Sustainability of Digital Formats: Planning for Library of Congress Collections

RAR, or the Roshal ARchive format thanks to its namesake creator software developer Eugene Roshal, is a proprietary archive file format that supports data compression, error recovery and file spanning. There are at least six main versions and subversions of the RAR format and because the early versions have little formal documentation, only selected versions currently are described at this site. The early versions RAR1.3 and especially RAR1.5 are thought to be the base for later versions but no public documentation is available about these early versions. Comments welcome. RAR4 is also known as RAR version 2.9 in some documentation. The current version as of this writing is RAR5. See Relationships below for more details.

Similar in purpose to ZIP files, RAR files are data containers in which one or more files are stored in compressed form. RAR is a proprietary format under copyright, along with RAR's compression applications and libraries to Alexander Roshal, brother of Eugene Roshal. RAR files are the native format for WinRAR software and can only be created through this tool which is licensed to win.rar GmbH although there are several options to open RAR files. See External Dependencies for details.

Structurally, a RAR file is comprised of variable length blocks of required and optional data. The precise composition of the blocks evolved over time with the versions. At its core, a RAR file is comprised of a marker or introductory block, an archive block which includes the archive header and file header, and closing block which includes additional comments or other information needed to correctly process the file. The order of these blocks may vary, but the first block must be a marker block followed by an archive header block. The archive block is the most complex because it contains the headers for the archive itself as well as the file headers:

• Self-extracting module (optional): Also known as SFX, this means any data preceding the file signature and the block size and contents are not defined.
• The RAR file signature is specific to the version of the format and must be searched for from the beginning of the file past the maximum SFX module size. According to the ad hoc description at RARLab and to PRONOM, the file signatures for RAR2.0, RAR3, RAR4, and RAR5 all begin with "Rar!" (Hex: 52 61 72 21). See subtypes for more details. See File Signatures.
• Archive encryption header (optional): Present only in archives with encrypted headers. Every next header after this one is started from 16 byte AES-256 initialization vector followed by encrypted header data. Size of encrypted header data block is aligned to 16 byte boundary. Encryption version is declared in the tag of the same name with only AES-256 (value = 0) supported for RAR5.
• Main archive header which, among other things, includes the optional Locator tag to quickly access the positions of different service blocks without scanning the entire archive.

After the main archive header but still within the archive block comes one or more file headers, one for each file within the archive. File headers are followed by optional Service headers.

• File header: Includes among other data the Compression Record (stored as values 0 - 5 where 0 means no compression) and an optional Hash record for the standard CRC32 checksum. If another hash algorithm is used, it is stored in the extra area record.
• Service headers: Optional headers that store supplementary information.

The End of Archive Marker follows the last File Header to close out the archive block, after which RAR does not read anything to permit third party tools to add extra information such as a digital signature to archive.

According to the RARLab website, RAR has several advantages over ZIP files including "more convenient multipart (multivolume) archives, tight compression including special solid, multimedia and text modes, strong AES-128 encryption, recovery records helping to repair an archive even in case of physical data damage, Unicode support to process non-English file names and a lot more." In addition, RAR archives usually provide a noticeably higher compression ratio than ZIP file format.