1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1990, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * This code is derived from software contributed to Berkeley by
8 * Chris Torek.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 *
34 * From: Id: vfscanf.c,v 1.13 1998/09/25 12:20:27 obrien Exp
35 * From: static char sccsid[] = "@(#)strtol.c 8.1 (Berkeley) 6/4/93";
36 * From: static char sccsid[] = "@(#)strtoul.c 8.1 (Berkeley) 6/4/93";
37 */
38
39 #include <sys/cdefs.h>
40 #include <sys/param.h>
41 #include <sys/systm.h>
42 #include <sys/ctype.h>
43 #include <sys/limits.h>
44 #include <sys/stddef.h>
45
46 /*
47 * Note that stdarg.h and the ANSI style va_start macro is used for both
48 * ANSI and traditional C compilers.
49 */
50 #include <machine/stdarg.h>
51
52 #define BUF 32 /* Maximum length of numeric string. */
53
54 /*
55 * Flags used during conversion.
56 */
57 #define LONG 0x01 /* l: long or double */
58 #define SHORT 0x04 /* h: short */
59 #define SUPPRESS 0x08 /* suppress assignment */
60 #define POINTER 0x10 /* weird %p pointer (`fake hex') */
61 #define NOSKIP 0x20 /* do not skip blanks */
62 #define QUAD 0x400
63 #define INTMAXT 0x800 /* j: intmax_t */
64 #define PTRDIFFT 0x1000 /* t: ptrdiff_t */
65 #define SIZET 0x2000 /* z: size_t */
66 #define SHORTSHORT 0x4000 /** hh: char */
67
68 /*
69 * The following are used in numeric conversions only:
70 * SIGNOK, NDIGITS, DPTOK, and EXPOK are for floating point;
71 * SIGNOK, NDIGITS, PFXOK, and NZDIGITS are for integral.
72 */
73 #define SIGNOK 0x40 /* +/- is (still) legal */
74 #define NDIGITS 0x80 /* no digits detected */
75
76 #define DPTOK 0x100 /* (float) decimal point is still legal */
77 #define EXPOK 0x200 /* (float) exponent (e+3, etc) still legal */
78
79 #define PFXOK 0x100 /* 0x prefix is (still) legal */
80 #define NZDIGITS 0x200 /* no zero digits detected */
81
82 /*
83 * Conversion types.
84 */
85 #define CT_CHAR 0 /* %c conversion */
86 #define CT_CCL 1 /* %[...] conversion */
87 #define CT_STRING 2 /* %s conversion */
88 #define CT_INT 3 /* integer, i.e., strtoq or strtouq */
89 typedef u_quad_t (*ccfntype)(const char *, char **, int);
90
91 static const u_char *__sccl(char *, const u_char *);
92
93 int
sscanf(const char * ibuf,const char * fmt,...)94 sscanf(const char *ibuf, const char *fmt, ...)
95 {
96 va_list ap;
97 int ret;
98
99 va_start(ap, fmt);
100 ret = vsscanf(ibuf, fmt, ap);
101 va_end(ap);
102 return(ret);
103 }
104
105 int
vsscanf(const char * inp,char const * fmt0,va_list ap)106 vsscanf(const char *inp, char const *fmt0, va_list ap)
107 {
108 int inr;
109 const u_char *fmt = (const u_char *)fmt0;
110 int c; /* character from format, or conversion */
111 size_t width; /* field width, or 0 */
112 char *p; /* points into all kinds of strings */
113 int n; /* handy integer */
114 int flags; /* flags as defined above */
115 char *p0; /* saves original value of p when necessary */
116 int nassigned; /* number of fields assigned */
117 int nconversions; /* number of conversions */
118 int nread; /* number of characters consumed from fp */
119 int base; /* base argument to strtoq/strtouq */
120 ccfntype ccfn; /* conversion function (strtoq/strtouq) */
121 char ccltab[256]; /* character class table for %[...] */
122 char buf[BUF]; /* buffer for numeric conversions */
123
124 /* `basefix' is used to avoid `if' tests in the integer scanner */
125 static short basefix[17] =
126 { 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 };
127
128 inr = strlen(inp);
129
130 nassigned = 0;
131 nconversions = 0;
132 nread = 0;
133 base = 0; /* XXX just to keep gcc happy */
134 ccfn = NULL; /* XXX just to keep gcc happy */
135 for (;;) {
136 c = *fmt++;
137 if (c == 0)
138 return (nassigned);
139 if (isspace(c)) {
140 while (inr > 0 && isspace(*inp))
141 nread++, inr--, inp++;
142 continue;
143 }
144 if (c != '%')
145 goto literal;
146 width = 0;
147 flags = 0;
148 /*
149 * switch on the format. continue if done;
150 * break once format type is derived.
151 */
152 again: c = *fmt++;
153 switch (c) {
154 case '%':
155 literal:
156 if (inr <= 0)
157 goto input_failure;
158 if (*inp != c)
159 goto match_failure;
160 inr--, inp++;
161 nread++;
162 continue;
163
164 case '*':
165 flags |= SUPPRESS;
166 goto again;
167 case 'j':
168 flags |= INTMAXT;
169 goto again;
170 case 'l':
171 if (flags & LONG){
172 flags &= ~LONG;
173 flags |= QUAD;
174 } else {
175 flags |= LONG;
176 }
177 goto again;
178 case 'q':
179 flags |= QUAD;
180 goto again;
181 case 't':
182 flags |= PTRDIFFT;
183 goto again;
184 case 'z':
185 flags |= SIZET;
186 goto again;
187 case 'h':
188 if (flags & SHORT){
189 flags &= ~SHORT;
190 flags |= SHORTSHORT;
191 } else {
192 flags |= SHORT;
193 }
194 goto again;
195
196 case '0': case '1': case '2': case '3': case '4':
197 case '5': case '6': case '7': case '8': case '9':
198 width = width * 10 + c - '0';
199 goto again;
200
201 /*
202 * Conversions.
203 *
204 */
205 case 'd':
206 c = CT_INT;
207 ccfn = (ccfntype)strtoq;
208 base = 10;
209 break;
210
211 case 'i':
212 c = CT_INT;
213 ccfn = (ccfntype)strtoq;
214 base = 0;
215 break;
216
217 case 'o':
218 c = CT_INT;
219 ccfn = strtouq;
220 base = 8;
221 break;
222
223 case 'u':
224 c = CT_INT;
225 ccfn = strtouq;
226 base = 10;
227 break;
228
229 case 'x':
230 flags |= PFXOK; /* enable 0x prefixing */
231 c = CT_INT;
232 ccfn = strtouq;
233 base = 16;
234 break;
235
236 case 's':
237 c = CT_STRING;
238 break;
239
240 case '[':
241 fmt = __sccl(ccltab, fmt);
242 flags |= NOSKIP;
243 c = CT_CCL;
244 break;
245
246 case 'c':
247 flags |= NOSKIP;
248 c = CT_CHAR;
249 break;
250
251 case 'p': /* pointer format is like hex */
252 flags |= POINTER | PFXOK;
253 c = CT_INT;
254 ccfn = strtouq;
255 base = 16;
256 break;
257
258 case 'n':
259 nconversions++;
260 if (flags & SUPPRESS) /* ??? */
261 continue;
262 if (flags & SHORTSHORT)
263 *va_arg(ap, char *) = nread;
264 else if (flags & SHORT)
265 *va_arg(ap, short *) = nread;
266 else if (flags & LONG)
267 *va_arg(ap, long *) = nread;
268 else if (flags & QUAD)
269 *va_arg(ap, quad_t *) = nread;
270 else if (flags & INTMAXT)
271 *va_arg(ap, intmax_t *) = nread;
272 else if (flags & SIZET)
273 *va_arg(ap, size_t *) = nread;
274 else if (flags & PTRDIFFT)
275 *va_arg(ap, ptrdiff_t *) = nread;
276 else
277 *va_arg(ap, int *) = nread;
278 continue;
279 }
280
281 /*
282 * We have a conversion that requires input.
283 */
284 if (inr <= 0)
285 goto input_failure;
286
287 /*
288 * Consume leading white space, except for formats
289 * that suppress this.
290 */
291 if ((flags & NOSKIP) == 0) {
292 while (isspace(*inp)) {
293 nread++;
294 if (--inr > 0)
295 inp++;
296 else
297 goto input_failure;
298 }
299 /*
300 * Note that there is at least one character in
301 * the buffer, so conversions that do not set NOSKIP
302 * can no longer result in an input failure.
303 */
304 }
305
306 /*
307 * Do the conversion.
308 */
309 switch (c) {
310 case CT_CHAR:
311 /* scan arbitrary characters (sets NOSKIP) */
312 if (width == 0)
313 width = 1;
314 if (flags & SUPPRESS) {
315 size_t sum = 0;
316 for (;;) {
317 if ((n = inr) < width) {
318 sum += n;
319 width -= n;
320 inp += n;
321 if (sum == 0)
322 goto input_failure;
323 break;
324 } else {
325 sum += width;
326 inr -= width;
327 inp += width;
328 break;
329 }
330 }
331 nread += sum;
332 } else {
333 bcopy(inp, va_arg(ap, char *), width);
334 inr -= width;
335 inp += width;
336 nread += width;
337 nassigned++;
338 }
339 nconversions++;
340 break;
341
342 case CT_CCL:
343 /* scan a (nonempty) character class (sets NOSKIP) */
344 if (width == 0)
345 width = (size_t)~0; /* `infinity' */
346 /* take only those things in the class */
347 if (flags & SUPPRESS) {
348 n = 0;
349 while (ccltab[(unsigned char)*inp]) {
350 n++, inr--, inp++;
351 if (--width == 0)
352 break;
353 if (inr <= 0) {
354 if (n == 0)
355 goto input_failure;
356 break;
357 }
358 }
359 if (n == 0)
360 goto match_failure;
361 } else {
362 p0 = p = va_arg(ap, char *);
363 while (ccltab[(unsigned char)*inp]) {
364 inr--;
365 *p++ = *inp++;
366 if (--width == 0)
367 break;
368 if (inr <= 0) {
369 if (p == p0)
370 goto input_failure;
371 break;
372 }
373 }
374 n = p - p0;
375 if (n == 0)
376 goto match_failure;
377 *p = 0;
378 nassigned++;
379 }
380 nread += n;
381 nconversions++;
382 break;
383
384 case CT_STRING:
385 /* like CCL, but zero-length string OK, & no NOSKIP */
386 if (width == 0)
387 width = (size_t)~0;
388 if (flags & SUPPRESS) {
389 n = 0;
390 while (!isspace(*inp)) {
391 n++, inr--, inp++;
392 if (--width == 0)
393 break;
394 if (inr <= 0)
395 break;
396 }
397 nread += n;
398 } else {
399 p0 = p = va_arg(ap, char *);
400 while (!isspace(*inp)) {
401 inr--;
402 *p++ = *inp++;
403 if (--width == 0)
404 break;
405 if (inr <= 0)
406 break;
407 }
408 *p = 0;
409 nread += p - p0;
410 nassigned++;
411 }
412 nconversions++;
413 continue;
414
415 case CT_INT:
416 /* scan an integer as if by strtoq/strtouq */
417 #ifdef hardway
418 if (width == 0 || width > sizeof(buf) - 1)
419 width = sizeof(buf) - 1;
420 #else
421 /* size_t is unsigned, hence this optimisation */
422 if (--width > sizeof(buf) - 2)
423 width = sizeof(buf) - 2;
424 width++;
425 #endif
426 flags |= SIGNOK | NDIGITS | NZDIGITS;
427 for (p = buf; width; width--) {
428 c = *inp;
429 /*
430 * Switch on the character; `goto ok'
431 * if we accept it as a part of number.
432 */
433 switch (c) {
434 /*
435 * The digit 0 is always legal, but is
436 * special. For %i conversions, if no
437 * digits (zero or nonzero) have been
438 * scanned (only signs), we will have
439 * base==0. In that case, we should set
440 * it to 8 and enable 0x prefixing.
441 * Also, if we have not scanned zero digits
442 * before this, do not turn off prefixing
443 * (someone else will turn it off if we
444 * have scanned any nonzero digits).
445 */
446 case '0':
447 if (base == 0) {
448 base = 8;
449 flags |= PFXOK;
450 }
451 if (flags & NZDIGITS)
452 flags &= ~(SIGNOK|NZDIGITS|NDIGITS);
453 else
454 flags &= ~(SIGNOK|PFXOK|NDIGITS);
455 goto ok;
456
457 /* 1 through 7 always legal */
458 case '1': case '2': case '3':
459 case '4': case '5': case '6': case '7':
460 base = basefix[base];
461 flags &= ~(SIGNOK | PFXOK | NDIGITS);
462 goto ok;
463
464 /* digits 8 and 9 ok iff decimal or hex */
465 case '8': case '9':
466 base = basefix[base];
467 if (base <= 8)
468 break; /* not legal here */
469 flags &= ~(SIGNOK | PFXOK | NDIGITS);
470 goto ok;
471
472 /* letters ok iff hex */
473 case 'A': case 'B': case 'C':
474 case 'D': case 'E': case 'F':
475 case 'a': case 'b': case 'c':
476 case 'd': case 'e': case 'f':
477 /* no need to fix base here */
478 if (base <= 10)
479 break; /* not legal here */
480 flags &= ~(SIGNOK | PFXOK | NDIGITS);
481 goto ok;
482
483 /* sign ok only as first character */
484 case '+': case '-':
485 if (flags & SIGNOK) {
486 flags &= ~SIGNOK;
487 goto ok;
488 }
489 break;
490
491 /* x ok iff flag still set & 2nd char */
492 case 'x': case 'X':
493 if (flags & PFXOK && p == buf + 1) {
494 base = 16; /* if %i */
495 flags &= ~PFXOK;
496 goto ok;
497 }
498 break;
499 }
500
501 /*
502 * If we got here, c is not a legal character
503 * for a number. Stop accumulating digits.
504 */
505 break;
506 ok:
507 /*
508 * c is legal: store it and look at the next.
509 */
510 *p++ = c;
511 if (--inr > 0)
512 inp++;
513 else
514 break; /* end of input */
515 }
516 /*
517 * If we had only a sign, it is no good; push
518 * back the sign. If the number ends in `x',
519 * it was [sign] '0' 'x', so push back the x
520 * and treat it as [sign] '0'.
521 */
522 if (flags & NDIGITS) {
523 if (p > buf) {
524 inp--;
525 inr++;
526 }
527 goto match_failure;
528 }
529 c = ((u_char *)p)[-1];
530 if (c == 'x' || c == 'X') {
531 --p;
532 inp--;
533 inr++;
534 }
535 if ((flags & SUPPRESS) == 0) {
536 u_quad_t res;
537
538 *p = 0;
539 res = (*ccfn)(buf, (char **)NULL, base);
540 if (flags & POINTER)
541 *va_arg(ap, void **) =
542 (void *)(uintptr_t)res;
543 else if (flags & SHORTSHORT)
544 *va_arg(ap, char *) = res;
545 else if (flags & SHORT)
546 *va_arg(ap, short *) = res;
547 else if (flags & LONG)
548 *va_arg(ap, long *) = res;
549 else if (flags & QUAD)
550 *va_arg(ap, quad_t *) = res;
551 else if (flags & INTMAXT)
552 *va_arg(ap, intmax_t *) = res;
553 else if (flags & PTRDIFFT)
554 *va_arg(ap, ptrdiff_t *) = res;
555 else if (flags & SIZET)
556 *va_arg(ap, size_t *) = res;
557 else
558 *va_arg(ap, int *) = res;
559 nassigned++;
560 }
561 nread += p - buf;
562 nconversions++;
563 break;
564 }
565 }
566 input_failure:
567 return (nconversions != 0 ? nassigned : -1);
568 match_failure:
569 return (nassigned);
570 }
571
572 /*
573 * Fill in the given table from the scanset at the given format
574 * (just after `['). Return a pointer to the character past the
575 * closing `]'. The table has a 1 wherever characters should be
576 * considered part of the scanset.
577 */
578 static const u_char *
__sccl(char * tab,const u_char * fmt)579 __sccl(char *tab, const u_char *fmt)
580 {
581 int c, n, v;
582
583 /* first `clear' the whole table */
584 c = *fmt++; /* first char hat => negated scanset */
585 if (c == '^') {
586 v = 1; /* default => accept */
587 c = *fmt++; /* get new first char */
588 } else
589 v = 0; /* default => reject */
590
591 /* XXX: Will not work if sizeof(tab*) > sizeof(char) */
592 for (n = 0; n < 256; n++)
593 tab[n] = v; /* memset(tab, v, 256) */
594
595 if (c == 0)
596 return (fmt - 1);/* format ended before closing ] */
597
598 /*
599 * Now set the entries corresponding to the actual scanset
600 * to the opposite of the above.
601 *
602 * The first character may be ']' (or '-') without being special;
603 * the last character may be '-'.
604 */
605 v = 1 - v;
606 for (;;) {
607 tab[c] = v; /* take character c */
608 doswitch:
609 n = *fmt++; /* and examine the next */
610 switch (n) {
611 case 0: /* format ended too soon */
612 return (fmt - 1);
613
614 case '-':
615 /*
616 * A scanset of the form
617 * [01+-]
618 * is defined as `the digit 0, the digit 1,
619 * the character +, the character -', but
620 * the effect of a scanset such as
621 * [a-zA-Z0-9]
622 * is implementation defined. The V7 Unix
623 * scanf treats `a-z' as `the letters a through
624 * z', but treats `a-a' as `the letter a, the
625 * character -, and the letter a'.
626 *
627 * For compatibility, the `-' is not considered
628 * to define a range if the character following
629 * it is either a close bracket (required by ANSI)
630 * or is not numerically greater than the character
631 * we just stored in the table (c).
632 */
633 n = *fmt;
634 if (n == ']' || n < c) {
635 c = '-';
636 break; /* resume the for(;;) */
637 }
638 fmt++;
639 /* fill in the range */
640 do {
641 tab[++c] = v;
642 } while (c < n);
643 c = n;
644 /*
645 * Alas, the V7 Unix scanf also treats formats
646 * such as [a-c-e] as `the letters a through e'.
647 * This too is permitted by the standard....
648 */
649 goto doswitch;
650 break;
651
652 case ']': /* end of scanset */
653 return (fmt);
654
655 default: /* just another character */
656 c = n;
657 break;
658 }
659 }
660 /* NOTREACHED */
661 }
662