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