LibGfx+LibCompress: Extract the LZW decoder and move it to LibCompress
Let's put this state-of-the-art decoder from the 80's in its own file in order to reuse it with other formats, such as TIFF or PDF.
This commit is contained in:
Lucas CHOLLET
Tim Schumacher
parent
a86c0ac003
commit
00ad8419cf
Notes:
sideshowbarker
2024-07-16 21:51:02 +09:00
Author: https://github.com/LucasChollet Commit: https://github.com/SerenityOS/serenity/commit/00ad8419cf Pull-request: https://github.com/SerenityOS/serenity/pull/21803 Reviewed-by: https://github.com/nico ✅ Reviewed-by: https://github.com/timschumi ✅
2 changed files with 153 additions and 136 deletions
Userland/Libraries
151
Userland/Libraries/LibCompress/LZWDecoder.h
Normal file
151
Userland/Libraries/LibCompress/LZWDecoder.h
Normal file
|
|
@ -0,0 +1,151 @@
|
|||
/*
|
||||
* Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
|
||||
* Copyright (c) 2022, the SerenityOS developers.
|
||||
*
|
||||
* SPDX-License-Identifier: BSD-2-Clause
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <AK/Debug.h>
|
||||
#include <AK/Format.h>
|
||||
#include <AK/IntegralMath.h>
|
||||
#include <AK/Vector.h>
|
||||
|
||||
namespace Compress {
|
||||
|
||||
class LZWDecoder {
|
||||
private:
|
||||
static constexpr int max_code_size = 12;
|
||||
|
||||
public:
|
||||
explicit LZWDecoder(Vector<u8> const& lzw_bytes, u8 min_code_size)
|
||||
: m_lzw_bytes(lzw_bytes)
|
||||
, m_code_size(min_code_size)
|
||||
, m_original_code_size(min_code_size)
|
||||
, m_table_capacity(AK::exp2<u32>(min_code_size))
|
||||
{
|
||||
init_code_table();
|
||||
}
|
||||
|
||||
u16 add_control_code()
|
||||
{
|
||||
u16 const control_code = m_code_table.size();
|
||||
m_code_table.append(Vector<u8> {});
|
||||
m_original_code_table.append(Vector<u8> {});
|
||||
if (m_code_table.size() >= m_table_capacity && m_code_size < max_code_size) {
|
||||
++m_code_size;
|
||||
++m_original_code_size;
|
||||
m_table_capacity *= 2;
|
||||
}
|
||||
return control_code;
|
||||
}
|
||||
|
||||
void reset()
|
||||
{
|
||||
m_code_table.clear();
|
||||
m_code_table.extend(m_original_code_table);
|
||||
m_code_size = m_original_code_size;
|
||||
m_table_capacity = AK::exp2<u32>(m_code_size);
|
||||
m_output.clear();
|
||||
}
|
||||
|
||||
ErrorOr<u16> next_code()
|
||||
{
|
||||
size_t current_byte_index = m_current_bit_index / 8;
|
||||
if (current_byte_index >= m_lzw_bytes.size()) {
|
||||
return Error::from_string_literal("LZWDecoder tries to read ouf of bounds");
|
||||
}
|
||||
|
||||
// Extract the code bits using a 32-bit mask to cover the possibility that if
|
||||
// the current code size > 9 bits then the code can span 3 bytes.
|
||||
u8 current_bit_offset = m_current_bit_index % 8;
|
||||
u32 mask = (u32)(m_table_capacity - 1) << current_bit_offset;
|
||||
|
||||
// Make a padded copy of the final bytes in the data to ensure we don't read past the end.
|
||||
if (current_byte_index + sizeof(mask) > m_lzw_bytes.size()) {
|
||||
u8 padded_last_bytes[sizeof(mask)] = { 0 };
|
||||
for (int i = 0; current_byte_index + i < m_lzw_bytes.size(); ++i) {
|
||||
padded_last_bytes[i] = m_lzw_bytes[current_byte_index + i];
|
||||
}
|
||||
u32 const* addr = (u32 const*)&padded_last_bytes;
|
||||
m_current_code = (*addr & mask) >> current_bit_offset;
|
||||
} else {
|
||||
u32 tmp_word;
|
||||
memcpy(&tmp_word, &m_lzw_bytes.at(current_byte_index), sizeof(u32));
|
||||
m_current_code = (tmp_word & mask) >> current_bit_offset;
|
||||
}
|
||||
|
||||
if (m_current_code > m_code_table.size()) {
|
||||
dbgln_if(GIF_DEBUG, "Corrupted LZW stream, invalid code: {} at bit index {}, code table size: {}",
|
||||
m_current_code,
|
||||
m_current_bit_index,
|
||||
m_code_table.size());
|
||||
return Error::from_string_literal("Corrupted LZW stream, invalid code");
|
||||
} else if (m_current_code == m_code_table.size() && m_output.is_empty()) {
|
||||
dbgln_if(GIF_DEBUG, "Corrupted LZW stream, valid new code but output buffer is empty: {} at bit index {}, code table size: {}",
|
||||
m_current_code,
|
||||
m_current_bit_index,
|
||||
m_code_table.size());
|
||||
return Error::from_string_literal("Corrupted LZW stream, valid new code but output buffer is empty");
|
||||
}
|
||||
|
||||
m_current_bit_index += m_code_size;
|
||||
|
||||
return m_current_code;
|
||||
}
|
||||
|
||||
Vector<u8>& get_output()
|
||||
{
|
||||
VERIFY(m_current_code <= m_code_table.size());
|
||||
if (m_current_code < m_code_table.size()) {
|
||||
Vector<u8> new_entry = m_output;
|
||||
m_output = m_code_table.at(m_current_code);
|
||||
new_entry.append(m_output[0]);
|
||||
extend_code_table(new_entry);
|
||||
} else if (m_current_code == m_code_table.size()) {
|
||||
VERIFY(!m_output.is_empty());
|
||||
m_output.append(m_output[0]);
|
||||
extend_code_table(m_output);
|
||||
}
|
||||
return m_output;
|
||||
}
|
||||
|
||||
private:
|
||||
void init_code_table()
|
||||
{
|
||||
m_code_table.ensure_capacity(m_table_capacity);
|
||||
for (u16 i = 0; i < m_table_capacity; ++i) {
|
||||
m_code_table.unchecked_append({ (u8)i });
|
||||
}
|
||||
m_original_code_table = m_code_table;
|
||||
}
|
||||
|
||||
void extend_code_table(Vector<u8> const& entry)
|
||||
{
|
||||
if (entry.size() > 1 && m_code_table.size() < 4096) {
|
||||
m_code_table.append(entry);
|
||||
if (m_code_table.size() >= m_table_capacity && m_code_size < max_code_size) {
|
||||
++m_code_size;
|
||||
m_table_capacity *= 2;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Vector<u8> const& m_lzw_bytes;
|
||||
|
||||
int m_current_bit_index { 0 };
|
||||
|
||||
Vector<Vector<u8>> m_code_table {};
|
||||
Vector<Vector<u8>> m_original_code_table {};
|
||||
|
||||
u8 m_code_size { 0 };
|
||||
u8 m_original_code_size { 0 };
|
||||
|
||||
u32 m_table_capacity { 0 };
|
||||
|
||||
u16 m_current_code { 0 };
|
||||
Vector<u8> m_output {};
|
||||
};
|
||||
|
||||
}
|
||||
|
|
@ -13,6 +13,7 @@
|
|||
#include <AK/Memory.h>
|
||||
#include <AK/MemoryStream.h>
|
||||
#include <AK/Try.h>
|
||||
#include <LibCompress/LZWDecoder.h>
|
||||
#include <LibGfx/ImageFormats/GIFLoader.h>
|
||||
#include <string.h>
|
||||
|
||||
|
|
@ -110,141 +111,6 @@ static ErrorOr<GIFFormat> decode_gif_header(Stream& stream)
|
|||
return Error::from_string_literal("GIF header unknown");
|
||||
}
|
||||
|
||||
class LZWDecoder {
|
||||
private:
|
||||
static constexpr int max_code_size = 12;
|
||||
|
||||
public:
|
||||
explicit LZWDecoder(Vector<u8> const& lzw_bytes, u8 min_code_size)
|
||||
: m_lzw_bytes(lzw_bytes)
|
||||
, m_code_size(min_code_size)
|
||||
, m_original_code_size(min_code_size)
|
||||
, m_table_capacity(AK::exp2<u32>(min_code_size))
|
||||
{
|
||||
init_code_table();
|
||||
}
|
||||
|
||||
u16 add_control_code()
|
||||
{
|
||||
u16 const control_code = m_code_table.size();
|
||||
m_code_table.append(Vector<u8> {});
|
||||
m_original_code_table.append(Vector<u8> {});
|
||||
if (m_code_table.size() >= m_table_capacity && m_code_size < max_code_size) {
|
||||
|
||||
++m_code_size;
|
||||
++m_original_code_size;
|
||||
m_table_capacity *= 2;
|
||||
}
|
||||
return control_code;
|
||||
}
|
||||
|
||||
void reset()
|
||||
{
|
||||
m_code_table.clear();
|
||||
m_code_table.extend(m_original_code_table);
|
||||
m_code_size = m_original_code_size;
|
||||
m_table_capacity = AK::exp2<u32>(m_code_size);
|
||||
m_output.clear();
|
||||
}
|
||||
|
||||
ErrorOr<u16> next_code()
|
||||
{
|
||||
size_t current_byte_index = m_current_bit_index / 8;
|
||||
if (current_byte_index >= m_lzw_bytes.size()) {
|
||||
return Error::from_string_literal("LZWDecoder tries to read ouf of bounds");
|
||||
}
|
||||
|
||||
// Extract the code bits using a 32-bit mask to cover the possibility that if
|
||||
// the current code size > 9 bits then the code can span 3 bytes.
|
||||
u8 current_bit_offset = m_current_bit_index % 8;
|
||||
u32 mask = (u32)(m_table_capacity - 1) << current_bit_offset;
|
||||
|
||||
// Make a padded copy of the final bytes in the data to ensure we don't read past the end.
|
||||
if (current_byte_index + sizeof(mask) > m_lzw_bytes.size()) {
|
||||
u8 padded_last_bytes[sizeof(mask)] = { 0 };
|
||||
for (int i = 0; current_byte_index + i < m_lzw_bytes.size(); ++i) {
|
||||
padded_last_bytes[i] = m_lzw_bytes[current_byte_index + i];
|
||||
}
|
||||
u32 const* addr = (u32 const*)&padded_last_bytes;
|
||||
m_current_code = (*addr & mask) >> current_bit_offset;
|
||||
} else {
|
||||
u32 tmp_word;
|
||||
memcpy(&tmp_word, &m_lzw_bytes.at(current_byte_index), sizeof(u32));
|
||||
m_current_code = (tmp_word & mask) >> current_bit_offset;
|
||||
}
|
||||
|
||||
if (m_current_code > m_code_table.size()) {
|
||||
dbgln_if(GIF_DEBUG, "Corrupted LZW stream, invalid code: {} at bit index {}, code table size: {}",
|
||||
m_current_code,
|
||||
m_current_bit_index,
|
||||
m_code_table.size());
|
||||
return Error::from_string_literal("Corrupted LZW stream, invalid code");
|
||||
} else if (m_current_code == m_code_table.size() && m_output.is_empty()) {
|
||||
dbgln_if(GIF_DEBUG, "Corrupted LZW stream, valid new code but output buffer is empty: {} at bit index {}, code table size: {}",
|
||||
m_current_code,
|
||||
m_current_bit_index,
|
||||
m_code_table.size());
|
||||
return Error::from_string_literal("Corrupted LZW stream, valid new code but output buffer is empty");
|
||||
}
|
||||
|
||||
m_current_bit_index += m_code_size;
|
||||
|
||||
return m_current_code;
|
||||
}
|
||||
|
||||
Vector<u8>& get_output()
|
||||
{
|
||||
VERIFY(m_current_code <= m_code_table.size());
|
||||
if (m_current_code < m_code_table.size()) {
|
||||
Vector<u8> new_entry = m_output;
|
||||
m_output = m_code_table.at(m_current_code);
|
||||
new_entry.append(m_output[0]);
|
||||
extend_code_table(new_entry);
|
||||
} else if (m_current_code == m_code_table.size()) {
|
||||
VERIFY(!m_output.is_empty());
|
||||
m_output.append(m_output[0]);
|
||||
extend_code_table(m_output);
|
||||
}
|
||||
return m_output;
|
||||
}
|
||||
|
||||
private:
|
||||
void init_code_table()
|
||||
{
|
||||
m_code_table.ensure_capacity(m_table_capacity);
|
||||
for (u16 i = 0; i < m_table_capacity; ++i) {
|
||||
m_code_table.unchecked_append({ (u8)i });
|
||||
}
|
||||
m_original_code_table = m_code_table;
|
||||
}
|
||||
|
||||
void extend_code_table(Vector<u8> const& entry)
|
||||
{
|
||||
if (entry.size() > 1 && m_code_table.size() < 4096) {
|
||||
m_code_table.append(entry);
|
||||
if (m_code_table.size() >= m_table_capacity && m_code_size < max_code_size) {
|
||||
++m_code_size;
|
||||
m_table_capacity *= 2;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Vector<u8> const& m_lzw_bytes;
|
||||
|
||||
int m_current_bit_index { 0 };
|
||||
|
||||
Vector<Vector<u8>> m_code_table {};
|
||||
Vector<Vector<u8>> m_original_code_table {};
|
||||
|
||||
u8 m_code_size { 0 };
|
||||
u8 m_original_code_size { 0 };
|
||||
|
||||
u32 m_table_capacity { 0 };
|
||||
|
||||
u16 m_current_code { 0 };
|
||||
Vector<u8> m_output {};
|
||||
};
|
||||
|
||||
static void copy_frame_buffer(Bitmap& dest, Bitmap const& src)
|
||||
{
|
||||
VERIFY(dest.size_in_bytes() == src.size_in_bytes());
|
||||
|
|
@ -317,7 +183,7 @@ static ErrorOr<void> decode_frame(GIFLoadingContext& context, size_t frame_index
|
|||
if (image->lzw_min_code_size > 8)
|
||||
return Error::from_string_literal("LZW minimum code size is greater than 8");
|
||||
|
||||
LZWDecoder decoder(image->lzw_encoded_bytes, image->lzw_min_code_size);
|
||||
Compress::LZWDecoder decoder(image->lzw_encoded_bytes, image->lzw_min_code_size);
|
||||
|
||||
// Add GIF-specific control codes
|
||||
int const clear_code = decoder.add_control_code();
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue