argpar/argpar.c

   1 /*
   2  * SPDX-License-Identifier: MIT
   3  *
   4  * Copyright (c) 2019-2021 Philippe Proulx <pproulx@efficios.com>
   5  * Copyright (c) 2020-2021 Simon Marchi <simon.marchi@efficios.com>
   6  */
   7
   8 #include <assert.h>
   9 #include <stdarg.h>
  10 #include <stdbool.h>
  11 #include <stdio.h>
  12 #include <stdlib.h>
  13 #include <string.h>
  14
  15 #include "argpar.h"
  16
  17 #define ARGPAR_REALLOC(_ptr, _type, _nmemb)                             \
  18         ((_type *) realloc(_ptr, (_nmemb) * sizeof(_type)))
  19
  20 #define ARGPAR_CALLOC(_type, _nmemb)                                    \
  21         ((_type *) calloc((_nmemb), sizeof(_type)))
  22
  23 #define ARGPAR_ZALLOC(_type) ARGPAR_CALLOC(_type, 1)
  24
  25 #define ARGPAR_ASSERT(_cond) assert(_cond)
  26
  27 #ifdef __MINGW_PRINTF_FORMAT
  28 # define ARGPAR_PRINTF_FORMAT __MINGW_PRINTF_FORMAT
  29 #else
  30 # define ARGPAR_PRINTF_FORMAT printf
  31 #endif
  32
  33 /*
  34  * An argpar iterator.
  35  *
  36  * Such a structure contains the state of an iterator between calls to
  37  * argpar_iter_next().
  38  */
  39 struct argpar_iter {
  40         /*
  41          * Data provided by the user to argpar_iter_create(); immutable
  42          * afterwards.
  43          */
  44         unsigned int argc;
  45         const char * const *argv;
  46         const struct argpar_opt_descr *descrs;
  47
  48         /*
  49          * Index of the argument to process in the next
  50          * argpar_iter_next() call.
  51          */
  52         unsigned int i;
  53
  54         /* Counter of non-option arguments */
  55         int non_opt_index;
  56
  57         /*
  58          * Current character of the current short option group: if it's
  59          * not `NULL`, the parser is in within a short option group,
  60          * therefore it must resume there in the next
  61          * argpar_iter_next() call.
  62          */
  63         const char *short_opt_ch;
  64
  65         /* Temporary character buffer which only grows */
  66         struct {
  67                 size_t size;
  68                 char *data;
  69         } tmp_buf;
  70 };
  71
  72 /* Base parsing item */
  73 struct argpar_item {
  74         enum argpar_item_type type;
  75 };
  76
  77 /* Option parsing item */
  78 struct argpar_item_opt {
  79         struct argpar_item base;
  80
  81         /* Corresponding descriptor */
  82         const struct argpar_opt_descr *descr;
  83
  84         /* Argument, or `NULL` if none; owned by this */
  85         char *arg;
  86 };
  87
  88 /* Non-option parsing item */
  89 struct argpar_item_non_opt {
  90         struct argpar_item base;
  91
  92         /*
  93          * Complete argument, pointing to one of the entries of the
  94          * original arguments (`argv`).
  95          */
  96         const char *arg;
  97
  98         /*
  99          * Index of this argument amongst all original arguments
 100          * (`argv`).
 101          */
 102         unsigned int orig_index;
 103
 104         /* Index of this argument amongst other non-option arguments */
 105         unsigned int non_opt_index;
 106 };
 107
 108 static __attribute__((format(ARGPAR_PRINTF_FORMAT, 1, 0)))
 109 char *argpar_vasprintf(const char * const fmt, va_list args)
 110 {
 111         int len1, len2;
 112         char *str;
 113         va_list args2;
 114
 115         va_copy(args2, args);
 116         len1 = vsnprintf(NULL, 0, fmt, args);
 117         if (len1 < 0) {
 118                 str = NULL;
 119                 goto end;
 120         }
 121
 122         str = malloc(len1 + 1);
 123         if (!str) {
 124                 goto end;
 125         }
 126
 127         len2 = vsnprintf(str, len1 + 1, fmt, args2);
 128         ARGPAR_ASSERT(len1 == len2);
 129
 130 end:
 131         va_end(args2);
 132         return str;
 133 }
 134
 135
 136 static __attribute__((format(ARGPAR_PRINTF_FORMAT, 1, 2)))
 137 char *argpar_asprintf(const char * const fmt, ...)
 138 {
 139         va_list args;
 140         char *str;
 141
 142         va_start(args, fmt);
 143         str = argpar_vasprintf(fmt, args);
 144         va_end(args);
 145         return str;
 146 }
 147
 148 static __attribute__((format(ARGPAR_PRINTF_FORMAT, 2, 3)))
 149 bool try_append_string_printf(char ** const str, const char *fmt, ...)
 150 {
 151         char *new_str = NULL;
 152         char *addendum = NULL;
 153         bool success;
 154         va_list args;
 155
 156         if (!str) {
 157                 success = true;
 158                 goto end;
 159         }
 160
 161         ARGPAR_ASSERT(str);
 162         va_start(args, fmt);
 163         addendum = argpar_vasprintf(fmt, args);
 164         va_end(args);
 165
 166         if (!addendum) {
 167                 success = false;
 168                 goto end;
 169         }
 170
 171         new_str = argpar_asprintf("%s%s", *str ? *str : "", addendum);
 172         if (!new_str) {
 173                 success = false;
 174                 goto end;
 175         }
 176
 177         free(*str);
 178         *str = new_str;
 179         success = true;
 180
 181 end:
 182         free(addendum);
 183         return success;
 184 }
 185
 186 ARGPAR_HIDDEN
 187 enum argpar_item_type argpar_item_type(const struct argpar_item * const item)
 188 {
 189         ARGPAR_ASSERT(item);
 190         return item->type;
 191 }
 192
 193 ARGPAR_HIDDEN
 194 const struct argpar_opt_descr *argpar_item_opt_descr(
 195                 const struct argpar_item * const item)
 196 {
 197         ARGPAR_ASSERT(item);
 198         ARGPAR_ASSERT(item->type == ARGPAR_ITEM_TYPE_OPT);
 199         return ((const struct argpar_item_opt *) item)->descr;
 200 }
 201
 202 ARGPAR_HIDDEN
 203 const char *argpar_item_opt_arg(const struct argpar_item * const item)
 204 {
 205         ARGPAR_ASSERT(item);
 206         ARGPAR_ASSERT(item->type == ARGPAR_ITEM_TYPE_OPT);
 207         return ((const struct argpar_item_opt *) item)->arg;
 208 }
 209
 210 ARGPAR_HIDDEN
 211 const char *argpar_item_non_opt_arg(const struct argpar_item * const item)
 212 {
 213         ARGPAR_ASSERT(item);
 214         ARGPAR_ASSERT(item->type == ARGPAR_ITEM_TYPE_NON_OPT);
 215         return ((const struct argpar_item_non_opt *) item)->arg;
 216 }
 217
 218 ARGPAR_HIDDEN
 219 unsigned int argpar_item_non_opt_orig_index(
 220                 const struct argpar_item * const item)
 221 {
 222         ARGPAR_ASSERT(item);
 223         ARGPAR_ASSERT(item->type == ARGPAR_ITEM_TYPE_NON_OPT);
 224         return ((const struct argpar_item_non_opt *) item)->orig_index;
 225 }
 226
 227 ARGPAR_HIDDEN
 228 unsigned int argpar_item_non_opt_non_opt_index(
 229                 const struct argpar_item * const item)
 230 {
 231         ARGPAR_ASSERT(item);
 232         ARGPAR_ASSERT(item->type == ARGPAR_ITEM_TYPE_NON_OPT);
 233         return ((const struct argpar_item_non_opt *) item)->non_opt_index;
 234 }
 235
 236 ARGPAR_HIDDEN
 237 void argpar_item_destroy(const struct argpar_item * const item)
 238 {
 239         if (!item) {
 240                 goto end;
 241         }
 242
 243         if (item->type == ARGPAR_ITEM_TYPE_OPT) {
 244                 struct argpar_item_opt * const opt_item =
 245                         (struct argpar_item_opt *) item;
 246
 247                 free(opt_item->arg);
 248         }
 249
 250         free((void *) item);
 251
 252 end:
 253         return;
 254 }
 255
 256 static
 257 struct argpar_item_opt *create_opt_item(
 258                 const struct argpar_opt_descr * const descr,
 259                 const char * const arg)
 260 {
 261         struct argpar_item_opt *opt_item =
 262                 ARGPAR_ZALLOC(struct argpar_item_opt);
 263
 264         if (!opt_item) {
 265                 goto end;
 266         }
 267
 268         opt_item->base.type = ARGPAR_ITEM_TYPE_OPT;
 269         opt_item->descr = descr;
 270
 271         if (arg) {
 272                 opt_item->arg = strdup(arg);
 273                 if (!opt_item->arg) {
 274                         goto error;
 275                 }
 276         }
 277
 278         goto end;
 279
 280 error:
 281         argpar_item_destroy(&opt_item->base);
 282         opt_item = NULL;
 283
 284 end:
 285         return opt_item;
 286 }
 287
 288 static
 289 struct argpar_item_non_opt *create_non_opt_item(const char * const arg,
 290                 const unsigned int orig_index,
 291                 const unsigned int non_opt_index)
 292 {
 293         struct argpar_item_non_opt * const non_opt_item =
 294                 ARGPAR_ZALLOC(struct argpar_item_non_opt);
 295
 296         if (!non_opt_item) {
 297                 goto end;
 298         }
 299
 300         non_opt_item->base.type = ARGPAR_ITEM_TYPE_NON_OPT;
 301         non_opt_item->arg = arg;
 302         non_opt_item->orig_index = orig_index;
 303         non_opt_item->non_opt_index = non_opt_index;
 304
 305 end:
 306         return non_opt_item;
 307 }
 308
 309 static
 310 const struct argpar_opt_descr *find_descr(
 311                 const struct argpar_opt_descr * const descrs,
 312                 const char short_name, const char * const long_name)
 313 {
 314         const struct argpar_opt_descr *descr;
 315
 316         for (descr = descrs; descr->short_name || descr->long_name; descr++) {
 317                 if (short_name && descr->short_name &&
 318                                 short_name == descr->short_name) {
 319                         goto end;
 320                 }
 321
 322                 if (long_name && descr->long_name &&
 323                                 strcmp(long_name, descr->long_name) == 0) {
 324                         goto end;
 325                 }
 326         }
 327
 328 end:
 329         return !descr->short_name && !descr->long_name ? NULL : descr;
 330 }
 331
 332 enum parse_orig_arg_opt_ret {
 333         PARSE_ORIG_ARG_OPT_RET_OK,
 334         PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT = -1,
 335         PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG = -2,
 336         PARSE_ORIG_ARG_OPT_RET_ERROR_UNEXPECTED_OPT_ARG = -4,
 337         PARSE_ORIG_ARG_OPT_RET_ERROR_MEMORY = -5,
 338 };
 339
 340 static
 341 enum parse_orig_arg_opt_ret parse_short_opts(const char * const short_opts,
 342                 const char * const next_orig_arg,
 343                 const struct argpar_opt_descr * const descrs,
 344                 struct argpar_iter * const iter,
 345                 char ** const error, struct argpar_item ** const item)
 346 {
 347         enum parse_orig_arg_opt_ret ret = PARSE_ORIG_ARG_OPT_RET_OK;
 348         bool used_next_orig_arg = false;
 349         const char *opt_arg = NULL;
 350         const struct argpar_opt_descr *descr;
 351         struct argpar_item_opt *opt_item;
 352
 353         ARGPAR_ASSERT(strlen(short_opts) != 0);
 354
 355         if (!iter->short_opt_ch) {
 356                 iter->short_opt_ch = short_opts;
 357         }
 358
 359         /* Find corresponding option descriptor */
 360         descr = find_descr(descrs, *iter->short_opt_ch, NULL);
 361         if (!descr) {
 362                 try_append_string_printf(error, "Unknown option `-%c`",
 363                         *iter->short_opt_ch);
 364                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT;
 365                 goto error;
 366         }
 367
 368         if (descr->with_arg) {
 369                 if (iter->short_opt_ch[1]) {
 370                         /* `-oarg` form */
 371                         opt_arg = &iter->short_opt_ch[1];
 372                 } else {
 373                         /* `-o arg` form */
 374                         opt_arg = next_orig_arg;
 375                         used_next_orig_arg = true;
 376                 }
 377
 378                 /*
 379                  * We accept `-o ''` (empty option argument), but not
 380                  * `-o` alone if an option argument is expected.
 381                  */
 382                 if (!opt_arg || (iter->short_opt_ch[1] &&
 383                                 strlen(opt_arg) == 0)) {
 384                         try_append_string_printf(error,
 385                                 "Missing required argument for option `-%c`",
 386                                 *iter->short_opt_ch);
 387                         used_next_orig_arg = false;
 388                         ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG;
 389                         goto error;
 390                 }
 391         }
 392
 393         /* Create and append option argument */
 394         opt_item = create_opt_item(descr, opt_arg);
 395         if (!opt_item) {
 396                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MEMORY;
 397                 goto error;
 398         }
 399
 400         *item = &opt_item->base;
 401         iter->short_opt_ch++;
 402
 403         if (descr->with_arg || !*iter->short_opt_ch) {
 404                 /* Option has an argument: no more options */
 405                 iter->short_opt_ch = NULL;
 406
 407                 if (used_next_orig_arg) {
 408                         iter->i += 2;
 409                 } else {
 410                         iter->i++;
 411                 }
 412         }
 413
 414         goto end;
 415
 416 error:
 417         ARGPAR_ASSERT(ret != PARSE_ORIG_ARG_OPT_RET_OK);
 418
 419 end:
 420         return ret;
 421 }
 422
 423 static
 424 enum parse_orig_arg_opt_ret parse_long_opt(const char * const long_opt_arg,
 425                 const char * const next_orig_arg,
 426                 const struct argpar_opt_descr * const descrs,
 427                 struct argpar_iter * const iter,
 428                 char ** const error, struct argpar_item ** const item)
 429 {
 430         enum parse_orig_arg_opt_ret ret = PARSE_ORIG_ARG_OPT_RET_OK;
 431         const struct argpar_opt_descr *descr;
 432         struct argpar_item_opt *opt_item;
 433         bool used_next_orig_arg = false;
 434
 435         /* Option's argument, if any */
 436         const char *opt_arg = NULL;
 437
 438         /* Position of first `=`, if any */
 439         const char *eq_pos;
 440
 441         /* Option name */
 442         const char *long_opt_name = long_opt_arg;
 443
 444         ARGPAR_ASSERT(strlen(long_opt_arg) != 0);
 445
 446         /* Find the first `=` in original argument */
 447         eq_pos = strchr(long_opt_arg, '=');
 448         if (eq_pos) {
 449                 const size_t long_opt_name_size = eq_pos - long_opt_arg;
 450
 451                 /* Isolate the option name */
 452                 while (long_opt_name_size > iter->tmp_buf.size - 1) {
 453                         iter->tmp_buf.size *= 2;
 454                         iter->tmp_buf.data = ARGPAR_REALLOC(iter->tmp_buf.data,
 455                                 char, iter->tmp_buf.size);
 456                         if (!iter->tmp_buf.data) {
 457                                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MEMORY;
 458                                 goto error;
 459                         }
 460                 }
 461
 462                 memcpy(iter->tmp_buf.data, long_opt_arg, long_opt_name_size);
 463                 iter->tmp_buf.data[long_opt_name_size] = '\0';
 464                 long_opt_name = iter->tmp_buf.data;
 465         }
 466
 467         /* Find corresponding option descriptor */
 468         descr = find_descr(descrs, '\0', long_opt_name);
 469         if (!descr) {
 470                 try_append_string_printf(error, "Unknown option `--%s`",
 471                         long_opt_name);
 472                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT;
 473                 goto error;
 474         }
 475
 476         /* Find option's argument if any */
 477         if (descr->with_arg) {
 478                 if (eq_pos) {
 479                         /* `--long-opt=arg` style */
 480                         opt_arg = eq_pos + 1;
 481                 } else {
 482                         /* `--long-opt arg` style */
 483                         if (!next_orig_arg) {
 484                                 try_append_string_printf(error,
 485                                         "Missing required argument for option `--%s`",
 486                                         long_opt_name);
 487                                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG;
 488                                 goto error;
 489                         }
 490
 491                         opt_arg = next_orig_arg;
 492                         used_next_orig_arg = true;
 493                 }
 494         } else if (eq_pos) {
 495                 /*
 496                  * Unexpected `--opt=arg` style for a long option which
 497                  * doesn't accept an argument.
 498                  */
 499                 try_append_string_printf(error,
 500                         "Unexpected argument for option `--%s`", long_opt_name);
 501                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_UNEXPECTED_OPT_ARG;
 502                 goto error;
 503         }
 504
 505         /* Create and append option argument */
 506         opt_item = create_opt_item(descr, opt_arg);
 507         if (!opt_item) {
 508                 goto error;
 509         }
 510
 511         if (used_next_orig_arg) {
 512                 iter->i += 2;
 513         } else {
 514                 iter->i++;
 515         }
 516
 517         *item = &opt_item->base;
 518         goto end;
 519
 520 error:
 521         ARGPAR_ASSERT(ret != PARSE_ORIG_ARG_OPT_RET_OK);
 522
 523 end:
 524         return ret;
 525 }
 526
 527 static
 528 enum parse_orig_arg_opt_ret parse_orig_arg_opt(const char * const orig_arg,
 529                 const char * const next_orig_arg,
 530                 const struct argpar_opt_descr * const descrs,
 531                 struct argpar_iter * const iter, char ** const error,
 532                 struct argpar_item ** const item)
 533 {
 534         enum parse_orig_arg_opt_ret ret = PARSE_ORIG_ARG_OPT_RET_OK;
 535
 536         ARGPAR_ASSERT(orig_arg[0] == '-');
 537
 538         if (orig_arg[1] == '-') {
 539                 /* Long option */
 540                 ret = parse_long_opt(&orig_arg[2],
 541                         next_orig_arg, descrs, iter, error, item);
 542         } else {
 543                 /* Short option */
 544                 ret = parse_short_opts(&orig_arg[1],
 545                         next_orig_arg, descrs, iter, error, item);
 546         }
 547
 548         return ret;
 549 }
 550
 551 static
 552 bool try_prepend_while_parsing_arg_to_error(char ** const error,
 553                 const unsigned int i, const char * const arg)
 554 {
 555         char *new_error;
 556         bool success;
 557
 558         if (!error) {
 559                 success = true;
 560                 goto end;
 561         }
 562
 563         ARGPAR_ASSERT(*error);
 564         new_error = argpar_asprintf("While parsing argument #%u (`%s`): %s",
 565                 i + 1, arg, *error);
 566         if (!new_error) {
 567                 success = false;
 568                 goto end;
 569         }
 570
 571         free(*error);
 572         *error = new_error;
 573         success = true;
 574
 575 end:
 576         return success;
 577 }
 578
 579 ARGPAR_HIDDEN
 580 struct argpar_iter *argpar_iter_create(const unsigned int argc,
 581                 const char * const * const argv,
 582                 const struct argpar_opt_descr * const descrs)
 583 {
 584         struct argpar_iter *iter = ARGPAR_ZALLOC(struct argpar_iter);
 585
 586         if (!iter) {
 587                 goto end;
 588         }
 589
 590         iter->argc = argc;
 591         iter->argv = argv;
 592         iter->descrs = descrs;
 593         iter->tmp_buf.size = 128;
 594         iter->tmp_buf.data = ARGPAR_CALLOC(char, iter->tmp_buf.size);
 595         if (!iter->tmp_buf.data) {
 596                 argpar_iter_destroy(iter);
 597                 iter = NULL;
 598                 goto end;
 599         }
 600
 601 end:
 602         return iter;
 603 }
 604
 605 ARGPAR_HIDDEN
 606 void argpar_iter_destroy(struct argpar_iter * const iter)
 607 {
 608         if (iter) {
 609                 free(iter->tmp_buf.data);
 610                 free(iter);
 611         }
 612 }
 613
 614 ARGPAR_HIDDEN
 615 enum argpar_iter_next_status argpar_iter_next(
 616                 struct argpar_iter * const iter,
 617                 const struct argpar_item ** const item, char ** const error)
 618 {
 619         enum argpar_iter_next_status status;
 620         enum parse_orig_arg_opt_ret parse_orig_arg_opt_ret;
 621         const char *orig_arg;
 622         const char *next_orig_arg;
 623
 624         ARGPAR_ASSERT(iter->i <= iter->argc);
 625
 626         if (error) {
 627                 *error = NULL;
 628         }
 629
 630         if (iter->i == iter->argc) {
 631                 status = ARGPAR_ITER_NEXT_STATUS_END;
 632                 goto end;
 633         }
 634
 635         orig_arg = iter->argv[iter->i];
 636         next_orig_arg =
 637                 iter->i < (iter->argc - 1) ? iter->argv[iter->i + 1] : NULL;
 638
 639         if (strcmp(orig_arg, "-") == 0 || strcmp(orig_arg, "--") == 0 ||
 640                         orig_arg[0] != '-') {
 641                 /* Non-option argument */
 642                 const struct argpar_item_non_opt * const non_opt_item =
 643                         create_non_opt_item(orig_arg, iter->i,
 644                                 iter->non_opt_index);
 645
 646                 if (!non_opt_item) {
 647                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_MEMORY;
 648                         goto end;
 649                 }
 650
 651                 iter->non_opt_index++;
 652                 iter->i++;
 653                 *item = &non_opt_item->base;
 654                 status = ARGPAR_ITER_NEXT_STATUS_OK;
 655                 goto end;
 656         }
 657
 658         /* Option argument */
 659         parse_orig_arg_opt_ret = parse_orig_arg_opt(orig_arg,
 660                 next_orig_arg, iter->descrs, iter, error,
 661                 (struct argpar_item **) item);
 662         switch (parse_orig_arg_opt_ret) {
 663         case PARSE_ORIG_ARG_OPT_RET_OK:
 664                 status = ARGPAR_ITER_NEXT_STATUS_OK;
 665                 break;
 666         case PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT:
 667         case PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG:
 668         case PARSE_ORIG_ARG_OPT_RET_ERROR_UNEXPECTED_OPT_ARG:
 669                 try_prepend_while_parsing_arg_to_error(error, iter->i,
 670                         orig_arg);
 671
 672                 switch (parse_orig_arg_opt_ret) {
 673                 case PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT:
 674                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_UNKNOWN_OPT;
 675                         break;
 676                 case PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG:
 677                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_MISSING_OPT_ARG;
 678                         break;
 679                 case PARSE_ORIG_ARG_OPT_RET_ERROR_UNEXPECTED_OPT_ARG:
 680                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_UNEXPECTED_OPT_ARG;
 681                         break;
 682                 default:
 683                         abort();
 684                 }
 685
 686                 break;
 687         case PARSE_ORIG_ARG_OPT_RET_ERROR_MEMORY:
 688                 status = ARGPAR_ITER_NEXT_STATUS_ERROR_MEMORY;
 689                 break;
 690         default:
 691                 abort();
 692         }
 693
 694 end:
 695         return status;
 696 }
 697
 698 ARGPAR_HIDDEN
 699 unsigned int argpar_iter_ingested_orig_args(
 700                 const struct argpar_iter * const iter)
 701 {
 702         return iter->i;
 703 }