1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296
//! Train a dictionary from various sources.
//!
//! A dictionary can help improve the compression of small files.
//! The dictionary must be present during decompression,
//! but can be shared across multiple "similar" files.
//!
//! Creating a dictionary using the `zstd` C library,
//! using the `zstd` command-line interface, using this library,
//! or using the `train` binary provided, should give the same result,
//! and are therefore completely compatible.
//!
//! To use, see [`Encoder::with_dictionary`] or [`Decoder::with_dictionary`].
//!
//! [`Encoder::with_dictionary`]: ../struct.Encoder.html#method.with_dictionary
//! [`Decoder::with_dictionary`]: ../struct.Decoder.html#method.with_dictionary
#[cfg(feature = "zdict_builder")]
use std::io::{self, Read};
pub use zstd_safe::{CDict, DDict};
/// Prepared dictionary for compression
///
/// A dictionary can include its own copy of the data (if it is `'static`), or it can merely point
/// to a separate buffer (if it has another lifetime).
pub struct EncoderDictionary<'a> {
cdict: CDict<'a>,
}
impl EncoderDictionary<'static> {
/// Creates a prepared dictionary for compression.
///
/// This will copy the dictionary internally.
pub fn copy(dictionary: &[u8], level: i32) -> Self {
Self {
cdict: zstd_safe::create_cdict(dictionary, level),
}
}
}
impl<'a> EncoderDictionary<'a> {
#[cfg(feature = "experimental")]
#[cfg_attr(feature = "doc-cfg", doc(cfg(feature = "experimental")))]
/// Create prepared dictionary for compression
///
/// A level of `0` uses zstd's default (currently `3`).
///
/// Only available with the `experimental` feature. Use `EncoderDictionary::copy` otherwise.
pub fn new(dictionary: &'a [u8], level: i32) -> Self {
Self {
cdict: zstd_safe::CDict::create_by_reference(dictionary, level),
}
}
/// Returns reference to `CDict` inner object
pub fn as_cdict(&self) -> &CDict<'a> {
&self.cdict
}
}
/// Prepared dictionary for decompression
pub struct DecoderDictionary<'a> {
ddict: DDict<'a>,
}
impl DecoderDictionary<'static> {
/// Create a prepared dictionary for decompression.
///
/// This will copy the dictionary internally.
pub fn copy(dictionary: &[u8]) -> Self {
Self {
ddict: zstd_safe::DDict::create(dictionary),
}
}
}
impl<'a> DecoderDictionary<'a> {
#[cfg(feature = "experimental")]
#[cfg_attr(feature = "doc-cfg", doc(cfg(feature = "experimental")))]
/// Create prepared dictionary for decompression
///
/// Only available with the `experimental` feature. Use `DecoderDictionary::copy` otherwise.
pub fn new(dict: &'a [u8]) -> Self {
Self {
ddict: zstd_safe::DDict::create_by_reference(dict),
}
}
/// Returns reference to `DDict` inner object
pub fn as_ddict(&self) -> &DDict<'a> {
&self.ddict
}
}
/// Train a dictionary from a big continuous chunk of data, with all samples
/// contiguous in memory.
///
/// This is the most efficient way to train a dictionary,
/// since this is directly fed into `zstd`.
///
/// * `sample_data` is the concatenation of all sample data.
/// * `sample_sizes` is the size of each sample in `sample_data`.
/// The sum of all `sample_sizes` should equal the length of `sample_data`.
/// * `max_size` is the maximum size of the dictionary to generate.
///
/// The result is the dictionary data. You can, for example, feed it to [`CDict::create`].
#[cfg(feature = "zdict_builder")]
#[cfg_attr(feature = "doc-cfg", doc(cfg(feature = "zdict_builder")))]
pub fn from_continuous(
sample_data: &[u8],
sample_sizes: &[usize],
max_size: usize,
) -> io::Result<Vec<u8>> {
use crate::map_error_code;
// Complain if the lengths don't add up to the entire data.
if sample_sizes.iter().sum::<usize>() != sample_data.len() {
return Err(io::Error::new(
io::ErrorKind::Other,
"sample sizes don't add up".to_string(),
));
}
let mut result = Vec::with_capacity(max_size);
zstd_safe::train_from_buffer(&mut result, sample_data, sample_sizes)
.map_err(map_error_code)?;
Ok(result)
}
/// Train a dictionary from multiple samples.
///
/// The samples will internally be copied to a single continuous buffer,
/// so make sure you have enough memory available.
///
/// If you need to stretch your system's limits,
/// [`from_continuous`] directly uses the given slice.
///
/// [`from_continuous`]: ./fn.from_continuous.html
///
/// * `samples` is a list of individual samples to train on.
/// * `max_size` is the maximum size of the dictionary to generate.
///
/// The result is the dictionary data. You can, for example, feed it to [`CDict::create`].
#[cfg(feature = "zdict_builder")]
#[cfg_attr(feature = "doc-cfg", doc(cfg(feature = "zdict_builder")))]
pub fn from_samples<S: AsRef<[u8]>>(
samples: &[S],
max_size: usize,
) -> io::Result<Vec<u8>> {
// Pre-allocate the entire required size.
let total_length: usize =
samples.iter().map(|sample| sample.as_ref().len()).sum();
let mut data = Vec::with_capacity(total_length);
// Copy every sample to a big chunk of memory
data.extend(samples.iter().flat_map(|s| s.as_ref()).cloned());
let sizes: Vec<_> = samples.iter().map(|s| s.as_ref().len()).collect();
from_continuous(&data, &sizes, max_size)
}
/// Train a dictionary from multiple samples.
///
/// Unlike [`from_samples`], this does not require having a list of all samples.
/// It also allows running into an error when iterating through the samples.
///
/// They will still be copied to a continuous array and fed to [`from_continuous`].
///
/// * `samples` is an iterator of individual samples to train on.
/// * `max_size` is the maximum size of the dictionary to generate.
///
/// The result is the dictionary data. You can, for example, feed it to [`CDict::create`].
///
/// # Examples
///
/// ```rust,no_run
/// // Train from a couple of json files.
/// let dict_buffer = zstd::dict::from_sample_iterator(
/// ["file_a.json", "file_b.json"]
/// .into_iter()
/// .map(|filename| std::fs::File::open(filename)),
/// 10_000, // 10kB dictionary
/// ).unwrap();
/// ```
///
/// ```rust,no_run
/// use std::io::BufRead as _;
/// // Treat each line from stdin as a separate sample.
/// let dict_buffer = zstd::dict::from_sample_iterator(
/// std::io::stdin().lock().lines().map(|line: std::io::Result<String>| {
/// // Transform each line into a `Cursor<Vec<u8>>` so they implement Read.
/// line.map(String::into_bytes)
/// .map(std::io::Cursor::new)
/// }),
/// 10_000, // 10kB dictionary
/// ).unwrap();
/// ```
#[cfg(feature = "zdict_builder")]
#[cfg_attr(feature = "doc-cfg", doc(cfg(feature = "zdict_builder")))]
pub fn from_sample_iterator<I, R>(
samples: I,
max_size: usize,
) -> io::Result<Vec<u8>>
where
I: IntoIterator<Item = io::Result<R>>,
R: Read,
{
let mut data = Vec::new();
let mut sizes = Vec::new();
for sample in samples {
let mut sample = sample?;
let len = sample.read_to_end(&mut data)?;
sizes.push(len);
}
from_continuous(&data, &sizes, max_size)
}
/// Train a dict from a list of files.
///
/// * `filenames` is an iterator of files to load. Each file will be treated as an individual
/// sample.
/// * `max_size` is the maximum size of the dictionary to generate.
///
/// The result is the dictionary data. You can, for example, feed it to [`CDict::create`].
#[cfg(feature = "zdict_builder")]
#[cfg_attr(feature = "doc-cfg", doc(cfg(feature = "zdict_builder")))]
pub fn from_files<I, P>(filenames: I, max_size: usize) -> io::Result<Vec<u8>>
where
P: AsRef<std::path::Path>,
I: IntoIterator<Item = P>,
{
from_sample_iterator(
filenames
.into_iter()
.map(|filename| std::fs::File::open(filename)),
max_size,
)
}
#[cfg(test)]
#[cfg(feature = "zdict_builder")]
mod tests {
use std::fs;
use std::io;
use std::io::Read;
use walkdir;
#[test]
fn test_dict_training() {
// Train a dictionary
let paths: Vec<_> = walkdir::WalkDir::new("src")
.into_iter()
.map(|entry| entry.unwrap())
.map(|entry| entry.into_path())
.filter(|path| path.to_str().unwrap().ends_with(".rs"))
.collect();
let dict = super::from_files(&paths, 4000).unwrap();
for path in paths {
let mut buffer = Vec::new();
let mut file = fs::File::open(path).unwrap();
let mut content = Vec::new();
file.read_to_end(&mut content).unwrap();
io::copy(
&mut &content[..],
&mut crate::stream::Encoder::with_dictionary(
&mut buffer,
1,
&dict,
)
.unwrap()
.auto_finish(),
)
.unwrap();
let mut result = Vec::new();
io::copy(
&mut crate::stream::Decoder::with_dictionary(
&buffer[..],
&dict[..],
)
.unwrap(),
&mut result,
)
.unwrap();
assert_eq!(&content, &result);
}
}
}