Crunch

Revision as of 13:37, 27 September 2020

This article is about the CP/M compressed file format. See the disambiguation section for other "Crunch" formats.

Crunch was a method of compressing single files popular on CP/M, devised by Steve Greenberg circa 1986. It superseded Squeeze and was succeeded by CrLZH, and crunched files were common in LBR archives. The underlying compression uses the LZW algorithm, combined with run-length encoding.

Similar to Squeeze, crunched files were signified in CP/M's 8.3 filename format by replacing the middle letter of the extension with Z (.?Z?), with the extension .ZZZ used for corner cases such as a blank extension.

There are two versions of the compressed data format, and not all decompressors support both. The new format is apparently more common.

Disambiguation

Not to be confused with:

• The "crunched" compression methods used in ARC format. These methods are possibly related to those used by CP/M Crunch, but the file formats are not otherwise similar.
• Crunch-Mania - An Amiga file compression utility
• Cruncher - An executable compression utility for DOS, by Ori Berger [1]
• CRUNCH - A compression optimization utility for DOS, by Bruce Gavin. It does not appear to have any native file formats. [2]
• Crunch - An old ARC compression utility by Richard P. Byrne [3]
• CRUNCH - A PKARC automation utility by Chuck Zulker [4]

See also

• Squeeze - predecessor
• CrLZH - successor
• LBR - container
• ZSQ (LZW compression)‎ - Possibly related

Format details

Note that, as explained in the format documentation, the "filename" field contains not only the filename, but also extension data. If extension data exists, the filename extension is padded with spaces until it is exactly three characters long.

One quirk that does not seem to be documented is that the high bit of every byte in the filename field must be ignored, and treated as if it were 0. It's unclear why, and it's unclear if this rule also applies to the terminating 0x00 (/0x80) byte.

Identification

Files begin with bytes 76 fe.

Specifications

• The file header is described in the text file LZDEF20.DOC shipped with CRUNCH20.LBR.
  • An extracted copy is provided here.
  • Note that the file header follows a similar/compatible structure to CrLZH. It was derived from Squeeze, but bears only a little resemblance to it.
• FIXME: is the exact compression algorithm documented anywhere?
  • CRUNCH20.DOC shipped in CRUNCH20.LBR says: It embodies all of the concepts employed in the UNIX COMPRESS / ARC512 algorithm, but is additionally enhanced by a "metastatic code reassignment" facility. This is one of several concepts I am developing as part of an effort to advance data compression techniques beyond current performance limits. I believe this is the first time this principle has been proposed or implemented.
  • See also "Technical Abstract" by Steven Greenberg, 16 November 1986: CRUNCH.ABS

Tools

• CFX (DOS/Unix)
• lbrate by Russell Marks, c. 2001 (Unix, GPL2)
• On CP/M (or emulators):
  • The canonical tools were CRUNCH and UNCR. Possibly Greenberg's last version (Feb 1988) is v2.4: CRUNCH24.LBR, CRNCH24S.LBR (source code).
  • The later LT31 deals with extracting from all of Squeeze, Crunch, CrLZH and LBR formats. Widely available in CP/M archives, e.g. LT31.LBR
• The Unarchiver

Sample files

• OAK CP/M archive → .../*.?z?
• Found in many LBR files. Note that you may have to tell your LBR utility not to decompress them (e.g. lbrate -n).