1 files changed, 173 insertions, 0 deletions
diff --git a/compiler/tfkit/src/PackCommand.cpp b/compiler/tfkit/src/PackCommand.cpp
new file mode 100644
index 000000000..36bf5a71c
--- /dev/null
+++ b/compiler/tfkit/src/PackCommand.cpp
@@ -0,0 +1,173 @@
+/*
+ * Copyright (c) 2019 Samsung Electronics Co., Ltd. All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "PackCommand.hpp"
+#include "Support.hpp"
+
+#include <tensorflow/core/framework/graph.pb.h>
+
+#include <google/protobuf/io/coded_stream.h>
+#include <google/protobuf/io/zero_copy_stream_impl.h>
+#include <google/protobuf/text_format.h>
+
+#include <cassert>
+#include <stdexcept>
+#include <vector>
+
+namespace
+{
+
+template <typename T> void pack(tensorflow::TensorProto *);
+
+template <> void pack<float>(tensorflow::TensorProto *input_tensor)
+{
+  const auto &input_shape = input_tensor->tensor_shape();
+  assert(input_shape.dim_size() <= 6);
+  int input_flat_size = tfkit::tf::GetElementCount(input_shape);
+
+  // Adjust where shape is not set but actual value exist
+  if (input_tensor->float_val().size() > 0 && input_flat_size == -1)
+  {
+    input_flat_size = input_tensor->float_val().size();
+  }
+
+  if (input_tensor->float_val().size() == 0)
+  {
+    // There may be tensor_content and we don't need to do anything as it is
+    // already packed format
+  }
+  else if (input_tensor->float_val().size() == input_flat_size)
+  {
+    // clang-format off
+  // TODO fix indentation
+  input_tensor->clear_tensor_content();
+
+  std::vector<float> tensor_content;
+  for (int i = 0; i < input_flat_size; ++i)
+  {
+    tensor_content.push_back(input_tensor->float_val(i));
+  }
+
+  input_tensor->set_tensor_content(std::string(
+      reinterpret_cast<const char *>(tensor_content.data()), sizeof(float) * input_flat_size));
+
+  input_tensor->clear_float_val();
+    // clang-format on
+  }
+  else
+  {
+    throw std::runtime_error{"Number of elements mismatch in pack<float>."};
+    // TODO: support for these
+  }
+}
+
+template <> void pack<int32_t>(tensorflow::TensorProto *input_tensor)
+{
+  const auto &input_shape = input_tensor->tensor_shape();
+  assert(input_shape.dim_size() <= 6);
+  int input_flat_size = tfkit::tf::GetElementCount(input_shape);
+
+  // Adjust where shape is not set but actual value exist
+  if (input_tensor->int_val().size() > 0 && input_flat_size == -1)
+  {
+    input_flat_size = input_tensor->int_val().size();
+  }
+
+  if (input_tensor->int_val().size() == 0)
+  {
+    // There may be tensor_content and we don't need to do anything as it is
+    // already packed format
+  }
+  else if (input_tensor->int_val().size() == input_flat_size)
+  {
+    input_tensor->clear_tensor_content();
+
+    std::vector<int32_t> tensor_content;
+    for (int i = 0; i < input_flat_size; ++i)
+    {
+      tensor_content.push_back(input_tensor->int_val(i));
+    }
+
+    input_tensor->set_tensor_content(std::string(
+        reinterpret_cast<const char *>(tensor_content.data()), sizeof(int32_t) * input_flat_size));
+
+    input_tensor->clear_int_val();
+  }
+  else
+  {
+    throw std::runtime_error{"Number of elements mismatch in pack<int32_t>."};
+    // TODO: support for these
+  }
+}
+
+void pack(tensorflow::GraphDef &graph_def)
+{
+  auto nodes = graph_def.mutable_node();
+  for (int i = 0; i < nodes->size(); ++i)
+  {
+    tensorflow::NodeDef *n = nodes->Mutable(i);
+    // TODO: handle for other operators
+    if (n->op() == "Const")
+    {
+      const auto dtype = tfkit::tf::GetDataTypeAttr(*n, "dtype");
+      tensorflow::TensorProto *tensor = tfkit::tf::GetTensorAttr(*n, "value");
+
+      switch (dtype)
+      {
+        case tensorflow::DT_FLOAT:
+          pack<float>(tensor);
+          break;
+        case tensorflow::DT_INT32:
+          pack<int32_t>(tensor);
+          break;
+        default:
+          throw std::runtime_error{"Unsupported dtype"};
+      }
+    }
+  }
+}
+
+} // namespace
+
+namespace tfkit
+{
+
+int PackCommand::run(int argc, const char *const *argv) const
+{
+  tensorflow::GraphDef graph_def;
+
+  CmdArguments cmdargs(argc, argv);
+
+  auto ioconfig = make_ioconfig(cmdargs);
+
+  google::protobuf::io::IstreamInputStream is{ioconfig->in()};
+
+  if (!google::protobuf::TextFormat::Parse(&is, &graph_def))
+  {
+    std::cerr << "ERROR: Failed to parse prototxt" << std::endl;
+    return 255;
+  }
+
+  // convert float_val to tensor_content
+  pack(graph_def);
+
+  google::protobuf::io::OstreamOutputStream os{ioconfig->out()};
+  google::protobuf::TextFormat::Print(graph_def, &os);
+
+  return 0;
+}
+
+} // namespace tfkit