]>
Commit | Line | Data |
---|---|---|
688903eb | 1 | /* Copyright (C) 1991-2018 Free Software Foundation, Inc. |
2c6fe0bd | 2 | This file is part of the GNU C Library. |
28f540f4 | 3 | |
2c6fe0bd | 4 | The GNU C Library is free software; you can redistribute it and/or |
41bdb6e2 AJ |
5 | modify it under the terms of the GNU Lesser General Public |
6 | License as published by the Free Software Foundation; either | |
7 | version 2.1 of the License, or (at your option) any later version. | |
28f540f4 | 8 | |
2c6fe0bd UD |
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 | |
41bdb6e2 | 12 | Lesser General Public License for more details. |
28f540f4 | 13 | |
41bdb6e2 | 14 | You should have received a copy of the GNU Lesser General Public |
59ba27a6 PE |
15 | License along with the GNU C Library; if not, see |
16 | <http://www.gnu.org/licenses/>. */ | |
28f540f4 | 17 | |
d64b6ad0 | 18 | #include <assert.h> |
28f540f4 RM |
19 | #include <errno.h> |
20 | #include <limits.h> | |
21 | #include <ctype.h> | |
22 | #include <stdarg.h> | |
bea9b193 | 23 | #include <stdbool.h> |
28f540f4 | 24 | #include <stdio.h> |
e852e889 | 25 | #include <stdint.h> |
28f540f4 RM |
26 | #include <stdlib.h> |
27 | #include <string.h> | |
8d8c6efa | 28 | #include <wchar.h> |
2c6fe0bd | 29 | #include <wctype.h> |
9090848d | 30 | #include <libc-diag.h> |
ec999b8e | 31 | #include <libc-lock.h> |
14ea22e9 | 32 | #include <locale/localeinfo.h> |
95e83974 | 33 | #include <scratch_buffer.h> |
28f540f4 | 34 | |
28f540f4 | 35 | #ifdef __GNUC__ |
14ea22e9 UD |
36 | # define HAVE_LONGLONG |
37 | # define LONGLONG long long | |
28f540f4 | 38 | #else |
14ea22e9 | 39 | # define LONGLONG long |
28f540f4 RM |
40 | #endif |
41 | ||
96d0213e UD |
42 | /* Determine whether we have to handle `long long' at all. */ |
43 | #if LONG_MAX == LONG_LONG_MAX | |
44 | # define need_longlong 0 | |
45 | #else | |
46 | # define need_longlong 1 | |
47 | #endif | |
48 | ||
859a09cf UD |
49 | /* Determine whether we have to handle `long'. */ |
50 | #if INT_MAX == LONG_MAX | |
51 | # define need_long 0 | |
52 | #else | |
e4899e7b | 53 | # define need_long 1 |
859a09cf UD |
54 | #endif |
55 | ||
96aa2d94 | 56 | /* Those are flags in the conversion format. */ |
4c02bf1a UD |
57 | #define LONG 0x0001 /* l: long or double */ |
58 | #define LONGDBL 0x0002 /* L: long long or long double */ | |
59 | #define SHORT 0x0004 /* h: short */ | |
60 | #define SUPPRESS 0x0008 /* *: suppress assignment */ | |
61 | #define POINTER 0x0010 /* weird %p pointer (`fake hex') */ | |
62 | #define NOSKIP 0x0020 /* do not skip blanks */ | |
63 | #define NUMBER_SIGNED 0x0040 /* signed integer */ | |
64 | #define GROUP 0x0080 /* ': group numbers */ | |
267c54dc | 65 | #define GNU_MALLOC 0x0100 /* a: malloc strings */ |
4c02bf1a UD |
66 | #define CHAR 0x0200 /* hh: char */ |
67 | #define I18N 0x0400 /* I: use locale's digits */ | |
68 | #define HEXA_FLOAT 0x0800 /* hexadecimal float */ | |
69 | #define READ_POINTER 0x1000 /* this is a pointer value */ | |
267c54dc UD |
70 | #define POSIX_MALLOC 0x2000 /* m: malloc strings */ |
71 | #define MALLOC (GNU_MALLOC | POSIX_MALLOC) | |
96aa2d94 | 72 | |
e3b22ad3 UD |
73 | #include <locale/localeinfo.h> |
74 | #include <libioP.h> | |
05be689b | 75 | |
e3b22ad3 UD |
76 | #ifdef COMPILE_WSCANF |
77 | # define ungetc(c, s) ((void) (c == WEOF \ | |
d64b6ad0 | 78 | || (--read_in, \ |
d18ea0c5 | 79 | _IO_sputbackwc (s, c)))) |
e3b22ad3 | 80 | # define ungetc_not_eof(c, s) ((void) (--read_in, \ |
d18ea0c5 | 81 | _IO_sputbackwc (s, c))) |
e3b22ad3 | 82 | # define inchar() (c == WEOF ? ((errno = inchar_errno), WEOF) \ |
d64b6ad0 | 83 | : ((c = _IO_getwc_unlocked (s)), \ |
7d68a8ec | 84 | (void) (c != WEOF \ |
6dd67bd5 UD |
85 | ? ++read_in \ |
86 | : (size_t) (inchar_errno = errno)), c)) | |
d64b6ad0 | 87 | |
e3b22ad3 UD |
88 | # define ISSPACE(Ch) iswspace (Ch) |
89 | # define ISDIGIT(Ch) iswdigit (Ch) | |
90 | # define ISXDIGIT(Ch) iswxdigit (Ch) | |
91 | # define TOLOWER(Ch) towlower (Ch) | |
92 | # define ORIENT if (_IO_fwide (s, 1) != 1) return WEOF | |
93 | # define __strtoll_internal __wcstoll_internal | |
94 | # define __strtoull_internal __wcstoull_internal | |
95 | # define __strtol_internal __wcstol_internal | |
96 | # define __strtoul_internal __wcstoul_internal | |
97 | # define __strtold_internal __wcstold_internal | |
98 | # define __strtod_internal __wcstod_internal | |
99 | # define __strtof_internal __wcstof_internal | |
100 | ||
101 | # define L_(Str) L##Str | |
e4899e7b | 102 | # define CHAR_T wchar_t |
e3b22ad3 | 103 | # define UCHAR_T unsigned int |
e4899e7b | 104 | # define WINT_T wint_t |
e3b22ad3 UD |
105 | # undef EOF |
106 | # define EOF WEOF | |
107 | #else | |
108 | # define ungetc(c, s) ((void) ((int) c == EOF \ | |
c764b9a7 | 109 | || (--read_in, \ |
d18ea0c5 | 110 | _IO_sputbackc (s, (unsigned char) c)))) |
e3b22ad3 | 111 | # define ungetc_not_eof(c, s) ((void) (--read_in, \ |
d18ea0c5 | 112 | _IO_sputbackc (s, (unsigned char) c))) |
e3b22ad3 | 113 | # define inchar() (c == EOF ? ((errno = inchar_errno), EOF) \ |
b112c02f | 114 | : ((c = _IO_getc_unlocked (s)), \ |
7d68a8ec | 115 | (void) (c != EOF \ |
6dd67bd5 UD |
116 | ? ++read_in \ |
117 | : (size_t) (inchar_errno = errno)), c)) | |
e3b22ad3 UD |
118 | # define ISSPACE(Ch) __isspace_l (Ch, loc) |
119 | # define ISDIGIT(Ch) __isdigit_l (Ch, loc) | |
120 | # define ISXDIGIT(Ch) __isxdigit_l (Ch, loc) | |
68dc4dcb | 121 | # define TOLOWER(Ch) __tolower_l ((unsigned char) (Ch), loc) |
e3b22ad3 | 122 | # define ORIENT if (_IO_vtable_offset (s) == 0 \ |
6b3f2b3d | 123 | && _IO_fwide (s, -1) != -1) \ |
411adb10 | 124 | return EOF |
d64b6ad0 | 125 | |
e3b22ad3 | 126 | # define L_(Str) Str |
e4899e7b | 127 | # define CHAR_T char |
e3b22ad3 UD |
128 | # define UCHAR_T unsigned char |
129 | # define WINT_T int | |
130 | #endif | |
d64b6ad0 | 131 | |
726d48ec GG |
132 | #include "printf-parse.h" /* Use read_int. */ |
133 | ||
e3b22ad3 UD |
134 | #define encode_error() do { \ |
135 | errval = 4; \ | |
2c6fe0bd | 136 | __set_errno (EILSEQ); \ |
e3b22ad3 | 137 | goto errout; \ |
2c6fe0bd | 138 | } while (0) |
e3b22ad3 UD |
139 | #define conv_error() do { \ |
140 | errval = 2; \ | |
141 | goto errout; \ | |
7c713e28 | 142 | } while (0) |
e3b22ad3 UD |
143 | #define input_error() do { \ |
144 | errval = 1; \ | |
145 | if (done == 0) done = EOF; \ | |
146 | goto errout; \ | |
7c713e28 | 147 | } while (0) |
267c54dc UD |
148 | #define add_ptr_to_free(ptr) \ |
149 | do \ | |
150 | { \ | |
151 | if (ptrs_to_free == NULL \ | |
152 | || ptrs_to_free->count == (sizeof (ptrs_to_free->ptrs) \ | |
153 | / sizeof (ptrs_to_free->ptrs[0]))) \ | |
154 | { \ | |
155 | struct ptrs_to_free *new_ptrs = alloca (sizeof (*ptrs_to_free)); \ | |
156 | new_ptrs->count = 0; \ | |
157 | new_ptrs->next = ptrs_to_free; \ | |
158 | ptrs_to_free = new_ptrs; \ | |
159 | } \ | |
160 | ptrs_to_free->ptrs[ptrs_to_free->count++] = (ptr); \ | |
161 | } \ | |
162 | while (0) | |
e3b22ad3 | 163 | #define ARGCHECK(s, format) \ |
7c713e28 RM |
164 | do \ |
165 | { \ | |
166 | /* Check file argument for consistence. */ \ | |
167 | CHECK_FILE (s, EOF); \ | |
f8b87ef0 UD |
168 | if (s->_flags & _IO_NO_READS) \ |
169 | { \ | |
170 | __set_errno (EBADF); \ | |
171 | return EOF; \ | |
172 | } \ | |
173 | else if (format == NULL) \ | |
174 | { \ | |
30bfee26 | 175 | __set_errno (EINVAL); \ |
f8b87ef0 UD |
176 | return EOF; \ |
177 | } \ | |
96aa2d94 | 178 | } while (0) |
e3b22ad3 | 179 | #define LOCK_STREAM(S) \ |
0dce3d15 | 180 | __libc_cleanup_region_start (1, (void (*) (void *)) &_IO_funlockfile, (S)); \ |
edf5b2d7 | 181 | _IO_flockfile (S) |
e3b22ad3 | 182 | #define UNLOCK_STREAM(S) \ |
c0fb8a56 UD |
183 | _IO_funlockfile (S); \ |
184 | __libc_cleanup_region_end (0) | |
28f540f4 | 185 | |
267c54dc UD |
186 | struct ptrs_to_free |
187 | { | |
188 | size_t count; | |
189 | struct ptrs_to_free *next; | |
190 | char **ptrs[32]; | |
191 | }; | |
28f540f4 | 192 | |
95e83974 FW |
193 | struct char_buffer { |
194 | CHAR_T *current; | |
195 | CHAR_T *end; | |
196 | struct scratch_buffer scratch; | |
197 | }; | |
198 | ||
199 | /* Returns a pointer to the first CHAR_T object in the buffer. Only | |
200 | valid if char_buffer_add (BUFFER, CH) has been called and | |
201 | char_buffer_error (BUFFER) is false. */ | |
202 | static inline CHAR_T * | |
203 | char_buffer_start (const struct char_buffer *buffer) | |
204 | { | |
205 | return (CHAR_T *) buffer->scratch.data; | |
206 | } | |
207 | ||
208 | /* Returns the number of CHAR_T objects in the buffer. Only valid if | |
209 | char_buffer_error (BUFFER) is false. */ | |
210 | static inline size_t | |
211 | char_buffer_size (const struct char_buffer *buffer) | |
212 | { | |
213 | return buffer->current - char_buffer_start (buffer); | |
214 | } | |
215 | ||
216 | /* Reinitializes BUFFER->current and BUFFER->end to cover the entire | |
217 | scratch buffer. */ | |
218 | static inline void | |
219 | char_buffer_rewind (struct char_buffer *buffer) | |
220 | { | |
221 | buffer->current = char_buffer_start (buffer); | |
222 | buffer->end = buffer->current + buffer->scratch.length / sizeof (CHAR_T); | |
223 | } | |
224 | ||
225 | /* Returns true if a previous call to char_buffer_add (BUFFER, CH) | |
226 | failed. */ | |
227 | static inline bool | |
228 | char_buffer_error (const struct char_buffer *buffer) | |
229 | { | |
230 | return __glibc_unlikely (buffer->current == NULL); | |
231 | } | |
232 | ||
233 | /* Slow path for char_buffer_add. */ | |
234 | static void | |
235 | char_buffer_add_slow (struct char_buffer *buffer, CHAR_T ch) | |
236 | { | |
237 | if (char_buffer_error (buffer)) | |
238 | return; | |
239 | size_t offset = buffer->end - (CHAR_T *) buffer->scratch.data; | |
240 | if (!scratch_buffer_grow_preserve (&buffer->scratch)) | |
241 | { | |
242 | buffer->current = NULL; | |
243 | buffer->end = NULL; | |
244 | return; | |
245 | } | |
246 | char_buffer_rewind (buffer); | |
247 | buffer->current += offset; | |
248 | *buffer->current++ = ch; | |
249 | } | |
250 | ||
251 | /* Adds CH to BUFFER. This function does not report any errors, check | |
252 | for them with char_buffer_error. */ | |
253 | static inline void | |
254 | char_buffer_add (struct char_buffer *buffer, CHAR_T ch) | |
255 | __attribute__ ((always_inline)); | |
256 | static inline void | |
257 | char_buffer_add (struct char_buffer *buffer, CHAR_T ch) | |
258 | { | |
259 | if (__glibc_unlikely (buffer->current == buffer->end)) | |
260 | char_buffer_add_slow (buffer, ch); | |
261 | else | |
262 | *buffer->current++ = ch; | |
263 | } | |
264 | ||
28f540f4 RM |
265 | /* Read formatted input from S according to the format string |
266 | FORMAT, using the argument list in ARG. | |
267 | Return the number of assignments made, or -1 for an input error. */ | |
e3b22ad3 | 268 | #ifdef COMPILE_WSCANF |
d64b6ad0 | 269 | int |
9964a145 | 270 | _IO_vfwscanf (FILE *s, const wchar_t *format, va_list argptr, |
c6251f03 | 271 | int *errp) |
e3b22ad3 | 272 | #else |
28f540f4 | 273 | int |
9964a145 | 274 | _IO_vfscanf_internal (FILE *s, const char *format, va_list argptr, |
c6251f03 | 275 | int *errp) |
96aa2d94 | 276 | #endif |
28f540f4 | 277 | { |
4cca6b86 | 278 | va_list arg; |
2e09a79a JM |
279 | const CHAR_T *f = format; |
280 | UCHAR_T fc; /* Current character of the format. */ | |
281 | WINT_T done = 0; /* Assignments done. */ | |
282 | size_t read_in = 0; /* Chars read in. */ | |
283 | WINT_T c = 0; /* Last char read. */ | |
284 | int width; /* Maximum field width. */ | |
285 | int flags; /* Modifiers for current format element. */ | |
e3b22ad3 UD |
286 | int errval = 0; |
287 | #ifndef COMPILE_WSCANF | |
af85385f | 288 | locale_t loc = _NL_CURRENT_LOCALE; |
f095bb72 | 289 | struct __locale_data *const curctype = loc->__locales[LC_CTYPE]; |
e3b22ad3 | 290 | #endif |
0793d348 | 291 | |
7d68a8ec UD |
292 | /* Errno of last failed inchar call. */ |
293 | int inchar_errno = 0; | |
28f540f4 | 294 | /* Status for reading F-P nums. */ |
6ecec3b6 | 295 | char got_digit, got_dot, got_e, negative; |
28f540f4 | 296 | /* If a [...] is a [^...]. */ |
d64b6ad0 | 297 | CHAR_T not_in; |
377a515b | 298 | #define exp_char not_in |
28f540f4 RM |
299 | /* Base for integral numbers. */ |
300 | int base; | |
77a58cad | 301 | /* Decimal point character. */ |
72acaddf | 302 | #ifdef COMPILE_WSCANF |
52a7f7c0 | 303 | wint_t decimal; |
72acaddf UD |
304 | #else |
305 | const char *decimal; | |
306 | #endif | |
0793d348 | 307 | /* The thousands character of the current locale. */ |
72acaddf | 308 | #ifdef COMPILE_WSCANF |
52a7f7c0 | 309 | wint_t thousands; |
72acaddf UD |
310 | #else |
311 | const char *thousands; | |
312 | #endif | |
267c54dc | 313 | struct ptrs_to_free *ptrs_to_free = NULL; |
8d8c6efa UD |
314 | /* State for the conversions. */ |
315 | mbstate_t state; | |
28f540f4 | 316 | /* Integral holding variables. */ |
f0bf9cb9 RM |
317 | union |
318 | { | |
319 | long long int q; | |
320 | unsigned long long int uq; | |
321 | long int l; | |
322 | unsigned long int ul; | |
323 | } num; | |
28f540f4 | 324 | /* Character-buffer pointer. */ |
ba1ffaa1 | 325 | char *str = NULL; |
2c6fe0bd | 326 | wchar_t *wstr = NULL; |
ba1ffaa1 | 327 | char **strptr = NULL; |
54188057 | 328 | ssize_t strsize = 0; |
05be689b RM |
329 | /* We must not react on white spaces immediately because they can |
330 | possibly be matched even if in the input stream no character is | |
331 | available anymore. */ | |
332 | int skip_space = 0; | |
28f540f4 | 333 | /* Workspace. */ |
d64b6ad0 | 334 | CHAR_T *tw; /* Temporary pointer. */ |
95e83974 FW |
335 | struct char_buffer charbuf; |
336 | scratch_buffer_init (&charbuf.scratch); | |
28f540f4 | 337 | |
4cca6b86 UD |
338 | #ifdef __va_copy |
339 | __va_copy (arg, argptr); | |
340 | #else | |
341 | arg = (va_list) argptr; | |
342 | #endif | |
343 | ||
d64b6ad0 UD |
344 | #ifdef ORIENT |
345 | ORIENT; | |
346 | #endif | |
347 | ||
96aa2d94 | 348 | ARGCHECK (s, format); |
28f540f4 | 349 | |
e3b22ad3 UD |
350 | { |
351 | #ifndef COMPILE_WSCANF | |
f095bb72 | 352 | struct __locale_data *const curnumeric = loc->__locales[LC_NUMERIC]; |
e3b22ad3 UD |
353 | #endif |
354 | ||
355 | /* Figure out the decimal point character. */ | |
72acaddf | 356 | #ifdef COMPILE_WSCANF |
e3b22ad3 | 357 | decimal = _NL_CURRENT_WORD (LC_NUMERIC, _NL_NUMERIC_DECIMAL_POINT_WC); |
72acaddf | 358 | #else |
e3b22ad3 | 359 | decimal = curnumeric->values[_NL_ITEM_INDEX (DECIMAL_POINT)].string; |
72acaddf | 360 | #endif |
e3b22ad3 | 361 | /* Figure out the thousands separator character. */ |
72acaddf | 362 | #ifdef COMPILE_WSCANF |
e3b22ad3 | 363 | thousands = _NL_CURRENT_WORD (LC_NUMERIC, _NL_NUMERIC_THOUSANDS_SEP_WC); |
72acaddf | 364 | #else |
e3b22ad3 UD |
365 | thousands = curnumeric->values[_NL_ITEM_INDEX (THOUSANDS_SEP)].string; |
366 | if (*thousands == '\0') | |
367 | thousands = NULL; | |
72acaddf | 368 | #endif |
e3b22ad3 | 369 | } |
28f540f4 | 370 | |
7c713e28 | 371 | /* Lock the stream. */ |
edf5b2d7 | 372 | LOCK_STREAM (s); |
aa1075ea | 373 | |
8d8c6efa | 374 | |
72acaddf | 375 | #ifndef COMPILE_WSCANF |
8d8c6efa UD |
376 | /* From now on we use `state' to convert the format string. */ |
377 | memset (&state, '\0', sizeof (state)); | |
d64b6ad0 | 378 | #endif |
8d8c6efa | 379 | |
28f540f4 RM |
380 | /* Run through the format string. */ |
381 | while (*f != '\0') | |
382 | { | |
be69ea41 RM |
383 | unsigned int argpos; |
384 | /* Extract the next argument, which is of type TYPE. | |
385 | For a %N$... spec, this is the Nth argument from the beginning; | |
386 | otherwise it is the next argument after the state now in ARG. */ | |
4cca6b86 | 387 | #ifdef __va_copy |
96aa2d94 | 388 | # define ARG(type) (argpos == 0 ? va_arg (arg, type) : \ |
4cca6b86 UD |
389 | ({ unsigned int pos = argpos; \ |
390 | va_list arg; \ | |
391 | __va_copy (arg, argptr); \ | |
392 | while (--pos > 0) \ | |
393 | (void) va_arg (arg, void *); \ | |
394 | va_arg (arg, type); \ | |
395 | })) | |
396 | #else | |
397 | # if 0 | |
398 | /* XXX Possible optimization. */ | |
399 | # define ARG(type) (argpos == 0 ? va_arg (arg, type) : \ | |
96aa2d94 RM |
400 | ({ va_list arg = (va_list) argptr; \ |
401 | arg = (va_list) ((char *) arg \ | |
402 | + (argpos - 1) \ | |
403 | * __va_rounded_size (void *)); \ | |
404 | va_arg (arg, type); \ | |
405 | })) | |
4cca6b86 UD |
406 | # else |
407 | # define ARG(type) (argpos == 0 ? va_arg (arg, type) : \ | |
be69ea41 RM |
408 | ({ unsigned int pos = argpos; \ |
409 | va_list arg = (va_list) argptr; \ | |
410 | while (--pos > 0) \ | |
411 | (void) va_arg (arg, void *); \ | |
412 | va_arg (arg, type); \ | |
413 | })) | |
4cca6b86 | 414 | # endif |
96aa2d94 | 415 | #endif |
be69ea41 | 416 | |
72acaddf | 417 | #ifndef COMPILE_WSCANF |
44f8759b | 418 | if (!isascii ((unsigned char) *f)) |
28f540f4 RM |
419 | { |
420 | /* Non-ASCII, may be a multibyte. */ | |
8d8c6efa | 421 | int len = __mbrlen (f, strlen (f), &state); |
28f540f4 RM |
422 | if (len > 0) |
423 | { | |
3867ee64 RM |
424 | do |
425 | { | |
426 | c = inchar (); | |
a1ffb40e | 427 | if (__glibc_unlikely (c == EOF)) |
3867ee64 | 428 | input_error (); |
44f8759b | 429 | else if (c != (unsigned char) *f++) |
3867ee64 | 430 | { |
44f8759b | 431 | ungetc_not_eof (c, s); |
3867ee64 RM |
432 | conv_error (); |
433 | } | |
434 | } | |
435 | while (--len > 0); | |
28f540f4 RM |
436 | continue; |
437 | } | |
438 | } | |
d64b6ad0 | 439 | #endif |
28f540f4 RM |
440 | |
441 | fc = *f++; | |
442 | if (fc != '%') | |
443 | { | |
05be689b | 444 | /* Remember to skip spaces. */ |
d64b6ad0 | 445 | if (ISSPACE (fc)) |
05be689b RM |
446 | { |
447 | skip_space = 1; | |
448 | continue; | |
449 | } | |
450 | ||
3867ee64 RM |
451 | /* Read a character. */ |
452 | c = inchar (); | |
453 | ||
28f540f4 | 454 | /* Characters other than format specs must just match. */ |
a1ffb40e | 455 | if (__glibc_unlikely (c == EOF)) |
96aa2d94 | 456 | input_error (); |
05be689b | 457 | |
2114e10a RM |
458 | /* We saw white space char as the last character in the format |
459 | string. Now it's time to skip all leading white space. */ | |
05be689b | 460 | if (skip_space) |
28f540f4 | 461 | { |
d64b6ad0 | 462 | while (ISSPACE (c)) |
a1ffb40e | 463 | if (__glibc_unlikely (inchar () == EOF)) |
d3b52028 | 464 | input_error (); |
05be689b | 465 | skip_space = 0; |
28f540f4 | 466 | } |
05be689b | 467 | |
a1ffb40e | 468 | if (__glibc_unlikely (c != fc)) |
3867ee64 | 469 | { |
655c0697 | 470 | ungetc (c, s); |
3867ee64 RM |
471 | conv_error (); |
472 | } | |
05be689b | 473 | |
28f540f4 RM |
474 | continue; |
475 | } | |
476 | ||
05be689b | 477 | /* This is the start of the conversion string. */ |
96aa2d94 | 478 | flags = 0; |
96aa2d94 | 479 | |
be69ea41 RM |
480 | /* Initialize state of modifiers. */ |
481 | argpos = 0; | |
be69ea41 | 482 | |
05be689b | 483 | /* Prepare temporary buffer. */ |
95e83974 | 484 | char_buffer_rewind (&charbuf); |
05be689b | 485 | |
be69ea41 | 486 | /* Check for a positional parameter specification. */ |
44f8759b | 487 | if (ISDIGIT ((UCHAR_T) *f)) |
be69ea41 | 488 | { |
726d48ec | 489 | argpos = read_int ((const UCHAR_T **) &f); |
d64b6ad0 | 490 | if (*f == L_('$')) |
be69ea41 RM |
491 | ++f; |
492 | else | |
493 | { | |
494 | /* Oops; that was actually the field width. */ | |
495 | width = argpos; | |
496 | argpos = 0; | |
497 | goto got_width; | |
498 | } | |
499 | } | |
500 | ||
a1d84548 UD |
501 | /* Check for the assignment-suppressing, the number grouping flag, |
502 | and the signal to use the locale's digit representation. */ | |
503 | while (*f == L_('*') || *f == L_('\'') || *f == L_('I')) | |
f0bf9cb9 RM |
504 | switch (*f++) |
505 | { | |
d64b6ad0 | 506 | case L_('*'): |
0793d348 | 507 | flags |= SUPPRESS; |
f0bf9cb9 | 508 | break; |
d64b6ad0 | 509 | case L_('\''): |
3979024a UD |
510 | #ifdef COMPILE_WSCANF |
511 | if (thousands != L'\0') | |
512 | #else | |
513 | if (thousands != NULL) | |
514 | #endif | |
515 | flags |= GROUP; | |
f0bf9cb9 | 516 | break; |
a1d84548 UD |
517 | case L_('I'): |
518 | flags |= I18N; | |
519 | break; | |
f0bf9cb9 RM |
520 | } |
521 | ||
28f540f4 RM |
522 | /* Find the maximum field width. */ |
523 | width = 0; | |
726d48ec GG |
524 | if (ISDIGIT ((UCHAR_T) *f)) |
525 | width = read_int ((const UCHAR_T **) &f); | |
be69ea41 | 526 | got_width: |
28f540f4 RM |
527 | if (width == 0) |
528 | width = -1; | |
529 | ||
530 | /* Check for type modifiers. */ | |
c3966b88 UD |
531 | switch (*f++) |
532 | { | |
d64b6ad0 | 533 | case L_('h'): |
c3966b88 | 534 | /* ints are short ints or chars. */ |
d64b6ad0 | 535 | if (*f == L_('h')) |
c3966b88 UD |
536 | { |
537 | ++f; | |
538 | flags |= CHAR; | |
539 | } | |
540 | else | |
541 | flags |= SHORT; | |
542 | break; | |
d64b6ad0 UD |
543 | case L_('l'): |
544 | if (*f == L_('l')) | |
c3966b88 UD |
545 | { |
546 | /* A double `l' is equivalent to an `L'. */ | |
547 | ++f; | |
859a09cf | 548 | flags |= LONGDBL | LONG; |
c3966b88 UD |
549 | } |
550 | else | |
551 | /* ints are long ints. */ | |
552 | flags |= LONG; | |
553 | break; | |
d64b6ad0 UD |
554 | case L_('q'): |
555 | case L_('L'): | |
c3966b88 | 556 | /* doubles are long doubles, and ints are long long ints. */ |
859a09cf | 557 | flags |= LONGDBL | LONG; |
c3966b88 | 558 | break; |
d64b6ad0 | 559 | case L_('a'): |
c3966b88 UD |
560 | /* The `a' is used as a flag only if followed by `s', `S' or |
561 | `['. */ | |
d64b6ad0 | 562 | if (*f != L_('s') && *f != L_('S') && *f != L_('[')) |
c3966b88 UD |
563 | { |
564 | --f; | |
565 | break; | |
566 | } | |
874aa523 UD |
567 | /* In __isoc99_*scanf %as, %aS and %a[ extension is not |
568 | supported at all. */ | |
569 | if (s->_flags2 & _IO_FLAGS2_SCANF_STD) | |
570 | { | |
571 | --f; | |
572 | break; | |
573 | } | |
c3966b88 UD |
574 | /* String conversions (%s, %[) take a `char **' |
575 | arg and fill it in with a malloc'd pointer. */ | |
267c54dc | 576 | flags |= GNU_MALLOC; |
c3966b88 | 577 | break; |
874aa523 UD |
578 | case L_('m'): |
579 | flags |= POSIX_MALLOC; | |
580 | if (*f == L_('l')) | |
581 | { | |
582 | ++f; | |
583 | flags |= LONG; | |
584 | } | |
585 | break; | |
d64b6ad0 | 586 | case L_('z'): |
96d0213e | 587 | if (need_longlong && sizeof (size_t) > sizeof (unsigned long int)) |
c3966b88 UD |
588 | flags |= LONGDBL; |
589 | else if (sizeof (size_t) > sizeof (unsigned int)) | |
590 | flags |= LONG; | |
591 | break; | |
d64b6ad0 | 592 | case L_('j'): |
96d0213e | 593 | if (need_longlong && sizeof (uintmax_t) > sizeof (unsigned long int)) |
c3966b88 UD |
594 | flags |= LONGDBL; |
595 | else if (sizeof (uintmax_t) > sizeof (unsigned int)) | |
596 | flags |= LONG; | |
597 | break; | |
d64b6ad0 | 598 | case L_('t'): |
96d0213e | 599 | if (need_longlong && sizeof (ptrdiff_t) > sizeof (long int)) |
c3966b88 UD |
600 | flags |= LONGDBL; |
601 | else if (sizeof (ptrdiff_t) > sizeof (int)) | |
602 | flags |= LONG; | |
603 | break; | |
604 | default: | |
605 | /* Not a recognized modifier. Backup. */ | |
606 | --f; | |
607 | break; | |
608 | } | |
28f540f4 RM |
609 | |
610 | /* End of the format string? */ | |
a1ffb40e | 611 | if (__glibc_unlikely (*f == L_('\0'))) |
96aa2d94 | 612 | conv_error (); |
28f540f4 RM |
613 | |
614 | /* Find the conversion specifier. */ | |
28f540f4 | 615 | fc = *f++; |
d64b6ad0 | 616 | if (skip_space || (fc != L_('[') && fc != L_('c') |
f9a06dc1 | 617 | && fc != L_('C') && fc != L_('n'))) |
05be689b RM |
618 | { |
619 | /* Eat whitespace. */ | |
14ea22e9 | 620 | int save_errno = errno; |
0923a2c8 | 621 | __set_errno (0); |
3867ee64 | 622 | do |
0923a2c8 UD |
623 | /* We add the additional test for EOF here since otherwise |
624 | inchar will restore the old errno value which might be | |
625 | EINTR but does not indicate an interrupt since nothing | |
626 | was read at this time. */ | |
627 | if (__builtin_expect ((c == EOF || inchar () == EOF) | |
628 | && errno == EINTR, 0)) | |
f8b87ef0 | 629 | input_error (); |
d64b6ad0 | 630 | while (ISSPACE (c)); |
0923a2c8 | 631 | __set_errno (save_errno); |
655c0697 | 632 | ungetc (c, s); |
05be689b RM |
633 | skip_space = 0; |
634 | } | |
635 | ||
28f540f4 RM |
636 | switch (fc) |
637 | { | |
d64b6ad0 | 638 | case L_('%'): /* Must match a literal '%'. */ |
14ea22e9 | 639 | c = inchar (); |
a1ffb40e | 640 | if (__glibc_unlikely (c == EOF)) |
dc30f461 | 641 | input_error (); |
a1ffb40e | 642 | if (__glibc_unlikely (c != fc)) |
3867ee64 | 643 | { |
44f8759b | 644 | ungetc_not_eof (c, s); |
3867ee64 RM |
645 | conv_error (); |
646 | } | |
28f540f4 RM |
647 | break; |
648 | ||
d64b6ad0 | 649 | case L_('n'): /* Answer number of assignments done. */ |
01cdeca0 RM |
650 | /* Corrigendum 1 to ISO C 1990 describes the allowed flags |
651 | with the 'n' conversion specifier. */ | |
0793d348 | 652 | if (!(flags & SUPPRESS)) |
5290baf0 UD |
653 | { |
654 | /* Don't count the read-ahead. */ | |
96d0213e | 655 | if (need_longlong && (flags & LONGDBL)) |
5290baf0 | 656 | *ARG (long long int *) = read_in; |
859a09cf | 657 | else if (need_long && (flags & LONG)) |
5290baf0 UD |
658 | *ARG (long int *) = read_in; |
659 | else if (flags & SHORT) | |
660 | *ARG (short int *) = read_in; | |
859a09cf | 661 | else if (!(flags & CHAR)) |
5290baf0 | 662 | *ARG (int *) = read_in; |
859a09cf UD |
663 | else |
664 | *ARG (char *) = read_in; | |
8d57beea UD |
665 | |
666 | #ifdef NO_BUG_IN_ISO_C_CORRIGENDUM_1 | |
667 | /* We have a severe problem here. The ISO C standard | |
668 | contradicts itself in explaining the effect of the %n | |
669 | format in `scanf'. While in ISO C:1990 and the ISO C | |
670 | Amendement 1:1995 the result is described as | |
671 | ||
672 | Execution of a %n directive does not effect the | |
673 | assignment count returned at the completion of | |
674 | execution of the f(w)scanf function. | |
675 | ||
676 | in ISO C Corrigendum 1:1994 the following was added: | |
677 | ||
678 | Subclause 7.9.6.2 | |
679 | Add the following fourth example: | |
680 | In: | |
681 | #include <stdio.h> | |
682 | int d1, d2, n1, n2, i; | |
683 | i = sscanf("123", "%d%n%n%d", &d1, &n1, &n2, &d2); | |
684 | the value 123 is assigned to d1 and the value3 to n1. | |
685 | Because %n can never get an input failure the value | |
686 | of 3 is also assigned to n2. The value of d2 is not | |
687 | affected. The value 3 is assigned to i. | |
688 | ||
14ea22e9 | 689 | We go for now with the historically correct code from ISO C, |
8d57beea UD |
690 | i.e., we don't count the %n assignments. When it ever |
691 | should proof to be wrong just remove the #ifdef above. */ | |
5290baf0 | 692 | ++done; |
8d57beea | 693 | #endif |
5290baf0 | 694 | } |
28f540f4 RM |
695 | break; |
696 | ||
d64b6ad0 | 697 | case L_('c'): /* Match characters. */ |
2c6fe0bd | 698 | if ((flags & LONG) == 0) |
28f540f4 | 699 | { |
874aa523 UD |
700 | if (width == -1) |
701 | width = 1; | |
702 | ||
703 | #define STRING_ARG(Str, Type, Width) \ | |
704 | do if (!(flags & SUPPRESS)) \ | |
705 | { \ | |
706 | if (flags & MALLOC) \ | |
707 | { \ | |
708 | /* The string is to be stored in a malloc'd buffer. */ \ | |
709 | /* For %mS using char ** is actually wrong, but \ | |
710 | shouldn't make a difference on any arch glibc \ | |
711 | supports and would unnecessarily complicate \ | |
712 | things. */ \ | |
713 | strptr = ARG (char **); \ | |
714 | if (strptr == NULL) \ | |
715 | conv_error (); \ | |
716 | /* Allocate an initial buffer. */ \ | |
717 | strsize = Width; \ | |
718 | *strptr = (char *) malloc (strsize * sizeof (Type)); \ | |
719 | Str = (Type *) *strptr; \ | |
720 | if (Str != NULL) \ | |
721 | add_ptr_to_free (strptr); \ | |
722 | else if (flags & POSIX_MALLOC) \ | |
3f8cc204 UD |
723 | { \ |
724 | done = EOF; \ | |
725 | goto errout; \ | |
726 | } \ | |
874aa523 UD |
727 | } \ |
728 | else \ | |
729 | Str = ARG (Type *); \ | |
730 | if (Str == NULL) \ | |
731 | conv_error (); \ | |
732 | } while (0) | |
733 | #ifdef COMPILE_WSCANF | |
734 | STRING_ARG (str, char, 100); | |
735 | #else | |
736 | STRING_ARG (str, char, (width > 1024 ? 1024 : width)); | |
737 | #endif | |
28f540f4 | 738 | |
14ea22e9 | 739 | c = inchar (); |
a1ffb40e | 740 | if (__glibc_unlikely (c == EOF)) |
2c6fe0bd | 741 | input_error (); |
28f540f4 | 742 | |
72acaddf | 743 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
744 | /* We have to convert the wide character(s) into multibyte |
745 | characters and store the result. */ | |
746 | memset (&state, '\0', sizeof (state)); | |
747 | ||
748 | do | |
749 | { | |
750 | size_t n; | |
751 | ||
874aa523 | 752 | if (!(flags & SUPPRESS) && (flags & POSIX_MALLOC) |
9dd87afb | 753 | && *strptr + strsize - str <= MB_LEN_MAX) |
874aa523 UD |
754 | { |
755 | /* We have to enlarge the buffer if the `m' flag | |
756 | was given. */ | |
757 | size_t strleng = str - *strptr; | |
758 | char *newstr; | |
759 | ||
760 | newstr = (char *) realloc (*strptr, strsize * 2); | |
761 | if (newstr == NULL) | |
762 | { | |
763 | /* Can't allocate that much. Last-ditch effort. */ | |
764 | newstr = (char *) realloc (*strptr, | |
9dd87afb | 765 | strleng + MB_LEN_MAX); |
874aa523 | 766 | if (newstr == NULL) |
3f8cc204 UD |
767 | { |
768 | /* c can't have `a' flag, only `m'. */ | |
769 | done = EOF; | |
770 | goto errout; | |
771 | } | |
874aa523 UD |
772 | else |
773 | { | |
774 | *strptr = newstr; | |
775 | str = newstr + strleng; | |
9dd87afb | 776 | strsize = strleng + MB_LEN_MAX; |
874aa523 UD |
777 | } |
778 | } | |
779 | else | |
780 | { | |
781 | *strptr = newstr; | |
782 | str = newstr + strleng; | |
783 | strsize *= 2; | |
784 | } | |
785 | } | |
786 | ||
4aebaa6b | 787 | n = __wcrtomb (!(flags & SUPPRESS) ? str : NULL, c, &state); |
a1ffb40e | 788 | if (__glibc_unlikely (n == (size_t) -1)) |
d64b6ad0 UD |
789 | /* No valid wide character. */ |
790 | input_error (); | |
791 | ||
792 | /* Increment the output pointer. Even if we don't | |
793 | write anything. */ | |
794 | str += n; | |
795 | } | |
796 | while (--width > 0 && inchar () != EOF); | |
797 | #else | |
2c6fe0bd UD |
798 | if (!(flags & SUPPRESS)) |
799 | { | |
800 | do | |
874aa523 UD |
801 | { |
802 | if ((flags & MALLOC) | |
803 | && (char *) str == *strptr + strsize) | |
804 | { | |
805 | /* Enlarge the buffer. */ | |
806 | size_t newsize | |
807 | = strsize | |
808 | + (strsize >= width ? width - 1 : strsize); | |
809 | ||
810 | str = (char *) realloc (*strptr, newsize); | |
811 | if (str == NULL) | |
812 | { | |
813 | /* Can't allocate that much. Last-ditch | |
814 | effort. */ | |
815 | str = (char *) realloc (*strptr, strsize + 1); | |
816 | if (str == NULL) | |
3f8cc204 UD |
817 | { |
818 | /* c can't have `a' flag, only `m'. */ | |
819 | done = EOF; | |
820 | goto errout; | |
821 | } | |
874aa523 UD |
822 | else |
823 | { | |
824 | *strptr = (char *) str; | |
825 | str += strsize; | |
826 | ++strsize; | |
827 | } | |
828 | } | |
829 | else | |
830 | { | |
831 | *strptr = (char *) str; | |
832 | str += strsize; | |
833 | strsize = newsize; | |
834 | } | |
835 | } | |
836 | *str++ = c; | |
837 | } | |
2c6fe0bd UD |
838 | while (--width > 0 && inchar () != EOF); |
839 | } | |
840 | else | |
841 | while (--width > 0 && inchar () != EOF); | |
d64b6ad0 | 842 | #endif |
2c6fe0bd | 843 | |
2c6fe0bd | 844 | if (!(flags & SUPPRESS)) |
874aa523 UD |
845 | { |
846 | if ((flags & MALLOC) && str - *strptr != strsize) | |
847 | { | |
848 | char *cp = (char *) realloc (*strptr, str - *strptr); | |
849 | if (cp != NULL) | |
850 | *strptr = cp; | |
851 | } | |
852 | strptr = NULL; | |
853 | ++done; | |
854 | } | |
2c6fe0bd UD |
855 | |
856 | break; | |
28f540f4 | 857 | } |
2c6fe0bd | 858 | /* FALLTHROUGH */ |
d64b6ad0 | 859 | case L_('C'): |
874aa523 UD |
860 | if (width == -1) |
861 | width = 1; | |
862 | ||
863 | STRING_ARG (wstr, wchar_t, (width > 1024 ? 1024 : width)); | |
d64b6ad0 UD |
864 | |
865 | c = inchar (); | |
a1ffb40e | 866 | if (__glibc_unlikely (c == EOF)) |
d64b6ad0 UD |
867 | input_error (); |
868 | ||
72acaddf | 869 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
870 | /* Just store the incoming wide characters. */ |
871 | if (!(flags & SUPPRESS)) | |
872 | { | |
873 | do | |
874aa523 UD |
874 | { |
875 | if ((flags & MALLOC) | |
876 | && wstr == (wchar_t *) *strptr + strsize) | |
877 | { | |
878 | size_t newsize | |
879 | = strsize + (strsize > width ? width - 1 : strsize); | |
880 | /* Enlarge the buffer. */ | |
881 | wstr = (wchar_t *) realloc (*strptr, | |
882 | newsize * sizeof (wchar_t)); | |
883 | if (wstr == NULL) | |
884 | { | |
885 | /* Can't allocate that much. Last-ditch effort. */ | |
886 | wstr = (wchar_t *) realloc (*strptr, | |
887 | (strsize + 1) | |
888 | * sizeof (wchar_t)); | |
889 | if (wstr == NULL) | |
3f8cc204 UD |
890 | { |
891 | /* C or lc can't have `a' flag, only `m' | |
892 | flag. */ | |
893 | done = EOF; | |
894 | goto errout; | |
895 | } | |
874aa523 UD |
896 | else |
897 | { | |
898 | *strptr = (char *) wstr; | |
899 | wstr += strsize; | |
900 | ++strsize; | |
901 | } | |
902 | } | |
903 | else | |
904 | { | |
905 | *strptr = (char *) wstr; | |
906 | wstr += strsize; | |
907 | strsize = newsize; | |
908 | } | |
909 | } | |
910 | *wstr++ = c; | |
911 | } | |
d64b6ad0 UD |
912 | while (--width > 0 && inchar () != EOF); |
913 | } | |
914 | else | |
915 | while (--width > 0 && inchar () != EOF); | |
916 | #else | |
2c6fe0bd | 917 | { |
d64b6ad0 UD |
918 | /* We have to convert the multibyte input sequence to wide |
919 | characters. */ | |
68bd3326 | 920 | char buf[1]; |
d64b6ad0 | 921 | mbstate_t cstate; |
28f540f4 | 922 | |
d64b6ad0 | 923 | memset (&cstate, '\0', sizeof (cstate)); |
edf5b2d7 | 924 | |
2c6fe0bd UD |
925 | do |
926 | { | |
d64b6ad0 UD |
927 | /* This is what we present the mbrtowc function first. */ |
928 | buf[0] = c; | |
d64b6ad0 | 929 | |
874aa523 UD |
930 | if (!(flags & SUPPRESS) && (flags & MALLOC) |
931 | && wstr == (wchar_t *) *strptr + strsize) | |
932 | { | |
933 | size_t newsize | |
934 | = strsize + (strsize > width ? width - 1 : strsize); | |
935 | /* Enlarge the buffer. */ | |
936 | wstr = (wchar_t *) realloc (*strptr, | |
937 | newsize * sizeof (wchar_t)); | |
938 | if (wstr == NULL) | |
939 | { | |
940 | /* Can't allocate that much. Last-ditch effort. */ | |
941 | wstr = (wchar_t *) realloc (*strptr, | |
942 | ((strsize + 1) | |
943 | * sizeof (wchar_t))); | |
944 | if (wstr == NULL) | |
3f8cc204 UD |
945 | { |
946 | /* C or lc can't have `a' flag, only `m' flag. */ | |
947 | done = EOF; | |
948 | goto errout; | |
949 | } | |
874aa523 UD |
950 | else |
951 | { | |
952 | *strptr = (char *) wstr; | |
953 | wstr += strsize; | |
954 | ++strsize; | |
955 | } | |
956 | } | |
957 | else | |
958 | { | |
959 | *strptr = (char *) wstr; | |
960 | wstr += strsize; | |
961 | strsize = newsize; | |
962 | } | |
963 | } | |
964 | ||
d64b6ad0 UD |
965 | while (1) |
966 | { | |
967 | size_t n; | |
968 | ||
969 | n = __mbrtowc (!(flags & SUPPRESS) ? wstr : NULL, | |
2f278c94 | 970 | buf, 1, &cstate); |
d64b6ad0 UD |
971 | |
972 | if (n == (size_t) -2) | |
973 | { | |
974 | /* Possibly correct character, just not enough | |
975 | input. */ | |
a1ffb40e | 976 | if (__glibc_unlikely (inchar () == EOF)) |
d64b6ad0 UD |
977 | encode_error (); |
978 | ||
2f278c94 | 979 | buf[0] = c; |
d64b6ad0 UD |
980 | continue; |
981 | } | |
982 | ||
a1ffb40e | 983 | if (__glibc_unlikely (n != 1)) |
d64b6ad0 UD |
984 | encode_error (); |
985 | ||
986 | /* We have a match. */ | |
987 | break; | |
988 | } | |
989 | ||
990 | /* Advance the result pointer. */ | |
991 | ++wstr; | |
992 | } | |
993 | while (--width > 0 && inchar () != EOF); | |
2c6fe0bd | 994 | } |
d64b6ad0 | 995 | #endif |
28f540f4 | 996 | |
d64b6ad0 | 997 | if (!(flags & SUPPRESS)) |
874aa523 UD |
998 | { |
999 | if ((flags & MALLOC) && wstr - (wchar_t *) *strptr != strsize) | |
1000 | { | |
1001 | wchar_t *cp = (wchar_t *) realloc (*strptr, | |
1002 | ((wstr | |
1003 | - (wchar_t *) *strptr) | |
1004 | * sizeof (wchar_t))); | |
1005 | if (cp != NULL) | |
1006 | *strptr = (char *) cp; | |
1007 | } | |
1008 | strptr = NULL; | |
1009 | ||
1010 | ++done; | |
1011 | } | |
2c6fe0bd | 1012 | |
d64b6ad0 UD |
1013 | break; |
1014 | ||
1015 | case L_('s'): /* Read a string. */ | |
1016 | if (!(flags & LONG)) | |
1017 | { | |
874aa523 | 1018 | STRING_ARG (str, char, 100); |
28f540f4 | 1019 | |
d64b6ad0 | 1020 | c = inchar (); |
a1ffb40e | 1021 | if (__glibc_unlikely (c == EOF)) |
d64b6ad0 | 1022 | input_error (); |
28f540f4 | 1023 | |
72acaddf | 1024 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
1025 | memset (&state, '\0', sizeof (state)); |
1026 | #endif | |
1027 | ||
1028 | do | |
3867ee64 | 1029 | { |
d64b6ad0 UD |
1030 | if (ISSPACE (c)) |
1031 | { | |
44f8759b | 1032 | ungetc_not_eof (c, s); |
d64b6ad0 UD |
1033 | break; |
1034 | } | |
1035 | ||
72acaddf | 1036 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
1037 | /* This is quite complicated. We have to convert the |
1038 | wide characters into multibyte characters and then | |
1039 | store them. */ | |
1040 | { | |
1041 | size_t n; | |
1042 | ||
1043 | if (!(flags & SUPPRESS) && (flags & MALLOC) | |
9dd87afb | 1044 | && *strptr + strsize - str <= MB_LEN_MAX) |
d64b6ad0 | 1045 | { |
874aa523 UD |
1046 | /* We have to enlarge the buffer if the `a' or `m' |
1047 | flag was given. */ | |
d90e1b42 UD |
1048 | size_t strleng = str - *strptr; |
1049 | char *newstr; | |
1050 | ||
1051 | newstr = (char *) realloc (*strptr, strsize * 2); | |
1052 | if (newstr == NULL) | |
d64b6ad0 UD |
1053 | { |
1054 | /* Can't allocate that much. Last-ditch | |
1055 | effort. */ | |
d90e1b42 | 1056 | newstr = (char *) realloc (*strptr, |
9dd87afb | 1057 | strleng + MB_LEN_MAX); |
d90e1b42 | 1058 | if (newstr == NULL) |
d64b6ad0 | 1059 | { |
267c54dc | 1060 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
1061 | { |
1062 | done = EOF; | |
1063 | goto errout; | |
1064 | } | |
d64b6ad0 UD |
1065 | /* We lose. Oh well. Terminate the |
1066 | string and stop converting, | |
1067 | so at least we don't skip any input. */ | |
d90e1b42 | 1068 | ((char *) (*strptr))[strleng] = '\0'; |
267c54dc | 1069 | strptr = NULL; |
d64b6ad0 UD |
1070 | ++done; |
1071 | conv_error (); | |
1072 | } | |
1073 | else | |
1074 | { | |
d90e1b42 UD |
1075 | *strptr = newstr; |
1076 | str = newstr + strleng; | |
9dd87afb | 1077 | strsize = strleng + MB_LEN_MAX; |
d64b6ad0 UD |
1078 | } |
1079 | } | |
1080 | else | |
1081 | { | |
d90e1b42 UD |
1082 | *strptr = newstr; |
1083 | str = newstr + strleng; | |
d64b6ad0 UD |
1084 | strsize *= 2; |
1085 | } | |
1086 | } | |
1087 | ||
4aebaa6b UD |
1088 | n = __wcrtomb (!(flags & SUPPRESS) ? str : NULL, c, |
1089 | &state); | |
a1ffb40e | 1090 | if (__glibc_unlikely (n == (size_t) -1)) |
d64b6ad0 UD |
1091 | encode_error (); |
1092 | ||
9dd87afb | 1093 | assert (n <= MB_LEN_MAX); |
d64b6ad0 UD |
1094 | str += n; |
1095 | } | |
1096 | #else | |
1097 | /* This is easy. */ | |
1098 | if (!(flags & SUPPRESS)) | |
1099 | { | |
1100 | *str++ = c; | |
1101 | if ((flags & MALLOC) | |
1102 | && (char *) str == *strptr + strsize) | |
1103 | { | |
1104 | /* Enlarge the buffer. */ | |
1105 | str = (char *) realloc (*strptr, 2 * strsize); | |
1106 | if (str == NULL) | |
1107 | { | |
1108 | /* Can't allocate that much. Last-ditch | |
1109 | effort. */ | |
1110 | str = (char *) realloc (*strptr, strsize + 1); | |
1111 | if (str == NULL) | |
1112 | { | |
267c54dc | 1113 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
1114 | { |
1115 | done = EOF; | |
1116 | goto errout; | |
1117 | } | |
d64b6ad0 UD |
1118 | /* We lose. Oh well. Terminate the |
1119 | string and stop converting, | |
1120 | so at least we don't skip any input. */ | |
1121 | ((char *) (*strptr))[strsize - 1] = '\0'; | |
267c54dc | 1122 | strptr = NULL; |
d64b6ad0 UD |
1123 | ++done; |
1124 | conv_error (); | |
1125 | } | |
1126 | else | |
1127 | { | |
1128 | *strptr = (char *) str; | |
1129 | str += strsize; | |
1130 | ++strsize; | |
1131 | } | |
1132 | } | |
1133 | else | |
1134 | { | |
1135 | *strptr = (char *) str; | |
1136 | str += strsize; | |
1137 | strsize *= 2; | |
1138 | } | |
1139 | } | |
1140 | } | |
1141 | #endif | |
28f540f4 | 1142 | } |
d64b6ad0 | 1143 | while ((width <= 0 || --width > 0) && inchar () != EOF); |
28f540f4 | 1144 | |
d64b6ad0 UD |
1145 | if (!(flags & SUPPRESS)) |
1146 | { | |
72acaddf | 1147 | #ifdef COMPILE_WSCANF |
44f8759b | 1148 | /* We have to emit the code to get into the initial |
d64b6ad0 UD |
1149 | state. */ |
1150 | char buf[MB_LEN_MAX]; | |
4aebaa6b | 1151 | size_t n = __wcrtomb (buf, L'\0', &state); |
d64b6ad0 UD |
1152 | if (n > 0 && (flags & MALLOC) |
1153 | && str + n >= *strptr + strsize) | |
1154 | { | |
1155 | /* Enlarge the buffer. */ | |
d90e1b42 UD |
1156 | size_t strleng = str - *strptr; |
1157 | char *newstr; | |
1158 | ||
1159 | newstr = (char *) realloc (*strptr, strleng + n + 1); | |
1160 | if (newstr == NULL) | |
d64b6ad0 | 1161 | { |
267c54dc | 1162 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
1163 | { |
1164 | done = EOF; | |
1165 | goto errout; | |
1166 | } | |
d64b6ad0 UD |
1167 | /* We lose. Oh well. Terminate the string |
1168 | and stop converting, so at least we don't | |
1169 | skip any input. */ | |
d90e1b42 | 1170 | ((char *) (*strptr))[strleng] = '\0'; |
267c54dc | 1171 | strptr = NULL; |
d64b6ad0 UD |
1172 | ++done; |
1173 | conv_error (); | |
1174 | } | |
1175 | else | |
1176 | { | |
d90e1b42 UD |
1177 | *strptr = newstr; |
1178 | str = newstr + strleng; | |
1179 | strsize = strleng + n + 1; | |
d64b6ad0 UD |
1180 | } |
1181 | } | |
1182 | ||
1183 | str = __mempcpy (str, buf, n); | |
1184 | #endif | |
d90e1b42 | 1185 | *str++ = '\0'; |
d64b6ad0 UD |
1186 | |
1187 | if ((flags & MALLOC) && str - *strptr != strsize) | |
1188 | { | |
1189 | char *cp = (char *) realloc (*strptr, str - *strptr); | |
1190 | if (cp != NULL) | |
1191 | *strptr = cp; | |
1192 | } | |
267c54dc | 1193 | strptr = NULL; |
d64b6ad0 UD |
1194 | |
1195 | ++done; | |
1196 | } | |
1197 | break; | |
28f540f4 | 1198 | } |
d64b6ad0 | 1199 | /* FALLTHROUGH */ |
28f540f4 | 1200 | |
d64b6ad0 | 1201 | case L_('S'): |
2c6fe0bd | 1202 | { |
72acaddf | 1203 | #ifndef COMPILE_WSCANF |
d64b6ad0 UD |
1204 | mbstate_t cstate; |
1205 | #endif | |
1206 | ||
1207 | /* Wide character string. */ | |
874aa523 | 1208 | STRING_ARG (wstr, wchar_t, 100); |
2c6fe0bd | 1209 | |
d64b6ad0 | 1210 | c = inchar (); |
694b49ca | 1211 | if (__builtin_expect (c == EOF, 0)) |
d64b6ad0 UD |
1212 | input_error (); |
1213 | ||
72acaddf | 1214 | #ifndef COMPILE_WSCANF |
d64b6ad0 UD |
1215 | memset (&cstate, '\0', sizeof (cstate)); |
1216 | #endif | |
1217 | ||
2c6fe0bd UD |
1218 | do |
1219 | { | |
d64b6ad0 | 1220 | if (ISSPACE (c)) |
2c6fe0bd | 1221 | { |
44f8759b | 1222 | ungetc_not_eof (c, s); |
2c6fe0bd UD |
1223 | break; |
1224 | } | |
1225 | ||
72acaddf | 1226 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
1227 | /* This is easy. */ |
1228 | if (!(flags & SUPPRESS)) | |
1229 | { | |
1230 | *wstr++ = c; | |
1231 | if ((flags & MALLOC) | |
1232 | && wstr == (wchar_t *) *strptr + strsize) | |
1233 | { | |
1234 | /* Enlarge the buffer. */ | |
1235 | wstr = (wchar_t *) realloc (*strptr, | |
1236 | (2 * strsize) | |
1237 | * sizeof (wchar_t)); | |
1238 | if (wstr == NULL) | |
1239 | { | |
1240 | /* Can't allocate that much. Last-ditch | |
874aa523 | 1241 | effort. */ |
d64b6ad0 | 1242 | wstr = (wchar_t *) realloc (*strptr, |
d90e1b42 UD |
1243 | (strsize + 1) |
1244 | * sizeof (wchar_t)); | |
d64b6ad0 UD |
1245 | if (wstr == NULL) |
1246 | { | |
267c54dc | 1247 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
1248 | { |
1249 | done = EOF; | |
1250 | goto errout; | |
1251 | } | |
d64b6ad0 UD |
1252 | /* We lose. Oh well. Terminate the string |
1253 | and stop converting, so at least we don't | |
1254 | skip any input. */ | |
1255 | ((wchar_t *) (*strptr))[strsize - 1] = L'\0'; | |
267c54dc | 1256 | strptr = NULL; |
d64b6ad0 UD |
1257 | ++done; |
1258 | conv_error (); | |
1259 | } | |
1260 | else | |
1261 | { | |
1262 | *strptr = (char *) wstr; | |
1263 | wstr += strsize; | |
1264 | ++strsize; | |
1265 | } | |
1266 | } | |
1267 | else | |
1268 | { | |
1269 | *strptr = (char *) wstr; | |
1270 | wstr += strsize; | |
1271 | strsize *= 2; | |
1272 | } | |
1273 | } | |
1274 | } | |
1275 | #else | |
1276 | { | |
68bd3326 | 1277 | char buf[1]; |
d64b6ad0 UD |
1278 | |
1279 | buf[0] = c; | |
d64b6ad0 UD |
1280 | |
1281 | while (1) | |
1282 | { | |
1283 | size_t n; | |
1284 | ||
1285 | n = __mbrtowc (!(flags & SUPPRESS) ? wstr : NULL, | |
2f278c94 | 1286 | buf, 1, &cstate); |
d64b6ad0 UD |
1287 | |
1288 | if (n == (size_t) -2) | |
1289 | { | |
1290 | /* Possibly correct character, just not enough | |
1291 | input. */ | |
a1ffb40e | 1292 | if (__glibc_unlikely (inchar () == EOF)) |
d64b6ad0 UD |
1293 | encode_error (); |
1294 | ||
2f278c94 | 1295 | buf[0] = c; |
d64b6ad0 UD |
1296 | continue; |
1297 | } | |
1298 | ||
a1ffb40e | 1299 | if (__glibc_unlikely (n != 1)) |
d64b6ad0 UD |
1300 | encode_error (); |
1301 | ||
1302 | /* We have a match. */ | |
2f278c94 | 1303 | ++wstr; |
d64b6ad0 UD |
1304 | break; |
1305 | } | |
1306 | ||
1307 | if (!(flags & SUPPRESS) && (flags & MALLOC) | |
1308 | && wstr == (wchar_t *) *strptr + strsize) | |
1309 | { | |
1310 | /* Enlarge the buffer. */ | |
1311 | wstr = (wchar_t *) realloc (*strptr, | |
1312 | (2 * strsize | |
1313 | * sizeof (wchar_t))); | |
1314 | if (wstr == NULL) | |
1315 | { | |
1316 | /* Can't allocate that much. Last-ditch effort. */ | |
1317 | wstr = (wchar_t *) realloc (*strptr, | |
1318 | ((strsize + 1) | |
1319 | * sizeof (wchar_t))); | |
1320 | if (wstr == NULL) | |
1321 | { | |
267c54dc | 1322 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
1323 | { |
1324 | done = EOF; | |
1325 | goto errout; | |
1326 | } | |
d64b6ad0 UD |
1327 | /* We lose. Oh well. Terminate the |
1328 | string and stop converting, so at | |
1329 | least we don't skip any input. */ | |
1330 | ((wchar_t *) (*strptr))[strsize - 1] = L'\0'; | |
267c54dc | 1331 | strptr = NULL; |
d64b6ad0 UD |
1332 | ++done; |
1333 | conv_error (); | |
1334 | } | |
1335 | else | |
1336 | { | |
1337 | *strptr = (char *) wstr; | |
1338 | wstr += strsize; | |
1339 | ++strsize; | |
1340 | } | |
1341 | } | |
1342 | else | |
1343 | { | |
1344 | *strptr = (char *) wstr; | |
1345 | wstr += strsize; | |
1346 | strsize *= 2; | |
1347 | } | |
1348 | } | |
1349 | } | |
1350 | #endif | |
2c6fe0bd | 1351 | } |
d64b6ad0 | 1352 | while ((width <= 0 || --width > 0) && inchar () != EOF); |
2c6fe0bd UD |
1353 | |
1354 | if (!(flags & SUPPRESS)) | |
1355 | { | |
d64b6ad0 UD |
1356 | *wstr++ = L'\0'; |
1357 | ||
1358 | if ((flags & MALLOC) && wstr - (wchar_t *) *strptr != strsize) | |
1359 | { | |
1360 | wchar_t *cp = (wchar_t *) realloc (*strptr, | |
1361 | ((wstr | |
1362 | - (wchar_t *) *strptr) | |
1363 | * sizeof(wchar_t))); | |
1364 | if (cp != NULL) | |
1365 | *strptr = (char *) cp; | |
1366 | } | |
267c54dc | 1367 | strptr = NULL; |
d64b6ad0 | 1368 | |
2c6fe0bd UD |
1369 | ++done; |
1370 | } | |
1371 | } | |
1372 | break; | |
1373 | ||
d64b6ad0 UD |
1374 | case L_('x'): /* Hexadecimal integer. */ |
1375 | case L_('X'): /* Ditto. */ | |
28f540f4 | 1376 | base = 16; |
28f540f4 RM |
1377 | goto number; |
1378 | ||
d64b6ad0 | 1379 | case L_('o'): /* Octal integer. */ |
28f540f4 | 1380 | base = 8; |
28f540f4 RM |
1381 | goto number; |
1382 | ||
d64b6ad0 | 1383 | case L_('u'): /* Unsigned decimal integer. */ |
28f540f4 | 1384 | base = 10; |
28f540f4 RM |
1385 | goto number; |
1386 | ||
d64b6ad0 | 1387 | case L_('d'): /* Signed decimal integer. */ |
28f540f4 | 1388 | base = 10; |
3979024a | 1389 | flags |= NUMBER_SIGNED; |
28f540f4 RM |
1390 | goto number; |
1391 | ||
d64b6ad0 | 1392 | case L_('i'): /* Generic number. */ |
28f540f4 | 1393 | base = 0; |
3979024a | 1394 | flags |= NUMBER_SIGNED; |
28f540f4 RM |
1395 | |
1396 | number: | |
14ea22e9 | 1397 | c = inchar (); |
a1ffb40e | 1398 | if (__glibc_unlikely (c == EOF)) |
96aa2d94 | 1399 | input_error (); |
28f540f4 RM |
1400 | |
1401 | /* Check for a sign. */ | |
d64b6ad0 | 1402 | if (c == L_('-') || c == L_('+')) |
28f540f4 | 1403 | { |
95e83974 | 1404 | char_buffer_add (&charbuf, c); |
28f540f4 RM |
1405 | if (width > 0) |
1406 | --width; | |
3867ee64 | 1407 | c = inchar (); |
28f540f4 RM |
1408 | } |
1409 | ||
1410 | /* Look for a leading indication of base. */ | |
d64b6ad0 | 1411 | if (width != 0 && c == L_('0')) |
28f540f4 RM |
1412 | { |
1413 | if (width > 0) | |
1414 | --width; | |
28f540f4 | 1415 | |
95e83974 | 1416 | char_buffer_add (&charbuf, c); |
3867ee64 | 1417 | c = inchar (); |
28f540f4 | 1418 | |
d64b6ad0 | 1419 | if (width != 0 && TOLOWER (c) == L_('x')) |
28f540f4 RM |
1420 | { |
1421 | if (base == 0) | |
1422 | base = 16; | |
1423 | if (base == 16) | |
1424 | { | |
1425 | if (width > 0) | |
1426 | --width; | |
3867ee64 | 1427 | c = inchar (); |
28f540f4 RM |
1428 | } |
1429 | } | |
1430 | else if (base == 0) | |
1431 | base = 8; | |
1432 | } | |
1433 | ||
1434 | if (base == 0) | |
1435 | base = 10; | |
1436 | ||
694b49ca | 1437 | if (base == 10 && __builtin_expect ((flags & I18N) != 0, 0)) |
28f540f4 | 1438 | { |
a1d84548 UD |
1439 | int from_level; |
1440 | int to_level; | |
eac4282f | 1441 | int level; |
72acaddf | 1442 | #ifdef COMPILE_WSCANF |
e87f0c33 | 1443 | const wchar_t *wcdigits[10]; |
f448d449 | 1444 | const wchar_t *wcdigits_extended[10]; |
a1d84548 UD |
1445 | #else |
1446 | const char *mbdigits[10]; | |
f448d449 | 1447 | const char *mbdigits_extended[10]; |
a1d84548 | 1448 | #endif |
f448d449 UD |
1449 | /* "to_inpunct" is a map from ASCII digits to their |
1450 | equivalent in locale. This is defined for locales | |
1451 | which use an extra digits set. */ | |
1452 | wctrans_t map = __wctrans ("to_inpunct"); | |
a1d84548 | 1453 | int n; |
05be689b | 1454 | |
a1d84548 | 1455 | from_level = 0; |
72acaddf | 1456 | #ifdef COMPILE_WSCANF |
a1d84548 UD |
1457 | to_level = _NL_CURRENT_WORD (LC_CTYPE, |
1458 | _NL_CTYPE_INDIGITS_WC_LEN) - 1; | |
1459 | #else | |
e3b22ad3 | 1460 | to_level = (uint32_t) curctype->values[_NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN)].word - 1; |
a1d84548 UD |
1461 | #endif |
1462 | ||
f448d449 | 1463 | /* Get the alternative digit forms if there are any. */ |
a1ffb40e | 1464 | if (__glibc_unlikely (map != NULL)) |
f448d449 UD |
1465 | { |
1466 | /* Adding new level for extra digits set in locale file. */ | |
1467 | ++to_level; | |
1468 | ||
1469 | for (n = 0; n < 10; ++n) | |
1470 | { | |
1471 | #ifdef COMPILE_WSCANF | |
1472 | wcdigits[n] = (const wchar_t *) | |
f095bb72 | 1473 | _NL_CURRENT (LC_CTYPE, _NL_CTYPE_INDIGITS0_WC + n); |
f448d449 UD |
1474 | |
1475 | wchar_t *wc_extended = (wchar_t *) | |
1476 | alloca ((to_level + 2) * sizeof (wchar_t)); | |
1477 | __wmemcpy (wc_extended, wcdigits[n], to_level); | |
1478 | wc_extended[to_level] = __towctrans (L'0' + n, map); | |
1479 | wc_extended[to_level + 1] = '\0'; | |
1480 | wcdigits_extended[n] = wc_extended; | |
1481 | #else | |
1482 | mbdigits[n] | |
f095bb72 | 1483 | = curctype->values[_NL_CTYPE_INDIGITS0_MB + n].string; |
f448d449 UD |
1484 | |
1485 | /* Get the equivalent wide char in map. */ | |
1486 | wint_t extra_wcdigit = __towctrans (L'0' + n, map); | |
1487 | ||
1488 | /* Convert it to multibyte representation. */ | |
1489 | mbstate_t state; | |
1490 | memset (&state, '\0', sizeof (state)); | |
1491 | ||
1492 | char extra_mbdigit[MB_LEN_MAX]; | |
1493 | size_t mblen | |
1494 | = __wcrtomb (extra_mbdigit, extra_wcdigit, &state); | |
1495 | ||
1496 | if (mblen == (size_t) -1) | |
1497 | { | |
1498 | /* Ignore this new level. */ | |
1499 | map = NULL; | |
1500 | break; | |
1501 | } | |
1502 | ||
1503 | /* Calculate the length of mbdigits[n]. */ | |
1504 | const char *last_char = mbdigits[n]; | |
1505 | for (level = 0; level < to_level; ++level) | |
1506 | last_char = strchr (last_char, '\0') + 1; | |
1507 | ||
1508 | size_t mbdigits_len = last_char - mbdigits[n]; | |
1509 | ||
1510 | /* Allocate memory for extended multibyte digit. */ | |
1511 | char *mb_extended; | |
1512 | mb_extended = (char *) alloca (mbdigits_len + mblen + 1); | |
1513 | ||
1514 | /* And get the mbdigits + extra_digit string. */ | |
1515 | *(char *) __mempcpy (__mempcpy (mb_extended, mbdigits[n], | |
1516 | mbdigits_len), | |
1517 | extra_mbdigit, mblen) = '\0'; | |
1518 | mbdigits_extended[n] = mb_extended; | |
1519 | #endif | |
1520 | } | |
1521 | } | |
1522 | ||
eac4282f UD |
1523 | /* Read the number into workspace. */ |
1524 | while (c != EOF && width != 0) | |
a1d84548 | 1525 | { |
eac4282f UD |
1526 | /* In this round we get the pointer to the digit strings |
1527 | and also perform the first round of comparisons. */ | |
1528 | for (n = 0; n < 10; ++n) | |
1529 | { | |
1530 | /* Get the string for the digits with value N. */ | |
72acaddf | 1531 | #ifdef COMPILE_WSCANF |
2a9feb92 CM |
1532 | |
1533 | /* wcdigits_extended[] is fully set in the loop | |
1534 | above, but the test for "map != NULL" is done | |
1535 | inside the loop here and outside the loop there. */ | |
1536 | DIAG_PUSH_NEEDS_COMMENT; | |
1537 | DIAG_IGNORE_NEEDS_COMMENT (4.7, "-Wmaybe-uninitialized"); | |
1538 | ||
a1ffb40e | 1539 | if (__glibc_unlikely (map != NULL)) |
f448d449 UD |
1540 | wcdigits[n] = wcdigits_extended[n]; |
1541 | else | |
1542 | wcdigits[n] = (const wchar_t *) | |
1543 | _NL_CURRENT (LC_CTYPE, _NL_CTYPE_INDIGITS0_WC + n); | |
eac4282f | 1544 | wcdigits[n] += from_level; |
a1d84548 | 1545 | |
2a9feb92 CM |
1546 | DIAG_POP_NEEDS_COMMENT; |
1547 | ||
52a7f7c0 | 1548 | if (c == (wint_t) *wcdigits[n]) |
eac4282f UD |
1549 | { |
1550 | to_level = from_level; | |
1551 | break; | |
1552 | } | |
1553 | ||
1554 | /* Advance the pointer to the next string. */ | |
1555 | ++wcdigits[n]; | |
a1d84548 | 1556 | #else |
eac4282f UD |
1557 | const char *cmpp; |
1558 | int avail = width > 0 ? width : INT_MAX; | |
e87f0c33 | 1559 | |
a1ffb40e | 1560 | if (__glibc_unlikely (map != NULL)) |
f448d449 UD |
1561 | mbdigits[n] = mbdigits_extended[n]; |
1562 | else | |
1563 | mbdigits[n] | |
1564 | = curctype->values[_NL_CTYPE_INDIGITS0_MB + n].string; | |
a1d84548 | 1565 | |
eac4282f UD |
1566 | for (level = 0; level < from_level; level++) |
1567 | mbdigits[n] = strchr (mbdigits[n], '\0') + 1; | |
a1d84548 | 1568 | |
eac4282f | 1569 | cmpp = mbdigits[n]; |
ecb72bd5 | 1570 | while ((unsigned char) *cmpp == c && avail >= 0) |
eac4282f UD |
1571 | { |
1572 | if (*++cmpp == '\0') | |
1573 | break; | |
1574 | else | |
1575 | { | |
ecb72bd5 | 1576 | if (avail == 0 || inchar () == EOF) |
eac4282f UD |
1577 | break; |
1578 | --avail; | |
1579 | } | |
1580 | } | |
1581 | ||
1582 | if (*cmpp == '\0') | |
1583 | { | |
1584 | if (width > 0) | |
1585 | width = avail; | |
1586 | to_level = from_level; | |
1587 | break; | |
1588 | } | |
a1d84548 | 1589 | |
eac4282f UD |
1590 | /* We are pushing all read characters back. */ |
1591 | if (cmpp > mbdigits[n]) | |
1592 | { | |
1593 | ungetc (c, s); | |
1594 | while (--cmpp > mbdigits[n]) | |
44f8759b UD |
1595 | ungetc_not_eof ((unsigned char) *cmpp, s); |
1596 | c = (unsigned char) *cmpp; | |
eac4282f UD |
1597 | } |
1598 | ||
1599 | /* Advance the pointer to the next string. */ | |
1600 | mbdigits[n] = strchr (mbdigits[n], '\0') + 1; | |
a1d84548 | 1601 | #endif |
eac4282f | 1602 | } |
a1d84548 | 1603 | |
eac4282f | 1604 | if (n == 10) |
a1d84548 | 1605 | { |
eac4282f UD |
1606 | /* Have not yet found the digit. */ |
1607 | for (level = from_level + 1; level <= to_level; ++level) | |
a1d84548 | 1608 | { |
eac4282f UD |
1609 | /* Search all ten digits of this level. */ |
1610 | for (n = 0; n < 10; ++n) | |
1611 | { | |
72acaddf | 1612 | #ifdef COMPILE_WSCANF |
52a7f7c0 | 1613 | if (c == (wint_t) *wcdigits[n]) |
eac4282f | 1614 | break; |
a1d84548 | 1615 | |
eac4282f UD |
1616 | /* Advance the pointer to the next string. */ |
1617 | ++wcdigits[n]; | |
a1d84548 | 1618 | #else |
eac4282f UD |
1619 | const char *cmpp; |
1620 | int avail = width > 0 ? width : INT_MAX; | |
1621 | ||
1622 | cmpp = mbdigits[n]; | |
ecb72bd5 | 1623 | while ((unsigned char) *cmpp == c && avail >= 0) |
eac4282f UD |
1624 | { |
1625 | if (*++cmpp == '\0') | |
1626 | break; | |
1627 | else | |
1628 | { | |
ecb72bd5 | 1629 | if (avail == 0 || inchar () == EOF) |
eac4282f UD |
1630 | break; |
1631 | --avail; | |
1632 | } | |
1633 | } | |
a1d84548 | 1634 | |
eac4282f UD |
1635 | if (*cmpp == '\0') |
1636 | { | |
1637 | if (width > 0) | |
1638 | width = avail; | |
1639 | break; | |
1640 | } | |
1641 | ||
1642 | /* We are pushing all read characters back. */ | |
1643 | if (cmpp > mbdigits[n]) | |
1644 | { | |
1645 | ungetc (c, s); | |
1646 | while (--cmpp > mbdigits[n]) | |
44f8759b UD |
1647 | ungetc_not_eof ((unsigned char) *cmpp, s); |
1648 | c = (unsigned char) *cmpp; | |
eac4282f UD |
1649 | } |
1650 | ||
1651 | /* Advance the pointer to the next string. */ | |
1652 | mbdigits[n] = strchr (mbdigits[n], '\0') + 1; | |
1653 | #endif | |
1654 | } | |
1655 | ||
1656 | if (n < 10) | |
a1d84548 | 1657 | { |
eac4282f UD |
1658 | /* Found it. */ |
1659 | from_level = level; | |
1660 | to_level = level; | |
1661 | break; | |
a1d84548 | 1662 | } |
eac4282f UD |
1663 | } |
1664 | } | |
a1d84548 | 1665 | |
eac4282f UD |
1666 | if (n < 10) |
1667 | c = L_('0') + n; | |
3979024a | 1668 | else if (flags & GROUP) |
eac4282f UD |
1669 | { |
1670 | /* Try matching against the thousands separator. */ | |
1671 | #ifdef COMPILE_WSCANF | |
1672 | if (c != thousands) | |
1673 | break; | |
1674 | #else | |
1675 | const char *cmpp = thousands; | |
1676 | int avail = width > 0 ? width : INT_MAX; | |
1677 | ||
ecb72bd5 | 1678 | while ((unsigned char) *cmpp == c && avail >= 0) |
eac4282f | 1679 | { |
95e83974 | 1680 | char_buffer_add (&charbuf, c); |
eac4282f | 1681 | if (*++cmpp == '\0') |
a1d84548 | 1682 | break; |
eac4282f UD |
1683 | else |
1684 | { | |
ecb72bd5 | 1685 | if (avail == 0 || inchar () == EOF) |
eac4282f UD |
1686 | break; |
1687 | --avail; | |
1688 | } | |
1689 | } | |
a1d84548 | 1690 | |
95e83974 FW |
1691 | if (char_buffer_error (&charbuf)) |
1692 | { | |
1693 | __set_errno (ENOMEM); | |
1694 | done = EOF; | |
1695 | goto errout; | |
1696 | } | |
1697 | ||
eac4282f UD |
1698 | if (*cmpp != '\0') |
1699 | { | |
1700 | /* We are pushing all read characters back. */ | |
1701 | if (cmpp > thousands) | |
1702 | { | |
95e83974 | 1703 | charbuf.current -= cmpp - thousands; |
eac4282f UD |
1704 | ungetc (c, s); |
1705 | while (--cmpp > thousands) | |
44f8759b UD |
1706 | ungetc_not_eof ((unsigned char) *cmpp, s); |
1707 | c = (unsigned char) *cmpp; | |
eac4282f UD |
1708 | } |
1709 | break; | |
a1d84548 UD |
1710 | } |
1711 | ||
eac4282f UD |
1712 | if (width > 0) |
1713 | width = avail; | |
a1d84548 | 1714 | |
eac4282f | 1715 | /* The last thousands character will be added back by |
95e83974 | 1716 | the char_buffer_add below. */ |
996b03dc | 1717 | --charbuf.current; |
eac4282f | 1718 | #endif |
a1d84548 | 1719 | } |
eac4282f UD |
1720 | else |
1721 | break; | |
a1d84548 | 1722 | |
95e83974 | 1723 | char_buffer_add (&charbuf, c); |
eac4282f UD |
1724 | if (width > 0) |
1725 | --width; | |
a1d84548 | 1726 | |
eac4282f UD |
1727 | c = inchar (); |
1728 | } | |
7752137a | 1729 | } |
a1d84548 UD |
1730 | else |
1731 | /* Read the number into workspace. */ | |
1732 | while (c != EOF && width != 0) | |
1733 | { | |
72acaddf UD |
1734 | if (base == 16) |
1735 | { | |
1736 | if (!ISXDIGIT (c)) | |
1737 | break; | |
1738 | } | |
52a7f7c0 | 1739 | else if (!ISDIGIT (c) || (int) (c - L_('0')) >= base) |
72acaddf | 1740 | { |
3979024a | 1741 | if (base == 10 && (flags & GROUP)) |
72acaddf UD |
1742 | { |
1743 | /* Try matching against the thousands separator. */ | |
1744 | #ifdef COMPILE_WSCANF | |
1745 | if (c != thousands) | |
1746 | break; | |
1747 | #else | |
1748 | const char *cmpp = thousands; | |
1749 | int avail = width > 0 ? width : INT_MAX; | |
1750 | ||
ecb72bd5 | 1751 | while ((unsigned char) *cmpp == c && avail >= 0) |
eac4282f | 1752 | { |
95e83974 | 1753 | char_buffer_add (&charbuf, c); |
eac4282f UD |
1754 | if (*++cmpp == '\0') |
1755 | break; | |
1756 | else | |
1757 | { | |
ecb72bd5 | 1758 | if (avail == 0 || inchar () == EOF) |
eac4282f UD |
1759 | break; |
1760 | --avail; | |
1761 | } | |
1762 | } | |
72acaddf | 1763 | |
95e83974 FW |
1764 | if (char_buffer_error (&charbuf)) |
1765 | { | |
1766 | __set_errno (ENOMEM); | |
1767 | done = EOF; | |
1768 | goto errout; | |
1769 | } | |
1770 | ||
72acaddf UD |
1771 | if (*cmpp != '\0') |
1772 | { | |
eac4282f | 1773 | /* We are pushing all read characters back. */ |
72acaddf UD |
1774 | if (cmpp > thousands) |
1775 | { | |
95e83974 | 1776 | charbuf.current -= cmpp - thousands; |
72acaddf UD |
1777 | ungetc (c, s); |
1778 | while (--cmpp > thousands) | |
44f8759b UD |
1779 | ungetc_not_eof ((unsigned char) *cmpp, s); |
1780 | c = (unsigned char) *cmpp; | |
72acaddf UD |
1781 | } |
1782 | break; | |
1783 | } | |
eac4282f | 1784 | |
72acaddf | 1785 | if (width > 0) |
eac4282f UD |
1786 | width = avail; |
1787 | ||
1788 | /* The last thousands character will be added back by | |
95e83974 FW |
1789 | the char_buffer_add below. */ |
1790 | --charbuf.current; | |
72acaddf UD |
1791 | #endif |
1792 | } | |
1793 | else | |
1794 | break; | |
1795 | } | |
95e83974 | 1796 | char_buffer_add (&charbuf, c); |
a1d84548 UD |
1797 | if (width > 0) |
1798 | --width; | |
1799 | ||
1800 | c = inchar (); | |
1801 | } | |
28f540f4 | 1802 | |
95e83974 FW |
1803 | if (char_buffer_error (&charbuf)) |
1804 | { | |
1805 | __set_errno (ENOMEM); | |
1806 | done = EOF; | |
1807 | goto errout; | |
1808 | } | |
1809 | ||
1810 | if (char_buffer_size (&charbuf) == 0 | |
1811 | || (char_buffer_size (&charbuf) == 1 | |
1812 | && (char_buffer_start (&charbuf)[0] == L_('+') | |
1813 | || char_buffer_start (&charbuf)[0] == L_('-')))) | |
c53a89d4 UD |
1814 | { |
1815 | /* There was no number. If we are supposed to read a pointer | |
1816 | we must recognize "(nil)" as well. */ | |
95e83974 | 1817 | if (__builtin_expect (char_buffer_size (&charbuf) == 0 |
4c02bf1a | 1818 | && (flags & READ_POINTER) |
728dab0e | 1819 | && (width < 0 || width >= 5) |
694b49ca UD |
1820 | && c == '(' |
1821 | && TOLOWER (inchar ()) == L_('n') | |
1822 | && TOLOWER (inchar ()) == L_('i') | |
1823 | && TOLOWER (inchar ()) == L_('l') | |
1824 | && inchar () == L_(')'), 1)) | |
c53a89d4 UD |
1825 | /* We must produce the value of a NULL pointer. A single |
1826 | '0' digit is enough. */ | |
95e83974 | 1827 | char_buffer_add (&charbuf, L_('0')); |
c53a89d4 UD |
1828 | else |
1829 | { | |
1830 | /* The last read character is not part of the number | |
1831 | anymore. */ | |
655c0697 | 1832 | ungetc (c, s); |
c53a89d4 UD |
1833 | |
1834 | conv_error (); | |
1835 | } | |
1836 | } | |
1837 | else | |
1838 | /* The just read character is not part of the number anymore. */ | |
655c0697 | 1839 | ungetc (c, s); |
28f540f4 RM |
1840 | |
1841 | /* Convert the number. */ | |
95e83974 FW |
1842 | char_buffer_add (&charbuf, L_('\0')); |
1843 | if (char_buffer_error (&charbuf)) | |
1844 | { | |
1845 | __set_errno (ENOMEM); | |
1846 | done = EOF; | |
1847 | goto errout; | |
1848 | } | |
96d0213e | 1849 | if (need_longlong && (flags & LONGDBL)) |
f0bf9cb9 | 1850 | { |
3979024a | 1851 | if (flags & NUMBER_SIGNED) |
95e83974 FW |
1852 | num.q = __strtoll_internal |
1853 | (char_buffer_start (&charbuf), &tw, base, flags & GROUP); | |
f0bf9cb9 | 1854 | else |
95e83974 FW |
1855 | num.uq = __strtoull_internal |
1856 | (char_buffer_start (&charbuf), &tw, base, flags & GROUP); | |
f0bf9cb9 | 1857 | } |
28f540f4 | 1858 | else |
f0bf9cb9 | 1859 | { |
3979024a | 1860 | if (flags & NUMBER_SIGNED) |
95e83974 FW |
1861 | num.l = __strtol_internal |
1862 | (char_buffer_start (&charbuf), &tw, base, flags & GROUP); | |
f0bf9cb9 | 1863 | else |
95e83974 FW |
1864 | num.ul = __strtoul_internal |
1865 | (char_buffer_start (&charbuf), &tw, base, flags & GROUP); | |
f0bf9cb9 | 1866 | } |
95e83974 | 1867 | if (__glibc_unlikely (char_buffer_start (&charbuf) == tw)) |
28f540f4 RM |
1868 | conv_error (); |
1869 | ||
0793d348 | 1870 | if (!(flags & SUPPRESS)) |
28f540f4 | 1871 | { |
3979024a | 1872 | if (flags & NUMBER_SIGNED) |
28f540f4 | 1873 | { |
96d0213e | 1874 | if (need_longlong && (flags & LONGDBL)) |
3979024a | 1875 | *ARG (LONGLONG int *) = num.q; |
859a09cf | 1876 | else if (need_long && (flags & LONG)) |
3979024a | 1877 | *ARG (long int *) = num.l; |
0793d348 | 1878 | else if (flags & SHORT) |
3979024a | 1879 | *ARG (short int *) = (short int) num.l; |
859a09cf | 1880 | else if (!(flags & CHAR)) |
3979024a | 1881 | *ARG (int *) = (int) num.l; |
859a09cf | 1882 | else |
3979024a | 1883 | *ARG (signed char *) = (signed char) num.ul; |
28f540f4 RM |
1884 | } |
1885 | else | |
1886 | { | |
96d0213e | 1887 | if (need_longlong && (flags & LONGDBL)) |
3979024a | 1888 | *ARG (unsigned LONGLONG int *) = num.uq; |
859a09cf | 1889 | else if (need_long && (flags & LONG)) |
3979024a | 1890 | *ARG (unsigned long int *) = num.ul; |
0793d348 | 1891 | else if (flags & SHORT) |
3979024a UD |
1892 | *ARG (unsigned short int *) |
1893 | = (unsigned short int) num.ul; | |
859a09cf | 1894 | else if (!(flags & CHAR)) |
3979024a | 1895 | *ARG (unsigned int *) = (unsigned int) num.ul; |
859a09cf | 1896 | else |
3979024a | 1897 | *ARG (unsigned char *) = (unsigned char) num.ul; |
28f540f4 RM |
1898 | } |
1899 | ++done; | |
1900 | } | |
1901 | break; | |
1902 | ||
d64b6ad0 UD |
1903 | case L_('e'): /* Floating-point numbers. */ |
1904 | case L_('E'): | |
1905 | case L_('f'): | |
233f5f68 | 1906 | case L_('F'): |
d64b6ad0 UD |
1907 | case L_('g'): |
1908 | case L_('G'): | |
1909 | case L_('a'): | |
1910 | case L_('A'): | |
14ea22e9 | 1911 | c = inchar (); |
ac2ca022 UD |
1912 | if (width > 0) |
1913 | --width; | |
a1ffb40e | 1914 | if (__glibc_unlikely (c == EOF)) |
96aa2d94 | 1915 | input_error (); |
28f540f4 | 1916 | |
6ecec3b6 | 1917 | got_digit = got_dot = got_e = 0; |
918b198d | 1918 | |
28f540f4 | 1919 | /* Check for a sign. */ |
d64b6ad0 | 1920 | if (c == L_('-') || c == L_('+')) |
28f540f4 | 1921 | { |
d64b6ad0 | 1922 | negative = c == L_('-'); |
a1ffb40e | 1923 | if (__glibc_unlikely (width == 0 || inchar () == EOF)) |
28f540f4 | 1924 | /* EOF is only an input error before we read any chars. */ |
96aa2d94 | 1925 | conv_error (); |
28f540f4 RM |
1926 | if (width > 0) |
1927 | --width; | |
1928 | } | |
01cdeca0 RM |
1929 | else |
1930 | negative = 0; | |
28f540f4 | 1931 | |
9b26f5c4 | 1932 | /* Take care for the special arguments "nan" and "inf". */ |
d64b6ad0 | 1933 | if (TOLOWER (c) == L_('n')) |
9b26f5c4 UD |
1934 | { |
1935 | /* Maybe "nan". */ | |
95e83974 | 1936 | char_buffer_add (&charbuf, c); |
694b49ca UD |
1937 | if (__builtin_expect (width == 0 |
1938 | || inchar () == EOF | |
1939 | || TOLOWER (c) != L_('a'), 0)) | |
f4efd068 | 1940 | conv_error (); |
9df54168 UD |
1941 | if (width > 0) |
1942 | --width; | |
95e83974 | 1943 | char_buffer_add (&charbuf, c); |
694b49ca UD |
1944 | if (__builtin_expect (width == 0 |
1945 | || inchar () == EOF | |
1946 | || TOLOWER (c) != L_('n'), 0)) | |
f4efd068 | 1947 | conv_error (); |
9df54168 UD |
1948 | if (width > 0) |
1949 | --width; | |
95e83974 | 1950 | char_buffer_add (&charbuf, c); |
9b26f5c4 UD |
1951 | /* It is "nan". */ |
1952 | goto scan_float; | |
1953 | } | |
d64b6ad0 | 1954 | else if (TOLOWER (c) == L_('i')) |
9b26f5c4 UD |
1955 | { |
1956 | /* Maybe "inf" or "infinity". */ | |
95e83974 | 1957 | char_buffer_add (&charbuf, c); |
694b49ca UD |
1958 | if (__builtin_expect (width == 0 |
1959 | || inchar () == EOF | |
1960 | || TOLOWER (c) != L_('n'), 0)) | |
f4efd068 | 1961 | conv_error (); |
9df54168 UD |
1962 | if (width > 0) |
1963 | --width; | |
95e83974 | 1964 | char_buffer_add (&charbuf, c); |
694b49ca UD |
1965 | if (__builtin_expect (width == 0 |
1966 | || inchar () == EOF | |
1967 | || TOLOWER (c) != L_('f'), 0)) | |
f4efd068 | 1968 | conv_error (); |
9df54168 UD |
1969 | if (width > 0) |
1970 | --width; | |
95e83974 | 1971 | char_buffer_add (&charbuf, c); |
9b26f5c4 | 1972 | /* It is as least "inf". */ |
9df54168 | 1973 | if (width != 0 && inchar () != EOF) |
9b26f5c4 | 1974 | { |
d64b6ad0 | 1975 | if (TOLOWER (c) == L_('i')) |
9b26f5c4 | 1976 | { |
9df54168 UD |
1977 | if (width > 0) |
1978 | --width; | |
8d8c6efa | 1979 | /* Now we have to read the rest as well. */ |
95e83974 | 1980 | char_buffer_add (&charbuf, c); |
694b49ca UD |
1981 | if (__builtin_expect (width == 0 |
1982 | || inchar () == EOF | |
1983 | || TOLOWER (c) != L_('n'), 0)) | |
f4efd068 | 1984 | conv_error (); |
9df54168 UD |
1985 | if (width > 0) |
1986 | --width; | |
95e83974 | 1987 | char_buffer_add (&charbuf, c); |
694b49ca UD |
1988 | if (__builtin_expect (width == 0 |
1989 | || inchar () == EOF | |
1990 | || TOLOWER (c) != L_('i'), 0)) | |
f4efd068 | 1991 | conv_error (); |
9df54168 UD |
1992 | if (width > 0) |
1993 | --width; | |
95e83974 | 1994 | char_buffer_add (&charbuf, c); |
694b49ca UD |
1995 | if (__builtin_expect (width == 0 |
1996 | || inchar () == EOF | |
1997 | || TOLOWER (c) != L_('t'), 0)) | |
f4efd068 | 1998 | conv_error (); |
9df54168 UD |
1999 | if (width > 0) |
2000 | --width; | |
95e83974 | 2001 | char_buffer_add (&charbuf, c); |
694b49ca UD |
2002 | if (__builtin_expect (width == 0 |
2003 | || inchar () == EOF | |
2004 | || TOLOWER (c) != L_('y'), 0)) | |
f4efd068 | 2005 | conv_error (); |
9df54168 UD |
2006 | if (width > 0) |
2007 | --width; | |
95e83974 | 2008 | char_buffer_add (&charbuf, c); |
9b26f5c4 UD |
2009 | } |
2010 | else | |
2011 | /* Never mind. */ | |
655c0697 | 2012 | ungetc (c, s); |
9b26f5c4 UD |
2013 | } |
2014 | goto scan_float; | |
2015 | } | |
2016 | ||
d64b6ad0 | 2017 | exp_char = L_('e'); |
9df54168 | 2018 | if (width != 0 && c == L_('0')) |
377a515b | 2019 | { |
95e83974 | 2020 | char_buffer_add (&charbuf, c); |
63551311 | 2021 | c = inchar (); |
9df54168 UD |
2022 | if (width > 0) |
2023 | --width; | |
2024 | if (width != 0 && TOLOWER (c) == L_('x')) | |
63551311 UD |
2025 | { |
2026 | /* It is a number in hexadecimal format. */ | |
95e83974 | 2027 | char_buffer_add (&charbuf, c); |
377a515b | 2028 | |
3979024a | 2029 | flags |= HEXA_FLOAT; |
d64b6ad0 | 2030 | exp_char = L_('p'); |
377a515b | 2031 | |
63551311 UD |
2032 | /* Grouping is not allowed. */ |
2033 | flags &= ~GROUP; | |
2034 | c = inchar (); | |
9df54168 UD |
2035 | if (width > 0) |
2036 | --width; | |
63551311 | 2037 | } |
a4966c61 AS |
2038 | else |
2039 | got_digit = 1; | |
377a515b UD |
2040 | } |
2041 | ||
ac2ca022 | 2042 | while (1) |
28f540f4 | 2043 | { |
95e83974 FW |
2044 | if (char_buffer_error (&charbuf)) |
2045 | { | |
2046 | __set_errno (ENOMEM); | |
2047 | done = EOF; | |
2048 | goto errout; | |
2049 | } | |
d64b6ad0 | 2050 | if (ISDIGIT (c)) |
6ecec3b6 | 2051 | { |
95e83974 | 2052 | char_buffer_add (&charbuf, c); |
6ecec3b6 AS |
2053 | got_digit = 1; |
2054 | } | |
3979024a | 2055 | else if (!got_e && (flags & HEXA_FLOAT) && ISXDIGIT (c)) |
6ecec3b6 | 2056 | { |
95e83974 | 2057 | char_buffer_add (&charbuf, c); |
6ecec3b6 AS |
2058 | got_digit = 1; |
2059 | } | |
95e83974 | 2060 | else if (got_e && charbuf.current[-1] == exp_char |
d64b6ad0 | 2061 | && (c == L_('-') || c == L_('+'))) |
95e83974 | 2062 | char_buffer_add (&charbuf, c); |
6ecec3b6 | 2063 | else if (got_digit && !got_e |
52a7f7c0 | 2064 | && (CHAR_T) TOLOWER (c) == exp_char) |
28f540f4 | 2065 | { |
95e83974 | 2066 | char_buffer_add (&charbuf, exp_char); |
28f540f4 RM |
2067 | got_e = got_dot = 1; |
2068 | } | |
28f540f4 | 2069 | else |
377a515b | 2070 | { |
72acaddf UD |
2071 | #ifdef COMPILE_WSCANF |
2072 | if (! got_dot && c == decimal) | |
2073 | { | |
95e83974 | 2074 | char_buffer_add (&charbuf, c); |
72acaddf UD |
2075 | got_dot = 1; |
2076 | } | |
3979024a | 2077 | else if ((flags & GROUP) != 0 && ! got_dot && c == thousands) |
95e83974 | 2078 | char_buffer_add (&charbuf, c); |
72acaddf UD |
2079 | else |
2080 | { | |
2081 | /* The last read character is not part of the number | |
2082 | anymore. */ | |
2083 | ungetc (c, s); | |
2084 | break; | |
2085 | } | |
2086 | #else | |
2087 | const char *cmpp = decimal; | |
2088 | int avail = width > 0 ? width : INT_MAX; | |
2089 | ||
2090 | if (! got_dot) | |
2091 | { | |
ecb72bd5 | 2092 | while ((unsigned char) *cmpp == c && avail >= 0) |
72acaddf UD |
2093 | if (*++cmpp == '\0') |
2094 | break; | |
2095 | else | |
2096 | { | |
ecb72bd5 | 2097 | if (avail == 0 || inchar () == EOF) |
72acaddf UD |
2098 | break; |
2099 | --avail; | |
2100 | } | |
2101 | } | |
2102 | ||
2103 | if (*cmpp == '\0') | |
2104 | { | |
2105 | /* Add all the characters. */ | |
2106 | for (cmpp = decimal; *cmpp != '\0'; ++cmpp) | |
95e83974 | 2107 | char_buffer_add (&charbuf, (unsigned char) *cmpp); |
72acaddf | 2108 | if (width > 0) |
eac4282f | 2109 | width = avail; |
72acaddf UD |
2110 | got_dot = 1; |
2111 | } | |
2112 | else | |
2113 | { | |
2114 | /* Figure out whether it is a thousands separator. | |
2115 | There is one problem: we possibly read more than | |
2116 | one character. We cannot push them back but since | |
2117 | we know that parts of the `decimal' string matched, | |
2118 | we can compare against it. */ | |
2119 | const char *cmp2p = thousands; | |
2120 | ||
3979024a | 2121 | if ((flags & GROUP) != 0 && ! got_dot) |
72acaddf | 2122 | { |
eb35b097 | 2123 | while (cmp2p - thousands < cmpp - decimal |
72acaddf UD |
2124 | && *cmp2p == decimal[cmp2p - thousands]) |
2125 | ++cmp2p; | |
eb35b097 | 2126 | if (cmp2p - thousands == cmpp - decimal) |
72acaddf | 2127 | { |
ecb72bd5 | 2128 | while ((unsigned char) *cmp2p == c && avail >= 0) |
72acaddf UD |
2129 | if (*++cmp2p == '\0') |
2130 | break; | |
2131 | else | |
2132 | { | |
ecb72bd5 | 2133 | if (avail == 0 || inchar () == EOF) |
72acaddf UD |
2134 | break; |
2135 | --avail; | |
2136 | } | |
2137 | } | |
2138 | } | |
2139 | ||
2140 | if (cmp2p != NULL && *cmp2p == '\0') | |
2141 | { | |
2142 | /* Add all the characters. */ | |
2143 | for (cmpp = thousands; *cmpp != '\0'; ++cmpp) | |
95e83974 | 2144 | char_buffer_add (&charbuf, (unsigned char) *cmpp); |
72acaddf | 2145 | if (width > 0) |
eac4282f | 2146 | width = avail; |
72acaddf UD |
2147 | } |
2148 | else | |
2149 | { | |
2150 | /* The last read character is not part of the number | |
2151 | anymore. */ | |
2152 | ungetc (c, s); | |
2153 | break; | |
2154 | } | |
2155 | } | |
2156 | #endif | |
377a515b | 2157 | } |
ac2ca022 UD |
2158 | |
2159 | if (width == 0 || inchar () == EOF) | |
2160 | break; | |
2161 | ||
28f540f4 RM |
2162 | if (width > 0) |
2163 | --width; | |
01cdeca0 | 2164 | } |
3867ee64 | 2165 | |
95e83974 FW |
2166 | if (char_buffer_error (&charbuf)) |
2167 | { | |
2168 | __set_errno (ENOMEM); | |
2169 | done = EOF; | |
2170 | goto errout; | |
2171 | } | |
2172 | ||
ecb72bd5 UD |
2173 | wctrans_t map; |
2174 | if (__builtin_expect ((flags & I18N) != 0, 0) | |
2175 | /* Hexadecimal floats make no sense, fixing localized | |
2176 | digits with ASCII letters. */ | |
3979024a | 2177 | && !(flags & HEXA_FLOAT) |
ecb72bd5 | 2178 | /* Minimum requirement. */ |
95e83974 | 2179 | && (char_buffer_size (&charbuf) == 0 || got_dot) |
ecb72bd5 UD |
2180 | && (map = __wctrans ("to_inpunct")) != NULL) |
2181 | { | |
2182 | /* Reget the first character. */ | |
2183 | inchar (); | |
2184 | ||
2185 | /* Localized digits, decimal points, and thousands | |
2186 | separator. */ | |
2187 | wint_t wcdigits[12]; | |
2188 | ||
2189 | /* First get decimal equivalent to check if we read it | |
2190 | or not. */ | |
2191 | wcdigits[11] = __towctrans (L'.', map); | |
2192 | ||
2193 | /* If we have not read any character or have just read | |
f095bb72 UD |
2194 | locale decimal point which matches the decimal point |
2195 | for localized FP numbers, then we may have localized | |
2196 | digits. Note, we test GOT_DOT above. */ | |
ecb72bd5 | 2197 | #ifdef COMPILE_WSCANF |
95e83974 FW |
2198 | if (char_buffer_size (&charbuf) == 0 |
2199 | || (char_buffer_size (&charbuf) == 1 | |
2200 | && wcdigits[11] == decimal)) | |
ecb72bd5 UD |
2201 | #else |
2202 | char mbdigits[12][MB_LEN_MAX + 1]; | |
2203 | ||
2204 | mbstate_t state; | |
2205 | memset (&state, '\0', sizeof (state)); | |
2206 | ||
95e83974 | 2207 | bool match_so_far = char_buffer_size (&charbuf) == 0; |
ecb72bd5 UD |
2208 | size_t mblen = __wcrtomb (mbdigits[11], wcdigits[11], &state); |
2209 | if (mblen != (size_t) -1) | |
2210 | { | |
2211 | mbdigits[11][mblen] = '\0'; | |
95e83974 FW |
2212 | match_so_far |= |
2213 | (char_buffer_size (&charbuf) == strlen (decimal) | |
2214 | && strcmp (decimal, mbdigits[11]) == 0); | |
ecb72bd5 UD |
2215 | } |
2216 | else | |
2217 | { | |
2218 | size_t decimal_len = strlen (decimal); | |
2219 | /* This should always be the case but the data comes | |
2220 | from a file. */ | |
2221 | if (decimal_len <= MB_LEN_MAX) | |
2222 | { | |
95e83974 | 2223 | match_so_far |= char_buffer_size (&charbuf) == decimal_len; |
ecb72bd5 UD |
2224 | memcpy (mbdigits[11], decimal, decimal_len + 1); |
2225 | } | |
2226 | else | |
2227 | match_so_far = false; | |
2228 | } | |
2229 | ||
2230 | if (match_so_far) | |
2231 | #endif | |
2232 | { | |
3979024a UD |
2233 | bool have_locthousands = (flags & GROUP) != 0; |
2234 | ||
ecb72bd5 | 2235 | /* Now get the digits and the thousands-sep equivalents. */ |
f095bb72 | 2236 | for (int n = 0; n < 11; ++n) |
ecb72bd5 UD |
2237 | { |
2238 | if (n < 10) | |
2239 | wcdigits[n] = __towctrans (L'0' + n, map); | |
2240 | else if (n == 10) | |
3979024a UD |
2241 | { |
2242 | wcdigits[10] = __towctrans (L',', map); | |
2243 | have_locthousands &= wcdigits[10] != L'\0'; | |
2244 | } | |
ecb72bd5 UD |
2245 | |
2246 | #ifndef COMPILE_WSCANF | |
2247 | memset (&state, '\0', sizeof (state)); | |
2248 | ||
2249 | size_t mblen = __wcrtomb (mbdigits[n], wcdigits[n], | |
2250 | &state); | |
2251 | if (mblen == (size_t) -1) | |
2252 | { | |
2253 | if (n == 10) | |
2254 | { | |
3979024a | 2255 | if (have_locthousands) |
ecb72bd5 UD |
2256 | { |
2257 | size_t thousands_len = strlen (thousands); | |
2258 | if (thousands_len <= MB_LEN_MAX) | |
2259 | memcpy (mbdigits[10], thousands, | |
2260 | thousands_len + 1); | |
2261 | else | |
2262 | have_locthousands = false; | |
2263 | } | |
2264 | } | |
2265 | else | |
2266 | /* Ignore checking against localized digits. */ | |
2267 | goto no_i18nflt; | |
2268 | } | |
2269 | else | |
2270 | mbdigits[n][mblen] = '\0'; | |
2271 | #endif | |
2272 | } | |
2273 | ||
2274 | /* Start checking against localized digits, if | |
382466e0 | 2275 | conversion is done correctly. */ |
ecb72bd5 UD |
2276 | while (1) |
2277 | { | |
95e83974 FW |
2278 | if (char_buffer_error (&charbuf)) |
2279 | { | |
2280 | __set_errno (ENOMEM); | |
2281 | done = EOF; | |
2282 | goto errout; | |
2283 | } | |
2284 | if (got_e && charbuf.current[-1] == exp_char | |
ecb72bd5 | 2285 | && (c == L_('-') || c == L_('+'))) |
95e83974 FW |
2286 | char_buffer_add (&charbuf, c); |
2287 | else if (char_buffer_size (&charbuf) > 0 && !got_e | |
ecb72bd5 UD |
2288 | && (CHAR_T) TOLOWER (c) == exp_char) |
2289 | { | |
95e83974 | 2290 | char_buffer_add (&charbuf, exp_char); |
ecb72bd5 UD |
2291 | got_e = got_dot = 1; |
2292 | } | |
2293 | else | |
2294 | { | |
2295 | /* Check against localized digits, decimal point, | |
2296 | and thousands separator. */ | |
2297 | int n; | |
2298 | for (n = 0; n < 12; ++n) | |
2299 | { | |
2300 | #ifdef COMPILE_WSCANF | |
2301 | if (c == wcdigits[n]) | |
2302 | { | |
2303 | if (n < 10) | |
95e83974 | 2304 | char_buffer_add (&charbuf, L_('0') + n); |
ecb72bd5 UD |
2305 | else if (n == 11 && !got_dot) |
2306 | { | |
95e83974 | 2307 | char_buffer_add (&charbuf, decimal); |
ecb72bd5 UD |
2308 | got_dot = 1; |
2309 | } | |
2310 | else if (n == 10 && have_locthousands | |
2311 | && ! got_dot) | |
95e83974 | 2312 | char_buffer_add (&charbuf, thousands); |
ecb72bd5 UD |
2313 | else |
2314 | /* The last read character is not part | |
2315 | of the number anymore. */ | |
2316 | n = 12; | |
2317 | ||
2318 | break; | |
2319 | } | |
2320 | #else | |
2321 | const char *cmpp = mbdigits[n]; | |
2322 | int avail = width > 0 ? width : INT_MAX; | |
2323 | ||
2324 | while ((unsigned char) *cmpp == c && avail >= 0) | |
2325 | if (*++cmpp == '\0') | |
2326 | break; | |
2327 | else | |
2328 | { | |
2329 | if (avail == 0 || inchar () == EOF) | |
2330 | break; | |
2331 | --avail; | |
2332 | } | |
2333 | if (*cmpp == '\0') | |
2334 | { | |
2335 | if (width > 0) | |
2336 | width = avail; | |
2337 | ||
2338 | if (n < 10) | |
95e83974 | 2339 | char_buffer_add (&charbuf, L_('0') + n); |
ecb72bd5 UD |
2340 | else if (n == 11 && !got_dot) |
2341 | { | |
2342 | /* Add all the characters. */ | |
2343 | for (cmpp = decimal; *cmpp != '\0'; | |
2344 | ++cmpp) | |
95e83974 FW |
2345 | char_buffer_add (&charbuf, |
2346 | (unsigned char) *cmpp); | |
ecb72bd5 UD |
2347 | |
2348 | got_dot = 1; | |
2349 | } | |
2350 | else if (n == 10 && (flags & GROUP) != 0 | |
4c02bf1a | 2351 | && ! got_dot) |
ecb72bd5 UD |
2352 | { |
2353 | /* Add all the characters. */ | |
2354 | for (cmpp = thousands; *cmpp != '\0'; | |
2355 | ++cmpp) | |
95e83974 FW |
2356 | char_buffer_add (&charbuf, |
2357 | (unsigned char) *cmpp); | |
ecb72bd5 UD |
2358 | } |
2359 | else | |
2360 | /* The last read character is not part | |
2361 | of the number anymore. */ | |
2362 | n = 12; | |
2363 | ||
2364 | break; | |
2365 | } | |
2366 | ||
2367 | /* We are pushing all read characters back. */ | |
2368 | if (cmpp > mbdigits[n]) | |
2369 | { | |
2370 | ungetc (c, s); | |
2371 | while (--cmpp > mbdigits[n]) | |
2372 | ungetc_not_eof ((unsigned char) *cmpp, s); | |
2373 | c = (unsigned char) *cmpp; | |
2374 | } | |
2375 | #endif | |
2376 | } | |
2377 | ||
2378 | if (n >= 12) | |
2379 | { | |
2380 | /* The last read character is not part | |
2381 | of the number anymore. */ | |
2382 | ungetc (c, s); | |
2383 | break; | |
2384 | } | |
2385 | } | |
2386 | ||
2387 | if (width == 0 || inchar () == EOF) | |
2388 | break; | |
2389 | ||
2390 | if (width > 0) | |
2391 | --width; | |
2392 | } | |
2393 | } | |
2394 | ||
2395 | #ifndef COMPILE_WSCANF | |
2396 | no_i18nflt: | |
2397 | ; | |
2398 | #endif | |
2399 | } | |
2400 | ||
95e83974 FW |
2401 | if (char_buffer_error (&charbuf)) |
2402 | { | |
2403 | __set_errno (ENOMEM); | |
2404 | done = EOF; | |
2405 | goto errout; | |
2406 | } | |
2407 | ||
377a515b UD |
2408 | /* Have we read any character? If we try to read a number |
2409 | in hexadecimal notation and we have read only the `0x' | |
bd9be679 | 2410 | prefix this is an error. */ |
95e83974 FW |
2411 | if (__glibc_unlikely (char_buffer_size (&charbuf) == 0 |
2412 | || ((flags & HEXA_FLOAT) | |
2413 | && char_buffer_size (&charbuf) == 2))) | |
96aa2d94 | 2414 | conv_error (); |
28f540f4 | 2415 | |
9b26f5c4 | 2416 | scan_float: |
28f540f4 | 2417 | /* Convert the number. */ |
95e83974 FW |
2418 | char_buffer_add (&charbuf, L_('\0')); |
2419 | if (char_buffer_error (&charbuf)) | |
2420 | { | |
2421 | __set_errno (ENOMEM); | |
2422 | done = EOF; | |
2423 | goto errout; | |
2424 | } | |
c6251f03 | 2425 | if ((flags & LONGDBL) && !__ldbl_is_dbl) |
28f540f4 | 2426 | { |
95e83974 FW |
2427 | long double d = __strtold_internal |
2428 | (char_buffer_start (&charbuf), &tw, flags & GROUP); | |
2429 | if (!(flags & SUPPRESS) && tw != char_buffer_start (&charbuf)) | |
01cdeca0 | 2430 | *ARG (long double *) = negative ? -d : d; |
28f540f4 | 2431 | } |
c6251f03 | 2432 | else if (flags & (LONG | LONGDBL)) |
28f540f4 | 2433 | { |
95e83974 FW |
2434 | double d = __strtod_internal |
2435 | (char_buffer_start (&charbuf), &tw, flags & GROUP); | |
2436 | if (!(flags & SUPPRESS) && tw != char_buffer_start (&charbuf)) | |
01cdeca0 | 2437 | *ARG (double *) = negative ? -d : d; |
28f540f4 RM |
2438 | } |
2439 | else | |
2440 | { | |
95e83974 FW |
2441 | float d = __strtof_internal |
2442 | (char_buffer_start (&charbuf), &tw, flags & GROUP); | |
2443 | if (!(flags & SUPPRESS) && tw != char_buffer_start (&charbuf)) | |
01cdeca0 | 2444 | *ARG (float *) = negative ? -d : d; |
28f540f4 RM |
2445 | } |
2446 | ||
95e83974 | 2447 | if (__glibc_unlikely (tw == char_buffer_start (&charbuf))) |
28f540f4 RM |
2448 | conv_error (); |
2449 | ||
0793d348 | 2450 | if (!(flags & SUPPRESS)) |
28f540f4 RM |
2451 | ++done; |
2452 | break; | |
2453 | ||
d64b6ad0 | 2454 | case L_('['): /* Character class. */ |
2c6fe0bd | 2455 | if (flags & LONG) |
874aa523 | 2456 | STRING_ARG (wstr, wchar_t, 100); |
2c6fe0bd | 2457 | else |
874aa523 | 2458 | STRING_ARG (str, char, 100); |
28f540f4 | 2459 | |
d64b6ad0 | 2460 | if (*f == L_('^')) |
28f540f4 RM |
2461 | { |
2462 | ++f; | |
2463 | not_in = 1; | |
2464 | } | |
2465 | else | |
2466 | not_in = 0; | |
2467 | ||
d64b6ad0 UD |
2468 | if (width < 0) |
2469 | /* There is no width given so there is also no limit on the | |
2470 | number of characters we read. Therefore we set width to | |
2471 | a very high value to make the algorithm easier. */ | |
2472 | width = INT_MAX; | |
2473 | ||
72acaddf | 2474 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
2475 | /* Find the beginning and the end of the scanlist. We are not |
2476 | creating a lookup table since it would have to be too large. | |
2477 | Instead we search each time through the string. This is not | |
2478 | a constant lookup time but who uses this feature deserves to | |
2479 | be punished. */ | |
2480 | tw = (wchar_t *) f; /* Marks the beginning. */ | |
2481 | ||
cc7f258f | 2482 | if (*f == L']') |
d64b6ad0 UD |
2483 | ++f; |
2484 | ||
2485 | while ((fc = *f++) != L'\0' && fc != L']'); | |
2486 | ||
a1ffb40e | 2487 | if (__glibc_unlikely (fc == L'\0')) |
d64b6ad0 | 2488 | conv_error (); |
4c6b2202 | 2489 | wchar_t *twend = (wchar_t *) f - 1; |
d64b6ad0 | 2490 | #else |
b17178fc RM |
2491 | /* Fill WP with byte flags indexed by character. |
2492 | We will use this flag map for matching input characters. */ | |
95e83974 FW |
2493 | if (!scratch_buffer_set_array_size |
2494 | (&charbuf.scratch, UCHAR_MAX + 1, 1)) | |
b17178fc | 2495 | { |
95e83974 FW |
2496 | done = EOF; |
2497 | goto errout; | |
b17178fc | 2498 | } |
95e83974 | 2499 | memset (charbuf.scratch.data, '\0', UCHAR_MAX + 1); |
b17178fc RM |
2500 | |
2501 | fc = *f; | |
2502 | if (fc == ']' || fc == '-') | |
2503 | { | |
2504 | /* If ] or - appears before any char in the set, it is not | |
2505 | the terminator or separator, but the first char in the | |
2506 | set. */ | |
95e83974 | 2507 | ((char *)charbuf.scratch.data)[fc] = 1; |
b17178fc RM |
2508 | ++f; |
2509 | } | |
02ac66c5 | 2510 | |
28f540f4 | 2511 | while ((fc = *f++) != '\0' && fc != ']') |
3f651a4d | 2512 | if (fc == '-' && *f != '\0' && *f != ']' |
d64b6ad0 UD |
2513 | && (unsigned char) f[-2] <= (unsigned char) *f) |
2514 | { | |
2515 | /* Add all characters from the one before the '-' | |
2516 | up to (but not including) the next format char. */ | |
44f8759b | 2517 | for (fc = (unsigned char) f[-2]; fc < (unsigned char) *f; ++fc) |
95e83974 | 2518 | ((char *)charbuf.scratch.data)[fc] = 1; |
d64b6ad0 UD |
2519 | } |
2520 | else | |
2521 | /* Add the character to the flag map. */ | |
95e83974 | 2522 | ((char *)charbuf.scratch.data)[fc] = 1; |
d64b6ad0 | 2523 | |
a1ffb40e | 2524 | if (__glibc_unlikely (fc == '\0')) |
d64b6ad0 UD |
2525 | conv_error(); |
2526 | #endif | |
28f540f4 | 2527 | |
2c6fe0bd | 2528 | if (flags & LONG) |
28f540f4 | 2529 | { |
d64b6ad0 | 2530 | size_t now = read_in; |
72acaddf | 2531 | #ifdef COMPILE_WSCANF |
a1ffb40e | 2532 | if (__glibc_unlikely (inchar () == WEOF)) |
d64e603a UD |
2533 | input_error (); |
2534 | ||
d64b6ad0 UD |
2535 | do |
2536 | { | |
2537 | wchar_t *runp; | |
2538 | ||
d64b6ad0 UD |
2539 | /* Test whether it's in the scanlist. */ |
2540 | runp = tw; | |
4c6b2202 | 2541 | while (runp < twend) |
d64b6ad0 | 2542 | { |
4c6b2202 UD |
2543 | if (runp[0] == L'-' && runp[1] != '\0' |
2544 | && runp + 1 != twend | |
d64b6ad0 UD |
2545 | && runp != tw |
2546 | && (unsigned int) runp[-1] <= (unsigned int) runp[1]) | |
2547 | { | |
2548 | /* Match against all characters in between the | |
2549 | first and last character of the sequence. */ | |
2550 | wchar_t wc; | |
2551 | ||
cc7f258f | 2552 | for (wc = runp[-1] + 1; wc <= runp[1]; ++wc) |
52a7f7c0 | 2553 | if ((wint_t) wc == c) |
d64b6ad0 UD |
2554 | break; |
2555 | ||
cc7f258f | 2556 | if (wc <= runp[1] && !not_in) |
d64b6ad0 | 2557 | break; |
cc7f258f | 2558 | if (wc <= runp[1] && not_in) |
d64b6ad0 UD |
2559 | { |
2560 | /* The current character is not in the | |
f095bb72 | 2561 | scanset. */ |
73f1b067 | 2562 | ungetc (c, s); |
d64b6ad0 UD |
2563 | goto out; |
2564 | } | |
cc7f258f UD |
2565 | |
2566 | runp += 2; | |
d64b6ad0 UD |
2567 | } |
2568 | else | |
2569 | { | |
52a7f7c0 | 2570 | if ((wint_t) *runp == c && !not_in) |
d64b6ad0 | 2571 | break; |
52a7f7c0 | 2572 | if ((wint_t) *runp == c && not_in) |
d64b6ad0 | 2573 | { |
73f1b067 | 2574 | ungetc (c, s); |
d64b6ad0 UD |
2575 | goto out; |
2576 | } | |
cc7f258f UD |
2577 | |
2578 | ++runp; | |
d64b6ad0 | 2579 | } |
cc7f258f | 2580 | } |
d64b6ad0 | 2581 | |
4c6b2202 | 2582 | if (runp == twend && !not_in) |
cc7f258f | 2583 | { |
73f1b067 | 2584 | ungetc (c, s); |
cc7f258f | 2585 | goto out; |
d64b6ad0 UD |
2586 | } |
2587 | ||
2588 | if (!(flags & SUPPRESS)) | |
2589 | { | |
2590 | *wstr++ = c; | |
2591 | ||
2592 | if ((flags & MALLOC) | |
2593 | && wstr == (wchar_t *) *strptr + strsize) | |
2594 | { | |
2595 | /* Enlarge the buffer. */ | |
2596 | wstr = (wchar_t *) realloc (*strptr, | |
2597 | (2 * strsize) | |
2598 | * sizeof (wchar_t)); | |
2599 | if (wstr == NULL) | |
2600 | { | |
2601 | /* Can't allocate that much. Last-ditch | |
2602 | effort. */ | |
2603 | wstr = (wchar_t *) | |
d90e1b42 UD |
2604 | realloc (*strptr, (strsize + 1) |
2605 | * sizeof (wchar_t)); | |
d64b6ad0 UD |
2606 | if (wstr == NULL) |
2607 | { | |
267c54dc | 2608 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
2609 | { |
2610 | done = EOF; | |
2611 | goto errout; | |
2612 | } | |
d64b6ad0 UD |
2613 | /* We lose. Oh well. Terminate the string |
2614 | and stop converting, so at least we don't | |
2615 | skip any input. */ | |
2616 | ((wchar_t *) (*strptr))[strsize - 1] = L'\0'; | |
267c54dc | 2617 | strptr = NULL; |
d64b6ad0 UD |
2618 | ++done; |
2619 | conv_error (); | |
2620 | } | |
2621 | else | |
2622 | { | |
2623 | *strptr = (char *) wstr; | |
2624 | wstr += strsize; | |
2625 | ++strsize; | |
2626 | } | |
2627 | } | |
2628 | else | |
2629 | { | |
2630 | *strptr = (char *) wstr; | |
2631 | wstr += strsize; | |
2632 | strsize *= 2; | |
2633 | } | |
2634 | } | |
2635 | } | |
2636 | } | |
d64e603a | 2637 | while (--width > 0 && inchar () != WEOF); |
d64b6ad0 UD |
2638 | out: |
2639 | #else | |
2640 | char buf[MB_LEN_MAX]; | |
2641 | size_t cnt = 0; | |
2642 | mbstate_t cstate; | |
2643 | ||
a1ffb40e | 2644 | if (__glibc_unlikely (inchar () == EOF)) |
d64e603a UD |
2645 | input_error (); |
2646 | ||
d64b6ad0 | 2647 | memset (&cstate, '\0', sizeof (cstate)); |
2c6fe0bd UD |
2648 | |
2649 | do | |
3867ee64 | 2650 | { |
95e83974 | 2651 | if (((char *) charbuf.scratch.data)[c] == not_in) |
2c6fe0bd | 2652 | { |
44f8759b | 2653 | ungetc_not_eof (c, s); |
2c6fe0bd UD |
2654 | break; |
2655 | } | |
d64b6ad0 UD |
2656 | |
2657 | /* This is easy. */ | |
2658 | if (!(flags & SUPPRESS)) | |
2659 | { | |
2660 | size_t n; | |
2661 | ||
2662 | /* Convert it into a wide character. */ | |
bed38142 UD |
2663 | buf[0] = c; |
2664 | n = __mbrtowc (wstr, buf, 1, &cstate); | |
d64b6ad0 UD |
2665 | |
2666 | if (n == (size_t) -2) | |
2667 | { | |
2668 | /* Possibly correct character, just not enough | |
2669 | input. */ | |
bed38142 | 2670 | ++cnt; |
9dd87afb | 2671 | assert (cnt < MB_LEN_MAX); |
d64e603a | 2672 | continue; |
d64b6ad0 | 2673 | } |
0f740059 | 2674 | cnt = 0; |
d64b6ad0 | 2675 | |
d64b6ad0 UD |
2676 | ++wstr; |
2677 | if ((flags & MALLOC) | |
2678 | && wstr == (wchar_t *) *strptr + strsize) | |
2679 | { | |
2680 | /* Enlarge the buffer. */ | |
2681 | wstr = (wchar_t *) realloc (*strptr, | |
2682 | (2 * strsize | |
2683 | * sizeof (wchar_t))); | |
2684 | if (wstr == NULL) | |
2685 | { | |
2686 | /* Can't allocate that much. Last-ditch | |
2687 | effort. */ | |
2688 | wstr = (wchar_t *) | |
2689 | realloc (*strptr, ((strsize + 1) | |
2690 | * sizeof (wchar_t))); | |
2691 | if (wstr == NULL) | |
2692 | { | |
267c54dc | 2693 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
2694 | { |
2695 | done = EOF; | |
2696 | goto errout; | |
2697 | } | |
d64b6ad0 UD |
2698 | /* We lose. Oh well. Terminate the |
2699 | string and stop converting, | |
2700 | so at least we don't skip any input. */ | |
2701 | ((wchar_t *) (*strptr))[strsize - 1] = L'\0'; | |
267c54dc | 2702 | strptr = NULL; |
d64b6ad0 UD |
2703 | ++done; |
2704 | conv_error (); | |
2705 | } | |
2706 | else | |
2707 | { | |
2708 | *strptr = (char *) wstr; | |
2709 | wstr += strsize; | |
2710 | ++strsize; | |
2711 | } | |
2712 | } | |
2713 | else | |
2714 | { | |
2715 | *strptr = (char *) wstr; | |
2716 | wstr += strsize; | |
2717 | strsize *= 2; | |
2718 | } | |
2719 | } | |
2720 | } | |
d64e603a UD |
2721 | |
2722 | if (--width <= 0) | |
2723 | break; | |
3867ee64 | 2724 | } |
d64e603a | 2725 | while (inchar () != EOF); |
d64b6ad0 | 2726 | |
a1ffb40e | 2727 | if (__glibc_unlikely (cnt != 0)) |
d64b6ad0 UD |
2728 | /* We stopped in the middle of recognizing another |
2729 | character. That's a problem. */ | |
2730 | encode_error (); | |
2731 | #endif | |
3867ee64 | 2732 | |
a1ffb40e | 2733 | if (__glibc_unlikely (now == read_in)) |
d64b6ad0 | 2734 | /* We haven't succesfully read any character. */ |
2c6fe0bd | 2735 | conv_error (); |
28f540f4 | 2736 | |
2c6fe0bd UD |
2737 | if (!(flags & SUPPRESS)) |
2738 | { | |
d64b6ad0 UD |
2739 | *wstr++ = L'\0'; |
2740 | ||
2741 | if ((flags & MALLOC) | |
2742 | && wstr - (wchar_t *) *strptr != strsize) | |
2743 | { | |
2744 | wchar_t *cp = (wchar_t *) | |
2745 | realloc (*strptr, ((wstr - (wchar_t *) *strptr) | |
2746 | * sizeof(wchar_t))); | |
2747 | if (cp != NULL) | |
2748 | *strptr = (char *) cp; | |
2749 | } | |
267c54dc | 2750 | strptr = NULL; |
d64b6ad0 | 2751 | |
2c6fe0bd UD |
2752 | ++done; |
2753 | } | |
2754 | } | |
2755 | else | |
28f540f4 | 2756 | { |
d64b6ad0 | 2757 | size_t now = read_in; |
d64e603a | 2758 | |
a1ffb40e | 2759 | if (__glibc_unlikely (inchar () == EOF)) |
d64e603a UD |
2760 | input_error (); |
2761 | ||
72acaddf | 2762 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
2763 | |
2764 | memset (&state, '\0', sizeof (state)); | |
2765 | ||
2766 | do | |
2767 | { | |
2768 | wchar_t *runp; | |
2769 | size_t n; | |
2770 | ||
d64b6ad0 UD |
2771 | /* Test whether it's in the scanlist. */ |
2772 | runp = tw; | |
4c6b2202 | 2773 | while (runp < twend) |
d64b6ad0 | 2774 | { |
4c6b2202 UD |
2775 | if (runp[0] == L'-' && runp[1] != '\0' |
2776 | && runp + 1 != twend | |
d64b6ad0 UD |
2777 | && runp != tw |
2778 | && (unsigned int) runp[-1] <= (unsigned int) runp[1]) | |
2779 | { | |
2780 | /* Match against all characters in between the | |
2781 | first and last character of the sequence. */ | |
2782 | wchar_t wc; | |
2783 | ||
cc7f258f | 2784 | for (wc = runp[-1] + 1; wc <= runp[1]; ++wc) |
52a7f7c0 | 2785 | if ((wint_t) wc == c) |
d64b6ad0 UD |
2786 | break; |
2787 | ||
cc7f258f | 2788 | if (wc <= runp[1] && !not_in) |
d64b6ad0 | 2789 | break; |
cc7f258f | 2790 | if (wc <= runp[1] && not_in) |
d64b6ad0 UD |
2791 | { |
2792 | /* The current character is not in the | |
f095bb72 | 2793 | scanset. */ |
73f1b067 | 2794 | ungetc (c, s); |
d64b6ad0 UD |
2795 | goto out2; |
2796 | } | |
cc7f258f UD |
2797 | |
2798 | runp += 2; | |
d64b6ad0 UD |
2799 | } |
2800 | else | |
2801 | { | |
52a7f7c0 | 2802 | if ((wint_t) *runp == c && !not_in) |
d64b6ad0 | 2803 | break; |
52a7f7c0 | 2804 | if ((wint_t) *runp == c && not_in) |
d64b6ad0 | 2805 | { |
73f1b067 | 2806 | ungetc (c, s); |
d64b6ad0 UD |
2807 | goto out2; |
2808 | } | |
cc7f258f UD |
2809 | |
2810 | ++runp; | |
d64b6ad0 | 2811 | } |
cc7f258f | 2812 | } |
d64b6ad0 | 2813 | |
4c6b2202 | 2814 | if (runp == twend && !not_in) |
cc7f258f | 2815 | { |
73f1b067 | 2816 | ungetc (c, s); |
cc7f258f | 2817 | goto out2; |
d64b6ad0 UD |
2818 | } |
2819 | ||
2820 | if (!(flags & SUPPRESS)) | |
2821 | { | |
2822 | if ((flags & MALLOC) | |
9dd87afb | 2823 | && *strptr + strsize - str <= MB_LEN_MAX) |
d64b6ad0 UD |
2824 | { |
2825 | /* Enlarge the buffer. */ | |
d90e1b42 UD |
2826 | size_t strleng = str - *strptr; |
2827 | char *newstr; | |
2828 | ||
2829 | newstr = (char *) realloc (*strptr, 2 * strsize); | |
2830 | if (newstr == NULL) | |
d64b6ad0 UD |
2831 | { |
2832 | /* Can't allocate that much. Last-ditch | |
2833 | effort. */ | |
d90e1b42 | 2834 | newstr = (char *) realloc (*strptr, |
9dd87afb | 2835 | strleng + MB_LEN_MAX); |
d90e1b42 | 2836 | if (newstr == NULL) |
d64b6ad0 | 2837 | { |
267c54dc | 2838 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
2839 | { |
2840 | done = EOF; | |
2841 | goto errout; | |
2842 | } | |
d64b6ad0 UD |
2843 | /* We lose. Oh well. Terminate the string |
2844 | and stop converting, so at least we don't | |
2845 | skip any input. */ | |
d90e1b42 | 2846 | ((char *) (*strptr))[strleng] = '\0'; |
267c54dc | 2847 | strptr = NULL; |
d64b6ad0 UD |
2848 | ++done; |
2849 | conv_error (); | |
2850 | } | |
2851 | else | |
2852 | { | |
d90e1b42 UD |
2853 | *strptr = newstr; |
2854 | str = newstr + strleng; | |
9dd87afb | 2855 | strsize = strleng + MB_LEN_MAX; |
d64b6ad0 UD |
2856 | } |
2857 | } | |
2858 | else | |
2859 | { | |
d90e1b42 UD |
2860 | *strptr = newstr; |
2861 | str = newstr + strleng; | |
d64b6ad0 UD |
2862 | strsize *= 2; |
2863 | } | |
2864 | } | |
2865 | } | |
2866 | ||
4aebaa6b | 2867 | n = __wcrtomb (!(flags & SUPPRESS) ? str : NULL, c, &state); |
a1ffb40e | 2868 | if (__glibc_unlikely (n == (size_t) -1)) |
d64b6ad0 UD |
2869 | encode_error (); |
2870 | ||
9dd87afb | 2871 | assert (n <= MB_LEN_MAX); |
d64b6ad0 UD |
2872 | str += n; |
2873 | } | |
d64e603a | 2874 | while (--width > 0 && inchar () != WEOF); |
d64b6ad0 UD |
2875 | out2: |
2876 | #else | |
2c6fe0bd UD |
2877 | do |
2878 | { | |
95e83974 | 2879 | if (((char *) charbuf.scratch.data)[c] == not_in) |
2c6fe0bd | 2880 | { |
44f8759b | 2881 | ungetc_not_eof (c, s); |
2c6fe0bd UD |
2882 | break; |
2883 | } | |
d64b6ad0 UD |
2884 | |
2885 | /* This is easy. */ | |
2886 | if (!(flags & SUPPRESS)) | |
2887 | { | |
2888 | *str++ = c; | |
2889 | if ((flags & MALLOC) | |
2890 | && (char *) str == *strptr + strsize) | |
2891 | { | |
2892 | /* Enlarge the buffer. */ | |
e3b22ad3 UD |
2893 | size_t newsize = 2 * strsize; |
2894 | ||
2895 | allocagain: | |
2896 | str = (char *) realloc (*strptr, newsize); | |
d64b6ad0 UD |
2897 | if (str == NULL) |
2898 | { | |
2899 | /* Can't allocate that much. Last-ditch | |
2900 | effort. */ | |
e3b22ad3 | 2901 | if (newsize > strsize + 1) |
d64b6ad0 | 2902 | { |
e3b22ad3 UD |
2903 | newsize = strsize + 1; |
2904 | goto allocagain; | |
d64b6ad0 | 2905 | } |
267c54dc | 2906 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
2907 | { |
2908 | done = EOF; | |
2909 | goto errout; | |
2910 | } | |
e3b22ad3 UD |
2911 | /* We lose. Oh well. Terminate the |
2912 | string and stop converting, | |
2913 | so at least we don't skip any input. */ | |
2914 | ((char *) (*strptr))[strsize - 1] = '\0'; | |
267c54dc | 2915 | strptr = NULL; |
e3b22ad3 UD |
2916 | ++done; |
2917 | conv_error (); | |
d64b6ad0 UD |
2918 | } |
2919 | else | |
2920 | { | |
2921 | *strptr = (char *) str; | |
2922 | str += strsize; | |
e3b22ad3 | 2923 | strsize = newsize; |
d64b6ad0 UD |
2924 | } |
2925 | } | |
2926 | } | |
2c6fe0bd | 2927 | } |
d64e603a | 2928 | while (--width > 0 && inchar () != EOF); |
d64b6ad0 | 2929 | #endif |
2c6fe0bd | 2930 | |
a1ffb40e | 2931 | if (__glibc_unlikely (now == read_in)) |
d64b6ad0 | 2932 | /* We haven't succesfully read any character. */ |
2c6fe0bd UD |
2933 | conv_error (); |
2934 | ||
2935 | if (!(flags & SUPPRESS)) | |
2936 | { | |
72acaddf | 2937 | #ifdef COMPILE_WSCANF |
44f8759b | 2938 | /* We have to emit the code to get into the initial |
d64b6ad0 UD |
2939 | state. */ |
2940 | char buf[MB_LEN_MAX]; | |
4aebaa6b | 2941 | size_t n = __wcrtomb (buf, L'\0', &state); |
d64b6ad0 UD |
2942 | if (n > 0 && (flags & MALLOC) |
2943 | && str + n >= *strptr + strsize) | |
2944 | { | |
2945 | /* Enlarge the buffer. */ | |
d90e1b42 UD |
2946 | size_t strleng = str - *strptr; |
2947 | char *newstr; | |
2948 | ||
2949 | newstr = (char *) realloc (*strptr, strleng + n + 1); | |
2950 | if (newstr == NULL) | |
d64b6ad0 | 2951 | { |
267c54dc | 2952 | if (flags & POSIX_MALLOC) |
3f8cc204 UD |
2953 | { |
2954 | done = EOF; | |
2955 | goto errout; | |
2956 | } | |
d64b6ad0 UD |
2957 | /* We lose. Oh well. Terminate the string |
2958 | and stop converting, so at least we don't | |
2959 | skip any input. */ | |
d90e1b42 | 2960 | ((char *) (*strptr))[strleng] = '\0'; |
267c54dc | 2961 | strptr = NULL; |
d64b6ad0 UD |
2962 | ++done; |
2963 | conv_error (); | |
2964 | } | |
2965 | else | |
2966 | { | |
d90e1b42 UD |
2967 | *strptr = newstr; |
2968 | str = newstr + strleng; | |
2969 | strsize = strleng + n + 1; | |
d64b6ad0 UD |
2970 | } |
2971 | } | |
2972 | ||
2973 | str = __mempcpy (str, buf, n); | |
2974 | #endif | |
d90e1b42 | 2975 | *str++ = '\0'; |
d64b6ad0 UD |
2976 | |
2977 | if ((flags & MALLOC) && str - *strptr != strsize) | |
2978 | { | |
2979 | char *cp = (char *) realloc (*strptr, str - *strptr); | |
2980 | if (cp != NULL) | |
2981 | *strptr = cp; | |
2982 | } | |
267c54dc | 2983 | strptr = NULL; |
d64b6ad0 | 2984 | |
2c6fe0bd UD |
2985 | ++done; |
2986 | } | |
28f540f4 RM |
2987 | } |
2988 | break; | |
2989 | ||
d64b6ad0 | 2990 | case L_('p'): /* Generic pointer. */ |
28f540f4 RM |
2991 | base = 16; |
2992 | /* A PTR must be the same size as a `long int'. */ | |
0793d348 | 2993 | flags &= ~(SHORT|LONGDBL); |
859a09cf UD |
2994 | if (need_long) |
2995 | flags |= LONG; | |
4c02bf1a | 2996 | flags |= READ_POINTER; |
28f540f4 | 2997 | goto number; |
47b856a9 UD |
2998 | |
2999 | default: | |
3000 | /* If this is an unknown format character punt. */ | |
27658f20 | 3001 | conv_error (); |
28f540f4 RM |
3002 | } |
3003 | } | |
3004 | ||
14ea22e9 | 3005 | /* The last thing we saw int the format string was a white space. |
05be689b RM |
3006 | Consume the last white spaces. */ |
3007 | if (skip_space) | |
3867ee64 RM |
3008 | { |
3009 | do | |
3010 | c = inchar (); | |
655c0697 | 3011 | while (ISSPACE (c)); |
3867ee64 RM |
3012 | ungetc (c, s); |
3013 | } | |
05be689b | 3014 | |
e3b22ad3 | 3015 | errout: |
7c713e28 | 3016 | /* Unlock stream. */ |
c0fb8a56 | 3017 | UNLOCK_STREAM (s); |
aa1075ea | 3018 | |
95e83974 | 3019 | scratch_buffer_free (&charbuf.scratch); |
e3b22ad3 UD |
3020 | if (errp != NULL) |
3021 | *errp |= errval; | |
3022 | ||
a1ffb40e | 3023 | if (__glibc_unlikely (done == EOF)) |
267c54dc | 3024 | { |
a1ffb40e | 3025 | if (__glibc_unlikely (ptrs_to_free != NULL)) |
267c54dc UD |
3026 | { |
3027 | struct ptrs_to_free *p = ptrs_to_free; | |
3028 | while (p != NULL) | |
3029 | { | |
3030 | for (size_t cnt = 0; cnt < p->count; ++cnt) | |
3031 | { | |
3032 | free (*p->ptrs[cnt]); | |
3033 | *p->ptrs[cnt] = NULL; | |
3034 | } | |
3035 | p = p->next; | |
267c54dc UD |
3036 | ptrs_to_free = p; |
3037 | } | |
3038 | } | |
267c54dc | 3039 | } |
a1ffb40e | 3040 | else if (__glibc_unlikely (strptr != NULL)) |
267c54dc UD |
3041 | { |
3042 | free (*strptr); | |
3043 | *strptr = NULL; | |
3044 | } | |
3867ee64 | 3045 | return done; |
28f540f4 RM |
3046 | } |
3047 | ||
e3b22ad3 | 3048 | #ifdef COMPILE_WSCANF |
d64b6ad0 UD |
3049 | int |
3050 | __vfwscanf (FILE *s, const wchar_t *format, va_list argptr) | |
3051 | { | |
3052 | return _IO_vfwscanf (s, format, argptr, NULL); | |
3053 | } | |
c6251f03 | 3054 | ldbl_weak_alias (__vfwscanf, vfwscanf) |
e3b22ad3 | 3055 | #else |
b2518f55 | 3056 | int |
c6251f03 | 3057 | ___vfscanf (FILE *s, const char *format, va_list argptr) |
b2518f55 | 3058 | { |
c6251f03 | 3059 | return _IO_vfscanf_internal (s, format, argptr, NULL); |
b2518f55 | 3060 | } |
c6251f03 | 3061 | ldbl_strong_alias (_IO_vfscanf_internal, _IO_vfscanf) |
d18ea0c5 | 3062 | ldbl_hidden_def (_IO_vfscanf_internal, _IO_vfscanf) |
c6251f03 RM |
3063 | ldbl_strong_alias (___vfscanf, __vfscanf) |
3064 | ldbl_hidden_def (___vfscanf, __vfscanf) | |
3065 | ldbl_weak_alias (___vfscanf, vfscanf) | |
d64b6ad0 | 3066 | #endif |