]>
Commit | Line | Data |
---|---|---|
fea681da MK |
1 | .\" From Henry Spencer's regex package (as found in the apache |
2 | .\" distribution). The package carries the following copyright: | |
3 | .\" | |
4 | .\" Copyright 1992, 1993, 1994 Henry Spencer. All rights reserved. | |
5 | .\" This software is not subject to any license of the American Telephone | |
6 | .\" and Telegraph Company or of the Regents of the University of California. | |
c13182ef | 7 | .\" |
fea681da MK |
8 | .\" Permission is granted to anyone to use this software for any purpose |
9 | .\" on any computer system, and to alter it and redistribute it, subject | |
10 | .\" to the following restrictions: | |
c13182ef | 11 | .\" |
fea681da MK |
12 | .\" 1. The author is not responsible for the consequences of use of this |
13 | .\" software, no matter how awful, even if they arise from flaws in it. | |
c13182ef | 14 | .\" |
fea681da MK |
15 | .\" 2. The origin of this software must not be misrepresented, either by |
16 | .\" explicit claim or by omission. Since few users ever read sources, | |
17 | .\" credits must appear in the documentation. | |
c13182ef | 18 | .\" |
fea681da MK |
19 | .\" 3. Altered versions must be plainly marked as such, and must not be |
20 | .\" misrepresented as being the original software. Since few users | |
21 | .\" ever read sources, credits must appear in the documentation. | |
c13182ef | 22 | .\" |
fea681da | 23 | .\" 4. This notice may not be removed or altered. |
c13182ef | 24 | .\" |
fea681da MK |
25 | .\" In order to comply with `credits must appear in the documentation' |
26 | .\" I added an AUTHOR paragraph below - aeb. | |
27 | .\" | |
28 | .\" In the default nroff environment there is no dagger \(dg. | |
bf6c69c9 MK |
29 | .\" |
30 | .\" 2005-05-11 Removed discussion of `[[:<:]]' and `[[:>:]]', which | |
31 | .\" appear not to be in the glibc implementation of regcomp | |
32 | .\" | |
fea681da MK |
33 | .ie t .ds dg \(dg |
34 | .el .ds dg (!) | |
5a25ff94 | 35 | .TH REGEX 7 2007-12-12 "" "Linux Programmer's Manual" |
fea681da | 36 | .SH NAME |
4dec66f9 | 37 | regex \- POSIX.2 regular expressions |
fea681da | 38 | .SH DESCRIPTION |
324633ae | 39 | Regular expressions ("RE"s), |
4dec66f9 | 40 | as defined in POSIX.2, come in two forms: |
fea681da MK |
41 | modern REs (roughly those of |
42 | .IR egrep ; | |
324633ae | 43 | POSIX.2 calls these "extended" REs) |
fea681da MK |
44 | and obsolete REs (roughly those of |
45 | .BR ed (1); | |
324633ae | 46 | POSIX.2 "basic" REs). |
fea681da MK |
47 | Obsolete REs mostly exist for backward compatibility in some old programs; |
48 | they will be discussed at the end. | |
fa203d85 | 49 | POSIX.2 leaves some aspects of RE syntax and semantics open; |
333a424b | 50 | "\*(dg" marks decisions on these aspects that |
fa203d85 | 51 | may not be fully portable to other POSIX.2 implementations. |
fea681da | 52 | .PP |
1f04cc97 | 53 | A (modern) RE is one\*(dg or more non-empty\*(dg \fIbranches\fR, |
333a424b | 54 | separated by \(aq|\(aq. |
fea681da MK |
55 | It matches anything that matches one of the branches. |
56 | .PP | |
57 | A branch is one\*(dg or more \fIpieces\fR, concatenated. | |
58 | It matches a match for the first, followed by a match for the second, etc. | |
59 | .PP | |
60 | A piece is an \fIatom\fR possibly followed | |
333a424b | 61 | by a single\*(dg \(aq*\(aq, \(aq+\(aq, \(aq?\(aq, or \fIbound\fR. |
c45660d7 MK |
62 | An atom followed by \(aq*\(aq |
63 | matches a sequence of 0 or more matches of the atom. | |
64 | An atom followed by \(aq+\(aq | |
65 | matches a sequence of 1 or more matches of the atom. | |
66 | An atom followed by \(aq?\(aq | |
67 | matches a sequence of 0 or 1 matches of the atom. | |
fea681da | 68 | .PP |
333a424b MK |
69 | A \fIbound\fR is \(aq{\(aq followed by an unsigned decimal integer, |
70 | possibly followed by \(aq,\(aq | |
fea681da | 71 | possibly followed by another unsigned decimal integer, |
333a424b | 72 | always followed by \(aq}\(aq. |
097585ed MK |
73 | The integers must lie between 0 and |
74 | .B RE_DUP_MAX | |
75 | (255\*(dg) inclusive, | |
fea681da MK |
76 | and if there are two of them, the first may not exceed the second. |
77 | An atom followed by a bound containing one integer \fIi\fR | |
78 | and no comma matches | |
79 | a sequence of exactly \fIi\fR matches of the atom. | |
80 | An atom followed by a bound | |
81 | containing one integer \fIi\fR and a comma matches | |
82 | a sequence of \fIi\fR or more matches of the atom. | |
83 | An atom followed by a bound | |
84 | containing two integers \fIi\fR and \fIj\fR matches | |
85 | a sequence of \fIi\fR through \fIj\fR (inclusive) matches of the atom. | |
86 | .PP | |
c45660d7 MK |
87 | An atom is a regular expression enclosed in "\fI()\fP" |
88 | (matching a match for the regular expression), | |
333a424b MK |
89 | an empty set of "\fI()\fP" (matching the null string)\*(dg, |
90 | a \fIbracket expression\fR (see below), \(aq.\(aq | |
91 | (matching any single character), \(aq^\(aq (matching the null string at the | |
92 | beginning of a line), \(aq$\(aq (matching the null string at the | |
93 | end of a line), a \(aq\e\(aq followed by one of the characters | |
94 | "\fI^.[$()|*+?{\e\fP" | |
fea681da | 95 | (matching that character taken as an ordinary character), |
333a424b | 96 | a \(aq\e\(aq followed by any other character\*(dg |
fea681da | 97 | (matching that character taken as an ordinary character, |
333a424b | 98 | as if the \(aq\e\(aq had not been present\*(dg), |
fea681da | 99 | or a single character with no other significance (matching that character). |
333a424b | 100 | A \(aq{\(aq followed by a character other than a digit is an ordinary |
fea681da | 101 | character, not the beginning of a bound\*(dg. |
333a424b | 102 | It is illegal to end an RE with \(aq\e\(aq. |
fea681da | 103 | .PP |
333a424b | 104 | A \fIbracket expression\fR is a list of characters enclosed in "\fI[]\fP". |
fea681da | 105 | It normally matches any single character from the list (but see below). |
333a424b | 106 | If the list begins with \(aq^\(aq, |
fea681da MK |
107 | it matches any single character |
108 | (but see below) \fInot\fR from the rest of the list. | |
333a424b | 109 | If two characters in the list are separated by \(aq\-\(aq, this is shorthand |
fea681da MK |
110 | for the full \fIrange\fR of characters between those two (inclusive) in the |
111 | collating sequence, | |
333a424b | 112 | for example, "\fI[0\-9]\fP" in ASCII matches any decimal digit. |
fea681da | 113 | It is illegal\*(dg for two ranges to share an |
333a424b | 114 | endpoint, for example, "\fIa-c-e\fP". |
fea681da MK |
115 | Ranges are very collating-sequence-dependent, |
116 | and portable programs should avoid relying on them. | |
117 | .PP | |
333a424b MK |
118 | To include a literal \(aq]\(aq in the list, make it the first character |
119 | (following a possible \(aq^\(aq). | |
120 | To include a literal \(aq\-\(aq, make it the first or last character, | |
fea681da | 121 | or the second endpoint of a range. |
333a424b | 122 | To use a literal \(aq\-\(aq as the first endpoint of a range, |
c45660d7 MK |
123 | enclose it in "\fI[.\fP" and "\fI.]\fP" |
124 | to make it a collating element (see below). | |
333a424b MK |
125 | With the exception of these and some combinations using \(aq[\(aq (see next |
126 | paragraphs), all other special characters, including \(aq\e\(aq, lose their | |
fea681da MK |
127 | special significance within a bracket expression. |
128 | .PP | |
129 | Within a bracket expression, a collating element (a character, | |
130 | a multi-character sequence that collates as if it were a single character, | |
131 | or a collating-sequence name for either) | |
333a424b | 132 | enclosed in "\fI[.\fP" and "\fI.]\fP" stands for the |
fea681da MK |
133 | sequence of characters of that collating element. |
134 | The sequence is a single element of the bracket expression's list. | |
c13182ef | 135 | A bracket expression containing a multi-character collating element |
fea681da | 136 | can thus match more than one character, |
333a424b MK |
137 | for example, if the collating sequence includes a "ch" collating element, |
138 | then the RE "\fI[[.ch.]]*c\fP" matches the first five characters | |
139 | of "chchcc". | |
fea681da | 140 | .PP |
333a424b MK |
141 | Within a bracket expression, a collating element enclosed in "\fI[=\fP" and |
142 | "\fI=]\fP" is an equivalence class, standing for the sequences of characters | |
fea681da MK |
143 | of all collating elements equivalent to that one, including itself. |
144 | (If there are no other equivalent collating elements, | |
c45660d7 MK |
145 | the treatment is as if the enclosing delimiters |
146 | were "\fI[.\fP" and "\fI.]\fP".) | |
fea681da | 147 | For example, if o and \o'o^' are the members of an equivalence class, |
c45660d7 MK |
148 | then "\fI[[=o=]]\fP", "\fI[[=\o'o^'=]]\fP", |
149 | and "\fI[o\o'o^']\fP" are all synonymous. | |
fea681da MK |
150 | An equivalence class may not\*(dg be an endpoint |
151 | of a range. | |
152 | .PP | |
153 | Within a bracket expression, the name of a \fIcharacter class\fR enclosed | |
c45660d7 MK |
154 | in "\fI[:\fP" and "\fI:]\fP" stands for the list |
155 | of all characters belonging to that | |
fea681da MK |
156 | class. |
157 | Standard character class names are: | |
158 | .PP | |
159 | .RS | |
160 | .nf | |
161 | .ta 3c 6c 9c | |
162 | alnum digit punct | |
163 | alpha graph space | |
164 | blank lower upper | |
165 | cntrl print xdigit | |
166 | .fi | |
167 | .RE | |
168 | .PP | |
169 | These stand for the character classes defined in | |
170 | .BR wctype (3). | |
171 | A locale may provide others. | |
172 | A character class may not be used as an endpoint of a range. | |
bf6c69c9 MK |
173 | .\" As per http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=295666 |
174 | .\" The following does not seem to apply in the glibc implementation | |
175 | .\" .PP | |
176 | .\" There are two special cases\*(dg of bracket expressions: | |
c45660d7 MK |
177 | .\" the bracket expressions "\fI[[:<:]]\fP" and "\fI[[:>:]]\fP" match |
178 | .\" the null string at the beginning and end of a word respectively. | |
bf6c69c9 MK |
179 | .\" A word is defined as a sequence of |
180 | .\" word characters | |
181 | .\" which is neither preceded nor followed by | |
182 | .\" word characters. | |
183 | .\" A word character is an | |
184 | .\" .I alnum | |
185 | .\" character (as defined by | |
186 | .\" .BR wctype (3)) | |
187 | .\" or an underscore. | |
188 | .\" This is an extension, | |
4dec66f9 | 189 | .\" compatible with but not specified by POSIX.2, |
bf6c69c9 MK |
190 | .\" and should be used with |
191 | .\" caution in software intended to be portable to other systems. | |
fea681da MK |
192 | .PP |
193 | In the event that an RE could match more than one substring of a given | |
194 | string, | |
195 | the RE matches the one starting earliest in the string. | |
196 | If the RE could match more than one substring starting at that point, | |
197 | it matches the longest. | |
198 | Subexpressions also match the longest possible substrings, subject to | |
199 | the constraint that the whole match be as long as possible, | |
200 | with subexpressions starting earlier in the RE taking priority over | |
201 | ones starting later. | |
202 | Note that higher-level subexpressions thus take priority over | |
203 | their lower-level component subexpressions. | |
204 | .PP | |
205 | Match lengths are measured in characters, not collating elements. | |
206 | A null string is considered longer than no match at all. | |
207 | For example, | |
333a424b | 208 | "\fIbb*\fP" matches the three middle characters of "abbbc", |
c45660d7 MK |
209 | "\fI(wee|week)(knights|nights)\fP" |
210 | matches all ten characters of "weeknights", | |
333a424b | 211 | when "\fI(.*).*\fP" is matched against "abc" the parenthesized subexpression |
fea681da | 212 | matches all three characters, and |
c45660d7 MK |
213 | when "\fI(a*)*\fP" is matched against "bc" |
214 | both the whole RE and the parenthesized | |
fea681da MK |
215 | subexpression match the null string. |
216 | .PP | |
217 | If case-independent matching is specified, | |
218 | the effect is much as if all case distinctions had vanished from the | |
219 | alphabet. | |
220 | When an alphabetic that exists in multiple cases appears as an | |
221 | ordinary character outside a bracket expression, it is effectively | |
222 | transformed into a bracket expression containing both cases, | |
333a424b | 223 | for example, \(aqx\(aq becomes "\fI[xX]\fP". |
fea681da | 224 | When it appears inside a bracket expression, all case counterparts |
333a424b MK |
225 | of it are added to the bracket expression, so that, for example, "\fI[x]\fP" |
226 | becomes "\fI[xX]\fP" and "\fI[^x]\fP" becomes "\fI[^xX]\fP". | |
fea681da MK |
227 | .PP |
228 | No particular limit is imposed on the length of REs\*(dg. | |
229 | Programs intended to be portable should not employ REs longer | |
230 | than 256 bytes, | |
231 | as an implementation can refuse to accept such REs and remain | |
232 | POSIX-compliant. | |
233 | .PP | |
324633ae | 234 | Obsolete ("basic") regular expressions differ in several respects. |
c45660d7 MK |
235 | \(aq|\(aq, \(aq+\(aq, and \(aq?\(aq are |
236 | ordinary characters and there is no equivalent | |
fea681da | 237 | for their functionality. |
333a424b MK |
238 | The delimiters for bounds are "\fI\e{\fP" and "\fI\e}\fP", |
239 | with \(aq{\(aq and \(aq}\(aq by themselves ordinary characters. | |
240 | The parentheses for nested subexpressions are "\fI\e(\fP" and "\fI\e)\fP", | |
241 | with \(aq(\(aq and \(aq)\(aq by themselves ordinary characters. | |
242 | \(aq^\(aq is an ordinary character except at the beginning of the | |
fea681da | 243 | RE or\*(dg the beginning of a parenthesized subexpression, |
333a424b | 244 | \(aq$\(aq is an ordinary character except at the end of the |
fea681da | 245 | RE or\*(dg the end of a parenthesized subexpression, |
333a424b | 246 | and \(aq*\(aq is an ordinary character if it appears at the beginning of the |
fea681da | 247 | RE or the beginning of a parenthesized subexpression |
333a424b | 248 | (after a possible leading \(aq^\(aq). |
4f020e78 | 249 | .PP |
fea681da | 250 | Finally, there is one new type of atom, a \fIback reference\fR: |
333a424b | 251 | \(aq\e\(aq followed by a non-zero decimal digit \fId\fR |
fea681da MK |
252 | matches the same sequence of characters |
253 | matched by the \fId\fRth parenthesized subexpression | |
254 | (numbering subexpressions by the positions of their opening parentheses, | |
255 | left to right), | |
333a424b | 256 | so that, for example, "\fI\e([bc]\e)\e1\fP" matches "bb" or "cc" but not "bc". |
fea681da MK |
257 | .SH BUGS |
258 | Having two kinds of REs is a botch. | |
259 | .PP | |
333a424b MK |
260 | The current POSIX.2 spec says that \(aq)\(aq is an ordinary character in |
261 | the absence of an unmatched \(aq(\(aq; | |
fea681da MK |
262 | this was an unintentional result of a wording error, |
263 | and change is likely. | |
264 | Avoid relying on it. | |
265 | .PP | |
266 | Back references are a dreadful botch, | |
267 | posing major problems for efficient implementations. | |
268 | They are also somewhat vaguely defined | |
269 | (does | |
333a424b | 270 | "\fIa\e(\e(b\e)*\e2\e)*d\fP" match "abbbd"?). |
fea681da MK |
271 | Avoid using them. |
272 | .PP | |
fa203d85 | 273 | POSIX.2's specification of case-independent matching is vague. |
324633ae | 274 | The "one case implies all cases" definition given above |
fea681da | 275 | is current consensus among implementors as to the right interpretation. |
4f020e78 MK |
276 | .\" As per http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=295666 |
277 | .\" The following does not seem to apply in the glibc implementation | |
278 | .\" .PP | |
279 | .\" The syntax for word boundaries is incredibly ugly. | |
fd7f0a7f MK |
280 | .\" .SH AUTHOR |
281 | .\" This page was taken from Henry Spencer's regex package. | |
e37e3282 | 282 | .SH "SEE ALSO" |
845d36d6 | 283 | .BR grep (1), |
e37e3282 MK |
284 | .BR regex (3) |
285 | .PP | |
286 | POSIX.2, section 2.8 (Regular Expression Notation). |