jpreprocess_njd/
lib.rs

1mod contrib;
2mod node;
3mod open_jtalk;
4
5use jpreprocess_core::{token::Token, word_entry::WordEntry, JPreprocessResult};
6use jpreprocess_window::{IterQuintMut, IterQuintMutTrait};
7
8pub use contrib::*;
9pub use node::*;
10pub use open_jtalk::*;
11
12#[derive(Clone, Debug, PartialEq)]
13pub struct NJD {
14    pub nodes: Vec<NJDNode>,
15}
16
17impl NJD {
18    pub fn remove_silent_node(&mut self) {
19        self.nodes.retain(|node| !node.get_pron().is_empty())
20    }
21
22    pub fn from_tokens<'a, T: Token>(
23        tokens: impl 'a + IntoIterator<Item = T>,
24    ) -> JPreprocessResult<Self> {
25        let mut nodes = Vec::new();
26        for mut token in tokens {
27            let (string, entry) = token.fetch()?;
28            nodes.extend(NJDNode::load(string, &entry));
29        }
30
31        Ok(Self { nodes })
32    }
33    pub fn from_strings(njd_features: Vec<String>) -> Self {
34        Self {
35            nodes: njd_features
36                .iter()
37                .flat_map(|feature| NJDNode::load_csv(feature))
38                .collect(),
39        }
40    }
41
42    pub fn preprocess(&mut self) {
43        use open_jtalk::*;
44
45        pronunciation::njd_set_pronunciation(self);
46        digit_sequence::njd_digit_sequence(self);
47        digit::njd_set_digit(self);
48        accent_phrase::njd_set_accent_phrase(self);
49        accent_type::njd_set_accent_type(self);
50        unvoiced_vowel::njd_set_unvoiced_vowel(self);
51        // long vowel estimator is deprecated
52        // long_vowel::njd_set_long_vowel(self);
53    }
54}
55
56impl<'a> FromIterator<(&'a str, &'a WordEntry)> for NJD {
57    fn from_iter<I: IntoIterator<Item = (&'a str, &'a WordEntry)>>(iter: I) -> Self {
58        let nodes = iter
59            .into_iter()
60            .flat_map(|(text, word_entry)| NJDNode::load(text, word_entry))
61            .collect();
62        Self { nodes }
63    }
64}
65impl<'a> FromIterator<&'a str> for NJD {
66    fn from_iter<I: IntoIterator<Item = &'a str>>(iter: I) -> Self {
67        let nodes = iter.into_iter().flat_map(NJDNode::load_csv).collect();
68        Self { nodes }
69    }
70}
71
72impl IterQuintMutTrait for NJD {
73    type Item = NJDNode;
74    fn iter_quint_mut(&mut self) -> IterQuintMut<'_, Self::Item> {
75        IterQuintMut::new(&mut self.nodes)
76    }
77    fn iter_quint_mut_range(&mut self, start: usize, end: usize) -> IterQuintMut<'_, Self::Item> {
78        IterQuintMut::new(&mut self.nodes[start..end])
79    }
80}
81
82impl From<NJD> for Vec<String> {
83    fn from(njd: NJD) -> Self {
84        njd.nodes.into_iter().map(|node| node.to_string()).collect()
85    }
86}
jpreprocess_njd/lib.rs

jpreprocess_njd/
lib.rs
