Technical Reference: Base Operating System and Extensions , Volume 2

regexec Subroutine

Purpose

Compares the null-terminated string specified by the value of the String parameter against the compiled basic or extended regular expression Preg, which must have previously been compiled by a call to the regcomp subroutine.

Library

Standard C Library (libc. a)

Syntax

#include <regex.h>

int regexec (Preg, String, NMatch, PMatch, EFlags)
const regex_t * Preg;
const char * String;
size_t  NMatch;
regmatch_t * PMatch;
int  EFlags;

Description

The regexec subroutine compares the null-terminated string in the String parameter with the compiled basic or extended regular expression in the Preg parameter initialized by a previous call to the regcomp subroutine. If a match is found, the regexec subroutine returns a value of 0. The regexec subroutine returns a nonzero value if it finds no match or it finds an error.

If the NMatch parameter has a value of 0, or if the REG_NOSUB flag was set on the call to the regcomp subroutine, the regexec subroutine ignores the PMatch parameter. Otherwise, the PMatch parameter points to an array of at least the number of elements specified by the NMatch parameter. The regexec subroutine fills in the elements of the array pointed to by the PMatch parameter with offsets of the substrings of the String parameter. The offsets correspond to the parenthetic subexpressions of the original pattern parameter that was specified to the regcomp subroutine.

The pmatch.rm_so structure is the byte offset of the beginning of the substring, and the pmatch.rm_eo structure is one greater than the byte offset of the end of the substring. Subexpression i begins at the i th matched open parenthesis, counting from 1. The 0 element of the array corresponds to the entire pattern. Unused elements of the PMatch parameter, up to the value PMatch[NMatch-1], are filled with -1. If more than the number of subexpressions specified by the NMatch parameter (the pattern parameter itself counts as a subexpression), only the first NMatch-1 subexpressions are recorded.

When a basic or extended regular expression is being matched, any given parenthetic subexpression of the pattern parameter might match several different substrings of the String parameter. Otherwise, it might not match any substring even though the pattern as a whole did match.

The following rules are used to determine which substrings to report in the PMatch parameter when regular expressions are matched:

• If a subexpression in a regular expression participated in the match several times, the offset of the last matching substring is reported in the PMatch parameter.
• If a subexpression did not participate in a match, the byte offset in the PMatch parameter is a value of -1. A subexpression does not participate in a match if any of the following are true:
  • An * (asterisk) or \{\} (backslash, left brace, backslash, right brace) appears immediately after the subexpression in a basic regular expression.
  • An * (asterisk), ? (question mark), or { } (left and right braces) appears immediately after the subexpression in an extended regular expression and the subexpression did not match (matched 0 times).
  • A | (pipe) is used in an extended regular expression to select either the subexpression that didn't match or another subexpression, and the other subexpression matched.
• If a subexpression is contained in a subexpression, the data in the PMatch parameter refers to the last such subexpression.
• If a subexpression is contained in a subexpression and the byte offsets in the PMatch parameter have a value of -1, the pointers in the PMatch parameter also have a value of -1.
• If a subexpression matched a zero-length string, the offsets in the PMatch parameter refer to the byte immediately following the matching string.

If the REG_NOSUB flag was set in the cflags parameter in the call to the regcomp subroutine, and the NMatch parameter is not equal to 0 in the call to the regexec subroutine, the content of the PMatch array is unspecified.

If the REG_NEWLINE flag was not set in the cflags parameter when the regcomp subroutine was called, then a new-line character in the pattern or String parameter is treated as an ordinary character. If the REG_NEWLINE flag was set when the regcomp subroutine was called, the new-line character is treated as an ordinary character except as follows:

• A new-line character in the String parameter is not matched by a period outside of a bracket expression or by any form of a nonmatching list. A nonmatching list expression begins with a ^ (circumflex) and specifies a list that matches any character or collating element and the expression in the list after the leading caret. For example, the regular expression [^abc] matches any character except a, b, or c. The circumflex has this special meaning only when it is the first character in the list, immediately following the left bracket.
• A ^ (circumflex) in the pattern parameter, when used to specify expression anchoring, matches the zero-length string immediately after a new-line character in the String parameter, regardless of the setting of the REG_NOTBOL flag.
• A $ (dollar sign) in the pattern parameter, when used to specify expression anchoring, matches the zero-length string immediately before a new-line character in the String parameter, regardless of the setting of the REG_NOTEOL flag.

Parameters

Return Values

On successful completion, the regexec subroutine returns a value of 0 to indicate that the contents of the String parameter matched the contents of the pattern parameter, or to indicate that no match occurred. The REG_NOMATCH error is defined in the regex.h file.

Error Codes

If the regexec subroutine is unsuccessful, it returns a nonzero value indicating the type of problem. The following macros for possible error codes that can be returned are defined in the regex.h file:

If the value of the Preg parameter to the regexec subroutine is not a compiled basic or extended regular expression returned by the regcomp subroutine, the result is undefined.

Examples

The following example demonstrates how the REG_NOTBOL flag can be used with the regexec subroutine to find all substrings in a line that match a pattern supplied by a user. (For simplicity, very little error-checking is done in this example.)

(void) regcomp (&re, pattern, 0) ;
/* this call to regexec finds the first match on the line */
error = regexec (&re, &buffer[0], 1, &pm, 0) ;
while (error = = 0) {   /* while matches found */
<subString found between pm.r._sp and pm.rm_ep>
/* This call to regexec finds the next match */
error = regexec (&re, pm.rm_ep, 1, &pm, REG_NOTBOL) ;

Implementation Specifics

This subroutine is part of Base Operating System (BOS) Runtime.

Related Information

The regcomp (regcomp Subroutine) subroutine, regerror (regerror Subroutine) subroutine, regfree (regfree Subroutine) subroutine.

Subroutines Overview and Understanding Internationalized Regular Expression Subroutines in AIX 5L Version 5.1 General Programming Concepts: Writing and Debugging Programs.