Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
att
GitHub Repository: att/ast
 Path: blob/master/src/lib/libast/regex/regcoll.c
1810 views
1
/***********************************************************************

2
*                                                                      *

3
*               This software is part of the ast package               *

4
*          Copyright (c) 1985-2012 AT&T Intellectual Property          *

5
*                      and is licensed under the                       *

6
*                 Eclipse Public License, Version 1.0                  *

7
*                    by AT&T Intellectual Property                     *

8
*                                                                      *

9
*                A copy of the License is available at                 *

10
*          http://www.eclipse.org/org/documents/epl-v10.html           *

11
*         (with md5 checksum b35adb5213ca9657e911e9befb180842)         *

12
*                                                                      *

13
*              Information and Software Systems Research               *

14
*                            AT&T Research                             *

15
*                           Florham Park NJ                            *

16
*                                                                      *

17
*                 Glenn Fowler <[email protected]>                  *

18
*                  David Korn <[email protected]>                   *

19
*                   Phong Vo <[email protected]>                    *

20
*                                                                      *

21
***********************************************************************/

22
#pragma prototyped

23
/*

24
 * regex collation symbol support

25
 */

26


27
#include "reglib.h"

28


29
/*

30
 * return the collating symbol delimited by [c c], where c is either '=' or '.'

31
 * s points to the first char after the initial [

32
 * if e!=0 it is set to point to the next char in s on return

33
 *

34
 * the collating symbol is converted to multibyte in <buf,size>

35
 * the return value is:

36
 *	-1	syntax error / invalid collating element

37
 *	>=0	size with 0-terminated mb character (*wc != 0)

38
 *		or collating element (*wc == 0) in buf

39
 */

40


41
int

42
regcollate(register const char* s, char** e, char* buf, size_t size, wchar_t* wc)

43
{

44
	register int			c;

45
	register char*			b;

46
	register char*			x;

47
	const char*			t;

48
	int				i;

49
	int				r;

50
	int				term;

51
	wchar_t				w;

52
	char				xfm[256];

53
	char				tmp[sizeof(xfm)];

54


55
	if (size < 2 || (term = *s) != '.' && term != '=' || !*++s || *s == term && *(s + 1) == ']')

56
		goto nope;

57
	t = s;

58
	w = mbchar(s);

59
	if ((r = (s - t)) > 1)

60
	{

61
		if (*s++ != term || *s++ != ']')

62
			goto oops;

63
		goto done;

64
	}

65
	if (*s == term && *(s + 1) == ']')

66
	{

67
		s += 2;

68
		goto done;

69
	}

70
	b = buf;

71
	x = buf + size - 2;

72
	s = t;

73
	for (;;)

74
	{

75
		if (!(c = *s++))

76
			goto oops;

77
		if (c == term)

78
		{

79
			if (!(c = *s++))

80
				goto oops;

81
			if (c != term)

82
			{

83
				if (c != ']')

84
					goto oops;

85
				break;

86
			}

87
		}

88
		if (b < x)

89
			*b++ = c;

90
	}

91
	r = s - t - 2;

92
	w = 0;

93
	if (b >= x)

94
		goto done;

95
	*b = 0;

96
	for (i = 0; i < r && i < sizeof(tmp) - 1; i++)

97
		tmp[i] = '0';

98
	tmp[i] = 0;

99
	if (mbxfrm(xfm, buf, sizeof(xfm)) >= mbxfrm(xfm, tmp, sizeof(xfm)))

100
		goto nope;

101
	t = (const char*)buf;

102
 done:

103
	if (r <= size && (char*)t != buf)

104
	{

105
		memcpy(buf, t, r);

106
		if (r < size)

107
			buf[r] = 0;

108
	}

109
	if (wc)

110
		*wc = w;

111
	if (e)

112
		*e = (char*)s;

113
	return r;

114
 oops:

115
 	s--;

116
 nope:

117
	if (e)

118
		*e = (char*)s;

119
	return -1;

120
}

121


122