date.py 2.9 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889
  1. import pynini
  2. from fun_text_processing.text_normalization.en.graph_utils import (
  3. DAMO_NOT_QUOTE,
  4. DAMO_SIGMA,
  5. GraphFst,
  6. delete_extra_space,
  7. delete_space,
  8. )
  9. from pynini.examples import plurals
  10. from pynini.lib import pynutil
  11. class DateFst(GraphFst):
  12. """
  13. Finite state transducer for verbalizing date, e.g.
  14. date { month: "february" day: "five" year: "twenty twelve" preserve_order: true } -> february fifth twenty twelve
  15. date { day: "five" month: "february" year: "twenty twelve" preserve_order: true } -> the fifth of february twenty twelve
  16. Args:
  17. ordinal: OrdinalFst
  18. deterministic: if True will provide a single transduction option,
  19. for False multiple transduction are generated (used for audio-based normalization)
  20. """
  21. def __init__(self, ordinal: GraphFst, deterministic: bool = True, lm: bool = False):
  22. super().__init__(name="date", kind="verbalize", deterministic=deterministic)
  23. month = pynini.closure(DAMO_NOT_QUOTE, 1)
  24. day_cardinal = (
  25. pynutil.delete("day:")
  26. + delete_space
  27. + pynutil.delete("\"")
  28. + pynini.closure(DAMO_NOT_QUOTE, 1)
  29. + pynutil.delete("\"")
  30. )
  31. day = day_cardinal @ ordinal.suffix
  32. month = pynutil.delete("month:") + delete_space + pynutil.delete("\"") + month + pynutil.delete("\"")
  33. year = (
  34. pynutil.delete("year:")
  35. + delete_space
  36. + pynutil.delete("\"")
  37. + pynini.closure(DAMO_NOT_QUOTE, 1)
  38. + delete_space
  39. + pynutil.delete("\"")
  40. )
  41. # month (day) year
  42. graph_mdy = (
  43. month + pynini.closure(delete_extra_space + day, 0, 1) + pynini.closure(delete_extra_space + year, 0, 1)
  44. )
  45. # may 5 -> may five
  46. if not deterministic and not lm:
  47. graph_mdy |= (
  48. month
  49. + pynini.closure(delete_extra_space + day_cardinal, 0, 1)
  50. + pynini.closure(delete_extra_space + year, 0, 1)
  51. )
  52. # day month year
  53. graph_dmy = (
  54. pynutil.insert("the ")
  55. + day
  56. + delete_extra_space
  57. + pynutil.insert("of ")
  58. + month
  59. + pynini.closure(delete_extra_space + year, 0, 1)
  60. )
  61. optional_preserve_order = pynini.closure(
  62. pynutil.delete("preserve_order:") + delete_space + pynutil.delete("true") + delete_space
  63. | pynutil.delete("field_order:")
  64. + delete_space
  65. + pynutil.delete("\"")
  66. + DAMO_NOT_QUOTE
  67. + pynutil.delete("\"")
  68. + delete_space
  69. )
  70. final_graph = (
  71. (plurals._priority_union(graph_mdy, pynutil.add_weight(graph_dmy, 0.0001), DAMO_SIGMA) | year)
  72. + delete_space
  73. + optional_preserve_order
  74. )
  75. delete_tokens = self.delete_tokens(final_graph)
  76. self.fst = delete_tokens.optimize()