Fix _arrange_cols: stop mutating the caller's properties list

`_arrange_cols` did `properties.append("geometry")` and
`properties[properties.index("id")] = output_id` in place. Calling any
getter twice with the same `properties=[...]` therefore caused that
list to grow unboundedly and have `id` rewritten across calls — a
real action-at-a-distance defect for any code that re-uses a
`properties` list across requests.

Take a local copy of the list and mutate that. Caller's list is
untouched.

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

Author: thodson-usgs

dataretrieval/waterdata/utils.py

@@ -687,15 +687,17 @@ def _arrange_cols(
     # If properties are provided, filter to only those columns
     # plus geometry if skip_geometry is False
     if properties and not all(pd.isna(properties)):
+        # Work on a local copy so we don't mutate the caller's list across calls.
+        local_properties = list(properties)
         # Make sure geometry stays in the dataframe if skip_geometry is False
-        if "geometry" in df.columns and "geometry" not in properties:
-            properties.append("geometry")
+        if "geometry" in df.columns and "geometry" not in local_properties:
+            local_properties.append("geometry")
         # id is technically a valid column from the service, but these
         # functions make the name more specific. So, if someone requests
         # 'id', give them the output_id column
-        if "id" in properties:
-            properties[properties.index("id")] = output_id
-        df = df.loc[:, [col for col in properties if col in df.columns]]
+        if "id" in local_properties:
+            local_properties[local_properties.index("id")] = output_id
+        df = df.loc[:, [col for col in local_properties if col in df.columns]]
 
     # Move meaningless-to-user, extra id columns to the end
     # of the dataframe, if they exist

tests/waterdata_utils_test.py

@@ -1,8 +1,10 @@
 from unittest import mock
 
+import pandas as pd
 import requests
 
 from dataretrieval.waterdata.utils import (
+    _arrange_cols,
     _get_args,
     _walk_pages,
 )
@@ -80,3 +82,47 @@ def test_walk_pages_multiple_mocked():
     assert mock_client.send.called
     assert mock_client.request.called
     assert mock_client.request.call_args[0][1] == "https://example.com/page2"
+
+
+# --- _arrange_cols ----------------------------------------------------------
+
+
+def test_arrange_cols_does_not_mutate_caller_properties():
+    """`_arrange_cols` must not mutate the caller's `properties` list.
+
+    Regression: previously the function did
+    ``properties.append("geometry")`` and
+    ``properties[properties.index("id")] = output_id`` in place, so the
+    caller's list grew and was rewritten across successive calls.
+    """
+    df = pd.DataFrame(
+        {
+            "id": ["a", "b"],
+            "value": [1.0, 2.0],
+            "geometry": ["p1", "p2"],
+        }
+    )
+    properties = ["id", "value"]
+    snapshot = list(properties)
+
+    _arrange_cols(df, properties, output_id="daily_id")
+    _arrange_cols(df, properties, output_id="daily_id")
+
+    assert properties == snapshot, (
+        f"caller's properties list was mutated: {properties!r} != {snapshot!r}"
+    )
+
+
+def test_arrange_cols_swaps_id_in_returned_columns():
+    """`'id'` in `properties` should still resolve to the output_id column."""
+    df = pd.DataFrame({"id": ["a"], "value": [1.0]})
+    result = _arrange_cols(df, ["id", "value"], output_id="daily_id")
+    assert "daily_id" in result.columns
+    assert "id" not in result.columns
+
+
+def test_arrange_cols_keeps_geometry_when_present():
+    """Geometry must come along even if the caller didn't list it."""
+    df = pd.DataFrame({"id": ["a"], "value": [1.0], "geometry": ["p1"]})
+    result = _arrange_cols(df, ["value"], output_id="daily_id")
+    assert "geometry" in result.columns