Merge pull request #6558 from markotoplak/dask-zero-len

DaskTable: fix loading zero length dask arrays
biolab · Aug 30, 2023 · 1431a72 · 1431a72
2 parents 51beee7 + ef4fd0c
commit 1431a72
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 3 deletions.
diff --git a/Orange/data/dask.py b/Orange/data/dask.py
@@ -86,10 +86,10 @@ def from_arrays(cls, domain, X=None, Y=None, metas=None):
         assert size is not None
 
         if X is None:
-            X = da.zeros((size, 0), chunks=(size, 0))
+            X = da.zeros((size, len(domain.attributes)), chunks=(size, -1))
 
         if Y is None:
-            Y = da.zeros((size, 0), chunks=(size, 0))
+            Y = da.zeros((size, len(domain.class_vars)), chunks=(size, -1))
 
         if metas is None:
             metas = np.zeros((size, 0))

diff --git a/Orange/tests/test_dasktable.py b/Orange/tests/test_dasktable.py
@@ -90,7 +90,7 @@ def same_tables(self, table, dasktable):
         numpy.testing.assert_equal(dasktable.Y, table.Y)
         numpy.testing.assert_equal(dasktable.metas, table.metas)
 
-    def test_zero_size_dask_arrays(self):
+    def test_zero_width_dask_arrays(self):
         empty = Table.from_numpy(Domain([], [ContinuousVariable("y")]),
                                 X=np.ones((10**5, 0)),
                                 Y=np.ones((10**5, 1)))
@@ -111,6 +111,30 @@ def test_zero_size_dask_arrays(self):
             self.assertEqual(data.X.shape, (10**5, 1))
             self.assertEqual(data._Y.shape, (10**5, 0))
 
+    def test_zero_len_dask_arrays(self):
+        atts = [ContinuousVariable("x"), ContinuousVariable("y")]
+        empty = Table.from_numpy(Domain([], atts),
+                                 X=np.ones((0, 0)),
+                                 Y=np.ones((0, 2)))
+
+        with open_as_dask(empty) as data:
+            self.assertIsInstance(data, DaskTable)
+            self.assertIsInstance(data.X, da.Array)
+            self.assertIsInstance(data._Y, da.Array)
+            self.assertEqual(data.X.shape, (0, 0))
+            self.assertEqual(data._Y.shape, (0, 2))
+
+        empty = Table.from_numpy(Domain(atts, []),
+                                 X=np.ones((0, 2)),
+                                 Y=np.ones((0, 0)))
+
+        with open_as_dask(empty) as data:
+            self.assertIsInstance(data, DaskTable)
+            self.assertIsInstance(data.X, da.Array)
+            self.assertIsInstance(data._Y, da.Array)
+            self.assertEqual(data.X.shape, (0, 2))
+            self.assertEqual(data._Y.shape, (0, 0))
+
     def test_compute(self):
         zoo = Table('zoo')
         with named_file('', suffix='.hdf5') as fn: