module-regex.md

Module: (curry regex)

v1.2.2 — 2026-06-07

POSIX extended regular expressions using the system <regex.h> (glibc, musl, Darwin). No external library required.

Import

(import (curry regex))

Compiling patterns

(regex-compile pattern [flags]) → regex

Compile a regex pattern string. Returns an opaque regex handle.

flags defaults to REG_EXTENDED. Pass one of the REG_* constants, or combine them with bitwise-or.

(define rx (regex-compile "[A-Z][a-z]+"))
(define rx-i (regex-compile "hello" REG_ICASE))

(regex-free rx)

Release the compiled pattern and free memory. Call this when the regex is no longer needed.

(regex? x) → boolean

Returns #t if x is a regex handle.

Matching

(regex-match rx string [eflags]) → #f | list of (start . end)

Test whether string matches rx. Returns #f on no match, or a list of (start . end) byte-offset pairs — one per capture group, with the full match first (group 0).

(define rx (regex-compile "([a-z]+)@([a-z]+)\\.([a-z]+)"))
(regex-match rx "user@example.com")
; → ((0 . 15) (0 . 4) (5 . 12) (13 . 16))

(regex-match-string rx string [eflags]) → #f | list of strings

Like regex-match but returns matched substrings instead of byte offsets. Unmatched optional groups return #f.

(regex-match-string rx "user@example.com")
; → ("user@example.com" "user" "example" "com")

Replacement

(regex-replace rx string replacement [all?]) → string

Replace the first match (or all matches if all? is #t) with replacement. Use \1–\9 in the replacement string to reference capture groups.

(define rx (regex-compile "([a-z]+)@([a-z]+)"))
(regex-replace rx "foo@bar.com" "\\2@\\1")
; → "bar@foo.com"
(regex-replace (regex-compile "[aeiou]") "hello world" "*" #t)
; → "h*ll* w*rld"

Splitting

(regex-split rx string) → list of strings

Split string on occurrences of rx. Returns a list of the substrings between separators.

(regex-split (regex-compile "[,;]+") "a,b;;c,d")
; → ("a" "b" "c" "d")

Flag constants

Constant	Meaning
REG_EXTENDED	POSIX extended syntax (default)
REG_ICASE	Case-insensitive matching
REG_NOSUB	Do not record match offsets
REG_NEWLINE	^/$ match at embedded newlines
REG_NOTBOL	^ does not match at start of string
REG_NOTEOL	$ does not match at end of string

Resource management

Regex handles are plain pair/bytevector values with no GC finalizer. The underlying regex_t is heap-allocated with malloc. Call (regex-free rx) to release it; if you don't, the memory leaks when the Scheme handle is collected.

Requirements

No external library required — uses the POSIX <regex.h> present on all supported platforms.