feat(legacy): port the v0.4 decoder to Rust

Move the frozen v0.4 frame, entropy, streaming, and dictionary decoder
implementation to Rust while retaining the C translation unit as an ABI
anchor. Register the decoder behind the matching legacy feature and keep its
historical behavior isolated from newer formats.

Test Plan:
- rustfmt +nightly --check --edition 2021 rust/src/legacy/zstd_v04.rs rust/src/legacy/mod.rs
- RUSTC_WRAPPER= CARGO_BUILD_RUSTC_WRAPPER= cargo test --manifest-path rust/Cargo.toml --no-default-features --features decompression,legacy-v04 legacy::zstd_v04
- RUSTC_WRAPPER= CARGO_BUILD_RUSTC_WRAPPER= cargo clippy --manifest-path rust/Cargo.toml --all-targets --no-default-features --features decompression,legacy-v04 -- -D warnings
- C/Rust v0.4 ABI and decompression build checks
- git diff --cached --check
This commit is contained in:
2026-07-12 10:43:32 +02:00
parent b8ccfd7b87
commit 393b2bec45
3 changed files with 2906 additions and 3585 deletions
+3
View File
@@ -44,3 +44,6 @@ pub mod zstd_v02;
#[cfg(feature = "legacy-v03")]
pub mod zstd_v03;
#[cfg(feature = "legacy-v04")]
pub mod zstd_v04;
+2897
View File
@@ -0,0 +1,2897 @@
#![allow(non_snake_case)]
//! Frozen decoder for the zstd v0.4 format.
//!
//! `lib/legacy/zstd_v04.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 `ZSTDv04_Dctx`.
use crate::errors::{ERR_isError, ZstdErrorCode, ERROR};
use std::os::raw::{c_char, c_uint, c_void};
use std::ptr;
const ZSTD_MAGIC_NUMBER: u32 = 0xFD2F_B524;
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 IS_RAW: u8 = 1;
const IS_RLE: u8 = 2;
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 write_le16(dst: *mut u8, value: u16) {
let bytes = value.to_le_bytes();
ptr::copy_nonoverlapping(bytes.as_ptr(), dst, 2);
}
#[inline]
unsafe fn zstd_copy8(dst: *mut u8, src: *const u8) {
ptr::copy(src, dst, 8);
}
#[inline]
unsafe fn zstd_copy4(dst: *mut u8, src: *const u8) {
ptr::copy(src, dst, 4);
}
unsafe fn zstd_wildcopy(dst: *mut u8, src: *const u8, length: isize) {
let mut op = dst;
let mut ip = src;
let end = if length >= 0 {
(dst as usize).wrapping_add(length as usize)
} else {
(dst as usize).wrapping_sub(length.wrapping_neg() as usize)
};
loop {
zstd_copy8(op, ip);
op = op.add(8);
ip = ip.add(8);
if (op as usize) >= end {
break;
}
}
}
#[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 v0.4 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 v0.4 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
}
#[repr(C)]
#[derive(Clone, Copy)]
struct HufDEltX4 {
sequence: u16,
nb_bits: u8,
length: u8,
}
#[repr(C)]
#[derive(Clone, Copy)]
struct SortedSymbol {
symbol: u8,
weight: u8,
}
type RankVal = [[u32; HUF_ABSOLUTE_MAX_TABLELOG + 1]; HUF_ABSOLUTE_MAX_TABLELOG];
#[inline]
unsafe fn huf_set_dtable_x4(
dtable: *mut HufDEltX4,
index: usize,
sequence: u16,
nb_bits: u32,
length: u8,
) {
let entry = dtable.add(index);
write_le16(entry.cast::<u8>(), sequence);
(*entry).nb_bits = nb_bits as u8;
(*entry).length = length;
}
#[allow(clippy::too_many_arguments)]
unsafe fn huf_fill_dtable_x4_level2(
dtable: *mut HufDEltX4,
size_log: u32,
consumed: u32,
rank_val_origin: &[u32; HUF_ABSOLUTE_MAX_TABLELOG + 1],
min_weight: usize,
sorted_symbols: &[SortedSymbol; HUF_MAX_SYMBOL_VALUE + 1],
sorted_start: usize,
sorted_list_size: usize,
nb_bits_baseline: u32,
base_seq: u16,
) {
let mut rank_val = *rank_val_origin;
if min_weight > 1 {
let skip_size = rank_val[min_weight] as usize;
for index in 0..skip_size {
huf_set_dtable_x4(dtable, index, base_seq, consumed, 1);
}
}
for item in 0..sorted_list_size {
let sorted = sorted_symbols[sorted_start + item];
let symbol = sorted.symbol as u32;
let weight = sorted.weight as usize;
let nb_bits = nb_bits_baseline - weight as u32;
let length = 1u32 << (size_log - nb_bits);
let start = rank_val[weight] as usize;
let end = start + length as usize;
let sequence = base_seq.wrapping_add((symbol << 8) as u16);
for index in start..end {
huf_set_dtable_x4(dtable, index, sequence, nb_bits + consumed, 2);
}
rank_val[weight] = rank_val[weight].wrapping_add(length);
}
}
#[allow(clippy::too_many_arguments)]
unsafe fn huf_fill_dtable_x4(
dtable: *mut HufDEltX4,
target_log: u32,
sorted_list: &[SortedSymbol; HUF_MAX_SYMBOL_VALUE + 1],
sorted_list_size: usize,
rank_start0: &[u32; HUF_ABSOLUTE_MAX_TABLELOG + 2],
rank_val_origin: &RankVal,
max_weight: u32,
nb_bits_baseline: u32,
) {
let mut rank_val = rank_val_origin[0];
let scale_log = nb_bits_baseline as i32 - target_log as i32;
let min_bits = nb_bits_baseline - max_weight;
for item in 0..sorted_list_size {
let sorted = sorted_list[item];
let symbol = sorted.symbol as u16;
let weight = sorted.weight as usize;
let nb_bits = nb_bits_baseline - weight as u32;
let start = rank_val[weight] as usize;
let length = 1u32 << (target_log - nb_bits);
if target_log - nb_bits >= min_bits {
let mut min_weight = nb_bits as i32 + scale_log;
if min_weight < 1 {
min_weight = 1;
}
let min_weight = min_weight as usize;
let sorted_rank = rank_start0[min_weight + 1] as usize;
huf_fill_dtable_x4_level2(
dtable.add(start),
target_log - nb_bits,
nb_bits,
&rank_val_origin[nb_bits as usize],
min_weight,
sorted_list,
sorted_rank,
sorted_list_size - sorted_rank,
nb_bits_baseline,
symbol,
);
} else {
let end = start + length as usize;
for index in start..end {
huf_set_dtable_x4(dtable, index, symbol, nb_bits, 1);
}
}
rank_val[weight] = rank_val[weight].wrapping_add(length);
}
}
#[allow(clippy::needless_range_loop)]
unsafe fn huf_read_dtable_x4(
dtable: &mut [u32; 1 + (1 << HUF_MAX_TABLELOG)],
src: *const u8,
src_size: usize,
) -> usize {
let mut weight_list = [0u8; HUF_MAX_SYMBOL_VALUE + 1];
let mut sorted_symbols = [SortedSymbol {
symbol: 0,
weight: 0,
}; HUF_MAX_SYMBOL_VALUE + 1];
let mut rank_stats = [0u32; HUF_ABSOLUTE_MAX_TABLELOG + 1];
let mut rank_start0 = [0u32; HUF_ABSOLUTE_MAX_TABLELOG + 2];
let mut rank_val = [[0u32; HUF_ABSOLUTE_MAX_TABLELOG + 1]; HUF_ABSOLUTE_MAX_TABLELOG];
let mut nb_symbols = 0u32;
let mut table_log = 0u32;
let mem_log = dtable[0];
if mem_log > HUF_ABSOLUTE_MAX_TABLELOG as u32 {
return ERROR(ZstdErrorCode::TableLogTooLarge);
}
let i_size = huf_read_stats(
&mut weight_list,
&mut rank_stats,
&mut nb_symbols,
&mut table_log,
src,
src_size,
);
if ERR_isError(i_size) {
return i_size;
}
if table_log > mem_log {
return ERROR(ZstdErrorCode::TableLogTooLarge);
}
let mut max_weight = table_log as usize;
loop {
if rank_stats[max_weight] != 0 {
break;
}
if max_weight == 0 {
return ERROR(ZstdErrorCode::Generic);
}
max_weight -= 1;
}
let mut next_rank_start = 0u32;
for weight in 1..=max_weight {
let current = next_rank_start;
next_rank_start = next_rank_start.wrapping_add(rank_stats[weight]);
rank_start0[weight + 1] = current;
}
rank_start0[0] = next_rank_start;
let size_of_sort = next_rank_start as usize;
for symbol in 0..nb_symbols as usize {
let weight = weight_list[symbol] as usize;
let rank = rank_start0[weight + 1] as usize;
sorted_symbols[rank] = SortedSymbol {
symbol: symbol as u8,
weight: weight as u8,
};
rank_start0[weight + 1] = rank as u32 + 1;
}
rank_start0[1] = 0;
let min_bits = table_log + 1 - max_weight as u32;
let rescale = (mem_log as i32 - table_log as i32) - 1;
let mut next_rank_val = 0u32;
for weight in 1..=max_weight {
let current = next_rank_val;
let shift = (weight as i32 + rescale) as u32;
next_rank_val = next_rank_val.wrapping_add(rank_stats[weight] << shift);
rank_val[0][weight] = current;
}
if min_bits <= mem_log.saturating_sub(min_bits) {
for consumed in min_bits..=mem_log - min_bits {
for weight in 1..=max_weight {
rank_val[consumed as usize][weight] = rank_val[0][weight] >> consumed;
}
}
}
let table = dtable.as_mut_ptr().add(1) as *mut HufDEltX4;
huf_fill_dtable_x4(
table,
mem_log,
&sorted_symbols,
size_of_sort,
&rank_start0,
&rank_val,
max_weight as u32,
table_log + 1,
);
i_size
}
#[inline]
unsafe fn huf_decode_symbol_x4(
op: &mut *mut u8,
stream: &mut DStream,
dtable: *const HufDEltX4,
table_log: u32,
) -> u32 {
let value = look_bits_fast(stream, table_log);
let entry = dtable.add(value);
ptr::copy(entry.cast::<u8>(), *op, 2);
skip_bits(stream, (*entry).nb_bits as u32);
let length = (*entry).length as u32;
*op = (*op).add(length as usize);
length
}
#[inline]
unsafe fn huf_decode_last_symbol_x4(
op: *mut u8,
stream: &mut DStream,
dtable: *const HufDEltX4,
table_log: u32,
) -> u32 {
let value = look_bits_fast(stream, table_log);
let entry = dtable.add(value);
*op = read_le16(entry.cast::<u8>()) as u8;
if (*entry).length == 1 {
skip_bits(stream, (*entry).nb_bits as u32);
} else if stream.bits_consumed < USIZE_BITS {
skip_bits(stream, (*entry).nb_bits as u32);
if stream.bits_consumed > USIZE_BITS {
stream.bits_consumed = USIZE_BITS;
}
}
1
}
#[inline]
unsafe fn huf_decode_symbol_x4_0(
op: &mut *mut u8,
stream: &mut DStream,
dtable: *const HufDEltX4,
table_log: u32,
) {
huf_decode_symbol_x4(op, stream, dtable, table_log);
}
#[inline]
unsafe fn huf_decode_symbol_x4_1(
op: &mut *mut u8,
stream: &mut DStream,
dtable: *const HufDEltX4,
table_log: u32,
) {
if USIZE_BITS == 64 || HUF_MAX_TABLELOG <= 12 {
huf_decode_symbol_x4(op, stream, dtable, table_log);
}
}
#[inline]
unsafe fn huf_decode_symbol_x4_2(
op: &mut *mut u8,
stream: &mut DStream,
dtable: *const HufDEltX4,
table_log: u32,
) {
if USIZE_BITS == 64 {
huf_decode_symbol_x4(op, stream, dtable, table_log);
}
}
unsafe fn huf_decode_stream_x4(
mut p: *mut u8,
stream: &mut DStream,
p_end: *mut u8,
dtable: *const HufDEltX4,
table_log: u32,
) -> usize {
let p_start = p;
while reload_dstream(stream) == DSTREAM_UNFINISHED
&& (p as usize) < (p_end as usize).wrapping_sub(7)
{
huf_decode_symbol_x4_2(&mut p, stream, dtable, table_log);
huf_decode_symbol_x4_1(&mut p, stream, dtable, table_log);
huf_decode_symbol_x4_2(&mut p, stream, dtable, table_log);
huf_decode_symbol_x4_0(&mut p, stream, dtable, table_log);
}
while reload_dstream(stream) == DSTREAM_UNFINISHED
&& (p as usize) <= (p_end as usize).wrapping_sub(2)
{
huf_decode_symbol_x4_0(&mut p, stream, dtable, table_log);
}
while (p as usize) <= (p_end as usize).wrapping_sub(2) {
huf_decode_symbol_x4_0(&mut p, stream, dtable, table_log);
}
if (p as usize) < p_end as usize {
p = p.add(huf_decode_last_symbol_x4(p, stream, dtable, table_log) as usize);
}
(p as usize).wrapping_sub(p_start as usize)
}
unsafe fn huf_decompress4x4_using_dtable(
dst: *mut u8,
dst_size: usize,
c_src: *const u8,
c_src_size: usize,
dtable: &[u32; 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 total = length1
.wrapping_add(length2)
.wrapping_add(length3)
.wrapping_add(6);
let length4 = c_src_size.wrapping_sub(total);
if length4 > c_src_size {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
let streams = [
c_src.add(6),
c_src.add(6 + length1),
c_src.add(6 + length1 + length2),
c_src.add(6 + length1 + length2 + length3),
];
let lengths = [length1, length2, length3, length4];
let segment = dst_size.wrapping_add(3) / 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 table = dtable.as_ptr().add(1) as *const HufDEltX4;
let table_log = dtable[0];
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, streams[index], lengths[index]);
if ERR_isError(error) {
return error;
}
let decoded = huf_decode_stream_x4(
starts[index],
&mut stream,
starts[index].add(sizes[index]),
table,
table_log,
);
if decoded != sizes[index] {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
}
dst_size
}
unsafe fn huf_decompress4x4(
dst: *mut u8,
dst_size: usize,
c_src: *const u8,
c_src_size: usize,
) -> usize {
let mut dtable = [0u32; 1 + (1 << HUF_MAX_TABLELOG)];
dtable[0] = HUF_MAX_TABLELOG as u32;
let header_size = huf_read_dtable_x4(&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_decompress4x4_using_dtable(
dst,
dst_size,
c_src.add(header_size),
c_src_size - header_size,
&dtable,
)
}
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(c_src, dst, dst_size);
return dst_size;
}
if c_src_size == 1 {
ptr::write_bytes(dst, *c_src, dst_size);
return dst_size;
}
const ALGO_TIME: [[[u32; 2]; 3]; 16] = [
[[0, 0], [1, 1], [2, 2]],
[[0, 0], [1, 1], [2, 2]],
[[38, 130], [1313, 74], [2151, 38]],
[[448, 128], [1353, 74], [2238, 41]],
[[556, 128], [1353, 74], [2238, 47]],
[[714, 128], [1418, 74], [2436, 53]],
[[883, 128], [1437, 74], [2464, 61]],
[[897, 128], [1515, 75], [2622, 68]],
[[926, 128], [1613, 75], [2730, 75]],
[[947, 128], [1729, 77], [3359, 77]],
[[1107, 128], [2083, 81], [4006, 84]],
[[1177, 128], [2379, 87], [4785, 88]],
[[1242, 128], [2415, 93], [5155, 84]],
[[1349, 128], [2644, 106], [5260, 106]],
[[1455, 128], [2422, 124], [4174, 124]],
[[722, 128], [1891, 145], [1936, 146]],
];
let q = c_src_size.wrapping_mul(16) / dst_size;
let d256 = (dst_size >> 8) as u32;
let mut dtime = [0u32; 3];
for index in 0..3 {
dtime[index] =
ALGO_TIME[q][index][0].wrapping_add(ALGO_TIME[q][index][1].wrapping_mul(d256));
}
dtime[1] = dtime[1].wrapping_add(dtime[1] >> 4);
dtime[2] = dtime[2].wrapping_add(dtime[2] >> 3);
if dtime[1] < dtime[0] {
huf_decompress4x4(dst, dst_size, c_src, c_src_size)
} else {
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,
)
}
}
/* ******************************************
* v0.4 frame decoder
********************************************/
const FRAME_HEADER_SIZE_MIN: usize = 5;
const FRAME_HEADER_SIZE_MAX: usize = 5;
const BLOCK_HEADER_SIZE: usize = 3;
const WINDOWLOG_ABSOLUTE_MIN: u32 = 11;
const STAGE_GET_FRAME_HEADER_SIZE: u32 = 0;
const STAGE_DECODE_FRAME_HEADER: u32 = 1;
const STAGE_DECODE_BLOCK_HEADER: u32 = 2;
const STAGE_DECOMPRESS_BLOCK: u32 = 3;
#[repr(C)]
#[derive(Clone, Copy)]
struct ZstdParameters {
src_size: u64,
window_log: u32,
content_log: u32,
hash_log: u32,
search_log: u32,
search_length: u32,
strategy: u32,
}
#[repr(C)]
pub struct ZSTDv04_Dctx {
ll_table: [u32; 1 + (1 << LL_FSE_LOG)],
off_table: [u32; 1 + (1 << OFF_FSE_LOG)],
ml_table: [u32; 1 + (1 << ML_FSE_LOG)],
previous_dst_end: *const u8,
base: *const u8,
v_base: *const u8,
dict_end: *const u8,
expected: usize,
header_size: usize,
params: ZstdParameters,
b_type: u32,
stage: u32,
lit_ptr: *const u8,
lit_size: usize,
lit_buffer: [u8; BLOCKSIZE + 8],
header_buffer: [u8; FRAME_HEADER_SIZE_MAX],
}
#[derive(Clone, Copy)]
struct BlockProperties {
block_type: u32,
orig_size: u32,
}
#[inline]
unsafe fn reset_dctx(dctx: &mut ZSTDv04_Dctx) -> usize {
dctx.expected = FRAME_HEADER_SIZE_MIN;
dctx.stage = STAGE_GET_FRAME_HEADER_SIZE;
dctx.previous_dst_end = ptr::null();
dctx.base = ptr::null();
dctx.v_base = ptr::null();
dctx.dict_end = ptr::null();
0
}
#[inline]
unsafe fn create_dctx() -> *mut ZSTDv04_Dctx {
let dctx = libc::malloc(std::mem::size_of::<ZSTDv04_Dctx>()) as *mut ZSTDv04_Dctx;
if dctx.is_null() {
return ptr::null_mut();
}
reset_dctx(&mut *dctx);
dctx
}
#[inline]
unsafe fn free_dctx(dctx: *mut ZSTDv04_Dctx) -> usize {
libc::free(dctx.cast::<c_void>());
0
}
unsafe fn decode_frame_header_part1(
dctx: &mut ZSTDv04_Dctx,
src: *const u8,
src_size: usize,
) -> usize {
if src_size != FRAME_HEADER_SIZE_MIN {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
if read_le32(src) != ZSTD_MAGIC_NUMBER {
return ERROR(ZstdErrorCode::PrefixUnknown);
}
dctx.header_size = FRAME_HEADER_SIZE_MIN;
dctx.header_size
}
unsafe fn get_frame_params(params: &mut ZstdParameters, src: *const u8, src_size: usize) -> usize {
if src_size < FRAME_HEADER_SIZE_MIN {
return FRAME_HEADER_SIZE_MAX;
}
if read_le32(src) != ZSTD_MAGIC_NUMBER {
return ERROR(ZstdErrorCode::PrefixUnknown);
}
*params = ZstdParameters {
src_size: 0,
window_log: 0,
content_log: 0,
hash_log: 0,
search_log: 0,
search_length: 0,
strategy: 0,
};
let descriptor = *src.add(4);
params.window_log = (descriptor & 15) as u32 + WINDOWLOG_ABSOLUTE_MIN;
if descriptor >> 4 != 0 {
return ERROR(ZstdErrorCode::FrameParameterUnsupported);
}
0
}
unsafe fn decode_frame_header_part2(
dctx: &mut ZSTDv04_Dctx,
src: *const u8,
src_size: usize,
) -> usize {
if src_size != dctx.header_size {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
let result = get_frame_params(&mut dctx.params, src, src_size);
if USIZE_BITS == 32 && dctx.params.window_log > 25 {
return ERROR(ZstdErrorCode::FrameParameterUnsupported);
}
result
}
unsafe fn get_block_size(
src: *const u8,
src_size: usize,
properties: &mut BlockProperties,
) -> usize {
if src_size < BLOCK_HEADER_SIZE {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
let header_flags = *src;
let c_size = *src.add(2) as usize
| ((*src.add(1) as usize) << 8)
| (((header_flags as usize) & 7) << 16);
properties.block_type = (header_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_raw_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(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.wrapping_add(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.wrapping_add(5)
}
unsafe fn decode_literals_block(dctx: &mut ZSTDv04_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;
let offset = lit_size.min(BLOCKSIZE);
if offset + 8 <= dctx.lit_buffer.len() {
dctx.lit_buffer[offset..offset + 8].fill(0);
}
read_size
}
IS_RAW => {
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(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);
return lit_size + 3;
}
dctx.lit_ptr = src.add(3);
dctx.lit_size = lit_size;
lit_size + 3
}
IS_RLE => {
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
}
_ => ERROR(ZstdErrorCode::CorruptionDetected),
}
}
unsafe fn decode_seq_headers(
dctx: &mut ZSTDv04_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.wrapping_add(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;
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 | ((types as usize & 1) << 8);
ip = ip.add(2);
}
*dumps = ip;
ip = ip.add(dump_size);
*dumps_length = dump_size;
if (ip as usize) > end.wrapping_sub(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.wrapping_sub(ip as usize),
);
if ERR_isError(size) {
return ERROR(ZstdErrorCode::Generic);
}
if log > LL_FSE_LOG {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
ip = ip.add(size);
fse_build_dtable(&mut dctx.ll_table, &norm, max, log);
}
}
match off_type {
BT_RLE => {
if (ip as usize) > end.wrapping_sub(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.wrapping_sub(ip as usize),
);
if ERR_isError(size) {
return ERROR(ZstdErrorCode::Generic);
}
if log > OFF_FSE_LOG {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
ip = ip.add(size);
fse_build_dtable(&mut dctx.off_table, &norm, max, log);
}
}
match ml_type {
BT_RLE => {
if (ip as usize) > end.wrapping_sub(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.wrapping_sub(ip as usize),
);
if ERR_isError(size) {
return ERROR(ZstdErrorCode::Generic);
}
if log > ML_FSE_LOG {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
ip = ip.add(size);
fse_build_dtable(&mut dctx.ml_table, &norm, max, log);
}
}
(ip as usize).wrapping_sub(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 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
};
let mut lit_length = lit_length;
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);
}
if (dumps as usize) >= dumps_end as usize {
dumps = dumps_end.wrapping_sub(1);
}
}
let offset_code = fse_decode_symbol(&mut state.state_off, &mut state.stream, false) as u32;
if USIZE_BITS == 32 {
reload_dstream(&mut state.stream);
}
let nb_bits = offset_code.wrapping_sub(1);
let nb_bits = if offset_code == 0 { 0 } else { nb_bits };
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 USIZE_BITS == 32 {
reload_dstream(&mut state.stream);
}
if offset_code == 0 {
offset = previous_offset;
}
if offset_code != 0 || lit_length == 0 {
state.prev_offset = sequence.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);
}
if (dumps as usize) >= dumps_end as usize {
dumps = dumps_end.wrapping_sub(1);
}
}
sequence.lit_length = lit_length;
sequence.offset = offset;
sequence.match_length = match_length + MINMATCH;
state.dumps = dumps;
}
#[allow(clippy::too_many_arguments)]
unsafe fn exec_sequence(
mut op: *mut u8,
mut sequence: Sequence,
lit_ptr: &mut *const u8,
lit_limit: *const u8,
base: *const u8,
v_base: *const u8,
dict_end: *const u8,
oend: *mut u8,
) -> usize {
let o_lit_end = (op as usize).wrapping_add(sequence.lit_length);
let sequence_length = sequence.lit_length.wrapping_add(sequence.match_length);
let o_match_end = o_lit_end.wrapping_add(sequence.match_length);
let oend_addr = oend as usize;
let oend_8 = oend_addr.wrapping_sub(8);
let lit_end = (*lit_ptr as usize).wrapping_add(sequence.lit_length);
let mut match_addr = o_lit_end.wrapping_sub(sequence.offset);
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_8 {
return ERROR(ZstdErrorCode::DstSizeTooSmall);
}
if o_match_end > oend_addr {
return ERROR(ZstdErrorCode::DstSizeTooSmall);
}
if lit_end > lit_limit as usize {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
zstd_wildcopy(op, *lit_ptr, sequence.lit_length as isize);
op = o_lit_end as *mut u8;
*lit_ptr = lit_end as *const u8;
if sequence.offset > o_lit_end.wrapping_sub(base as usize) {
if sequence.offset > o_lit_end.wrapping_sub(v_base as usize) {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
match_addr = (dict_end as usize).wrapping_sub((base as usize).wrapping_sub(match_addr));
if match_addr.wrapping_add(sequence.match_length) <= dict_end as usize {
ptr::copy(
match_addr as *const u8,
o_lit_end as *mut u8,
sequence.match_length,
);
return sequence_length;
}
let length1 = (dict_end as usize).wrapping_sub(match_addr);
ptr::copy(match_addr as *const u8, o_lit_end as *mut u8, length1);
op = o_lit_end.wrapping_add(length1) as *mut u8;
sequence.match_length = sequence.match_length.wrapping_sub(length1);
match_addr = base as usize;
if (op as usize) > oend_8 || sequence.match_length < MINMATCH {
let mut out = op;
let mut m = match_addr as *const u8;
while (out as usize) < o_match_end {
*out = *m;
out = out.add(1);
m = m.add(1);
}
return sequence_length;
}
}
if sequence.offset < 8 {
let dec32 = [0usize, 1, 2, 1, 4, 4, 4, 4];
let dec64 = [8usize, 8, 8, 7, 8, 9, 10, 11];
*op = *(match_addr as *const u8);
*op.add(1) = *(match_addr as *const u8).add(1);
*op.add(2) = *(match_addr as *const u8).add(2);
*op.add(3) = *(match_addr as *const u8).add(3);
let match_ptr = (match_addr as *const u8).add(dec32[sequence.offset]);
zstd_copy4(op.add(4), match_ptr);
match_addr = match_addr
.wrapping_add(8)
.wrapping_sub(dec64[sequence.offset]);
} else {
zstd_copy8(op, match_addr as *const u8);
match_addr = match_addr.wrapping_add(8);
}
op = op.add(8);
match_addr = match_addr.wrapping_add(8);
if o_match_end > oend_addr.wrapping_sub(16 - MINMATCH) {
if (op as usize) < oend_8 {
let dist = oend_8 - op as usize;
zstd_wildcopy(op, match_addr as *const u8, dist as isize);
match_addr = match_addr.wrapping_add(dist);
op = oend_8 as *mut u8;
}
while (op as usize) < o_match_end {
*op = *(match_addr as *const u8);
op = op.add(1);
match_addr = match_addr.wrapping_add(1);
}
} else {
zstd_wildcopy(
op,
match_addr as *const u8,
sequence.match_length.wrapping_sub(8) as isize,
);
}
sequence_length
}
unsafe fn decompress_sequences(
dctx: &mut ZSTDv04_Dctx,
dst: *mut u8,
max_dst_size: usize,
seq_start: *const u8,
seq_size: usize,
) -> usize {
let ip = seq_start;
let iend = (seq_start as usize).wrapping_add(seq_size);
let ostart = dst;
let mut op = dst;
let oend = (dst as usize).wrapping_add(max_dst_size) as *mut u8;
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;
}
let 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, iend.wrapping_sub(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 reload_dstream(&mut state.stream) <= DSTREAM_COMPLETED && nb_seq != 0 {
nb_seq -= 1;
decode_sequence(&mut sequence, &mut state);
let produced = exec_sequence(
op,
sequence,
&mut dctx.lit_ptr,
lit_limit,
dctx.base,
dctx.v_base,
dctx.dict_end,
oend,
);
if ERR_isError(produced) {
return produced;
}
op = op.add(produced);
}
if !end_of_dstream(&state.stream) {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
let last_literal_size = (lit_limit as usize).wrapping_sub(dctx.lit_ptr as usize);
if dctx.lit_ptr as usize > lit_limit as usize {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
if (op as usize).wrapping_add(last_literal_size) > oend as usize {
return ERROR(ZstdErrorCode::DstSizeTooSmall);
}
if last_literal_size != 0 {
if !ptr::eq(op.cast_const(), dctx.lit_ptr) {
ptr::copy(dctx.lit_ptr, op, last_literal_size);
}
op = op.add(last_literal_size);
}
(op as usize).wrapping_sub(ostart as usize)
}
unsafe fn check_continuity(dctx: &mut ZSTDv04_Dctx, dst: *const u8) {
if dst != dctx.previous_dst_end {
dctx.dict_end = dctx.previous_dst_end;
let delta = (dctx.previous_dst_end as usize).wrapping_sub(dctx.base as usize);
dctx.v_base = (dst as usize).wrapping_sub(delta) as *const u8;
dctx.base = dst;
dctx.previous_dst_end = dst;
}
}
unsafe fn decompress_block_internal(
dctx: &mut ZSTDv04_Dctx,
dst: *mut u8,
max_dst_size: usize,
src: *const u8,
src_size: usize,
) -> usize {
if src_size > BLOCKSIZE {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
let literal_size = decode_literals_block(dctx, src, src_size);
if ERR_isError(literal_size) {
return literal_size;
}
decompress_sequences(
dctx,
dst,
max_dst_size,
src.add(literal_size),
src_size - literal_size,
)
}
unsafe fn decompress_insert_dictionary(dctx: &mut ZSTDv04_Dctx, dict: *const u8, dict_size: usize) {
dctx.dict_end = dctx.previous_dst_end;
let delta = (dctx.previous_dst_end as usize).wrapping_sub(dctx.base as usize);
dctx.v_base = (dict as usize).wrapping_sub(delta) as *const u8;
dctx.base = dict;
dctx.previous_dst_end = (dict as usize).wrapping_add(dict_size) as *const u8;
}
unsafe fn decompress_using_dict(
ctx: &mut ZSTDv04_Dctx,
dst: *mut u8,
max_dst_size: usize,
src: *const u8,
src_size: usize,
dict: *const u8,
dict_size: usize,
) -> usize {
let mut ip = src;
let ostart = dst;
let mut op = dst;
let oend_addr = (dst as usize).wrapping_add(max_dst_size);
let mut remaining_size = src_size;
let mut block_properties = BlockProperties {
block_type: BT_END,
orig_size: 0,
};
reset_dctx(ctx);
if !dict.is_null() {
decompress_insert_dictionary(ctx, dict, dict_size);
ctx.dict_end = ctx.previous_dst_end;
let delta = (ctx.previous_dst_end as usize).wrapping_sub(ctx.base as usize);
ctx.v_base = (dst as usize).wrapping_sub(delta) as *const u8;
ctx.base = dst;
} else {
ctx.v_base = dst;
ctx.base = dst;
ctx.dict_end = dst;
}
if src_size < FRAME_HEADER_SIZE_MIN + BLOCK_HEADER_SIZE {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
let header_size = decode_frame_header_part1(ctx, src, FRAME_HEADER_SIZE_MIN);
if ERR_isError(header_size) {
return header_size;
}
if src_size < header_size + BLOCK_HEADER_SIZE {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
ip = ip.add(header_size);
remaining_size -= header_size;
let header_result = decode_frame_header_part2(ctx, src, header_size);
if ERR_isError(header_result) {
return header_result;
}
loop {
let block_size = get_block_size(ip, remaining_size, &mut block_properties);
if ERR_isError(block_size) {
return block_size;
}
ip = ip.add(BLOCK_HEADER_SIZE);
remaining_size -= BLOCK_HEADER_SIZE;
if block_size > remaining_size {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
let decoded_size = match block_properties.block_type {
BT_COMPRESSED => decompress_block_internal(
ctx,
op,
oend_addr.wrapping_sub(op as usize),
ip,
block_size,
),
BT_RAW => copy_raw_block(op, oend_addr.wrapping_sub(op as usize), ip, block_size),
BT_RLE => return ERROR(ZstdErrorCode::Generic),
BT_END => {
if remaining_size != 0 {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
0
}
_ => return ERROR(ZstdErrorCode::Generic),
};
if block_size == 0 {
break;
}
if ERR_isError(decoded_size) {
return decoded_size;
}
op = op.add(decoded_size);
ip = ip.add(block_size);
remaining_size -= block_size;
}
(op as usize).wrapping_sub(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;
}
unsafe fn find_frame_size_info(
src: *const u8,
src_size: usize,
c_size: *mut usize,
d_bound: *mut u64,
) {
let mut ip = src;
let mut remaining_size = src_size;
let mut nb_blocks = 0usize;
let mut block_properties = BlockProperties {
block_type: BT_END,
orig_size: 0,
};
if src_size < FRAME_HEADER_SIZE_MIN {
error_frame_size_info(c_size, d_bound, ERROR(ZstdErrorCode::SrcSizeWrong));
return;
}
if read_le32(src) != ZSTD_MAGIC_NUMBER {
error_frame_size_info(c_size, d_bound, ERROR(ZstdErrorCode::PrefixUnknown));
return;
}
ip = ip.add(FRAME_HEADER_SIZE_MIN);
remaining_size -= FRAME_HEADER_SIZE_MIN;
loop {
let block_size = get_block_size(ip, remaining_size, &mut block_properties);
if ERR_isError(block_size) {
error_frame_size_info(c_size, d_bound, block_size);
return;
}
ip = ip.add(BLOCK_HEADER_SIZE);
remaining_size -= BLOCK_HEADER_SIZE;
if block_size > remaining_size {
error_frame_size_info(c_size, d_bound, ERROR(ZstdErrorCode::SrcSizeWrong));
return;
}
if block_size == 0 {
break;
}
ip = ip.add(block_size);
remaining_size -= block_size;
nb_blocks = nb_blocks.wrapping_add(1);
}
*c_size = (ip as usize).wrapping_sub(src as usize);
*d_bound = (nb_blocks.wrapping_mul(BLOCKSIZE)) as u64;
}
unsafe fn next_src_size_to_decompress(dctx: &ZSTDv04_Dctx) -> usize {
dctx.expected
}
unsafe fn decompress_continue(
ctx: &mut ZSTDv04_Dctx,
dst: *mut u8,
max_dst_size: usize,
src: *const u8,
src_size: usize,
) -> usize {
if src_size != ctx.expected {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
check_continuity(ctx, dst);
match ctx.stage {
STAGE_GET_FRAME_HEADER_SIZE => {
if src_size != FRAME_HEADER_SIZE_MIN {
return ERROR(ZstdErrorCode::SrcSizeWrong);
}
ctx.header_size = decode_frame_header_part1(ctx, src, FRAME_HEADER_SIZE_MIN);
if ERR_isError(ctx.header_size) {
return ctx.header_size;
}
ptr::copy_nonoverlapping(src, ctx.header_buffer.as_mut_ptr(), FRAME_HEADER_SIZE_MIN);
if ctx.header_size > FRAME_HEADER_SIZE_MIN {
return ERROR(ZstdErrorCode::Generic);
}
ctx.expected = 0;
let result =
decode_frame_header_part2(ctx, ctx.header_buffer.as_ptr(), ctx.header_size);
if ERR_isError(result) {
return result;
}
ctx.expected = BLOCK_HEADER_SIZE;
ctx.stage = STAGE_DECODE_BLOCK_HEADER;
0
}
STAGE_DECODE_FRAME_HEADER => {
let result =
decode_frame_header_part2(ctx, ctx.header_buffer.as_ptr(), ctx.header_size);
if ERR_isError(result) {
return result;
}
ctx.expected = BLOCK_HEADER_SIZE;
ctx.stage = STAGE_DECODE_BLOCK_HEADER;
0
}
STAGE_DECODE_BLOCK_HEADER => {
let mut properties = BlockProperties {
block_type: BT_END,
orig_size: 0,
};
let block_size = get_block_size(src, BLOCK_HEADER_SIZE, &mut properties);
if ERR_isError(block_size) {
return block_size;
}
if properties.block_type == BT_END {
ctx.expected = 0;
ctx.stage = STAGE_GET_FRAME_HEADER_SIZE;
} else {
ctx.expected = block_size;
ctx.b_type = properties.block_type;
ctx.stage = STAGE_DECOMPRESS_BLOCK;
}
0
}
STAGE_DECOMPRESS_BLOCK => {
let result = match ctx.b_type {
BT_COMPRESSED => decompress_block_internal(ctx, dst, max_dst_size, src, src_size),
BT_RAW => copy_raw_block(dst, max_dst_size, src, src_size),
BT_RLE => return ERROR(ZstdErrorCode::Generic),
BT_END => 0,
_ => return ERROR(ZstdErrorCode::Generic),
};
ctx.stage = STAGE_DECODE_BLOCK_HEADER;
ctx.expected = BLOCK_HEADER_SIZE;
if ERR_isError(result) {
return result;
}
ctx.previous_dst_end = (dst as usize).wrapping_add(result) as *const u8;
result
}
_ => ERROR(ZstdErrorCode::Generic),
}
}
#[repr(C)]
pub struct ZBUFFv04_DCtx {
zc: *mut ZSTDv04_Dctx,
params: ZstdParameters,
in_buff: *mut u8,
in_buff_size: usize,
in_pos: usize,
out_buff: *mut u8,
out_buff_size: usize,
out_start: usize,
out_end: usize,
h_pos: usize,
dict: *const u8,
dict_size: usize,
stage: u32,
header_buffer: [u8; FRAME_HEADER_SIZE_MAX],
}
const BUFF_INIT: u32 = 0;
const BUFF_READ_HEADER: u32 = 1;
const BUFF_LOAD_HEADER: u32 = 2;
const BUFF_DECODE_HEADER: u32 = 3;
const BUFF_READ: u32 = 4;
const BUFF_LOAD: u32 = 5;
const BUFF_FLUSH: u32 = 6;
unsafe fn buff_create_dctx() -> *mut ZBUFFv04_DCtx {
let zbc = libc::malloc(std::mem::size_of::<ZBUFFv04_DCtx>()) as *mut ZBUFFv04_DCtx;
if zbc.is_null() {
return ptr::null_mut();
}
ptr::write_bytes(zbc.cast::<u8>(), 0, std::mem::size_of::<ZBUFFv04_DCtx>());
(*zbc).zc = create_dctx();
(*zbc).stage = BUFF_INIT;
zbc
}
unsafe fn buff_free_dctx(zbc: *mut ZBUFFv04_DCtx) -> usize {
if zbc.is_null() {
return 0;
}
free_dctx((*zbc).zc);
libc::free((*zbc).in_buff.cast::<c_void>());
libc::free((*zbc).out_buff.cast::<c_void>());
libc::free(zbc.cast::<c_void>());
0
}
unsafe fn buff_decompress_init(zbc: &mut ZBUFFv04_DCtx) -> usize {
zbc.stage = BUFF_READ_HEADER;
zbc.h_pos = 0;
zbc.in_pos = 0;
zbc.out_start = 0;
zbc.out_end = 0;
zbc.dict_size = 0;
reset_dctx(&mut *zbc.zc)
}
unsafe fn buff_decompress_with_dictionary(
zbc: &mut ZBUFFv04_DCtx,
dict: *const u8,
dict_size: usize,
) -> usize {
zbc.dict = dict;
zbc.dict_size = dict_size;
0
}
unsafe fn buff_limit_copy(
dst: *mut u8,
max_dst_size: usize,
src: *const u8,
src_size: usize,
) -> usize {
let length = max_dst_size.min(src_size);
if length != 0 {
ptr::copy(src, dst, length);
}
length
}
unsafe fn buff_decompress_continue(
zbc: &mut ZBUFFv04_DCtx,
dst: *mut u8,
max_dst_size_ptr: *mut usize,
src: *const u8,
src_size_ptr: *mut usize,
) -> usize {
let istart = src as usize;
let mut ip = istart;
let iend = istart.wrapping_add(*src_size_ptr);
let ostart = dst as usize;
let mut op = ostart;
let oend = ostart.wrapping_add(*max_dst_size_ptr);
let mut not_done = true;
while not_done {
match zbc.stage {
BUFF_INIT => return ERROR(ZstdErrorCode::InitMissing),
BUFF_READ_HEADER => {
let header_size = get_frame_params(&mut zbc.params, src, *src_size_ptr);
if ERR_isError(header_size) {
return header_size;
}
if header_size != 0 {
ptr::copy(
src,
zbc.header_buffer.as_mut_ptr().add(zbc.h_pos),
*src_size_ptr,
);
zbc.h_pos += *src_size_ptr;
*max_dst_size_ptr = 0;
zbc.stage = BUFF_LOAD_HEADER;
return header_size - zbc.h_pos;
}
zbc.stage = BUFF_DECODE_HEADER;
}
BUFF_LOAD_HEADER => {
let header_size = buff_limit_copy(
zbc.header_buffer.as_mut_ptr().add(zbc.h_pos),
FRAME_HEADER_SIZE_MAX - zbc.h_pos,
src,
*src_size_ptr,
);
zbc.h_pos += header_size;
ip = ip.wrapping_add(header_size);
let header_size =
get_frame_params(&mut zbc.params, zbc.header_buffer.as_ptr(), zbc.h_pos);
if ERR_isError(header_size) {
return header_size;
}
if header_size != 0 {
*max_dst_size_ptr = 0;
return header_size - zbc.h_pos;
}
zbc.stage = BUFF_DECODE_HEADER;
}
BUFF_DECODE_HEADER => {
let needed_out_size = 1usize << zbc.params.window_log;
let needed_in_size = BLOCKSIZE;
if zbc.in_buff_size < needed_in_size {
libc::free(zbc.in_buff.cast::<c_void>());
zbc.in_buff_size = needed_in_size;
zbc.in_buff = libc::malloc(needed_in_size) as *mut u8;
if zbc.in_buff.is_null() {
return ERROR(ZstdErrorCode::MemoryAllocation);
}
}
if zbc.out_buff_size < needed_out_size {
libc::free(zbc.out_buff.cast::<c_void>());
zbc.out_buff_size = needed_out_size;
zbc.out_buff = libc::malloc(needed_out_size) as *mut u8;
if zbc.out_buff.is_null() {
return ERROR(ZstdErrorCode::MemoryAllocation);
}
}
if zbc.dict_size != 0 {
decompress_insert_dictionary(&mut *zbc.zc, zbc.dict, zbc.dict_size);
}
if zbc.h_pos != 0 {
ptr::copy(zbc.header_buffer.as_ptr(), zbc.in_buff, zbc.h_pos);
zbc.in_pos = zbc.h_pos;
zbc.h_pos = 0;
zbc.stage = BUFF_LOAD;
} else {
zbc.stage = BUFF_READ;
}
}
BUFF_READ => {
let needed_in_size = next_src_size_to_decompress(&*zbc.zc);
if needed_in_size == 0 {
zbc.stage = BUFF_INIT;
not_done = false;
continue;
}
if iend.wrapping_sub(ip) >= needed_in_size {
let decoded_size = decompress_continue(
&mut *zbc.zc,
zbc.out_buff.add(zbc.out_start),
zbc.out_buff_size - zbc.out_start,
ip as *const u8,
needed_in_size,
);
if ERR_isError(decoded_size) {
return decoded_size;
}
ip = ip.wrapping_add(needed_in_size);
if decoded_size == 0 {
continue;
}
zbc.out_end = zbc.out_start + decoded_size;
zbc.stage = BUFF_FLUSH;
continue;
}
if ip == iend {
not_done = false;
continue;
}
zbc.stage = BUFF_LOAD;
}
BUFF_LOAD => {
let needed_in_size = next_src_size_to_decompress(&*zbc.zc);
let to_load = needed_in_size.wrapping_sub(zbc.in_pos);
if to_load > zbc.in_buff_size.wrapping_sub(zbc.in_pos) {
return ERROR(ZstdErrorCode::CorruptionDetected);
}
let loaded_size = buff_limit_copy(
zbc.in_buff.add(zbc.in_pos),
to_load,
ip as *const u8,
iend.wrapping_sub(ip),
);
ip = ip.wrapping_add(loaded_size);
zbc.in_pos += loaded_size;
if loaded_size < to_load {
not_done = false;
continue;
}
let decoded_size = decompress_continue(
&mut *zbc.zc,
zbc.out_buff.add(zbc.out_start),
zbc.out_buff_size - zbc.out_start,
zbc.in_buff,
needed_in_size,
);
if ERR_isError(decoded_size) {
return decoded_size;
}
zbc.in_pos = 0;
if decoded_size == 0 {
zbc.stage = BUFF_READ;
continue;
}
zbc.out_end = zbc.out_start + decoded_size;
zbc.stage = BUFF_FLUSH;
}
BUFF_FLUSH => {
let to_flush_size = zbc.out_end.wrapping_sub(zbc.out_start);
let flushed_size = buff_limit_copy(
op as *mut u8,
oend.wrapping_sub(op),
zbc.out_buff.add(zbc.out_start),
to_flush_size,
);
op = op.wrapping_add(flushed_size);
zbc.out_start += flushed_size;
if flushed_size == to_flush_size {
zbc.stage = BUFF_READ;
if zbc.out_start + BLOCKSIZE > zbc.out_buff_size {
zbc.out_start = 0;
zbc.out_end = 0;
}
} else {
not_done = false;
}
}
_ => return ERROR(ZstdErrorCode::Generic),
}
}
*src_size_ptr = ip.wrapping_sub(istart);
*max_dst_size_ptr = op.wrapping_sub(ostart);
let mut next_src_size_hint = next_src_size_to_decompress(&*zbc.zc);
if next_src_size_hint > 3 {
next_src_size_hint = next_src_size_hint.wrapping_add(3);
}
next_src_size_hint.wrapping_sub(zbc.in_pos)
}
#[no_mangle]
pub extern "C" fn ZSTDv04_isError(code: usize) -> c_uint {
ERR_isError(code) as c_uint
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_decompress(
dst: *mut c_void,
max_original_size: usize,
src: *const c_void,
compressed_size: usize,
) -> usize {
let dctx = create_dctx();
if dctx.is_null() {
return ERROR(ZstdErrorCode::MemoryAllocation);
}
let result = decompress_using_dict(
&mut *dctx,
dst.cast::<u8>(),
max_original_size,
src.cast::<u8>(),
compressed_size,
ptr::null(),
0,
);
free_dctx(dctx);
result
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_decompressDCtx(
dctx: *mut ZSTDv04_Dctx,
dst: *mut c_void,
max_original_size: usize,
src: *const c_void,
compressed_size: usize,
) -> usize {
decompress_using_dict(
&mut *dctx,
dst.cast::<u8>(),
max_original_size,
src.cast::<u8>(),
compressed_size,
ptr::null(),
0,
)
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_findFrameSizeInfoLegacy(
src: *const c_void,
src_size: usize,
c_size: *mut usize,
d_bound: *mut u64,
) {
find_frame_size_info(src.cast::<u8>(), src_size, c_size, d_bound);
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_resetDCtx(dctx: *mut ZSTDv04_Dctx) -> usize {
reset_dctx(&mut *dctx)
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_createDCtx() -> *mut ZSTDv04_Dctx {
create_dctx()
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_freeDCtx(dctx: *mut ZSTDv04_Dctx) -> usize {
free_dctx(dctx)
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_nextSrcSizeToDecompress(dctx: *mut ZSTDv04_Dctx) -> usize {
next_src_size_to_decompress(&*dctx)
}
#[no_mangle]
pub unsafe extern "C" fn ZSTDv04_decompressContinue(
dctx: *mut ZSTDv04_Dctx,
dst: *mut c_void,
max_dst_size: usize,
src: *const c_void,
src_size: usize,
) -> usize {
decompress_continue(
&mut *dctx,
dst.cast::<u8>(),
max_dst_size,
src.cast::<u8>(),
src_size,
)
}
#[no_mangle]
pub extern "C" fn ZBUFFv04_isError(code: usize) -> c_uint {
ERR_isError(code) as c_uint
}
#[no_mangle]
pub extern "C" fn ZBUFFv04_getErrorName(code: usize) -> *const c_char {
crate::errors::ERR_getErrorName(code)
}
#[no_mangle]
pub extern "C" fn ZBUFFv04_recommendedDInSize() -> usize {
BLOCKSIZE + 3
}
#[no_mangle]
pub extern "C" fn ZBUFFv04_recommendedDOutSize() -> usize {
BLOCKSIZE
}
#[no_mangle]
pub unsafe extern "C" fn ZBUFFv04_createDCtx() -> *mut ZBUFFv04_DCtx {
buff_create_dctx()
}
#[no_mangle]
pub unsafe extern "C" fn ZBUFFv04_freeDCtx(dctx: *mut ZBUFFv04_DCtx) -> usize {
buff_free_dctx(dctx)
}
#[no_mangle]
pub unsafe extern "C" fn ZBUFFv04_decompressInit(dctx: *mut ZBUFFv04_DCtx) -> usize {
buff_decompress_init(&mut *dctx)
}
#[no_mangle]
pub unsafe extern "C" fn ZBUFFv04_decompressWithDictionary(
dctx: *mut ZBUFFv04_DCtx,
dict: *const c_void,
dict_size: usize,
) -> usize {
buff_decompress_with_dictionary(&mut *dctx, dict.cast::<u8>(), dict_size)
}
#[no_mangle]
pub unsafe extern "C" fn ZBUFFv04_decompressContinue(
dctx: *mut ZBUFFv04_DCtx,
dst: *mut c_void,
max_dst_size_ptr: *mut usize,
src: *const c_void,
src_size_ptr: *mut usize,
) -> usize {
buff_decompress_continue(
&mut *dctx,
dst.cast::<u8>(),
max_dst_size_ptr,
src.cast::<u8>(),
src_size_ptr,
)
}
#[cfg(test)]
mod tests {
use super::*;
const RAW_FRAME: &[u8] = &[
0x24, 0xB5, 0x2F, 0xFD, 0x00, 0x40, 0x00, 0x0B, b'r', b'a', b'w', b' ', b'v', b'0', b'.',
b'4', b'!', b'!', b'!', 0xC0, 0x00, 0x00,
];
const COMPRESSED_FRAME: &[u8] = &[
0x24, 0xB5, 0x2F, 0xFD, 0x00, 0x00, 0x00, 0xBB, 0xB0, 0x02, 0xC0, 0x10, 0x00, 0x1E, 0xB0,
0x01, 0x02, 0x00, 0x00, 0x80, 0x00, 0xE8, 0x92, 0x34, 0x12, 0x97, 0xC8, 0xDF, 0xE9, 0xF3,
0xEF, 0x53, 0xEA, 0x1D, 0x27, 0x4F, 0x0C, 0x44, 0x90, 0x0C, 0x8D, 0xF1, 0xB4, 0x89, 0x17,
0x00, 0x18, 0x00, 0x18, 0x00, 0x3F, 0xE6, 0xE2, 0xE3, 0x74, 0xD6, 0xEC, 0xC9, 0x4A, 0xE0,
0x71, 0x71, 0x42, 0x3E, 0x64, 0x4F, 0x6A, 0x45, 0x4E, 0x78, 0xEC, 0x49, 0x03, 0x3F, 0xC6,
0x80, 0xAB, 0x8F, 0x75, 0x5E, 0x6F, 0x2E, 0x3E, 0x7E, 0xC6, 0xDC, 0x45, 0x69, 0x6C, 0xC5,
0xFD, 0xC7, 0x40, 0xB8, 0x84, 0x8A, 0x01, 0xEB, 0xA8, 0xD1, 0x40, 0x39, 0x90, 0x4C, 0x64,
0xF8, 0xEB, 0x53, 0xE6, 0x18, 0x0B, 0x67, 0x12, 0xAD, 0xB8, 0x99, 0xB3, 0x5A, 0x6F, 0x8A,
0x19, 0x03, 0x01, 0x50, 0x67, 0x56, 0xF5, 0x9F, 0x35, 0x84, 0x60, 0xA0, 0x60, 0x91, 0xC9,
0x0A, 0xDC, 0xAB, 0xAB, 0xE0, 0xE2, 0x81, 0xFA, 0xCF, 0xC6, 0xBA, 0x01, 0x0E, 0x00, 0x54,
0x00, 0x00, 0x19, 0x00, 0x00, 0x54, 0x14, 0x00, 0x24, 0x24, 0x04, 0xFE, 0x04, 0x84, 0x4E,
0x41, 0x00, 0x27, 0xE2, 0x02, 0xC4, 0xB1, 0x00, 0xD2, 0x51, 0x00, 0x79, 0x58, 0x41, 0x28,
0x00, 0xE0, 0x0C, 0x01, 0x68, 0x65, 0x00, 0x04, 0x13, 0x0C, 0xDA, 0x0C, 0x80, 0x22, 0x06,
0xC0, 0x00, 0x00,
];
unsafe fn decode(frame: &[u8], capacity: usize) -> Result<Vec<u8>, usize> {
let mut output = vec![0u8; capacity];
let size = ZSTDv04_decompress(
output.as_mut_ptr().cast::<c_void>(),
output.len(),
frame.as_ptr().cast::<c_void>(),
frame.len(),
);
if ZSTDv04_isError(size) != 0 {
Err(size)
} else {
output.truncate(size);
Ok(output)
}
}
#[test]
fn decodes_raw_v04_frames() {
unsafe {
assert_eq!(decode(RAW_FRAME, 32).unwrap(), b"raw v0.4!!!");
}
}
#[test]
fn decodes_repository_compressed_v04_frame() {
let expected =
b"snowden is snowed in / he's now then in his snow den / when does the snow end?\n\
goodbye little dog / you dug some holes in your day / they'll be hard to fill.\n\
when life shuts a door, / just open it. it\xE2\x80\x99s a door. / that is how doors work.\n";
unsafe {
assert_eq!(decode(COMPRESSED_FRAME, expected.len()).unwrap(), expected);
}
}
#[test]
fn reports_frame_size_and_errors() {
unsafe {
let mut c_size = 0;
let mut bound = 0;
ZSTDv04_findFrameSizeInfoLegacy(
RAW_FRAME.as_ptr().cast::<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(RAW_FRAME, 1).unwrap_err(),
ERROR(ZstdErrorCode::DstSizeTooSmall)
);
}
}
#[test]
fn streaming_raw_frame_matches_one_shot() {
let expected = b"raw v0.4!!!";
let mut output = vec![0u8; expected.len()];
let mut input_offset = 0;
let mut output_offset = 0;
unsafe {
let dctx = ZSTDv04_createDCtx();
assert!(!dctx.is_null());
loop {
let needed = ZSTDv04_nextSrcSizeToDecompress(dctx);
if needed == 0 {
break;
}
assert!(input_offset + needed <= RAW_FRAME.len());
let produced = ZSTDv04_decompressContinue(
dctx,
output.as_mut_ptr().add(output_offset).cast::<c_void>(),
output.len() - output_offset,
RAW_FRAME.as_ptr().add(input_offset).cast::<c_void>(),
needed,
);
assert_eq!(ZSTDv04_isError(produced), 0);
input_offset += needed;
output_offset += produced;
}
assert_eq!(ZSTDv04_freeDCtx(dctx), 0);
}
assert_eq!(input_offset, RAW_FRAME.len());
assert_eq!(output_offset, expected.len());
assert_eq!(output, expected);
}
#[test]
fn buffered_streaming_raw_frame_matches_one_shot() {
let expected = b"raw v0.4!!!";
let mut output = vec![0u8; expected.len()];
let mut input_size = RAW_FRAME.len();
let mut output_size = output.len();
unsafe {
let dctx = ZBUFFv04_createDCtx();
assert!(!dctx.is_null());
assert_eq!(ZBUFFv04_decompressInit(dctx), 0);
let hint = ZBUFFv04_decompressContinue(
dctx,
output.as_mut_ptr().cast::<c_void>(),
&mut output_size,
RAW_FRAME.as_ptr().cast::<c_void>(),
&mut input_size,
);
assert_eq!(ZBUFFv04_isError(hint), 0);
assert_eq!(&output[..output_size], expected);
assert_eq!(input_size, RAW_FRAME.len());
assert_eq!(ZBUFFv04_freeDCtx(dctx), 0);
assert_eq!(ZBUFFv04_freeDCtx(ptr::null_mut()), 0);
}
}
}