| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566 |
- import pynini
- from fun_text_processing.text_normalization.en.graph_utils import (
- DAMO_NOT_QUOTE,
- DAMO_SIGMA,
- GraphFst,
- delete_preserve_order,
- insert_space,
- )
- from fun_text_processing.text_normalization.es.utils import get_abs_path
- from pynini.lib import pynutil
- digit_no_zero = pynini.invert(pynini.string_file(get_abs_path("data/numbers/digit.tsv")))
- zero = pynini.invert(pynini.string_file(get_abs_path("data/numbers/zero.tsv")))
- graph_symbols = pynini.string_file(get_abs_path("data/electronic/symbols.tsv"))
- server_common = pynini.string_file(get_abs_path("data/electronic/server_name.tsv"))
- domain_common = pynini.string_file(get_abs_path("data/electronic/domain.tsv"))
- class ElectronicFst(GraphFst):
- """
- Finite state transducer for verbalizing electronic
- e.g. electronic { username: "abc" domain: "hotmail.com" } -> "a b c arroba hotmail punto com"
- -> "a b c arroba h o t m a i l punto c o m"
- -> "a b c arroba hotmail punto c o m"
- -> "a b c at h o t m a i l punto com"
- 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="electronic", kind="verbalize", deterministic=deterministic)
- graph_digit_no_zero = (
- digit_no_zero @ pynini.cdrewrite(pynini.cross("un", "uno"), "", "", DAMO_SIGMA).optimize()
- )
- graph_digit = graph_digit_no_zero | zero
- def add_space_after_char():
- return pynini.closure(DAMO_NOT_QUOTE - pynini.accep(" ") + insert_space) + (
- DAMO_NOT_QUOTE - pynini.accep(" ")
- )
- verbalize_characters = pynini.cdrewrite(graph_symbols | graph_digit, "", "", DAMO_SIGMA)
- user_name = pynutil.delete("username: \"") + add_space_after_char() + pynutil.delete("\"")
- user_name @= verbalize_characters
- convert_defaults = pynutil.add_weight(DAMO_NOT_QUOTE, weight=0.0001) | domain_common | server_common
- domain = convert_defaults + pynini.closure(insert_space + convert_defaults)
- domain @= verbalize_characters
- domain = pynutil.delete("domain: \"") + domain + pynutil.delete("\"")
- protocol = (
- pynutil.delete("protocol: \"")
- + add_space_after_char() @ pynini.cdrewrite(graph_symbols, "", "", DAMO_SIGMA)
- + pynutil.delete("\"")
- )
- self.graph = (pynini.closure(protocol + pynini.accep(" "), 0, 1) + domain) | (
- user_name + pynini.accep(" ") + pynutil.insert("arroba ") + domain
- )
- delete_tokens = self.delete_tokens(self.graph + delete_preserve_order)
- self.fst = delete_tokens.optimize()
|