From d1e1e05546f883f245be3c869cad383753aac790 Mon Sep 17 00:00:00 2001
From: Oceania2018 <haiping008@gmail.com>
Date: Sun, 27 Oct 2019 10:46:49 -0500
Subject: [PATCH] inputs for rnn/while/TensorArrayReadV3 are incorrect #433

---
 .../Operations/NnOps/rnn.cs                   | 33 +++++++++++++++----
 1 file changed, 26 insertions(+), 7 deletions(-)

diff --git a/src/TensorFlowNET.Core/Operations/NnOps/rnn.cs b/src/TensorFlowNET.Core/Operations/NnOps/rnn.cs
index 41516bb8..41a4622a 100644
--- a/src/TensorFlowNET.Core/Operations/NnOps/rnn.cs
+++ b/src/TensorFlowNET.Core/Operations/NnOps/rnn.cs
@@ -14,6 +14,7 @@
    limitations under the License.
 ******************************************************************************/
 
+using NumSharp;
 using System;
 using System.Collections.Generic;
 using System.Linq;
@@ -24,7 +25,7 @@ namespace Tensorflow.Operations
 {
     internal class rnn
     {
-        public static (Tensor, Tensor) dynamic_rnn(RNNCell cell, Tensor inputs_tensor,
+        public static (Tensor, Tensor) dynamic_rnn(RnnCell cell, Tensor inputs_tensor,
             Tensor sequence_length = null, Tensor initial_state = null, 
             TF_DataType dtype = TF_DataType.DtInvalid,
             int? parallel_iterations = null, bool swap_memory = false, bool time_major = false)
@@ -79,7 +80,7 @@ namespace Tensorflow.Operations
         /// <param name="sequence_length"></param>
         /// <param name="dtype"></param>
         /// <returns></returns>
-        private static (Tensor, Tensor) _dynamic_rnn_loop(RNNCell cell, Tensor inputs, Tensor initial_state,
+        private static (Tensor, Tensor) _dynamic_rnn_loop(RnnCell cell, Tensor inputs, Tensor initial_state,
             int parallel_iterations, bool swap_memory, Tensor sequence_length = null, TF_DataType dtype = TF_DataType.DtInvalid)
         {
             var state = initial_state;
@@ -170,11 +171,11 @@ namespace Tensorflow.Operations
                         flat_input_i.dtype));
                 }
 
-                for (int i = 0; i < input_ta.Count; i++)
+                input_ta = zip(input_ta, flat_input).Select(x =>
                 {
-                    var (ta, input_) = (input_ta[i], flat_input[i]);
-                    ta.unstack(input_);
-                }
+                    var (ta, input_) = (x.Item1, x.Item2);
+                    return ta.unstack(input_);
+                }).ToList();
             }
 
             // Make sure that we run at least 1 step, if necessary, to ensure
@@ -192,11 +193,29 @@ namespace Tensorflow.Operations
             // Take a time step of the dynamic RNN.
             Func<BodyItemInRnnWhileLoop, BodyItemInRnnWhileLoop> _time_step = (item) =>
             {
+                Tensor[] input_t = null;
+                var (time1, output_ta_t, state1) = (item.time, item.output_ta_t, item.state);
                 if (in_graph_mode)
                 {
-                    input_ta.Select(ta => ta.read(time)).ToArray();
+                    input_t = input_ta.Select(ta => ta.read(time1)).ToArray();
+                    // Restore some shape information
+                    foreach (var (input_, shape) in zip(input_t, inputs_got_shape))
+                        input_.set_shape(shape[new Slice(1)]);
+                }
+                else
+                {
+                    // input_t = tuple(ta[time.numpy()] for ta in input_ta)
                 }
 
+                var input_t_t = nest.pack_sequence_as2(structure: inputs, flat_sequence: input_t);
+                // Keras RNN cells only accept state as list, even if it's a single tensor.
+                // var is_keras_rnn_cell = _is_keras_rnn_cell(cell);
+                (Tensor, Tensor) a = (null, null);
+                if (sequence_length != null)
+                    throw new NotImplementedException("sequence_length != null");
+                else
+                    a = cell.__call__(input_t_t, state1);
+
                 return item;
             };