1 /* Internal header for proving correct grouping in strings of numbers.
2    Copyright (C) 1995-2022 Free Software Foundation, Inc.
3    This file is part of the GNU C Library.
4 
5    The GNU C Library is free software; you can redistribute it and/or
6    modify it under the terms of the GNU Lesser General Public
7    License as published by the Free Software Foundation; either
8    version 2.1 of the License, or (at your option) any later version.
9 
10    The GNU C Library is distributed in the hope that it will be useful,
11    but WITHOUT ANY WARRANTY; without even the implied warranty of
12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13    Lesser General Public License for more details.
14 
15    You should have received a copy of the GNU Lesser General Public
16    License along with the GNU C Library; if not, see
17    <https://www.gnu.org/licenses/>.  */
18 
19 #include <limits.h>
20 #include <stddef.h>
21 #include <string.h>
22 
23 #ifndef MAX
24 #define MAX(a,b)	({ typeof(a) _a = (a); typeof(b) _b = (b); \
25 			   _a > _b ? _a : _b; })
26 #endif
27 
28 #ifdef USE_WIDE_CHAR
29 # include <wctype.h>
30 # define L_(Ch) L##Ch
31 # define UCHAR_TYPE wint_t
32 # define STRING_TYPE wchar_t
33 #else
34 # define L_(Ch) Ch
35 # define UCHAR_TYPE unsigned char
36 # define STRING_TYPE char
37 #endif
38 
39 #include "grouping.h"
40 
41 /* Find the maximum prefix of the string between BEGIN and END which
42    satisfies the grouping rules.  It is assumed that at least one digit
43    follows BEGIN directly.  */
44 
45 const STRING_TYPE *
46 #ifdef USE_WIDE_CHAR
__correctly_grouped_prefixwc(const STRING_TYPE * begin,const STRING_TYPE * end,wchar_t thousands,const char * grouping)47 __correctly_grouped_prefixwc (const STRING_TYPE *begin, const STRING_TYPE *end,
48 			      wchar_t thousands,
49 #else
50 __correctly_grouped_prefixmb (const STRING_TYPE *begin, const STRING_TYPE *end,
51 			      const char *thousands,
52 #endif
53 			      const char *grouping)
54 {
55 #ifndef USE_WIDE_CHAR
56   size_t thousands_len;
57   int cnt;
58 #endif
59 
60   if (grouping == NULL)
61     return end;
62 
63 #ifndef USE_WIDE_CHAR
64   thousands_len = strlen (thousands);
65 #endif
66 
67   while (end > begin)
68     {
69       const STRING_TYPE *cp = end - 1;
70       const char *gp = grouping;
71 
72       /* Check first group.  */
73       while (cp >= begin)
74 	{
75 #ifdef USE_WIDE_CHAR
76 	  if (*cp == thousands)
77 	    break;
78 #else
79 	  if (cp[thousands_len - 1] == *thousands)
80 	    {
81 	      for (cnt = 1; thousands[cnt] != '\0'; ++cnt)
82 		if (thousands[cnt] != cp[thousands_len - 1 - cnt])
83 		  break;
84 	      if (thousands[cnt] == '\0')
85 		break;
86 	    }
87 #endif
88 	  --cp;
89 	}
90 
91       /* We allow the representation to contain no grouping at all even if
92 	 the locale specifies we can have grouping.  */
93       if (cp < begin)
94 	return end;
95 
96       if (end - cp == (int) *gp + 1)
97 	{
98 	  /* This group matches the specification.  */
99 
100 	  const STRING_TYPE *new_end;
101 
102 	  if (cp < begin)
103 	    /* There is just one complete group.  We are done.  */
104 	    return end;
105 
106 	  /* CP points to a thousands separator character.  The preceding
107 	     remainder of the string from BEGIN to NEW_END is the part we
108 	     will consider if there is a grouping error in this trailing
109 	     portion from CP to END.  */
110 	  new_end = cp - 1;
111 
112 	  /* Loop while the grouping is correct.  */
113 	  while (1)
114 	    {
115 	      /* Get the next grouping rule.  */
116 	      ++gp;
117 	      if (*gp == 0)
118 		/* If end is reached use last rule.  */
119 	        --gp;
120 
121 	      /* Skip the thousands separator.  */
122 	      --cp;
123 
124 	      if (*gp == CHAR_MAX
125 #if CHAR_MIN < 0
126 		  || *gp < 0
127 #endif
128 		  )
129 	        {
130 	          /* No more thousands separators are allowed to follow.  */
131 	          while (cp >= begin)
132 		    {
133 #ifdef USE_WIDE_CHAR
134 		      if (*cp == thousands)
135 			break;
136 #else
137 		      for (cnt = 0; thousands[cnt] != '\0'; ++cnt)
138 			if (thousands[cnt] != cp[thousands_len - cnt - 1])
139 			  break;
140 		      if (thousands[cnt] == '\0')
141 			break;
142 #endif
143 		      --cp;
144 		    }
145 
146 	          if (cp < begin)
147 		    /* OK, only digits followed.  */
148 		    return end;
149 	        }
150 	      else
151 	        {
152 		  /* Check the next group.  */
153 	          const STRING_TYPE *group_end = cp;
154 
155 		  while (cp >= begin)
156 		    {
157 #ifdef USE_WIDE_CHAR
158 		      if (*cp == thousands)
159 			break;
160 #else
161 		      for (cnt = 0; thousands[cnt] != '\0'; ++cnt)
162 			if (thousands[cnt] != cp[thousands_len - cnt - 1])
163 			  break;
164 		      if (thousands[cnt] == '\0')
165 			break;
166 #endif
167 		      --cp;
168 		    }
169 
170 		  if (cp < begin && group_end - cp <= (int) *gp)
171 		    /* Final group is correct.  */
172 		    return end;
173 
174 		  if (cp < begin || group_end - cp != (int) *gp)
175 		    /* Incorrect group.  Punt.  */
176 		    break;
177 		}
178 	    }
179 
180 	  /* The trailing portion of the string starting at NEW_END
181 	     contains a grouping error.  So we will look for a correctly
182 	     grouped number in the preceding portion instead.  */
183 	  end = new_end;
184 	}
185       else
186 	{
187 	  /* Even the first group was wrong; determine maximum shift.  */
188 	  if (end - cp > (int) *gp + 1)
189 	    end = cp + (int) *gp + 1;
190 	  else if (cp < begin)
191 	    /* This number does not fill the first group, but is correct.  */
192 	    return end;
193 	  else
194 	    /* CP points to a thousands separator character.  */
195 	    end = cp;
196 	}
197     }
198 
199   return MAX (begin, end);
200 }
201