from tensorflow.python import pywrap_tensorflow as _pywrap_tensorflow  # pylint: disable=unused-import
from tensorflow.python.tools import module_util as _module_util
from tensorflow.python.util.lazy_loader import KerasLazyLoader as _KerasLazyLoader

# Make sure code inside the TensorFlow codebase can use tf2.enabled() at import.
_os.environ["TF2_BEHAVIOR"] = "1"
from tensorflow.python import tf2 as _tf2

  TracingContext* tracing_ctx = dyn_cast<TracingContext>(unwrap(ctx));
  if (!tracing_ctx) {
    tsl::Set_TF_Status_from_Status(
        s, tensorflow::errors::InvalidArgument(
               "Only TracingContext can be converted into a function."));
    return nullptr;
  }
  tsl::Set_TF_Status_from_Status(s,
                                 tracing_ctx->Finalize(unwrap(outputs), &func));
  TF_DeleteExecutionContext(ctx);

      ->set_disable_optimize_for_static_graph(true);
}

void TF_TensorFromProto(const TF_Buffer* from, TF_Tensor* to,
                        TF_Status* status) {
  TF_SetStatus(status, TF_OK, "");
  tensorflow::TensorProto from_tensor_proto;
  status->status = BufferToMessage(from, &from_tensor_proto);
  if (!status->status.ok()) {
    return;
  }
  status->status =

     *
     *    dA = U
     *    dbias = reduceSum(U, dims = channel_dim)
     */

    AbstractTensorHandle* upstream_grad = grad_outputs[0];
    DCHECK(upstream_grad);

    // Recover data format from forward pass for gradient.
    std::string data_format;
    TF_RETURN_IF_ERROR(forward_attrs_.Get("data_format", &data_format));

    // Grad for A
    grad_inputs[0] = upstream_grad;
    grad_inputs[0]->Ref();

  for (int i{}; i < inputs.size(); ++i) {
    theta_inputs[i] = inputs[i];
  }

  AbstractTensorHandle* theta =
      theta_inputs[input_index];  // parameter we are grad checking

  // Convert from AbstractTensor to TF_Tensor.
  TF_Tensor* theta_tensor;
  TF_RETURN_IF_ERROR(GetValue(theta, &theta_tensor));

  // Get number of elements and fill data.
  int num_elems = TF_TensorElementCount(theta_tensor);

  tensorflow::FunctionDef fdef;
  status->status = tensorflow::GraphToFunctionDef(
      fn_body->graph, fn_name, append_hash_to_fn_name != 0,
      /*set_stateful_from_nodes=*/true,
      /*copy_placeholder_attrs_from_nodes=*/true, body_nodes, input_tensors,
      output_tensors, output_names_vec, control_output_nodes,
      control_output_names_vec, description, &fdef);
  if (TF_GetCode(status) != TF_OK) {

                    CancellationManager& cancellation_manager,
                    absl::optional<int64_t> step_id = absl::nullopt);
  // Block until the previous `StartExecute` operation has executed. Forwards
  // the status from `TFE_Execute` and returns outputs if the status is OK.
  std::vector<TensorHandlePtr> Join(TF_Status* status);

  // Block until all Ops finished running on the thread.
  void AsyncWait(TF_Status* status);

 private:

      TestScalarTensorHandle<T, datatype>(eager_ctx, value);
  *tensor =
      unwrap(TF_CreateAbstractTensorFromEagerTensor(input_eager, status.get()));
  return absl::OkStatus();
}

// Places data from `t` into *result_tensor.
Status GetValue(AbstractTensorHandle* t, TF_Tensor** result_tensor);
}  // namespace tensorflow

  TF_Graph* g1 = TF_NewGraph();
  TF_Graph* g2 = TF_NewGraph();

  TF_Operation* feed = Placeholder(g1, s);
  ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);

  // Attempt to create node in g2 with input from g1
  Neg(feed, g2, s);
  EXPECT_EQ(TF_INVALID_ARGUMENT, TF_GetCode(s));
  EXPECT_STREQ("foo", TF_Message(s));

  TF_DeleteGraph(g1);
  TF_DeleteGraph(g2);
  TF_DeleteStatus(s);
}
*/

"""configure script to get build parameters from user."""

import argparse
import errno
import glob
import json
import os
import platform
import re
import subprocess
import sys

# pylint: disable=g-import-not-at-top
try:
  from shutil import which
except ImportError:
  from distutils.spawn import find_executable as which
# pylint: enable=g-import-not-at-top