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