lindera_dictionary/builder/
connection_cost_matrix.rs

1use std::borrow::Cow;
2use std::fs::File;
3use std::io::{self, Write};
4use std::path::Path;
5use std::str::FromStr;
6
7use byteorder::{LittleEndian, WriteBytesExt};
8use derive_builder::Builder;
9use log::debug;
10
11use crate::LinderaResult;
12use crate::decompress::Algorithm;
13use crate::error::LinderaErrorKind;
14use crate::util::{compress_write, read_file_with_encoding};
15
16#[derive(Builder, Debug)]
17#[builder(name = ConnectionCostMatrixBuilderOptions)]
18#[builder(build_fn(name = "builder"))]
19pub struct ConnectionCostMatrixBuilder {
20    #[builder(default = "\"UTF-8\".into()", setter(into))]
21    encoding: Cow<'static, str>,
22    #[builder(default = "Algorithm::Deflate")]
23    compress_algorithm: Algorithm,
24}
25
26impl ConnectionCostMatrixBuilder {
27    pub fn build(&self, input_dir: &Path, output_dir: &Path) -> LinderaResult<()> {
28        let matrix_data_path = input_dir.join("matrix.def");
29        debug!("reading {matrix_data_path:?}");
30        let matrix_data = read_file_with_encoding(&matrix_data_path, &self.encoding)?;
31
32        let mut lines = Vec::new();
33        for line in matrix_data.lines() {
34            let fields: Vec<i32> = line
35                .split_whitespace()
36                .map(i32::from_str)
37                .collect::<Result<_, _>>()
38                .map_err(|err| LinderaErrorKind::Parse.with_error(anyhow::anyhow!(err)))?;
39            lines.push(fields);
40        }
41        let mut lines_it = lines.into_iter();
42        let header = lines_it.next().ok_or_else(|| {
43            LinderaErrorKind::Content.with_error(anyhow::anyhow!("unknown error"))
44        })?;
45        let forward_size = header[0] as u32;
46        let backward_size = header[1] as u32;
47        let len = 3 + (forward_size * backward_size) as usize;
48        let mut costs = vec![i16::MAX; len];
49        costs[0] = -1; // Version flag for transposed layout
50        costs[1] = forward_size as i16;
51        costs[2] = backward_size as i16;
52        for fields in lines_it {
53            let forward_id = fields[0] as u32;
54            let backward_id = fields[1] as u32;
55            let cost = fields[2] as u16;
56            costs[3 + (forward_id + backward_id * forward_size) as usize] = cost as i16;
57        }
58
59        let wtr_matrix_mtx_path = output_dir.join(Path::new("matrix.mtx"));
60        let mut wtr_matrix_mtx = io::BufWriter::new(
61            File::create(wtr_matrix_mtx_path)
62                .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?,
63        );
64        let mut matrix_mtx_buffer = Vec::new();
65        for cost in costs {
66            matrix_mtx_buffer
67                .write_i16::<LittleEndian>(cost)
68                .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
69        }
70
71        compress_write(
72            &matrix_mtx_buffer,
73            self.compress_algorithm,
74            &mut wtr_matrix_mtx,
75        )?;
76
77        wtr_matrix_mtx
78            .flush()
79            .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
80
81        Ok(())
82    }
83}
lindera_dictionary/builder/connection_cost_matrix.rs

lindera_dictionary/builder/
connection_cost_matrix.rs
