| 1 | /* -*-c-*- |
| 2 | * |
| 3 | * `printf'-style formatting for dynamic strings |
| 4 | * |
| 5 | * (c) 1999 Straylight/Edgeware |
| 6 | */ |
| 7 | |
| 8 | /*----- Licensing notice --------------------------------------------------* |
| 9 | * |
| 10 | * This file is part of the mLib utilities library. |
| 11 | * |
| 12 | * mLib is free software; you can redistribute it and/or modify |
| 13 | * it under the terms of the GNU Library General Public License as |
| 14 | * published by the Free Software Foundation; either version 2 of the |
| 15 | * License, or (at your option) any later version. |
| 16 | * |
| 17 | * mLib is distributed in the hope that it will be useful, |
| 18 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 19 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 20 | * GNU Library General Public License for more details. |
| 21 | * |
| 22 | * You should have received a copy of the GNU Library General Public |
| 23 | * License along with mLib; if not, write to the Free |
| 24 | * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, |
| 25 | * MA 02111-1307, USA. |
| 26 | */ |
| 27 | |
| 28 | /*----- Header files ------------------------------------------------------*/ |
| 29 | |
| 30 | #include "config.h" |
| 31 | |
| 32 | #include <assert.h> |
| 33 | #include <ctype.h> |
| 34 | #include <math.h> |
| 35 | #include <stdarg.h> |
| 36 | #include <stdio.h> |
| 37 | #include <stdlib.h> |
| 38 | #include <string.h> |
| 39 | |
| 40 | #ifdef HAVE_FLOAT_H |
| 41 | # include <float.h> |
| 42 | #endif |
| 43 | |
| 44 | #include "darray.h" |
| 45 | #include "dstr.h" |
| 46 | |
| 47 | /*----- Tunable constants -------------------------------------------------*/ |
| 48 | |
| 49 | /* |
| 50 | * For each format specifier, at least @PUTFSTEP@ bytes are ensured before |
| 51 | * writing the formatted result. |
| 52 | */ |
| 53 | |
| 54 | #define PUTFSTEP 64 /* Buffer size for @putf@ */ |
| 55 | |
| 56 | /*----- Preliminary definitions -------------------------------------------*/ |
| 57 | |
| 58 | #define OUTPUT_FMTTYPES(_) \ |
| 59 | _(i, unsigned int) \ |
| 60 | _(li, unsigned long) \ |
| 61 | _(s, char *) \ |
| 62 | _(p, void *) \ |
| 63 | _(f, double) \ |
| 64 | _(Lf, long double) |
| 65 | |
| 66 | #define PERCENT_N_FMTTYPES(_) \ |
| 67 | _(hn, short *) \ |
| 68 | _(n, int *) \ |
| 69 | _(ln, long *) |
| 70 | |
| 71 | #define FMTTYPES(_) \ |
| 72 | OUTPUT_FMTTYPES(_) \ |
| 73 | PERCENT_N_FMTTYPES(_) |
| 74 | |
| 75 | enum { |
| 76 | fmt_unset = 0, |
| 77 | #define CODE(code, ty) fmt_##code, |
| 78 | FMTTYPES(CODE) |
| 79 | #undef CODE |
| 80 | fmt__limit |
| 81 | }; |
| 82 | |
| 83 | typedef struct { |
| 84 | int fmt; |
| 85 | union { |
| 86 | #define MEMB(code, ty) ty code; |
| 87 | FMTTYPES(MEMB) |
| 88 | #undef MEMB |
| 89 | } u; |
| 90 | } fmtarg; |
| 91 | |
| 92 | DA_DECL(fmtarg_v, fmtarg); |
| 93 | |
| 94 | enum { |
| 95 | len_std = 0, |
| 96 | len_h, |
| 97 | len_l, |
| 98 | len_ll, |
| 99 | len_L |
| 100 | }; |
| 101 | |
| 102 | #define f_len 0x000fu |
| 103 | #define f_wd 0x0010u |
| 104 | #define f_wdarg 0x0020u |
| 105 | #define f_prec 0x0040u |
| 106 | #define f_precarg 0x0080u |
| 107 | #define f_plus 0x0100u |
| 108 | #define f_minus 0x0200u |
| 109 | #define f_sharp 0x0400u |
| 110 | #define f_zero 0x0800u |
| 111 | #define f_posarg 0x1000u |
| 112 | |
| 113 | typedef struct { |
| 114 | const char *p; |
| 115 | size_t n; |
| 116 | unsigned f; |
| 117 | int fmt, ch; |
| 118 | int wd, prec; |
| 119 | int arg; |
| 120 | } fmtspec; |
| 121 | |
| 122 | DA_DECL(fmtspec_v, fmtspec); |
| 123 | |
| 124 | /*----- Main code ---------------------------------------------------------*/ |
| 125 | |
| 126 | /* --- @dstr_vputf@ --- * |
| 127 | * |
| 128 | * Arguments: @dstr *d@ = pointer to a dynamic string block |
| 129 | * @const char *p@ = pointer to @printf@-style format string |
| 130 | * @va_list *ap@ = argument handle |
| 131 | * |
| 132 | * Returns: The number of characters written to the string. |
| 133 | * |
| 134 | * Use: As for @dstr_putf@, but may be used as a back-end to user- |
| 135 | * supplied functions with @printf@-style interfaces. |
| 136 | */ |
| 137 | |
| 138 | static void set_arg(fmtarg_v *av, size_t i, int fmt) |
| 139 | { |
| 140 | size_t j, n; |
| 141 | |
| 142 | n = DA_LEN(av); |
| 143 | if (i >= n) { |
| 144 | DA_ENSURE(av, i + 1 - n); |
| 145 | for (j = n; j <= i; j++) DA(av)[j].fmt = fmt_unset; |
| 146 | DA_UNSAFE_EXTEND(av, i + 1 - n); |
| 147 | } |
| 148 | |
| 149 | if (DA(av)[i].fmt == fmt_unset) DA(av)[i].fmt = fmt; |
| 150 | else assert(DA(av)[i].fmt == fmt); |
| 151 | } |
| 152 | |
| 153 | int dstr_vputf(dstr *d, const char *p, va_list *ap) |
| 154 | { |
| 155 | size_t n = d->len; |
| 156 | size_t sz, mx; |
| 157 | dstr dd = DSTR_INIT; |
| 158 | fmtspec_v sv = DA_INIT; |
| 159 | fmtarg_v av = DA_INIT; |
| 160 | fmtarg *fa, *fal; |
| 161 | fmtspec *fs, *fsl; |
| 162 | unsigned f; |
| 163 | int i, anext; |
| 164 | int wd, prec; |
| 165 | |
| 166 | /* --- Initial pass through the input, parsing format specifiers --- * |
| 167 | * |
| 168 | * We essentially compile the format string into a vector of @fmtspec@ |
| 169 | * objects, each of which represnts a chunk of literal text followed by a |
| 170 | * (possibly imaginary, in the case of the final one) formatting directive. |
| 171 | * Output then simply consists of interpreting these specifiers in order. |
| 172 | */ |
| 173 | |
| 174 | anext = 0; |
| 175 | |
| 176 | while (*p) { |
| 177 | f = 0; |
| 178 | DA_ENSURE(&sv, 1); |
| 179 | fs = &DA(&sv)[DA_LEN(&sv)]; |
| 180 | DA_UNSAFE_EXTEND(&sv, 1); |
| 181 | |
| 182 | /* --- Find the end of this literal portion --- */ |
| 183 | |
| 184 | fs->p = p; |
| 185 | while (*p && *p != '%') p++; |
| 186 | fs->n = p - fs->p; |
| 187 | |
| 188 | /* --- Some simple cases --- * |
| 189 | * |
| 190 | * We might have reached the end of the string, or maybe a `%%' escape. |
| 191 | */ |
| 192 | |
| 193 | if (!*p) { fs->fmt = fmt_unset; fs->ch = 0; break; } |
| 194 | p++; |
| 195 | if (*p == '%') { fs->fmt = fmt_unset; fs->ch = '%'; p++; continue; } |
| 196 | |
| 197 | /* --- Pick up initial flags --- */ |
| 198 | |
| 199 | flags: |
| 200 | for (;;) { |
| 201 | switch (*p) { |
| 202 | case '+': f |= f_plus; break; |
| 203 | case '-': f |= f_minus; break; |
| 204 | case '#': f |= f_sharp; break; |
| 205 | case '0': f |= f_zero; break; |
| 206 | default: goto done_flags; |
| 207 | } |
| 208 | p++; |
| 209 | } |
| 210 | |
| 211 | /* --- Pick up the field width --- */ |
| 212 | |
| 213 | done_flags: |
| 214 | i = 0; |
| 215 | while (isdigit((unsigned char)*p)) i = 10*i + *p++ - '0'; |
| 216 | |
| 217 | /* --- Snag: this might have been an argument position indicator --- */ |
| 218 | |
| 219 | if (i && *p == '$' && (!f || f == f_zero)) { |
| 220 | f |= f_posarg; |
| 221 | fs->arg = i - 1; |
| 222 | p++; |
| 223 | goto flags; |
| 224 | } |
| 225 | |
| 226 | /* --- Set the field width --- * |
| 227 | * |
| 228 | * If @i@ is nonzero here then we have a numeric field width. Otherwise |
| 229 | * it might be `*', maybe with an explicit argument number. |
| 230 | */ |
| 231 | |
| 232 | if (i) { |
| 233 | f |= f_wd; |
| 234 | fs->wd = i; |
| 235 | } else if (*p == '*') { |
| 236 | p++; |
| 237 | if (!isdigit((unsigned char)*p)) |
| 238 | i = anext++; |
| 239 | else { |
| 240 | i = *p++ - '0'; |
| 241 | while (isdigit((unsigned char)*p)) i = 10*i + *p++ - '0'; |
| 242 | assert(*p == '$'); p++; |
| 243 | assert(i > 0); i--; |
| 244 | } |
| 245 | f |= f_wd | f_wdarg; |
| 246 | set_arg(&av, i, fmt_i); fs->wd = i; |
| 247 | } |
| 248 | |
| 249 | /* --- Maybe we have a precision spec --- */ |
| 250 | |
| 251 | if (*p == '.') { |
| 252 | p++; |
| 253 | f |= f_prec; |
| 254 | if (isdigit((unsigned char)*p)) { |
| 255 | i = *p++ - '0'; |
| 256 | while (isdigit((unsigned char)*p)) i = 10*i + *p++ - '0'; |
| 257 | fs->prec = i; |
| 258 | } else if (*p != '*') |
| 259 | fs->prec = 0; |
| 260 | else { |
| 261 | p++; |
| 262 | if (!isdigit((unsigned char)*p)) |
| 263 | i = anext++; |
| 264 | else { |
| 265 | i = *p++ - '0'; |
| 266 | while (isdigit((unsigned char)*p)) i = 10*i + *p++ - '0'; |
| 267 | assert(*p == '$'); p++; |
| 268 | assert(i > 0); i--; |
| 269 | } |
| 270 | f |= f_precarg; |
| 271 | set_arg(&av, i, fmt_i); fs->prec = i; |
| 272 | } |
| 273 | } |
| 274 | |
| 275 | /* --- Maybe some length flags --- */ |
| 276 | |
| 277 | switch (*p) { |
| 278 | case 'h': f |= len_h; p++; break; |
| 279 | case 'l': f |= len_l; p++; break; |
| 280 | case 'L': f |= len_L; p++; break; |
| 281 | } |
| 282 | |
| 283 | /* --- The flags are now ready --- */ |
| 284 | |
| 285 | fs->f = f; |
| 286 | |
| 287 | /* --- At the end, an actual directive --- */ |
| 288 | |
| 289 | fs->ch = *p; |
| 290 | switch (*p++) { |
| 291 | case '%': |
| 292 | fs->fmt = fmt_unset; |
| 293 | break; |
| 294 | case 'd': case 'i': case 'x': case 'X': case 'o': case 'u': |
| 295 | switch (f & f_len) { |
| 296 | case len_l: fs->fmt = fmt_li; break; |
| 297 | default: fs->fmt = fmt_i; |
| 298 | } |
| 299 | break; |
| 300 | case 'e': case 'E': case 'f': case 'F': case 'g': case 'G': |
| 301 | fs->fmt = (f & f_len) == len_L ? fmt_Lf : fmt_f; |
| 302 | break; |
| 303 | case 'c': |
| 304 | fs->fmt = fmt_i; |
| 305 | break; |
| 306 | case 's': |
| 307 | fs->fmt = fmt_s; |
| 308 | break; |
| 309 | case 'p': |
| 310 | fs->fmt = fmt_p; |
| 311 | break; |
| 312 | case 'n': |
| 313 | switch (f & f_len) { |
| 314 | case len_h: fs->fmt = fmt_hn; break; |
| 315 | case len_l: fs->fmt = fmt_ln; break; |
| 316 | default: fs->fmt = fmt_n; |
| 317 | } |
| 318 | break; |
| 319 | default: |
| 320 | fprintf(stderr, |
| 321 | "FATAL dstr_vputf: unknown format specifier `%c'\n", p[-1]); |
| 322 | abort(); |
| 323 | } |
| 324 | |
| 325 | /* --- Finally sort out the argument --- * |
| 326 | * |
| 327 | * If we don't have explicit argument positions then this comes after the |
| 328 | * width and precision; and we don't know the type code until we've |
| 329 | * parsed the specifier, so this seems the right place to handle it. |
| 330 | */ |
| 331 | |
| 332 | if (!(f & f_posarg)) fs->arg = anext++; |
| 333 | set_arg(&av, fs->arg, fs->fmt); |
| 334 | } |
| 335 | |
| 336 | /* --- Quick pass over the argument vector to collect the arguments --- */ |
| 337 | |
| 338 | for (fa = DA(&av), fal = fa + DA_LEN(&av); fa < fal; fa++) { |
| 339 | switch (fa->fmt) { |
| 340 | #define CASE(code, ty) case fmt_##code: fa->u.code = va_arg(*ap, ty); break; |
| 341 | FMTTYPES(CASE) |
| 342 | #undef CASE |
| 343 | default: abort(); |
| 344 | } |
| 345 | } |
| 346 | |
| 347 | /* --- Final pass through the format string to produce output --- */ |
| 348 | |
| 349 | fa = DA(&av); |
| 350 | for (fs = DA(&sv), fsl = fs + DA_LEN(&sv); fs < fsl; fs++) { |
| 351 | f = fs->f; |
| 352 | |
| 353 | /* --- Output the literal portion --- */ |
| 354 | |
| 355 | if (fs->n) DPUTM(d, fs->p, fs->n); |
| 356 | |
| 357 | /* --- And now the variable portion --- */ |
| 358 | |
| 359 | if (fs->fmt == fmt_unset) { |
| 360 | switch (fs->ch) { |
| 361 | case 0: break; |
| 362 | case '%': DPUTC(d, '%'); break; |
| 363 | default: abort(); |
| 364 | } |
| 365 | continue; |
| 366 | } |
| 367 | |
| 368 | DRESET(&dd); |
| 369 | DPUTC(&dd, '%'); |
| 370 | |
| 371 | /* --- Resolve the width and precision --- */ |
| 372 | |
| 373 | if (!(f & f_wd)) |
| 374 | wd = 0; |
| 375 | else { |
| 376 | wd = (fs->f & f_wdarg) ? *(int *)&fa[fs->wd].u.i : fs->wd; |
| 377 | if (wd < 0) { wd = -wd; f |= f_minus; } |
| 378 | } |
| 379 | |
| 380 | if (!(f & f_prec)) |
| 381 | prec = 0; |
| 382 | else { |
| 383 | prec = (fs->f & f_precarg) ? *(int *)&fa[fs->prec].u.i : fs->prec; |
| 384 | if (prec < 0) { prec = 0; f &= ~f_prec; } |
| 385 | } |
| 386 | |
| 387 | /* --- Write out the flags, width and precision --- */ |
| 388 | |
| 389 | if (f & f_plus) DPUTC(&dd, '+'); |
| 390 | if (f & f_minus) DPUTC(&dd, '-'); |
| 391 | if (f & f_sharp) DPUTC(&dd, '#'); |
| 392 | if (f & f_zero) DPUTC(&dd, '0'); |
| 393 | |
| 394 | if (f & f_wd) { |
| 395 | DENSURE(&dd, PUTFSTEP); |
| 396 | dd.len += sprintf(dd.buf + dd.len, "%d", wd); |
| 397 | } |
| 398 | |
| 399 | if (f & f_prec) { |
| 400 | DENSURE(&dd, PUTFSTEP + 1); |
| 401 | dd.len += sprintf(dd.buf + dd.len, ".%d", prec); |
| 402 | } |
| 403 | |
| 404 | /* --- Write out the length gadget --- */ |
| 405 | |
| 406 | switch (f & f_len) { |
| 407 | case len_h: DPUTC(&dd, 'h'); break; |
| 408 | case len_l: DPUTC(&dd, 'l'); break; |
| 409 | case len_L: DPUTC(&dd, 'L'); break; |
| 410 | case len_std: break; |
| 411 | default: abort(); |
| 412 | } |
| 413 | |
| 414 | /* --- And finally the actually important bit --- */ |
| 415 | |
| 416 | DPUTC(&dd, fs->ch); |
| 417 | DPUTZ(&dd); |
| 418 | |
| 419 | /* --- Make sure we have enough space for the output --- */ |
| 420 | |
| 421 | sz = PUTFSTEP; |
| 422 | if (sz < wd) sz = wd; |
| 423 | if (sz < prec + 16) sz = prec + 16; |
| 424 | switch (fs->ch) { |
| 425 | case 'a': case 'A': |
| 426 | case 'e': case 'E': case 'f': case 'F': case 'g': case 'G': |
| 427 | #ifdef HAVE_FLOAT_H |
| 428 | if (fs->ch == 'f') { |
| 429 | mx = ((fs->f & f_len) == len_L ? |
| 430 | LDBL_MAX_10_EXP : DBL_MAX_10_EXP) + 16; |
| 431 | if (sz < mx) sz = mx; |
| 432 | } |
| 433 | break; |
| 434 | #else |
| 435 | DPUTS(d, "<no float support>"); |
| 436 | continue; |
| 437 | #endif |
| 438 | case 's': |
| 439 | if (!(f & f_prec)) { |
| 440 | n = strlen(fa[fs->arg].u.s); |
| 441 | if (sz < n) sz = n; |
| 442 | } |
| 443 | break; |
| 444 | case 'n': |
| 445 | switch (fs->fmt) { |
| 446 | #define CASE(code, ty) \ |
| 447 | case fmt_##code: *fa[fs->arg].u.code = d->len - n; break; |
| 448 | PERCENT_N_FMTTYPES(CASE) |
| 449 | #undef CASE |
| 450 | default: abort(); |
| 451 | } |
| 452 | continue; |
| 453 | } |
| 454 | |
| 455 | /* --- Finally do the output stage --- */ |
| 456 | |
| 457 | DENSURE(d, sz + 1); |
| 458 | switch (fs->fmt) { |
| 459 | #ifdef HAVE_SNPRINTF |
| 460 | # define CASE(code, ty) case fmt_##code: \ |
| 461 | i = snprintf(d->buf + d->len, sz + 1, dd.buf, fa[fs->arg].u.code); \ |
| 462 | break; |
| 463 | #else |
| 464 | # define CASE(code, ty) case fmt_##code: \ |
| 465 | i = sprintf(d->buf + d->len, dd.buf, fa[fs->arg].u.code); \ |
| 466 | break; |
| 467 | #endif |
| 468 | OUTPUT_FMTTYPES(CASE) |
| 469 | #undef CASE |
| 470 | default: abort(); |
| 471 | } |
| 472 | assert(0 <= i && i <= sz); d->len += i; |
| 473 | } |
| 474 | |
| 475 | /* --- We're done --- */ |
| 476 | |
| 477 | DPUTZ(d); |
| 478 | DDESTROY(&dd); |
| 479 | return (d->len - n); |
| 480 | } |
| 481 | |
| 482 | /* --- @dstr_putf@ --- * |
| 483 | * |
| 484 | * Arguments: @dstr *d@ = pointer to a dynamic string block |
| 485 | * @const char *p@ = pointer to @printf@-style format string |
| 486 | * @...@ = argument handle |
| 487 | * |
| 488 | * Returns: The number of characters written to the string. |
| 489 | * |
| 490 | * Use: Writes a piece of text to a dynamic string, doing @printf@- |
| 491 | * style substitutions as it goes. Intended to be robust if |
| 492 | * faced with malicious arguments, but not if the format string |
| 493 | * itself is malicious. |
| 494 | */ |
| 495 | |
| 496 | int dstr_putf(dstr *d, const char *p, ...) |
| 497 | { |
| 498 | int n; |
| 499 | va_list ap; |
| 500 | va_start(ap, p); |
| 501 | n = dstr_vputf(d, p, &ap); |
| 502 | va_end(ap); |
| 503 | return (n); |
| 504 | } |
| 505 | |
| 506 | /*----- That's all, folks -------------------------------------------------*/ |