1 /*-
2 * Copyright (c) 1990, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * This code is derived from software contributed to Berkeley by
6 * Chris Torek.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the University nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 *
32 * $FreeBSD: src/sys/kern/subr_scanf.c,v 1.13 1999/11/24 01:03:01 archie Exp $
33 * $DragonFly: src/sys/kern/subr_scanf.c,v 1.4 2006/12/13 21:58:50 dillon Exp $
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/param.h>
40 #include <sys/systm.h>
41 #include <sys/ctype.h>
42 #include <machine/limits.h>
43
44 /*
45 * Note that stdarg.h and the ANSI style va_start macro is used for both
46 * ANSI and traditional C compilers.
47 */
48 #include <machine/stdarg.h>
49
50 #define BUF 32 /* Maximum length of numeric string. */
51
52 /*
53 * Flags used during conversion.
54 */
55 #define LONG 0x01 /* l: long or double */
56 #define SHORT 0x04 /* h: short */
57 #define SUPPRESS 0x08 /* suppress assignment */
58 #define POINTER 0x10 /* weird %p pointer (`fake hex') */
59 #define NOSKIP 0x20 /* do not skip blanks */
60 #define QUAD 0x400
61 #define SHORTSHORT 0x4000 /* hh: char */
62
63 /*
64 * The following are used in numeric conversions only:
65 * SIGNOK, NDIGITS, DPTOK, and EXPOK are for floating point;
66 * SIGNOK, NDIGITS, PFXOK, and NZDIGITS are for integral.
67 */
68 #define SIGNOK 0x40 /* +/- is (still) legal */
69 #define NDIGITS 0x80 /* no digits detected */
70
71 #define DPTOK 0x100 /* (float) decimal point is still legal */
72 #define EXPOK 0x200 /* (float) exponent (e+3, etc) still legal */
73
74 #define PFXOK 0x100 /* 0x prefix is (still) legal */
75 #define NZDIGITS 0x200 /* no zero digits detected */
76
77 /*
78 * Conversion types.
79 */
80 #define CT_CHAR 0 /* %c conversion */
81 #define CT_CCL 1 /* %[...] conversion */
82 #define CT_STRING 2 /* %s conversion */
83 #define CT_INT 3 /* integer, i.e., strtoq or strtouq */
84 typedef u_quad_t (*ccfntype)(const char *, char **, int);
85
86 static const u_char *__sccl(char *, const u_char *);
87
88 int
ksscanf(const char * ibuf,const char * fmt,...)89 ksscanf(const char *ibuf, const char *fmt, ...)
90 {
91 __va_list ap;
92 int ret;
93
94 __va_start(ap, fmt);
95 ret = kvsscanf(ibuf, fmt, ap);
96 __va_end(ap);
97 return(ret);
98 }
99
100 int
kvsscanf(const char * inp,char const * fmt0,__va_list ap)101 kvsscanf(const char *inp, char const *fmt0, __va_list ap)
102 {
103 int inr;
104 const u_char *fmt = (const u_char *)fmt0;
105 int c; /* character from format, or conversion */
106 size_t width; /* field width, or 0 */
107 char *p; /* points into all kinds of strings */
108 int n; /* handy integer */
109 int flags; /* flags as defined above */
110 char *p0; /* saves original value of p when necessary */
111 int nassigned; /* number of fields assigned */
112 int nconversions; /* number of conversions */
113 int nread; /* number of characters consumed from fp */
114 int base; /* base argument to strtoq/strtouq */
115 ccfntype ccfn; /* conversion function (strtoq/strtouq) */
116 char ccltab[256]; /* character class table for %[...] */
117 char buf[BUF]; /* buffer for numeric conversions */
118
119 /* `basefix' is used to avoid `if' tests in the integer scanner */
120 static short basefix[17] =
121 { 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 };
122
123 inr = strlen(inp);
124
125 nassigned = 0;
126 nconversions = 0;
127 nread = 0;
128 base = 0; /* XXX just to keep gcc happy */
129 ccfn = NULL; /* XXX just to keep gcc happy */
130 for (;;) {
131 c = *fmt++;
132 if (c == 0)
133 return (nassigned);
134 if (isspace(c)) {
135 while (inr > 0 && isspace(*inp))
136 nread++, inr--, inp++;
137 continue;
138 }
139 if (c != '%')
140 goto literal;
141 width = 0;
142 flags = 0;
143 /*
144 * switch on the format. continue if done;
145 * break once format type is derived.
146 */
147 again: c = *fmt++;
148 switch (c) {
149 case '%':
150 literal:
151 if (inr <= 0)
152 goto input_failure;
153 if (*inp != c)
154 goto match_failure;
155 inr--, inp++;
156 nread++;
157 continue;
158
159 case '*':
160 flags |= SUPPRESS;
161 goto again;
162 case 'l':
163 flags |= LONG;
164 goto again;
165 case 'q':
166 flags |= QUAD;
167 goto again;
168 case 'h':
169 if (flags & SHORT) {
170 flags &= ~SHORT;
171 flags |= SHORTSHORT;
172 } else
173 flags |= SHORT;
174 goto again;
175
176 case '0': case '1': case '2': case '3': case '4':
177 case '5': case '6': case '7': case '8': case '9':
178 width = width * 10 + c - '0';
179 goto again;
180
181 /*
182 * Conversions.
183 *
184 */
185 case 'd':
186 c = CT_INT;
187 ccfn = (ccfntype)strtoq;
188 base = 10;
189 break;
190
191 case 'i':
192 c = CT_INT;
193 ccfn = (ccfntype)strtoq;
194 base = 0;
195 break;
196
197 case 'o':
198 c = CT_INT;
199 ccfn = strtouq;
200 base = 8;
201 break;
202
203 case 'u':
204 c = CT_INT;
205 ccfn = strtouq;
206 base = 10;
207 break;
208
209 case 'x':
210 flags |= PFXOK; /* enable 0x prefixing */
211 c = CT_INT;
212 ccfn = strtouq;
213 base = 16;
214 break;
215
216 case 's':
217 c = CT_STRING;
218 break;
219
220 case '[':
221 fmt = __sccl(ccltab, fmt);
222 flags |= NOSKIP;
223 c = CT_CCL;
224 break;
225
226 case 'c':
227 flags |= NOSKIP;
228 c = CT_CHAR;
229 break;
230
231 case 'p': /* pointer format is like hex */
232 flags |= POINTER | PFXOK;
233 c = CT_INT;
234 ccfn = strtouq;
235 base = 16;
236 break;
237
238 case 'n':
239 nconversions++;
240 if (flags & SUPPRESS) /* ??? */
241 continue;
242 if (flags & SHORTSHORT)
243 *__va_arg(ap, char *) = nread;
244 else if (flags & SHORT)
245 *__va_arg(ap, short *) = nread;
246 else if (flags & LONG)
247 *__va_arg(ap, long *) = nread;
248 else if (flags & QUAD)
249 *__va_arg(ap, quad_t *) = nread;
250 else
251 *__va_arg(ap, int *) = nread;
252 continue;
253 }
254
255 /*
256 * We have a conversion that requires input.
257 */
258 if (inr <= 0)
259 goto input_failure;
260
261 /*
262 * Consume leading white space, except for formats
263 * that suppress this.
264 */
265 if ((flags & NOSKIP) == 0) {
266 while (isspace(*inp)) {
267 nread++;
268 if (--inr > 0)
269 inp++;
270 else
271 goto input_failure;
272 }
273 /*
274 * Note that there is at least one character in
275 * the buffer, so conversions that do not set NOSKIP
276 * can no longer result in an input failure.
277 */
278 }
279
280 /*
281 * Do the conversion.
282 */
283 switch (c) {
284
285 case CT_CHAR:
286 /* scan arbitrary characters (sets NOSKIP) */
287 if (width == 0)
288 width = 1;
289 if (flags & SUPPRESS) {
290 size_t sum = 0;
291 for (;;) {
292 if ((n = inr) < width) {
293 sum += n;
294 width -= n;
295 inp += n;
296 if (sum == 0)
297 goto input_failure;
298 break;
299 } else {
300 sum += width;
301 inr -= width;
302 inp += width;
303 break;
304 }
305 }
306 nread += sum;
307 } else {
308 bcopy(inp, __va_arg(ap, char *), width);
309 inr -= width;
310 inp += width;
311 nread += width;
312 nassigned++;
313 }
314 nconversions++;
315 break;
316
317 case CT_CCL:
318 /* scan a (nonempty) character class (sets NOSKIP) */
319 if (width == 0)
320 width = (size_t)~0; /* `infinity' */
321 /* take only those things in the class */
322 if (flags & SUPPRESS) {
323 n = 0;
324 while (ccltab[(unsigned char)*inp]) {
325 n++, inr--, inp++;
326 if (--width == 0)
327 break;
328 if (inr <= 0) {
329 if (n == 0)
330 goto input_failure;
331 break;
332 }
333 }
334 if (n == 0)
335 goto match_failure;
336 } else {
337 p0 = p = __va_arg(ap, char *);
338 while (ccltab[(unsigned char)*inp]) {
339 inr--;
340 *p++ = *inp++;
341 if (--width == 0)
342 break;
343 if (inr <= 0) {
344 if (p == p0)
345 goto input_failure;
346 break;
347 }
348 }
349 n = p - p0;
350 if (n == 0)
351 goto match_failure;
352 *p = 0;
353 nassigned++;
354 }
355 nread += n;
356 nconversions++;
357 break;
358
359 case CT_STRING:
360 /* like CCL, but zero-length string OK, & no NOSKIP */
361 if (width == 0)
362 width = (size_t)~0;
363 if (flags & SUPPRESS) {
364 n = 0;
365 while (!isspace(*inp)) {
366 n++, inr--, inp++;
367 if (--width == 0)
368 break;
369 if (inr <= 0)
370 break;
371 }
372 nread += n;
373 } else {
374 p0 = p = __va_arg(ap, char *);
375 while (!isspace(*inp)) {
376 inr--;
377 *p++ = *inp++;
378 if (--width == 0)
379 break;
380 if (inr <= 0)
381 break;
382 }
383 *p = 0;
384 nread += p - p0;
385 nassigned++;
386 }
387 nconversions++;
388 continue;
389
390 case CT_INT:
391 /* scan an integer as if by strtoq/strtouq */
392 #ifdef hardway
393 if (width == 0 || width > sizeof(buf) - 1)
394 width = sizeof(buf) - 1;
395 #else
396 /* size_t is unsigned, hence this optimisation */
397 if (--width > sizeof(buf) - 2)
398 width = sizeof(buf) - 2;
399 width++;
400 #endif
401 flags |= SIGNOK | NDIGITS | NZDIGITS;
402 for (p = buf; width; width--) {
403 c = *inp;
404 /*
405 * Switch on the character; `goto ok'
406 * if we accept it as a part of number.
407 */
408 switch (c) {
409
410 /*
411 * The digit 0 is always legal, but is
412 * special. For %i conversions, if no
413 * digits (zero or nonzero) have been
414 * scanned (only signs), we will have
415 * base==0. In that case, we should set
416 * it to 8 and enable 0x prefixing.
417 * Also, if we have not scanned zero digits
418 * before this, do not turn off prefixing
419 * (someone else will turn it off if we
420 * have scanned any nonzero digits).
421 */
422 case '0':
423 if (base == 0) {
424 base = 8;
425 flags |= PFXOK;
426 }
427 if (flags & NZDIGITS)
428 flags &= ~(SIGNOK|NZDIGITS|NDIGITS);
429 else
430 flags &= ~(SIGNOK|PFXOK|NDIGITS);
431 goto ok;
432
433 /* 1 through 7 always legal */
434 case '1': case '2': case '3':
435 case '4': case '5': case '6': case '7':
436 base = basefix[base];
437 flags &= ~(SIGNOK | PFXOK | NDIGITS);
438 goto ok;
439
440 /* digits 8 and 9 ok iff decimal or hex */
441 case '8': case '9':
442 base = basefix[base];
443 if (base <= 8)
444 break; /* not legal here */
445 flags &= ~(SIGNOK | PFXOK | NDIGITS);
446 goto ok;
447
448 /* letters ok iff hex */
449 case 'A': case 'B': case 'C':
450 case 'D': case 'E': case 'F':
451 case 'a': case 'b': case 'c':
452 case 'd': case 'e': case 'f':
453 /* no need to fix base here */
454 if (base <= 10)
455 break; /* not legal here */
456 flags &= ~(SIGNOK | PFXOK | NDIGITS);
457 goto ok;
458
459 /* sign ok only as first character */
460 case '+': case '-':
461 if (flags & SIGNOK) {
462 flags &= ~SIGNOK;
463 goto ok;
464 }
465 break;
466
467 /* x ok iff flag still set & 2nd char */
468 case 'x': case 'X':
469 if (flags & PFXOK && p == buf + 1) {
470 base = 16; /* if %i */
471 flags &= ~PFXOK;
472 goto ok;
473 }
474 break;
475 }
476
477 /*
478 * If we got here, c is not a legal character
479 * for a number. Stop accumulating digits.
480 */
481 break;
482 ok:
483 /*
484 * c is legal: store it and look at the next.
485 */
486 *p++ = c;
487 if (--inr > 0)
488 inp++;
489 else
490 break; /* end of input */
491 }
492 /*
493 * If we had only a sign, it is no good; push
494 * back the sign. If the number ends in `x',
495 * it was [sign] '0' 'x', so push back the x
496 * and treat it as [sign] '0'.
497 */
498 if (flags & NDIGITS) {
499 if (p > buf) {
500 inp--;
501 inr++;
502 }
503 goto match_failure;
504 }
505 c = ((u_char *)p)[-1];
506 if (c == 'x' || c == 'X') {
507 --p;
508 inp--;
509 inr++;
510 }
511 if ((flags & SUPPRESS) == 0) {
512 u_quad_t res;
513
514 *p = 0;
515 res = (*ccfn)(buf, NULL, base);
516 if (flags & POINTER)
517 *__va_arg(ap, void **) =
518 (void *)(uintptr_t)res;
519 else if (flags & SHORTSHORT)
520 *__va_arg(ap, char *) = res;
521 else if (flags & SHORT)
522 *__va_arg(ap, short *) = res;
523 else if (flags & LONG)
524 *__va_arg(ap, long *) = res;
525 else if (flags & QUAD)
526 *__va_arg(ap, quad_t *) = res;
527 else
528 *__va_arg(ap, int *) = res;
529 nassigned++;
530 }
531 nread += p - buf;
532 nconversions++;
533 break;
534
535 }
536 }
537 input_failure:
538 return (nconversions != 0 ? nassigned : -1);
539 match_failure:
540 return (nassigned);
541 }
542
543 /*
544 * Fill in the given table from the scanset at the given format
545 * (just after `['). Return a pointer to the character past the
546 * closing `]'. The table has a 1 wherever characters should be
547 * considered part of the scanset.
548 */
549 static const u_char *
__sccl(char * tab,const u_char * fmt)550 __sccl(char *tab, const u_char *fmt)
551 {
552 int c, n, v;
553
554 /* first `clear' the whole table */
555 c = *fmt++; /* first char hat => negated scanset */
556 if (c == '^') {
557 v = 1; /* default => accept */
558 c = *fmt++; /* get new first char */
559 } else
560 v = 0; /* default => reject */
561
562 /* XXX: Will not work if sizeof(tab*) > sizeof(char) */
563 for (n = 0; n < 256; n++)
564 tab[n] = v; /* memset(tab, v, 256) */
565
566 if (c == 0)
567 return (fmt - 1);/* format ended before closing ] */
568
569 /*
570 * Now set the entries corresponding to the actual scanset
571 * to the opposite of the above.
572 *
573 * The first character may be ']' (or '-') without being special;
574 * the last character may be '-'.
575 */
576 v = 1 - v;
577 for (;;) {
578 tab[c] = v; /* take character c */
579 doswitch:
580 n = *fmt++; /* and examine the next */
581 switch (n) {
582
583 case 0: /* format ended too soon */
584 return (fmt - 1);
585
586 case '-':
587 /*
588 * A scanset of the form
589 * [01+-]
590 * is defined as `the digit 0, the digit 1,
591 * the character +, the character -', but
592 * the effect of a scanset such as
593 * [a-zA-Z0-9]
594 * is implementation defined. The V7 Unix
595 * scanf treats `a-z' as `the letters a through
596 * z', but treats `a-a' as `the letter a, the
597 * character -, and the letter a'.
598 *
599 * For compatibility, the `-' is not considerd
600 * to define a range if the character following
601 * it is either a close bracket (required by ANSI)
602 * or is not numerically greater than the character
603 * we just stored in the table (c).
604 */
605 n = *fmt;
606 if (n == ']' || n < c) {
607 c = '-';
608 break; /* resume the for(;;) */
609 }
610 fmt++;
611 /* fill in the range */
612 do {
613 tab[++c] = v;
614 } while (c < n);
615 c = n;
616 /*
617 * Alas, the V7 Unix scanf also treats formats
618 * such as [a-c-e] as `the letters a through e'.
619 * This too is permitted by the standard....
620 */
621 goto doswitch;
622 break;
623
624 case ']': /* end of scanset */
625 return (fmt);
626
627 default: /* just another character */
628 c = n;
629 break;
630 }
631 }
632 /* NOTREACHED */
633 }
634
635