1/* Convert string representation of a number into an integer value.
2
3   Copyright (C) 1991, 1992, 1994, 1995, 1996, 1997, 1998, 1999, 2003, 2005,
4   2006, 2007
5   Free Software Foundation, Inc.
6
7   NOTE: The canonical source of this file is maintained with the GNU C
8   Library.  Bugs can be reported to bug-glibc@gnu.org.
9
10   This program is free software: you can redistribute it and/or modify it
11   under the terms of the GNU General Public License as published by the
12   Free Software Foundation; either version 3 of the License, or any
13   later version.
14
15   This program is distributed in the hope that it will be useful,
16   but WITHOUT ANY WARRANTY; without even the implied warranty of
17   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18   GNU General Public License for more details.
19
20   You should have received a copy of the GNU General Public License
21   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
22
23#ifdef _LIBC
24# define USE_NUMBER_GROUPING
25#else
26# include <config.h>
27#endif
28
29#include <ctype.h>
30#include <errno.h>
31#ifndef __set_errno
32# define __set_errno(Val) errno = (Val)
33#endif
34
35#include <limits.h>
36#include <stddef.h>
37#include <stdlib.h>
38#include <string.h>
39
40#ifdef USE_NUMBER_GROUPING
41# include "../locale/localeinfo.h"
42#endif
43
44/* Nonzero if we are defining `strtoul' or `strtoull', operating on
45   unsigned integers.  */
46#ifndef UNSIGNED
47# define UNSIGNED 0
48# define INT LONG int
49#else
50# define INT unsigned LONG int
51#endif
52
53/* Determine the name.  */
54#ifdef USE_IN_EXTENDED_LOCALE_MODEL
55# if UNSIGNED
56#  ifdef USE_WIDE_CHAR
57#   ifdef QUAD
58#    define strtol __wcstoull_l
59#   else
60#    define strtol __wcstoul_l
61#   endif
62#  else
63#   ifdef QUAD
64#    define strtol __strtoull_l
65#   else
66#    define strtol __strtoul_l
67#   endif
68#  endif
69# else
70#  ifdef USE_WIDE_CHAR
71#   ifdef QUAD
72#    define strtol __wcstoll_l
73#   else
74#    define strtol __wcstol_l
75#   endif
76#  else
77#   ifdef QUAD
78#    define strtol __strtoll_l
79#   else
80#    define strtol __strtol_l
81#   endif
82#  endif
83# endif
84#else
85# if UNSIGNED
86#  ifdef USE_WIDE_CHAR
87#   ifdef QUAD
88#    define strtol wcstoull
89#   else
90#    define strtol wcstoul
91#   endif
92#  else
93#   ifdef QUAD
94#    define strtol strtoull
95#   else
96#    define strtol strtoul
97#   endif
98#  endif
99# else
100#  ifdef USE_WIDE_CHAR
101#   ifdef QUAD
102#    define strtol wcstoll
103#   else
104#    define strtol wcstol
105#   endif
106#  else
107#   ifdef QUAD
108#    define strtol strtoll
109#   endif
110#  endif
111# endif
112#endif
113
114/* If QUAD is defined, we are defining `strtoll' or `strtoull',
115   operating on `long long int's.  */
116#ifdef QUAD
117# define LONG long long
118# define STRTOL_LONG_MIN LONG_LONG_MIN
119# define STRTOL_LONG_MAX LONG_LONG_MAX
120# define STRTOL_ULONG_MAX ULONG_LONG_MAX
121
122/* The extra casts in the following macros work around compiler bugs,
123   e.g., in Cray C 5.0.3.0.  */
124
125/* True if negative values of the signed integer type T use two's
126   complement, ones' complement, or signed magnitude representation,
127   respectively.  Much GNU code assumes two's complement, but some
128   people like to be portable to all possible C hosts.  */
129# define TYPE_TWOS_COMPLEMENT(t) ((t) ~ (t) 0 == (t) -1)
130# define TYPE_ONES_COMPLEMENT(t) ((t) ~ (t) 0 == 0)
131# define TYPE_SIGNED_MAGNITUDE(t) ((t) ~ (t) 0 < (t) -1)
132
133/* True if the arithmetic type T is signed.  */
134# define TYPE_SIGNED(t) (! ((t) 0 < (t) -1))
135
136/* The maximum and minimum values for the integer type T.  These
137   macros have undefined behavior if T is signed and has padding bits.
138   If this is a problem for you, please let us know how to fix it for
139   your host.  */
140# define TYPE_MINIMUM(t) \
141   ((t) (! TYPE_SIGNED (t) \
142	 ? (t) 0 \
143	 : TYPE_SIGNED_MAGNITUDE (t) \
144	 ? ~ (t) 0 \
145	 : ~ (t) 0 << (sizeof (t) * CHAR_BIT - 1)))
146# define TYPE_MAXIMUM(t) \
147   ((t) (! TYPE_SIGNED (t) \
148	 ? (t) -1 \
149	 : ~ (~ (t) 0 << (sizeof (t) * CHAR_BIT - 1))))
150
151# ifndef ULONG_LONG_MAX
152#  define ULONG_LONG_MAX TYPE_MAXIMUM (unsigned long long)
153# endif
154# ifndef LONG_LONG_MAX
155#  define LONG_LONG_MAX TYPE_MAXIMUM (long long int)
156# endif
157# ifndef LONG_LONG_MIN
158#  define LONG_LONG_MIN TYPE_MINIMUM (long long int)
159# endif
160
161# if __GNUC__ == 2 && __GNUC_MINOR__ < 7
162   /* Work around gcc bug with using this constant.  */
163   static const unsigned long long int maxquad = ULONG_LONG_MAX;
164#  undef STRTOL_ULONG_MAX
165#  define STRTOL_ULONG_MAX maxquad
166# endif
167#else
168# define LONG long
169# define STRTOL_LONG_MIN LONG_MIN
170# define STRTOL_LONG_MAX LONG_MAX
171# define STRTOL_ULONG_MAX ULONG_MAX
172#endif
173
174
175/* We use this code also for the extended locale handling where the
176   function gets as an additional argument the locale which has to be
177   used.  To access the values we have to redefine the _NL_CURRENT
178   macro.  */
179#ifdef USE_IN_EXTENDED_LOCALE_MODEL
180# undef _NL_CURRENT
181# define _NL_CURRENT(category, item) \
182  (current->values[_NL_ITEM_INDEX (item)].string)
183# define LOCALE_PARAM , loc
184# define LOCALE_PARAM_PROTO , __locale_t loc
185#else
186# define LOCALE_PARAM
187# define LOCALE_PARAM_PROTO
188#endif
189
190#include <wchar.h>
191
192#ifdef USE_WIDE_CHAR
193# include <wctype.h>
194# define L_(Ch) L##Ch
195# define UCHAR_TYPE wint_t
196# define STRING_TYPE wchar_t
197# ifdef USE_IN_EXTENDED_LOCALE_MODEL
198#  define ISSPACE(Ch) __iswspace_l ((Ch), loc)
199#  define ISALPHA(Ch) __iswalpha_l ((Ch), loc)
200#  define TOUPPER(Ch) __towupper_l ((Ch), loc)
201# else
202#  define ISSPACE(Ch) iswspace (Ch)
203#  define ISALPHA(Ch) iswalpha (Ch)
204#  define TOUPPER(Ch) towupper (Ch)
205# endif
206#else
207# define L_(Ch) Ch
208# define UCHAR_TYPE unsigned char
209# define STRING_TYPE char
210# ifdef USE_IN_EXTENDED_LOCALE_MODEL
211#  define ISSPACE(Ch) __isspace_l ((Ch), loc)
212#  define ISALPHA(Ch) __isalpha_l ((Ch), loc)
213#  define TOUPPER(Ch) __toupper_l ((Ch), loc)
214# else
215#  define ISSPACE(Ch) isspace (Ch)
216#  define ISALPHA(Ch) isalpha (Ch)
217#  define TOUPPER(Ch) toupper (Ch)
218# endif
219#endif
220
221#define INTERNAL(X) INTERNAL1(X)
222#define INTERNAL1(X) __##X##_internal
223#define WEAKNAME(X) WEAKNAME1(X)
224
225#ifdef USE_NUMBER_GROUPING
226/* This file defines a function to check for correct grouping.  */
227# include "grouping.h"
228#endif
229
230
231
232/* Convert NPTR to an `unsigned long int' or `long int' in base BASE.
233   If BASE is 0 the base is determined by the presence of a leading
234   zero, indicating octal or a leading "0x" or "0X", indicating hexadecimal.
235   If BASE is < 2 or > 36, it is reset to 10.
236   If ENDPTR is not NULL, a pointer to the character after the last
237   one converted is stored in *ENDPTR.  */
238
239INT
240INTERNAL (strtol) (const STRING_TYPE *nptr, STRING_TYPE **endptr,
241		   int base, int group LOCALE_PARAM_PROTO)
242{
243  int negative;
244  register unsigned LONG int cutoff;
245  register unsigned int cutlim;
246  register unsigned LONG int i;
247  register const STRING_TYPE *s;
248  register UCHAR_TYPE c;
249  const STRING_TYPE *save, *end;
250  int overflow;
251
252#ifdef USE_NUMBER_GROUPING
253# ifdef USE_IN_EXTENDED_LOCALE_MODEL
254  struct locale_data *current = loc->__locales[LC_NUMERIC];
255# endif
256  /* The thousands character of the current locale.  */
257  wchar_t thousands = L'\0';
258  /* The numeric grouping specification of the current locale,
259     in the format described in <locale.h>.  */
260  const char *grouping;
261
262  if (group)
263    {
264      grouping = _NL_CURRENT (LC_NUMERIC, GROUPING);
265      if (*grouping <= 0 || *grouping == CHAR_MAX)
266	grouping = NULL;
267      else
268	{
269	  /* Figure out the thousands separator character.  */
270# if defined _LIBC || defined _HAVE_BTOWC
271	  thousands = __btowc (*_NL_CURRENT (LC_NUMERIC, THOUSANDS_SEP));
272	  if (thousands == WEOF)
273	    thousands = L'\0';
274# endif
275	  if (thousands == L'\0')
276	    grouping = NULL;
277	}
278    }
279  else
280    grouping = NULL;
281#endif
282
283  if (base < 0 || base == 1 || base > 36)
284    {
285      __set_errno (EINVAL);
286      return 0;
287    }
288
289  save = s = nptr;
290
291  /* Skip white space.  */
292  while (ISSPACE (*s))
293    ++s;
294  if (*s == L_('\0'))
295    goto noconv;
296
297  /* Check for a sign.  */
298  if (*s == L_('-'))
299    {
300      negative = 1;
301      ++s;
302    }
303  else if (*s == L_('+'))
304    {
305      negative = 0;
306      ++s;
307    }
308  else
309    negative = 0;
310
311  /* Recognize number prefix and if BASE is zero, figure it out ourselves.  */
312  if (*s == L_('0'))
313    {
314      if ((base == 0 || base == 16) && TOUPPER (s[1]) == L_('X'))
315	{
316	  s += 2;
317	  base = 16;
318	}
319      else if (base == 0)
320	base = 8;
321    }
322  else if (base == 0)
323    base = 10;
324
325  /* Save the pointer so we can check later if anything happened.  */
326  save = s;
327
328#ifdef USE_NUMBER_GROUPING
329  if (group)
330    {
331      /* Find the end of the digit string and check its grouping.  */
332      end = s;
333      for (c = *end; c != L_('\0'); c = *++end)
334	if ((wchar_t) c != thousands
335	    && ((wchar_t) c < L_('0') || (wchar_t) c > L_('9'))
336	    && (!ISALPHA (c) || (int) (TOUPPER (c) - L_('A') + 10) >= base))
337	  break;
338      if (*s == thousands)
339	end = s;
340      else
341	end = correctly_grouped_prefix (s, end, thousands, grouping);
342    }
343  else
344#endif
345    end = NULL;
346
347  cutoff = STRTOL_ULONG_MAX / (unsigned LONG int) base;
348  cutlim = STRTOL_ULONG_MAX % (unsigned LONG int) base;
349
350  overflow = 0;
351  i = 0;
352  for (c = *s; c != L_('\0'); c = *++s)
353    {
354      if (s == end)
355	break;
356      if (c >= L_('0') && c <= L_('9'))
357	c -= L_('0');
358      else if (ISALPHA (c))
359	c = TOUPPER (c) - L_('A') + 10;
360      else
361	break;
362      if ((int) c >= base)
363	break;
364      /* Check for overflow.  */
365      if (i > cutoff || (i == cutoff && c > cutlim))
366	overflow = 1;
367      else
368	{
369	  i *= (unsigned LONG int) base;
370	  i += c;
371	}
372    }
373
374  /* Check if anything actually happened.  */
375  if (s == save)
376    goto noconv;
377
378  /* Store in ENDPTR the address of one character
379     past the last character we converted.  */
380  if (endptr != NULL)
381    *endptr = (STRING_TYPE *) s;
382
383#if !UNSIGNED
384  /* Check for a value that is within the range of
385     `unsigned LONG int', but outside the range of `LONG int'.  */
386  if (overflow == 0
387      && i > (negative
388	      ? -((unsigned LONG int) (STRTOL_LONG_MIN + 1)) + 1
389	      : (unsigned LONG int) STRTOL_LONG_MAX))
390    overflow = 1;
391#endif
392
393  if (overflow)
394    {
395      __set_errno (ERANGE);
396#if UNSIGNED
397      return STRTOL_ULONG_MAX;
398#else
399      return negative ? STRTOL_LONG_MIN : STRTOL_LONG_MAX;
400#endif
401    }
402
403  /* Return the result of the appropriate sign.  */
404  return negative ? -i : i;
405
406noconv:
407  /* We must handle a special case here: the base is 0 or 16 and the
408     first two characters are '0' and 'x', but the rest are no
409     hexadecimal digits.  This is no error case.  We return 0 and
410     ENDPTR points to the `x`.  */
411  if (endptr != NULL)
412    {
413      if (save - nptr >= 2 && TOUPPER (save[-1]) == L_('X')
414	  && save[-2] == L_('0'))
415	*endptr = (STRING_TYPE *) &save[-1];
416      else
417	/*  There was no number to convert.  */
418	*endptr = (STRING_TYPE *) nptr;
419    }
420
421  return 0L;
422}
423
424/* External user entry point.  */
425
426
427INT
428#ifdef weak_function
429weak_function
430#endif
431strtol (const STRING_TYPE *nptr, STRING_TYPE **endptr,
432	int base LOCALE_PARAM_PROTO)
433{
434  return INTERNAL (strtol) (nptr, endptr, base, 0 LOCALE_PARAM);
435}
436