1 /* Copyright (C) 1991, 92, 93, 94, 95, 96, 97 Free Software Foundation, Inc.
2 This file is part of the GNU C Library.
4 The GNU C Library is free software; you can redistribute it and/or
5 modify it under the terms of the GNU Library General Public License as
6 published by the Free Software Foundation; either version 2 of the
7 License, or (at your option) any later version.
9 The GNU C Library is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 Library General Public License for more details.
14 You should have received a copy of the GNU Library General Public
15 License along with the GNU C Library; see the file COPYING.LIB. If not,
16 write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
17 Boston, MA 02111-1307, USA. */
19 #include "../locale/localeinfo.h"
28 #include <libc-lock.h>
32 #define LONGLONG long long
37 /* Those are flags in the conversion format. */
38 # define LONG 0x001 /* l: long or double */
39 # define LONGDBL 0x002 /* L: long long or long double */
40 # define SHORT 0x004 /* h: short */
41 # define SUPPRESS 0x008 /* *: suppress assignment */
42 # define POINTER 0x010 /* weird %p pointer (`fake hex') */
43 # define NOSKIP 0x020 /* do not skip blanks */
44 # define WIDTH 0x040 /* width was given */
45 # define GROUP 0x080 /* ': group numbers */
46 # define MALLOC 0x100 /* a: malloc strings */
48 # define TYPEMOD (LONG|LONGDBL|SHORT)
56 # define va_list _IO_va_list
57 # define ungetc(c, s) (--read_in, _IO_ungetc (c, s))
58 # define inchar() ((c = _IO_getc_unlocked (s)), (void) ++read_in, c)
59 # define encode_error() do { \
60 if (errp != NULL) *errp |= 4; \
61 _IO_funlockfile (s); \
62 __set_errno (EILSEQ); \
65 # define conv_error() do { \
66 if (errp != NULL) *errp |= 2; \
67 _IO_funlockfile (s); \
70 # define input_error() do { \
71 _IO_funlockfile (s); \
72 if (errp != NULL) *errp |= 1; \
75 # define memory_error() do { \
76 _IO_funlockfile (s); \
77 __set_errno (ENOMEM); \
80 # define ARGCHECK(s, format) \
83 /* Check file argument for consistence. */ \
84 CHECK_FILE (s, EOF); \
85 if (s->_flags & _IO_NO_READS) \
87 __set_errno (EBADF); \
90 else if (format == NULL) \
96 # define LOCK_STREAM(S) \
97 __libc_cleanup_region_start ((void (*) (void *)) &_IO_funlockfile, (S)); \
99 # define UNLOCK_STREAM __libc_cleanup_region_end (1)
101 # define ungetc(c, s) (--read_in, ungetc (c, s))
102 # define inchar() ((c = getc (s)), (void) ++read_in, c)
103 # define encode_error() do { \
105 __set_errno (EILSEQ); \
108 # define conv_error() do { \
112 # define input_error() do { \
114 return done ?: EOF; \
116 # define memory_error() do { \
118 __set_errno (ENOMEM); \
121 # define ARGCHECK(s, format) \
124 /* Check file argument for consistence. */ \
125 if (!__validfp (s) || !s->__mode.__read) \
127 __set_errno (EBADF); \
130 else if (format == NULL) \
132 __set_errno (EINVAL); \
137 /* XXX For now !!! */
138 # define flockfile(S) /* nothing */
139 # define funlockfile(S) /* nothing */
140 # define LOCK_STREAM(S)
141 # define UNLOCK_STREAM
143 # define LOCK_STREAM(S) \
144 __libc_cleanup_region_start (&__funlockfile, (S)); \
146 # define UNLOCK_STREAM __libc_cleanup_region_end (1)
151 /* Read formatted input from S according to the format string
152 FORMAT, using the argument list in ARG.
153 Return the number of assignments made, or -1 for an input error. */
156 _IO_vfscanf (s, format, argptr, errp)
163 __vfscanf (FILE *s, const char *format, va_list argptr)
166 va_list arg = (va_list) argptr;
168 register const char *f = format;
169 register unsigned char fc; /* Current character of the format. */
170 register size_t done = 0; /* Assignments done. */
171 register size_t read_in = 0; /* Chars read in. */
172 register int c = 0; /* Last char read. */
173 register int width; /* Maximum field width. */
174 register int flags; /* Modifiers for current format element. */
176 /* Status for reading F-P nums. */
177 char got_dot, got_e, negative;
178 /* If a [...] is a [^...]. */
180 /* Base for integral numbers. */
182 /* Signedness for integral numbers. */
184 /* Decimal point character. */
186 /* The thousands character of the current locale. */
188 /* Integral holding variables. */
192 unsigned long long int uq;
194 unsigned long int ul;
196 /* Character-buffer pointer. */
198 wchar_t *wstr = NULL;
199 char **strptr = NULL;
201 /* We must not react on white spaces immediately because they can
202 possibly be matched even if in the input stream no character is
203 available anymore. */
206 char *tw; /* Temporary pointer. */
207 char *wp = NULL; /* Workspace. */
208 size_t wpmax = 0; /* Maximal size of workspace. */
209 size_t wpsize; /* Currently used bytes in workspace. */
213 if (wpsize == wpmax) \
216 wpmax = UCHAR_MAX > 2 * wpmax ? UCHAR_MAX : 2 * wpmax; \
217 wp = (char *) alloca (wpmax); \
219 memcpy (wp, old, wpsize); \
221 wp[wpsize++] = (Ch); \
225 ARGCHECK (s, format);
227 /* Figure out the decimal point character. */
228 if (mbtowc (&decimal, _NL_CURRENT (LC_NUMERIC, DECIMAL_POINT),
229 strlen (_NL_CURRENT (LC_NUMERIC, DECIMAL_POINT))) <= 0)
230 decimal = (wchar_t) *_NL_CURRENT (LC_NUMERIC, DECIMAL_POINT);
231 /* Figure out the thousands separator character. */
232 if (mbtowc (&thousands, _NL_CURRENT (LC_NUMERIC, THOUSANDS_SEP),
233 strlen (_NL_CURRENT (LC_NUMERIC, THOUSANDS_SEP))) <= 0)
234 thousands = (wchar_t) *_NL_CURRENT (LC_NUMERIC, THOUSANDS_SEP);
236 /* Lock the stream. */
239 /* Run through the format string. */
243 /* Extract the next argument, which is of type TYPE.
244 For a %N$... spec, this is the Nth argument from the beginning;
245 otherwise it is the next argument after the state now in ARG. */
247 /* XXX Possible optimization. */
248 # define ARG(type) (argpos == 0 ? va_arg (arg, type) : \
249 ({ va_list arg = (va_list) argptr; \
250 arg = (va_list) ((char *) arg \
252 * __va_rounded_size (void *)); \
253 va_arg (arg, type); \
256 # define ARG(type) (argpos == 0 ? va_arg (arg, type) : \
257 ({ unsigned int pos = argpos; \
258 va_list arg = (va_list) argptr; \
260 (void) va_arg (arg, void *); \
261 va_arg (arg, type); \
267 /* Non-ASCII, may be a multibyte. */
268 int len = mblen (f, strlen (f));
290 /* Remember to skip spaces. */
297 /* Read a character. */
300 /* Characters other than format specs must just match. */
304 /* We saw white space char as the last character in the format
305 string. Now it's time to skip all leading white space. */
309 if (inchar () == EOF && errno == EINTR)
323 /* This is the start of the conversion string. */
326 /* Initialize state of modifiers. */
329 /* Prepare temporary buffer. */
332 /* Check for a positional parameter specification. */
337 argpos = argpos * 10 + (*f++ - '0');
342 /* Oops; that was actually the field width. */
350 /* Check for the assignment-suppressing and the number grouping flag. */
351 while (*f == '*' || *f == '\'')
362 /* We have seen width. */
366 /* Find the maximum field width. */
377 /* Check for type modifiers. */
378 while (*f == 'h' || *f == 'l' || *f == 'L' || *f == 'a' || *f == 'q')
382 /* int's are short int's. */
384 /* Signal illegal format element. */
389 if (flags & (SHORT|LONGDBL))
391 else if (flags & LONG)
393 /* A double `l' is equivalent to an `L'. */
398 /* int's are long int's. */
403 /* double's are long double's, and int's are long long int's. */
405 /* Signal illegal format element. */
411 /* Signal illegal format element. */
413 /* String conversions (%s, %[) take a `char **'
414 arg and fill it in with a malloc'd pointer. */
419 /* End of the format string? */
423 /* We must take care for EINTR errors. */
424 if (c == EOF && errno == EINTR)
427 /* Find the conversion specifier. */
429 if (skip_space || (fc != '[' && fc != 'c' && fc != 'C' && fc != 'n'))
431 /* Eat whitespace. */
433 if (inchar () == EOF && errno == EINTR)
442 case '%': /* Must match a literal '%'. */
451 case 'n': /* Answer number of assignments done. */
452 /* Corrigendum 1 to ISO C 1990 describes the allowed flags
453 with the 'n' conversion specifier. */
454 if (!(flags & SUPPRESS))
456 /* Don't count the read-ahead. */
458 *ARG (long long int *) = read_in;
459 else if (flags & LONG)
460 *ARG (long int *) = read_in;
461 else if (flags & SHORT)
462 *ARG (short int *) = read_in;
464 *ARG (int *) = read_in;
469 case 'c': /* Match characters. */
470 if ((flags & LONG) == 0)
472 if (!(flags & SUPPRESS))
486 if (!(flags & SUPPRESS))
490 while (--width > 0 && inchar () != EOF);
493 while (--width > 0 && inchar () != EOF);
496 /* I.e., EOF was read. */
499 if (!(flags & SUPPRESS))
506 /* Get UTF-8 encoded wide character. Here we assume (as in
507 other parts of the libc) that we only have to handle
514 if (!(flags & SUPPRESS))
516 wstr = ARG (wchar_t *);
523 #define NEXT_WIDE_CHAR(First) \
526 /* EOF is only an error for the first character. */ \
537 if ((c & 0xc0) == 0x80 || (c & 0xfe) == 0xfe) \
539 if ((c & 0xe0) == 0xc0) \
541 /* We expect two bytes. */ \
545 else if ((c & 0xf0) == 0xe0) \
547 /* We expect three bytes. */ \
551 else if ((c & 0xf8) == 0xf0) \
553 /* We expect four bytes. */ \
557 else if ((c & 0xfc) == 0xf8) \
559 /* We expect five bytes. */ \
565 /* We expect six bytes. */ \
574 || (c & 0xc0) == 0x80 || (c & 0xfe) == 0xfe) \
582 if (!(flags & SUPPRESS)) \
586 NEXT_WIDE_CHAR (first);
591 /* I.e., EOF was read. */
594 if (!(flags & SUPPRESS))
599 case 's': /* Read a string. */
601 /* We have to process a wide character string. */
602 goto wide_char_string;
604 #define STRING_ARG(Str, Type) \
605 if (!(flags & SUPPRESS)) \
607 if (flags & MALLOC) \
609 /* The string is to be stored in a malloc'd buffer. */ \
610 strptr = ARG (char **); \
611 if (strptr == NULL) \
613 /* Allocate an initial buffer. */ \
615 *strptr = malloc (strsize * sizeof (Type)); \
616 Str = (Type *) *strptr; \
619 Str = ARG (Type *); \
623 STRING_ARG (str, char);
636 #define STRING_ADD_CHAR(Str, c, Type) \
637 if (!(flags & SUPPRESS)) \
640 if ((flags & MALLOC) && (char *) Str == *strptr + strsize) \
642 /* Enlarge the buffer. */ \
643 Str = realloc (*strptr, strsize * 2 * sizeof (Type)); \
646 /* Can't allocate that much. Last-ditch effort. */\
647 Str = realloc (*strptr, \
648 (strsize + 1) * sizeof (Type)); \
651 /* We lose. Oh well. \
652 Terminate the string and stop converting, \
653 so at least we don't skip any input. */ \
654 ((Type *) (*strptr))[strsize] = '\0'; \
660 *strptr = (char *) Str; \
661 Str = ((Type *) *strptr) + strsize; \
667 *strptr = (char *) Str; \
668 Str = ((Type *) *strptr) + strsize; \
673 STRING_ADD_CHAR (str, c, char);
674 } while ((width <= 0 || --width > 0) && inchar () != EOF);
679 if (!(flags & SUPPRESS))
687 /* Wide character string. */
692 STRING_ARG (wstr, wchar_t);
697 NEXT_WIDE_CHAR (first);
701 /* XXX We would have to push back the whole wide char
702 with possibly many bytes. But since scanf does
703 not make a difference for white space characters
704 we can simply push back a simple <SP> which is
705 guaranteed to be in the [:space:] class. */
710 STRING_ADD_CHAR (wstr, val, wchar_t);
713 while (width <= 0 || --width > 0);
715 if (!(flags & SUPPRESS))
723 case 'x': /* Hexadecimal integer. */
724 case 'X': /* Ditto. */
729 case 'o': /* Octal integer. */
734 case 'u': /* Unsigned decimal integer. */
739 case 'd': /* Signed decimal integer. */
744 case 'i': /* Generic number. */
753 /* Check for a sign. */
754 if (c == '-' || c == '+')
762 /* Look for a leading indication of base. */
763 if (width != 0 && c == '0')
771 if (width != 0 && tolower (c) == 'x')
789 /* Read the number into workspace. */
790 while (c != EOF && width != 0)
792 if (base == 16 ? !isxdigit (c) :
793 ((!isdigit (c) || c - '0' >= base) &&
794 !((flags & GROUP) && base == 10 && c == thousands)))
803 /* The just read character is not part of the number anymore. */
807 (wpsize == 1 && (wp[0] == '+' || wp[0] == '-')))
808 /* There was no number. */
811 /* Convert the number. */
816 num.q = __strtoq_internal (wp, &tw, base, flags & GROUP);
818 num.uq = __strtouq_internal (wp, &tw, base, flags & GROUP);
823 num.l = __strtol_internal (wp, &tw, base, flags & GROUP);
825 num.ul = __strtoul_internal (wp, &tw, base, flags & GROUP);
830 if (!(flags & SUPPRESS))
835 *ARG (unsigned LONGLONG int *) = num.uq;
836 else if (flags & LONG)
837 *ARG (unsigned long int *) = num.ul;
838 else if (flags & SHORT)
839 *ARG (unsigned short int *)
840 = (unsigned short int) num.ul;
842 *ARG (unsigned int *) = (unsigned int) num.ul;
847 *ARG (LONGLONG int *) = num.q;
848 else if (flags & LONG)
849 *ARG (long int *) = num.l;
850 else if (flags & SHORT)
851 *ARG (short int *) = (short int) num.l;
853 *ARG (int *) = (int) num.l;
859 case 'e': /* Floating-point numbers. */
868 /* Check for a sign. */
869 if (c == '-' || c == '+')
872 if (inchar () == EOF)
873 /* EOF is only an input error before we read any chars. */
886 else if (got_e && wp[wpsize - 1] == 'e'
887 && (c == '-' || c == '+'))
889 else if (wpsize > 0 && !got_e && tolower (c) == 'e')
894 else if (c == decimal && !got_dot)
899 else if ((flags & GROUP) && c == thousands && !got_dot)
906 while (inchar () != EOF && width != 0);
908 /* The last read character is not part of the number anymore. */
914 /* Convert the number. */
918 long double d = __strtold_internal (wp, &tw, flags & GROUP);
919 if (!(flags & SUPPRESS) && tw != wp)
920 *ARG (long double *) = negative ? -d : d;
922 else if (flags & LONG)
924 double d = __strtod_internal (wp, &tw, flags & GROUP);
925 if (!(flags & SUPPRESS) && tw != wp)
926 *ARG (double *) = negative ? -d : d;
930 float d = __strtof_internal (wp, &tw, flags & GROUP);
931 if (!(flags & SUPPRESS) && tw != wp)
932 *ARG (float *) = negative ? -d : d;
938 if (!(flags & SUPPRESS))
942 case '[': /* Character class. */
945 STRING_ARG (wstr, wchar_t);
946 c = '\0'; /* This is to keep gcc quiet. */
950 STRING_ARG (str, char);
965 /* Fill WP with byte flags indexed by character.
966 We will use this flag map for matching input characters. */
967 if (wpmax < UCHAR_MAX)
970 wp = (char *) alloca (wpmax);
972 memset (wp, 0, UCHAR_MAX);
975 if (fc == ']' || fc == '-')
977 /* If ] or - appears before any char in the set, it is not
978 the terminator or separator, but the first char in the
984 while ((fc = *f++) != '\0' && fc != ']')
986 if (fc == '-' && *f != '\0' && *f != ']' &&
987 (unsigned char) f[-2] <= (unsigned char) *f)
989 /* Add all characters from the one before the '-'
990 up to (but not including) the next format char. */
991 for (fc = f[-2]; fc < *f; ++fc)
995 /* Add the character to the flag map. */
1000 if (!(flags & LONG))
1013 NEXT_WIDE_CHAR (first);
1014 if (val > 255 || wp[val] == not_in)
1016 /* XXX We have a problem here. We read a wide
1017 character and this possibly took several
1018 bytes. But we can only push back one single
1019 character. To be sure we don't create wrong
1020 input we push it back only in case it is
1021 representable within one byte. */
1026 STRING_ADD_CHAR (wstr, val, wchar_t);
1036 if (!(flags & SUPPRESS))
1044 num.ul = read_in - 1; /* -1 because we already read one char. */
1047 if (wp[c] == not_in)
1052 STRING_ADD_CHAR (str, c, char);
1056 while (width != 0 && inchar () != EOF);
1058 if (read_in == num.ul)
1061 if (!(flags & SUPPRESS))
1069 case 'p': /* Generic pointer. */
1071 /* A PTR must be the same size as a `long int'. */
1072 flags &= ~(SHORT|LONGDBL);
1079 /* The last thing we saw int the format string was a white space.
1080 Consume the last white spaces. */
1085 while (isspace (c));
1089 /* Unlock stream. */
1097 __vfscanf (FILE *s, const char *format, va_list argptr)
1099 return _IO_vfscanf (s, format, argptr, NULL);
1103 weak_alias (__vfscanf, vfscanf)