time.py 2.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475
  1. import pynini
  2. from fun_text_processing.inverse_text_normalization.vi.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: "3" } -> 3h
  14. time { hours: "12" minutes: "30" } -> 12:30
  15. time { hours: "1" minutes: "12" second: "22"} -> 1:12:22
  16. time { minutes: "36" second: "45"} -> 36p45s
  17. time { hours: "2" zone: "gmt" } -> 2h gmt
  18. """
  19. def __init__(self):
  20. super().__init__(name="time", kind="verbalize")
  21. add_leading_zero_to_double_digit = (DAMO_DIGIT + DAMO_DIGIT) | (pynutil.insert("0") + DAMO_DIGIT)
  22. hour = (
  23. pynutil.delete("hours:")
  24. + delete_space
  25. + pynutil.delete('"')
  26. + pynini.closure(DAMO_DIGIT, 1)
  27. + pynutil.delete('"')
  28. )
  29. minute = (
  30. pynutil.delete("minutes:")
  31. + delete_space
  32. + pynutil.delete('"')
  33. + pynini.closure(DAMO_DIGIT, 1)
  34. + pynutil.delete('"')
  35. )
  36. second = (
  37. pynutil.delete("seconds:")
  38. + delete_space
  39. + pynutil.delete('"')
  40. + pynini.closure(DAMO_DIGIT, 1)
  41. + pynutil.delete('"')
  42. )
  43. zone = (
  44. delete_space
  45. + insert_space
  46. + pynutil.delete("zone:")
  47. + delete_space
  48. + pynutil.delete('"')
  49. + pynini.closure(DAMO_CHAR - " ", 1)
  50. + pynutil.delete('"')
  51. )
  52. optional_zone = pynini.closure(zone, 0, 1)
  53. optional_second = pynini.closure(
  54. delete_space + pynutil.insert(":") + (second @ add_leading_zero_to_double_digit), 0, 1,
  55. )
  56. graph_h = hour + pynutil.insert("h")
  57. graph_hms = (
  58. hour + delete_space + pynutil.insert(":") + (minute @ add_leading_zero_to_double_digit) + optional_second
  59. )
  60. graph_ms = (
  61. minute
  62. + delete_space
  63. + pynutil.insert("p")
  64. + (second @ add_leading_zero_to_double_digit)
  65. + pynutil.insert("s")
  66. )
  67. graph = (graph_h | graph_ms | graph_hms) + optional_zone
  68. delete_tokens = self.delete_tokens(graph)
  69. self.fst = delete_tokens.optimize()