fraction.py 1.4 KB

12345678910111213141516171819202122232425262728293031323334353637383940
  1. import pynini
  2. from fun_text_processing.inverse_text_normalization.ja.graph_utils import DAMO_NOT_QUOTE, GraphFst, delete_space, DAMO_SIGMA, delete_extra_space
  3. from pynini.lib import pynutil
  4. class FractionFst(GraphFst):
  5. """
  6. Finite state transducer for verbalizing fraction,
  7. e.g. fraction { numerator: "2" denominator: "3" } } -> 2/3
  8. e.g. fraction { numerator: "20" denominator: "3" negative: "true"} } -> 2/3
  9. """
  10. def __init__(self):
  11. super().__init__(name="fraction", kind="verbalize")
  12. optional_sign = pynini.closure(pynini.cross('negative: "true"', "-") + delete_space, 0, 1)
  13. numerator = pynutil.delete('numerator: "') + pynini.closure(DAMO_NOT_QUOTE, 1) + pynutil.delete('"')
  14. denominator = (
  15. pynutil.insert("/")
  16. + pynutil.delete('denominator: "')
  17. + pynini.closure(DAMO_NOT_QUOTE, 1)
  18. + pynutil.delete('"')
  19. )
  20. graph = (numerator + delete_space + denominator).optimize()
  21. # numerator = pynutil.delete('numerator: "') + DAMO_NOT_QUOTE + pynutil.delete('"')
  22. #
  23. # denominator = (
  24. # pynutil.delete('denominator: "')
  25. # + DAMO_NOT_QUOTE
  26. # + pynutil.delete('"')
  27. # )
  28. #
  29. # graph = (numerator + pynutil.insert("/") + denominator).optimize()
  30. self.numbers = graph
  31. delete_tokens = self.delete_tokens(optional_sign + graph)
  32. self.fst = delete_tokens.optimize()