decimal.py 2.1 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455
  1. import pynini
  2. from fun_text_processing.text_normalization.en.graph_utils import DAMO_NOT_QUOTE, GraphFst, delete_space, insert_space
  3. from pynini.lib import pynutil
  4. class DecimalFst(GraphFst):
  5. """
  6. Finite state transducer for verbalizing decimal, e.g.
  7. decimal { negative: "true" integer_part: "twelve" fractional_part: "five o o six" quantity: "billion" } -> minus twelve point five o o six billion
  8. Args:
  9. deterministic: if True will provide a single transduction option,
  10. for False multiple transduction are generated (used for audio-based normalization)
  11. """
  12. def __init__(self, cardinal, deterministic: bool = True):
  13. super().__init__(name="decimal", kind="verbalize", deterministic=deterministic)
  14. self.optional_sign = pynini.cross("negative: \"true\"", "minus ")
  15. if not deterministic:
  16. self.optional_sign |= pynini.cross("negative: \"true\"", "negative ")
  17. self.optional_sign = pynini.closure(self.optional_sign + delete_space, 0, 1)
  18. self.integer = pynutil.delete("integer_part:") + cardinal.integer
  19. self.optional_integer = pynini.closure(self.integer + delete_space + insert_space, 0, 1)
  20. self.fractional_default = (
  21. pynutil.delete("fractional_part:")
  22. + delete_space
  23. + pynutil.delete("\"")
  24. + pynini.closure(DAMO_NOT_QUOTE, 1)
  25. + pynutil.delete("\"")
  26. )
  27. self.fractional = pynutil.insert("point ") + self.fractional_default
  28. self.quantity = (
  29. delete_space
  30. + insert_space
  31. + pynutil.delete("quantity:")
  32. + delete_space
  33. + pynutil.delete("\"")
  34. + pynini.closure(DAMO_NOT_QUOTE, 1)
  35. + pynutil.delete("\"")
  36. )
  37. self.optional_quantity = pynini.closure(self.quantity, 0, 1)
  38. graph = self.optional_sign + (
  39. self.integer
  40. | (self.integer + self.quantity)
  41. | (self.optional_integer + self.fractional + self.optional_quantity)
  42. )
  43. self.numbers = graph
  44. delete_tokens = self.delete_tokens(graph)
  45. self.fst = delete_tokens.optimize()