RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-27 22:27:35 +00:00
Code Issues Activity Actions

LibCompress: Extract the LZMA state to a separate class

Browse source 
We will also need this in the compressor, as it needs to do the exact
same calculations in reverse.
This commit is contained in:
Tim Schumacher 2023-05-01 13:01:06 +02:00 • committed by Andreas Kling
parent 221b91ff61
commit 440d8f908f
2 changed files with 134 additions and 110 deletions
Download patch file Download diff file Expand all files Collapse all files

137
Userland/Libraries/LibCompress/Lzma.h
View file

@ -48,7 +48,81 @@ private:
};
static_assert(sizeof(LzmaHeader) == 13);
class LzmaDecompressor : public Stream {
class LzmaState {
protected:
// LZMA uses 11-bit probability counters, but they are usually stored in 16-bit variables.
// Therefore, we can model probabilities with a resolution of up to 1 / 2^11 (which is equal to 1 / 2048).
// The default probability for most counters is 0.5.
using Probability = u16;
static constexpr size_t probability_bit_count = 11;
static constexpr Probability default_probability = (1 << probability_bit_count) / 2;
static void initialize_to_default_probability(Span<Probability>);
LzmaState(FixedArray<Probability> literal_probabilities);
u64 m_total_processed_bytes { 0 };
static constexpr size_t literal_probability_table_size = 0x300;
FixedArray<Probability> m_literal_probabilities;
struct LzmaLengthCoderState {
public:
LzmaLengthCoderState();
Probability m_first_choice_probability { default_probability };
Probability m_second_choice_probability { default_probability };
static constexpr size_t maximum_number_of_position_bits = 4;
Array<Array<Probability, (1 << 3)>, (1 << maximum_number_of_position_bits)> m_low_length_probabilities;
Array<Array<Probability, (1 << 3)>, (1 << maximum_number_of_position_bits)> m_medium_length_probabilities;
Array<Probability, (1 << 8)> m_high_length_probabilities;
};
LzmaLengthCoderState m_length_coder;
LzmaLengthCoderState m_rep_length_coder;
static constexpr u16 normalized_to_real_match_length_offset = 2;
static constexpr u32 normalized_to_real_match_distance_offset = 1;
static constexpr size_t number_of_length_to_position_states = 4;
Array<Array<Probability, (1 << 6)>, number_of_length_to_position_states> m_length_to_position_states;
static constexpr size_t first_position_slot_with_binary_tree_bits = 4;
static constexpr size_t first_position_slot_with_direct_encoded_bits = 14;
// This is a bit wasteful on memory and not in the specification, but it makes the math easier.
static constexpr size_t number_of_binary_tree_distance_slots = first_position_slot_with_direct_encoded_bits - first_position_slot_with_binary_tree_bits;
static constexpr size_t largest_number_of_binary_tree_distance_bits = 5;
Array<Array<Probability, (1 << largest_number_of_binary_tree_distance_bits)>, number_of_binary_tree_distance_slots> m_binary_tree_distance_probabilities;
static constexpr size_t number_of_alignment_bits = 4;
Array<Probability, (1 << number_of_alignment_bits)> m_alignment_bit_probabilities;
// LZ state tracking.
u16 m_state { 0 };
u32 m_rep0 { 0 };
u32 m_rep1 { 0 };
u32 m_rep2 { 0 };
u32 m_rep3 { 0 };
u32 current_repetition_offset() const;
void update_state_after_literal();
void update_state_after_match();
void update_state_after_rep();
void update_state_after_short_rep();
static constexpr size_t maximum_number_of_position_bits = 4;
static constexpr size_t number_of_states = 12;
Array<Probability, (number_of_states << maximum_number_of_position_bits)> m_is_match_probabilities;
Array<Probability, number_of_states> m_is_rep_probabilities;
Array<Probability, number_of_states> m_is_rep_g0_probabilities;
Array<Probability, number_of_states> m_is_rep_g1_probabilities;
Array<Probability, number_of_states> m_is_rep_g2_probabilities;
Array<Probability, (number_of_states << maximum_number_of_position_bits)> m_is_rep0_long_probabilities;
};
class LzmaDecompressor : public Stream
, LzmaState {
public:
/// Creates a decompressor from a standalone LZMA container (.lzma file extension, occasionally known as an LZMA 'archive').
static ErrorOr<NonnullOwnPtr<LzmaDecompressor>> create_from_container(MaybeOwned<Stream>, Optional<MaybeOwned<CircularBuffer>> dictionary = {});
@ -65,14 +139,6 @@ public:
virtual void close() override;
private:
// LZMA uses 11-bit probability counters, but they are usually stored in 16-bit variables.
// Therefore, we can model probabilities with a resolution of up to 1 / 2^11 (which is equal to 1 / 2048).
// The default probability for most counters is 0.5.
using Probability = u16;
static constexpr size_t probability_bit_count = 11;
static constexpr Probability default_probability = (1 << probability_bit_count) / 2;
static void initialize_to_default_probability(Span<Probability>);
LzmaDecompressor(MaybeOwned<Stream>, LzmaDecompressorOptions, MaybeOwned<CircularBuffer>, FixedArray<Probability> literal_probabilities);
MaybeOwned<Stream> m_stream;
@ -80,7 +146,6 @@ private:
// This doubles as an output buffer, since we have to write all of our results into this anyways.
MaybeOwned<CircularBuffer> m_dictionary;
u64 m_total_decoded_bytes { 0 };
bool m_found_end_of_stream_marker { false };
bool is_range_decoder_in_clean_state() const;
bool has_reached_expected_data_size() const;
@ -101,62 +166,12 @@ private:
ErrorOr<u16> decode_symbol_using_reverse_bit_tree(size_t bit_count, Span<Probability> probability_tree);
ErrorOr<void> decode_literal_to_output_buffer();
static constexpr size_t literal_probability_table_size = 0x300;
FixedArray<Probability> m_literal_probabilities;
struct LzmaLengthDecoderState {
public:
LzmaLengthDecoderState();
Probability m_first_choice_probability { default_probability };
Probability m_second_choice_probability { default_probability };
static constexpr size_t maximum_number_of_position_bits = 4;
Array<Array<Probability, (1 << 3)>, (1 << maximum_number_of_position_bits)> m_low_length_probabilities;
Array<Array<Probability, (1 << 3)>, (1 << maximum_number_of_position_bits)> m_medium_length_probabilities;
Array<Probability, (1 << 8)> m_high_length_probabilities;
};
LzmaLengthDecoderState m_length_decoder;
LzmaLengthDecoderState m_rep_length_decoder;
static constexpr u16 normalized_to_real_match_length_offset = 2;
ErrorOr<u16> decode_normalized_match_length(LzmaLengthDecoderState&);
static constexpr size_t number_of_length_to_position_states = 4;
Array<Array<Probability, (1 << 6)>, number_of_length_to_position_states> m_length_to_position_states;
static constexpr size_t first_position_slot_with_binary_tree_bits = 4;
static constexpr size_t first_position_slot_with_direct_encoded_bits = 14;
// This is a bit wasteful on memory and not in the specification, but it makes the math easier.
static constexpr size_t number_of_binary_tree_distance_slots = first_position_slot_with_direct_encoded_bits - first_position_slot_with_binary_tree_bits;
static constexpr size_t largest_number_of_binary_tree_distance_bits = 5;
Array<Array<Probability, (1 << largest_number_of_binary_tree_distance_bits)>, number_of_binary_tree_distance_slots> m_binary_tree_distance_probabilities;
static constexpr size_t number_of_alignment_bits = 4;
Array<Probability, (1 << number_of_alignment_bits)> m_alignment_bit_probabilities;
ErrorOr<u16> decode_normalized_match_length(LzmaLengthCoderState&);
// This deviates from the specification, which states that "unsigned" is at least 16-bit.
// However, the match distance needs to be at least 32-bit, at the very least to hold the 0xFFFFFFFF end marker value.
static constexpr u32 normalized_to_real_match_distance_offset = 1;
ErrorOr<u32> decode_normalized_match_distance(u16 normalized_match_length);
// LZ state tracking.
u16 m_state { 0 };
u32 m_rep0 { 0 };
u32 m_rep1 { 0 };
u32 m_rep2 { 0 };
u32 m_rep3 { 0 };
u32 current_repetition_offset() const;
static constexpr size_t maximum_number_of_position_bits = 4;
static constexpr size_t number_of_states = 12;
Array<Probability, (number_of_states << maximum_number_of_position_bits)> m_is_match_probabilities;
Array<Probability, number_of_states> m_is_rep_probabilities;
Array<Probability, number_of_states> m_is_rep_g0_probabilities;
Array<Probability, number_of_states> m_is_rep_g1_probabilities;
Array<Probability, number_of_states> m_is_rep_g2_probabilities;
Array<Probability, (number_of_states << maximum_number_of_position_bits)> m_is_rep0_long_probabilities;
};
}