| 1 | #include <stdio.h> |
| 2 | #include <stdlib.h> |
| 3 | #include <stdarg.h> |
| 4 | #include <ctype.h> |
| 5 | #include <wchar.h> |
| 6 | #include <wctype.h> |
| 7 | #include <limits.h> |
| 8 | #include <string.h> |
| 9 | |
| 10 | #include "stdio_impl.h" |
| 11 | #include "shgetc.h" |
| 12 | #include "intscan.h" |
| 13 | #include "floatscan.h" |
| 14 | |
| 15 | #define SIZE_hh -2 |
| 16 | #define SIZE_h -1 |
| 17 | #define SIZE_def 0 |
| 18 | #define SIZE_l 1 |
| 19 | #define SIZE_L 2 |
| 20 | #define SIZE_ll 3 |
| 21 | |
| 22 | static void store_int(void *dest, int size, unsigned long long i) |
| 23 | { |
| 24 | 	if (!dest) return; |
| 25 | 	switch (size) { |
| 26 | 	case SIZE_hh: |
| 27 | 		*(char *)dest = i; |
| 28 | 		break; |
| 29 | 	case SIZE_h: |
| 30 | 		*(short *)dest = i; |
| 31 | 		break; |
| 32 | 	case SIZE_def: |
| 33 | 		*(int *)dest = i; |
| 34 | 		break; |
| 35 | 	case SIZE_l: |
| 36 | 		*(long *)dest = i; |
| 37 | 		break; |
| 38 | 	case SIZE_ll: |
| 39 | 		*(long long *)dest = i; |
| 40 | 		break; |
| 41 | 	} |
| 42 | } |
| 43 | |
| 44 | static void *arg_n(va_list ap, unsigned int n) |
| 45 | { |
| 46 | 	void *p; |
| 47 | 	unsigned int i; |
| 48 | 	va_list ap2; |
| 49 | 	va_copy(ap2, ap); |
| 50 | 	for (i=n; i>1; i--) va_arg(ap2, void *); |
| 51 | 	p = va_arg(ap2, void *); |
| 52 | 	va_end(ap2); |
| 53 | 	return p; |
| 54 | } |
| 55 | |
| 56 | static int in_set(const wchar_t *set, int c) |
| 57 | { |
| 58 | 	int j; |
| 59 | 	const wchar_t *p = set; |
| 60 | 	if (*p == '-') { |
| 61 | 		if (c=='-') return 1; |
| 62 | 		p++; |
| 63 | 	} else if (*p == ']') { |
| 64 | 		if (c==']') return 1; |
| 65 | 		p++; |
| 66 | 	} |
| 67 | 	for (; *p && *p != ']'; p++) { |
| 68 | 		if (*p=='-' && p[1] && p[1] != ']') |
| 69 | 			for (j=p++[-1]; j<*p; j++) |
| 70 | 				if (c==j) return 1; |
| 71 | 		if (c==*p) return 1; |
| 72 | 	} |
| 73 | 	return 0; |
| 74 | } |
| 75 | |
| 76 | #if 1 |
| 77 | #undef getwc |
| 78 | #define getwc(f) \ |
| 79 | 	((f)->rpos != (f)->rend && *(f)->rpos < 128 ? *(f)->rpos++ : (getwc)(f)) |
| 80 | |
| 81 | #undef ungetwc |
| 82 | #define ungetwc(c,f) \ |
| 83 | 	((f)->rend && (c)<128U ? *--(f)->rpos : ungetwc((c),(f))) |
| 84 | #endif |
| 85 | |
| 86 | int vfwscanf(FILE *restrict f, const wchar_t *restrict fmt, va_list ap) |
| 87 | { |
| 88 | 	int width; |
| 89 | 	int size; |
| 90 | 	int alloc; |
| 91 | 	const wchar_t *p; |
| 92 | 	int c, t; |
| 93 | 	char *s; |
| 94 | 	wchar_t *wcs; |
| 95 | 	void *dest=NULL; |
| 96 | 	int invert; |
| 97 | 	int matches=0; |
| 98 | 	off_t pos = 0, cnt; |
| 99 | 	static const char size_pfx[][3] = { "hh", "h", "", "l", "L", "ll" }; |
| 100 | 	char tmp[3*sizeof(int)+10]; |
| 101 | 	const wchar_t *set; |
| 102 | 	size_t i, k; |
| 103 | |
| 104 | 	FLOCK(f); |
| 105 | |
| 106 | 	fwide(f, 1); |
| 107 | |
| 108 | 	for (p=fmt; *p; p++) { |
| 109 | |
| 110 | 		alloc = 0; |
| 111 | |
| 112 | 		if (iswspace(*p)) { |
| 113 | 			while (iswspace(p[1])) p++; |
| 114 | 			while (iswspace((c=getwc(f)))) pos++; |
| 115 | 			ungetwc(c, f); |
| 116 | 			continue; |
| 117 | 		} |
| 118 | 		if (*p != '%' || p[1] == '%') { |
| 119 | 			if (*p == '%') { |
| 120 | 				p++; |
| 121 | 				while (iswspace((c=getwc(f)))) pos++; |
| 122 | 			} else { |
| 123 | 				c = getwc(f); |
| 124 | 			} |
| 125 | 			if (c!=*p) { |
| 126 | 				ungetwc(c, f); |
| 127 | 				if (c<0) goto input_fail; |
| 128 | 				goto match_fail; |
| 129 | 			} |
| 130 | 			pos++; |
| 131 | 			continue; |
| 132 | 		} |
| 133 | |
| 134 | 		p++; |
| 135 | 		if (*p=='*') { |
| 136 | 			dest = 0; p++; |
| 137 | 		} else if (iswdigit(*p) && p[1]=='$') { |
| 138 | 			dest = arg_n(ap, *p-'0'); p+=2; |
| 139 | 		} else { |
| 140 | 			dest = va_arg(ap, void *); |
| 141 | 		} |
| 142 | |
| 143 | 		for (width=0; iswdigit(*p); p++) { |
| 144 | 			width = 10*width + *p - '0'; |
| 145 | 		} |
| 146 | |
| 147 | 		if (*p=='m') { |
| 148 | 			wcs = 0; |
| 149 | 			s = 0; |
| 150 | 			alloc = !!dest; |
| 151 | 			p++; |
| 152 | 		} else { |
| 153 | 			alloc = 0; |
| 154 | 		} |
| 155 | |
| 156 | 		size = SIZE_def; |
| 157 | 		switch (*p++) { |
| 158 | 		case 'h': |
| 159 | 			if (*p == 'h') p++, size = SIZE_hh; |
| 160 | 			else size = SIZE_h; |
| 161 | 			break; |
| 162 | 		case 'l': |
| 163 | 			if (*p == 'l') p++, size = SIZE_ll; |
| 164 | 			else size = SIZE_l; |
| 165 | 			break; |
| 166 | 		case 'j': |
| 167 | 			size = SIZE_ll; |
| 168 | 			break; |
| 169 | 		case 'z': |
| 170 | 		case 't': |
| 171 | 			size = SIZE_l; |
| 172 | 			break; |
| 173 | 		case 'L': |
| 174 | 			size = SIZE_L; |
| 175 | 			break; |
| 176 | 		case 'd': case 'i': case 'o': case 'u': case 'x': |
| 177 | 		case 'a': case 'e': case 'f': case 'g': |
| 178 | 		case 'A': case 'E': case 'F': case 'G': case 'X': |
| 179 | 		case 's': case 'c': case '[': |
| 180 | 		case 'S': case 'C': |
| 181 | 		case 'p': case 'n': |
| 182 | 			p--; |
| 183 | 			break; |
| 184 | 		default: |
| 185 | 			goto fmt_fail; |
| 186 | 		} |
| 187 | |
| 188 | 		t = *p; |
| 189 | |
| 190 | 		/* Transform S,C -> ls,lc */ |
| 191 | 		if ((t&0x2f)==3) { |
| 192 | 			size = SIZE_l; |
| 193 | 			t |= 32; |
| 194 | 		} |
| 195 | |
| 196 | 		if (t != 'n') { |
| 197 | 			if (t != '[' && (t|32) != 'c') |
| 198 | 				while (iswspace((c=getwc(f)))) pos++; |
| 199 | 			else |
| 200 | 				c=getwc(f); |
| 201 | 			if (c < 0) goto input_fail; |
| 202 | 			ungetwc(c, f); |
| 203 | 		} |
| 204 | |
| 205 | 		switch (t) { |
| 206 | 		case 'n': |
| 207 | 			store_int(dest, size, pos); |
| 208 | 			/* do not increment match count, etc! */ |
| 209 | 			continue; |
| 210 | |
| 211 | 		case 's': |
| 212 | 		case 'c': |
| 213 | 		case '[': |
| 214 | 			if (t == 'c') { |
| 215 | 				if (width<1) width = 1; |
| 216 | 				invert = 1; |
| 217 | 				set = L""; |
| 218 | 			} else if (t == 's') { |
| 219 | 				invert = 1; |
| 220 | 				static const wchar_t spaces[] = { |
| 221 | 					' ', '\t', '\n', '\r', 11, 12, 0x0085, |
| 222 | 					0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, |
| 223 | 					0x2006, 0x2008, 0x2009, 0x200a, |
| 224 | 					0x2028, 0x2029, 0x205f, 0x3000, 0 }; |
| 225 | 				set = spaces; |
| 226 | 			} else { |
| 227 | 				if (*++p == '^') p++, invert = 1; |
| 228 | 				else invert = 0; |
| 229 | 				set = p; |
| 230 | 				if (*p==']') p++; |
| 231 | 				while (*p!=']') { |
| 232 | 					if (!*p) goto fmt_fail; |
| 233 | 					p++; |
| 234 | 				} |
| 235 | 			} |
| 236 | |
| 237 | 			s = (size == SIZE_def) ? dest : 0; |
| 238 | 			wcs = (size == SIZE_l) ? dest : 0; |
| 239 | |
| 240 | 			int gotmatch = 0; |
| 241 | |
| 242 | 			if (width < 1) width = -1; |
| 243 | |
| 244 | 			i = 0; |
| 245 | 			if (alloc) { |
| 246 | 				k = t=='c' ? width+1U : 31; |
| 247 | 				if (size == SIZE_l) { |
| 248 | 					wcs = malloc(k*sizeof(wchar_t)); |
| 249 | 					if (!wcs) goto alloc_fail; |
| 250 | 				} else { |
| 251 | 					s = malloc(k); |
| 252 | 					if (!s) goto alloc_fail; |
| 253 | 				} |
| 254 | 			} |
| 255 | 			while (width) { |
| 256 | 				if ((c=getwc(f))<0) break; |
| 257 | 				if (in_set(set, c) == invert) |
| 258 | 					break; |
| 259 | 				if (wcs) { |
| 260 | 					wcs[i++] = c; |
| 261 | 					if (alloc && i==k) { |
| 262 | 						k += k+1; |
| 263 | 						wchar_t *tmp = realloc(wcs, k*sizeof(wchar_t)); |
| 264 | 						if (!tmp) goto alloc_fail; |
| 265 | 						wcs = tmp; |
| 266 | 					} |
| 267 | 				} else if (size != SIZE_l) { |
| 268 | 					int l = wctomb(s?s+i:tmp, c); |
| 269 | 					if (l<0) goto input_fail; |
| 270 | 					i += l; |
| 271 | 					if (alloc && i > k-4) { |
| 272 | 						k += k+1; |
| 273 | 						char *tmp = realloc(s, k); |
| 274 | 						if (!tmp) goto alloc_fail; |
| 275 | 						s = tmp; |
| 276 | 					} |
| 277 | 				} |
| 278 | 				pos++; |
| 279 | 				width-=(width>0); |
| 280 | 				gotmatch=1; |
| 281 | 			} |
| 282 | 			if (width) { |
| 283 | 				ungetwc(c, f); |
| 284 | 				if (t == 'c' || !gotmatch) goto match_fail; |
| 285 | 			} |
| 286 | |
| 287 | 			if (alloc) { |
| 288 | 				if (size == SIZE_l) *(wchar_t **)dest = wcs; |
| 289 | 				else *(char **)dest = s; |
| 290 | 			} |
| 291 | 			if (t != 'c') { |
| 292 | 				if (wcs) wcs[i] = 0; |
| 293 | 				if (s) s[i] = 0; |
| 294 | 			} |
| 295 | 			break; |
| 296 | |
| 297 | 		case 'd': case 'i': case 'o': case 'u': case 'x': |
| 298 | 		case 'a': case 'e': case 'f': case 'g': |
| 299 | 		case 'A': case 'E': case 'F': case 'G': case 'X': |
| 300 | 		case 'p': |
| 301 | 			if (width < 1) width = 0; |
| 302 | 			snprintf(tmp, sizeof tmp, "%.*s%.0d%s%c%%lln", |
| 303 | 				1+!dest, "%*", width, size_pfx[size+2], t); |
| 304 | 			cnt = 0; |
| 305 | 			if (fscanf(f, tmp, dest?dest:&cnt, &cnt) == -1) |
| 306 | 				goto input_fail; |
| 307 | 			else if (!cnt) |
| 308 | 				goto match_fail; |
| 309 | 			pos += cnt; |
| 310 | 			break; |
| 311 | 		default: |
| 312 | 			goto fmt_fail; |
| 313 | 		} |
| 314 | |
| 315 | 		if (dest) matches++; |
| 316 | 	} |
| 317 | 	if (0) { |
| 318 | fmt_fail: |
| 319 | alloc_fail: |
| 320 | input_fail: |
| 321 | 		if (!matches) matches--; |
| 322 | match_fail: |
| 323 | 		if (alloc) { |
| 324 | 			free(s); |
| 325 | 			free(wcs); |
| 326 | 		} |
| 327 | 	} |
| 328 | 	FUNLOCK(f); |
| 329 | 	return matches; |
| 330 | } |
| 331 | |
| 332 | weak_alias(vfwscanf,__isoc99_vfwscanf); |