usual/regex.h File Reference

POSIX regular expession API, provided by either libc or internally. More...

Data Structures
struct	regex_t
	Compiled regex. More...
struct	regmatch_t
	Match location. More...
Defines
Standard flags to regcomp()
#define	REG_EXTENDED
	Use POSIX Extended Regex Syntax instead of Basic Syntax.
#define	REG_ICASE
	Do case-insensitive matching.
#define	REG_NOSUB
	Do case-insensitive matching.
#define	REG_NEWLINE
	Do case-insensitive matching.
Standard flags to regexec()
#define	REG_NOTBOL
	The start of string is not beginning of line, so ^ should not match.
#define	REG_NOTEOL
	The end of string is not end of line, so $ should not match.
Standard error codes
#define	REG_NOMATCH
	Match not found.
#define	REG_BADBR
	Bad {} repeat specification.
#define	REG_BADPAT
	General problem with regular expression.
#define	REG_BADRPT
	Repeat used without preceding non-repeat element.
#define	REG_EBRACE
	Syntax error with {}.
#define	REG_EBRACK
	Syntax error with [].
#define	REG_ECOLLATE
	Bad collation reference.
#define	REG_ECTYPE
	Bad character class reference.
#define	REG_EESCAPE
	Trailing backslack.
#define	REG_EPAREN
	Syntax error with ()
#define	REG_ERANGE
	Bad endpoint in range.
#define	REG_ESPACE
	No memory.
#define	REG_ESUBREG
	Bad subgroup reference.
Other defines
#define	RE_DUP_MAX
	Max count user can enter via {}.
Non-standard flags for regcomp()
#define	REG_RELAXED_SYNTAX
	Allow few common non-standard escapes:
#define	REG_RELAXED_MATCHING
	Dont permute groups in attempt to get longest match.
#define	REG_RELAXED
	Turn on both REG_RELAXED_SYNTAX and REG_RELAXED_MATCHING.
Typedefs
typedef long	regoff_t
	Type for offset in match.
Functions
int	regcomp (regex_t *rx, const char *re, int flags)
	Compile regex.
int	regexec (const regex_t *rx, const char *str, size_t nmatch, regmatch_t pmatch[], int eflags)
	Execute regex on a string.
size_t	regerror (int err, const regex_t *rx, char *dst, size_t dstlen)
	Give error description.
void	regfree (regex_t *rx)
	Free resources allocated by regcomp().

---

Detailed Description

POSIX regular expession API, provided by either libc or internally.

The internal regex engine is only activated if OS does not provide <regex.h> (eg. Windows) or if --with-internal-regex is used when configuring libusual.

Features of internal regex (uregex).

Simple recursive matcher, only features are small size and POSIX compatibility. Supports both Extended Regular Expressions (ERE) and Basic Regular Expressions (BRE).

Supported syntax

   Both: . * ^ $ [] [[:cname:]]
   ERE: () {} | + ?
   BRE: \(\) \{\} \1-9

With REG_RELAXED_SYNTAX, following common escapes will be available:

    Both: \b\B\d\D\s\S\w\W
    BRE:  \|
    ERE:  \1-9

With REG_RELAXED_MATCHING it returns the first match found after applying leftmost-longest to all elements. It skips the combinatorics to turn it into guaranteed-longest match.

Skipped POSIX features

• collation classes: [[. .]]
• equivalence classes: [[= =]]
• char ranges by locale order: [a-z] (byte order will be used)
• multi-byte chars: UTF-8

Global defines

• USUAL_RELAXED_REGEX
• USE_INTERNAL_REGEX

Compatibility

• POSIX-2008 <regex.h> spec - by default uRegex run in mode where only features specified by POSIX are available.

• AT&T Research regex(3) regression tests - uRegex follows the interpretation given there and fully passes the testsuite.

---

Define Documentation

#define REG_EXTENDED

Use POSIX Extended Regex Syntax instead of Basic Syntax.

#define REG_ICASE

Do case-insensitive matching.

#define REG_NOSUB

Do case-insensitive matching.

#define REG_NEWLINE

Do case-insensitive matching.

#define REG_NOTBOL

The start of string is not beginning of line, so ^ should not match.

#define REG_NOTEOL

The end of string is not end of line, so $ should not match.

#define REG_NOMATCH

Match not found.

#define REG_BADBR

Bad {} repeat specification.

#define REG_BADPAT

General problem with regular expression.

#define REG_BADRPT

Repeat used without preceding non-repeat element.

#define REG_EBRACE

Syntax error with {}.

#define REG_EBRACK

Syntax error with [].

#define REG_ECOLLATE

Bad collation reference.

#define REG_ECTYPE

Bad character class reference.

#define REG_EESCAPE

Trailing backslack.

#define REG_EPAREN

Syntax error with ()

#define REG_ERANGE

Bad endpoint in range.

#define REG_ESPACE

No memory.

#define REG_ESUBREG

Bad subgroup reference.

#define RE_DUP_MAX

Max count user can enter via {}.

#define REG_RELAXED_SYNTAX

Allow few common non-standard escapes:

   \b - word-change
   \B - not word change
   \d - digit
   \D - non-digit
   \s - space
   \S - non-space
   \w - word char
   \W - non-word char
   \/ - /

#define REG_RELAXED_MATCHING

Dont permute groups in attempt to get longest match.

May give minor speed win at the expense of strict POSIX compatibility.

#define REG_RELAXED

Turn on both REG_RELAXED_SYNTAX and REG_RELAXED_MATCHING.

---

Typedef Documentation

typedef long regoff_t

Type for offset in match.

---

Function Documentation

int regcomp	(	regex_t *	rx,
		const char *	re,
		int	flags
	)

Compile regex.

Parameters:

rx	Pre-allocated regex_t structure to fill.
re	Regex as zero-terminated string.
flags	See above for regcomp() flags.

int regexec	(	const regex_t *	rx,
		const char *	str,
		size_t	nmatch,
		regmatch_t	pmatch[],
		int	eflags
	)

Execute regex on a string.

Parameters:

rx	Regex previously initialized with regcomp()
str	Zero-terminated string to match
nmatch	Number of matches in pmatch
pmatch	Array of matches.
eflags	Execution flags. Supported flags: REG_NOTBOL, REG_NOTEOL

size_t regerror	(	int	err,
		const regex_t *	rx,
		char *	dst,
		size_t	dstlen
	)

Give error description.

Parameters:

err	Error code returned by regcomp() or regexec()
rx	Regex structure used in regcomp() or regexec()
dst	Destination buffer
dstlen	Size of dst

void regfree

(

regex_t *

rx

)

Free resources allocated by regcomp().

Parameters:

rx	Regex previously filled by regcomp()