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 within the current short option group: if
  59          * it's not `NULL`, the parser is within a short option group,
  60          * therefore it must resume there in the next argpar_iter_next()
  61          * call.
  62          */
  63         const char *short_opt_group_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_opt_group(
 342                 const char * const short_opt_group,
 343                 const char * const next_orig_arg,
 344                 const struct argpar_opt_descr * const descrs,
 345                 struct argpar_iter * const iter,
 346                 char ** const error, struct argpar_item ** const item)
 347 {
 348         enum parse_orig_arg_opt_ret ret = PARSE_ORIG_ARG_OPT_RET_OK;
 349         bool used_next_orig_arg = false;
 350         const char *opt_arg = NULL;
 351         const struct argpar_opt_descr *descr;
 352         struct argpar_item_opt *opt_item;
 353
 354         ARGPAR_ASSERT(strlen(short_opt_group) != 0);
 355
 356         if (!iter->short_opt_group_ch) {
 357                 iter->short_opt_group_ch = short_opt_group;
 358         }
 359
 360         /* Find corresponding option descriptor */
 361         descr = find_descr(descrs, *iter->short_opt_group_ch, NULL);
 362         if (!descr) {
 363                 try_append_string_printf(error, "Unknown option `-%c`",
 364                         *iter->short_opt_group_ch);
 365                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT;
 366                 goto error;
 367         }
 368
 369         if (descr->with_arg) {
 370                 if (iter->short_opt_group_ch[1]) {
 371                         /* `-oarg` form */
 372                         opt_arg = &iter->short_opt_group_ch[1];
 373                 } else {
 374                         /* `-o arg` form */
 375                         opt_arg = next_orig_arg;
 376                         used_next_orig_arg = true;
 377                 }
 378
 379                 /*
 380                  * We accept `-o ''` (empty option argument), but not
 381                  * `-o` alone if an option argument is expected.
 382                  */
 383                 if (!opt_arg || (iter->short_opt_group_ch[1] &&
 384                                 strlen(opt_arg) == 0)) {
 385                         try_append_string_printf(error,
 386                                 "Missing required argument for option `-%c`",
 387                                 *iter->short_opt_group_ch);
 388                         used_next_orig_arg = false;
 389                         ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG;
 390                         goto error;
 391                 }
 392         }
 393
 394         /* Create and append option argument */
 395         opt_item = create_opt_item(descr, opt_arg);
 396         if (!opt_item) {
 397                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MEMORY;
 398                 goto error;
 399         }
 400
 401         *item = &opt_item->base;
 402         iter->short_opt_group_ch++;
 403
 404         if (descr->with_arg || !*iter->short_opt_group_ch) {
 405                 /* Option has an argument: no more options */
 406                 iter->short_opt_group_ch = NULL;
 407
 408                 if (used_next_orig_arg) {
 409                         iter->i += 2;
 410                 } else {
 411                         iter->i++;
 412                 }
 413         }
 414
 415         goto end;
 416
 417 error:
 418         ARGPAR_ASSERT(ret != PARSE_ORIG_ARG_OPT_RET_OK);
 419
 420 end:
 421         return ret;
 422 }
 423
 424 static
 425 enum parse_orig_arg_opt_ret parse_long_opt(const char * const long_opt_arg,
 426                 const char * const next_orig_arg,
 427                 const struct argpar_opt_descr * const descrs,
 428                 struct argpar_iter * const iter,
 429                 char ** const error, struct argpar_item ** const item)
 430 {
 431         enum parse_orig_arg_opt_ret ret = PARSE_ORIG_ARG_OPT_RET_OK;
 432         const struct argpar_opt_descr *descr;
 433         struct argpar_item_opt *opt_item;
 434         bool used_next_orig_arg = false;
 435
 436         /* Option's argument, if any */
 437         const char *opt_arg = NULL;
 438
 439         /* Position of first `=`, if any */
 440         const char *eq_pos;
 441
 442         /* Option name */
 443         const char *long_opt_name = long_opt_arg;
 444
 445         ARGPAR_ASSERT(strlen(long_opt_arg) != 0);
 446
 447         /* Find the first `=` in original argument */
 448         eq_pos = strchr(long_opt_arg, '=');
 449         if (eq_pos) {
 450                 const size_t long_opt_name_size = eq_pos - long_opt_arg;
 451
 452                 /* Isolate the option name */
 453                 while (long_opt_name_size > iter->tmp_buf.size - 1) {
 454                         iter->tmp_buf.size *= 2;
 455                         iter->tmp_buf.data = ARGPAR_REALLOC(iter->tmp_buf.data,
 456                                 char, iter->tmp_buf.size);
 457                         if (!iter->tmp_buf.data) {
 458                                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MEMORY;
 459                                 goto error;
 460                         }
 461                 }
 462
 463                 memcpy(iter->tmp_buf.data, long_opt_arg, long_opt_name_size);
 464                 iter->tmp_buf.data[long_opt_name_size] = '\0';
 465                 long_opt_name = iter->tmp_buf.data;
 466         }
 467
 468         /* Find corresponding option descriptor */
 469         descr = find_descr(descrs, '\0', long_opt_name);
 470         if (!descr) {
 471                 try_append_string_printf(error, "Unknown option `--%s`",
 472                         long_opt_name);
 473                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT;
 474                 goto error;
 475         }
 476
 477         /* Find option's argument if any */
 478         if (descr->with_arg) {
 479                 if (eq_pos) {
 480                         /* `--long-opt=arg` style */
 481                         opt_arg = eq_pos + 1;
 482                 } else {
 483                         /* `--long-opt arg` style */
 484                         if (!next_orig_arg) {
 485                                 try_append_string_printf(error,
 486                                         "Missing required argument for option `--%s`",
 487                                         long_opt_name);
 488                                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG;
 489                                 goto error;
 490                         }
 491
 492                         opt_arg = next_orig_arg;
 493                         used_next_orig_arg = true;
 494                 }
 495         } else if (eq_pos) {
 496                 /*
 497                  * Unexpected `--opt=arg` style for a long option which
 498                  * doesn't accept an argument.
 499                  */
 500                 try_append_string_printf(error,
 501                         "Unexpected argument for option `--%s`", long_opt_name);
 502                 ret = PARSE_ORIG_ARG_OPT_RET_ERROR_UNEXPECTED_OPT_ARG;
 503                 goto error;
 504         }
 505
 506         /* Create and append option argument */
 507         opt_item = create_opt_item(descr, opt_arg);
 508         if (!opt_item) {
 509                 goto error;
 510         }
 511
 512         if (used_next_orig_arg) {
 513                 iter->i += 2;
 514         } else {
 515                 iter->i++;
 516         }
 517
 518         *item = &opt_item->base;
 519         goto end;
 520
 521 error:
 522         ARGPAR_ASSERT(ret != PARSE_ORIG_ARG_OPT_RET_OK);
 523
 524 end:
 525         return ret;
 526 }
 527
 528 static
 529 enum parse_orig_arg_opt_ret parse_orig_arg_opt(const char * const orig_arg,
 530                 const char * const next_orig_arg,
 531                 const struct argpar_opt_descr * const descrs,
 532                 struct argpar_iter * const iter, char ** const error,
 533                 struct argpar_item ** const item)
 534 {
 535         enum parse_orig_arg_opt_ret ret = PARSE_ORIG_ARG_OPT_RET_OK;
 536
 537         ARGPAR_ASSERT(orig_arg[0] == '-');
 538
 539         if (orig_arg[1] == '-') {
 540                 /* Long option */
 541                 ret = parse_long_opt(&orig_arg[2],
 542                         next_orig_arg, descrs, iter, error, item);
 543         } else {
 544                 /* Short option */
 545                 ret = parse_short_opt_group(&orig_arg[1],
 546                         next_orig_arg, descrs, iter, error, item);
 547         }
 548
 549         return ret;
 550 }
 551
 552 static
 553 bool try_prepend_while_parsing_arg_to_error(char ** const error,
 554                 const unsigned int i, const char * const arg)
 555 {
 556         char *new_error;
 557         bool success;
 558
 559         if (!error) {
 560                 success = true;
 561                 goto end;
 562         }
 563
 564         ARGPAR_ASSERT(*error);
 565         new_error = argpar_asprintf("While parsing argument #%u (`%s`): %s",
 566                 i + 1, arg, *error);
 567         if (!new_error) {
 568                 success = false;
 569                 goto end;
 570         }
 571
 572         free(*error);
 573         *error = new_error;
 574         success = true;
 575
 576 end:
 577         return success;
 578 }
 579
 580 ARGPAR_HIDDEN
 581 struct argpar_iter *argpar_iter_create(const unsigned int argc,
 582                 const char * const * const argv,
 583                 const struct argpar_opt_descr * const descrs)
 584 {
 585         struct argpar_iter *iter = ARGPAR_ZALLOC(struct argpar_iter);
 586
 587         if (!iter) {
 588                 goto end;
 589         }
 590
 591         iter->argc = argc;
 592         iter->argv = argv;
 593         iter->descrs = descrs;
 594         iter->tmp_buf.size = 128;
 595         iter->tmp_buf.data = ARGPAR_CALLOC(char, iter->tmp_buf.size);
 596         if (!iter->tmp_buf.data) {
 597                 argpar_iter_destroy(iter);
 598                 iter = NULL;
 599                 goto end;
 600         }
 601
 602 end:
 603         return iter;
 604 }
 605
 606 ARGPAR_HIDDEN
 607 void argpar_iter_destroy(struct argpar_iter * const iter)
 608 {
 609         if (iter) {
 610                 free(iter->tmp_buf.data);
 611                 free(iter);
 612         }
 613 }
 614
 615 ARGPAR_HIDDEN
 616 enum argpar_iter_next_status argpar_iter_next(
 617                 struct argpar_iter * const iter,
 618                 const struct argpar_item ** const item, char ** const error)
 619 {
 620         enum argpar_iter_next_status status;
 621         enum parse_orig_arg_opt_ret parse_orig_arg_opt_ret;
 622         const char *orig_arg;
 623         const char *next_orig_arg;
 624
 625         ARGPAR_ASSERT(iter->i <= iter->argc);
 626
 627         if (error) {
 628                 *error = NULL;
 629         }
 630
 631         if (iter->i == iter->argc) {
 632                 status = ARGPAR_ITER_NEXT_STATUS_END;
 633                 goto end;
 634         }
 635
 636         orig_arg = iter->argv[iter->i];
 637         next_orig_arg =
 638                 iter->i < (iter->argc - 1) ? iter->argv[iter->i + 1] : NULL;
 639
 640         if (strcmp(orig_arg, "-") == 0 || strcmp(orig_arg, "--") == 0 ||
 641                         orig_arg[0] != '-') {
 642                 /* Non-option argument */
 643                 const struct argpar_item_non_opt * const non_opt_item =
 644                         create_non_opt_item(orig_arg, iter->i,
 645                                 iter->non_opt_index);
 646
 647                 if (!non_opt_item) {
 648                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_MEMORY;
 649                         goto end;
 650                 }
 651
 652                 iter->non_opt_index++;
 653                 iter->i++;
 654                 *item = &non_opt_item->base;
 655                 status = ARGPAR_ITER_NEXT_STATUS_OK;
 656                 goto end;
 657         }
 658
 659         /* Option argument */
 660         parse_orig_arg_opt_ret = parse_orig_arg_opt(orig_arg,
 661                 next_orig_arg, iter->descrs, iter, error,
 662                 (struct argpar_item **) item);
 663         switch (parse_orig_arg_opt_ret) {
 664         case PARSE_ORIG_ARG_OPT_RET_OK:
 665                 status = ARGPAR_ITER_NEXT_STATUS_OK;
 666                 break;
 667         case PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT:
 668         case PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG:
 669         case PARSE_ORIG_ARG_OPT_RET_ERROR_UNEXPECTED_OPT_ARG:
 670                 try_prepend_while_parsing_arg_to_error(error, iter->i,
 671                         orig_arg);
 672
 673                 switch (parse_orig_arg_opt_ret) {
 674                 case PARSE_ORIG_ARG_OPT_RET_ERROR_UNKNOWN_OPT:
 675                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_UNKNOWN_OPT;
 676                         break;
 677                 case PARSE_ORIG_ARG_OPT_RET_ERROR_MISSING_OPT_ARG:
 678                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_MISSING_OPT_ARG;
 679                         break;
 680                 case PARSE_ORIG_ARG_OPT_RET_ERROR_UNEXPECTED_OPT_ARG:
 681                         status = ARGPAR_ITER_NEXT_STATUS_ERROR_UNEXPECTED_OPT_ARG;
 682                         break;
 683                 default:
 684                         abort();
 685                 }
 686
 687                 break;
 688         case PARSE_ORIG_ARG_OPT_RET_ERROR_MEMORY:
 689                 status = ARGPAR_ITER_NEXT_STATUS_ERROR_MEMORY;
 690                 break;
 691         default:
 692                 abort();
 693         }
 694
 695 end:
 696         return status;
 697 }
 698
 699 ARGPAR_HIDDEN
 700 unsigned int argpar_iter_ingested_orig_args(
 701                 const struct argpar_iter * const iter)
 702 {
 703         return iter->i;
 704 }