lindera_dictionary/
util.rs1use 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;
14use crate::error::LinderaErrorKind;
15
16use rkyv::{Archive, Deserialize as RkyvDeserialize, Serialize as RkyvSerialize};
17
18pub fn write_data<W: Write>(buffer: &[u8], writer: &mut W) -> LinderaResult<()> {
20 writer.write_all(buffer).map_err(|err| {
21 LinderaErrorKind::Io
22 .with_error(err)
23 .add_context("Failed to write data to output")
24 })?;
25 Ok(())
26}
27
28pub fn read_file(filename: &Path) -> LinderaResult<Vec<u8>> {
29 let mut input_read = File::open(filename).map_err(|err| {
30 LinderaErrorKind::Io
31 .with_error(err)
32 .add_context(format!("Failed to open file: {}", filename.display()))
33 })?;
34 let mut buffer = Vec::new();
35 input_read.read_to_end(&mut buffer).map_err(|err| {
36 LinderaErrorKind::Io.with_error(err).add_context(format!(
37 "Failed to read file contents: {}",
38 filename.display()
39 ))
40 })?;
41 Ok(buffer)
42}
43
44pub fn read_aligned_file(filename: &Path) -> LinderaResult<rkyv::util::AlignedVec<16>> {
47 let raw_data = read_file(filename)?;
48
49 let mut aligned_data = rkyv::util::AlignedVec::<16>::new();
50 aligned_data.extend_from_slice(&raw_data);
51
52 Ok(aligned_data)
53}
54
55#[cfg(feature = "mmap")]
56pub fn mmap_file(filename: &Path) -> LinderaResult<Mmap> {
57 let file = File::open(filename).map_err(|err| {
58 LinderaErrorKind::Io.with_error(err).add_context(format!(
59 "Failed to open file for memory mapping: {}",
60 filename.display()
61 ))
62 })?;
63 let mmap = unsafe { Mmap::map(&file) }.map_err(|err| {
64 LinderaErrorKind::Io
65 .with_error(err)
66 .add_context(format!("Failed to memory map file: {}", filename.display()))
67 })?;
68 Ok(mmap)
69}
70
71pub fn read_file_with_encoding(filepath: &Path, encoding_name: &str) -> LinderaResult<String> {
72 let encoding = Encoding::for_label_no_replacement(encoding_name.as_bytes());
73 let encoding = encoding.ok_or_else(|| {
74 LinderaErrorKind::Decode.with_error(anyhow!("Invalid encoding: {encoding_name}"))
75 })?;
76
77 let buffer = read_file(filepath)?;
78 Ok(encoding.decode(&buffer).0.into_owned())
79}
80
81use std::sync::Arc;
82
83#[derive(Clone)]
84pub enum Data {
85 Static(&'static [u8]),
86 Vec(Vec<u8>),
87 #[cfg(feature = "mmap")]
88 Map(Arc<Mmap>),
89}
90
91impl Archive for Data {
92 type Archived = rkyv::vec::ArchivedVec<u8>;
93 type Resolver = rkyv::vec::VecResolver;
94
95 fn resolve(&self, resolver: Self::Resolver, out: rkyv::Place<Self::Archived>) {
96 rkyv::vec::ArchivedVec::resolve_from_slice(self.deref(), resolver, out);
97 }
98}
99
100impl<S> RkyvSerialize<S> for Data
101where
102 S: rkyv::rancor::Fallible + rkyv::ser::Writer + rkyv::ser::Allocator + ?Sized,
103{
104 fn serialize(&self, serializer: &mut S) -> Result<Self::Resolver, S::Error> {
105 rkyv::vec::ArchivedVec::serialize_from_slice(self.deref(), serializer)
106 }
107}
108
109impl<D: rkyv::rancor::Fallible + ?Sized> RkyvDeserialize<Data, D> for rkyv::vec::ArchivedVec<u8> {
110 fn deserialize(&self, _deserializer: &mut D) -> Result<Data, D::Error> {
111 let mut vec = Vec::with_capacity(self.len());
112 vec.extend_from_slice(self.as_slice());
113 Ok(Data::Vec(vec))
114 }
115}
116
117impl Deref for Data {
118 type Target = [u8];
119 fn deref(&self) -> &Self::Target {
120 match self {
121 Data::Static(s) => s,
122 Data::Vec(v) => v,
123 #[cfg(feature = "mmap")]
124 Data::Map(m) => m,
125 }
126 }
127}
128
129impl From<&'static [u8]> for Data {
130 fn from(s: &'static [u8]) -> Self {
131 Self::Static(s)
132 }
133}
134
135impl<T: Deref<Target = [u8]>> From<&'static T> for Data {
136 fn from(t: &'static T) -> Self {
137 Self::Static(t)
138 }
139}
140
141impl From<Vec<u8>> for Data {
142 fn from(v: Vec<u8>) -> Self {
143 Self::Vec(v)
144 }
145}
146
147#[cfg(feature = "mmap")]
148impl From<Mmap> for Data {
149 fn from(m: Mmap) -> Self {
150 Self::Map(Arc::new(m))
151 }
152}
153
154impl Serialize for Data {
155 fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
156 where
157 S: serde::Serializer,
158 {
159 serializer.serialize_bytes(self.deref())
160 }
161}
162
163impl<'de> Deserialize<'de> for Data {
164 fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
165 where
166 D: serde::Deserializer<'de>,
167 {
168 let v = <Vec<u8> as serde::Deserialize>::deserialize(deserializer)?;
169 Ok(Data::Vec(v))
170 }
171}