Overview
csplit (context split) cuts a file where you tell it to: at a line number or at the first line that matches a pattern. The pieces are named xx00, xx01, and so on. Where split cuts by size, csplit cuts by content, which suits files made of sections.
Background
A csplit command appeared in PWB UNIX, according to the macOS manual, which also says it conforms to POSIX.1-2001. Today the name covers GNU csplit and the csplit that macOS ships. The BusyBox 1.37.0 build used for this page has no csplit applet. They agree on line-number and /pattern/ arguments and on -s and -f.
Examples
Cut before line 4: lines 1 to 3 go in xx00 and the rest in xx01; -s hides the byte counts:
seq 6 > n6.txt && csplit -s n6.txt 4 && paste -sd, xx00 && paste -sd, xx01
Choose the prefix of the file names with -f:
seq 6 > n6.txt && csplit -s -f part_ n6.txt 4 && ls part_*
Cut before the first line that matches a pattern:
printf 'a\nb\nc\nb\n' > words.txt && csplit -s words.txt '/c/' && paste -sd, xx00 && paste -sd, xx01
Options
-s
Do not print the size of each piece.
-f
Use this prefix instead of xx for the names of the pieces.
--suppress-matched
Do not include the matching line in either piece.
Common mistakes
Without -s, csplit prints the size in bytes of every piece, which can look like noise in a script. If the pattern is not found, csplit fails and may leave pieces behind. The pattern is a regular expression and a line number counts from 1. The first piece ends just before the line you name, so the matching line starts the next piece.
Portability
This page has no BusyBox view: BusyBox has no csplit applet. The {*} repeat count and --suppress-matched are GNU features; macOS rejects {*} and needs a number such as {1}. The Common examples were executed on GNU coreutils 9.11 and macOS system tools (see the development guide for the exact check).
About GNU csplit
GNU csplit is part of GNU coreutils, and this page's checks ran coreutils 9.11. {*} repeats the pattern for as long as it matches.
GNU examples
Cut at every line that matches, with a repeat count of {*}:
printf 'a\nb\na\nb\n' > ab.txt && csplit -s ab.txt '/a/' '{*}' && ls xx* | grep -c .
References
Original explanations, with upstream documentation for further reading. FreeBSD references are comparison material, not a claim of Apple version equivalence.