feat(waterdata.xarray): always drop hash IDs; ignore include_hash

The xarray path never surfaces hash columns (neither the per-record OGC
feature id nor the per-series join key), so route every wrapper through a
_fetch helper that pops include_hash before calling the underlying getter.
This makes the flag inert in the xarray path -- passing include_hash=True no
longer triggers a fetch of a column we'd only discard -- and documents that
hash IDs are always omitted here. The pandas getters keep include_hash as
the opt-in flag; only the xarray layer drops them unconditionally.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: thodson-usgs

dataretrieval/waterdata/xarray.py

@@ -329,22 +329,33 @@ def _sites(df):
     return []
 
 
+def _fetch(func, args, kwargs):
+    """Call the underlying getter with hash IDs forced off.
+
+    The xarray path never surfaces hash columns (neither the per-record
+    UUID nor the per-series join key), so ``include_hash`` is dropped here:
+    passing it has no effect and we avoid fetching a column we'd discard.
+    """
+    kwargs.pop("include_hash", None)
+    return func(*args, **kwargs)
+
+
 def _xr_doc(func):
     """Prepend an xarray note to the wrapped getter's docstring."""
     note = (
         "    xarray wrapper: same arguments as "
         f"``dataretrieval.waterdata.{func.__name__}``, but returns a\n"
-        "    CF-conventions ``xarray.Dataset`` with series metadata populated.\n\n"
+        "    CF-conventions ``xarray.Dataset`` with series metadata populated.\n"
+        "    Hash-valued ID columns are always omitted here; the\n"
+        "    ``include_hash`` flag does not apply.\n\n"
     )
     return note + (func.__doc__ or "")
 
 
 def _timeseries_wrapper(func, *, service, default_cell_method=None):
     @_wraps(func)
     def wrapper(*args, **kwargs):
-        # Default (hash-free) fetch: the per-record UUID is never requested or
-        # materialized; CF attributes come from the surviving columns.
-        df, base_meta = func(*args, **kwargs)
+        df, base_meta = _fetch(func, args, kwargs)
         return _build_timeseries(
             df,
             base_meta,
@@ -360,7 +371,7 @@ def wrapper(*args, **kwargs):
 def _field_wrapper(func, *, service):
     @_wraps(func)
     def wrapper(*args, **kwargs):
-        df, base_meta = func(*args, **kwargs)
+        df, base_meta = _fetch(func, args, kwargs)
         return _build_timeseries(
             df,
             base_meta,
@@ -377,7 +388,7 @@ def wrapper(*args, **kwargs):
 def _stats_wrapper(func, *, service):
     @_wraps(func)
     def wrapper(*args, **kwargs):
-        df, base_meta = func(*args, **kwargs)
+        df, base_meta = _fetch(func, args, kwargs)
         return _build_stats(df, base_meta, service)
 
     wrapper.__doc__ = _xr_doc(func)

tests/waterdata_xarray_test.py

@@ -178,3 +178,20 @@ def test_public_wrappers_exist_and_are_documented():
         fn = getattr(wdx, name)
         assert callable(fn)
         assert "xarray wrapper" in (fn.__doc__ or "")
+
+
+def test_fetch_strips_include_hash():
+    # The xarray path never surfaces hash columns, so include_hash must be
+    # dropped before the underlying getter is called (no wasted fetch).
+    captured = {}
+
+    def fake_getter(*args, **kwargs):
+        captured.update(kwargs)
+        return "df", "meta"
+
+    df, meta = wdx._fetch(
+        fake_getter, (), {"include_hash": True, "parameter_code": "00060"}
+    )
+    assert (df, meta) == ("df", "meta")
+    assert "include_hash" not in captured
+    assert captured == {"parameter_code": "00060"}