date.py 2.6 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768
  1. import pynini
  2. from fun_text_processing.text_normalization.en.graph_utils import (
  3. DAMO_NOT_QUOTE,
  4. DAMO_SIGMA,
  5. DAMO_SPACE,
  6. GraphFst,
  7. delete_preserve_order,
  8. )
  9. from fun_text_processing.text_normalization.es.graph_utils import strip_cardinal_apocope
  10. from fun_text_processing.text_normalization.es.taggers.date import articles
  11. from pynini.lib import pynutil
  12. class DateFst(GraphFst):
  13. """
  14. Finite state transducer for verbalizing date, e.g.
  15. date { day: "treinta y uno" month: "marzo" year: "dos mil" } -> "treinta y uno de marzo de dos mil"
  16. date { day: "uno" month: "mayo" year: "del mil novecientos noventa" } -> "primero de mayo del mil novecientos noventa"
  17. Args:
  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, deterministic: bool = True):
  22. super().__init__(name="date", kind="verbalize", deterministic=deterministic)
  23. day_cardinal = pynutil.delete("day: \"") + pynini.closure(DAMO_NOT_QUOTE, 1) + pynutil.delete("\"")
  24. day = strip_cardinal_apocope(day_cardinal)
  25. primero = pynini.cdrewrite(pynini.cross("uno", "primero"), "[BOS]", "[EOS]", DAMO_SIGMA)
  26. day = (
  27. (day @ primero) if deterministic else pynini.union(day, day @ primero)
  28. ) # Primero for first day is traditional, but will vary depending on region
  29. month = pynutil.delete("month: \"") + pynini.closure(DAMO_NOT_QUOTE, 1) + pynutil.delete("\"")
  30. year = (
  31. pynutil.delete("year: \"")
  32. + articles
  33. + DAMO_SPACE
  34. + pynini.closure(DAMO_NOT_QUOTE, 1)
  35. + pynutil.delete("\"")
  36. )
  37. # Insert preposition if wasn't originally with the year. This would mean a space was present
  38. year = pynutil.add_weight(year, -0.001)
  39. year |= (
  40. pynutil.delete("year: \"")
  41. + pynutil.insert("de ")
  42. + pynini.closure(DAMO_NOT_QUOTE, 1)
  43. + pynutil.delete("\"")
  44. )
  45. # day month year
  46. graph_dmy = day + pynini.cross(DAMO_SPACE, " de ") + month + pynini.closure(pynini.accep(" ") + year, 0, 1)
  47. graph_mdy = month + DAMO_SPACE + day + pynini.closure(DAMO_SPACE + year, 0, 1)
  48. if deterministic:
  49. graph_mdy += pynutil.delete(" preserve_order: true") # Only accepts this if was explicitly passed
  50. self.graph = graph_dmy | graph_mdy
  51. final_graph = self.graph + delete_preserve_order
  52. delete_tokens = self.delete_tokens(final_graph)
  53. self.fst = delete_tokens.optimize()