path: root/man7/regex.7

'\" t
.\" From Henry Spencer's regex package (as found in the apache
.\" distribution). The package carries the following copyright:
.\"
.\"  Copyright 1992, 1993, 1994 Henry Spencer.  All rights reserved.
.\" %%%LICENSE_START(MISC)
.\"  This software is not subject to any license of the American Telephone
.\"  and Telegraph Company or of the Regents of the University of California.
.\"
.\"  Permission is granted to anyone to use this software for any purpose
.\"  on any computer system, and to alter it and redistribute it, subject
.\"  to the following restrictions:
.\"
.\"  1. The author is not responsible for the consequences of use of this
.\"     software, no matter how awful, even if they arise from flaws in it.
.\"
.\"  2. The origin of this software must not be misrepresented, either by
.\"     explicit claim or by omission.  Since few users ever read sources,
.\"     credits must appear in the documentation.
.\"
.\"  3. Altered versions must be plainly marked as such, and must not be
.\"     misrepresented as being the original software.  Since few users
.\"     ever read sources, credits must appear in the documentation.
.\"
.\"  4. This notice may not be removed or altered.
.\" %%%LICENSE_END
.\"
.\" In order to comply with `credits must appear in the documentation'
.\" I added an AUTHOR paragraph below - aeb.
.\"
.\" In the default nroff environment there is no dagger \(dg.
.\"
.\" 2005-05-11 Removed discussion of `[[:<:]]' and `[[:>:]]', which
.\" 	appear not to be in the glibc implementation of regcomp
.\"
.ie t .ds dg \(dg
.el .ds dg (!)
.TH regex 7 2023-11-01 "Linux man-pages 6.7"
.SH NAME
regex \- POSIX.2 regular expressions
.SH DESCRIPTION
Regular expressions ("RE"s),
as defined in POSIX.2, come in two forms:
modern REs (roughly those of
.BR egrep (1);
POSIX.2 calls these "extended" REs)
and obsolete REs (roughly those of
.BR ed (1);
POSIX.2 "basic" REs).
Obsolete REs mostly exist for backward compatibility in some old programs;
they will be discussed at the end.
POSIX.2 leaves some aspects of RE syntax and semantics open;
"\*(dg" marks decisions on these aspects that
may not be fully portable to other POSIX.2 implementations.
.P
A (modern) RE is one\*(dg or more nonempty\*(dg \fIbranches\fR,
separated by \[aq]|\[aq].
It matches anything that matches one of the branches.
.P
A branch is one\*(dg or more \fIpieces\fR, concatenated.
It matches a match for the first, followed by a match for the second,
and so on.
.P
A piece is an \fIatom\fR possibly followed
by a single\*(dg \[aq]*\[aq], \[aq]+\[aq], \[aq]?\[aq], or \fIbound\fR.
An atom followed by \[aq]*\[aq]
matches a sequence of 0 or more matches of the atom.
An atom followed by \[aq]+\[aq]
matches a sequence of 1 or more matches of the atom.
An atom followed by \[aq]?\[aq]
matches a sequence of 0 or 1 matches of the atom.
.P
A \fIbound\fR is \[aq]{\[aq] followed by an unsigned decimal integer,
possibly followed by \[aq],\[aq]
possibly followed by another unsigned decimal integer,
always followed by \[aq]}\[aq].
The integers must lie between 0 and
.B RE_DUP_MAX
(255\*(dg) inclusive,
and if there are two of them, the first may not exceed the second.
An atom followed by a bound containing one integer \fIi\fR
and no comma matches
a sequence of exactly \fIi\fR matches of the atom.
An atom followed by a bound
containing one integer \fIi\fR and a comma matches
a sequence of \fIi\fR or more matches of the atom.
An atom followed by a bound
containing two integers \fIi\fR and \fIj\fR matches
a sequence of \fIi\fR through \fIj\fR (inclusive) matches of the atom.
.P
An atom is a regular expression enclosed in "\fI()\fP"
(matching a match for the regular expression),
an empty set of "\fI()\fP" (matching the null string)\*(dg,
a \fIbracket expression\fR (see below),
\[aq].\[aq] (matching any single character),
\[aq]\[ha]\[aq] (matching the null string at the beginning of a line),
\[aq]$\[aq] (matching the null string at the end of a line),
a \[aq]\e\[aq] followed by one of the characters "\fI\[ha].[$()|*+?{\e\fP"
(matching that character taken as an ordinary character),
a \[aq]\e\[aq] followed by any other character\*(dg
(matching that character taken as an ordinary character,
as if the \[aq]\e\[aq] had not been present\*(dg),
or a single character with no other significance (matching that character).
A \[aq]{\[aq] followed by a character other than a digit
is an ordinary character,
not the beginning of a bound\*(dg.
It is illegal to end an RE with \[aq]\e\[aq].
.P
A \fIbracket expression\fR is a list of characters enclosed in "\fI[]\fP".
It normally matches any single character from the list (but see below).
If the list begins with \[aq]\[ha]\[aq],
it matches any single character
(but see below) \fInot\fR from the rest of the list.
If two characters in the list are separated by \[aq]\-\[aq], this is shorthand
for the full \fIrange\fR of characters between those two (inclusive) in the
collating sequence,
for example, "\fI[0\-9]\fP" in ASCII matches any decimal digit.
It is illegal\*(dg for two ranges to share an
endpoint, for example, "\fIa\-c\-e\fP".
Ranges are very collating-sequence-dependent,
and portable programs should avoid relying on them.
.P
To include a literal \[aq]]\[aq] in the list, make it the first character
(following a possible \[aq]\[ha]\[aq]).
To include a literal \[aq]\-\[aq], make it the first or last character,
or the second endpoint of a range.
To use a literal \[aq]\-\[aq] as the first endpoint of a range,
enclose it in "\fI[.\fP" and "\fI.]\fP"
to make it a collating element (see below).
With the exception of these and some combinations using \[aq][\[aq] (see next
paragraphs), all other special characters, including \[aq]\e\[aq], lose their
special significance within a bracket expression.
.P
Within a bracket expression, a collating element (a character,
a multicharacter sequence that collates as if it were a single character,
or a collating-sequence name for either)
enclosed in "\fI[.\fP" and "\fI.]\fP" stands for the
sequence of characters of that collating element.
The sequence is a single element of the bracket expression's list.
A bracket expression containing a multicharacter collating element
can thus match more than one character,
for example, if the collating sequence includes a "ch" collating element,
then the RE "\fI[[.ch.]]*c\fP" matches the first five characters
of "chchcc".
.P
Within a bracket expression, a collating element enclosed in "\fI[=\fP" and
"\fI=]\fP" is an equivalence class, standing for the sequences of characters
of all collating elements equivalent to that one, including itself.
(If there are no other equivalent collating elements,
the treatment is as if the enclosing delimiters
were "\fI[.\fP" and "\fI.]\fP".)
For example, if o and \(^o are the members of an equivalence class,
then "\fI[[=o=]]\fP", "\fI[[=\(^o=]]\fP",
and "\fI[o\(^o]\fP" are all synonymous.
An equivalence class may not\*(dg be an endpoint
of a range.
.P
Within a bracket expression, the name of a \fIcharacter class\fR enclosed
in "\fI[:\fP" and "\fI:]\fP" stands for the list
of all characters belonging to that
class.
Standard character class names are:
.P
.RS
.TS
l l l.
alnum	digit	punct
alpha	graph	space
blank	lower	upper
cntrl	print	xdigit
.TE
.RE
.P
These stand for the character classes defined in
.BR wctype (3).
A locale may provide others.
A character class may not be used as an endpoint of a range.
.\" As per http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=295666
.\" The following does not seem to apply in the glibc implementation
.\" .P
.\" There are two special cases\*(dg of bracket expressions:
.\" the bracket expressions "\fI[[:<:]]\fP" and "\fI[[:>:]]\fP" match
.\" the null string at the beginning and end of a word respectively.
.\" A word is defined as a sequence of
.\" word characters
.\" which is neither preceded nor followed by
.\" word characters.
.\" A word character is an
.\" .I alnum
.\" character (as defined by
.\" .BR wctype (3))
.\" or an underscore.
.\" This is an extension,
.\" compatible with but not specified by POSIX.2,
.\" and should be used with
.\" caution in software intended to be portable to other systems.
.P
In the event that an RE could match more than one substring of a given
string,
the RE matches the one starting earliest in the string.
If the RE could match more than one substring starting at that point,
it matches the longest.
Subexpressions also match the longest possible substrings, subject to
the constraint that the whole match be as long as possible,
with subexpressions starting earlier in the RE taking priority over
ones starting later.
Note that higher-level subexpressions thus take priority over
their lower-level component subexpressions.
.P
Match lengths are measured in characters, not collating elements.
A null string is considered longer than no match at all.
For example,
"\fIbb*\fP" matches the three middle characters of "abbbc",
"\fI(wee|week)(knights|nights)\fP"
matches all ten characters of "weeknights",
when "\fI(.*).*\fP" is matched against "abc" the parenthesized subexpression
matches all three characters, and
when "\fI(a*)*\fP" is matched against "bc"
both the whole RE and the parenthesized
subexpression match the null string.
.P
If case-independent matching is specified,
the effect is much as if all case distinctions had vanished from the
alphabet.
When an alphabetic that exists in multiple cases appears as an
ordinary character outside a bracket expression, it is effectively
transformed into a bracket expression containing both cases,
for example, \[aq]x\[aq] becomes "\fI[xX]\fP".
When it appears inside a bracket expression, all case counterparts
of it are added to the bracket expression, so that, for example, "\fI[x]\fP"
becomes "\fI[xX]\fP" and "\fI[\[ha]x]\fP" becomes "\fI[\[ha]xX]\fP".
.P
No particular limit is imposed on the length of REs\*(dg.
Programs intended to be portable should not employ REs longer
than 256 bytes,
as an implementation can refuse to accept such REs and remain
POSIX-compliant.
.P
Obsolete ("basic") regular expressions differ in several respects.
\[aq]|\[aq], \[aq]+\[aq], and \[aq]?\[aq] are
ordinary characters and there is no equivalent
for their functionality.
The delimiters for bounds are "\fI\e{\fP" and "\fI\e}\fP",
with \[aq]{\[aq] and \[aq]}\[aq] by themselves ordinary characters.
The parentheses for nested subexpressions are "\fI\e(\fP" and "\fI\e)\fP",
with \[aq](\[aq] and \[aq])\[aq] by themselves ordinary characters.
\[aq]\[ha]\[aq] is an ordinary character except at the beginning of the
RE or\*(dg the beginning of a parenthesized subexpression,
\[aq]$\[aq] is an ordinary character except at the end of the
RE or\*(dg the end of a parenthesized subexpression,
and \[aq]*\[aq] is an ordinary character if it appears at the beginning of the
RE or the beginning of a parenthesized subexpression
(after a possible leading \[aq]\[ha]\[aq]).
.P
Finally, there is one new type of atom, a \fIback reference\fR:
\[aq]\e\[aq] followed by a nonzero decimal digit \fId\fR
matches the same sequence of characters
matched by the \fId\fRth parenthesized subexpression
(numbering subexpressions by the positions of their opening parentheses,
left to right),
so that, for example, "\fI\e([bc]\e)\e1\fP" matches "bb" or "cc" but not "bc".
.SH BUGS
Having two kinds of REs is a botch.
.P
The current POSIX.2 spec says that \[aq])\[aq] is an ordinary character in
the absence of an unmatched \[aq](\[aq];
this was an unintentional result of a wording error,
and change is likely.
Avoid relying on it.
.P
Back references are a dreadful botch,
posing major problems for efficient implementations.
They are also somewhat vaguely defined
(does
"\fIa\e(\e(b\e)*\e2\e)*d\fP" match "abbbd"?).
Avoid using them.
.P
POSIX.2's specification of case-independent matching is vague.
The "one case implies all cases" definition given above
is current consensus among implementors as to the right interpretation.
.\" As per http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=295666
.\" The following does not seem to apply in the glibc implementation
.\" .P
.\" The syntax for word boundaries is incredibly ugly.
.SH AUTHOR
.\" Sigh... The page license means we must have the author's name
.\" in the formatted output.
This page was taken from Henry Spencer's regex package.
.SH SEE ALSO
.BR grep (1),
.BR regex (3)
.P
POSIX.2, section 2.8 (Regular Expression Notation).