tensorflow: Issue with retrain.py, "could not convert string to float" when creating bottlenecks

So I placed some checks in the retrain.py file and its working fine now. From What I understand this is not because of a faulty image file but a faulty bottleneck file. As soon as the error was caught by my checks, the script started creating new bottleneck files instead of checking the cached ones.

In other words, if an incomplete bottleneck file exists in the cache then this error arises (this file can be produced because of sudden system shutdown).

Please add some checks to the code. I’ll paste my solution below:

#retrain.py
...
def get_or_create_bottleneck(sess, image_lists, label_name, index, image_dir,
                             category, bottleneck_dir, jpeg_data_tensor,
                             bottleneck_tensor):
...
  with open(bottleneck_path, 'r') as bottleneck_file:
    bottleneck_string = bottleneck_file.read()
  
  #avoiding non float values
    try:
      bottleneck_values = [float(x) for x in bottleneck_string.split(',')]
    except:
      print("Invalid float found, sending None instead")
      return None

  return bottleneck_values

...
def cache_bottlenecks(sess, image_lists, image_dir, bottleneck_dir,
                      jpeg_data_tensor, bottleneck_tensor):
...
      for index, unused_base_name in enumerate(category_list):
        bNeck = get_or_create_bottleneck(sess, image_lists, label_name, index,
                                 image_dir, category, bottleneck_dir,
                                 jpeg_data_tensor, bottleneck_tensor)
        if bNeck is not None:
          how_many_bottlenecks += 1
          if how_many_bottlenecks % 100 == 0:
            print(str(how_many_bottlenecks) + ' bottleneck files created.')

...
def get_random_cached_bottlenecks(sess, image_lists, how_many, category,
                                  bottleneck_dir, image_dir, jpeg_data_tensor,
                                  bottleneck_tensor):
...
  for unused_i in range(how_many):
    label_index = random.randrange(class_count)
    label_name = list(image_lists.keys())[label_index]
    image_index = random.randrange(MAX_NUM_IMAGES_PER_CLASS + 1)
    bottleneck = get_or_create_bottleneck(sess, image_lists, label_name,
                                          image_index, image_dir, category,
                                          bottleneck_dir, jpeg_data_tensor,
                                          bottleneck_tensor)
    if bottleneck is not None:
      ground_truth = np.zeros(class_count, dtype=np.float32)
      ground_truth[label_index] = 1.0
      bottlenecks.append(bottleneck)
      ground_truths.append(ground_truth)
  return bottlenecks, ground_truths

...
def get_random_distorted_bottlenecks(
    sess, image_lists, how_many, category, image_dir, input_jpeg_tensor,
    distorted_image, resized_input_tensor, bottleneck_tensor):
...
    bottleneck = run_bottleneck_on_image(sess, distorted_image_data,
                                         resized_input_tensor,
                                         bottleneck_tensor)
    if bottleneck is not None:
      ground_truth = np.zeros(class_count, dtype=np.float32)
      ground_truth[label_index] = 1.0
      bottlenecks.append(bottleneck)
      ground_truths.append(ground_truth)
  return bottlenecks, ground_truths