MLBEDSW-3301: Vela fails ungracefully when reading string buffers

When encountering a sparse string buffer, Vela fails
both due to missing a mapping for a Numpy string type
and also for not being able to read sparse buffers.
The failing line is attempting to reshape a [100]
buffer into a [3, 5] tensor which does not work due
to Vela treating the buffer as non-sparse.
The solution here is to simply not do the reshape
for string buffers (which all appear to be sparse)
since it is not something that will be supported in
the future anyway.
The related operator can then be pushed to the CPU
as expected.

Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: Iea0af6cd60a691f975209014b6aa098dde8d6a4b

3 files changed, 38 insertions, 3 deletions

diff --git a/ethosu/vela/test/test_tflite_reader.py b/ethosu/vela/test/test_tflite_reader.py
index 23abb4a..14c9b20 100644
--- a/ethosu/vela/test/test_tflite_reader.py
+++ b/ethosu/vela/test/test_tflite_reader.py
@@ -18,9 +18,11 @@
 from unittest.mock import MagicMock
 from unittest.mock import patch
 
+import numpy as np
 import pytest
 
 from ethosu.vela.operation import Op
+from ethosu.vela.tflite.TensorType import TensorType
 from ethosu.vela.tflite_reader import TFLiteSubgraph
 
 
@@ -79,3 +81,34 @@ class TestTFLiteSubgraph:
             assert len(created_op.inputs) == expected
             assert created_op.outputs[0].name == "tensor_{}".format(output)
             assert inputs[-1] != -1 or not created_op.inputs[-1]
+
+    string_buffer_testdata = [
+        (np.array([np.random.randint(256) for _ in range(100)], dtype=np.uint8), [3, 5]),
+        (np.array([np.random.randint(256) for _ in range(100)], dtype=np.int16), [10, 10]),
+        (np.array([np.random.randint(256) for _ in range(100)], dtype=np.float32), [100]),
+        (np.array([], dtype=np.int8), [30]),
+    ]
+
+    @pytest.mark.parametrize("buffer, tens_shape", string_buffer_testdata)
+    def test_parse_tensor_with_string_buffer(self, buffer, tens_shape):
+        tens_data = MagicMock()
+        tens_data.ShapeAsNumpy = MagicMock(return_value=np.array(tens_shape), dtype=np.int32)
+        tens_data.Name = MagicMock(return_value=b"test_data")
+        tens_data.Type = MagicMock(return_value=TensorType.STRING)
+        tens_data.Quantization = MagicMock(return_value=None)
+        tens_data.Buffer = MagicMock(return_value=0)
+
+        tfl_sg = MagicMock()
+        tfl_sg.Name = MagicMock(return_value=b"test_sg")
+        tfl_sg.TensorsLength = MagicMock(return_value=0)
+        tfl_sg.OperatorsLength = MagicMock(return_value=0)
+        tfl_sg.OutputsAsNumpy = MagicMock(return_value=[])
+        tfl_sg.InputsAsNumpy = MagicMock(return_value=[])
+
+        graph = MagicMock()
+        graph.buffers = [buffer]
+
+        subgraph = TFLiteSubgraph(graph, tfl_sg)
+
+        tens = subgraph.parse_tensor(tens_data)
+        assert tens.values is None
diff --git a/ethosu/vela/tflite_mapping.py b/ethosu/vela/tflite_mapping.py
index bbd44b0..44ecedc 100644
--- a/ethosu/vela/tflite_mapping.py
+++ b/ethosu/vela/tflite_mapping.py
@@ -171,6 +171,7 @@ datatype_map_numpy = {
     TensorType.BOOL: np.bool,
     TensorType.COMPLEX64: np.complex64,
     TensorType.COMPLEX128: np.complex128,
+    TensorType.STRING: np.dtype("S1"),
 }
 
 
diff --git a/ethosu/vela/tflite_reader.py b/ethosu/vela/tflite_reader.py
index b3b0720..c190f7e 100644
--- a/ethosu/vela/tflite_reader.py
+++ b/ethosu/vela/tflite_reader.py
@@ -106,7 +106,8 @@ class TFLiteSubgraph:
         np_shape = tens_data.ShapeAsNumpy()
         shape = list(np_shape) if type(np_shape) is np.ndarray else []
         name = decode_str(tens_data.Name())
-        dtype = datatype_map[tens_data.Type()]
+        tens_dtype = tens_data.Type()
+        dtype = datatype_map[tens_dtype]
         tens = Tensor(shape, dtype, name)
         quant = tens_data.Quantization()
 
@@ -129,8 +130,8 @@ class TFLiteSubgraph:
 
         tens.values = None
         buf = self.graph.buffers[tens_data.Buffer()]
-        if buf is not None:
-            tens.values = np.array(buf.view(datatype_map_numpy[tens_data.Type()]).reshape(shape))
+        if buf is not None and dtype != DataType.string:
+            tens.values = np.array(buf.view(datatype_map_numpy[tens_dtype]).reshape(shape))
             if tens.quantization is not None:
                 tens.quant_values = tens.values
                 tens.values = tens.quantization.dequantize(tens.quant_values)