kongdeqiang
7 天以前 28ccfbfc51068a663a80764e14074df5edf2b5ba
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
import pynini
from fun_text_processing.text_normalization.en.graph_utils import (
    DAMO_ALPHA,
    DAMO_DIGIT,
    DAMO_SIGMA,
    GraphFst,
    delete_extra_space,
    delete_space,
    insert_space,
    plurals,
)
from fun_text_processing.text_normalization.en.utils import get_abs_path
from pynini.lib import pynutil
 
 
class TelephoneFst(GraphFst):
    """
    Finite state transducer for classifying telephone, and IP, and SSN which includes country code, number part and extension
    country code optional: +***
    number part: ***-***-****, or (***) ***-****
    extension optional: 1-9999
    E.g
    +1 123-123-5678-1 -> telephone { country_code: "one" number_part: "one two three, one two three, five six seven eight" extension: "one" }
    1-800-GO-U-HAUL -> telephone { country_code: "one" number_part: "one, eight hundred GO U HAUL" }
    Args:
        deterministic: if True will provide a single transduction option,
            for False multiple transduction are generated (used for audio-based normalization)
    """
 
    def __init__(self, deterministic: bool = True):
        super().__init__(name="telephone", kind="classify", deterministic=deterministic)
 
        add_separator = pynutil.insert(", ")  # between components
        zero = pynini.cross("0", "zero")
        if not deterministic:
            zero |= pynini.cross("0", pynini.union("o", "oh"))
        digit = (
            pynini.invert(pynini.string_file(get_abs_path("data/number/digit.tsv"))).optimize()
            | zero
        )
 
        telephone_prompts = pynini.string_file(get_abs_path("data/telephone/telephone_prompt.tsv"))
        country_code = (
            pynini.closure(telephone_prompts + delete_extra_space, 0, 1)
            + pynini.closure(pynini.cross("+", "plus "), 0, 1)
            + pynini.closure(digit + insert_space, 0, 2)
            + digit
            + pynutil.insert(",")
        )
        country_code |= telephone_prompts
        country_code = pynutil.insert('country_code: "') + country_code + pynutil.insert('"')
        country_code = (
            country_code + pynini.closure(pynutil.delete("-"), 0, 1) + delete_space + insert_space
        )
 
        area_part_default = pynini.closure(digit + insert_space, 2, 2) + digit
        area_part = pynini.cross("800", "eight hundred") | pynini.compose(
            pynini.difference(DAMO_SIGMA, "800"), area_part_default
        )
 
        area_part = (
            (area_part + (pynutil.delete("-") | pynutil.delete(".")))
            | (
                pynutil.delete("(")
                + area_part
                + (
                    (pynutil.delete(")") + pynini.closure(pynutil.delete(" "), 0, 1))
                    | pynutil.delete(")-")
                )
            )
        ) + add_separator
 
        del_separator = pynini.closure(pynini.union("-", " ", "."), 0, 1)
        number_length = ((DAMO_DIGIT + del_separator) | (DAMO_ALPHA + del_separator)) ** 7
        number_words = pynini.closure(
            (DAMO_DIGIT @ digit) + (insert_space | (pynini.cross("-", ", ")))
            | DAMO_ALPHA
            | (DAMO_ALPHA + pynini.cross("-", " "))
        )
        number_words |= pynini.closure(
            (DAMO_DIGIT @ digit) + (insert_space | (pynini.cross(".", ", ")))
            | DAMO_ALPHA
            | (DAMO_ALPHA + pynini.cross(".", " "))
        )
        number_words = pynini.compose(number_length, number_words)
        number_part = area_part + number_words
        number_part = pynutil.insert('number_part: "') + number_part + pynutil.insert('"')
        extension = (
            pynutil.insert('extension: "')
            + pynini.closure(digit + insert_space, 0, 3)
            + digit
            + pynutil.insert('"')
        )
        extension = pynini.closure(insert_space + extension, 0, 1)
 
        graph = plurals._priority_union(
            country_code + number_part, number_part, DAMO_SIGMA
        ).optimize()
        graph = plurals._priority_union(
            country_code + number_part + extension, graph, DAMO_SIGMA
        ).optimize()
        graph = plurals._priority_union(number_part + extension, graph, DAMO_SIGMA).optimize()
 
        # ip
        ip_prompts = pynini.string_file(get_abs_path("data/telephone/ip_prompt.tsv"))
        digit_to_str_graph = digit + pynini.closure(pynutil.insert(" ") + digit, 0, 2)
        ip_graph = digit_to_str_graph + (pynini.cross(".", " dot ") + digit_to_str_graph) ** 3
        graph |= (
            pynini.closure(
                pynutil.insert('country_code: "')
                + ip_prompts
                + pynutil.insert('"')
                + delete_extra_space,
                0,
                1,
            )
            + pynutil.insert('number_part: "')
            + ip_graph.optimize()
            + pynutil.insert('"')
        )
        # ssn
        ssn_prompts = pynini.string_file(get_abs_path("data/telephone/ssn_prompt.tsv"))
        three_digit_part = digit + (pynutil.insert(" ") + digit) ** 2
        two_digit_part = digit + pynutil.insert(" ") + digit
        four_digit_part = digit + (pynutil.insert(" ") + digit) ** 3
        ssn_separator = pynini.cross("-", ", ")
        ssn_graph = (
            three_digit_part + ssn_separator + two_digit_part + ssn_separator + four_digit_part
        )
 
        graph |= (
            pynini.closure(
                pynutil.insert('country_code: "')
                + ssn_prompts
                + pynutil.insert('"')
                + delete_extra_space,
                0,
                1,
            )
            + pynutil.insert('number_part: "')
            + ssn_graph.optimize()
            + pynutil.insert('"')
        )
 
        final_graph = self.add_tokens(graph)
        self.fst = final_graph.optimize()