money.py 6.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155
  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 shift_cardinal_gender, strip_cardinal_apocope
  10. from fun_text_processing.text_normalization.es.utils import get_abs_path
  11. from pynini.lib import pynutil
  12. fem = pynini.string_file((get_abs_path("data/money/currency_plural_fem.tsv")))
  13. masc = pynini.string_file((get_abs_path("data/money/currency_plural_masc.tsv")))
  14. fem_singular = pynini.project(fem, "input")
  15. masc_singular = pynini.project(masc, "input")
  16. fem_plural = pynini.project(fem, "output")
  17. masc_plural = pynini.project(masc, "output")
  18. class MoneyFst(GraphFst):
  19. """
  20. Finite state transducer for verbalizing money, e.g.
  21. money { currency_maj: "euro" integer_part: "un"} -> "un euro"
  22. money { currency_maj: "euro" integer_part: "un" fractional_part: "cero cero un"} -> "uno coma cero cero uno euros"
  23. money { integer_part: "un" currency_maj: "libra" fractional_part: "cuarenta" preserve_order: true} -> "una libra cuarenta"
  24. money { integer_part: "un" currency_maj: "libra" fractional_part: "cuarenta" currency_min: "peniques" preserve_order: true} -> "una libra con cuarenta peniques"
  25. money { fractional_part: "un" currency_min: "penique" preserve_order: true} -> "un penique"
  26. Args:
  27. decimal: GraphFst
  28. deterministic: if True will provide a single transduction option,
  29. for False multiple transduction are generated (used for audio-based normalization)
  30. """
  31. def __init__(self, decimal: GraphFst, deterministic: bool = True):
  32. super().__init__(name="money", kind="verbalize", deterministic=deterministic)
  33. maj_singular_masc = (
  34. pynutil.delete("currency_maj: \"")
  35. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ masc_singular)
  36. + pynutil.delete("\"")
  37. )
  38. maj_singular_fem = (
  39. pynutil.delete("currency_maj: \"")
  40. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ fem_singular)
  41. + pynutil.delete("\"")
  42. )
  43. maj_plural_masc = (
  44. pynutil.delete("currency_maj: \"")
  45. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ masc_plural)
  46. + pynutil.delete("\"")
  47. )
  48. maj_plural_fem = (
  49. pynutil.delete("currency_maj: \"")
  50. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ fem_plural)
  51. + pynutil.delete("\"")
  52. )
  53. maj_masc = maj_plural_masc | maj_singular_masc # Tagger kept quantity resolution stable
  54. maj_fem = maj_plural_fem | maj_singular_fem
  55. min_singular_masc = (
  56. pynutil.delete("currency_min: \"")
  57. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ masc_singular)
  58. + pynutil.delete("\"")
  59. )
  60. min_singular_fem = (
  61. pynutil.delete("currency_min: \"")
  62. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ fem_singular)
  63. + pynutil.delete("\"")
  64. )
  65. min_plural_masc = (
  66. pynutil.delete("currency_min: \"")
  67. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ masc_plural)
  68. + pynutil.delete("\"")
  69. )
  70. min_plural_fem = (
  71. pynutil.delete("currency_min: \"")
  72. + (pynini.closure(DAMO_NOT_QUOTE, 1) @ fem_plural)
  73. + pynutil.delete("\"")
  74. )
  75. min_masc = min_plural_masc | min_singular_masc
  76. min_fem = min_plural_fem | min_singular_fem
  77. fractional_part = (
  78. pynutil.delete("fractional_part: \"") + pynini.closure(DAMO_NOT_QUOTE, 1) + pynutil.delete("\"")
  79. )
  80. integer_part = pynutil.delete("integer_part: \"") + pynini.closure(DAMO_NOT_QUOTE, 1) + pynutil.delete("\"")
  81. optional_add_and = pynini.closure(pynutil.insert(pynini.union("con ", "y ")), 0, 1)
  82. # *** currency_maj
  83. graph_integer_masc = integer_part + DAMO_SPACE + maj_masc
  84. graph_integer_fem = shift_cardinal_gender(integer_part) + DAMO_SPACE + maj_fem
  85. graph_integer = graph_integer_fem | graph_integer_masc
  86. # *** currency_maj + (***) | ((con) *** current_min)
  87. graph_integer_with_minor_masc = (
  88. graph_integer_masc
  89. + DAMO_SPACE
  90. + pynini.union(
  91. optional_add_and + strip_cardinal_apocope(fractional_part),
  92. (optional_add_and + fractional_part + DAMO_SPACE + min_masc),
  93. (optional_add_and + shift_cardinal_gender(fractional_part) + DAMO_SPACE + min_fem),
  94. ) # Could be minor currency that is different gender
  95. + delete_preserve_order
  96. )
  97. graph_integer_with_minor_fem = (
  98. graph_integer_fem
  99. + DAMO_SPACE
  100. + pynini.union(
  101. optional_add_and + shift_cardinal_gender(fractional_part),
  102. (optional_add_and + fractional_part + DAMO_SPACE + min_masc),
  103. (optional_add_and + shift_cardinal_gender(fractional_part) + DAMO_SPACE + min_fem),
  104. ) # Could be minor currency that is different gender
  105. + delete_preserve_order
  106. )
  107. graph_integer_with_minor = graph_integer_with_minor_fem | graph_integer_with_minor_masc
  108. ## *** coma *** currency_maj
  109. graph_decimal_masc = decimal.graph_masc + DAMO_SPACE + maj_masc
  110. graph_decimal_fem = decimal.graph_fem
  111. graph_decimal_fem |= decimal.numbers_only_quantity # can still have "x billions" with fem currency
  112. graph_decimal_fem += DAMO_SPACE + maj_fem
  113. graph_decimal = graph_decimal_fem | graph_decimal_masc
  114. graph_decimal = (
  115. pynini.cdrewrite(
  116. pynutil.insert(" de"), "quantity: \"" + pynini.closure(DAMO_NOT_QUOTE, 1), "\"", DAMO_SIGMA
  117. )
  118. @ graph_decimal
  119. ) # formally it's millones/billones de ***
  120. # *** current_min
  121. graph_minor_masc = fractional_part + DAMO_SPACE + min_masc + delete_preserve_order
  122. graph_minor_fem = shift_cardinal_gender(fractional_part) + DAMO_SPACE + min_fem + delete_preserve_order
  123. graph_minor = graph_minor_fem | graph_minor_masc
  124. graph = graph_integer | graph_integer_with_minor | graph_decimal | graph_minor
  125. delete_tokens = self.delete_tokens(graph)
  126. self.fst = delete_tokens.optimize()