Skip to content

acridotheres/hssp

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

55 Commits
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

HSSP

Huge Size Supporting Package


Welcome to the HSSP specification!

If you have any feature requests for future versions, please open an issue.

Read this in other languages:


✨ Now also available on Acridotheres for Developers! ✨


Table of contents


0. About this specification

This is the official HSSP specification, licensed under the MIT License. All dates are in ISO 8601 format (YYYY-MM-DD). If a MurmurHash3 checksum is used, it always uses the number 822616071 (decimal) as seed.

1. Introduction

HSSP is designed to support huge file sizes while being way better than other container formats like ZIP, RAR, etc.

2. Versions

HSSP has multiple versions, each with different features.

2.1. HSSP v1 (Wrongfolder/WFLD)

Release date
2023-03-14

Wrongfolder is the first version of HSSP. It's called Wrongfolder, because @Le0X8 published a wrong folder of the first HSSP implementation in JavaScript (β†’ official JavaScript library).

2.1.1. Features

  • Storing files
  • Content encryption
  • Integrity check
  • "Index file" (you can define a main file of the package)

2.1.2. File structure

The file structure of Wrongfolder is very simple. It consists of a 64 byte header and a body.

2.1.2.1. Header
Offset Length Description Type
0x0 0x4 Magic value: SFA\x00 UTF-8
0x4 0x4 MurmurHash3 checksum of the body (files) Uint32LE
0x8 0x4 Number of contained files Uint32LE
0xC 0x20 Password hash (double SHA-256) (filled with 0x0 if no password is set)
0x2C 0x10 Initialization vector (IV) (filled with 0x0 if no password is set)
0x3C 0x4 "Index file" (main file of the package, e. g. a README) Uint32LE
2.1.2.2. Body

The body consists of multiple files.

A file is structured like this:

Offset Length Description Type
0x0 0x8 File length in Bytes (= FL) Uint64LE
0x8 0x2 Name length in Bytes (= NL) Uint16LE
0xA NL Name of the file (if the file starts with //, the file will be parsed as a folder) UTF-8
0xA+NL FL File contents (if the file is a folder, this will be ignored.)
0xA+NL+FL NL Unused. You can put anything here.
2.1.2.3. Encryption

The encryption is done with AES-256-CBC. The password is hashed with SHA-256 and the result is hashed again with SHA-256. The result is the key.

When encrypting, you have to encrypt the whole body.

2.2. HSSP v2 (Rightfolder/RFLD)

Release date
2023-03-14

2.2.1. New features

none

2.2.2. File structure

The file structure of Rightfolder is very simple. It consists of a 64 byte header and a body.

2.2.2.1. Header
Offset Length Description Type
0x0 0x4 Magic value: HSSP UTF-8
0x4 0x4 MurmurHash3 checksum of the body (files) Uint32LE
0x8 0x4 Number of contained files Uint32LE
0xC 0x20 Password hash (double SHA-256) (filled with 0x0 if no password is set)
0x2C 0x10 Initialization vector (IV) (filled with 0x0 if no password is set)
0x3C 0x4 "Index file" (main file of the package, e. g. a README) Uint32LE
2.2.2.2. Body

The body consists of multiple files.

A file is structured like this:

Offset Length Description Type
0x0 0x8 File length in Bytes (= FL) Uint64LE
0x8 0x2 Name length in Bytes (= NL) Uint16LE
0xA NL Name of the file (if the file starts with //, the file will be parsed as a folder) UTF-8
0xA+NL FL File contents (if the file is a folder, this will be ignored.)
0xA+NL+FL NL Unused. You can put anything here.
2.2.2.3. Encryption

The encryption is done with AES-256-CBC. The password is hashed with SHA-256 and the result is hashed again with SHA-256. The result is the key.

When encrypting, you have to encrypt the whole body.

2.3. HSSP v3 (Doubleheader/DHDR)

Release date
2023-03-14

The header is now 128 bytes long.

2.3.1. New features

none

2.3.2. File structure

The file structure of Doubleheader is very simple. It consists of a 128 byte header and a body.

2.3.2.1. Header
Offset Length Description Type
0x0 0x4 Magic value: HSSP UTF-8
0x4 0x4 MurmurHash3 checksum of the body (files) Uint32LE
0x8 0x4 Number of contained files Uint32LE
0xC 0x20 Password hash (double SHA-256) (filled with 0x0 if no password is set)
0x2C 0x10 Initialization vector (IV) (filled with 0x0 if no password is set)
0x3C 0x4 "Index file" (main file of the package, e. g. a README) Uint32LE
0x40 0x40 Unused. You can put anything here.
2.3.2.2. Body

The body consists of multiple files.

A file is structured like this:

Offset Length Description Type
0x0 0x8 File length in Bytes (= FL) Uint64LE
0x8 0x2 Name length in Bytes (= NL) Uint16LE
0xA NL Name of the file (if the file starts with //, the file will be parsed as a folder) UTF-8
0xA+NL FL File contents (if the file is a folder, this will be ignored.)
0xA+NL+FL NL Unused. You can put anything here.
2.3.2.3. Encryption

The encryption is done with AES-256-CBC. The password is hashed with SHA-256 and the result is hashed again with SHA-256. The result is the key.

When encrypting, you have to encrypt the whole body.

2.4. HSSP v4 (Indexed/IDXD)

Release date
2023-07-11

HSSP Indexed is the first "real update" to HSSP. It now really uses the additional 64 bytes of the header and is more complex than the versions before. Make sure to decrypt the file before decompressing it.

2.4.1. New features

  • Indexing
  • Compression
  • Splitting

2.4.2. Removed features

  • "Index file" β†’ replaced by a "main file" file attribute in the index

2.4.3. File structure

2.4.3.1. Header
Offset Length Description Type
0x0 0x4 Magic value: HSSP UTF-8
0x4 0x1 File standard version (0x4) Uint8
0x5 0x3 Unused. You can put anything here.
0x8 0x4 Number of contained files Uint32LE
0xC 0x20 Password hash (double SHA-256)
0x2C 0x10 Initialization vector (IV)
0x3C 0x4 Used compression algorithm UTF-8
0x40 0x4 Checksum of the body (files) and the index Uint32LE
0x44 0x8 File count of all the files, even those that are not in this package! (may be 0x0 if not splitted) Uint64LE
0x4C 0x8 Offset of the file that has been splitted. (0x0 if not splitted) Uint64LE
0x54 0x4 Checksum of the previous package (0x0 if not splitted or not existing) Uint32LE
0x58 0x4 Checksum of the next package (0x0 if not splitted or not existing) Uint32LE
0x5C 0x4 ID of this package (0x0 if not splitted) (0 is the first part, 1 is the second, ...) Uint32LE
0x60 0x10 Comment UTF-8
0x70 0x10 Generator credits, the official JavaScript library puts hssp 5.0.0 @ npm in here. Acridotheres uses something like core v0.1.0. UTF-8
2.4.3.2. Index

The index is structured like this:

Offset Length Description Type
0x0 0x8 File length in Bytes Uint64LE
0x8 0x2 Name length in Bytes (= NL) Uint16LE
0xA NL Name of the file UTF-8
0xA+NL 0x2 Owner length in Bytes (= OL) Uint16LE
0xC+NL OL Owner of the file UTF-8
0xC+NL+OL 0x2 Owner group length in Bytes (= GL) Uint16LE
0xE+NL+OL GL Owner group of the file UTF-8
0xE+NL+OL+GL 0x4 Web link length in Bytes (= WL) Uint32LE
0x12+NL+OL+GL WL Web link of the file UTF-8
0x12+NL+OL+GL+WL 0x6 File creation time as Unix timestamp (ms) Uint48LE
0x18+NL+OL+GL+WL 0x6 File modification time as Unix timestamp (ms) Uint48LE
0x1E+NL+OL+GL+WL 0x6 File access time as Unix timestamp (ms) Uint48LE
0x24+NL+OL+GL+WL 0x1 + 1 bit File permissions (chmod format) binary
0x25+NL+OL+GL+WL + 1 bit 7 bit File attributes (is folder, is hidden, is system file, enable backup, require backup, is read-only, is main file) binary

Repeat this for every file.

2.4.3.3. Body

The body consists of multiple files.

A file is structured like this:

Offset Length Description Type
0x0 ? File contents

2.5. HSSP v5 (Flagged/FLGD)

Release date
2023-08-24

HSSP Flagged is a small, but important update to HSSP Indexed. It adds flags to determine if compression, encryption, splitting, etc. is used, which makes HSSP more robust because "hash equal to zero" bugs cannot happen anymore.

"Hash equal to zero" bugs happend when a parser thought that the file is not encrypted, because the password hash was equal to zero. This is not possible anymore, because the parser can now check securely if the file is encrypted.

2.5.1. New features

  • Flags

2.5.2. File structure

2.5.2.1. Header
Offset Length Description Type
0x0 0x4 Magic value: HSSP UTF-8
0x4 0x1 File standard version (0x4) Uint8
0x5 0x3 Flags
0x8 0x4 Number of contained files Uint32LE
0xC 0x20 Password hash (double SHA-256)
0x2C 0x10 Initialization vector (IV)
0x3C 0x4 Used compression algorithm UTF-8
0x40 0x4 Checksum of the body (files) Uint32LE
0x44 0x8 File count of all the files, even those that are not in this package! (may be 0x0 if not splitted) Uint64LE
0x4C 0x8 Offset of the file that has been splitted. (0x0 if not splitted) Uint64LE
0x54 0x4 Checksum of the previous package (0x0 if not splitted or not existing) Uint32LE
0x58 0x4 Checksum of the next package (0x0 if not splitted or not existing) Uint32LE
0x5C 0x4 ID of this package (0x0 if not splitted) (0 is the first part, 1 is the second, ...) Uint32LE
0x60 0x10 Comment UTF-8
0x70 0x10 Generator credits, the official JavaScript library puts hssp 5.0.0 @ npm in here. Acridotheres uses something like core v0.1.0. UTF-8
2.5.2.2. Flags
Bit Name Description
0 Is encrypted If this bit is set, the file is encrypted.
1 Is compressed If this bit is set, the file is compressed.
2 Is splitted If this bit is set, the file is splitted.
3 Continues previous split package If this bit is set, the file continues the previous split package.
4 Continues next split package If this bit is set, the file continues the next split package.
5 F6 Unused.
6 F7 Unused.
7 F8 Unused.
8 F9 Unused.
9 F10 Unused.
10 F11 Unused.
11 F12 Unused.
12 F13 Unused.
13 F14 Unused.
14 F15 Unused.
15 F16 Unused.
16 F17 Unused.
17 F18 Unused.
18 F19 Unused.
19 F20 Unused.
20 F21 Unused.
21 F22 Unused.
22 F23 Unused.
23 F24 Unused.
2.5.2.3. Index

The index is structured like this:

Offset Length Description Type
0x0 0x8 File length in Bytes Uint64LE
0x8 0x2 Name length in Bytes (= NL) Uint16LE
0xA NL Name of the file UTF-8
0xA+NL 0x2 Owner length in Bytes (= OL) Uint16LE
0xC+NL OL Owner of the file UTF-8
0xC+NL+OL 0x2 Owner group length in Bytes (= GL) Uint16LE
0xE+NL+OL GL Owner group of the file UTF-8
0xE+NL+OL+GL 0x4 Web link length in Bytes (= WL) Uint32LE
0x12+NL+OL+GL WL Web link of the file UTF-8
0x12+NL+OL+GL+WL 0x6 File creation time as Unix timestamp (ms) Uint48LE
0x18+NL+OL+GL+WL 0x6 File modification time as Unix timestamp (ms) Uint48LE
0x1E+NL+OL+GL+WL 0x6 File access time as Unix timestamp (ms) Uint48LE
0x24+NL+OL+GL+WL 0x1 + 1 bit File permissions (chmod format) binary
0x25+NL+OL+GL+WL + 1 bit 7 bit File attributes (is folder, is hidden, is system file, enable backup, require backup, is read-only, is main file) binary

Repeat this for every file.

2.5.2.4. Body

The body consists of multiple files.

A file is structured like this:

Offset Length Description Type
0x0 ? File contents

2.6. HSSP v6 (Separated/SPRD)

Release date
2023-08-31

This update is another big update to HSSP. It separates the index and the files, which opens possibilities for multithreaded parsing for a massive performance boost, because files can now be decrypted and decompressed at the same time.

Because of the new chaining feature, the file structure is now much more complex than before, but because this feature, the file generation is now much more flexible.

2.6.1. New features

  • Index and files are now encrypted and compressed separately, also every chain now has its own checksum
  • Calculation of the index length is now required
  • File chaining (files can now be chained for encryption and compression) to bundle multiple files together

2.6.2. Removed features

  • Flags F9-F24 because they were unused
  • HSSP v4 compression codes because they were longer than they should be

2.6.3. File structure

2.6.3.1. Header
Offset Length Description Type
0x0 0x4 Magic value: HSSP UTF-8
0x4 0x1 File standard version (0x4) Uint8
0x5 0x1 Flags
0x6 0x2 Index length in bytes (-IE*FC) =(IL) Uint16LE
0x8 0x4 Number of contained files (=FC) Uint32LE
0xC 0x20 Password hash (double SHA-256)
0x2C 0x10 Initialization vector (IV)
0x3C 0x2 Used compression algorithm
0x3E 0x2 Average index entry length (=IE) Uint16LE
0x40 0x4 Checksum of the index Uint32LE
0x44 0x8 File count of all the files, even those that are not in this package! (may be 0x0 if not splitted) Uint64LE
0x4C 0x8 Offset of the file that has been splitted. (0x0 if not splitted) Uint64LE
0x54 0x4 Checksum of the previous package (0x0 if not splitted or not existing) Uint32LE
0x58 0x4 Checksum of the next package (0x0 if not splitted or not existing) Uint32LE
0x5C 0x4 ID of this package (0x0 if not splitted) (0 is the first part, 1 is the second, ...) Uint32LE
0x60 0x10 Comment UTF-8
0x70 0x10 Generator credits, the official JavaScript library puts hssp 5.0.0 @ npm in here. Acridotheres uses something like core v0.1.0. UTF-8
2.6.3.2. Flags
Bit Name Description
0 Is encrypted If this bit is set, the file is encrypted.
1 Is compressed If this bit is set, the file is compressed.
2 Is splitted If this bit is set, the file is splitted.
3 Continues previous split package If this bit is set, the file continues the previous split package.
4 Continues next split package If this bit is set, the file continues the next split package.
5 Compression before encryption If this bit is set, the file is encrypted before instead after compression.
6 F7 Unused.
7 F8 Unused.
2.6.3.3. Index

The index length is calculated like this: IE \* FC + IL, this is now required due to index encryption & compression.

The index is structured like this:

Offset Length Description Type
0x0 0x8 File length in Bytes Uint64LE
0x8 0x2 Name length in Bytes (= NL) Uint16LE
0xA NL Name of the file UTF-8
0xA+NL 0x2 Owner length in Bytes (= OL) Uint16LE
0xC+NL OL Owner of the file UTF-8
0xC+NL+OL 0x2 Owner group length in Bytes (= GL) Uint16LE
0xE+NL+OL GL Owner group of the file UTF-8
0xE+NL+OL+GL 0x4 Web link length in Bytes (= WL) Uint32LE
0x12+NL+OL+GL WL Web link of the file UTF-8
0x12+NL+OL+GL+WL 0x6 File creation time as Unix timestamp (ms) Uint48LE
0x18+NL+OL+GL+WL 0x6 File modification time as Unix timestamp (ms) Uint48LE
0x1E+NL+OL+GL+WL 0x6 File access time as Unix timestamp (ms) Uint48LE
0x24+NL+OL+GL+WL 0x1 + 1 bit File permissions (chmod format) binary
0x25+NL+OL+GL+WL + 1 bit 7 bit File attributes (is folder, is hidden, is system file, enable backup, require backup, is read-only, is main file) binary
0x26+NL+OL+GL+WL 0x1 File attributes 2 (is encrypted, is compressed, connected for encryption, connected for compression, 4 unallocated bits) binary
0x27+NL+OL+GL+WL 0x4 Chain MurmurHash3 checksum binary
0x2B+NL+OL+GL+WL 0x8 File length after encryption & compression Uint64LE

Repeat this for every file.

2.6.3.4. Body

The body consists of multiple files.

A file is structured like this:

Offset Length Description Type
0x0 ? File contents

2.7. HSSP v7 (Cryptography Update/CRYP)

Release date
2024-08-31

This large update adds the option to use a different encryption algorithm than AES-256-CBC, signing, asymmetric encryption support and native support for Acridotheres WebAuth as well as better header organization & even more checksums to verify integrity even better.

Warning

HSSP7 has a very different header structure compared to the previous versions.

2.7.1. New features

  • Added support for different encryption algorithms
  • Added support for signing
  • Added WebAuth support
  • Added support for multiple compression & encryption algorithms in one file
  • Optimized header structure
  • Comments are optional and can be up to 65536 bytes long
  • Generator credits are now replaced by a generator ID (e. g. acridotheres.com β†’ 0x4163) and can be optional (by setting the ID to 0x0).

2.7.2. Start Header

The archive starts with a 9 byte header.

Offset Length Description Type
0x0 0x4 Magic value: HSSP UTF-8
0x4 0x1 File standard version (0x7) Uint8
0x5 0x4 Main header checksum Uint32LE

2.7.3. Main Header

Most of the header fields from the previous versions are now in the main header.

Offset Length Description Type
0x0 0x1 Flags
0x1 0x0/0x2 Index length in bytes (-IE*FC) =(IL); only set if the archive is encrypted or compressed Uint16LE
0x4 Number of contained files Uint32LE
0x0/0x2 Average index entry length (=IE); only set if the archive is encrypted or compressed Uint16LE
0x4 Checksum of the index Uint32LE
0x4 Chain count Uint32LE
0x4 Checksum of the chain list Uint32LE
0x0/0x4 Splitting header checksum; only set if the archive is splitted Uint32LE
0x0/0x4 Number of encryption headers; only set if the archive is encrypted Uint32LE
0x0/0x4 Checksum of the first encryption header; only set if the archive is encrypted Uint32LE
0x0/0x4 Number of compression headers; only set if the archive is compressed Uint32LE
0x0/0x4 Checksum of the first compression header; only set if the archive is compressed Uint32LE
0x0/0x4 Number of signing headers; only set if the archive is signed Uint32LE
0x0/0x4 Checksum of the first signing header; only set if the archive is signed Uint32LE

2.7.4. Flags

When bits 2 and 3 are set, the file is not splitted.

Bit Name Description
0 Is encrypted If this bit is set, the file is encrypted.
1 Is compressed If this bit is set, the file is compressed.
2 Is first If this bit is set, the file is the first part of a splitted archive.
3 Is last If this bit is set, the file is the last part of a splitted archive.
4 Continues previous split package If this bit is set, the file continues the previous split package.
5 Continued in next split package If this bit is set, the file continues the next split package.
6 Is signed If this bit is set, the file is signed.
7 Uses WebAuth If this bit is set, the file uses WebAuth.

2.7.5. Splitting Header

The splitting header is only present if the archive is splitted.

Offset Length Description Type
0x0 0x0/0x4 Checksum of the previous archive; only set if the archive is not the first part Uint32LE
0x0/0x4 Checksum of the next archive; only set if the archive is not the last part Uint32LE
0x0/0x8 Continued file length; only set if the archive is not the last part Uint64LE
0x0/0x8 Continuing file length; only set if the archive is not the first part Uint64LE

2.7.6. Encryption Headers

The encryption headers are only present if the archive is encrypted.

Offset Length Description Type
0x0 0x4 Encryption method ID (to identify encrption config later in the file) Uint32LE
0x4 0x2 Encryption algorithm (see here) Uint16LE
0x6 0x2 Key checksum algorithm (see here) Uint16LE
0x8 ? Key checksum, length depends on the algorithm
? IV, length depends on the algorithm
0x0/0x4 WebAuth URL length; only set if the archive uses WebAuth (has to be set to 0x0 if WebAuth is not used for this method) (=UL) Uint32LE
0x0/UL WebAuth URL UTF-8
0x0/0x4 Next encryption header checksum; only set if there is another encryption header Uint32LE

2.7.7. Compression Headers

The compression headers are only present if the archive is compressed.

Offset Length Description Type
0x0 0x2 Compression method ID (to identify compression config later in the file) Uint32LE
0x2 0x2 Compression algorithm (see here) Uint16LE
0x4 0x4 Next compression header checksum; only set if there is another compression header Uint32LE

2.7.8. Signing Headers

The signing headers are only present if the archive is signed.

Offset Length Description Type
0x0 0x4 Signing method ID (to identify signing config later in the file) Uint32LE
0x4 0x2 Signing algorithm (see here) Uint16LE
0x6 0x4 Public key length (=KL) Uint32LE
0xA KL Public key
0xA+KL 0x0/0x4 Next signing header checksum; only set if there is another signing header Uint32LE

2.7.9. Chain list

The chain list is a list of all chains in the archive.

Offset Length Description Type
0x0 0x4 Chain ID Uint32LE
0x4 0x4 Chain checksum Uint32LE
0x8 0x4 Chain encryption ID Uint32LE
0xC 0x2 Chain compression ID Uint32LE
0xE 0x4 Chain signing ID Uint32LE
0x12 0x4 Chain key checksum Uint32LE

2.7.10. Index

The index starts like this:

Offset Length Description Type
0x0 0x4 Encryption ID Uint32LE
0x4 0x4 Compression ID Uint32LE
0x8 0x4 Signing ID Uint32LE

A file in the index is structured like this:

Offset Length Description Type
0x0 0x8 File length in Bytes Uint64LE
0x8 0x2 Name length in Bytes (= NL) Uint16LE
0xA NL Name of the file UTF-8
0xA+NL 0x2 Owner length in Bytes (= OL) Uint16LE
0xC+NL OL Owner of the file UTF-8
0xC+NL+OL 0x2 Owner group length in Bytes (= GL) Uint16LE
0xE+NL+OL GL Owner group of the file UTF-8
0xE+NL+OL+GL 0x4 Web link length in Bytes (= WL) Uint32LE
0x12+NL+OL+GL WL Web link of the file UTF-8
0x12+NL+OL+GL+WL 0x6 File creation time as Unix timestamp (ms) Uint48LE
0x18+NL+OL+GL+WL 0x6 File modification time as Unix timestamp (ms) Uint48LE
0x1E+NL+OL+GL+WL 0x6 File access time as Unix timestamp (ms) Uint48LE
0x24+NL+OL+GL+WL 0x1 + 1 bit File permissions (chmod format) binary
0x25+NL+OL+GL+WL + 1 bit 7 bit File attributes (is folder, is hidden, is system file, enable backup, require backup, is read-only, is main file) binary
0x26+NL+OL+GL+WL 0x1 File attributes 2 (is encrypted, is compressed, connected for encryption, connected for compression, 4 unallocated bits) binary
0x27+NL+OL+GL+WL 0x4 Chain ID binary
0x2B+NL+OL+GL+WL 0x8 File length after encryption & compression Uint64LE
0x33+NL+OL+GL+WL 0x4 File checksum Uint32LE

2.7.11. Body

The body consists of multiple files.

A file is structured like this:

Offset Length Description Type
0x0 ? File contents

3. Compression algorithms

Name IDXD/FLGD code SPRD/CRYP code
No algorithm NONE not used anymore
LZMA LZMA 0x4950
DEFLATE DFLT 0x4446

4. Encryption algorithms

Name CRYP code Symmetric IV length
AES-256-CBC 0x4136 Yes 0x10

5. Checksum algorithms

Name CRYP code Length
SHA-256 0x5336 0x20
SHA-512 0x5332 0x40

6. Signing algorithms

Name CRYP code
RSA + SHA-512 0x5236

7. Contributors

Leonard Lesinski
Leonard Lesinski

🚧 πŸ“–

Releases

No releases published