lz

README

Module LZ

The LZ module provides sequencers that convert byte streams into blocks of Lempel-Ziv 77 sequences. It is designed to support multiple compression methods that differ in the way they are encoding those LZ77 sequences.

Documentation

Overview

Package lz supports encoding and decoding of LZ77 sequences. A sequence, as described in the Zstandard specification, consists of a literal copy command followed by a match copy command. The literal copy command is described by the length in literal bytes to be copied, and the match command consists of the distance of the match to copy and the length of the match in bytes.

A Parser converts a byte stream into blocks of sequences. The Decoder converts the block of sequences into the original decompressed byte stream.

The module provides multiple parser implementations that offer different combinations of encoding speed and compression ratios. Usually, a slower parser will generate a better compression ratio.

Parsers may use different matchers to provide their functionality. One Example is [greedyParser] which can use multiple Matcher implementations.

The library supports the implementation of parsers outside of this package.

Index

• Variables
• type Block
  • func (b *Block) Len() int64
  • func (b *Block) LenCheck() (n int64, err error)
• type Buffer
  • func (b *Buffer) ByteAt(off int64) (c byte, err error)
  • func (b *Buffer) Init(size, retentionSize int, shift ShiftFunc) error
  • func (b *Buffer) Parsable() int
  • func (b *Buffer) ReadAt(p []byte, off int64) (n int, err error)
  • func (b *Buffer) ReadFrom(r io.Reader) (n int64, err error)
  • func (b *Buffer) Reset(data []byte) error
  • func (b *Buffer) Write(p []byte) (n int, err error)
• type Configurator
  • func ParseJSON(data []byte) (Configurator, error)
• type Decoder
  • func (d *Decoder) ByteAtEnd(off int) byte
  • func (d *Decoder) Init(opts DecoderOptions) error
  • func (d *Decoder) Read(p []byte) (n int, err error)
  • func (d *Decoder) Reset()
  • func (d *Decoder) Write(p []byte) (n int, err error)
  • func (d *Decoder) WriteBlock(blk *Block) (n int, err error)
  • func (d *Decoder) WriteByte(c byte) error
  • func (d *Decoder) WriteMatch(mu, ou uint32) (n int, err error)
  • func (d *Decoder) WriteTo(w io.Writer) (n int64, err error)
• type DecoderOptions
  • func (opts DecoderOptions) NewDecoder() (*Decoder, error)
• type Entry
• type GenericMatcherOptions
  • func (opts *GenericMatcherOptions) MarshalJSON() ([]byte, error)
  • func (opts *GenericMatcherOptions) NewMatcher(windowSize, retentionSize, bufferSize int) (Matcher, error)
  • func (opts *GenericMatcherOptions) UnmarshalJSON(data []byte) error
• type GreedyParserOptions
  • func (gpo *GreedyParserOptions) MarshalJSON() ([]byte, error)
  • func (gpo *GreedyParserOptions) NewParser(windowSize, retentionSize, bufferSize int) (Parser, error)
  • func (gpo *GreedyParserOptions) UnmarshalJSON(data []byte) error
• type HashOptions
  • func (hopts *HashOptions) GetInputLen() int
  • func (hopts *HashOptions) MarshalJSON() ([]byte, error)
  • func (hopts *HashOptions) NewMapper() (Mapper, error)
• type Mapper
• type MapperConfigurator
  • func UnmarshalJSONMapperOptions(data []byte) (MapperConfigurator, error)
• type Matcher
• type MatcherConfigurator
  • func UnmarshalJSONMatcherOptions(data []byte) (MatcherConfigurator, error)
• type Parser
• type ParserFlags
• type Seq
  • func (s Seq) Len() int64
• type ShiftFunc

Variables

var ErrEndOfBuffer = errors.New("lz: end of buffer")

ErrEndOfBuffer is returned at the end of the buffer.

var ErrFullBuffer = errors.New("lz: full buffer")

ErrFullBuffer is returned when the buffer is full and no more data can be written to it.

var ErrOutOfBuffer = errors.New("lz: offset outside of buffer")

ErrOutOfBuffer is returned when the offset is outside of the buffer.

var ErrStartOfBuffer = errors.New("lz: start of buffer")

ErrStartOfBuffer is returned at the start of the buffer.

Types

type Block

type Block struct {
	Sequences []Seq
	Literals  []byte
}

Block stores sequences and literals. Note that the sequences stored in the Sequences slice might not consume the entire Literals slice. The remaining literal bytes must be added to the decoded text after all sequences have been decoded.

func (*Block) Len

func (b *Block) Len() int64

Len computes the length of the block in bytes. It assumes that the sum of the literal lengths in the sequences does not exceed the length of the Literals byte slice.

func (*Block) LenCheck

func (b *Block) LenCheck() (n int64, err error)

LenCheck computes the length of the block in bytes and verifies that the sum of the literal lengths in the sequences is less than the bytes in the Literals field. If that is not the case an error is returned.

type Buffer

type Buffer struct {
	Data []byte
	// Window end index
	W int
	// maximum buffer size
	Size int
	// RetentionSize is the number of bytes to keep when pruning the buffer.
	RetentionSize int
	// offset of Data
	Off int64

	// Shift will be called when the buffer is pruned to inform other
	// components about the number of bytes removed from the start of
	// the buffer.
	Shift ShiftFunc
}

Buffer is the buffer used for LZ parsing.

The Off field describes the offset of Data[0] in the original stream. The W points to the end of sliding window used for copying matches.

Data is not fully allocated at the beginning. It grows with the usage. There must be always 7 extra bytes allocated at the end of Data to allow easy reads of data from the buffer.

func (*Buffer) ByteAt

func (b *Buffer) ByteAt(off int64) (c byte, err error)

ByteAt returns the byte at offset off. If off is outside of the buffer, ErrOutOfBuffer is returned.

func (*Buffer) Init

func (b *Buffer) Init(size, retentionSize int, shift ShiftFunc) error

Init initializes the buffer. The old data slice is reused and the capacity might be larger than the new buffer size.

func (*Buffer) Parsable

func (b *Buffer) Parsable() int

Parsable returns the number of bytes that can be parsed from the buffer.

func (*Buffer) ReadAt

func (b *Buffer) ReadAt(p []byte, off int64) (n int, err error)

ReadAt reads len(p) bytes from the buffer starting at byte offset off. It returns the number of bytes read and any error encountered. If off is outside of the buffer, ErrOutOfBuffer is returned.

func (*Buffer) ReadFrom

func (b *Buffer) ReadFrom(r io.Reader) (n int64, err error)

ReadFrom reads data from r until EOF or error. It returns the number of bytes read and any error encountered.

func (*Buffer) Reset

func (b *Buffer) Reset(data []byte) error

Reset resets the buffer with the provided data slice. If the data slice is larger than the buffer size, the buffer size will be updated. Note that the data slice should have 7 extra bytes, len(data)+7 <= cap(data). Otherwise the old slice will be used or a new one need to be allocated.

func (*Buffer) Write

func (b *Buffer) Write(p []byte) (n int, err error)

Write writes data to the buffer. If not all data can be written, ErrFullBuffer is returned.

type Configurator

type Configurator interface {
	NewParser(windowSize, retentionSize, bufferSize int) (Parser, error)
}

Configurator creates a parser. Usually an Options type implements the interface.

func ParseJSON

func ParseJSON(data []byte) (Configurator, error)

ParseJSON unmarshals parser options from JSON data.

type Decoder

type Decoder struct {
	// Data is the actual buffer. The end of the slice is also the head of
	// the dictionary window.
	Data []byte
	// R tracks the position of the reads from the buffer and must be less
	// or equal to the length of the Data slice.
	R int
	// Off records the total offset and marks the end of the Data slice,
	// which is also the end of the dictionary window.
	Off int64

	// DecoderOptions provides the configuration parameters WindowSize and
	// BufferSize.
	DecoderOptions
}

Decoder provides a simple buffer for decoding LZ77 sequences. Data is the actual buffer. The end of the slice is also the head of the dictionary window. R tracks the read position in the buffer and must be less than or equal to the length of the Data slice. Off records the total offset and marks the end of the Data slice, which is also the end of the dictionary window. DecoderConfig provides the configuration parameters WindowSize and BufferSize.

func (*Decoder) ByteAtEnd

func (d *Decoder) ByteAtEnd(off int) byte

ByteAtEnd returns the byte at the end of the buffer.

func (*Decoder) Init

func (d *Decoder) Init(opts DecoderOptions) error

Init initializes the DecoderBuffer.

func (*Decoder) Read

func (d *Decoder) Read(p []byte) (n int, err error)

Read reads decoded data from the buffer.

func (*Decoder) Reset

func (d *Decoder) Reset()

Reset returns the DecoderBuffer to its initialized state.

func (*Decoder) Write

func (d *Decoder) Write(p []byte) (n int, err error)

Write inserts the slice into the buffer. The method will write the entire slice or return 0 and ErrFullBuffer.

func (*Decoder) WriteBlock

func (d *Decoder) WriteBlock(blk *Block) (n int, err error)

WriteBlock writes sequences from the block into the buffer. Each sequence is written atomically, as the block value is not modified. If there is not enough space in the buffer, ErrFullBuffer will be returned. All written sequences and literals will be removed from the block.

The capacity of the block slices will not be maintained. You have to keep a copy of the block to achieve that.

The growth of the array is limited to BufferSize.

The function returns the number of bytes written.

func (*Decoder) WriteByte

func (d *Decoder) WriteByte(c byte) error

WriteByte writes a single byte into the buffer.

func (*Decoder) WriteMatch

func (d *Decoder) WriteMatch(mu, ou uint32) (n int, err error)

WriteMatch appends the match to the end of the buffer. The match will be written completely, or n=0 and ErrFullBuffer will be returned.

func (*Decoder) WriteTo

func (d *Decoder) WriteTo(w io.Writer) (n int64, err error)

WriteTo writes the decoded data to the writer.

type DecoderOptions

type DecoderOptions struct {
	// Size of the sliding dictionary window in bytes.
	WindowSize int
	// Maximum size of the buffer in bytes.
	BufferSize int
}

DecoderOptions contains the parameters for the DecoderBuffer and decoder types. WindowSize must be smaller than BufferSize. It is recommended to set BufferSize to twice the WindowSize.

func (DecoderOptions) NewDecoder

func (opts DecoderOptions) NewDecoder() (*Decoder, error)

type Entry

type Entry struct {
	// contains filtered or unexported fields
}

Entry is returned by a Mapper for a found match.

type GenericMatcherOptions

type GenericMatcherOptions struct {
	MinMatchLen int
	MaxMatchLen int

	MapperOptions MapperConfigurator
}

GenericMatcherOptions provide the options for a generic matcher.

func (*GenericMatcherOptions) MarshalJSON

func (opts *GenericMatcherOptions) MarshalJSON() ([]byte, error)

MarshalJSON marshals the matcher options into JSON and adds the MatcherType field.

func (*GenericMatcherOptions) NewMatcher

func (opts *GenericMatcherOptions) NewMatcher(windowSize, retentionSize, bufferSize int) (Matcher, error)

NewMatcher creates a new generic matcher using the generic matcher options.

func (*GenericMatcherOptions) UnmarshalJSON

func (opts *GenericMatcherOptions) UnmarshalJSON(data []byte) error

type GreedyParserOptions

type GreedyParserOptions struct {
	MatcherOptions MatcherConfigurator
}

GreedyParserOptions defines the configuration options for a greedy parser.

func (*GreedyParserOptions) MarshalJSON

func (gpo *GreedyParserOptions) MarshalJSON() ([]byte, error)

MarshalJSON provides a custom JSON marshaller that adds a type field to the JSON structure.

func (*GreedyParserOptions) NewParser

func (gpo *GreedyParserOptions) NewParser(windowSize, retentionSize, bufferSize int) (Parser, error)

NewParser creates a new greedy parser using the greedy parser options.

func (*GreedyParserOptions) UnmarshalJSON

func (gpo *GreedyParserOptions) UnmarshalJSON(data []byte) error

UnmarshalJSON provides a custom JSON unmarshaler that parses the type field from the JSON structure.

type HashOptions

type HashOptions struct {
	InputLen int
	HashBits int
}

HashOptions provides the parameters for the Hash Mapper.

func (*HashOptions) GetInputLen

func (hopts *HashOptions) GetInputLen() int

GetInputLen returns the input length.

func (*HashOptions) MarshalJSON

func (hopts *HashOptions) MarshalJSON() ([]byte, error)

MarshalJSON generates the JSON representation of HashOptions by adding the Mapper field and set it to "hash".

func (*HashOptions) NewMapper

func (hopts *HashOptions) NewMapper() (Mapper, error)

NewMapper creates the hash mapper.

type Mapper

type Mapper interface {
	InputLen() int
	Reset()
	// Shift is called by the number of bytes pruned from the buffer.
	Shift(delta int)
	Put(a, w int, p []byte) int

	// Get returns all candidate entries for the provided hash value. The
	// entry value v contains the all 4 bytes stored a position i.
	Get(v uint64) []Entry
}

Mapper will be typically implemented by hash tables.

The Put method returns the number of trailing bytes that could not be hashed. Shift is called, when n bytes have been pruned from the buffer.

type MapperConfigurator

type MapperConfigurator interface {
	NewMapper() (Mapper, error)
}

MapperConfigurator creates a mapper, usually an Options type implements this function.

func UnmarshalJSONMapperOptions

func UnmarshalJSONMapperOptions(data []byte) (MapperConfigurator, error)

UnmarshalJSONMapperOptions unmarshals mapper options from JSON data. The function looks first for the MapperType field to determine the type of mapper to create.

type Matcher

type Matcher interface {
	Edges(n int) []Seq
	Skip(n int) (skipped int, err error)

	Write(p []byte) (n int, err error)
	ReadFrom(r io.Reader) (n int64, err error)

	ReadAt(p []byte, off int64) (n int, err error)
	ByteAt(off int64) (c byte, err error)

	Reset(data []byte) error

	Buf() *Buffer

	WindowSize() int
	Options() MatcherConfigurator
}

Matcher is responsible to find matches or Literal bytes in the byte stream.

type MatcherConfigurator

type MatcherConfigurator interface {
	NewMatcher(windowSize, retentionSize, bufferSize int) (Matcher, error)
}

MatcherConfigurator creates a matcher, usually an Options type implements the interface.

func UnmarshalJSONMatcherOptions

func UnmarshalJSONMatcherOptions(data []byte) (MatcherConfigurator, error)

UnmarshalJSONMatcherOptions unmarshals matcher options from JSON data. The function looks first for the MatcherType field to determine the type of matcher to create.

type Parser

type Parser interface {
	// Parse up to block size bytes from the internal buffer and provides
	// the sequences in the block structure. While slices will be reused,
	// not old information will be maintained.
	Parse(blk *Block, n int, flags ParserFlags) (parsed int, err error)

	// Write writes data into the internal buffer.
	Write(p []byte) (n int, err error)

	// ReadFrom reads data from the provided reader into the internal
	// buffer.
	ReadFrom(r io.Reader) (n int64, err error)

	// ReadAt reads len(p) bytes from the internal buffer at offset off.
	ReadAt(p []byte, off int64) (n int, err error)

	// ByteAt returns the byte at offset off in the internal buffer.
	ByteAt(off int64) (c byte, err error)

	// Reset resets the internal buffer to the provided data.
	Reset(data []byte) error

	// WindowSize returns the window size used by the parser.
	WindowSize() int

	// Options returns the options used to create the parser.
	Options() Configurator
}

Parser provides the possibility to parse a byte stream into LZ77 sequences.

type ParserFlags

type ParserFlags int

ParserFlags define optional parser behavior.

const (
	// NoTrailingLiterals indicates that the parser should not generate
	// trailing literal bytes in the output.
	NoTrailingLiterals ParserFlags = 1 << iota
)

type Seq

type Seq struct {
	LitLen   uint32
	MatchLen uint32
	Offset   uint32
	Aux      uint32
}

Seq represents a single Lempel-Ziv 77 sequence describing a match, consisting of the offset, the length of the match, and the number of literals preceding the match. The Aux field can be used in upper layers to store additional information.

func (Seq) Len

func (s Seq) Len() int64

Len returns the complete length of the sequence in bytes.

type ShiftFunc

type ShiftFunc func(delta int)

ShiftFunc defines the type of the shift function, called when the buffer is pruned to provide more available space.