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
use super::{Chunker, ChunkerError};
use std::io::Read;
/// Settings for a fastcdc `Chunker`
///
/// These are limited to `usize`, and not `u64`, because this implementation makes
/// extensive use of in memory buffers of size `max_size`
///
/// This chunker, unlike `BuzHash` does not support any attempted mitigation of
/// chunk based fingerprinting attacks. Those who are concerned about such an
/// attack may wish to use the `BuzHash` chunker until such a time that a better
/// repository format that does not leak information about chunk sizes can be
/// developed.
#[derive(Clone, Copy)]
pub struct FastCDC {
pub min_size: usize,
pub max_size: usize,
pub avg_size: usize,
}
impl Chunker for FastCDC {
type Chunks = FastCDCChunker;
fn chunk_boxed(&self, read: Box<dyn Read + Send + 'static>) -> Self::Chunks {
FastCDCChunker {
settings: *self,
buffer: vec![0_u8; self.max_size],
length: 0,
read,
eof: false,
}
}
}
impl Default for FastCDC {
fn default() -> Self {
FastCDC {
min_size: 32_768,
avg_size: 65_536,
max_size: 131_072,
}
}
}
pub struct FastCDCChunker {
/// The settings used for this `Chunker`
settings: FastCDC,
/// The in memory buffer used to hack the chosen FastCDC implementation into working
///
/// This must always be kept at a size of `max_size`
buffer: Vec<u8>,
/// The length of the data in the buffer
length: usize,
/// The reader this `Chunker` is slicing
read: Box<dyn Read + Send + 'static>,
/// Has the reader hit EoF?
eof: bool,
}
impl FastCDCChunker {
/// Drains a specified number of bytes from the reader, and refills it back up to `max_size` with
/// zeros.
///
/// Additionally updates the pointer to the new end of the vec
///
/// # Errors
///
/// Returns `ChunkerError::InternalError` if the given count of bytes to drain is greater than the
/// current size of the used region of the buffer.
///
/// # Panics
///
/// Panics if the internal buffer's length is not `max_size`. This is an invariant, and the end
/// consumer of the struct should never be exposed to this error.
fn drain_bytes(&mut self, count: usize) -> Result<Vec<u8>, ChunkerError> {
assert!(self.buffer.len() == self.settings.max_size);
if count > self.length {
Err(ChunkerError::InternalError(format!(
"Invalid count given to FastCDCChunker::drain_bytes. Count: {}, Length: {}",
count, self.length
)))
} else {
// Drain the bytes from the vec
let output = self.buffer.drain(..count).collect::<Vec<_>>();
// Update the length
self.length -= count;
// Resize the buffer back to max_size
self.buffer.resize(self.settings.max_size, 0_u8);
// Return the drained bytes
Ok(output)
}
}
/// Returns true if the internal buffer is empty
fn is_empty(&self) -> bool {
self.length == 0
}
/// Attempts fill the buffer back up with bytes from the read
///
/// Returns the number of bytes read. Will not attempt to read bytes if `EoF` has already been
/// encountered.
///
/// # Errors
///
/// Returns `ChunkerError::IOError` if the reader provides any error value during reading
///
/// # Panics
///
/// Panics if the internal buffer's length is not `max_size`. This is an invariant, and the end
/// consumer of the struct shuold never be exposed to this error.
fn read_bytes(&mut self) -> Result<usize, ChunkerError> {
assert!(self.buffer.len() == self.settings.max_size);
if self.eof {
Ok(0)
} else {
let mut total_bytes = 0;
// While we have not hit eof, and there is still room in the buffer, keep reading
while !self.eof && self.length < self.settings.max_size {
// read some bytes
let bytes_read = self.read.read(&mut self.buffer[self.length..])?;
// Update the length
self.length += bytes_read;
// If the number of bytes read was zero, set the eof flag
if bytes_read == 0 {
self.eof = true;
}
// Update the total
total_bytes += bytes_read;
}
Ok(total_bytes)
}
}
/// Uses the `FastCDC` algorithm to produce the next chunk of data.
///
/// # Errors
///
/// Returns `ChunkerError::Empty` if `EoF` has been hit
///
/// # Panics
///
/// Panics if the internal buffer's length is not `max_size`. This is an invariant, and the end
/// consumer of the struct should never be exposed to this error.
fn next_chunk(&mut self) -> Result<Vec<u8>, ChunkerError> {
assert_eq!(self.buffer.len(), self.settings.max_size);
// First, perform a read, to make sure the buffer is as full as it can be
self.read_bytes()?;
// Check to see if we are empty, if so, return early
if self.is_empty() {
Err(ChunkerError::Empty)
} else {
// Attempt to produce our slice
let mut slicer = fastcdc::FastCDC::new(
&self.buffer[..self.length],
self.settings.min_size,
self.settings.avg_size,
self.settings.max_size,
);
if let Some(chunk) = slicer.next() {
let result = self.drain_bytes(chunk.length)?;
Ok(result)
} else {
// We really shouldn't be here, since we ruled out the empty case, earlier but we
// will error anyway
Err(ChunkerError::Empty)
}
}
}
}
impl Iterator for FastCDCChunker {
type Item = Result<Vec<u8>, ChunkerError>;
fn next(&mut self) -> Option<Result<Vec<u8>, ChunkerError>> {
let slice = self.next_chunk();
if let Err(ChunkerError::Empty) = slice {
None
} else {
Some(slice)
}
}
}
#[cfg(test)]
mod tests {
use super::*;
use rand::prelude::*;
use std::io::Cursor;
// Provides a test slice 10 times the default max size in length
fn get_test_data() -> Vec<u8> {
let size = FastCDC::default().max_size * 10;
let mut vec = vec![0_u8; size];
rand::thread_rng().fill_bytes(&mut vec);
vec
}
// Data should be split into one or more chunks.
//
// In this case, the data is larger than `max_size`, so it should be more than one chunk
#[test]
fn one_or_more_chunks() {
let data = get_test_data();
let cursor = Cursor::new(data);
let chunker = FastCDC::default();
assert!(
chunker
.chunk(cursor)
.map(std::result::Result::unwrap)
.count()
> 1
);
}
// Data should be identical after reassembaly by simple concatenation
#[test]
fn reassemble_data() {
let data = get_test_data();
let cursor = Cursor::new(data.clone());
let chunks = FastCDC::default()
.chunk(cursor)
.map(std::result::Result::unwrap)
.collect::<Vec<_>>();
let rebuilt: Vec<u8> = chunks.concat();
assert_eq!(data, rebuilt);
}
// Running the chunker over the same data twice should result in identical chunks
#[test]
fn identical_chunks() {
let data = get_test_data();
let cursor1 = Cursor::new(data.clone());
let chunks1 = FastCDC::default()
.chunk(cursor1)
.map(std::result::Result::unwrap)
.collect::<Vec<_>>();
let cursor2 = Cursor::new(data);
let chunks2 = FastCDC::default()
.chunk(cursor2)
.map(std::result::Result::unwrap)
.collect::<Vec<_>>();
assert_eq!(chunks1, chunks2);
}
// Verifies that this `Chunker` does not produce chunks larger than its max size
#[test]
fn max_size() {
let data = get_test_data();
let max_size = FastCDC::default().max_size;
let chunks = FastCDC::default()
.chunk(Cursor::new(data))
.map(std::result::Result::unwrap)
.collect::<Vec<_>>();
for chunk in chunks {
assert!(chunk.len() <= max_size);
}
}
// Verifies that this `Chunker`, at most, produces 1 under-sized chunk
#[test]
fn min_size() {
let data = get_test_data();
let min_size = FastCDC::default().min_size;
let chunks = FastCDC::default()
.chunk(Cursor::new(data))
.map(std::result::Result::unwrap)
.collect::<Vec<_>>();
let mut undersized_count = 0;
for chunk in chunks {
if chunk.len() < min_size {
undersized_count += 1;
}
}
assert!(undersized_count <= 1);
}
}