swextract

A Python utility for extracting files from IRIX software packages.

Background

IRIX software is distributed as pairs of files: a .idb index file and one or more .sw archive files. The .idb describes every file in the package — its path, permissions, which archive it lives in, and its compressed size. The .sw files are sequential binary archives of LZW-compressed (.Z) or verbatim data, prefixed with a 13-byte file header and 2-byte per-entry overhead.

This tool parses the .idb and extracts each file from the appropriate .sw archive into a directory tree that mirrors the original IRIX filesystem layout.

Requirements

• Python 3.6+
• uncompress (the ncompress package on Debian/Ubuntu: sudo apt install ncompress)

Usage

python3 swextract.py <idb> <sw> <outdir>

Argument	Description
idb	Path to the .idb index file
sw	Path to one of the package's .sw archive files
outdir	Directory to extract files into (created if necessary)

Example

python3 swextract.py /path/to/c_fe.idb /path/to/c_fe.sw /path/to/irixroot

Multiple .sw files

A package may ship several .sw archives. The .idb refers to them by component name — for example, c_fe.sw.c, c_fe.sw64.lib, and c_fe.man.relnotes. The script resolves these by stripping the last extension (c_fe.sw.c → c_fe.sw, c_fe.man.relnotes → c_fe.man) and looking for the resulting filename in the same directory as the provided .sw file. Place all available archives in the same directory and point the script at any one of them; it will find the rest automatically.

Files belonging to an archive that is not present are skipped with a notice.

Output

Files are extracted into outdir preserving the full IRIX path (e.g. outdir/usr/lib/cfe, outdir/usr/lib32/cmplrs/fec). Directories are created as needed.

.sw archive format

Documented here for reference, as it does not appear to be described elsewhere.

• Bytes 0–12: 13-byte file header. The first 12 bytes are an ASCII version string (e.g. im001V620P02); the 13th byte is 0x00.
• Per entry: 2-byte entry header + path + compressed data.
  • Byte 0: entry type/flags (observed as 0x00).
  • Byte 1: path length as a single unsigned byte.
  • Bytes 2–(2+pathlen−1): file path, not null-terminated.
  • Remaining bytes: compressed file data, either LZW (.Z format, magic 1f 9d) or verbatim (when cmpsize is 0 in the .idb).
• Entries are sequential with no padding between them.
• The off() field sometimes present in .idb files gives the absolute byte offset of an entry's 2-byte header within the archive; when absent (as in the packages this tool was developed against), offsets are computed by accumulating 2 + pathlen + cmpsize per entry starting from 13.