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 //  Copyright (C) 2007 Lasse Collin
   9 //
  10 //  This program is free software; you can redistribute it and/or
  11 //  modify it under the terms of the GNU Lesser General Public
  12 //  License as published by the Free Software Foundation; either
  13 //  version 2.1 of the License, or (at your option) any later version.
  14 //
  15 //  This program is distributed in the hope that it will be useful,
  16 //  but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 //  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18 //  Lesser General Public License for more details.
  19 //
  20 ///////////////////////////////////////////////////////////////////////////////
  21
  22 #include "private.h"
  23
  24 #include "getopt.h"
  25 #include <ctype.h>
  26
  27
  28 bool opt_stdout = false;
  29 bool opt_force = false;
  30 bool opt_keep_original = false;
  31
  32 // We don't modify or free() this, but we need to assign it in some
  33 // non-const pointers.
  34 const char *stdin_filename = "(stdin)";
  35
  36
  37 static void
  38 parse_real(args_info *args, int argc, char **argv)
  39 {
  40         enum {
  41                 OPT_SUBBLOCK = INT_MIN,
  42                 OPT_X86,
  43                 OPT_POWERPC,
  44                 OPT_IA64,
  45                 OPT_ARM,
  46                 OPT_ARMTHUMB,
  47                 OPT_SPARC,
  48                 OPT_DELTA,
  49                 OPT_LZMA1,
  50                 OPT_LZMA2,
  51
  52                 OPT_FILES,
  53                 OPT_FILES0,
  54         };
  55
  56         static const char short_opts[] = "cC:defF:hHlkM:qrS:tT:vVz0123456789";
  57
  58         static const struct option long_opts[] = {
  59                 // Operation mode
  60                 { "compress",       no_argument,       NULL,  'z' },
  61                 { "decompress",     no_argument,       NULL,  'd' },
  62                 { "uncompress",     no_argument,       NULL,  'd' },
  63                 { "test",           no_argument,       NULL,  't' },
  64                 { "list",           no_argument,       NULL,  'l' },
  65                 { "info",           no_argument,       NULL,  'l' },
  66
  67                 // Operation modifiers
  68                 { "keep",           no_argument,       NULL,  'k' },
  69                 { "force",          no_argument,       NULL,  'f' },
  70                 { "stdout",         no_argument,       NULL,  'c' },
  71                 { "to-stdout",      no_argument,       NULL,  'c' },
  72                 { "suffix",         required_argument, NULL,  'S' },
  73                 // { "recursive",      no_argument,       NULL,  'r' }, // TODO
  74                 { "files",          optional_argument, NULL,  OPT_FILES },
  75                 { "files0",         optional_argument, NULL,  OPT_FILES0 },
  76
  77                 // Basic compression settings
  78                 { "format",         required_argument, NULL,  'F' },
  79                 { "check",          required_argument, NULL,  'C' },
  80                 { "memory",         required_argument, NULL,  'M' },
  81                 { "threads",        required_argument, NULL,  'T' },
  82
  83                 { "extreme",        no_argument,       NULL,  'e' },
  84                 { "fast",           no_argument,       NULL,  '0' },
  85                 { "best",           no_argument,       NULL,  '9' },
  86
  87                 // Filters
  88                 { "lzma1",          optional_argument, NULL,  OPT_LZMA1 },
  89                 { "lzma2",          optional_argument, NULL,  OPT_LZMA2 },
  90                 { "x86",            no_argument,       NULL,  OPT_X86 },
  91                 { "bcj",            no_argument,       NULL,  OPT_X86 },
  92                 { "powerpc",        no_argument,       NULL,  OPT_POWERPC },
  93                 { "ppc",            no_argument,       NULL,  OPT_POWERPC },
  94                 { "ia64",           no_argument,       NULL,  OPT_IA64 },
  95                 { "itanium",        no_argument,       NULL,  OPT_IA64 },
  96                 { "arm",            no_argument,       NULL,  OPT_ARM },
  97                 { "armthumb",       no_argument,       NULL,  OPT_ARMTHUMB },
  98                 { "sparc",          no_argument,       NULL,  OPT_SPARC },
  99                 { "delta",          optional_argument, NULL,  OPT_DELTA },
 100                 { "subblock",       optional_argument, NULL,  OPT_SUBBLOCK },
 101
 102                 // Other options
 103                 { "quiet",          no_argument,       NULL,  'q' },
 104                 { "verbose",        no_argument,       NULL,  'v' },
 105                 { "help",           no_argument,       NULL,  'h' },
 106                 { "long-help",      no_argument,       NULL,  'H' },
 107                 { "version",        no_argument,       NULL,  'V' },
 108
 109                 { NULL,                 0,                 NULL,   0 }
 110         };
 111
 112         int c;
 113
 114         while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
 115                         != -1) {
 116                 switch (c) {
 117                 // Compression preset (also for decompression if --format=raw)
 118                 case '0': case '1': case '2': case '3': case '4':
 119                 case '5': case '6': case '7': case '8': case '9':
 120                         coder_set_preset(c - '0');
 121                         break;
 122
 123                 // --memory
 124                 case 'M':
 125                         // On 32-bit systems, SIZE_MAX would make more sense
 126                         // than UINT64_MAX. But use UINT64_MAX still so that
 127                         // scripts that assume > 4 GiB values don't break.
 128                         hardware_memlimit_set(str_to_uint64(
 129                                         "memory", optarg, 0, UINT64_MAX));
 130                         break;
 131
 132                 // --suffix
 133                 case 'S':
 134                         suffix_set(optarg);
 135                         break;
 136
 137                 case 'T':
 138                         opt_threads = str_to_uint64("threads", optarg,
 139                                         1, SIZE_MAX);
 140                         break;
 141
 142                 // --version
 143                 case 'V':
 144                         // This doesn't return.
 145                         message_version();
 146
 147                 // --stdout
 148                 case 'c':
 149                         opt_stdout = true;
 150                         break;
 151
 152                 // --decompress
 153                 case 'd':
 154                         opt_mode = MODE_DECOMPRESS;
 155                         break;
 156
 157                 // --extreme
 158                 case 'e':
 159                         coder_set_extreme();
 160                         break;
 161
 162                 // --force
 163                 case 'f':
 164                         opt_force = true;
 165                         break;
 166
 167                 // --help
 168                 case 'h':
 169                         // This doesn't return.
 170                         message_help(false);
 171
 172                 // --long-help
 173                 case 'H':
 174                         // This doesn't return.
 175                         message_help(true);
 176
 177                 // --list
 178                 case 'l':
 179                         opt_mode = MODE_LIST;
 180                         break;
 181
 182                 // --keep
 183                 case 'k':
 184                         opt_keep_original = true;
 185                         break;
 186
 187                 // --quiet
 188                 case 'q':
 189                         message_verbosity_decrease();
 190                         break;
 191
 192                 case 't':
 193                         opt_mode = MODE_TEST;
 194                         break;
 195
 196                 // --verbose
 197                 case 'v':
 198                         message_verbosity_increase();
 199                         break;
 200
 201                 case 'z':
 202                         opt_mode = MODE_COMPRESS;
 203                         break;
 204
 205                 // Filter setup
 206
 207                 case OPT_SUBBLOCK:
 208                         coder_add_filter(LZMA_FILTER_SUBBLOCK,
 209                                         options_subblock(optarg));
 210                         break;
 211
 212                 case OPT_X86:
 213                         coder_add_filter(LZMA_FILTER_X86, NULL);
 214                         break;
 215
 216                 case OPT_POWERPC:
 217                         coder_add_filter(LZMA_FILTER_POWERPC, NULL);
 218                         break;
 219
 220                 case OPT_IA64:
 221                         coder_add_filter(LZMA_FILTER_IA64, NULL);
 222                         break;
 223
 224                 case OPT_ARM:
 225                         coder_add_filter(LZMA_FILTER_ARM, NULL);
 226                         break;
 227
 228                 case OPT_ARMTHUMB:
 229                         coder_add_filter(LZMA_FILTER_ARMTHUMB, NULL);
 230                         break;
 231
 232                 case OPT_SPARC:
 233                         coder_add_filter(LZMA_FILTER_SPARC, NULL);
 234                         break;
 235
 236                 case OPT_DELTA:
 237                         coder_add_filter(LZMA_FILTER_DELTA,
 238                                         options_delta(optarg));
 239                         break;
 240
 241                 case OPT_LZMA1:
 242                         coder_add_filter(LZMA_FILTER_LZMA1,
 243                                         options_lzma(optarg));
 244                         break;
 245
 246                 case OPT_LZMA2:
 247                         coder_add_filter(LZMA_FILTER_LZMA2,
 248                                         options_lzma(optarg));
 249                         break;
 250
 251                 // Other
 252
 253                 // --format
 254                 case 'F': {
 255                         // Just in case, support both "lzma" and "alone" since
 256                         // the latter was used for forward compatibility in
 257                         // LZMA Utils 4.32.x.
 258                         static const struct {
 259                                 char str[8];
 260                                 enum format_type format;
 261                         } types[] = {
 262                                 { "auto",   FORMAT_AUTO },
 263                                 { "xz",     FORMAT_XZ },
 264                                 { "lzma",   FORMAT_LZMA },
 265                                 { "alone",  FORMAT_LZMA },
 266                                 // { "gzip",   FORMAT_GZIP },
 267                                 // { "gz",     FORMAT_GZIP },
 268                                 { "raw",    FORMAT_RAW },
 269                         };
 270
 271                         size_t i = 0;
 272                         while (strcmp(types[i].str, optarg) != 0)
 273                                 if (++i == ARRAY_SIZE(types))
 274                                         message_fatal(_("%s: Unknown file "
 275                                                         "format type"),
 276                                                         optarg);
 277
 278                         opt_format = types[i].format;
 279                         break;
 280                 }
 281
 282                 // --check
 283                 case 'C': {
 284                         static const struct {
 285                                 char str[8];
 286                                 lzma_check check;
 287                         } types[] = {
 288                                 { "none",   LZMA_CHECK_NONE },
 289                                 { "crc32",  LZMA_CHECK_CRC32 },
 290                                 { "crc64",  LZMA_CHECK_CRC64 },
 291                                 { "sha256", LZMA_CHECK_SHA256 },
 292                         };
 293
 294                         size_t i = 0;
 295                         while (strcmp(types[i].str, optarg) != 0) {
 296                                 if (++i == ARRAY_SIZE(types))
 297                                         message_fatal(_("%s: Unknown integrity"
 298                                                         "check type"), optarg);
 299                         }
 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 }