ -2178,8 +2185,9 @@ def read_geotiff_gpu(source: str, *,
                      name: str | None = None,
                      chunks: int | tuple | None = None,
                      max_pixels: int | None = None,
-                     on_gpu_failure=_ON_GPU_FAILURE_SENTINEL,
-                     gpu=_GPU_DEPRECATED_SENTINEL) -> xr.DataArray:
+                     on_gpu_failure: str = _ON_GPU_FAILURE_SENTINEL,
+                     gpu: str = _GPU_DEPRECATED_SENTINEL,
+                     ) -> xr.DataArray:
     """Read a GeoTIFF with GPU-accelerated decompression via Numba CUDA.
 
     Decompresses all tiles in parallel on the GPU and returns a
@@ -2714,7 +2722,7 @@ def _read_once():
 
 
 def write_geotiff_gpu(data: xr.DataArray | cupy.ndarray | np.ndarray,
-                      path, *,
+                      path: str | BinaryIO, *,
                       crs: int | str | None = None,
                       nodata=None,
                       compression: str = 'zstd',
@@ -3069,7 +3077,8 @@ def _gpu_compress_to_part(gpu_arr, w, h, spp):
     _write_bytes(file_bytes, path)
 
 
-def read_vrt(source: str, *, dtype=None, window=None,
+def read_vrt(source: str, *, dtype=None,
+             window: tuple | None = None,
              band: int | None = None,
              name: str | None = None,
              chunks: int | tuple | None = None,

xrspatial/geotiff/tests/test_signature_annotations_1654.py

@@ -0,0 +1,131 @@
+"""Regression test for #1654: public geotiff API parameter annotations.
+
+The api-consistency sweep on 2026-05-12 flagged a MEDIUM type-annotation
+drift across the public ``xrspatial.geotiff`` surface. The same parameter
+was annotated on some sibling functions but missing on others:
+
+* ``window``: annotated on ``read_geotiff_dask`` and ``read_geotiff_gpu``
+  but missing on ``open_geotiff`` and ``read_vrt``.
+* ``path``: annotated on ``write_vrt.vrt_path`` (str-only) but missing
+  on ``to_geotiff`` and ``write_geotiff_gpu`` (str or binary file-like).
+* ``on_gpu_failure`` (and the deprecated ``gpu`` alias on
+  ``read_geotiff_gpu``): documented as ``{'auto', 'strict'}`` strings
+  but no annotation. The sentinel default did not preclude annotating
+  the user-visible value type.
+
+This module pins each annotation so future signature changes do not
+silently drop them.
+"""
+from __future__ import annotations
+
+import inspect
+
+from xrspatial.geotiff import (
+    open_geotiff,
+    read_geotiff_dask,
+    read_geotiff_gpu,
+    read_vrt,
+    to_geotiff,
+    write_geotiff_gpu,
+    write_vrt,
+)
+
+
+def _annotation(fn, param_name):
+    """Return the stringified annotation for ``fn``'s ``param_name``.
+
+    ``from __future__ import annotations`` keeps annotations as strings
+    at runtime, so the comparison works against the source literal.
+    """
+    sig = inspect.signature(fn)
+    p = sig.parameters[param_name]
+    assert p.annotation is not inspect.Parameter.empty, (
+        f"{fn.__name__}({param_name}=...) is missing a type annotation"
+    )
+    return str(p.annotation)
+
+
+# --- window: 4-tuple (r0, c0, r1, c1) or None ---
+
+
+def test_open_geotiff_window_annotated():
+    assert _annotation(open_geotiff, 'window') == 'tuple | None'
+
+
+def test_read_vrt_window_annotated():
+    assert _annotation(read_vrt, 'window') == 'tuple | None'
+
+
+def test_read_geotiff_dask_window_annotated():
+    """Pre-existing annotation -- keep it pinned so it does not regress."""
+    assert _annotation(read_geotiff_dask, 'window') == 'tuple | None'
+
+
+def test_read_geotiff_gpu_window_annotated():
+    """Pre-existing annotation -- keep it pinned so it does not regress."""
+    assert _annotation(read_geotiff_gpu, 'window') == 'tuple | None'
+
+
+# --- path: str or binary file-like (writer entry points) ---
+
+
+def test_to_geotiff_path_annotated():
+    """``to_geotiff(data, path, ...)`` ``path`` accepts str or BinaryIO."""
+    ann = _annotation(to_geotiff, 'path')
+    assert 'str' in ann
+    assert 'BinaryIO' in ann
+
+
+def test_write_geotiff_gpu_path_annotated():
+    """``write_geotiff_gpu(data, path, ...)`` ``path`` mirrors ``to_geotiff``."""
+    ann = _annotation(write_geotiff_gpu, 'path')
+    assert 'str' in ann
+    assert 'BinaryIO' in ann
+
+
+def test_write_vrt_vrt_path_annotated():
+    """``write_vrt(vrt_path, ...)`` stays str-only (VRT writes are
+    path-only by design; no file-like buffer support)."""
+    assert _annotation(write_vrt, 'vrt_path') == 'str'
+
+
+# --- on_gpu_failure: 'auto' | 'strict' (GPU failure policy) ---
+
+
+def test_open_geotiff_on_gpu_failure_annotated():
+    assert _annotation(open_geotiff, 'on_gpu_failure') == 'str'
+
+
+def test_read_geotiff_gpu_on_gpu_failure_annotated():
+    assert _annotation(read_geotiff_gpu, 'on_gpu_failure') == 'str'
+
+
+def test_read_geotiff_gpu_deprecated_gpu_alias_annotated():
+    """The deprecated ``gpu=`` alias on ``read_geotiff_gpu`` carries the
+    same ``str`` annotation as the new ``on_gpu_failure`` kwarg."""
+    assert _annotation(read_geotiff_gpu, 'gpu') == 'str'
+
+
+# --- Smoke: the new annotations do not break runtime call semantics ---
+
+
+def test_open_geotiff_window_kwarg_runtime(tmp_path):
+    """The annotated ``window`` kwarg still accepts a 4-tuple and returns
+    the requested sub-window. The test does not exercise ``on_gpu_failure``
+    because the runtime semantics are GPU-only; the annotation itself is
+    pinned by ``test_open_geotiff_on_gpu_failure_annotated``.
+    """
+    import numpy as np
+    import xarray as xr
+
+    arr = np.arange(64, dtype=np.float32).reshape(8, 8)
+    da = xr.DataArray(
+        arr, dims=['y', 'x'],
+        coords={'y': np.arange(8.0, 0, -1), 'x': np.arange(8.0)},
+        attrs={'crs': 4326, 'transform': (1.0, 0, 0.0, 0, -1.0, 8.0)},
+    )
+
+    path = str(tmp_path / 'window_kwarg.tif')
+    to_geotiff(da, path)
+    r = open_geotiff(path, window=(0, 0, 4, 4))
+    assert r.shape == (4, 4)