feat(legacy): port the v0.3 decoder to Rust
Move the frozen v0.3 frame decoder, entropy tables, and streaming context state into a feature-gated Rust module. Keep the historical C translation unit as an ABI shim so existing C callers retain the opaque context and error-code contracts while v0.4 and newer decoders remain unchanged. Test Plan: - RUSTC_WRAPPER= CARGO_BUILD_RUSTC_WRAPPER= cargo test --manifest-path rust/Cargo.toml --no-default-features --features legacy-v01,legacy-v02, legacy-v03 - RUSTC_WRAPPER= CARGO_BUILD_RUSTC_WRAPPER= cargo test --manifest-path rust/Cargo.toml --no-default-features --features compression,decompression, dict-builder,legacy-v01,legacy-v02,legacy-v03 - RUSTC_WRAPPER= CARGO_BUILD_RUSTC_WRAPPER= cargo clippy --manifest-path rust/Cargo.toml --all-targets --no-default-features --features compression,decompression,dict-builder,legacy-v01,legacy-v02,legacy-v03 -- -D warnings - make -B -C lib libzstd.a ZSTD_LEGACY_SUPPORT=3 V=1 - git diff --cached --check
This commit is contained in:
@@ -41,3 +41,6 @@ pub mod zstd_v01;
|
||||
|
||||
#[cfg(feature = "legacy-v02")]
|
||||
pub mod zstd_v02;
|
||||
|
||||
#[cfg(feature = "legacy-v03")]
|
||||
pub mod zstd_v03;
|
||||
|
||||
@@ -0,0 +1,1790 @@
|
||||
#![allow(non_snake_case)]
|
||||
|
||||
//! Frozen decoder for the zstd v0.3 format.
|
||||
//!
|
||||
//! `lib/legacy/zstd_v03.c` is an old, self-contained decoder. This module
|
||||
//! keeps that boundary: it owns its FSE, bit-stream, and Huffman state rather
|
||||
//! than depending on the current entropy implementations. The public entry
|
||||
//! points below retain the C ABI and the context is deliberately malloc/free
|
||||
//! allocated so C callers can continue to own an opaque `ZSTDv03_Dctx`.
|
||||
|
||||
use crate::errors::{ERR_isError, ZstdErrorCode, ERROR};
|
||||
use std::os::raw::{c_uint, c_void};
|
||||
use std::ptr;
|
||||
|
||||
const ZSTD_MAGIC_NUMBER: u32 = 0xFD2F_B523;
|
||||
const ZSTD_CONTENTSIZE_ERROR: u64 = u64::MAX - 1;
|
||||
|
||||
const BLOCKSIZE: usize = 128 * 1024;
|
||||
const MIN_SEQUENCES_SIZE: usize = 2 + 2 + 3 + 1;
|
||||
const MIN_CBLOCK_SIZE: usize = 3 + MIN_SEQUENCES_SIZE;
|
||||
const MINMATCH: usize = 4;
|
||||
|
||||
const ML_BITS: u32 = 7;
|
||||
const LL_BITS: u32 = 6;
|
||||
const OFF_BITS: u32 = 5;
|
||||
const MAX_ML: u32 = (1 << ML_BITS) - 1;
|
||||
const MAX_LL: u32 = (1 << LL_BITS) - 1;
|
||||
const MAX_OFF: u32 = 31;
|
||||
const ML_FSE_LOG: u32 = 10;
|
||||
const LL_FSE_LOG: u32 = 10;
|
||||
const OFF_FSE_LOG: u32 = 9;
|
||||
|
||||
const FSE_MAX_MEMORY_USAGE: u32 = 14;
|
||||
const FSE_MAX_SYMBOL_VALUE: u32 = 255;
|
||||
const FSE_MAX_TABLELOG: u32 = FSE_MAX_MEMORY_USAGE - 2;
|
||||
const FSE_MIN_TABLELOG: u32 = 5;
|
||||
const FSE_TABLELOG_ABSOLUTE_MAX: u32 = 15;
|
||||
|
||||
const HUF_MAX_SYMBOL_VALUE: usize = 255;
|
||||
const HUF_MAX_TABLELOG: usize = 12;
|
||||
const HUF_ABSOLUTE_MAX_TABLELOG: usize = 16;
|
||||
|
||||
const BT_COMPRESSED: u32 = 0;
|
||||
const BT_RAW: u32 = 1;
|
||||
const BT_RLE: u32 = 2;
|
||||
const BT_END: u32 = 3;
|
||||
|
||||
const DSTREAM_UNFINISHED: u32 = 0;
|
||||
const DSTREAM_END_OF_BUFFER: u32 = 1;
|
||||
const DSTREAM_COMPLETED: u32 = 2;
|
||||
const DSTREAM_TOO_FAR: u32 = 3;
|
||||
|
||||
const USIZE_BITS: u32 = usize::BITS;
|
||||
|
||||
#[repr(C)]
|
||||
#[derive(Clone, Copy)]
|
||||
struct FseDecode {
|
||||
new_state: u16,
|
||||
symbol: u8,
|
||||
nb_bits: u8,
|
||||
}
|
||||
|
||||
#[repr(C)]
|
||||
struct FseDTableHeader {
|
||||
table_log: u16,
|
||||
fast_mode: u16,
|
||||
}
|
||||
|
||||
#[derive(Clone, Copy)]
|
||||
struct DStream {
|
||||
bit_container: usize,
|
||||
bits_consumed: u32,
|
||||
ptr: *const u8,
|
||||
start: *const u8,
|
||||
}
|
||||
|
||||
#[derive(Clone, Copy)]
|
||||
struct FseDState {
|
||||
state: usize,
|
||||
table: *const FseDecode,
|
||||
}
|
||||
|
||||
#[inline]
|
||||
fn highbit32(value: u32) -> u32 {
|
||||
value.leading_zeros() ^ 31
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn read_le16(ptr: *const u8) -> u16 {
|
||||
u16::from_le_bytes(std::ptr::read_unaligned(ptr as *const [u8; 2]))
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn read_le24(ptr: *const u8) -> u32 {
|
||||
(read_le16(ptr) as u32) | ((*ptr.add(2) as u32) << 16)
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn read_le32(ptr: *const u8) -> u32 {
|
||||
u32::from_le_bytes(std::ptr::read_unaligned(ptr as *const [u8; 4]))
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn read_le_size(ptr: *const u8) -> usize {
|
||||
if std::mem::size_of::<usize>() == 4 {
|
||||
read_le32(ptr) as usize
|
||||
} else {
|
||||
u64::from_le_bytes(std::ptr::read_unaligned(ptr as *const [u8; 8])) as usize
|
||||
}
|
||||
}
|
||||
|
||||
/* ******************************************
|
||||
* Backward bit stream (the v0.3 snapshot)
|
||||
********************************************/
|
||||
|
||||
unsafe fn init_dstream(stream: &mut DStream, src: *const u8, src_size: usize) -> usize {
|
||||
let word = std::mem::size_of::<usize>();
|
||||
if src_size == 0 {
|
||||
*stream = DStream {
|
||||
bit_container: 0,
|
||||
bits_consumed: 0,
|
||||
ptr: ptr::null(),
|
||||
start: ptr::null(),
|
||||
};
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
|
||||
stream.start = src;
|
||||
if src_size >= word {
|
||||
stream.ptr = src.add(src_size - word);
|
||||
stream.bit_container = read_le_size(stream.ptr);
|
||||
let end_byte = *src.add(src_size - 1) as u32;
|
||||
if end_byte == 0 {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
stream.bits_consumed = 8 - highbit32(end_byte);
|
||||
} else {
|
||||
stream.ptr = src;
|
||||
stream.bit_container = *src as usize;
|
||||
if src_size >= 7 {
|
||||
stream.bit_container += (*src.add(6) as usize) << (USIZE_BITS as usize - 16);
|
||||
}
|
||||
if src_size >= 6 {
|
||||
stream.bit_container += (*src.add(5) as usize) << (USIZE_BITS as usize - 24);
|
||||
}
|
||||
if src_size >= 5 {
|
||||
stream.bit_container += (*src.add(4) as usize) << (USIZE_BITS as usize - 32);
|
||||
}
|
||||
if src_size >= 4 {
|
||||
stream.bit_container += (*src.add(3) as usize) << 24;
|
||||
}
|
||||
if src_size >= 3 {
|
||||
stream.bit_container += (*src.add(2) as usize) << 16;
|
||||
}
|
||||
if src_size >= 2 {
|
||||
stream.bit_container += (*src.add(1) as usize) << 8;
|
||||
}
|
||||
let end_byte = *src.add(src_size - 1) as u32;
|
||||
if end_byte == 0 {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
stream.bits_consumed = 8 - highbit32(end_byte);
|
||||
stream.bits_consumed += ((word - src_size) * 8) as u32;
|
||||
}
|
||||
src_size
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn look_bits(stream: &DStream, nb_bits: u32) -> usize {
|
||||
let mask = USIZE_BITS - 1;
|
||||
((stream.bit_container << (stream.bits_consumed & mask)) >> 1)
|
||||
>> (mask.wrapping_sub(nb_bits) & mask)
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn look_bits_fast(stream: &DStream, nb_bits: u32) -> usize {
|
||||
let mask = USIZE_BITS - 1;
|
||||
(stream.bit_container << (stream.bits_consumed & mask))
|
||||
>> ((mask + 1).wrapping_sub(nb_bits) & mask)
|
||||
}
|
||||
|
||||
#[inline]
|
||||
fn skip_bits(stream: &mut DStream, nb_bits: u32) {
|
||||
stream.bits_consumed = stream.bits_consumed.wrapping_add(nb_bits);
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn read_bits(stream: &mut DStream, nb_bits: u32) -> usize {
|
||||
let value = look_bits(stream, nb_bits);
|
||||
skip_bits(stream, nb_bits);
|
||||
value
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn read_bits_fast(stream: &mut DStream, nb_bits: u32) -> usize {
|
||||
let value = look_bits_fast(stream, nb_bits);
|
||||
skip_bits(stream, nb_bits);
|
||||
value
|
||||
}
|
||||
|
||||
unsafe fn reload_dstream(stream: &mut DStream) -> u32 {
|
||||
let word = std::mem::size_of::<usize>();
|
||||
if stream.bits_consumed > (word * 8) as u32 {
|
||||
return DSTREAM_TOO_FAR;
|
||||
}
|
||||
if (stream.ptr as usize) >= (stream.start as usize).wrapping_add(word) {
|
||||
stream.ptr = stream.ptr.sub((stream.bits_consumed >> 3) as usize);
|
||||
stream.bits_consumed &= 7;
|
||||
stream.bit_container = read_le_size(stream.ptr);
|
||||
return DSTREAM_UNFINISHED;
|
||||
}
|
||||
if stream.ptr == stream.start {
|
||||
if stream.bits_consumed < (word * 8) as u32 {
|
||||
return DSTREAM_END_OF_BUFFER;
|
||||
}
|
||||
return DSTREAM_COMPLETED;
|
||||
}
|
||||
|
||||
let mut nb_bytes = stream.bits_consumed >> 3;
|
||||
let mut result = DSTREAM_UNFINISHED;
|
||||
if (stream.ptr as usize).wrapping_sub(nb_bytes as usize) < stream.start as usize {
|
||||
nb_bytes = (stream.ptr as usize - stream.start as usize) as u32;
|
||||
result = DSTREAM_END_OF_BUFFER;
|
||||
}
|
||||
stream.ptr = stream.ptr.sub(nb_bytes as usize);
|
||||
stream.bits_consumed -= nb_bytes * 8;
|
||||
stream.bit_container = read_le_size(stream.ptr);
|
||||
result
|
||||
}
|
||||
|
||||
#[inline]
|
||||
fn end_of_dstream(stream: &DStream) -> bool {
|
||||
stream.ptr == stream.start && stream.bits_consumed == USIZE_BITS
|
||||
}
|
||||
|
||||
/* ******************************************
|
||||
* FSE decoding
|
||||
********************************************/
|
||||
|
||||
#[inline]
|
||||
fn fse_table_step(table_size: u32) -> u32 {
|
||||
(table_size >> 1) + (table_size >> 3) + 3
|
||||
}
|
||||
|
||||
#[allow(clippy::needless_range_loop)]
|
||||
unsafe fn fse_build_dtable(
|
||||
dt: &mut [u32],
|
||||
normalized_counter: &[i16; 256],
|
||||
max_symbol_value: u32,
|
||||
table_log: u32,
|
||||
) -> usize {
|
||||
if max_symbol_value > FSE_MAX_SYMBOL_VALUE {
|
||||
return ERROR(ZstdErrorCode::MaxSymbolValueTooLarge);
|
||||
}
|
||||
if table_log > FSE_MAX_TABLELOG {
|
||||
return ERROR(ZstdErrorCode::TableLogTooLarge);
|
||||
}
|
||||
|
||||
let table_size = 1u32 << table_log;
|
||||
let table_mask = table_size - 1;
|
||||
let step = fse_table_step(table_size);
|
||||
let mut symbol_next = [0u16; 256];
|
||||
let table_header = dt.as_mut_ptr() as *mut FseDTableHeader;
|
||||
let table_decode = dt.as_mut_ptr().add(1) as *mut FseDecode;
|
||||
let mut position = 0u32;
|
||||
let mut high_threshold = table_size - 1;
|
||||
let large_limit = (1i32 << (table_log - 1)) as i16;
|
||||
let mut no_large = 1u16;
|
||||
|
||||
(*table_header).table_log = table_log as u16;
|
||||
for symbol in 0..=max_symbol_value as usize {
|
||||
let count = normalized_counter[symbol];
|
||||
if count == -1 {
|
||||
(*table_decode.add(high_threshold as usize)).symbol = symbol as u8;
|
||||
high_threshold = high_threshold.wrapping_sub(1);
|
||||
symbol_next[symbol] = 1;
|
||||
} else {
|
||||
if count >= large_limit {
|
||||
no_large = 0;
|
||||
}
|
||||
symbol_next[symbol] = count as u16;
|
||||
}
|
||||
}
|
||||
|
||||
for symbol in 0..=max_symbol_value as usize {
|
||||
let count = normalized_counter[symbol];
|
||||
for _ in 0..count.max(0) {
|
||||
(*table_decode.add(position as usize)).symbol = symbol as u8;
|
||||
position = (position + step) & table_mask;
|
||||
while position > high_threshold {
|
||||
position = (position + step) & table_mask;
|
||||
}
|
||||
}
|
||||
}
|
||||
if position != 0 {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
|
||||
for index in 0..table_size as usize {
|
||||
let symbol = (*table_decode.add(index)).symbol as usize;
|
||||
let next_state = symbol_next[symbol];
|
||||
symbol_next[symbol] = symbol_next[symbol].wrapping_add(1);
|
||||
let nb_bits = (table_log - highbit32(next_state as u32)) as u8;
|
||||
(*table_decode.add(index)).nb_bits = nb_bits;
|
||||
(*table_decode.add(index)).new_state =
|
||||
(((next_state as u32) << nb_bits).wrapping_sub(table_size)) as u16;
|
||||
}
|
||||
(*table_header).fast_mode = no_large;
|
||||
0
|
||||
}
|
||||
|
||||
unsafe fn fse_read_ncount(
|
||||
normalized_counter: &mut [i16; 256],
|
||||
max_sv: &mut u32,
|
||||
table_log: &mut u32,
|
||||
header: *const u8,
|
||||
header_size: usize,
|
||||
) -> usize {
|
||||
if header_size < 4 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let start = header as usize;
|
||||
let end = start.wrapping_add(header_size);
|
||||
let mut ip = header;
|
||||
let mut char_num = 0u32;
|
||||
let mut previous_zero = false;
|
||||
let mut bit_stream = read_le32(ip);
|
||||
let mut nb_bits = ((bit_stream & 0xF) + FSE_MIN_TABLELOG) as i32;
|
||||
if nb_bits > FSE_TABLELOG_ABSOLUTE_MAX as i32 {
|
||||
return ERROR(ZstdErrorCode::TableLogTooLarge);
|
||||
}
|
||||
bit_stream >>= 4;
|
||||
let mut bit_count = 4i32;
|
||||
*table_log = nb_bits as u32;
|
||||
let mut remaining = (1i32 << nb_bits) + 1;
|
||||
let mut threshold = 1i32 << nb_bits;
|
||||
nb_bits += 1;
|
||||
|
||||
while remaining > 1 && char_num <= *max_sv {
|
||||
if previous_zero {
|
||||
let mut n0 = char_num;
|
||||
while bit_stream & 0xFFFF == 0xFFFF {
|
||||
n0 += 24;
|
||||
if (ip as usize) < end.wrapping_sub(5) {
|
||||
ip = ip.add(2);
|
||||
bit_stream = read_le32(ip) >> bit_count;
|
||||
} else {
|
||||
bit_stream >>= 16;
|
||||
bit_count += 16;
|
||||
}
|
||||
}
|
||||
while bit_stream & 3 == 3 {
|
||||
n0 += 3;
|
||||
bit_stream >>= 2;
|
||||
bit_count += 2;
|
||||
}
|
||||
n0 += bit_stream & 3;
|
||||
bit_count += 2;
|
||||
if n0 > *max_sv {
|
||||
return ERROR(ZstdErrorCode::MaxSymbolValueTooSmall);
|
||||
}
|
||||
while char_num < n0 {
|
||||
normalized_counter[char_num as usize] = 0;
|
||||
char_num += 1;
|
||||
}
|
||||
if (ip as usize) <= end.wrapping_sub(7)
|
||||
|| (ip as usize).wrapping_add((bit_count >> 3) as usize) <= end.wrapping_sub(4)
|
||||
{
|
||||
ip = ip.add((bit_count >> 3) as usize);
|
||||
bit_count &= 7;
|
||||
bit_stream = read_le32(ip) >> bit_count;
|
||||
} else {
|
||||
bit_stream >>= 2;
|
||||
}
|
||||
}
|
||||
|
||||
let max = ((2 * threshold - 1) - remaining) as i16;
|
||||
let mut count: i16;
|
||||
if (bit_stream & (threshold - 1) as u32) < max as i32 as u32 {
|
||||
count = (bit_stream & (threshold - 1) as u32) as u16 as i16;
|
||||
bit_count += nb_bits - 1;
|
||||
} else {
|
||||
count = (bit_stream & (2 * threshold - 1) as u32) as u16 as i16;
|
||||
if count as i32 >= threshold {
|
||||
count = (count as i32 - max as i32) as i16;
|
||||
}
|
||||
bit_count += nb_bits;
|
||||
}
|
||||
count = count.wrapping_sub(1);
|
||||
remaining -= (count as i32).abs();
|
||||
normalized_counter[char_num as usize] = count;
|
||||
char_num += 1;
|
||||
previous_zero = count == 0;
|
||||
while remaining < threshold {
|
||||
nb_bits -= 1;
|
||||
threshold >>= 1;
|
||||
}
|
||||
|
||||
if (ip as usize) <= end.wrapping_sub(7)
|
||||
|| (ip as usize).wrapping_add((bit_count >> 3) as usize) <= end.wrapping_sub(4)
|
||||
{
|
||||
ip = ip.add((bit_count >> 3) as usize);
|
||||
bit_count &= 7;
|
||||
} else {
|
||||
bit_count -= (8 * (end.wrapping_sub(4) as isize - ip as usize as isize)) as i32;
|
||||
ip = (end - 4) as *const u8;
|
||||
}
|
||||
bit_stream = read_le32(ip) >> (bit_count & 31);
|
||||
}
|
||||
|
||||
if remaining != 1 {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
*max_sv = char_num - 1;
|
||||
ip = ip.add(((bit_count + 7) >> 3) as usize);
|
||||
if (ip as usize).wrapping_sub(start) > header_size {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
(ip as usize).wrapping_sub(start)
|
||||
}
|
||||
|
||||
unsafe fn fse_build_dtable_rle(dt: &mut [u32], symbol: u8) -> usize {
|
||||
let header = dt.as_mut_ptr() as *mut FseDTableHeader;
|
||||
let cell = dt.as_mut_ptr().add(1) as *mut FseDecode;
|
||||
(*header).table_log = 0;
|
||||
(*header).fast_mode = 0;
|
||||
(*cell).new_state = 0;
|
||||
(*cell).symbol = symbol;
|
||||
(*cell).nb_bits = 0;
|
||||
0
|
||||
}
|
||||
|
||||
unsafe fn fse_build_dtable_raw(dt: &mut [u32], nb_bits: u32) -> usize {
|
||||
if nb_bits < 1 {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
let header = dt.as_mut_ptr() as *mut FseDTableHeader;
|
||||
let cells = dt.as_mut_ptr().add(1) as *mut FseDecode;
|
||||
let table_size = 1u32 << nb_bits;
|
||||
(*header).table_log = nb_bits as u16;
|
||||
(*header).fast_mode = 1;
|
||||
for symbol in 0..table_size {
|
||||
let cell = cells.add(symbol as usize);
|
||||
(*cell).new_state = 0;
|
||||
(*cell).symbol = symbol as u8;
|
||||
(*cell).nb_bits = nb_bits as u8;
|
||||
}
|
||||
0
|
||||
}
|
||||
|
||||
unsafe fn fse_init_dstate(state: &mut FseDState, stream: &mut DStream, dt: *const u32) {
|
||||
let header = dt as *const FseDTableHeader;
|
||||
state.state = read_bits(stream, (*header).table_log as u32);
|
||||
reload_dstream(stream);
|
||||
state.table = dt.add(1) as *const FseDecode;
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn fse_decode_symbol(state: &mut FseDState, stream: &mut DStream, fast: bool) -> u8 {
|
||||
let info = *state.table.add(state.state);
|
||||
let low_bits = if fast {
|
||||
read_bits_fast(stream, info.nb_bits as u32)
|
||||
} else {
|
||||
read_bits(stream, info.nb_bits as u32)
|
||||
};
|
||||
state.state = (info.new_state as usize).wrapping_add(low_bits);
|
||||
info.symbol
|
||||
}
|
||||
|
||||
unsafe fn fse_decompress_using_dtable(
|
||||
dst: *mut u8,
|
||||
max_dst_size: usize,
|
||||
c_src: *const u8,
|
||||
c_src_size: usize,
|
||||
dt: &[u32],
|
||||
) -> usize {
|
||||
let header = &*(dt.as_ptr() as *const FseDTableHeader);
|
||||
let fast = header.fast_mode != 0;
|
||||
let start = dst;
|
||||
let end_addr = (dst as usize).wrapping_add(max_dst_size);
|
||||
let limit_addr = end_addr.wrapping_sub(3);
|
||||
let mut op = dst;
|
||||
let mut stream = DStream {
|
||||
bit_container: 0,
|
||||
bits_consumed: 0,
|
||||
ptr: ptr::null(),
|
||||
start: ptr::null(),
|
||||
};
|
||||
let mut state1 = FseDState {
|
||||
state: 0,
|
||||
table: ptr::null(),
|
||||
};
|
||||
let mut state2 = state1;
|
||||
let error = init_dstream(&mut stream, c_src, c_src_size);
|
||||
if ERR_isError(error) {
|
||||
return error;
|
||||
}
|
||||
fse_init_dstate(&mut state1, &mut stream, dt.as_ptr());
|
||||
fse_init_dstate(&mut state2, &mut stream, dt.as_ptr());
|
||||
|
||||
const RELOAD_2: bool = FSE_MAX_TABLELOG * 2 + 7 > USIZE_BITS;
|
||||
const RELOAD_4: bool = FSE_MAX_TABLELOG * 4 + 7 > USIZE_BITS;
|
||||
while reload_dstream(&mut stream) == DSTREAM_UNFINISHED && (op as usize) < limit_addr {
|
||||
*op = fse_decode_symbol(&mut state1, &mut stream, fast);
|
||||
if RELOAD_2 {
|
||||
reload_dstream(&mut stream);
|
||||
}
|
||||
*op.add(1) = fse_decode_symbol(&mut state2, &mut stream, fast);
|
||||
if RELOAD_4 && reload_dstream(&mut stream) > DSTREAM_UNFINISHED {
|
||||
op = op.add(2);
|
||||
break;
|
||||
}
|
||||
*op.add(2) = fse_decode_symbol(&mut state1, &mut stream, fast);
|
||||
if RELOAD_2 {
|
||||
reload_dstream(&mut stream);
|
||||
}
|
||||
*op.add(3) = fse_decode_symbol(&mut state2, &mut stream, fast);
|
||||
op = op.add(4);
|
||||
}
|
||||
|
||||
loop {
|
||||
if reload_dstream(&mut stream) > DSTREAM_COMPLETED
|
||||
|| op as usize == end_addr
|
||||
|| (end_of_dstream(&stream) && (fast || state1.state == 0))
|
||||
{
|
||||
break;
|
||||
}
|
||||
*op = fse_decode_symbol(&mut state1, &mut stream, fast);
|
||||
op = op.add(1);
|
||||
if reload_dstream(&mut stream) > DSTREAM_COMPLETED
|
||||
|| op as usize == end_addr
|
||||
|| (end_of_dstream(&stream) && (fast || state2.state == 0))
|
||||
{
|
||||
break;
|
||||
}
|
||||
*op = fse_decode_symbol(&mut state2, &mut stream, fast);
|
||||
op = op.add(1);
|
||||
}
|
||||
|
||||
if end_of_dstream(&stream) && state1.state == 0 && state2.state == 0 {
|
||||
return (op as usize) - (start as usize);
|
||||
}
|
||||
if op as usize == end_addr {
|
||||
return ERROR(ZstdErrorCode::DstSizeTooSmall);
|
||||
}
|
||||
ERROR(ZstdErrorCode::CorruptionDetected)
|
||||
}
|
||||
|
||||
unsafe fn fse_decompress(
|
||||
dst: *mut u8,
|
||||
max_dst_size: usize,
|
||||
c_src: *const u8,
|
||||
c_src_size: usize,
|
||||
) -> usize {
|
||||
if c_src_size < 2 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let mut counters = [0i16; 256];
|
||||
let mut max_symbol = FSE_MAX_SYMBOL_VALUE;
|
||||
let mut table_log = 0;
|
||||
let header_size = fse_read_ncount(
|
||||
&mut counters,
|
||||
&mut max_symbol,
|
||||
&mut table_log,
|
||||
c_src,
|
||||
c_src_size,
|
||||
);
|
||||
if ERR_isError(header_size) {
|
||||
return header_size;
|
||||
}
|
||||
if header_size >= c_src_size {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let mut table = vec![0u32; 1 + (1usize << FSE_MAX_TABLELOG)];
|
||||
let error = fse_build_dtable(&mut table, &counters, max_symbol, table_log);
|
||||
if ERR_isError(error) {
|
||||
return error;
|
||||
}
|
||||
fse_decompress_using_dtable(
|
||||
dst,
|
||||
max_dst_size,
|
||||
c_src.add(header_size),
|
||||
c_src_size - header_size,
|
||||
&table,
|
||||
)
|
||||
}
|
||||
|
||||
/* ******************************************
|
||||
* Huffman decoding
|
||||
********************************************/
|
||||
|
||||
#[repr(C)]
|
||||
#[derive(Clone, Copy)]
|
||||
struct HufDEltX2 {
|
||||
byte: u8,
|
||||
nb_bits: u8,
|
||||
}
|
||||
|
||||
#[allow(clippy::manual_div_ceil)]
|
||||
unsafe fn huf_read_stats(
|
||||
huff_weight: &mut [u8; HUF_MAX_SYMBOL_VALUE + 1],
|
||||
rank_stats: &mut [u32; HUF_ABSOLUTE_MAX_TABLELOG + 1],
|
||||
nb_symbols: &mut u32,
|
||||
table_log: &mut u32,
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
if src_size == 0 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let mut i_size = *src as usize;
|
||||
let o_size: usize;
|
||||
if i_size >= 128 {
|
||||
if i_size >= 242 {
|
||||
const RLE_LENGTHS: [usize; 14] = [1, 2, 3, 4, 7, 8, 15, 16, 31, 32, 63, 64, 127, 128];
|
||||
let index = i_size - 242;
|
||||
if index >= RLE_LENGTHS.len() {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
o_size = RLE_LENGTHS[index];
|
||||
huff_weight.fill(1);
|
||||
i_size = 0;
|
||||
} else {
|
||||
o_size = i_size - 127;
|
||||
i_size = (o_size + 1) / 2;
|
||||
if i_size + 1 > src_size {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
if o_size >= huff_weight.len() {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
let weights = src.add(1);
|
||||
let mut n = 0;
|
||||
while n < o_size {
|
||||
huff_weight[n] = *weights.add(n / 2) >> 4;
|
||||
if n + 1 < huff_weight.len() {
|
||||
huff_weight[n + 1] = *weights.add(n / 2) & 15;
|
||||
}
|
||||
n += 2;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
if i_size + 1 > src_size {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let decoded = fse_decompress(
|
||||
huff_weight.as_mut_ptr(),
|
||||
huff_weight.len() - 1,
|
||||
src.add(1),
|
||||
i_size,
|
||||
);
|
||||
if ERR_isError(decoded) {
|
||||
return decoded;
|
||||
}
|
||||
o_size = decoded;
|
||||
}
|
||||
|
||||
rank_stats.fill(0);
|
||||
let mut weight_total = 0u32;
|
||||
for &weight in huff_weight.iter().take(o_size) {
|
||||
if weight as usize >= HUF_ABSOLUTE_MAX_TABLELOG {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
rank_stats[weight as usize] += 1;
|
||||
weight_total += (1u32 << weight) >> 1;
|
||||
}
|
||||
if weight_total == 0 {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
let calculated_log = highbit32(weight_total) + 1;
|
||||
if calculated_log as usize > HUF_ABSOLUTE_MAX_TABLELOG {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
let total = 1u32 << calculated_log;
|
||||
let rest = total - weight_total;
|
||||
if rest == 0 || (1u32 << highbit32(rest)) != rest {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
let last_weight = highbit32(rest) + 1;
|
||||
huff_weight[o_size] = last_weight as u8;
|
||||
rank_stats[last_weight as usize] += 1;
|
||||
if rank_stats[1] < 2 || (rank_stats[1] & 1) != 0 {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
*nb_symbols = (o_size + 1) as u32;
|
||||
*table_log = calculated_log;
|
||||
i_size + 1
|
||||
}
|
||||
|
||||
#[allow(clippy::needless_range_loop)]
|
||||
unsafe fn huf_read_dtable_x2(
|
||||
dtable: &mut [u16; 1 + (1 << HUF_MAX_TABLELOG)],
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
let mut huff_weight = [0u8; HUF_MAX_SYMBOL_VALUE + 1];
|
||||
let mut rank_val = [0u32; HUF_ABSOLUTE_MAX_TABLELOG + 1];
|
||||
let mut nb_symbols = 0u32;
|
||||
let mut table_log = 0u32;
|
||||
let i_size = huf_read_stats(
|
||||
&mut huff_weight,
|
||||
&mut rank_val,
|
||||
&mut nb_symbols,
|
||||
&mut table_log,
|
||||
src,
|
||||
src_size,
|
||||
);
|
||||
if ERR_isError(i_size) {
|
||||
return i_size;
|
||||
}
|
||||
if table_log as usize > dtable[0] as usize {
|
||||
return ERROR(ZstdErrorCode::TableLogTooLarge);
|
||||
}
|
||||
dtable[0] = table_log as u16;
|
||||
|
||||
let mut next_rank_start = 0u32;
|
||||
for weight in 1..=table_log as usize {
|
||||
let current = next_rank_start;
|
||||
next_rank_start += rank_val[weight] << (weight - 1);
|
||||
rank_val[weight] = current;
|
||||
}
|
||||
let cells = dtable.as_mut_ptr().add(1) as *mut HufDEltX2;
|
||||
for (symbol, &weight) in huff_weight.iter().enumerate().take(nb_symbols as usize) {
|
||||
let weight = weight as usize;
|
||||
let length = (1u32 << weight) >> 1;
|
||||
let entry = HufDEltX2 {
|
||||
byte: symbol as u8,
|
||||
nb_bits: (table_log + 1 - weight as u32) as u8,
|
||||
};
|
||||
for index in rank_val[weight]..rank_val[weight] + length {
|
||||
*cells.add(index as usize) = entry;
|
||||
}
|
||||
rank_val[weight] += length;
|
||||
}
|
||||
i_size
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn huf_decode_symbol(stream: &mut DStream, table: *const HufDEltX2, table_log: u32) -> u8 {
|
||||
let entry = *table.add(look_bits_fast(stream, table_log));
|
||||
skip_bits(stream, entry.nb_bits as u32);
|
||||
entry.byte
|
||||
}
|
||||
|
||||
unsafe fn huf_decode_stream(
|
||||
dst: *mut u8,
|
||||
dst_size: usize,
|
||||
stream: &mut DStream,
|
||||
table: *const HufDEltX2,
|
||||
table_log: u32,
|
||||
) -> usize {
|
||||
let start = dst;
|
||||
let end = dst.add(dst_size);
|
||||
let mut op = dst;
|
||||
|
||||
/* The v0.3 C decoder uses a 4-symbol unrolled loop followed by a tail.
|
||||
* Decoding one symbol at a time has the same state transitions and keeps
|
||||
* the same stop-bit validation while remaining easy to audit. */
|
||||
while op < end {
|
||||
let status = reload_dstream(stream);
|
||||
if status == DSTREAM_TOO_FAR || (status == DSTREAM_COMPLETED && op < end) {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
*op = huf_decode_symbol(stream, table, table_log);
|
||||
op = op.add(1);
|
||||
}
|
||||
if !end_of_dstream(stream) {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
(op as usize) - (start as usize)
|
||||
}
|
||||
|
||||
unsafe fn huf_decompress4x2_using_dtable(
|
||||
dst: *mut u8,
|
||||
dst_size: usize,
|
||||
c_src: *const u8,
|
||||
c_src_size: usize,
|
||||
dtable: &[u16; 1 + (1 << HUF_MAX_TABLELOG)],
|
||||
) -> usize {
|
||||
if c_src_size < 10 {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
let length1 = read_le16(c_src) as usize;
|
||||
let length2 = read_le16(c_src.add(2)) as usize;
|
||||
let length3 = read_le16(c_src.add(4)) as usize;
|
||||
let payload = length1
|
||||
.checked_add(length2)
|
||||
.and_then(|v| v.checked_add(length3))
|
||||
.and_then(|v| v.checked_add(6));
|
||||
let payload = match payload {
|
||||
Some(value) if value <= c_src_size => value,
|
||||
_ => return ERROR(ZstdErrorCode::CorruptionDetected),
|
||||
};
|
||||
let length4 = c_src_size - payload;
|
||||
let stream1 = c_src.add(6);
|
||||
let stream2 = stream1.add(length1);
|
||||
let stream3 = stream2.add(length2);
|
||||
let stream4 = stream3.add(length3);
|
||||
let segment = dst_size.div_ceil(4);
|
||||
let starts = [
|
||||
dst,
|
||||
dst.add(segment),
|
||||
dst.add(segment * 2),
|
||||
dst.add(segment * 3),
|
||||
];
|
||||
let sizes = [
|
||||
segment.min(dst_size),
|
||||
segment.min(dst_size.saturating_sub(segment)),
|
||||
segment.min(dst_size.saturating_sub(segment * 2)),
|
||||
dst_size.saturating_sub(segment * 3),
|
||||
];
|
||||
let lengths = [length1, length2, length3, length4];
|
||||
let sources = [stream1, stream2, stream3, stream4];
|
||||
let table = dtable.as_ptr().add(1) as *const HufDEltX2;
|
||||
let table_log = dtable[0] as u32;
|
||||
for index in 0..4 {
|
||||
let mut stream = DStream {
|
||||
bit_container: 0,
|
||||
bits_consumed: 0,
|
||||
ptr: ptr::null(),
|
||||
start: ptr::null(),
|
||||
};
|
||||
let error = init_dstream(&mut stream, sources[index], lengths[index]);
|
||||
if ERR_isError(error) {
|
||||
return error;
|
||||
}
|
||||
let decoded = huf_decode_stream(starts[index], sizes[index], &mut stream, table, table_log);
|
||||
if ERR_isError(decoded) {
|
||||
return decoded;
|
||||
}
|
||||
}
|
||||
dst_size
|
||||
}
|
||||
|
||||
unsafe fn huf_decompress(
|
||||
dst: *mut u8,
|
||||
dst_size: usize,
|
||||
c_src: *const u8,
|
||||
c_src_size: usize,
|
||||
) -> usize {
|
||||
if dst_size == 0 {
|
||||
return ERROR(ZstdErrorCode::DstSizeTooSmall);
|
||||
}
|
||||
if c_src_size > dst_size {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
if c_src_size == dst_size {
|
||||
ptr::copy_nonoverlapping(c_src, dst, dst_size);
|
||||
return dst_size;
|
||||
}
|
||||
if c_src_size == 1 {
|
||||
ptr::write_bytes(dst, *c_src, dst_size);
|
||||
return dst_size;
|
||||
}
|
||||
let mut dtable = [0u16; 1 + (1 << HUF_MAX_TABLELOG)];
|
||||
dtable[0] = HUF_MAX_TABLELOG as u16;
|
||||
let header_size = huf_read_dtable_x2(&mut dtable, c_src, c_src_size);
|
||||
if ERR_isError(header_size) {
|
||||
return header_size;
|
||||
}
|
||||
if header_size >= c_src_size {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
huf_decompress4x2_using_dtable(
|
||||
dst,
|
||||
dst_size,
|
||||
c_src.add(header_size),
|
||||
c_src_size - header_size,
|
||||
&dtable,
|
||||
)
|
||||
}
|
||||
|
||||
/* ******************************************
|
||||
* v0.3 frame decoder
|
||||
********************************************/
|
||||
|
||||
const LL_TABLE_SIZE: usize = 1 + (1 << LL_FSE_LOG);
|
||||
const OFF_TABLE_SIZE: usize = 1 + (1 << OFF_FSE_LOG);
|
||||
const ML_TABLE_SIZE: usize = 1 + (1 << ML_FSE_LOG);
|
||||
|
||||
#[repr(C)]
|
||||
pub struct ZSTDv03_Dctx {
|
||||
ll_table: [u32; LL_TABLE_SIZE],
|
||||
off_table: [u32; OFF_TABLE_SIZE],
|
||||
ml_table: [u32; ML_TABLE_SIZE],
|
||||
previous_dst_end: *mut u8,
|
||||
base: *mut u8,
|
||||
expected: usize,
|
||||
b_type: u32,
|
||||
phase: u32,
|
||||
lit_ptr: *const u8,
|
||||
lit_size: usize,
|
||||
lit_buffer: [u8; BLOCKSIZE + 8],
|
||||
}
|
||||
|
||||
#[derive(Clone, Copy)]
|
||||
struct BlockProperties {
|
||||
block_type: u32,
|
||||
orig_size: u32,
|
||||
}
|
||||
|
||||
#[inline]
|
||||
unsafe fn get_block_size(
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
properties: &mut BlockProperties,
|
||||
) -> usize {
|
||||
if src_size < 3 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let flags = *src;
|
||||
let c_size =
|
||||
*src.add(2) as usize | ((*src.add(1) as usize) << 8) | (((*src as usize) & 7) << 16);
|
||||
properties.block_type = (flags >> 6) as u32;
|
||||
properties.orig_size = if properties.block_type == BT_RLE {
|
||||
c_size as u32
|
||||
} else {
|
||||
0
|
||||
};
|
||||
if properties.block_type == BT_END {
|
||||
return 0;
|
||||
}
|
||||
if properties.block_type == BT_RLE {
|
||||
return 1;
|
||||
}
|
||||
c_size
|
||||
}
|
||||
|
||||
unsafe fn copy_uncompressed_block(
|
||||
dst: *mut u8,
|
||||
max_dst_size: usize,
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
if src_size > max_dst_size {
|
||||
return ERROR(ZstdErrorCode::DstSizeTooSmall);
|
||||
}
|
||||
if src_size != 0 {
|
||||
ptr::copy_nonoverlapping(src, dst, src_size);
|
||||
}
|
||||
src_size
|
||||
}
|
||||
|
||||
unsafe fn decompress_literals(
|
||||
dst: *mut u8,
|
||||
max_dst_size: &mut usize,
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
let lit_size = ((read_le32(src) & 0x1F_FFFF) >> 2) as usize;
|
||||
let lit_c_size = ((read_le32(src.add(2)) & 0xFF_FFFF) >> 5) as usize;
|
||||
if lit_size > *max_dst_size {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
if lit_c_size + 5 > src_size {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
let decoded = huf_decompress(dst, lit_size, src.add(5), lit_c_size);
|
||||
if ERR_isError(decoded) {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
*max_dst_size = lit_size;
|
||||
lit_c_size + 5
|
||||
}
|
||||
|
||||
unsafe fn decode_literals_block(dctx: &mut ZSTDv03_Dctx, src: *const u8, src_size: usize) -> usize {
|
||||
if src_size < MIN_CBLOCK_SIZE {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
match *src & 3 {
|
||||
0 => {
|
||||
let mut lit_size = BLOCKSIZE;
|
||||
let read_size =
|
||||
decompress_literals(dctx.lit_buffer.as_mut_ptr(), &mut lit_size, src, src_size);
|
||||
dctx.lit_ptr = dctx.lit_buffer.as_ptr();
|
||||
dctx.lit_size = lit_size;
|
||||
if !ERR_isError(read_size) {
|
||||
dctx.lit_buffer[lit_size..lit_size + 8].fill(0);
|
||||
}
|
||||
read_size
|
||||
}
|
||||
1 => {
|
||||
let lit_size = ((read_le32(src) & 0xFF_FFFF) >> 2) as usize;
|
||||
if lit_size > src_size.wrapping_sub(11) {
|
||||
if lit_size > BLOCKSIZE || lit_size > src_size - 3 {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
ptr::copy_nonoverlapping(src, dctx.lit_buffer.as_mut_ptr(), lit_size);
|
||||
dctx.lit_ptr = dctx.lit_buffer.as_ptr();
|
||||
dctx.lit_size = lit_size;
|
||||
dctx.lit_buffer[lit_size..lit_size + 8].fill(0);
|
||||
lit_size + 3
|
||||
} else {
|
||||
dctx.lit_ptr = src.add(3);
|
||||
dctx.lit_size = lit_size;
|
||||
lit_size + 3
|
||||
}
|
||||
}
|
||||
2 => {
|
||||
let lit_size = ((read_le32(src) & 0xFF_FFFF) >> 2) as usize;
|
||||
if lit_size > BLOCKSIZE {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
dctx.lit_buffer[..lit_size + 8].fill(*src.add(3));
|
||||
dctx.lit_ptr = dctx.lit_buffer.as_ptr();
|
||||
dctx.lit_size = lit_size;
|
||||
4
|
||||
}
|
||||
_ => unreachable!(),
|
||||
}
|
||||
}
|
||||
|
||||
unsafe fn decode_seq_headers(
|
||||
dctx: &mut ZSTDv03_Dctx,
|
||||
nb_seq: &mut i32,
|
||||
dumps: &mut *const u8,
|
||||
dumps_length: &mut usize,
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
if src_size < 5 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let start = src as usize;
|
||||
let end = start + src_size;
|
||||
let mut ip = src;
|
||||
*nb_seq = read_le16(ip) as i32;
|
||||
ip = ip.add(2);
|
||||
let types = *ip;
|
||||
let ll_type = (types >> 6) as u32;
|
||||
let off_type = ((types >> 4) & 3) as u32;
|
||||
let ml_type = ((types >> 2) & 3) as u32;
|
||||
let dump_size: usize;
|
||||
if types & 2 != 0 {
|
||||
dump_size = *ip.add(2) as usize | ((*ip.add(1) as usize) << 8);
|
||||
ip = ip.add(3);
|
||||
} else {
|
||||
dump_size = *ip.add(1) as usize | (((*ip as usize) & 1) << 8);
|
||||
ip = ip.add(2);
|
||||
}
|
||||
*dumps = ip;
|
||||
*dumps_length = dump_size;
|
||||
if dump_size > end.wrapping_sub(ip as usize) {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
ip = ip.add(dump_size);
|
||||
if end.wrapping_sub(ip as usize) < 3 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
|
||||
let mut norm = [0i16; 256];
|
||||
match ll_type {
|
||||
BT_RLE => {
|
||||
fse_build_dtable_rle(&mut dctx.ll_table, *ip);
|
||||
ip = ip.add(1);
|
||||
}
|
||||
BT_RAW => {
|
||||
fse_build_dtable_raw(&mut dctx.ll_table, LL_BITS);
|
||||
}
|
||||
_ => {
|
||||
let mut max = MAX_LL;
|
||||
let mut log = 0;
|
||||
let size = fse_read_ncount(&mut norm, &mut max, &mut log, ip, end - ip as usize);
|
||||
if ERR_isError(size) {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
if log > LL_FSE_LOG {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
ip = ip.add(size);
|
||||
let error = fse_build_dtable(&mut dctx.ll_table, &norm, max, log);
|
||||
if ERR_isError(error) {
|
||||
return error;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
match off_type {
|
||||
BT_RLE => {
|
||||
if end.wrapping_sub(ip as usize) < 2 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
fse_build_dtable_rle(&mut dctx.off_table, *ip & MAX_OFF as u8);
|
||||
ip = ip.add(1);
|
||||
}
|
||||
BT_RAW => {
|
||||
fse_build_dtable_raw(&mut dctx.off_table, OFF_BITS);
|
||||
}
|
||||
_ => {
|
||||
let mut max = MAX_OFF;
|
||||
let mut log = 0;
|
||||
let size = fse_read_ncount(&mut norm, &mut max, &mut log, ip, end - ip as usize);
|
||||
if ERR_isError(size) {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
if log > OFF_FSE_LOG {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
ip = ip.add(size);
|
||||
let error = fse_build_dtable(&mut dctx.off_table, &norm, max, log);
|
||||
if ERR_isError(error) {
|
||||
return error;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
match ml_type {
|
||||
BT_RLE => {
|
||||
if end.wrapping_sub(ip as usize) < 2 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
fse_build_dtable_rle(&mut dctx.ml_table, *ip);
|
||||
ip = ip.add(1);
|
||||
}
|
||||
BT_RAW => {
|
||||
fse_build_dtable_raw(&mut dctx.ml_table, ML_BITS);
|
||||
}
|
||||
_ => {
|
||||
let mut max = MAX_ML;
|
||||
let mut log = 0;
|
||||
let size = fse_read_ncount(&mut norm, &mut max, &mut log, ip, end - ip as usize);
|
||||
if ERR_isError(size) {
|
||||
return ERROR(ZstdErrorCode::Generic);
|
||||
}
|
||||
if log > ML_FSE_LOG {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
ip = ip.add(size);
|
||||
let error = fse_build_dtable(&mut dctx.ml_table, &norm, max, log);
|
||||
if ERR_isError(error) {
|
||||
return error;
|
||||
}
|
||||
}
|
||||
}
|
||||
(ip as usize) - start
|
||||
}
|
||||
|
||||
#[derive(Clone, Copy)]
|
||||
struct Sequence {
|
||||
lit_length: usize,
|
||||
offset: usize,
|
||||
match_length: usize,
|
||||
}
|
||||
|
||||
struct SequenceState {
|
||||
stream: DStream,
|
||||
state_ll: FseDState,
|
||||
state_off: FseDState,
|
||||
state_ml: FseDState,
|
||||
prev_offset: usize,
|
||||
dumps: *const u8,
|
||||
dumps_end: *const u8,
|
||||
}
|
||||
|
||||
unsafe fn decode_sequence(sequence: &mut Sequence, state: &mut SequenceState) {
|
||||
let mut dumps = state.dumps;
|
||||
let dumps_end = state.dumps_end;
|
||||
let mut lit_length = fse_decode_symbol(&mut state.state_ll, &mut state.stream, false) as usize;
|
||||
let previous_offset = if lit_length != 0 {
|
||||
sequence.offset
|
||||
} else {
|
||||
state.prev_offset
|
||||
};
|
||||
state.prev_offset = sequence.offset;
|
||||
if lit_length == MAX_LL as usize {
|
||||
let add = if (dumps as usize) < dumps_end as usize {
|
||||
let value = *dumps as usize;
|
||||
dumps = dumps.add(1);
|
||||
value
|
||||
} else {
|
||||
0
|
||||
};
|
||||
if add < 255 {
|
||||
lit_length += add;
|
||||
} else if (dumps as usize) <= (dumps_end as usize).wrapping_sub(3) {
|
||||
lit_length = read_le24(dumps) as usize;
|
||||
dumps = dumps.add(3);
|
||||
}
|
||||
}
|
||||
|
||||
let offset_code = fse_decode_symbol(&mut state.state_off, &mut state.stream, false) as u32;
|
||||
let nb_bits = offset_code.saturating_sub(1);
|
||||
let offset_prefix = [
|
||||
1usize, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384, 32768, 65536,
|
||||
131072, 262144, 524288, 1048576, 2097152, 4194304, 8388608, 16777216, 33554432, 1, 1, 1, 1,
|
||||
1,
|
||||
];
|
||||
let mut offset =
|
||||
offset_prefix[offset_code as usize].wrapping_add(read_bits(&mut state.stream, nb_bits));
|
||||
if offset_code == 0 {
|
||||
offset = previous_offset;
|
||||
}
|
||||
|
||||
let mut match_length =
|
||||
fse_decode_symbol(&mut state.state_ml, &mut state.stream, false) as usize;
|
||||
if match_length == MAX_ML as usize {
|
||||
let add = if (dumps as usize) < dumps_end as usize {
|
||||
let value = *dumps as usize;
|
||||
dumps = dumps.add(1);
|
||||
value
|
||||
} else {
|
||||
0
|
||||
};
|
||||
if add < 255 {
|
||||
match_length += add;
|
||||
} else if (dumps as usize) <= (dumps_end as usize).wrapping_sub(3) {
|
||||
match_length = read_le24(dumps) as usize;
|
||||
dumps = dumps.add(3);
|
||||
}
|
||||
}
|
||||
sequence.lit_length = lit_length;
|
||||
sequence.offset = offset;
|
||||
sequence.match_length = match_length + MINMATCH;
|
||||
state.dumps = dumps;
|
||||
}
|
||||
|
||||
unsafe fn exec_sequence(
|
||||
op: *mut u8,
|
||||
sequence: Sequence,
|
||||
lit_ptr: &mut *const u8,
|
||||
lit_limit: *const u8,
|
||||
base: *mut u8,
|
||||
oend: *mut u8,
|
||||
) -> usize {
|
||||
let ostart = op as usize;
|
||||
let o_lit_end = (op as usize).wrapping_add(sequence.lit_length);
|
||||
let o_match_end = o_lit_end.wrapping_add(sequence.match_length);
|
||||
let oend_addr = oend as usize;
|
||||
let lit_end = (*lit_ptr as usize).wrapping_add(sequence.lit_length);
|
||||
let sequence_length = sequence.lit_length.wrapping_add(sequence.match_length);
|
||||
|
||||
if sequence_length > oend_addr.wrapping_sub(op as usize) {
|
||||
return ERROR(ZstdErrorCode::DstSizeTooSmall);
|
||||
}
|
||||
if sequence.lit_length > (lit_limit as usize).wrapping_sub(*lit_ptr as usize) {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
if o_lit_end > oend_addr.wrapping_sub(8) {
|
||||
return ERROR(ZstdErrorCode::DstSizeTooSmall);
|
||||
}
|
||||
if sequence.offset > ((o_lit_end.wrapping_sub(base as usize)) as u32) as usize {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
if o_match_end > oend_addr {
|
||||
return ERROR(ZstdErrorCode::DstSizeTooSmall);
|
||||
}
|
||||
if lit_end > lit_limit as usize {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
|
||||
if sequence.lit_length != 0 {
|
||||
ptr::copy(*lit_ptr, op, sequence.lit_length);
|
||||
}
|
||||
*lit_ptr = lit_end as *const u8;
|
||||
let match_start = o_lit_end as *mut u8;
|
||||
let offset = sequence.offset;
|
||||
if offset == 0 || offset > o_lit_end.wrapping_sub(base as usize) {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
for index in 0..sequence.match_length {
|
||||
let source = if index >= offset {
|
||||
match_start.add(index - offset)
|
||||
} else {
|
||||
match_start.sub(offset - index)
|
||||
};
|
||||
*match_start.add(index) = *source;
|
||||
}
|
||||
o_match_end - ostart
|
||||
}
|
||||
|
||||
unsafe fn decompress_sequences(
|
||||
dctx: &mut ZSTDv03_Dctx,
|
||||
dst: *mut u8,
|
||||
max_dst_size: usize,
|
||||
seq_start: *const u8,
|
||||
seq_size: usize,
|
||||
) -> usize {
|
||||
let mut ip = seq_start;
|
||||
let seq_end = (seq_start as usize).wrapping_add(seq_size);
|
||||
let ostart = dst;
|
||||
let mut op = dst;
|
||||
let oend = dst.add(max_dst_size);
|
||||
let mut nb_seq = 0i32;
|
||||
let mut dumps = ptr::null();
|
||||
let mut dumps_length = 0usize;
|
||||
let header_size = decode_seq_headers(
|
||||
dctx,
|
||||
&mut nb_seq,
|
||||
&mut dumps,
|
||||
&mut dumps_length,
|
||||
ip,
|
||||
seq_size,
|
||||
);
|
||||
if ERR_isError(header_size) {
|
||||
return header_size;
|
||||
}
|
||||
ip = ip.add(header_size);
|
||||
let mut state = SequenceState {
|
||||
stream: DStream {
|
||||
bit_container: 0,
|
||||
bits_consumed: 0,
|
||||
ptr: ptr::null(),
|
||||
start: ptr::null(),
|
||||
},
|
||||
state_ll: FseDState {
|
||||
state: 0,
|
||||
table: ptr::null(),
|
||||
},
|
||||
state_off: FseDState {
|
||||
state: 0,
|
||||
table: ptr::null(),
|
||||
},
|
||||
state_ml: FseDState {
|
||||
state: 0,
|
||||
table: ptr::null(),
|
||||
},
|
||||
prev_offset: 4,
|
||||
dumps,
|
||||
dumps_end: dumps.add(dumps_length),
|
||||
};
|
||||
let lit_limit = dctx.lit_ptr.add(dctx.lit_size);
|
||||
let error = init_dstream(&mut state.stream, ip, seq_end - ip as usize);
|
||||
if ERR_isError(error) {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
fse_init_dstate(
|
||||
&mut state.state_ll,
|
||||
&mut state.stream,
|
||||
dctx.ll_table.as_ptr(),
|
||||
);
|
||||
fse_init_dstate(
|
||||
&mut state.state_off,
|
||||
&mut state.stream,
|
||||
dctx.off_table.as_ptr(),
|
||||
);
|
||||
fse_init_dstate(
|
||||
&mut state.state_ml,
|
||||
&mut state.stream,
|
||||
dctx.ml_table.as_ptr(),
|
||||
);
|
||||
|
||||
let mut sequence = Sequence {
|
||||
lit_length: 0,
|
||||
offset: 4,
|
||||
match_length: 0,
|
||||
};
|
||||
while nb_seq > 0 && reload_dstream(&mut state.stream) <= DSTREAM_COMPLETED {
|
||||
nb_seq -= 1;
|
||||
decode_sequence(&mut sequence, &mut state);
|
||||
let produced = exec_sequence(op, sequence, &mut dctx.lit_ptr, lit_limit, dctx.base, oend);
|
||||
if ERR_isError(produced) {
|
||||
return produced;
|
||||
}
|
||||
op = op.add(produced);
|
||||
}
|
||||
|
||||
if !end_of_dstream(&state.stream) || nb_seq < 0 {
|
||||
return ERROR(ZstdErrorCode::CorruptionDetected);
|
||||
}
|
||||
let last_literal_size = (lit_limit as usize).wrapping_sub(dctx.lit_ptr as usize);
|
||||
if op as usize + last_literal_size > oend as usize {
|
||||
return ERROR(ZstdErrorCode::DstSizeTooSmall);
|
||||
}
|
||||
if last_literal_size != 0 {
|
||||
if !ptr::eq(op, dctx.lit_ptr) {
|
||||
ptr::copy(dctx.lit_ptr, op, last_literal_size);
|
||||
}
|
||||
op = op.add(last_literal_size);
|
||||
}
|
||||
(op as usize) - (ostart as usize)
|
||||
}
|
||||
|
||||
unsafe fn decompress_block(
|
||||
dctx: &mut ZSTDv03_Dctx,
|
||||
dst: *mut u8,
|
||||
max_dst_size: usize,
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
let literals_size = decode_literals_block(dctx, src, src_size);
|
||||
if ERR_isError(literals_size) {
|
||||
return literals_size;
|
||||
}
|
||||
decompress_sequences(
|
||||
dctx,
|
||||
dst,
|
||||
max_dst_size,
|
||||
src.add(literals_size),
|
||||
src_size - literals_size,
|
||||
)
|
||||
}
|
||||
|
||||
unsafe fn decompress_dctx(
|
||||
dctx: &mut ZSTDv03_Dctx,
|
||||
dst: *mut u8,
|
||||
max_dst_size: usize,
|
||||
src: *const u8,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
if src_size < 4 + 3 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
if read_le32(src) != ZSTD_MAGIC_NUMBER {
|
||||
return ERROR(ZstdErrorCode::PrefixUnknown);
|
||||
}
|
||||
let mut ip = src.add(4);
|
||||
let mut remaining = src_size - 4;
|
||||
let ostart = dst;
|
||||
let mut op = dst;
|
||||
let oend = dst.add(max_dst_size);
|
||||
loop {
|
||||
let mut properties = BlockProperties {
|
||||
block_type: BT_END,
|
||||
orig_size: 0,
|
||||
};
|
||||
let block_size = get_block_size(ip, remaining, &mut properties);
|
||||
if ERR_isError(block_size) {
|
||||
return block_size;
|
||||
}
|
||||
ip = ip.add(3);
|
||||
remaining -= 3;
|
||||
if block_size > remaining {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
let decoded = match properties.block_type {
|
||||
BT_COMPRESSED => {
|
||||
decompress_block(dctx, op, oend as usize - op as usize, ip, block_size)
|
||||
}
|
||||
BT_RAW => copy_uncompressed_block(op, oend as usize - op as usize, ip, block_size),
|
||||
BT_RLE => return ERROR(ZstdErrorCode::Generic),
|
||||
BT_END => {
|
||||
if remaining != 0 {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
0
|
||||
}
|
||||
_ => return ERROR(ZstdErrorCode::Generic),
|
||||
};
|
||||
if block_size == 0 {
|
||||
break;
|
||||
}
|
||||
if ERR_isError(decoded) {
|
||||
return decoded;
|
||||
}
|
||||
op = op.add(decoded);
|
||||
ip = ip.add(block_size);
|
||||
remaining -= block_size;
|
||||
}
|
||||
(op as usize) - (ostart as usize)
|
||||
}
|
||||
|
||||
unsafe fn error_frame_size_info(c_size: *mut usize, d_bound: *mut u64, ret: usize) {
|
||||
*c_size = ret;
|
||||
*d_bound = ZSTD_CONTENTSIZE_ERROR;
|
||||
}
|
||||
|
||||
/* ******************************************
|
||||
* Exported C ABI
|
||||
********************************************/
|
||||
|
||||
#[no_mangle]
|
||||
pub extern "C" fn ZSTDv03_isError(code: usize) -> c_uint {
|
||||
ERR_isError(code) as c_uint
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_decompress(
|
||||
dst: *mut c_void,
|
||||
max_original_size: usize,
|
||||
src: *const c_void,
|
||||
compressed_size: usize,
|
||||
) -> usize {
|
||||
let mut dctx = std::mem::MaybeUninit::<ZSTDv03_Dctx>::zeroed();
|
||||
let dctx_ptr = dctx.as_mut_ptr();
|
||||
(*dctx_ptr).base = dst as *mut u8;
|
||||
decompress_dctx(
|
||||
&mut *dctx_ptr,
|
||||
dst as *mut u8,
|
||||
max_original_size,
|
||||
src as *const u8,
|
||||
compressed_size,
|
||||
)
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_decompressDCtx(
|
||||
ctx: *mut c_void,
|
||||
dst: *mut c_void,
|
||||
max_original_size: usize,
|
||||
src: *const c_void,
|
||||
compressed_size: usize,
|
||||
) -> usize {
|
||||
decompress_dctx(
|
||||
&mut *(ctx as *mut ZSTDv03_Dctx),
|
||||
dst as *mut u8,
|
||||
max_original_size,
|
||||
src as *const u8,
|
||||
compressed_size,
|
||||
)
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_findFrameSizeInfoLegacy(
|
||||
src: *const c_void,
|
||||
src_size: usize,
|
||||
c_size: *mut usize,
|
||||
d_bound: *mut u64,
|
||||
) {
|
||||
if src_size < 4 + 3 {
|
||||
error_frame_size_info(c_size, d_bound, ERROR(ZstdErrorCode::SrcSizeWrong));
|
||||
return;
|
||||
}
|
||||
if read_le32(src as *const u8) != ZSTD_MAGIC_NUMBER {
|
||||
error_frame_size_info(c_size, d_bound, ERROR(ZstdErrorCode::PrefixUnknown));
|
||||
return;
|
||||
}
|
||||
let start = src as *const u8;
|
||||
let mut ip = start.add(4);
|
||||
let mut remaining = src_size - 4;
|
||||
let mut nb_blocks = 0usize;
|
||||
loop {
|
||||
let mut properties = BlockProperties {
|
||||
block_type: BT_END,
|
||||
orig_size: 0,
|
||||
};
|
||||
let block_size = get_block_size(ip, remaining, &mut properties);
|
||||
if ERR_isError(block_size) {
|
||||
error_frame_size_info(c_size, d_bound, block_size);
|
||||
return;
|
||||
}
|
||||
ip = ip.add(3);
|
||||
remaining -= 3;
|
||||
if block_size > remaining {
|
||||
error_frame_size_info(c_size, d_bound, ERROR(ZstdErrorCode::SrcSizeWrong));
|
||||
return;
|
||||
}
|
||||
if block_size == 0 {
|
||||
break;
|
||||
}
|
||||
ip = ip.add(block_size);
|
||||
remaining -= block_size;
|
||||
nb_blocks += 1;
|
||||
}
|
||||
*c_size = ip as usize - start as usize;
|
||||
*d_bound = (nb_blocks * BLOCKSIZE) as u64;
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_resetDCtx(dctx: *mut ZSTDv03_Dctx) -> usize {
|
||||
(*dctx).expected = 4;
|
||||
(*dctx).phase = 0;
|
||||
(*dctx).previous_dst_end = ptr::null_mut();
|
||||
(*dctx).base = ptr::null_mut();
|
||||
0
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_createDCtx() -> *mut ZSTDv03_Dctx {
|
||||
let dctx = libc::malloc(std::mem::size_of::<ZSTDv03_Dctx>()) as *mut ZSTDv03_Dctx;
|
||||
if dctx.is_null() {
|
||||
return ptr::null_mut();
|
||||
}
|
||||
ZSTDv03_resetDCtx(dctx);
|
||||
dctx
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_freeDCtx(dctx: *mut ZSTDv03_Dctx) -> usize {
|
||||
libc::free(dctx as *mut c_void);
|
||||
0
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_nextSrcSizeToDecompress(dctx: *mut ZSTDv03_Dctx) -> usize {
|
||||
(*dctx).expected
|
||||
}
|
||||
|
||||
#[no_mangle]
|
||||
pub unsafe extern "C" fn ZSTDv03_decompressContinue(
|
||||
dctx: *mut ZSTDv03_Dctx,
|
||||
dst: *mut c_void,
|
||||
max_dst_size: usize,
|
||||
src: *const c_void,
|
||||
src_size: usize,
|
||||
) -> usize {
|
||||
let ctx = &mut *dctx;
|
||||
let dst = dst as *mut u8;
|
||||
let src = src as *const u8;
|
||||
if src_size != ctx.expected {
|
||||
return ERROR(ZstdErrorCode::SrcSizeWrong);
|
||||
}
|
||||
if dst != ctx.previous_dst_end {
|
||||
ctx.base = dst;
|
||||
}
|
||||
if ctx.phase == 0 {
|
||||
if read_le32(src) != ZSTD_MAGIC_NUMBER {
|
||||
return ERROR(ZstdErrorCode::PrefixUnknown);
|
||||
}
|
||||
ctx.phase = 1;
|
||||
ctx.expected = 3;
|
||||
return 0;
|
||||
}
|
||||
if ctx.phase == 1 {
|
||||
let mut properties = BlockProperties {
|
||||
block_type: BT_END,
|
||||
orig_size: 0,
|
||||
};
|
||||
let block_size = get_block_size(src, 3, &mut properties);
|
||||
if ERR_isError(block_size) {
|
||||
return block_size;
|
||||
}
|
||||
if properties.block_type == BT_END {
|
||||
ctx.expected = 0;
|
||||
ctx.phase = 0;
|
||||
} else {
|
||||
ctx.expected = block_size;
|
||||
ctx.b_type = properties.block_type;
|
||||
ctx.phase = 2;
|
||||
}
|
||||
return 0;
|
||||
}
|
||||
|
||||
let produced = match ctx.b_type {
|
||||
BT_COMPRESSED => decompress_block(ctx, dst, max_dst_size, src, src_size),
|
||||
BT_RAW => copy_uncompressed_block(dst, max_dst_size, src, src_size),
|
||||
BT_RLE => return ERROR(ZstdErrorCode::Generic),
|
||||
BT_END => 0,
|
||||
_ => return ERROR(ZstdErrorCode::Generic),
|
||||
};
|
||||
ctx.phase = 1;
|
||||
ctx.expected = 3;
|
||||
if ERR_isError(produced) {
|
||||
return produced;
|
||||
}
|
||||
ctx.previous_dst_end = dst.add(produced);
|
||||
produced
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
|
||||
const RAW_FRAME: &[u8] = &[
|
||||
0x23, 0xB5, 0x2F, 0xFD, 0x40, 0x00, 0x0B, b'r', b'a', b'w', b' ', b'v', b'0', b'.', b'3',
|
||||
b'!', b'!', b'!', 0xC0, 0x00, 0x00,
|
||||
];
|
||||
|
||||
const COMPRESSED_FRAME: &[u8] = &[
|
||||
0x23, 0xB5, 0x2F, 0xFD, 0x00, 0x00, 0x13, 0x1D, 0x00, 0x00, 0x68, 0x65, 0x6C, 0x6C, 0x6F,
|
||||
0x20, 0x0A, 0x01, 0x00, 0x54, 0x00, 0x00, 0x01, 0x30, 0x19, 0x46, 0xC0, 0x00, 0x00,
|
||||
];
|
||||
|
||||
const MULTI_BLOCK_FRAME: &[u8] = &[
|
||||
0x23, 0xB5, 0x2F, 0xFD, 0x00, 0x00, 0x42, 0xC5, 0x00, 0x00, 0x54, 0x68, 0x65, 0x20, 0x71,
|
||||
0x75, 0x69, 0x63, 0x6B, 0x20, 0x62, 0x72, 0x6F, 0x77, 0x6E, 0x20, 0x66, 0x6F, 0x78, 0x20,
|
||||
0x6A, 0x75, 0x6D, 0x70, 0x73, 0x20, 0x6F, 0x76, 0x65, 0x72, 0x20, 0x74, 0x68, 0x65, 0x20,
|
||||
0x6C, 0x61, 0x7A, 0x79, 0x20, 0x64, 0x6F, 0x67, 0x2E, 0x20, 0x65, 0x72, 0x20, 0x74, 0x01,
|
||||
0x00, 0x54, 0x04, 0xFF, 0xCB, 0xFF, 0x01, 0x80, 0x06, 0x80, 0xBF, 0x69, 0x03, 0x00, 0x00,
|
||||
0x13, 0x05, 0x00, 0x00, 0x20, 0x01, 0x00, 0x54, 0x04, 0xFF, 0xE3, 0x6E, 0x01, 0x00, 0xF0,
|
||||
0x7F, 0x00, 0xFC, 0x23, 0x10, 0xC0, 0x00, 0x00,
|
||||
];
|
||||
|
||||
unsafe fn decode(frame: &[u8], capacity: usize) -> Result<Vec<u8>, usize> {
|
||||
let mut output = vec![0u8; capacity];
|
||||
let size = ZSTDv03_decompress(
|
||||
output.as_mut_ptr() as *mut c_void,
|
||||
output.len(),
|
||||
frame.as_ptr() as *const c_void,
|
||||
frame.len(),
|
||||
);
|
||||
if ZSTDv03_isError(size) != 0 {
|
||||
Err(size)
|
||||
} else {
|
||||
output.truncate(size);
|
||||
Ok(output)
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn decodes_raw_and_compressed_v03_frames() {
|
||||
unsafe {
|
||||
assert_eq!(decode(RAW_FRAME, 32).unwrap(), b"raw v0.3!!!");
|
||||
assert_eq!(
|
||||
decode(COMPRESSED_FRAME, 64).unwrap(),
|
||||
b"hello hello hello hello hello\n"
|
||||
);
|
||||
let expected = b"The quick brown fox jumps over the lazy dog. ".repeat(5000);
|
||||
assert_eq!(decode(MULTI_BLOCK_FRAME, expected.len()).unwrap(), expected);
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn reports_frame_size_and_errors() {
|
||||
unsafe {
|
||||
let mut c_size = 0;
|
||||
let mut bound = 0;
|
||||
ZSTDv03_findFrameSizeInfoLegacy(
|
||||
RAW_FRAME.as_ptr() as *const c_void,
|
||||
RAW_FRAME.len(),
|
||||
&mut c_size,
|
||||
&mut bound,
|
||||
);
|
||||
assert_eq!(c_size, RAW_FRAME.len());
|
||||
assert_eq!(bound, BLOCKSIZE as u64);
|
||||
|
||||
let mut bad = RAW_FRAME.to_vec();
|
||||
bad[0] ^= 1;
|
||||
assert_eq!(
|
||||
decode(&bad, 32).unwrap_err(),
|
||||
ERROR(ZstdErrorCode::PrefixUnknown)
|
||||
);
|
||||
assert_eq!(
|
||||
decode(&RAW_FRAME[..RAW_FRAME.len() - 1], 32).unwrap_err(),
|
||||
ERROR(ZstdErrorCode::SrcSizeWrong)
|
||||
);
|
||||
assert_eq!(
|
||||
decode(COMPRESSED_FRAME, 1).unwrap_err(),
|
||||
ERROR(ZstdErrorCode::DstSizeTooSmall)
|
||||
);
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn streaming_raw_frame_matches_one_shot() {
|
||||
let expected = b"raw v0.3!!!";
|
||||
let mut output = vec![0u8; expected.len()];
|
||||
let mut input_offset = 0;
|
||||
let mut output_offset = 0;
|
||||
unsafe {
|
||||
let dctx = ZSTDv03_createDCtx();
|
||||
assert!(!dctx.is_null());
|
||||
loop {
|
||||
let needed = ZSTDv03_nextSrcSizeToDecompress(dctx);
|
||||
if needed == 0 {
|
||||
break;
|
||||
}
|
||||
assert!(input_offset + needed <= RAW_FRAME.len());
|
||||
let produced = ZSTDv03_decompressContinue(
|
||||
dctx,
|
||||
output.as_mut_ptr().add(output_offset) as *mut c_void,
|
||||
output.len() - output_offset,
|
||||
RAW_FRAME.as_ptr().add(input_offset) as *const c_void,
|
||||
needed,
|
||||
);
|
||||
assert_eq!(ZSTDv03_isError(produced), 0);
|
||||
input_offset += needed;
|
||||
output_offset += produced;
|
||||
}
|
||||
assert_eq!(ZSTDv03_freeDCtx(dctx), 0);
|
||||
}
|
||||
assert_eq!(input_offset, RAW_FRAME.len());
|
||||
assert_eq!(output_offset, expected.len());
|
||||
assert_eq!(output, expected);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn streaming_context_reset_and_null_free() {
|
||||
unsafe {
|
||||
let dctx = ZSTDv03_createDCtx();
|
||||
assert!(!dctx.is_null());
|
||||
assert_eq!(ZSTDv03_nextSrcSizeToDecompress(dctx), 4);
|
||||
|
||||
let bad_magic = [0u8; 4];
|
||||
assert_eq!(
|
||||
ZSTDv03_decompressContinue(
|
||||
dctx,
|
||||
ptr::null_mut(),
|
||||
0,
|
||||
bad_magic.as_ptr() as *const c_void,
|
||||
bad_magic.len(),
|
||||
),
|
||||
ERROR(ZstdErrorCode::PrefixUnknown)
|
||||
);
|
||||
assert_eq!(
|
||||
ZSTDv03_decompressContinue(
|
||||
dctx,
|
||||
ptr::null_mut(),
|
||||
0,
|
||||
bad_magic.as_ptr() as *const c_void,
|
||||
bad_magic.len() - 1,
|
||||
),
|
||||
ERROR(ZstdErrorCode::SrcSizeWrong)
|
||||
);
|
||||
assert_eq!(ZSTDv03_resetDCtx(dctx), 0);
|
||||
assert_eq!(ZSTDv03_nextSrcSizeToDecompress(dctx), 4);
|
||||
assert_eq!(ZSTDv03_freeDCtx(dctx), 0);
|
||||
assert_eq!(ZSTDv03_freeDCtx(ptr::null_mut()), 0);
|
||||
}
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user