1*b30d1939SAndy Fiddaman /***********************************************************************
2*b30d1939SAndy Fiddaman * *
3*b30d1939SAndy Fiddaman * This software is part of the ast package *
4*b30d1939SAndy Fiddaman * Copyright (c) 1985-2012 AT&T Intellectual Property *
5*b30d1939SAndy Fiddaman * and is licensed under the *
6*b30d1939SAndy Fiddaman * Eclipse Public License, Version 1.0 *
7*b30d1939SAndy Fiddaman * by AT&T Intellectual Property *
8*b30d1939SAndy Fiddaman * *
9*b30d1939SAndy Fiddaman * A copy of the License is available at *
10*b30d1939SAndy Fiddaman * http://www.eclipse.org/org/documents/epl-v10.html *
11*b30d1939SAndy Fiddaman * (with md5 checksum b35adb5213ca9657e911e9befb180842) *
12*b30d1939SAndy Fiddaman * *
13*b30d1939SAndy Fiddaman * Information and Software Systems Research *
14*b30d1939SAndy Fiddaman * AT&T Research *
15*b30d1939SAndy Fiddaman * Florham Park NJ *
16*b30d1939SAndy Fiddaman * *
17*b30d1939SAndy Fiddaman * Glenn Fowler <gsf@research.att.com> *
18*b30d1939SAndy Fiddaman * David Korn <dgk@research.att.com> *
19*b30d1939SAndy Fiddaman * Phong Vo <kpv@research.att.com> *
20*b30d1939SAndy Fiddaman * *
21*b30d1939SAndy Fiddaman ***********************************************************************/
22*b30d1939SAndy Fiddaman #pragma prototyped
23*b30d1939SAndy Fiddaman /*
24*b30d1939SAndy Fiddaman * regex collation symbol support
25*b30d1939SAndy Fiddaman */
26*b30d1939SAndy Fiddaman
27*b30d1939SAndy Fiddaman #include "reglib.h"
28*b30d1939SAndy Fiddaman
29*b30d1939SAndy Fiddaman /*
30*b30d1939SAndy Fiddaman * return the collating symbol delimited by [c c], where c is either '=' or '.'
31*b30d1939SAndy Fiddaman * s points to the first char after the initial [
32*b30d1939SAndy Fiddaman * if e!=0 it is set to point to the next char in s on return
33*b30d1939SAndy Fiddaman *
34*b30d1939SAndy Fiddaman * the collating symbol is converted to multibyte in <buf,size>
35*b30d1939SAndy Fiddaman * the return value is:
36*b30d1939SAndy Fiddaman * -1 syntax error / invalid collating element
37*b30d1939SAndy Fiddaman * >=0 size with 0-terminated mb character (*wc != 0)
38*b30d1939SAndy Fiddaman * or collating element (*wc == 0) in buf
39*b30d1939SAndy Fiddaman */
40*b30d1939SAndy Fiddaman
41*b30d1939SAndy Fiddaman int
regcollate(register const char * s,char ** e,char * buf,size_t size,wchar_t * wc)42*b30d1939SAndy Fiddaman regcollate(register const char* s, char** e, char* buf, size_t size, wchar_t* wc)
43*b30d1939SAndy Fiddaman {
44*b30d1939SAndy Fiddaman register int c;
45*b30d1939SAndy Fiddaman register char* b;
46*b30d1939SAndy Fiddaman register char* x;
47*b30d1939SAndy Fiddaman const char* t;
48*b30d1939SAndy Fiddaman int i;
49*b30d1939SAndy Fiddaman int r;
50*b30d1939SAndy Fiddaman int term;
51*b30d1939SAndy Fiddaman wchar_t w;
52*b30d1939SAndy Fiddaman char xfm[256];
53*b30d1939SAndy Fiddaman char tmp[sizeof(xfm)];
54*b30d1939SAndy Fiddaman
55*b30d1939SAndy Fiddaman if (size < 2 || (term = *s) != '.' && term != '=' || !*++s || *s == term && *(s + 1) == ']')
56*b30d1939SAndy Fiddaman goto nope;
57*b30d1939SAndy Fiddaman t = s;
58*b30d1939SAndy Fiddaman w = mbchar(s);
59*b30d1939SAndy Fiddaman if ((r = (s - t)) > 1)
60*b30d1939SAndy Fiddaman {
61*b30d1939SAndy Fiddaman if (*s++ != term || *s++ != ']')
62*b30d1939SAndy Fiddaman goto oops;
63*b30d1939SAndy Fiddaman goto done;
64*b30d1939SAndy Fiddaman }
65*b30d1939SAndy Fiddaman if (*s == term && *(s + 1) == ']')
66*b30d1939SAndy Fiddaman {
67*b30d1939SAndy Fiddaman s += 2;
68*b30d1939SAndy Fiddaman goto done;
69*b30d1939SAndy Fiddaman }
70*b30d1939SAndy Fiddaman b = buf;
71*b30d1939SAndy Fiddaman x = buf + size - 2;
72*b30d1939SAndy Fiddaman s = t;
73*b30d1939SAndy Fiddaman for (;;)
74*b30d1939SAndy Fiddaman {
75*b30d1939SAndy Fiddaman if (!(c = *s++))
76*b30d1939SAndy Fiddaman goto oops;
77*b30d1939SAndy Fiddaman if (c == term)
78*b30d1939SAndy Fiddaman {
79*b30d1939SAndy Fiddaman if (!(c = *s++))
80*b30d1939SAndy Fiddaman goto oops;
81*b30d1939SAndy Fiddaman if (c != term)
82*b30d1939SAndy Fiddaman {
83*b30d1939SAndy Fiddaman if (c != ']')
84*b30d1939SAndy Fiddaman goto oops;
85*b30d1939SAndy Fiddaman break;
86*b30d1939SAndy Fiddaman }
87*b30d1939SAndy Fiddaman }
88*b30d1939SAndy Fiddaman if (b < x)
89*b30d1939SAndy Fiddaman *b++ = c;
90*b30d1939SAndy Fiddaman }
91*b30d1939SAndy Fiddaman r = s - t - 2;
92*b30d1939SAndy Fiddaman w = 0;
93*b30d1939SAndy Fiddaman if (b >= x)
94*b30d1939SAndy Fiddaman goto done;
95*b30d1939SAndy Fiddaman *b = 0;
96*b30d1939SAndy Fiddaman for (i = 0; i < r && i < sizeof(tmp) - 1; i++)
97*b30d1939SAndy Fiddaman tmp[i] = '0';
98*b30d1939SAndy Fiddaman tmp[i] = 0;
99*b30d1939SAndy Fiddaman if (mbxfrm(xfm, buf, sizeof(xfm)) >= mbxfrm(xfm, tmp, sizeof(xfm)))
100*b30d1939SAndy Fiddaman goto nope;
101*b30d1939SAndy Fiddaman t = (const char*)buf;
102*b30d1939SAndy Fiddaman done:
103*b30d1939SAndy Fiddaman if (r <= size && (char*)t != buf)
104*b30d1939SAndy Fiddaman {
105*b30d1939SAndy Fiddaman memcpy(buf, t, r);
106*b30d1939SAndy Fiddaman if (r < size)
107*b30d1939SAndy Fiddaman buf[r] = 0;
108*b30d1939SAndy Fiddaman }
109*b30d1939SAndy Fiddaman if (wc)
110*b30d1939SAndy Fiddaman *wc = w;
111*b30d1939SAndy Fiddaman if (e)
112*b30d1939SAndy Fiddaman *e = (char*)s;
113*b30d1939SAndy Fiddaman return r;
114*b30d1939SAndy Fiddaman oops:
115*b30d1939SAndy Fiddaman s--;
116*b30d1939SAndy Fiddaman nope:
117*b30d1939SAndy Fiddaman if (e)
118*b30d1939SAndy Fiddaman *e = (char*)s;
119*b30d1939SAndy Fiddaman return -1;
120*b30d1939SAndy Fiddaman }
121