FunASR/fun_text_processing/text_normalization/zh/verbalizers/time.py

89 lines
2.8 KiB
Python
Raw Normal View History

2024-05-18 15:50:56 +08:00
import pynini
from fun_text_processing.text_normalization.zh.graph_utils import (
FUN_NOT_QUOTE,
GraphFst,
delete_space,
)
from fun_text_processing.text_normalization.zh.utils import UNIT_1e01, get_abs_path
from pynini.lib import pynutil
class Time(GraphFst):
"""
tokens { time { h: "1" m: "02" s: "36" } } -> 一点零二分三十六秒
tokens { time { suffix "am" hours: "1" minutes: "02" seconds: "36" } } -> 上午一点零二分三十六秒
"""
def __init__(self, deterministic: bool = True, lm: bool = False):
super().__init__(name="time", kind="verbalize", deterministic=deterministic)
graph_digit = pynini.string_file(get_abs_path("data/number/digit.tsv"))
graph_teen = pynini.string_file(get_abs_path("data/number/digit_teen.tsv"))
graph_zero = pynini.string_file(get_abs_path("data/number/zero.tsv"))
graph_no_zero = pynini.cross("0", "")
graph_digit_no_zero = graph_digit | graph_no_zero
graph_2_digit_zero_none = pynini.cross("0", "") + pynini.cross("0", "")
graph_2_digit_zero = pynini.cross("00", "")
graph_2_digit_time = (graph_teen + pynutil.insert(UNIT_1e01) + graph_digit_no_zero) | (
graph_zero + graph_digit
)
h = graph_2_digit_time | graph_2_digit_zero | graph_digit
m = graph_2_digit_time | graph_2_digit_zero
s = graph_2_digit_time | graph_2_digit_zero
# 6:25
h_m = (
pynutil.delete('hours: "')
+ h
+ pynutil.insert("")
+ pynutil.delete('"')
+ delete_space
+ pynutil.delete('minutes: "')
+ (graph_2_digit_time)
+ pynutil.insert("")
+ pynutil.delete('"')
)
# 23:00
h_00 = (
pynutil.delete('hours: "')
+ h
+ pynutil.insert("")
+ pynutil.delete('"')
+ delete_space
+ pynutil.delete('minutes: "')
+ (graph_2_digit_zero_none)
+ pynutil.delete('"')
)
# 9:12:52
h_m_s = (
pynutil.delete('hours: "')
+ h
+ pynutil.insert("")
+ pynutil.delete('"')
+ delete_space
+ pynutil.delete('minutes: "')
+ m
+ pynutil.insert("")
+ pynutil.delete('"')
+ delete_space
+ pynutil.delete('seconds: "')
+ s
+ pynutil.insert("")
+ pynutil.delete('"')
)
graph = h_m | h_m_s | h_00
graph_suffix = (
pynutil.delete('suffix: "')
+ pynini.closure(FUN_NOT_QUOTE)
+ pynutil.delete('"')
+ delete_space
+ graph
)
graph |= graph_suffix
self.fst = self.delete_tokens(graph).optimize()