[refactor-block] refactor DataBlock and DataBlockPath

This commit is contained in:
Alex Auvolat
2024-02-23 12:11:14 +01:00
parent 07c7895948
commit cd1069c1d4
3 changed files with 88 additions and 78 deletions
+64 -42
View File
@@ -13,77 +13,99 @@ pub enum DataBlockHeader {
Compressed,
}
/// A possibly compressed block of data
pub enum DataBlock {
/// Uncompressed data
Plain(Bytes),
/// Data compressed with zstd
Compressed(Bytes),
#[derive(Debug)]
pub struct DataBlockElem<T> {
header: DataBlockHeader,
elem: T,
}
#[derive(Debug)]
pub enum DataBlockPath {
/// Uncompressed data fail
Plain(PathBuf),
/// Compressed data fail
Compressed(PathBuf),
/// A possibly compressed block of data
pub type DataBlock = DataBlockElem<Bytes>;
/// A path to a possibly compressed block of data
pub type DataBlockPath = DataBlockElem<PathBuf>;
impl DataBlockHeader {
pub fn is_compressed(&self) -> bool {
matches!(self, DataBlockHeader::Compressed)
}
}
impl<T> DataBlockElem<T> {
pub fn from_parts(header: DataBlockHeader, elem: T) -> Self {
Self { header, elem }
}
pub fn plain(elem: T) -> Self {
Self {
header: DataBlockHeader::Plain,
elem,
}
}
pub fn compressed(elem: T) -> Self {
Self {
header: DataBlockHeader::Compressed,
elem,
}
}
pub fn into_parts(self) -> (DataBlockHeader, T) {
(self.header, self.elem)
}
pub fn as_parts_ref(&self) -> (DataBlockHeader, &T) {
(self.header, &self.elem)
}
/// Query whether this block is compressed
pub fn is_compressed(&self) -> bool {
self.header.is_compressed()
}
}
impl DataBlock {
/// Query whether this block is compressed
pub fn is_compressed(&self) -> bool {
matches!(self, DataBlock::Compressed(_))
}
/// Get the inner, possibly compressed buffer. You should probably use [`DataBlock::verify_get`]
/// instead
pub fn inner_buffer(&self) -> &[u8] {
use DataBlock::*;
let (Plain(ref res) | Compressed(ref res)) = self;
res
&self.elem
}
/// Verify data integrity. Does not return the buffer content.
pub fn verify(&self, hash: Hash) -> Result<(), Error> {
match self {
DataBlock::Plain(data) => {
if blake2sum(data) == hash {
match self.header {
DataBlockHeader::Plain => {
if blake2sum(&self.elem) == hash {
Ok(())
} else {
Err(Error::CorruptData(hash))
}
}
DataBlock::Compressed(data) => zstd::stream::copy_decode(&data[..], std::io::sink())
.map_err(|_| Error::CorruptData(hash)),
DataBlockHeader::Compressed => {
zstd::stream::copy_decode(&self.elem[..], std::io::sink())
.map_err(|_| Error::CorruptData(hash))
}
}
}
pub async fn from_buffer(data: Bytes, level: Option<i32>) -> DataBlock {
tokio::task::spawn_blocking(move || {
if let Some(level) = level {
if let Ok(data) = zstd_encode(&data[..], level) {
return DataBlock::Compressed(data.into());
if let Ok(data_compressed) = zstd_encode(&data[..], level) {
return DataBlock {
header: DataBlockHeader::Compressed,
elem: data_compressed.into(),
};
}
}
DataBlock::Plain(data)
DataBlock {
header: DataBlockHeader::Plain,
elem: data.into(),
}
})
.await
.unwrap()
}
pub fn into_parts(self) -> (DataBlockHeader, Bytes) {
match self {
DataBlock::Plain(data) => (DataBlockHeader::Plain, data),
DataBlock::Compressed(data) => (DataBlockHeader::Compressed, data),
}
}
pub fn from_parts(h: DataBlockHeader, bytes: Bytes) -> Self {
match h {
DataBlockHeader::Plain => DataBlock::Plain(bytes),
DataBlockHeader::Compressed => DataBlock::Compressed(bytes),
}
}
}
fn zstd_encode<R: std::io::Read>(mut source: R, level: i32) -> std::io::Result<Vec<u8>> {