cardinal.py 1.7 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849
  1. import pynini
  2. from fun_text_processing.text_normalization.en.graph_utils import DAMO_NOT_QUOTE, GraphFst
  3. from fun_text_processing.text_normalization.es.graph_utils import (
  4. add_cardinal_apocope_fem,
  5. shift_cardinal_gender,
  6. strip_cardinal_apocope,
  7. )
  8. from pynini.lib import pynutil
  9. class CardinalFst(GraphFst):
  10. """
  11. Finite state transducer for verbalizing cardinals
  12. e.g. cardinal { integer: "dos" } -> "dos"
  13. Args:
  14. deterministic: if True will provide a single transduction option,
  15. for False multiple transduction are generated (used for audio-based normalization)
  16. """
  17. def __init__(self, deterministic: bool = True):
  18. super().__init__(name="cardinal", kind="verbalize", deterministic=deterministic)
  19. optional_sign = pynini.closure(pynini.cross("negative: \"true\" ", "menos "), 0, 1)
  20. self.optional_sign = optional_sign
  21. integer = pynini.closure(DAMO_NOT_QUOTE, 1)
  22. self.integer = pynutil.delete(" \"") + integer + pynutil.delete("\"")
  23. integer = pynutil.delete("integer:") + self.integer
  24. graph_masc = optional_sign + integer
  25. graph_fem = shift_cardinal_gender(graph_masc)
  26. self.graph_masc = pynini.optimize(graph_masc)
  27. self.graph_fem = pynini.optimize(graph_fem)
  28. # Adding adjustment for fem gender (choice of gender will be random)
  29. graph = graph_masc | graph_fem
  30. if not deterministic:
  31. # For alternate renderings when apocope is omitted (i.e. cardinal stands alone)
  32. graph |= strip_cardinal_apocope(graph_masc)
  33. # "una" will drop to "un" in unique contexts
  34. graph |= add_cardinal_apocope_fem(graph_fem)
  35. delete_tokens = self.delete_tokens(graph)
  36. self.fst = delete_tokens.optimize()