time.py 2.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778
  1. import pynini
  2. from fun_text_processing.inverse_text_normalization.ko.graph_utils import (
  3. DAMO_CHAR,
  4. DAMO_DIGIT,
  5. GraphFst,
  6. delete_space,
  7. insert_space,
  8. )
  9. from pynini.lib import pynutil
  10. class TimeFst(GraphFst):
  11. """
  12. Finite state transducer for verbalizing time, e.g.
  13. time { hours: "12" minutes: "30" } -> 12:30
  14. time { hours: "1" minutes: "12" } -> 01:12
  15. time { hours: "2" suffix: "a.m." } -> 02:00 a.m.
  16. """
  17. def __init__(self):
  18. super().__init__(name="time", kind="verbalize")
  19. add_leading_zero_to_double_digit = (DAMO_DIGIT + DAMO_DIGIT) | (pynutil.insert("0") + DAMO_DIGIT)
  20. #hour
  21. hour = (
  22. pynutil.delete("hours:")
  23. + delete_space
  24. + pynutil.delete("\"")
  25. + pynini.closure(DAMO_DIGIT, 1)
  26. + pynutil.delete("\"")
  27. )
  28. #minute
  29. minute = (
  30. pynutil.delete("minutes:")
  31. + delete_space
  32. + pynutil.delete("\"")
  33. + pynini.closure(DAMO_DIGIT, 1)
  34. + pynutil.delete("\"")
  35. )
  36. #seconds
  37. second = (
  38. pynutil.delete("seconds:")
  39. + delete_space
  40. + pynutil.delete("\"")
  41. + pynini.closure(DAMO_DIGIT, 1)
  42. + pynutil.delete("\"")
  43. )
  44. suffix = (
  45. delete_space
  46. + insert_space
  47. + pynutil.delete("suffix:")
  48. + delete_space
  49. + pynutil.delete("\"")
  50. + pynini.closure(DAMO_CHAR - " ", 1)
  51. + pynutil.delete("\"")
  52. )
  53. optional_suffix = pynini.closure(suffix, 0, 1)
  54. zone = (
  55. delete_space
  56. + insert_space
  57. + pynutil.delete("zone:")
  58. + delete_space
  59. + pynutil.delete("\"")
  60. + pynini.closure(DAMO_CHAR - " ", 1)
  61. + pynutil.delete("\"")
  62. )
  63. optional_zone = pynini.closure(zone, 0, 1)
  64. #hms
  65. graph_hms = (hour @ add_leading_zero_to_double_digit) + delete_space + pynutil.insert(":") + (minute @ add_leading_zero_to_double_digit) + delete_space + pynutil.insert(":") + second
  66. #hm
  67. graph_hm = (hour @ add_leading_zero_to_double_digit) + delete_space + pynutil.insert(":") + (minute @ add_leading_zero_to_double_digit)
  68. graph = ( (graph_hms | graph_hm) + optional_suffix + optional_zone )
  69. delete_tokens = self.delete_tokens(graph)
  70. self.fst = delete_tokens.optimize()