Merge pull request #1150 from wright-group/issue-1118

safeguard join "method" kwarg

CHANGELOG.md

@@ -7,6 +7,10 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/).
 
 ### Fixed
 - `interact2D`: fixed bug where use_imshow broke the sliders
+- `data.join` ensures valid `method` is selected
+
+### Changed
+- `data.join` no longer supports `sum` method
 
 ## [3.5.0]
 

WrightTools/data/_join.py

@@ -62,7 +62,7 @@ def join(
         The name for the data object which is created. Default is 'join'.
     parent : WrightTools.Collection (optional)
         The location to place the joined data object. Default is new temp file at root.
-    method : {'first', 'last', 'min', 'max', 'sum', 'mean'}
+    method : {'first', 'last', 'min', 'max', 'mean'}
         Mode to use for merged points in the joined space.
         Default is 'first'.
     verbose : bool (optional)
@@ -76,6 +76,11 @@ def join(
     warnings.warn("join", category=wt_exceptions.EntireDatasetInMemoryWarning)
     if isinstance(datas, Collection):
         datas = datas.values()
+    valid_methods = ["first", "last", "min", "max", "mean"]
+    if method not in valid_methods:
+        if method == "sum":
+            raise ValueError(f"method 'sum' is deprecated; consider 'mean' instead.")
+        raise ValueError(f"invalid method {method!r}: expected {valid_methods}")
     datas = list(datas)
     if not isinstance(atol, collections.abc.Iterable):
         atol = [atol] * len(datas[0].axes)
@@ -177,7 +182,7 @@ def get_shape(out, datas, item_name):
         out.create_channel(
             shape=get_shape(out, datas, channel_name),
             **datas[0][channel_name].attrs,
-            dtype=datas[0][channel_name].dtype
+            dtype=datas[0][channel_name].dtype,
         )
         count[channel_name] = np.zeros_like(out[channel_name], dtype=int)
     for variable_name in variable_names:
@@ -186,7 +191,7 @@ def get_shape(out, datas, item_name):
             out.create_variable(
                 shape=get_shape(out, datas, variable_name),
                 **datas[0][variable_name].attrs,
-                dtype=datas[0][variable_name].dtype
+                dtype=datas[0][variable_name].dtype,
             )
             count[variable_name] = np.zeros_like(out[variable_name], dtype=int)
 

examples/join.py

@@ -25,12 +25,11 @@
 last = wt.data.join([a, b], method="last", name="last")
 min = wt.data.join([a, b], method="min", name="min")
 max = wt.data.join([a, b], method="max", name="max")
-sum = wt.data.join([a, b], method="sum", name="sum")
 mean = wt.data.join([a, b], method="mean", name="mean")
 
 # Plot the splits in columns
 fig, gs = wt.artists.create_figure(nrows=4, cols=[1, 1])
-for i, da in enumerate([a, b, first, last, min, max, sum, mean]):
+for i, da in enumerate([a, b, first, last, min, max, mean]):
     ax = plt.subplot(gs[i])
     ax.pcolor(da, vmin=0, vmax=6)
     wt.artists.corner_text(da.natural_name, ax=ax)

tests/data/join.py

@@ -515,30 +515,6 @@ def test_overlap_last():
     joined.close()
 
 
-def test_overlap_sum():
-    a = wt.Data()
-    b = wt.Data()
-
-    a.create_variable("x", np.linspace(0, 10, 11))
-    b.create_variable("x", np.linspace(5, 15, 11))
-    a.transform("x")
-    b.transform("x")
-    a.create_channel("y", np.ones_like(a.x[:]))
-    b.create_channel("y", np.ones_like(b.x[:]) * 2)
-
-    joined = wt.data.join([a, b], method="sum")
-
-    assert joined.shape == (16,)
-    assert np.allclose(joined.x.points, np.linspace(0, 15, 16))
-    assert np.isclose(joined.y[0], 1.0)
-    assert np.isclose(joined.y[10], 3.0)
-    assert np.isclose(joined.y[-1], 2.0)
-
-    a.close()
-    b.close()
-    joined.close()
-
-
 def test_overlap_max():
     a = wt.Data()
     b = wt.Data()