money.py 883 B

123456789101112131415161718192021222324
  1. import pynini
  2. from fun_text_processing.text_normalization.en.graph_utils import GraphFst
  3. from fun_text_processing.text_normalization.ru.alphabet import RU_ALPHA
  4. from pynini.lib import pynutil
  5. class MoneyFst(GraphFst):
  6. """
  7. Finite state transducer for verbalizing money, e.g.
  8. money { "пять рублей" } -> пять рублей
  9. Args:
  10. deterministic: if True will provide a single transduction option,
  11. for False multiple transduction are generated (used for audio-based normalization)
  12. """
  13. def __init__(self, deterministic: bool = True):
  14. super().__init__(name="money", kind="verbalize", deterministic=deterministic)
  15. graph = pynini.closure(RU_ALPHA | " ")
  16. delete_tokens = self.delete_tokens(pynutil.delete("integer_part: \"") + graph + pynutil.delete("\""))
  17. self.fst = delete_tokens.optimize()