Merge pull request #620 from lei-Qiao/improve_test_coverage

Improve test coverage

Author: wayuanho

tests/test_backend.py

@@ -108,24 +108,17 @@ def _run_test_case(self, output_names_with_port, feed_dict, **kwargs):
         kwargs["constant_fold"] = False
         return self.run_test_case(feed_dict, [], output_names_with_port, **kwargs)
 
-    def _test_expand_dims(self, idx):
+    def _test_expand_dims_known_rank(self, idx):
         tf.reset_default_graph()
         x_val = make_xval([3, 4])
         x = tf.placeholder(tf.float32, shape=x_val.shape, name=_TFINPUT)
         op = tf.expand_dims(x, idx)
         _ = tf.identity(op, name=_TFOUTPUT)
         self._run_test_case([_OUTPUT], {_INPUT: x_val})
 
-    def test_expand_dims(self):
+    def test_expand_dims_known_rank(self):
         for i in [-1, 0, 1, -2]:
-            self._test_expand_dims(i)
-
-    def test_expand_dims_dynamic_inputs(self):
-        x_val = make_xval([3, 4])
-        x = tf.placeholder(tf.float32, shape=[None, None], name=_TFINPUT)
-        op = tf.expand_dims(x, 0)
-        _ = tf.identity(op, name=_TFOUTPUT)
-        self._run_test_case([_OUTPUT], {_INPUT: x_val})
+            self._test_expand_dims_known_rank(i)
 
     def test_expand_dims_one_unknown_rank(self):
         tf.reset_default_graph()
@@ -135,14 +128,18 @@ def test_expand_dims_one_unknown_rank(self):
         _ = tf.identity(op, name=_TFOUTPUT)
         self._run_test_case([_OUTPUT], {_INPUT: x_val})
 
-    def test_expand_dims_more_unknown_rank(self):
+    def _test_expand_dims_more_unknown_rank(self, idx):
         tf.reset_default_graph()
         x_val = make_xval([3, 4])
         x = tf.placeholder(tf.float32, shape=[None, None], name=_TFINPUT)
-        op = tf.expand_dims(x, 0)
+        op = tf.expand_dims(x, idx)
         _ = tf.identity(op, name=_TFOUTPUT)
         self._run_test_case([_OUTPUT], {_INPUT: x_val})
 
+    def test_expand_dims_more_unknown_rank(self):
+        for i in [-1, 0, 1, -2]:
+            self._test_expand_dims_more_unknown_rank(i)
+
     @check_opset_min_version(9, "ConstantOfShape")
     def test_eye_non_const1(self):
         # tf.eye(num_rows), num_rows is not const here
@@ -1073,6 +1070,15 @@ def test_slice(self):
         _ = tf.identity(x_, name=_TFOUTPUT)
         self._run_test_case([_OUTPUT], {_INPUT: x_val})
 
+    def test_slice_neg_size(self):
+        x_val = np.array([[1, 2, 3, 4], [5, 6, 7, 8]], dtype=np.float32)
+        t1 = tf.constant([0, 1], dtype=tf.int32)
+        t2 = tf.constant([-1, 2], dtype=tf.int32)
+        x0 = tf.placeholder(tf.float32, x_val.shape, name=_TFINPUT)
+        x_ = tf.slice(x0, t1, t2)
+        _ = tf.identity(x_, name=_TFOUTPUT)
+        self._run_test_case([_OUTPUT], {_INPUT: x_val})
+
     @check_opset_min_version(10, "Slice in opset 10 can accept dymaic 'start' and 'ends'")
     def test_slice_with_non_const(self):
         x_val = np.array([[1, 2, 3, 4], [5, 6, 7, 8]], dtype=np.float32)
@@ -2253,7 +2259,6 @@ def test_sparse_softmax_cross_entropy_with_logits_large_class(self):
 
         self._run_test_case([_OUTPUT], {_INPUT: label_val, _INPUT1: logits_val}, rtol=1e-6)
 
-    @skip_onnxruntime_backend("onnxruntime Slice did not supported BOOL")
     def test_matrix_band_part(self):
         input_val = np.random.randint(0, 666, (10, 15)).astype(np.int32)
         input_x = tf.placeholder(dtype=tf.int32, shape=[None, None], name=_TFINPUT)
@@ -2263,7 +2268,6 @@ def test_matrix_band_part(self):
         _ = tf.identity(res1, name=_TFOUTPUT1)
         self._run_test_case([_OUTPUT, _OUTPUT1], {_INPUT: input_val})
 
-    @skip_onnxruntime_backend("onnxruntime Slice did not supported BOOL.")
     def test_matrix_band_part_2(self):
         input_val = np.random.randint(0, 666, (1, 1)).astype(np.int32)
         input_x = tf.placeholder(dtype=tf.int32, shape=[None, None], name=_TFINPUT)
@@ -2432,7 +2436,7 @@ def test_softsign(self):
 
     def test_batch_to_spacend(self):
         block_size = [2, 2]
-        crop = [[0, 1], [2, 1]]
+        crop = [[1, 0], [2, 1]]
 
         input_val = np.random.random_sample([40, 3, 5, 100]).astype(np.float32)
         input_x = tf.placeholder(dtype=tf.float32, shape=input_val.shape, name=_TFINPUT)  # NHWC

tf2onnx/onnx_opset/math.py

@@ -251,12 +251,9 @@ def version_1(cls, ctx, node, **kwargs):
         # ONNX: Each input value is divided by (bias+(alpha/size)*sum(xi^2 for every xi in the local region))^beta
         # TF: sqr_sum[a, b, c, d] = sum(input[a, b, c, d - depth_radius : d + depth_radius + 1] ** 2)
         #     output = input / (bias + alpha * sqr_sum) ** beta
-        depth_radius = node.get_attr("depth_radius")
-        if depth_radius:
-            size = depth_radius.i * 2 + 1
-        else:
-            # by default, depth_radius is 5 in tensorflow
-            size = 5 * 2 + 1
+
+        # by default, depth_radius is 5 in tensorflow
+        size = node.get_attr_value("depth_radius", 5) * 2 + 1
 
         node.set_attr("size", size)
         node.set_attr("alpha", size * node.get_attr("alpha").f)

tf2onnx/onnx_opset/tensor.py

@@ -1278,11 +1278,7 @@ def version_10(cls, ctx, node, **kwargs):
         seq_dim = node.get_attr("seq_dim")
         utils.make_sure(seq_dim is not None, "sequence dim must be given in {}".format(node.name))
         seq_dim = seq_dim.i
-        batch_dim = node.get_attr("batch_dim")
-        if batch_dim is not None:
-            batch_dim = batch_dim.i
-        else:
-            batch_dim = 0
+        batch_dim = node.get_attr_value("batch_dim", 0)
 
         ctx.remove_node(node.name)
         node = ctx.make_node(