How to use tf.keras.texts_to_sequences in tf.dataset

Question

I have a text classification task whose feeding files are lines of text. Before, I use keras and generator to handle with massive dataset. Now I want to use tf.dataset which