提交

kongdeqiang

2026-03-13 28ccfbfc51068a663a80764e14074df5edf2b5ba

 fun_text_processing/text_normalization/es/verbalizers/measure.py

@@ -1,5 +1,3 @@


import pynini
from fun_text_processing.text_normalization.en.graph_utils import (
    DAMO_NOT_QUOTE,
@@ -38,7 +36,9 @@
            for False multiple transduction are generated (used for audio-based normalization)
    """

    def __init__(self, decimal: GraphFst, cardinal: GraphFst, fraction: GraphFst, deterministic: bool):
    def __init__(
        self, decimal: GraphFst, cardinal: GraphFst, fraction: GraphFst, deterministic: bool
    ):
        super().__init__(name="measure", kind="verbalize", deterministic=deterministic)

        graph_decimal_masc = decimal.delete_tokens(decimal.graph_masc)
@@ -52,38 +52,59 @@
            DAMO_WHITE_SPACE + "por" + pynini.closure(DAMO_NOT_QUOTE, 1), 0, 1
        )
        unit_masc |= "por" + pynini.closure(DAMO_NOT_QUOTE, 1)
        unit_masc = pynutil.delete("units: \"") + (pynini.closure(DAMO_NOT_QUOTE) @ unit_masc) + pynutil.delete("\"")
        unit_masc = (
            pynutil.delete('units: "')
            + (pynini.closure(DAMO_NOT_QUOTE) @ unit_masc)
            + pynutil.delete('"')
        )

        unit_fem = (unit_plural_fem | unit_singular_fem) + pynini.closure(
            DAMO_WHITE_SPACE + "por" + pynini.closure(DAMO_NOT_QUOTE, 1), 0, 1
        )
        unit_fem = pynutil.delete("units: \"") + (pynini.closure(DAMO_NOT_QUOTE) @ unit_fem) + pynutil.delete("\"")
        unit_fem = (
            pynutil.delete('units: "')
            + (pynini.closure(DAMO_NOT_QUOTE) @ unit_fem)
            + pynutil.delete('"')
        )

        graph_masc = (graph_cardinal_masc | graph_decimal_masc) + DAMO_WHITE_SPACE + unit_masc
        graph_masc |= graph_fraction_masc + DAMO_WHITE_SPACE + pynutil.insert("de ") + unit_masc
        graph_masc |= pynutil.add_weight(
            graph_fraction_masc @ (DAMO_SIGMA + pynini.union("medio", "medios")) + DAMO_WHITE_SPACE + unit_masc, -0.001
            graph_fraction_masc @ (DAMO_SIGMA + pynini.union("medio", "medios"))
            + DAMO_WHITE_SPACE
            + unit_masc,
            -0.001,
        )  # "medio litro" not "medio de litro"

        graph_fem = (graph_cardinal_fem | graph_decimal_fem) + DAMO_WHITE_SPACE + unit_fem
        graph_fem |= graph_fraction_fem + DAMO_WHITE_SPACE + pynutil.insert("de ") + unit_fem
        graph_fem |= pynutil.add_weight(
            graph_fraction_fem @ (DAMO_SIGMA + pynini.union("media", "medias")) + DAMO_WHITE_SPACE + unit_fem, -0.001
            graph_fraction_fem @ (DAMO_SIGMA + pynini.union("media", "medias"))
            + DAMO_WHITE_SPACE
            + unit_fem,
            -0.001,
        )

        graph = graph_masc | graph_fem

        graph = (
            pynini.cdrewrite(
                pynutil.insert(" de"), "quantity: \"" + pynini.closure(DAMO_NOT_QUOTE, 1), "\"", DAMO_SIGMA
                pynutil.insert(" de"),
                'quantity: "' + pynini.closure(DAMO_NOT_QUOTE, 1),
                '"',
                DAMO_SIGMA,
            )
            @ graph
        )  # billones de xyz

        graph @= pynini.cdrewrite(pynini.cross(ones, "uno"), "", DAMO_WHITE_SPACE + "por", DAMO_SIGMA)
        graph @= pynini.cdrewrite(
            pynini.cross(ones, "uno"), "", DAMO_WHITE_SPACE + "por", DAMO_SIGMA
        )

        # To manage alphanumeric combonations ("a-8, 5x"), we let them use a weighted default path.
        alpha_num_unit = pynutil.delete("units: \"") + pynini.closure(DAMO_NOT_QUOTE) + pynutil.delete("\"")
        alpha_num_unit = (
            pynutil.delete('units: "') + pynini.closure(DAMO_NOT_QUOTE) + pynutil.delete('"')
        )
        graph_alpha_num = pynini.union(
            (graph_cardinal_masc | graph_decimal_masc) + DAMO_SPACE + alpha_num_unit,
            alpha_num_unit + delete_extra_space + (graph_cardinal_masc | graph_decimal_masc),