lindera_dictionary/
util.rs

1use std::fs::File;
2use std::io::{Read, Write};
3use std::ops::Deref;
4use std::path::Path;
5
6#[cfg(feature = "mmap")]
7use memmap2::Mmap;
8
9use anyhow::anyhow;
10use encoding_rs::Encoding;
11use serde::{Deserialize, Serialize};
12
13use crate::LinderaResult;
14#[cfg(feature = "compress")]
15use crate::compress::compress;
16use crate::decompress::Algorithm;
17use crate::error::LinderaErrorKind;
18
19#[cfg(feature = "compress")]
20pub fn compress_write<W: Write>(
21    buffer: &[u8],
22    algorithm: Algorithm,
23    writer: &mut W,
24) -> LinderaResult<()> {
25    let compressed = compress(buffer, algorithm)
26        .map_err(|err| LinderaErrorKind::Compress.with_error(anyhow::anyhow!(err)))?;
27    bincode::serde::encode_into_std_write(&compressed, writer, bincode::config::legacy())
28        .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
29
30    Ok(())
31}
32
33#[cfg(not(feature = "compress"))]
34pub fn compress_write<W: Write>(
35    buffer: &[u8],
36    _algorithm: Algorithm,
37    writer: &mut W,
38) -> LinderaResult<()> {
39    writer
40        .write_all(buffer)
41        .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
42
43    Ok(())
44}
45
46pub fn read_file(filename: &Path) -> LinderaResult<Vec<u8>> {
47    let mut input_read = File::open(filename)
48        .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
49    let mut buffer = Vec::new();
50    input_read
51        .read_to_end(&mut buffer)
52        .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
53    Ok(buffer)
54}
55
56#[cfg(feature = "mmap")]
57pub fn mmap_file(filename: &Path) -> LinderaResult<Mmap> {
58    let file = File::open(filename)
59        .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
60    let mmap = unsafe { Mmap::map(&file) }
61        .map_err(|err| LinderaErrorKind::Io.with_error(anyhow::anyhow!(err)))?;
62    Ok(mmap)
63}
64
65pub fn read_file_with_encoding(filepath: &Path, encoding_name: &str) -> LinderaResult<String> {
66    let encoding = Encoding::for_label_no_replacement(encoding_name.as_bytes());
67    let encoding = encoding.ok_or_else(|| {
68        LinderaErrorKind::Decode.with_error(anyhow!("Invalid encoding: {}", encoding_name))
69    })?;
70
71    let buffer = read_file(filepath)?;
72    Ok(encoding.decode(&buffer).0.into_owned())
73}
74
75pub enum Data {
76    Static(&'static [u8]),
77    Vec(Vec<u8>),
78    #[cfg(feature = "mmap")]
79    Map(Mmap),
80}
81
82impl Deref for Data {
83    type Target = [u8];
84    fn deref(&self) -> &Self::Target {
85        match self {
86            Data::Static(s) => s,
87            Data::Vec(v) => v,
88            #[cfg(feature = "mmap")]
89            Data::Map(m) => m,
90        }
91    }
92}
93
94impl From<&'static [u8]> for Data {
95    fn from(s: &'static [u8]) -> Self {
96        Self::Static(s)
97    }
98}
99
100impl<T: Deref<Target = [u8]>> From<&'static T> for Data {
101    fn from(t: &'static T) -> Self {
102        Self::Static(t)
103    }
104}
105
106impl From<Vec<u8>> for Data {
107    fn from(v: Vec<u8>) -> Self {
108        Self::Vec(v)
109    }
110}
111
112#[cfg(feature = "mmap")]
113impl From<Mmap> for Data {
114    fn from(m: Mmap) -> Self {
115        Self::Map(m)
116    }
117}
118
119impl Clone for Data {
120    fn clone(&self) -> Self {
121        match self {
122            Data::Static(s) => Data::Static(s),
123            Data::Vec(v) => Data::Vec(v.clone()),
124            #[cfg(feature = "mmap")]
125            Data::Map(m) => Data::Vec(m.to_vec()),
126        }
127    }
128}
129
130impl Serialize for Data {
131    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
132    where
133        S: serde::Serializer,
134    {
135        match self {
136            Self::Static(s) => serializer.serialize_bytes(s),
137            Self::Vec(v) => serializer.serialize_bytes(v),
138            #[cfg(feature = "mmap")]
139            Self::Map(m) => serializer.serialize_bytes(m),
140        }
141    }
142}
143
144impl<'de> Deserialize<'de> for Data {
145    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
146    where
147        D: serde::Deserializer<'de>,
148    {
149        Vec::<u8>::deserialize(deserializer).map(Self::Vec)
150    }
151}
lindera_dictionary/util.rs

lindera_dictionary/
util.rs
