mirror of
https://github.com/Xahau/xahaud.git
synced 2025-12-06 17:27:52 +00:00
b7ba509618
This adds codecs for snappy and lz4, and a new nodeobject codec. The nodeobject codec provides a highly efficient custom compression scheme for inner nodes, which make up the majority of nodestore databases. Non inner node objects are compressed using lz4. The NuDB backend is modified to use the nodeobject codec. This change is not backward compatible - older NuDB databases cannot be opened or imported.
513 lines
14 KiB
C++
513 lines
14 KiB
C++
//------------------------------------------------------------------------------
|
|
/*
|
|
This file is part of rippled: https://github.com/ripple/rippled
|
|
Copyright (c) 2012, 2013 Ripple Labs Inc.
|
|
|
|
Permission to use, copy, modify, and/or distribute this software for any
|
|
purpose with or without fee is hereby granted, provided that the above
|
|
copyright notice and this permission notice appear in all copies.
|
|
|
|
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
|
|
WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
|
|
MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
|
|
ANY SPECIAL , DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
|
|
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
|
|
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
|
|
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
|
|
*/
|
|
//==============================================================================
|
|
|
|
#ifndef RIPPLE_NODESTORE_CODEC_H_INCLUDED
|
|
#define RIPPLE_NODESTORE_CODEC_H_INCLUDED
|
|
|
|
#include <ripple/protocol/HashPrefix.h>
|
|
#include <beast/nudb/common.h>
|
|
#include <beast/nudb/detail/varint.h>
|
|
#include <lz4/lib/lz4.h>
|
|
#include <snappy.h>
|
|
#include <cstddef>
|
|
#include <cstring>
|
|
#include <utility>
|
|
|
|
namespace ripple {
|
|
namespace NodeStore {
|
|
|
|
namespace detail {
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
snappy_compress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf)
|
|
{
|
|
std::pair<void const*, std::size_t> result;
|
|
auto const out_max =
|
|
snappy::MaxCompressedLength(in_size);
|
|
void* const out = bf(out_max);
|
|
result.first = out;
|
|
snappy::RawCompress(
|
|
reinterpret_cast<char const*>(in),
|
|
in_size, reinterpret_cast<char*>(out),
|
|
&result.second);
|
|
return result;
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
snappy_decompress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf)
|
|
{
|
|
std::pair<void const*, std::size_t> result;
|
|
if (! snappy::GetUncompressedLength(
|
|
reinterpret_cast<char const*>(in),
|
|
in_size, &result.second))
|
|
throw beast::nudb::codec_error(
|
|
"snappy decompress");
|
|
void* const out = bf(result.second);
|
|
result.first = out;
|
|
if (! snappy::RawUncompress(
|
|
reinterpret_cast<char const*>(in), in_size,
|
|
reinterpret_cast<char*>(out)))
|
|
throw beast::nudb::codec_error(
|
|
"snappy decompress");
|
|
return result;
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
lz4_decompress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf)
|
|
{
|
|
using beast::nudb::codec_error;
|
|
using namespace beast::nudb::detail;
|
|
std::pair<void const*, std::size_t> result;
|
|
std::uint8_t const* p = reinterpret_cast<
|
|
std::uint8_t const*>(in);
|
|
auto const n = read_varint(
|
|
p, in_size, result.second);
|
|
if (n == 0)
|
|
throw codec_error(
|
|
"lz4 decompress");
|
|
void* const out = bf(result.second);
|
|
result.first = out;
|
|
if (LZ4_decompress_fast(
|
|
reinterpret_cast<char const*>(in) + n,
|
|
reinterpret_cast<char*>(out),
|
|
result.second) + n != in_size)
|
|
throw codec_error(
|
|
"lz4 decompress");
|
|
return result;
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
lz4_compress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf)
|
|
{
|
|
using beast::nudb::codec_error;
|
|
using namespace beast::nudb::detail;
|
|
std::pair<void const*, std::size_t> result;
|
|
std::array<std::uint8_t, varint_traits<
|
|
std::size_t>::max> vi;
|
|
auto const n = write_varint(
|
|
vi.data(), in_size);
|
|
auto const out_max =
|
|
LZ4_compressBound(in_size);
|
|
std::uint8_t* out = reinterpret_cast<
|
|
std::uint8_t*>(bf(n + out_max));
|
|
result.first = out;
|
|
std::memcpy(out, vi.data(), n);
|
|
auto const out_size = LZ4_compress(
|
|
reinterpret_cast<char const*>(in),
|
|
reinterpret_cast<char*>(out + n),
|
|
in_size);
|
|
if (out_size == 0)
|
|
throw codec_error(
|
|
"lz4 compress");
|
|
result.second = n + out_size;
|
|
return result;
|
|
}
|
|
|
|
//------------------------------------------------------------------------------
|
|
|
|
/*
|
|
object types:
|
|
|
|
0 = Uncompressed
|
|
1 = lz4 compressed
|
|
2 = inner node compressed
|
|
3 = full inner node
|
|
*/
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
nodeobject_decompress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf)
|
|
{
|
|
using beast::nudb::codec_error;
|
|
using namespace beast::nudb::detail;
|
|
|
|
std::uint8_t const* p = reinterpret_cast<
|
|
std::uint8_t const*>(in);
|
|
std::size_t type;
|
|
auto const vn = read_varint(
|
|
p, in_size, type);
|
|
if (vn == 0)
|
|
throw codec_error(
|
|
"nodeobject decompress");
|
|
p += vn;
|
|
in_size -= vn;
|
|
|
|
std::pair<void const*, std::size_t> result;
|
|
switch(type)
|
|
{
|
|
case 0: // uncompressed
|
|
{
|
|
result.first = p;
|
|
result.second = in_size;
|
|
break;
|
|
}
|
|
case 1: // lz4
|
|
{
|
|
result = lz4_decompress(
|
|
p, in_size, bf);
|
|
break;
|
|
}
|
|
case 2: // inner node
|
|
{
|
|
auto const hs =
|
|
field<std::uint16_t>::size; // Mask
|
|
if (in_size < hs + 32)
|
|
throw codec_error(
|
|
"nodeobject codec: short inner node");
|
|
istream is(p, in_size);
|
|
std::uint16_t mask;
|
|
read<std::uint16_t>(is, mask); // Mask
|
|
in_size -= hs;
|
|
result.second = 525;
|
|
void* const out = bf(result.second);
|
|
result.first = out;
|
|
ostream os(out, result.second);
|
|
write<std::uint32_t>(os, 0);
|
|
write<std::uint32_t>(os, 0);
|
|
write<std::uint8_t> (os, hotUNKNOWN);
|
|
write<std::uint32_t>(os, HashPrefix::innerNode);
|
|
if (mask == 0)
|
|
throw codec_error(
|
|
"nodeobject codec: empty inner node");
|
|
std::uint16_t bit = 0x8000;
|
|
for (int i = 16; i--; bit >>= 1)
|
|
{
|
|
if (mask & bit)
|
|
{
|
|
if (in_size < 32)
|
|
throw codec_error(
|
|
"nodeobject codec: short inner node");
|
|
std::memcpy(os.data(32), is(32), 32);
|
|
in_size -= 32;
|
|
}
|
|
else
|
|
{
|
|
std::memset(os.data(32), 0, 32);
|
|
}
|
|
}
|
|
if (in_size > 0)
|
|
throw codec_error(
|
|
"nodeobject codec: long inner node");
|
|
break;
|
|
}
|
|
case 3: // full inner node
|
|
{
|
|
if (in_size != 16 * 32) // hashes
|
|
throw codec_error(
|
|
"nodeobject codec: short full inner node");
|
|
istream is(p, in_size);
|
|
result.second = 525;
|
|
void* const out = bf(result.second);
|
|
result.first = out;
|
|
ostream os(out, result.second);
|
|
write<std::uint32_t>(os, 0);
|
|
write<std::uint32_t>(os, 0);
|
|
write<std::uint8_t> (os, hotUNKNOWN);
|
|
write<std::uint32_t>(os, HashPrefix::innerNode);
|
|
write(os, is(512), 512);
|
|
break;
|
|
}
|
|
default:
|
|
throw codec_error(
|
|
"nodeobject codec: bad type=" +
|
|
std::to_string(type));
|
|
};
|
|
return result;
|
|
}
|
|
|
|
template <class = void>
|
|
void const*
|
|
zero32()
|
|
{
|
|
static std::array<char, 32> v =
|
|
[]
|
|
{
|
|
std::array<char, 32> v;
|
|
v.fill(0);
|
|
return v;
|
|
}();
|
|
return v.data();
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
nodeobject_compress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf)
|
|
{
|
|
using beast::nudb::codec_error;
|
|
using namespace beast::nudb::detail;
|
|
|
|
std::size_t type = 1;
|
|
// Check for inner node
|
|
if (in_size == 525)
|
|
{
|
|
istream is(in, in_size);
|
|
std::uint32_t index;
|
|
std::uint32_t unused;
|
|
std::uint8_t kind;
|
|
std::uint32_t prefix;
|
|
read<std::uint32_t>(is, index);
|
|
read<std::uint32_t>(is, unused);
|
|
read<std::uint8_t> (is, kind);
|
|
read<std::uint32_t>(is, prefix);
|
|
if (prefix == HashPrefix::innerNode)
|
|
{
|
|
std::size_t n = 0;
|
|
std::uint16_t mask = 0;
|
|
std::array<
|
|
std::uint8_t, 512> vh;
|
|
for (unsigned bit = 0x8000;
|
|
bit; bit >>= 1)
|
|
{
|
|
void const* const h = is(32);
|
|
if (std::memcmp(
|
|
h, zero32(), 32) == 0)
|
|
continue;
|
|
std::memcpy(
|
|
vh.data() + 32 * n, h, 32);
|
|
mask |= bit;
|
|
++n;
|
|
}
|
|
std::pair<void const*,
|
|
std::size_t> result;
|
|
if (n < 16)
|
|
{
|
|
// 2 = inner node compressed
|
|
auto const type = 2U;
|
|
auto const vs = size_varint(type);
|
|
result.second =
|
|
vs +
|
|
field<std::uint16_t>::size + // mask
|
|
n * 32; // hashes
|
|
std::uint8_t* out = reinterpret_cast<
|
|
std::uint8_t*>(bf(result.second));
|
|
result.first = out;
|
|
ostream os(out, result.second);
|
|
write<varint>(os, type);
|
|
write<std::uint16_t>(os, mask);
|
|
write(os, vh.data(), n * 32);
|
|
return result;
|
|
}
|
|
// 3 = full inner node
|
|
auto const type = 3U;
|
|
auto const vs = size_varint(type);
|
|
result.second =
|
|
vs +
|
|
n * 32; // hashes
|
|
std::uint8_t* out = reinterpret_cast<
|
|
std::uint8_t*>(bf(result.second));
|
|
result.first = out;
|
|
ostream os(out, result.second);
|
|
write<varint>(os, type);
|
|
write(os, vh.data(), n * 32);
|
|
return result;
|
|
}
|
|
}
|
|
|
|
std::array<std::uint8_t, varint_traits<
|
|
std::size_t>::max> vi;
|
|
auto const vn = write_varint(
|
|
vi.data(), type);
|
|
std::pair<void const*, std::size_t> result;
|
|
switch(type)
|
|
{
|
|
case 0: // uncompressed
|
|
{
|
|
result.second = vn + in_size;
|
|
std::uint8_t* p = reinterpret_cast<
|
|
std::uint8_t*>(bf(result.second));
|
|
result.first = p;
|
|
std::memcpy(p, vi.data(), vn);
|
|
std::memcpy(p + vn, in, in_size);
|
|
break;
|
|
}
|
|
case 1: // lz4
|
|
{
|
|
std::uint8_t* p;
|
|
auto const lzr = lz4_compress(
|
|
in, in_size, [&p, &vn, &bf]
|
|
(std::size_t n)
|
|
{
|
|
p = reinterpret_cast<
|
|
std::uint8_t*>(
|
|
bf(vn + n));
|
|
return p + vn;
|
|
});
|
|
std::memcpy(p, vi.data(), vn);
|
|
result.first = p;
|
|
result.second = vn + lzr.second;
|
|
break;
|
|
}
|
|
default:
|
|
throw std::logic_error(
|
|
"nodeobject codec: unknown=" +
|
|
std::to_string(type));
|
|
};
|
|
return result;
|
|
}
|
|
|
|
} // detail
|
|
|
|
// Modifies an inner node to erase the ledger
|
|
// sequence and type information so the codec
|
|
// verification can pass.
|
|
//
|
|
template <class = void>
|
|
void
|
|
filter_inner (void* in, std::size_t in_size)
|
|
{
|
|
using beast::nudb::codec_error;
|
|
using namespace beast::nudb::detail;
|
|
|
|
// Check for inner node
|
|
if (in_size == 525)
|
|
{
|
|
istream is(in, in_size);
|
|
std::uint32_t index;
|
|
std::uint32_t unused;
|
|
std::uint8_t kind;
|
|
std::uint32_t prefix;
|
|
read<std::uint32_t>(is, index);
|
|
read<std::uint32_t>(is, unused);
|
|
read<std::uint8_t> (is, kind);
|
|
read<std::uint32_t>(is, prefix);
|
|
if (prefix == HashPrefix::innerNode)
|
|
{
|
|
ostream os(in, 9);
|
|
write<std::uint32_t>(os, 0);
|
|
write<std::uint32_t>(os, 0);
|
|
write<std::uint8_t> (os, hotUNKNOWN);
|
|
}
|
|
}
|
|
}
|
|
|
|
//------------------------------------------------------------------------------
|
|
|
|
class snappy_codec
|
|
{
|
|
public:
|
|
template <class... Args>
|
|
explicit
|
|
snappy_codec(Args&&... args)
|
|
{
|
|
}
|
|
|
|
char const*
|
|
name() const
|
|
{
|
|
return "snappy";
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
compress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf) const
|
|
{
|
|
return snappy_compress(in, in_size, bf);
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
decompress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf) const
|
|
{
|
|
return snappy_decompress(in, in_size, bf);
|
|
}
|
|
};
|
|
|
|
class lz4_codec
|
|
{
|
|
public:
|
|
template <class... Args>
|
|
explicit
|
|
lz4_codec(Args&&... args)
|
|
{
|
|
}
|
|
|
|
char const*
|
|
name() const
|
|
{
|
|
return "lz4";
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
decompress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf) const
|
|
{
|
|
return lz4_compress(in, in_size, bf);
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
compress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf) const
|
|
{
|
|
return lz4_compress(in, in_size, bf);
|
|
}
|
|
};
|
|
|
|
class nodeobject_codec
|
|
{
|
|
public:
|
|
template <class... Args>
|
|
explicit
|
|
nodeobject_codec(Args&&... args)
|
|
{
|
|
}
|
|
|
|
char const*
|
|
name() const
|
|
{
|
|
return "nodeobject";
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
decompress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf) const
|
|
{
|
|
return detail::nodeobject_decompress(
|
|
in, in_size, bf);
|
|
}
|
|
|
|
template <class BufferFactory>
|
|
std::pair<void const*, std::size_t>
|
|
compress (void const* in,
|
|
std::size_t in_size, BufferFactory&& bf) const
|
|
{
|
|
return detail::nodeobject_compress(
|
|
in, in_size, bf);
|
|
}
|
|
};
|
|
|
|
} // NodeStore
|
|
} // ripple
|
|
|
|
#endif
|