time.py 2.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778
  1. import pynini
  2. from fun_text_processing.inverse_text_normalization.zh.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. pynutil.delete("hour:")
  22. + delete_space
  23. + pynutil.delete("\"")
  24. + pynini.closure(DAMO_DIGIT, 1)
  25. + pynutil.delete("\"")
  26. )
  27. minute = (
  28. pynutil.delete("minute:")
  29. + delete_space
  30. + pynutil.delete("\"")
  31. + pynini.closure(DAMO_DIGIT, 1)
  32. + pynutil.delete("\"")
  33. )
  34. second = (
  35. pynutil.delete("second:")
  36. + delete_space
  37. + pynutil.delete("\"")
  38. + pynini.closure(DAMO_DIGIT, 1)
  39. + pynutil.delete("\"")
  40. )
  41. suffix = (
  42. delete_space
  43. + pynutil.insert(" ")
  44. + pynutil.delete("noon:")
  45. + delete_space
  46. + pynutil.delete("\"")
  47. + pynini.closure(DAMO_CHAR - " ", 1)
  48. + pynutil.delete("\"")
  49. )
  50. optional_suffix = pynini.closure(suffix, 0, 1)
  51. zone = (
  52. delete_space
  53. + pynutil.insert(" ")
  54. + pynutil.delete("zone:")
  55. + delete_space
  56. + pynutil.delete("\"")
  57. + pynini.closure(DAMO_CHAR - " ", 1)
  58. + pynutil.delete("\"")
  59. )
  60. optional_zone = pynini.closure(zone, 0, 1)
  61. graph = (
  62. hour @ add_leading_zero_to_double_digit
  63. + delete_space
  64. + pynutil.insert(":")
  65. + (minute @ add_leading_zero_to_double_digit)
  66. + (pynutil.insert(":") + second @ add_leading_zero_to_double_digit).ques
  67. + optional_suffix
  68. + optional_zone
  69. )
  70. delete_tokens = self.delete_tokens(graph)
  71. self.fst = delete_tokens.optimize()