src/xz/args.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       args.c
   4 /// \brief      Argument parsing
   5 ///
   6 /// \note       Filter-specific options parsing is in options.c.
   7 //
   8 //  Author:     Lasse Collin
   9 //
  10 //  This file has been put into the public domain.
  11 //  You can do whatever you want with this file.
  12 //
  13 ///////////////////////////////////////////////////////////////////////////////
  14
  15 #include "private.h"
  16
  17 #include "getopt.h"
  18 #include <ctype.h>
  19
  20
  21 bool opt_stdout = false;
  22 bool opt_force = false;
  23 bool opt_keep_original = false;
  24
  25 // We don't modify or free() this, but we need to assign it in some
  26 // non-const pointers.
  27 const char *stdin_filename = "(stdin)";
  28
  29
  30 static void
  31 parse_real(args_info *args, int argc, char **argv)
  32 {
  33         enum {
  34                 OPT_SUBBLOCK = INT_MIN,
  35                 OPT_X86,
  36                 OPT_POWERPC,
  37                 OPT_IA64,
  38                 OPT_ARM,
  39                 OPT_ARMTHUMB,
  40                 OPT_SPARC,
  41                 OPT_DELTA,
  42                 OPT_LZMA1,
  43                 OPT_LZMA2,
  44
  45                 OPT_FILES,
  46                 OPT_FILES0,
  47         };
  48
  49         static const char short_opts[]
  50                         = "cC:defF:hHlkM:qQrS:tT:vVz0123456789";
  51
  52         static const struct option long_opts[] = {
  53                 // Operation mode
  54                 { "compress",       no_argument,       NULL,  'z' },
  55                 { "decompress",     no_argument,       NULL,  'd' },
  56                 { "uncompress",     no_argument,       NULL,  'd' },
  57                 { "test",           no_argument,       NULL,  't' },
  58                 { "list",           no_argument,       NULL,  'l' },
  59
  60                 // Operation modifiers
  61                 { "keep",           no_argument,       NULL,  'k' },
  62                 { "force",          no_argument,       NULL,  'f' },
  63                 { "stdout",         no_argument,       NULL,  'c' },
  64                 { "to-stdout",      no_argument,       NULL,  'c' },
  65                 { "suffix",         required_argument, NULL,  'S' },
  66                 // { "recursive",      no_argument,       NULL,  'r' }, // TODO
  67                 { "files",          optional_argument, NULL,  OPT_FILES },
  68                 { "files0",         optional_argument, NULL,  OPT_FILES0 },
  69
  70                 // Basic compression settings
  71                 { "format",         required_argument, NULL,  'F' },
  72                 { "check",          required_argument, NULL,  'C' },
  73                 { "memory",         required_argument, NULL,  'M' },
  74                 { "threads",        required_argument, NULL,  'T' },
  75
  76                 { "extreme",        no_argument,       NULL,  'e' },
  77                 { "fast",           no_argument,       NULL,  '0' },
  78                 { "best",           no_argument,       NULL,  '9' },
  79
  80                 // Filters
  81                 { "lzma1",          optional_argument, NULL,  OPT_LZMA1 },
  82                 { "lzma2",          optional_argument, NULL,  OPT_LZMA2 },
  83                 { "x86",            no_argument,       NULL,  OPT_X86 },
  84                 { "bcj",            no_argument,       NULL,  OPT_X86 },
  85                 { "powerpc",        no_argument,       NULL,  OPT_POWERPC },
  86                 { "ppc",            no_argument,       NULL,  OPT_POWERPC },
  87                 { "ia64",           no_argument,       NULL,  OPT_IA64 },
  88                 { "itanium",        no_argument,       NULL,  OPT_IA64 },
  89                 { "arm",            no_argument,       NULL,  OPT_ARM },
  90                 { "armthumb",       no_argument,       NULL,  OPT_ARMTHUMB },
  91                 { "sparc",          no_argument,       NULL,  OPT_SPARC },
  92                 { "delta",          optional_argument, NULL,  OPT_DELTA },
  93                 { "subblock",       optional_argument, NULL,  OPT_SUBBLOCK },
  94
  95                 // Other options
  96                 { "quiet",          no_argument,       NULL,  'q' },
  97                 { "verbose",        no_argument,       NULL,  'v' },
  98                 { "no-warn",        no_argument,       NULL,  'Q' },
  99                 { "help",           no_argument,       NULL,  'h' },
 100                 { "long-help",      no_argument,       NULL,  'H' },
 101                 { "version",        no_argument,       NULL,  'V' },
 102
 103                 { NULL,                 0,                 NULL,   0 }
 104         };
 105
 106         int c;
 107
 108         while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
 109                         != -1) {
 110                 switch (c) {
 111                 // Compression preset (also for decompression if --format=raw)
 112                 case '0': case '1': case '2': case '3': case '4':
 113                 case '5': case '6': case '7': case '8': case '9':
 114                         coder_set_preset(c - '0');
 115                         break;
 116
 117                 // --memory
 118                 case 'M': {
 119                         // Support specifying the limit as a percentage of
 120                         // installed physical RAM.
 121                         size_t len = strlen(optarg);
 122                         if (len > 0 && optarg[len - 1] == '%') {
 123                                 optarg[len - 1] = '\0';
 124                                 hardware_memlimit_set_percentage(
 125                                                 str_to_uint64(
 126                                                 "memory%", optarg, 1, 100));
 127                         } else {
 128                                 // On 32-bit systems, SIZE_MAX would make more
 129                                 // sense than UINT64_MAX. But use UINT64_MAX
 130                                 // still so that scripts that assume > 4 GiB
 131                                 // values don't break.
 132                                 hardware_memlimit_set(str_to_uint64(
 133                                                 "memory", optarg,
 134                                                 0, UINT64_MAX));
 135                         }
 136
 137                         break;
 138                 }
 139
 140                 // --suffix
 141                 case 'S':
 142                         suffix_set(optarg);
 143                         break;
 144
 145                 case 'T':
 146                         hardware_threadlimit_set(str_to_uint64(
 147                                         "threads", optarg, 0, UINT32_MAX));
 148                         break;
 149
 150                 // --version
 151                 case 'V':
 152                         // This doesn't return.
 153                         message_version();
 154
 155                 // --stdout
 156                 case 'c':
 157                         opt_stdout = true;
 158                         break;
 159
 160                 // --decompress
 161                 case 'd':
 162                         opt_mode = MODE_DECOMPRESS;
 163                         break;
 164
 165                 // --extreme
 166                 case 'e':
 167                         coder_set_extreme();
 168                         break;
 169
 170                 // --force
 171                 case 'f':
 172                         opt_force = true;
 173                         break;
 174
 175                 // --help
 176                 case 'h':
 177                         // This doesn't return.
 178                         message_help(false);
 179
 180                 // --long-help
 181                 case 'H':
 182                         // This doesn't return.
 183                         message_help(true);
 184
 185                 // --list
 186                 case 'l':
 187                         opt_mode = MODE_LIST;
 188                         break;
 189
 190                 // --keep
 191                 case 'k':
 192                         opt_keep_original = true;
 193                         break;
 194
 195                 // --quiet
 196                 case 'q':
 197                         message_verbosity_decrease();
 198                         break;
 199
 200                 case 'Q':
 201                         set_exit_no_warn();
 202                         break;
 203
 204                 case 't':
 205                         opt_mode = MODE_TEST;
 206                         break;
 207
 208                 // --verbose
 209                 case 'v':
 210                         message_verbosity_increase();
 211                         break;
 212
 213                 case 'z':
 214                         opt_mode = MODE_COMPRESS;
 215                         break;
 216
 217                 // Filter setup
 218
 219                 case OPT_SUBBLOCK:
 220                         coder_add_filter(LZMA_FILTER_SUBBLOCK,
 221                                         options_subblock(optarg));
 222                         break;
 223
 224                 case OPT_X86:
 225                         coder_add_filter(LZMA_FILTER_X86, NULL);
 226                         break;
 227
 228                 case OPT_POWERPC:
 229                         coder_add_filter(LZMA_FILTER_POWERPC, NULL);
 230                         break;
 231
 232                 case OPT_IA64:
 233                         coder_add_filter(LZMA_FILTER_IA64, NULL);
 234                         break;
 235
 236                 case OPT_ARM:
 237                         coder_add_filter(LZMA_FILTER_ARM, NULL);
 238                         break;
 239
 240                 case OPT_ARMTHUMB:
 241                         coder_add_filter(LZMA_FILTER_ARMTHUMB, NULL);
 242                         break;
 243
 244                 case OPT_SPARC:
 245                         coder_add_filter(LZMA_FILTER_SPARC, NULL);
 246                         break;
 247
 248                 case OPT_DELTA:
 249                         coder_add_filter(LZMA_FILTER_DELTA,
 250                                         options_delta(optarg));
 251                         break;
 252
 253                 case OPT_LZMA1:
 254                         coder_add_filter(LZMA_FILTER_LZMA1,
 255                                         options_lzma(optarg));
 256                         break;
 257
 258                 case OPT_LZMA2:
 259                         coder_add_filter(LZMA_FILTER_LZMA2,
 260                                         options_lzma(optarg));
 261                         break;
 262
 263                 // Other
 264
 265                 // --format
 266                 case 'F': {
 267                         // Just in case, support both "lzma" and "alone" since
 268                         // the latter was used for forward compatibility in
 269                         // LZMA Utils 4.32.x.
 270                         static const struct {
 271                                 char str[8];
 272                                 enum format_type format;
 273                         } types[] = {
 274                                 { "auto",   FORMAT_AUTO },
 275                                 { "xz",     FORMAT_XZ },
 276                                 { "lzma",   FORMAT_LZMA },
 277                                 { "alone",  FORMAT_LZMA },
 278                                 // { "gzip",   FORMAT_GZIP },
 279                                 // { "gz",     FORMAT_GZIP },
 280                                 { "raw",    FORMAT_RAW },
 281                         };
 282
 283                         size_t i = 0;
 284                         while (strcmp(types[i].str, optarg) != 0)
 285                                 if (++i == ARRAY_SIZE(types))
 286                                         message_fatal(_("%s: Unknown file "
 287                                                         "format type"),
 288                                                         optarg);
 289
 290                         opt_format = types[i].format;
 291                         break;
 292                 }
 293
 294                 // --check
 295                 case 'C': {
 296                         static const struct {
 297                                 char str[8];
 298                                 lzma_check check;
 299                         } types[] = {
 300                                 { "none",   LZMA_CHECK_NONE },
 301                                 { "crc32",  LZMA_CHECK_CRC32 },
 302                                 { "crc64",  LZMA_CHECK_CRC64 },
 303                                 { "sha256", LZMA_CHECK_SHA256 },
 304                         };
 305
 306                         size_t i = 0;
 307                         while (strcmp(types[i].str, optarg) != 0) {
 308                                 if (++i == ARRAY_SIZE(types))
 309                                         message_fatal(_("%s: Unsupported "
 310                                                         "integrity "
 311                                                         "check type"), optarg);
 312                         }
 313
 314                         // Use a separate check in case we are using different
 315                         // liblzma than what was used to compile us.
 316                         if (!lzma_check_is_supported(types[i].check))
 317                                 message_fatal(_("%s: Unsupported integrity "
 318                                                 "check type"), optarg);
 319
 320                         coder_set_check(types[i].check);
 321                         break;
 322                 }
 323
 324                 case OPT_FILES:
 325                         args->files_delim = '\n';
 326
 327                 // Fall through
 328
 329                 case OPT_FILES0:
 330                         if (args->files_name != NULL)
 331                                 message_fatal(_("Only one file can be "
 332                                                 "specified with `--files'"
 333                                                 "or `--files0'."));
 334
 335                         if (optarg == NULL) {
 336                                 args->files_name = (char *)stdin_filename;
 337                                 args->files_file = stdin;
 338                         } else {
 339                                 args->files_name = optarg;
 340                                 args->files_file = fopen(optarg,
 341                                                 c == OPT_FILES ? "r" : "rb");
 342                                 if (args->files_file == NULL)
 343                                         message_fatal("%s: %s", optarg,
 344                                                         strerror(errno));
 345                         }
 346
 347                         break;
 348
 349                 default:
 350                         message_try_help();
 351                         my_exit(E_ERROR);
 352                 }
 353         }
 354
 355         return;
 356 }
 357
 358
 359 static void
 360 parse_environment(args_info *args, char *argv0)
 361 {
 362         char *env = getenv("XZ_OPT");
 363         if (env == NULL)
 364                 return;
 365
 366         // We modify the string, so make a copy of it.
 367         env = xstrdup(env);
 368
 369         // Calculate the number of arguments in env. argc stats at one
 370         // to include space for the program name.
 371         int argc = 1;
 372         bool prev_was_space = true;
 373         for (size_t i = 0; env[i] != '\0'; ++i) {
 374                 if (isspace(env[i])) {
 375                         prev_was_space = true;
 376                 } else if (prev_was_space) {
 377                         prev_was_space = false;
 378
 379                         // Keep argc small enough to fit into a singed int
 380                         // and to keep it usable for memory allocation.
 381                         if (++argc == MIN(INT_MAX, SIZE_MAX / sizeof(char *)))
 382                                 message_fatal(_("The environment variable "
 383                                                 "XZ_OPT contains too many "
 384                                                 "arguments"));
 385                 }
 386         }
 387
 388         // Allocate memory to hold pointers to the arguments. Add one to get
 389         // space for the terminating NULL (if some systems happen to need it).
 390         char **argv = xmalloc(((size_t)(argc) + 1) * sizeof(char *));
 391         argv[0] = argv0;
 392         argv[argc] = NULL;
 393
 394         // Go through the string again. Split the arguments using '\0'
 395         // characters and add pointers to the resulting strings to argv.
 396         argc = 1;
 397         prev_was_space = true;
 398         for (size_t i = 0; env[i] != '\0'; ++i) {
 399                 if (isspace(env[i])) {
 400                         prev_was_space = true;
 401                         env[i] = '\0';
 402                 } else if (prev_was_space) {
 403                         prev_was_space = false;
 404                         argv[argc++] = env + i;
 405                 }
 406         }
 407
 408         // Parse the argument list we got from the environment. All non-option
 409         // arguments i.e. filenames are ignored.
 410         parse_real(args, argc, argv);
 411
 412         // Reset the state of the getopt_long() so that we can parse the
 413         // command line options too. There are two incompatible ways to
 414         // do it.
 415 #ifdef HAVE_OPTRESET
 416         // BSD
 417         optind = 1;
 418         optreset = 1;
 419 #else
 420         // GNU, Solaris
 421         optind = 0;
 422 #endif
 423
 424         // We don't need the argument list from environment anymore.
 425         free(argv);
 426         free(env);
 427
 428         return;
 429 }
 430
 431
 432 extern void
 433 args_parse(args_info *args, int argc, char **argv)
 434 {
 435         // Initialize those parts of *args that we need later.
 436         args->files_name = NULL;
 437         args->files_file = NULL;
 438         args->files_delim = '\0';
 439
 440         // Check how we were called.
 441         {
 442 #ifdef DOSLIKE
 443                 // We adjusted argv[0] in the beginning of main() so we don't
 444                 // need to do anything here.
 445                 const char *name = argv[0];
 446 #else
 447                 // Remove the leading path name, if any.
 448                 const char *name = strrchr(argv[0], '/');
 449                 if (name == NULL)
 450                         name = argv[0];
 451                 else
 452                         ++name;
 453 #endif
 454
 455                 // NOTE: It's possible that name[0] is now '\0' if argv[0]
 456                 // is weird, but it doesn't matter here.
 457
 458                 // If the command name contains "lz",
 459                 // it implies --format=lzma.
 460                 if (strstr(name, "lz") != NULL)
 461                         opt_format = FORMAT_LZMA;
 462
 463                 // Operation mode
 464                 if (strstr(name, "cat") != NULL) {
 465                         // Imply --decompress --stdout
 466                         opt_mode = MODE_DECOMPRESS;
 467                         opt_stdout = true;
 468                 } else if (strstr(name, "un") != NULL) {
 469                         // Imply --decompress
 470                         opt_mode = MODE_DECOMPRESS;
 471                 }
 472         }
 473
 474         // First the flags from environment
 475         parse_environment(args, argv[0]);
 476
 477         // Then from the command line
 478         optind = 1;
 479         parse_real(args, argc, argv);
 480
 481         // Never remove the source file when the destination is not on disk.
 482         // In test mode the data is written nowhere, but setting opt_stdout
 483         // will make the rest of the code behave well.
 484         if (opt_stdout || opt_mode == MODE_TEST) {
 485                 opt_keep_original = true;
 486                 opt_stdout = true;
 487         }
 488
 489         // When compressing, if no --format flag was used, or it
 490         // was --format=auto, we compress to the .xz format.
 491         if (opt_mode == MODE_COMPRESS && opt_format == FORMAT_AUTO)
 492                 opt_format = FORMAT_XZ;
 493
 494         // Compression settings need to be validated (options themselves and
 495         // their memory usage) when compressing to any file format. It has to
 496         // be done also when uncompressing raw data, since for raw decoding
 497         // the options given on the command line are used to know what kind
 498         // of raw data we are supposed to decode.
 499         if (opt_mode == MODE_COMPRESS || opt_format == FORMAT_RAW)
 500                 coder_set_compression_settings();
 501
 502         // If no filenames are given, use stdin.
 503         if (argv[optind] == NULL && args->files_name == NULL) {
 504                 // We don't modify or free() the "-" constant. The caller
 505                 // modifies this so don't make the struct itself const.
 506                 static char *names_stdin[2] = { (char *)"-", NULL };
 507                 args->arg_names = names_stdin;
 508                 args->arg_count = 1;
 509         } else {
 510                 // We got at least one filename from the command line, or
 511                 // --files or --files0 was specified.
 512                 args->arg_names = argv + optind;
 513                 args->arg_count = argc - optind;
 514         }
 515
 516         return;
 517 }