Graphviz 14.0.2~dev.20251008.0253
Loading...
Searching...
No Matches
sfvscanf.c
Go to the documentation of this file.
1/*************************************************************************
2 * Copyright (c) 2011 AT&T Intellectual Property
3 * All rights reserved. This program and the accompanying materials
4 * are made available under the terms of the Eclipse Public License v1.0
5 * which accompanies this distribution, and is available at
6 * https://www.eclipse.org/legal/epl-v10.html
7 *
8 * Contributors: Details at https://graphviz.org
9 *************************************************************************/
10
11#include <assert.h>
12#include <inttypes.h>
13#include <limits.h>
14#include <sfio/sfhdr.h>
15#include <stdbool.h>
16#include <stddef.h>
17#include <stdio.h>
18#include <util/gv_ctype.h>
19
20/* The main engine for reading formatted data
21**
22** Written by Kiem-Phong Vo.
23*/
24
25#define MAXWIDTH INT_MAX // max amount to scan
26
31static const unsigned char *setclass(const unsigned char *form, bool *accept) {
32 int fmt, c;
33 bool yes;
34
35 if ((fmt = *form++) == '^') { /* we want the complement of this set */
36 yes = false;
37 fmt = *form++;
38 } else
39 yes = true;
40
41 for (c = 0; c <= UCHAR_MAX; ++c)
42 accept[c] = !yes;
43
44 if (fmt == ']' || fmt == '-') { /* special first char */
45 accept[fmt] = yes;
46 fmt = *form++;
47 }
48
49 for (; fmt != ']'; fmt = *form++) { /* done */
50 if (!fmt)
51 return form - 1;
52
53 /* interval */
54 if (fmt != '-' || form[0] == ']' || form[-2] > form[0])
55 accept[fmt] = yes;
56 else
57 for (c = form[-2] + 1; c < form[0]; ++c)
58 accept[c] = yes;
59 }
60
61 return form;
62}
63
68int sfvscanf(FILE *f, Sffmt_t *ft) {
69 int inp, shift, base, width;
70 ssize_t size;
71 int fmt, flags, dot, n_assign, v, n, n_input;
72 char *sp;
73 char accept[SF_MAXDIGITS];
74
75 Argv_t argv;
76
77 int argn;
78
79 void *value; /* location to assign scanned value */
80 const char *t_str;
81 ssize_t n_str;
82
83#define SFGETC(f,c) (((c) = getc(f)) < 0 ? c : (++n_input, c))
84#define SFUNGETC(f,c) do { \
85 ungetc((c), (f)); \
86 --n_input; \
87} while (0)
88
89 assert(f != NULL);
90
91 n_assign = n_input = 0;
92
93 inp = -1;
94
95 const char *form;
96 argv.ft = ft;
97
98 form = argv.ft->form;
99 argn = -1;
100
101 assert(ft != NULL && ft->extf != NULL);
102
103 loop_fmt:
104 while ((fmt = *form++)) {
105 if (fmt != '%') {
106 if (gv_isspace(fmt)) {
107 if (fmt != '\n')
108 fmt = -1;
109 for (;;) {
110 if (SFGETC(f, inp) < 0 || inp == fmt)
111 goto loop_fmt;
112 else if (!gv_isspace(inp)) {
113 SFUNGETC(f, inp);
114 goto loop_fmt;
115 }
116 }
117 } else {
118 match_1:
119 if (SFGETC(f, inp) != fmt) {
120 if (inp >= 0)
121 SFUNGETC(f, inp);
122 goto done;
123 }
124 }
125 continue;
126 }
127
128 if (*form == '%') {
129 form += 1;
130 goto match_1;
131 }
132
133 if (*form == '\0')
134 goto done;
135
136 if (*form == '*') {
138 form += 1;
139 } else
140 flags = 0;
141
142 /* matching some pattern */
143 base = 10;
144 size = -1;
145 width = dot = 0;
146 t_str = NULL;
147 n_str = 0;
148 value = NULL;
149
150 loop_flags: /* LOOP FOR FLAGS, WIDTH, BASE, TYPE */
151 switch ((fmt = *form++)) {
152 case LEFTP: /* get the type which is enclosed in balanced () */
153 t_str = form;
154 for (v = 1;;) {
155 switch (*form++) {
156 case 0: /* not balanceable, retract */
157 form = t_str;
158 t_str = NULL;
159 n_str = 0;
160 goto loop_flags;
161 case LEFTP: /* increasing nested level */
162 v += 1;
163 continue;
164 case RIGHTP: /* decreasing nested level */
165 if ((v -= 1) != 0)
166 continue;
167 if (*t_str != '*')
168 n_str = (form - 1) - t_str;
169 else {
170 t_str = _Sffmtintf(t_str + 1, &n);
171 ++argn;
172
173 FMTSET(ft, form, LEFTP, 0, 0, 0, 0, 0, NULL, 0);
174 n = ft->extf(&argv, ft);
175 if (n < 0)
176 goto done;
177 assert(ft->flags & SFFMT_VALUE);
178 if ((t_str = argv.s) && (n_str = (int)ft->size) < 0)
179 n_str = (ssize_t)strlen(t_str);
180 }
181 goto loop_flags;
182 default:
183 // skip over
184 break;
185 }
186 }
187
188 case '#': /* alternative format */
190 goto loop_flags;
191
192 case '.': /* width & base */
193 dot += 1;
194 if (gv_isdigit(*form)) {
195 fmt = *form++;
196 goto dot_size;
197 } else if (*form == '*') {
198 form = _Sffmtintf(form + 1, &n);
199 n = ++argn;
200
201 FMTSET(ft, form, '.', dot, 0, 0, 0, 0, NULL, 0);
202 if (ft->extf(&argv, ft) < 0)
203 goto done;
204 assert(ft->flags & SFFMT_VALUE);
205 v = argv.i;
206 if (v < 0)
207 v = 0;
208 goto dot_set;
209 } else
210 goto loop_flags;
211
212 case '0':
213 case '1':
214 case '2':
215 case '3':
216 case '4':
217 case '5':
218 case '6':
219 case '7':
220 case '8':
221 case '9':
222 dot_size:
223 for (v = fmt - '0'; gv_isdigit(*form); ++form)
224 v = v * 10 + (*form - '0');
225
226 dot_set:
227 if (dot == 0 || dot == 1)
228 width = v;
229 else if (dot == 2)
230 base = v;
231 goto loop_flags;
232
233 case 'I': /* object size */
234 size = 0;
235 flags = (flags & ~SFFMT_TYPES) | SFFMT_IFLAG;
236 if (gv_isdigit(*form)) {
237 for (n = *form; gv_isdigit(n); n = *++form)
238 size = size * 10 + (n - '0');
239 } else if (*form == '*') {
240 form = _Sffmtintf(form + 1, &n);
241 n = ++argn;
242
243 FMTSET(ft, form, 'I', sizeof(int), 0, 0, 0, 0, NULL, 0);
244 if (ft->extf(&argv, ft) < 0)
245 goto done;
246 assert(ft->flags & SFFMT_VALUE);
247 size = argv.i;
248 }
249 goto loop_flags;
250
251 case 'l':
252 size = -1;
253 flags &= ~SFFMT_TYPES;
254 if (*form == 'l') {
255 form += 1;
257 } else
258 flags |= SFFMT_LONG;
259 goto loop_flags;
260 case 'h':
261 size = -1;
262 flags &= ~SFFMT_TYPES;
263 if (*form == 'h') {
264 form += 1;
266 } else
268 goto loop_flags;
269 case 'L':
270 size = -1;
271 flags = (flags & ~SFFMT_TYPES) | SFFMT_LDOUBLE;
272 goto loop_flags;
273 case 'j':
274 size = -1;
275 flags = (flags & ~SFFMT_TYPES) | SFFMT_JFLAG;
276 goto loop_flags;
277 case 'z':
278 size = -1;
279 flags = (flags & ~SFFMT_TYPES) | SFFMT_ZFLAG;
280 goto loop_flags;
281 case 't':
282 size = -1;
283 flags = (flags & ~SFFMT_TYPES) | SFFMT_TFLAG;
284 goto loop_flags;
285 default: // continue with logic below
286 break;
287 }
288
289 /* set object size */
290 if (flags & (SFFMT_TYPES & ~SFFMT_IFLAG)) {
291 if ((_Sftype[fmt] & (SFFMT_INT | SFFMT_UINT)) || fmt == 'n') {
292 size = (flags & SFFMT_LLONG) ? (ssize_t)sizeof(long long) :
293 (flags & SFFMT_LONG) ? (ssize_t)sizeof(long) :
294 (flags & SFFMT_SHORT) ? (ssize_t)sizeof(short) :
295 (flags & SFFMT_SSHORT) ? (ssize_t)sizeof(char) :
296 (flags & SFFMT_JFLAG) ? (ssize_t)sizeof(long long) :
297 (flags & SFFMT_TFLAG) ? (ssize_t)sizeof(ptrdiff_t) :
298 (flags & SFFMT_ZFLAG) ? (ssize_t)sizeof(size_t) : -1;
299 } else if (_Sftype[fmt] & SFFMT_FLOAT) {
300 size = (flags & SFFMT_LDOUBLE) ? (ssize_t)sizeof(long double) :
301 (flags & (SFFMT_LONG | SFFMT_LLONG)) ? (ssize_t)sizeof(double) : -1;
302 }
303 }
304
305 ++argn;
306 FMTSET(ft, form, fmt, size, flags, width, 0, base, t_str, n_str);
307 v = ft->extf(&argv, ft);
308
309 if (v < 0)
310 goto done;
311 else if (v == 0) { // extf did not use input stream
312 FMTGET(ft, form, fmt, size, flags, width, n, base);
313 if ((ft->flags & SFFMT_VALUE) && !(ft->flags & SFFMT_SKIP))
314 value = argv.vp;
315 } else { // v > 0: number of input bytes consumed
316 n_input += v;
317 if (!(ft->flags & SFFMT_SKIP))
318 n_assign += 1;
319 continue;
320 }
321
322 if (_Sftype[fmt] == 0) /* unknown pattern */
323 continue;
324
325 assert(!(!value && !(flags & SFFMT_SKIP)));
326
327 if (fmt == 'n') { /* return length of consumed input */
328 if (sizeof(long) > sizeof(int) && FMTCMP(size, long, long long))
329 *((long *) value) = (long)n_input;
330 else if (sizeof(short) < sizeof(int) && FMTCMP(size, short, long long))
331 *((short *) value) = (short)n_input;
332 else if (size == sizeof(char))
333 *((char *) value) = (char)n_input;
334 else
335 *((int *) value) = (int)n_input;
336 continue;
337 }
338
339 /* if get here, start scanning input */
340 if (width == 0)
341 width = fmt == 'c' ? 1 : MAXWIDTH;
342
343 /* define the first input character */
344 if (fmt == 'c' || fmt == '[')
345 SFGETC(f, inp);
346 else {
347 do {
348 SFGETC(f, inp);
349 }
350 while (gv_isspace(inp)) // skip starting blanks
351 ;
352 }
353 if (inp < 0)
354 goto done;
355
356 if (_Sftype[fmt] == SFFMT_FLOAT) {
357 char *val;
358
359 val = accept;
360 if (width >= 0 && (size_t)width >= SF_MAXDIGITS)
361 width = SF_MAXDIGITS - 1;
362 int exponent = 0;
363 bool seen_dot = false;
364 do {
365 if (gv_isdigit(inp))
366 *val++ = inp;
367 else if (inp == '.') { /* too many dots */
368 if (seen_dot)
369 break;
370 seen_dot = true;
371 *val++ = '.';
372 } else if (inp == 'e' || inp == 'E') { /* too many e,E */
373 if (exponent++ > 0)
374 break;
375 *val++ = inp;
376 if (--width <= 0 || SFGETC(f, inp) < 0 ||
377 (inp != '-' && inp != '+' && !gv_isdigit(inp)))
378 break;
379 *val++ = inp;
380 } else if (inp == '-' || inp == '+') { /* too many signs */
381 if (val > accept)
382 break;
383 *val++ = inp;
384 } else
385 break;
386
387 } while (--width > 0 && SFGETC(f, inp) >= 0);
388
389 if (value) {
390 *val = '\0';
391 argv.d = strtod(accept, NULL);
392
393 n_assign += 1;
394 if (FMTCMP(size, double, long double))
395 *((double *) value) = argv.d;
396 else
397 *((float *) value) = (float) argv.d;
398 }
399 } else if (_Sftype[fmt] == SFFMT_UINT || fmt == 'p') {
400 if (inp == '-') {
401 SFUNGETC(f, inp);
402 goto done;
403 } else
404 goto int_cvt;
405 } else if (_Sftype[fmt] == SFFMT_INT) {
406 int_cvt:
407 if (inp == '-' || inp == '+') {
408 if (inp == '-')
410 while (--width > 0 && SFGETC(f, inp) >= 0)
411 if (!gv_isspace(inp))
412 break;
413 }
414 if (inp < 0)
415 goto done;
416
417 if (fmt == 'o')
418 base = 8;
419 else if (fmt == 'x' || fmt == 'p')
420 base = 16;
421 else if (fmt == 'i' && inp == '0') { /* self-described data */
422 base = 8;
423 if (width > 1) { /* peek to see if it's a base-16 */
424 if (SFGETC(f, inp) >= 0) {
425 if (inp == 'x' || inp == 'X')
426 base = 16;
427 SFUNGETC(f, inp);
428 }
429 inp = '0';
430 }
431 }
432
433 /* now convert */
434 argv.lu = 0;
435 if (base == 16) {
436 sp = (char *) _Sfcv36;
437 shift = 4;
438 if (sp[inp] >= 16) {
439 SFUNGETC(f, inp);
440 goto done;
441 }
442 if (inp == '0' && --width > 0) { /* skip leading 0x or 0X */
443 if (SFGETC(f, inp) >= 0 &&
444 (inp == 'x' || inp == 'X') && --width > 0)
445 SFGETC(f, inp);
446 }
447 if (inp >= 0 && sp[inp] < 16)
448 goto base_shift;
449 } else if (base == 10) { /* fast base 10 conversion */
450 if (inp < '0' || inp > '9') {
451 SFUNGETC(f, inp);
452 goto done;
453 }
454
455 do {
456 argv.lu =
457 (argv.lu << 3) + (argv.lu << 1) + (inp - '0');
458 } while (--width > 0 && SFGETC(f, inp) >= '0'
459 && inp <= '9');
460
461 if (fmt == 'i' && inp == '#' && !(flags & SFFMT_ALTER)) {
462 base = (int) argv.lu;
463 if (base < 2 || base > SF_RADIX)
464 goto done;
465 argv.lu = 0;
466 sp = base <= 36 ? (char *) _Sfcv36 : (char *) _Sfcv64;
467 if (--width > 0 &&
468 SFGETC(f, inp) >= 0 && sp[inp] < base)
469 goto base_conv;
470 }
471 } else { /* other bases */
472 sp = base <= 36 ? (char *) _Sfcv36 : (char *) _Sfcv64;
473 if (base < 2 || base > SF_RADIX || sp[inp] >= base) {
474 SFUNGETC(f, inp);
475 goto done;
476 }
477
478 base_conv: /* check for power of 2 conversions */
479 if ((base & ~(base - 1)) == base) {
480 if (base < 8)
481 shift = base < 4 ? 1 : 2;
482 else if (base < 32)
483 shift = base < 16 ? 3 : 4;
484 else
485 shift = base < 64 ? 5 : 6;
486
487 base_shift:do {
488 argv.lu = (argv.lu << shift) + sp[inp];
489 } while (--width > 0 &&
490 SFGETC(f, inp) >= 0 && sp[inp] < base);
491 } else {
492 do {
493 argv.lu = (argv.lu * base) + sp[inp];
494 } while (--width > 0 &&
495 SFGETC(f, inp) >= 0 && sp[inp] < base);
496 }
497 }
498
499 if (flags & SFFMT_MINUS)
500 argv.ll = -argv.ll;
501
502 if (value) {
503 n_assign += 1;
504
505 if (fmt == 'p') {
506 *((void **) value) = (void *)(uintptr_t)argv.lu;
507 } else if (sizeof(long) > sizeof(int) && FMTCMP(size, long, long long)) {
508 if (fmt == 'd' || fmt == 'i')
509 *((long *) value) = (long) argv.ll;
510 else
511 *((ulong *) value) = (ulong) argv.lu;
512 } else if (sizeof(short) < sizeof(int) && FMTCMP(size, short, long long)) {
513 if (fmt == 'd' || fmt == 'i')
514 *((short *) value) = (short) argv.ll;
515 else
516 *((ushort *) value) = (ushort) argv.lu;
517 } else if (size == sizeof(char)) {
518 if (fmt == 'd' || fmt == 'i')
519 *((char *) value) = (char) argv.ll;
520 else
521 *((uchar *) value) = (uchar) argv.lu;
522 } else {
523 if (fmt == 'd' || fmt == 'i')
524 *((int *) value) = (int) argv.ll;
525 else
526 *((unsigned*)value) = (unsigned)argv.lu;
527 }
528 }
529 } else if (fmt == 's' || fmt == 'c' || fmt == '[') {
530 if (size < 0)
531 size = MAXWIDTH;
532 if (value) {
533 argv.s = value;
534 if (fmt != 'c')
535 size -= 1;
536 } else
537 size = 0;
538
539 n = 0;
540 if (fmt == 's') {
541 do {
542 if (gv_isspace(inp))
543 break;
544 if ((n += 1) <= size)
545 *argv.s++ = inp;
546 } while (--width > 0 && SFGETC(f, inp) >= 0);
547 } else if (fmt == 'c') {
548 do {
549 if ((n += 1) <= size)
550 *argv.s++ = inp;
551 } while (--width > 0 && SFGETC(f, inp) >= 0);
552 } else { /* if(fmt == '[') */
553 bool accepted[UCHAR_MAX + 1];
554 form = (const char*)setclass((const unsigned char*)form, accepted);
555 do {
556 if (!accepted[inp]) {
557 if (n > 0 || (flags & SFFMT_ALTER))
558 break;
559 else {
560 SFUNGETC(f, inp);
561 goto done;
562 }
563 }
564 if ((n += 1) <= size)
565 *argv.s++ = inp;
566 } while (--width > 0 && SFGETC(f, inp) >= 0);
567 }
568
569 if (value && (n > 0 || fmt == '[')) {
570 n_assign += 1;
571 if (fmt != 'c' && size >= 0)
572 *argv.s = '\0';
573 }
574 }
575
576 if (width > 0 && inp >= 0)
577 SFUNGETC(f, inp);
578 }
579
580 done:
581
582 if (n_assign == 0 && inp < 0)
583 n_assign = -1;
584
585 return n_assign;
586}
static int flags
Definition gc.c:61
#define dot(v, w)
Definition geom.c:191
node NULL
Definition grammar.y:181
replacements for ctype.h functions
static bool gv_isdigit(int c)
Definition gv_ctype.h:41
static bool gv_isspace(int c)
Definition gv_ctype.h:55
#define RIGHTP
Definition sfhdr.h:83
#define FMTCMP(sz, type, maxtype)
Definition sfhdr.h:96
#define LEFTP
Definition sfhdr.h:82
#define FMTGET(ft, frm, fv, sz, flgs, wid, pr, bs)
Definition sfhdr.h:92
#define _Sfcv36
Definition sfhdr.h:131
#define SF_MAXDIGITS
Definition sfhdr.h:122
#define _Sfcv64
Definition sfhdr.h:132
#define _Sftype
Definition sfhdr.h:133
#define ushort
Definition sfhdr.h:47
#define ulong
Definition sfhdr.h:44
#define _Sffmtintf
Definition sfhdr.h:130
#define SF_RADIX
Definition sfhdr.h:116
#define SFFMT_TYPES
Definition sfhdr.h:104
#define uchar
Definition sfhdr.h:41
#define FMTSET(ft, frm, fv, sz, flgs, wid, pr, bs, ts, ns)
Definition sfhdr.h:86
#define SFFMT_MINUS
Definition sfhdr.h:102
#define SFFMT_INT
Definition sfhdr.h:109
#define SFFMT_UINT
Definition sfhdr.h:110
#define SFFMT_FLOAT
Definition sfhdr.h:111
#define SFFMT_ZFLAG
Definition sfio.h:49
#define SFFMT_ALTER
Definition sfio.h:55
#define SFFMT_LONG
Definition sfio.h:59
#define SFFMT_JFLAG
Definition sfio.h:65
#define SFFMT_SSHORT
Definition sfio.h:47
#define SFFMT_SHORT
Definition sfio.h:58
#define SFFMT_IFLAG
Definition sfio.h:64
#define SFFMT_LDOUBLE
Definition sfio.h:61
#define SFFMT_VALUE
Definition sfio.h:62
#define SFFMT_TFLAG
Definition sfio.h:48
#define SFFMT_SKIP
Definition sfio.h:57
#define SFFMT_LLONG
Definition sfio.h:60
static const unsigned char * setclass(const unsigned char *form, bool *accept)
Definition sfvscanf.c:31
#define SFUNGETC(f, c)
#define MAXWIDTH
Definition sfvscanf.c:25
int sfvscanf(FILE *f, Sffmt_t *ft)
Definition sfvscanf.c:68
#define SFGETC(f, c)
char * form
Definition sfio.h:34
int flags
Definition sfio.h:38
ssize_t size
Definition sfio.h:37
Sffmtext_f extf
Definition sfio.h:32
Definition sfhdr.h:65
unsigned long long lu
Definition sfhdr.h:73
Sffmt_t * ft
Definition sfhdr.h:79
void * vp
Definition sfhdr.h:78
long long ll
Definition sfhdr.h:72
double d
Definition sfhdr.h:75
char * s
Definition sfhdr.h:77
int i
Definition sfhdr.h:66