.7z format specification

Abstract

7-zip archive is one of popular files compression and archive formats. There has been no well-defined file format because there is no precise specification document in 20 years from its birth, so it has been considered as an application proprietary format.

There are some independent implementation of utility to handle 7-zip archives, precise documentation of specification is mandatory to keep compatibility and interoperability among implementations.

This specification defines an archive file format of .7z archive. A purpose of this document is to provide a concrete documentation to archive an interoperability among implementations.

Introduction

Purpose

This specification is intended to define a cross-platform, interoperable file storage and transfer format. The information here is meant to be a concise guide for those wishing to implement libraries and utility to handle 7-zip archive files.

This documentations is NOT a specification of any existed utilities and libraries. This documentation does not have some features which is implemented in an existed utility. It is because this document purpose is to keep interoperability.

Intended audience

This specification is intended for use by implementors of software to compress files into 7-zip format and/or decompress files from 7-zip format.

The text of the specification assumes a basic background in programming at the level of bits and other primitive data representations.

Scope

“7-zip archive” is one of popular files compression and archive formats. It is universally used to aggregate, compress, and encrypt files into a single interoperable container. No specific use or application need is defined by this format and no specific implementation guidance is provided. This document provides details on the storage format for creating 7-zip files. Information is provided on the records and fields that describe what a .7z file is.

This specification does not provide technical specification of compression methods such as LZMA, LZMA2, Delta, BCJ and every other methods. It also does not provide technical specification of encryption and hash methods such as AES and SHA256.

Trademarks

7-zip is a public-domain utility on Microsoft Windows platforms written by Igor Pavlov. 7-zip archive file format was originally produced and defined by 7-zip utility. p7zip is a cross-platform utility to handle 7zip archive file, which is a port of 7-zip to posix. py7zr is a library and utility written with pure python3 to handle 7zip archive, that is distributed under GNU Lessaer General Public License version 2.1 and later. xzutils is an file compression/decompression utility. liblzma is a library to provide LZMA and LZMA2 compression algorithm provided by xzutils project. Python is one of popular computer language and running platform copyrighted and licensed by Python Foundation. Python 3 provide lzma API deppend on liblzma.

Motivation

There are several file archive format and utilities. Many of them are born as proprietary format of archive utility software, because of its nature, only standardized formats are now alived as portable, stable for long time and freely usable specification. PKWare ZIP, GNU Tar and GZip are examples for it. Since 7-zip, its format and algorithm LZMA/LZMA2 are born as public-domain in 1999, it has been known as one of long lived file format.

There are two effort to make .7z archives as well-documented, portable, and long life. One is a documentation project here, and other is a software development project to be compatible with original 7zip and p7zip utility such as py7zr.

Notations

  • Use of the term SHALL indicates a required element.

  • MAY NOT or SHALL NOT indicates an element is prohibited from use.

  • SHOULD indicates a RECOMMENDED element.

  • SHOULD NOT indicates an element NOT RECOMMENDED for use.

  • MAY indicates an OPTIONAL element.

Data Representations

This chapter describes basic data representations used in 7-zip file.

BYTE

BYTE is a basic data type to store a char, Property ID or bitfield.

BYTEARRAY

BYTEARRAY is a sequence of BYTE. Its length SHALL be defined in another place.

String

There are two type of string data is used in 7-zip archive format.

  • UTF-16-LE

  • UTF-8

Integers

All integers that require more than one byte SHALL be in a little endian, Least significant byte (LSB) comes first, then more significant bytes in ascending order of significance (LSB MSB for two byte integers, B0 B1 B2 B3 for four bytes integers). The highest bit (value 128) of byte is number bit 7 and lowest bit (value 1) is number bit 0. Values are unsigned unless otherwise noted.

name

size

description

UINT32

4 bytes

integer at little endian
represent 0 to
4,294,967,295
(0xffffffff)

UINT64

8 bytes

integer at little endian
represent 0 to
18,446,744,073,709,551,615
(0xffffffffffffffff)

NUMBER

1-9
bytes
variable length integer
value represent 0 to
18,446,744,073,709,551,615
(0xffffffffffffffff)

NUMBER SHALL be a integer value encoded with the following scheme. in byte length between one byte to nine bytes.

Size of encoding sequence SHALL indicated at first byte. The rest bits of first byte represent a bits from MSB of value. Following bytes SHOULD be an integer as little endian.

First_Byte (binary )

Extra_Bytes

Value (y: little endian integer)

0xxxxxxx

(0b0xxxxxxx )

10xxxxxx

BYTE y[1]

(0b00xxxxxx << (8 * 1)) + y

110xxxxx

BYTE y[2]

(0b000xxxxx << (8 * 2)) + y

1110xxxx

BYTE y[3]

(0b0000xxxx << (8 * 3)) + y

11110xxx

BYTE y[4]

(0b00000xxx << (8 * 4)) + y

111110xx

BYTE y[5]

(0b000000xx << (8 * 5)) + y

1111110x

BYTE y[6]

(0b0000000x << (8 * 6)) + y

11111110

BYTE y[7]

y

11111111

BYTE y[8]

y

BitField

BitField represent eight boolean values in single BYTE.

The bit field is defined which order is from MSB to LSB, i.e. bit 7 (MSB) of first byte indicate a boolean for first stream, object or file, bit 6 of first byte indicate a boolean for second stream, object or file, and bit 0(LSB) of second byte indicate a boolean for 16th stream, object or file.

A length is vary according to a number of items to indicate. If a number of items is not multiple of eight, rest of bitfield SHOULD zero.

BooleanList

BooleanList is a list of boolean bit arrays. It has two field. First it defines an existence of boolean values for each items of number of files or objects. Then boolean bit fields continues. There is an extension of expression that indicate all boolean values is True, and skip boolean bit fields.

alldefined, BYTEboolean, BitField

File format

7-zip archive file format SHALL consist of three part. 7-zip archive file SHALL start with signature header. The data block SHOULD placed after the signature header. The data block is shown as Packed Streams.

A header database SHOULD be placed after the data block. The data block MAY be empty when no archived contents exists. So Packed Streams is optional.

Since Header database CAN be encoded then it SHOULD place after data block, that is Packed Streams for Headers. When Header database is encoded, Header encode Information SHALL placed instead of Header.

When Header database is placed as plain form, Packed Streams for Headers SHALL NOT exist.

Signature HeaderPacked StreamsPacked Streams for HeaderHeader Encode InformationHeader

Signature Header

Signature header SHALL consist in 32 bytes. Signature header SHALL start with Signature then continues with archive version. Start Header SHALL follow after archive version.

SignatureMajor Version, BYTE, '0x00'Minor Version, BYTE, '0x04'Start Header CRC, UINT32Next Header Offset, UINT64Next Header Size, UINT64Next Header CRC, UINT32

It can be observed as follows when taken hex dump.

address

0

1

2

3

4

5

6

7

8

9

A

B

C

D

E

F

0x0000

Signature

VN

S.H. CRC

N.H. offset

0x0010

offset(cont)

N.H. size

N.H. CRC

Signature

The first six bytes of a 7-zip file SHALL always contain b'7z\\xbc\\xaf\\x27\\x1c'.

Version Number

Version number SHALL consist with two bytes. Major version is 0x00, and minor version is 0x04 for now.

Start Header CRC

It SHALL be stored in form of UINT32. This CRC value SHALL be calculated from Next Header Offset, Next Header size and Next Header CRC.

Next Header offset

Next header offset SHALL be an offset from end of signature header to header database. Because signature header always consist with 32 bytes, the offset SHOULD be a value that absolute position of header database in archive file - 32 bytes. Next header offset SHALL be stored as UINT64.

Next Header size

Next header size SHALL be an size of a header database. Because a header database MAY be encoded, Next header size SHALL consist of encoded(packed) size, not a raw size. Next header size SHALL be stored as UINT64.

Next Header CRC

Next header CRC SHALL a CRC32 of Header that SHALL be stored in UINT32.

Property IDs

Information stored in Header SHALL be placed after Property ID. For example, Header Info block start with 0x01, which means Header, then continues data blocks, and 0x00, which is END, is placed at last. This structure can be recursive but there is a rules where particular ID can exist.

Property ID SHALL be a BYTE.

ID

Property

0x00

END

0x01

Header

0x02

ArchiveProperties

0x03

AdditionalStreamsInfo

0x04

MainStreamsInfo

0x05

FilesInfo

0x06

PackInfo

0x07

UnPackInfo

0x08

SubStreamsInfo

0x09

Size

0x0A

CRC

0x0B

Folder

0x0C

CodersUnPackSize

0x0D

NumUnPackStream

0x0E

EmptyStream

0x0F

EmptyFile

0x10

Anti

0x11

Name

0x12

CTime

0x13

ATime

0x14

MTime

0x15

Attributes

0x16

Comment

0x17

EncodedHeader

0x18

StartPos

0x19

Dummy

Header encode Information

Header encode Information is a Streams Information data for Header data as encoded data followed after ID 0x17, EncodedHeader Property.

EncodedHeader, Property IDStreams Information for Header, StreamsInfo

Pack Information

Pack Information SHALL start with one byte of id value; 0x06. Pack Information SHALL be const with Pack Position, Number of Pack Streams, a list of sizes of Pack Streams and a list of CRCs of pack streams. Pack position and Number of Pack streams SHALL be stored as variable length NUMBER form. Sizes of packed Streams SHALL stored as a list of NUMBER.

PackInfo, Property IDPack Position, NUMBERCount of Pack Streams, NUMBERSizes of Pack StreamsCRCs of Pack StreamsEND, Property ID

Pack Position

Pack Position SHALL indicate a position of encoded streams that value SHALL be an offset from the end of signature header. It MAY be a next position of end of signature header.

Count of Pack Streams

Count of Pack Streams SHALL indicate a number of encoded streams. LZMA and LZMA2 SHOULD have a single (one) stream. 7-zip CAN have encoding methods which produce multiple encoded streams. When there are multiple streams, a value of Number of Pack Streams SHALL indicate it.

Sizes of Pack Streams

Sizes of Pack Streams SHOULD be omitted when Number of Pack Streams is zero. This is an array of NUMBER values which length is as same as Count of Pack Streams. Size SHALL be positive integer and SHALL stored in NUMBER.

Size, Property IDsize, NUMBER

CRCs of Pack Streams

When Count of Pack Streams is zero, then CRCs of Pack Streams SHALL not exist. CRC CAN be exist and indicated as DigestDefined BooleanList. CRC SHALL be CRC32 and stored in UINT32.

CRC, Property IDDigestDefined, BooleanListcrc, UINT32

Coders Information

Coders Information SHALL located after Main Streams Information. It SHALL provide encoding and encryption filter parameters. It MAY be a single coder or multiple coders defined. It SHALL NOT be more than five coders. (Maximum four)

UnpackInfo, Property IDFolder, Property IDNumber of Folders, NUMBERNot Ext(0x00), BYTEFolderExt(0x01), BYTEData Stream Index, NUMBERCodersUnpackSize, Property IDUnpacksize, NUMBERUnpackDigest, Property IDUnpackDigest, UINT32END, Property ID

In default Folders information is placed inline, then External flag is 0x00.

UnpackSizes

UnpackSizes is a list of decompress sizes for each archived file data. When extract data from the archive, it SHALL be distilled from unpack streams and split chunk into defined sizes.

Filenames are defined in File Information block. An order of data chunks and a order of filenames SHALL be same, except for filenames which is defined as empty stream.

UnpackDigests

UnpackDigests is a list of CRC32 of decompress data digests for each folders. When extract data from the archive, it CAN check an integrity of data.

It SHALL be a list of NUMBER and its length SHALL be as same as number of folders. It MAY be skipped when Substreams Information defined.

Folders

Folder in 7-zip archive means a basic container unit for encoded data. It brings encoded data. The data chunk Packed Streams is defined as series of Folders.

Each Folder has coder information. CoderInfo is consist of flag, number of streams and properties.

Flag indicate the coder is simple i.e. single input and single output, or complex i.e. multiple input, multiple output.

When simple coder, number of streams is always one for input, and one for output, so it SHALL be skipped.

Number of Coders, NUMBERCoder Property

Number of coder SHALL be a NUMBER integer number. Coder Properties SHALL be a list of Coder Property with length SHALL be as same as Number of coder.

Coder Property

Coder Property is defined with flag which indicate coder types. According to flag that indicate coder is complex, the Coder Property MAY have a number of input and output streams of coder.

Flag is defined in one byte as following bit definitions.

  • bit 3-0: Codec ID size

  • bit 4: Is complex codec

  • bit 5: There are attributes

  • bit 6-7: Reserved, it SHOULD always be zero.

Flag, BYTECoder ID, BYTEARRAYNumInStreams, NUMBERNumOutStreams, NUMBERProperty Size, NUMBERProperty, BYTEARRAYInput Index, NUMBEROutout Index, NUMBERPacked Stream Index, NUMBER

BindPairs

BindPairs describe connection among coders when coder produce multiple output or required multiple input.

A coder property format is vary with flag. Following pseudo code indicate how each parameter located for informative purpose.

if (Is Complex Coder)
 {
   NUMBER ``NumInStreams``;
   NUMBER ``NumOutStreams``;
 }
 if (There Are Attributes)
 {
   NUMBER ``PropertiesSize``
   BYTE ``Properties[PropertiesSize]``
 }
}
NumBindPairs :  = ``NumOutStreamsTotal`` – 1;
for (``NumBindPairs``)
 {
   NUMBER ``InIndex``;
   NUMBER ``OutIndex``;
 }
NumPackedStreams : ``NumInStreamsTotal`` – ``NumBindPairs``;
 if (``NumPackedStreams`` > 1)
   for(``NumPackedStreams``)
   {
     NUMBER ``Index``;
   };

When using only simple codecs, which has one input stream and one output stream, coder property become as simple as follows;

Flag, BYTECoder ID, BYTEARRAYProperty Size, NUMBERProperty, BYTEARRAY

Here is an example of bytes of coder property when specifying LZMA.

  • b'\x23\x03\x01\x01\x05\x5D\x00\x10\x00\x00'

In this example, first byte 0x23 indicate that coder id size is three bytes, and it is not complex codec and there is a codec property. A coder ID is b'\x03\x01\x01' and property length is five and property is b'\x5D\x00\x10\x00\x00'.

Codec IDs

Conformant implementations SHALL support mandatory codecs that are COPY, LZMA, LZMA2, BCJ, and Delta. There are a variant of BCJ that are X86, PowerPC, SPARC, ARM, ARMTHUMB, and IA64. Conformant implementations SHOULD also support optional codecs that are AES, BZIP2, DEFLATE, BCJ2, and PPMd. Implementations MAY support additional codecs that are ZStandard, and LZ4. It MAY also support proprietary codec such as DEFLATE64.

Conformant implementations SHALL accept these codec IDs and when it does not support it, it SHOULD report it as not supported.

Here is a list of famous codec IDs.

NAME

ID

Note

COPY

0x00

DELTA

0x03

BCJ

0x04

LZMA

0x030101

P7Z_BCJ

0x03030103

P7Z_BCJ2

0x0303011b

[1] [2]

BCJ_PPC

0x03030205

BCJ_IA64

0x03030301

BCJ_ARM

0x03030501

BCJ_ARMT

0x03030701

BCJ_SPARC

0x03030805

LZMA2

0x21

BZIP2

0x040202

DEFLATE

0x040108

DEFLATE64

0x040109

[1] [3]

ZSTD

0x04f71101

BROTLI

0x04f71102

LZ4

0x04f71104

[1]

LZS

0x04f71105

[1]

LIZARD

0x04f71106

[1]

AES

0x06f10701

Footnotes

Substreams Information

Substream Information is an optional field that indicate substreams from each folder produces.

When the archive is not solid, there SHALL NOT be SubStreams information. When SubStreams Information is omitted, extractor still know a unpack size information as folder information.

Substreams Information hold an information about archived data blocks as in extracted form. It SHALL exist that number of unpack streams, size of each unpack streams, and CRC of each streams

SubStreamsInfo, Property IDNumUnpackStream, Property IDNumber of unpack streams, NUMBERSize, Property IDSize of unpack streams, NUMBERCRC, Property IDdigest, UINT32END, Property ID

Files Information

Files Information SHOULD hold a list of files, directories and symbolic links. Its order SHALL be as same as order of streams defined in packed information. A type of file is stored in Attribute field.

FileInfo, Property IDNumber of Files, NUMBEREmpty Stream, Property IDSize, NUMBERFlag of Empty Streams, BitFieldEmpty Files, Property IDSize, NUMBERFlag of Empty Files, BitFieldDummy, Property IDSize, NUMBER0x00Name, Property IDSize, NUMBERFileNamesExist, BooleanListNot External(0x00), BYTEFileName, UTF-16-LEExt(0x01), BYTEData Index, NUMBERMTime, Property IDSize, NUMBERTimeExist, BooleanListExternal, BYTE, 0x00FileTime, NUMBERExternal, BYTE, 0x01Data Index, NUMBERCTime, Property IDSize, NUMBERTimeExist, BooleanListExternal, BYTE, 0x00FileTime, NUMBERExternal, BYTE, 0x01Data Index, NUMBERATime, Property IDSize, NUMBERTimeExist, BooleanListExternal, BYTE, 0x00FileTime, NUMBERExternal, BYTE, 0x01Data Index, NUMBERAttribute, Property IDSize, NUMBERAttributeExist, BooleanListNot External(0x00), BYTEAttribute, UINT32Ext(0x01), BYTEData Index, NUMBEREND, Property ID

Size

Size field indicate a size of next data. For example, Name size means, a size in byte from a start of FileNamesExist field and an end of file names.

Empty Streams

Empty streams has a number of emptystreams and a boolean list to indicate which file entry does not have a packed stream.

Dummy

Dummy MAY be placed for alignment. When processing File Names, which is UTF-16-LE, it is better to be aligned in word barrier.

FileName

FileNam SHALL be a wide character string encoded with UTF-16-LE and follows wchar_t NULL character, i.e. 0x0000.

Path separator SHALL be normalized as ‘/’, which is as POSIX standard. FileName SHOULD be relative path notation.

Attribute

Attribute is a UINT32 integer value. From bit 0 to 15 are as same as Windows attributes. Bit 16 to 31 is used for storing unix attributes. When file is a symbolic link, it SHOULD has an attribute that UNIX_EXTENSION flag enabled, and link bit of unix attributes.

Attribute values

ID/Value

Description

FILE_ATTRIBUTE_READONLY 1 (0x1)

A file that is read-only.

FILE_ATTRIBUTE_HIDDEN 2 (0x2)

The file or directory is hidden.

FILE_ATTRIBUTE_DIRECTORY 16 (0x10)

It identifies a directory.

FILE_ATTRIBUTE_ARCHIVE 32 (0x20)

A file or directory that is an archive file or directory.

FILE_ATTRIBUTE_REPARSE_POINT 1024 (0x400)

file or directory that has an associated reparse point, or a file that is a symbolic link.

bit 16-31

UNIX file permissions and attributes. 16bit shift to left of permissions and attributes.

UNIX_EXTENSION (0x8000)

Indicate a unix permissions and file attributes are bundled when 1.

FileTime

FileTime are NUMBER values in 100-nanosecond intervals since 1601/01/01 (UTC)

File type and a way

Normal files

Normal files are stored with packed streams and ordinal file information. Its contents are stored into packed stream. It SHOULD have an attribute of Windows such as FILE_ATTRIBUTE_ARCHIVE. It MAY also have an attribute of UNIX such as rw_r__r__ permissions.

Empty files

Empty files, which size is zero, SHALL be stored without packed stream, and with flle information. It SHOULD have an attribute of Windows such as FILE_ATTRIBUTE_ARCHIVE. It MAY also have an attribute of UNIX such as rw_r__r__ permissions.

Directories

Directories are stored without packed streams. It have entries in file information. It SHALL have an attribute which is FILE_ATTRIBUTE_DIRECTORY. It MAY also have an attribute of UNIX such as rwxr_xr_x permissions.

Special Files

There is an extension to handle special files such as sockets, device files, and symbolic links. A type of special files is indicated as file attribute. Further attribute of special file is stored as a content.

Compliant client MAY skip record of special files on extraction.

REPARSE_POINT on Windows

Reparse point on windows SHOULD be stored with packed stream and file information. Its target link path, in absolute, are recorded into packed stream in UTF-8 character encoding. It SHALL have an attribute which is FILE_ATTRIBUTE_REPARSE_POINT.

Appendix: BNF expression (Informative)

This clause shows extended BNF expression of 7-zip file format.

7-zip archive       ::=  SignatureHeader, [PackedStreams],
                         [PackedStreamsForHeaders], Header | HeaderInfo
SignatureHeader     ::=  Signature, ArchiveVersion, StartHeader
Signature           ::=  ``b'7z\xBC\xAF\x27\x1C'``
ArchiveVersion      ::=  ``b'\x00\x04'``
StartHeader         ::=  StartHeaderCRC, NextHeaderOffset,
                         NextHeaderSize, NextHeaderCRC
NextHeaderOffset    ::=  `UINT64`
NextHeaderSize      ::=  `UINT64`
NextHeaderCRC       ::=  `UINT32`
StreamsInfo         ::=  PackInfo, CodersInfo, SubStreamsInfo
PackInfo            ::=  `0x06`, PackPos, NumPackStreams,
                         SizesOfPackStream, CRCsOfPackStreams
CodersInfo          ::=  `0x07`, FoldersInfo
Folders Information ::=  0x0B, NumFolders, FolderInfo,
                         CoderUnpackSizes, UnpackDigests, 0x00
FoldersInfo         ::=  `0x0B`, NumFolders, (`0x00`, Folders) | (`0x01`, DataStreamIndex)
                         [`0x0C`, UnPackSizes, [`0x0A`, UnpackDigests]], `0x00`
Folders             ::=  Folder{ Number of Folders }
UnpackSizes         ::=  UnPackSize { Sum of NumOutStreams for each Folders }
UnpackSize          ::=  `NUMBER`
UnpackDigests       ::=  CRC32 { Number of folders }
SubStreamsInfo      ::=  `0x08`, `0x0D`, NumUnPackStreamsInFolders{Num of Folders],
                         `0x09`, UnPackSize, `0x0A`,
                         Digests{Number of streams with unknown CRC}, 0x00
Folder              ::=  NumCoders, CoderData { NumCoders }
CoderData           ::=  CoderFlag, CoderID, NumCoderStreamInOut, Properties,
                         BinPairs, PackedStreamIndex
CoderFlag           ::=  BYTE(bit 0:3 CodecIdSize, 4: Is Complex Coder,
                         5: There Are Attributes, 6: Reserved, 7: 0)
CoderId             ::=  BYTE{CodecIdSize}
FilesInfo           ::=  `0x05`, NumFiles, FileInfo, [FileInfo]
FileInfo            ::=  NumFiles, [0x0E, bit array of IsEmptyStream],
                         [`0x0F`, bit array of IsEmptyFile],
                         [`0x11`, FileNames],
                         [`0x12`, FileTime], [`0x13`, FileTime], [`0x14`, FileTime],
                         [`0x15`, Attributes]
FileTime            ::=  (`0x00`, bit array of TimeDefined |  0x01),
                         (`0x00`, list of Time | 0x01, DataIndex)
FileNames           ::=  (`0x00`, list of each filename | 0x01, DataIndex)
filename            ::=  Name, `0x0000`
Name                ::=  UTF16-LE Char, [Name]
Attributes          ::=  (`0x00`, bit array of AttributesAreDefined |  `0x01`),
                         (`0x00`, list of Attribute | `0x01`, DataIndex)

A Coder flag affect a following CoderData existence as following algorithm;

if (Is Complex Coder)
 {
   NUMBER ``NumInStreams``;
   NUMBER ``NumOutStreams``;
 }
 if (There Are Attributes)
 {
   NUMBER ``PropertiesSize``
   BYTE ``Properties[PropertiesSize]``
 }
}
NumBindPairs :  = ``NumOutStreamsTotal`` – 1;
for (``NumBindPairs``)
 {
   NUMBER ``InIndex``;
   NUMBER ``OutIndex``;
 }
NumPackedStreams : ``NumInStreamsTotal`` – ``NumBindPairs``;
 if (``NumPackedStreams`` > 1)
   for(``NumPackedStreams``)
   {
     NUMBER ``Index``;
   };

Appendix: CRC algorithm (normative)

Chunk CRCs are calculated using standard CRC methods with pre and post conditioning, as defined by ISO 3309 [ISO-3309] or ITU-T V.42 [ITU-T-V42]. The CRC polynomial employed is

x^32+x^26+x^23+x^22+x^16+x^12+x^11+x^10+x^8+x^7+x^5+x^4+x^2+x+1

The 32-bit CRC register is initialized to all 1’s, and then the data from each byte is processed from the least significant bit (1) to the most significant bit (128). After all the data bytes are processed, the CRC register is inverted (its ones complement is taken). This value is transmitted (stored in the file) MSB first. For the purpose of separating into bytes and ordering, the least significant bit of the 32-bit CRC is defined to be the coefficient of the x31 term.

Practical calculation of the CRC always employs a precalculated table to greatly accelerate the computation

Appendix: Rationale

Byte order

It has been asked why 7-zip uses little endian byte order. It is a historical reason, that 7-zip was born as Microsoft Windows application in 1999, and its file format was a windows application format, when only little endian was used on target platform.

CRC32

CRC32 is a checksum.

Encode

Encode in this document express compressed, encrypted and/or filter data. When encoding, it should lead encoding metadata.

Extract

Extract in this document express decompress, decryption and/or filter data from archive.

UTF-16-LE

Unicode UTF-16 encoding uses 2 bytes or 4 bytes to represent Unicode characters. Because it is not one byte ordering, we need to consider endian, byte order. UTF-16-LE is a variant of UTF-16 definition which use Little-Endian for store data.

UTF-8

Unicode UTF-8 encoding uses a sequence of bytes, from 1 bytes to 4 bytes to represent Unicode characters. ISO 10646 defines it as 1 byts to 8 bytes encoding, so compliant implementation SHALL be able to handle 8bytes sequence and mark it as invalid.