Fix (#73270)

Author: co63oc

test/legacy_test/test_index_select_op.py

@@ -92,6 +92,55 @@ def init_dtype_type(self):
         self.index_size = 10
 
 
+class TestIndexSelectOp_ZeroSize(OpTest):
+    def setUp(self):
+        self.python_api = paddle.index_select
+        self.public_python_api = paddle.index_select
+        self.op_type = "index_select"
+        self.init_dtype_type()
+
+        index_np = np.random.randint(
+            low=-self.x_shape[self.dim],
+            high=self.x_shape[self.dim],
+            size=self.index_size,
+        )
+        x_np = np.random.random(self.x_shape).astype(self.x_type)
+        if self.dtype == np.complex64 or self.dtype == np.complex128:
+            x_np = (
+                np.random.random(self.x_shape)
+                + 1j * np.random.random(self.x_shape)
+            ).astype(self.x_type)
+        self.inputs = {'X': x_np, 'Index': index_np}
+        self.attrs = {'dim': self.dim}
+        outer_loop = np.prod(self.x_shape[: self.dim])
+        x_reshape = [outer_loop, *self.x_shape[self.dim :]]
+        x_np_reshape = np.reshape(x_np, tuple(x_reshape))
+        out_list = []
+        for i in range(outer_loop):
+            for j in range(self.index_size):
+                out_list.append(x_np_reshape[i, index_np[j]])
+        self.out_shape = list(self.x_shape)
+        self.out_shape[self.dim] = self.index_size
+        self.out_shape = tuple(self.out_shape)
+
+        out = np.reshape(out_list, self.out_shape)
+        self.outputs = {'Out': out}
+
+    def test_check_output(self):
+        self.check_output(check_pir=True)
+
+    def test_check_grad_normal(self):
+        self.check_grad(['X'], 'Out', check_pir=True)
+
+    def init_dtype_type(self):
+        self.x_type = np.float64
+        self.index_type = np.int64
+        self.dim = 1
+        # shape[dim] can not be 0.
+        self.x_shape = (0, 10, 0, 0)
+        self.index_size = 10
+
+
 class TestIndexSelectOpCaseSingleThread(TestIndexSelectOp):
     def init_dtype_type(self):
         if base.is_compiled_with_cuda():