date.py 886 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 DateFst(GraphFst):
  6. """
  7. Finite state transducer for verbalizing date, e.g.
  8. tokens { date { day: "первое мая" } } -> "первое мая"
  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="date", kind="verbalize", deterministic=deterministic)
  15. graph = pynutil.delete("day: \"") + pynini.closure(RU_ALPHA | " ", 1) + pynutil.delete("\"")
  16. delete_tokens = self.delete_tokens(graph)
  17. self.fst = delete_tokens.optimize()