src/lzma/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:dfF:hHlLkM:p:qrS:tT:vVz123456789";
  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                 { "preset",         required_argument, NULL,  'p' },
  81                 { "memory",         required_argument, NULL,  'M' },
  82                 { "threads",        required_argument, NULL,  'T' },
  83
  84                 { "fast",           no_argument,       NULL,  '1' },
  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                 // gzip-like options
 118
 119                 case '1': case '2': case '3': case '4':
 120                 case '5': case '6': case '7': case '8': case '9':
 121                         coder_set_preset(c - '0');
 122                         break;
 123
 124                 case 'p': {
 125                         const uint64_t preset = str_to_uint64(
 126                                         "preset", optarg, 1, 9);
 127                         coder_set_preset(preset);
 128                         break;
 129                 }
 130
 131                 // --memory
 132                 case 'M':
 133                         // On 32-bit systems, SIZE_MAX would make more sense
 134                         // than UINT64_MAX. But use UINT64_MAX still so that
 135                         // scripts that assume > 4 GiB values don't break.
 136                         hardware_memlimit_set(str_to_uint64(
 137                                         "memory", optarg, 0, UINT64_MAX));
 138                         break;
 139
 140                 // --suffix
 141                 case 'S':
 142                         suffix_set(optarg);
 143                         break;
 144
 145                 case 'T':
 146                         opt_threads = str_to_uint64("threads", optarg,
 147                                         1, SIZE_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                 // --force
 166                 case 'f':
 167                         opt_force = true;
 168                         break;
 169
 170                 // --help
 171                 case 'h':
 172                         // This doesn't return.
 173                         message_help(false);
 174
 175                 // --long-help
 176                 case 'H':
 177                         // This doesn't return.
 178                         message_help(true);
 179
 180                 // --list
 181                 case 'l':
 182                         opt_mode = MODE_LIST;
 183                         break;
 184
 185                 // --keep
 186                 case 'k':
 187                         opt_keep_original = true;
 188                         break;
 189
 190                 // --quiet
 191                 case 'q':
 192                         message_verbosity_decrease();
 193                         break;
 194
 195                 case 't':
 196                         opt_mode = MODE_TEST;
 197                         break;
 198
 199                 // --verbose
 200                 case 'v':
 201                         message_verbosity_increase();
 202                         break;
 203
 204                 case 'z':
 205                         opt_mode = MODE_COMPRESS;
 206                         break;
 207
 208                 // Filter setup
 209
 210                 case OPT_SUBBLOCK:
 211                         coder_add_filter(LZMA_FILTER_SUBBLOCK,
 212                                         options_subblock(optarg));
 213                         break;
 214
 215                 case OPT_X86:
 216                         coder_add_filter(LZMA_FILTER_X86, NULL);
 217                         break;
 218
 219                 case OPT_POWERPC:
 220                         coder_add_filter(LZMA_FILTER_POWERPC, NULL);
 221                         break;
 222
 223                 case OPT_IA64:
 224                         coder_add_filter(LZMA_FILTER_IA64, NULL);
 225                         break;
 226
 227                 case OPT_ARM:
 228                         coder_add_filter(LZMA_FILTER_ARM, NULL);
 229                         break;
 230
 231                 case OPT_ARMTHUMB:
 232                         coder_add_filter(LZMA_FILTER_ARMTHUMB, NULL);
 233                         break;
 234
 235                 case OPT_SPARC:
 236                         coder_add_filter(LZMA_FILTER_SPARC, NULL);
 237                         break;
 238
 239                 case OPT_DELTA:
 240                         coder_add_filter(LZMA_FILTER_DELTA,
 241                                         options_delta(optarg));
 242                         break;
 243
 244                 case OPT_LZMA1:
 245                         coder_add_filter(LZMA_FILTER_LZMA1,
 246                                         options_lzma(optarg));
 247                         break;
 248
 249                 case OPT_LZMA2:
 250                         coder_add_filter(LZMA_FILTER_LZMA2,
 251                                         options_lzma(optarg));
 252                         break;
 253
 254                 // Other
 255
 256                 // --format
 257                 case 'F': {
 258                         // Just in case, support both "lzma" and "alone" since
 259                         // the latter was used for forward compatibility in
 260                         // LZMA Utils 4.32.x.
 261                         static const struct {
 262                                 char str[8];
 263                                 enum format_type format;
 264                         } types[] = {
 265                                 { "auto",   FORMAT_AUTO },
 266                                 { "xz",     FORMAT_XZ },
 267                                 { "lzma",   FORMAT_LZMA },
 268                                 { "alone",  FORMAT_LZMA },
 269                                 // { "gzip",   FORMAT_GZIP },
 270                                 // { "gz",     FORMAT_GZIP },
 271                                 { "raw",    FORMAT_RAW },
 272                         };
 273
 274                         size_t i = 0;
 275                         while (strcmp(types[i].str, optarg) != 0)
 276                                 if (++i == ARRAY_SIZE(types))
 277                                         message_fatal(_("%s: Unknown file "
 278                                                         "format type"),
 279                                                         optarg);
 280
 281                         opt_format = types[i].format;
 282                         break;
 283                 }
 284
 285                 // --check
 286                 case 'C': {
 287                         static const struct {
 288                                 char str[8];
 289                                 lzma_check check;
 290                         } types[] = {
 291                                 { "none",   LZMA_CHECK_NONE },
 292                                 { "crc32",  LZMA_CHECK_CRC32 },
 293                                 { "crc64",  LZMA_CHECK_CRC64 },
 294                                 { "sha256", LZMA_CHECK_SHA256 },
 295                         };
 296
 297                         size_t i = 0;
 298                         while (strcmp(types[i].str, optarg) != 0) {
 299                                 if (++i == ARRAY_SIZE(types))
 300                                         message_fatal(_("%s: Unknown integrity"
 301                                                         "check type"), optarg);
 302                         }
 303
 304                         coder_set_check(types[i].check);
 305                         break;
 306                 }
 307
 308                 case OPT_FILES:
 309                         args->files_delim = '\n';
 310
 311                 // Fall through
 312
 313                 case OPT_FILES0:
 314                         if (args->files_name != NULL)
 315                                 message_fatal(_("Only one file can be "
 316                                                 "specified with `--files'"
 317                                                 "or `--files0'."));
 318
 319                         if (optarg == NULL) {
 320                                 args->files_name = (char *)stdin_filename;
 321                                 args->files_file = stdin;
 322                         } else {
 323                                 args->files_name = optarg;
 324                                 args->files_file = fopen(optarg,
 325                                                 c == OPT_FILES ? "r" : "rb");
 326                                 if (args->files_file == NULL)
 327                                         message_fatal("%s: %s", optarg,
 328                                                         strerror(errno));
 329                         }
 330
 331                         break;
 332
 333                 default:
 334                         message_try_help();
 335                         my_exit(E_ERROR);
 336                 }
 337         }
 338
 339         return;
 340 }
 341
 342
 343 static void
 344 parse_environment(args_info *args, char *argv0)
 345 {
 346         char *env = getenv("XZ_OPT");
 347         if (env == NULL)
 348                 return;
 349
 350         // We modify the string, so make a copy of it.
 351         env = xstrdup(env);
 352
 353         // Calculate the number of arguments in env. argc stats at one
 354         // to include space for the program name.
 355         int argc = 1;
 356         bool prev_was_space = true;
 357         for (size_t i = 0; env[i] != '\0'; ++i) {
 358                 if (isspace(env[i])) {
 359                         prev_was_space = true;
 360                 } else if (prev_was_space) {
 361                         prev_was_space = false;
 362
 363                         // Keep argc small enough to fit into a singed int
 364                         // and to keep it usable for memory allocation.
 365                         if (++argc == MIN(INT_MAX, SIZE_MAX / sizeof(char *)))
 366                                 message_fatal(_("The environment variable "
 367                                                 "XZ_OPT contains too many "
 368                                                 "arguments"));
 369                 }
 370         }
 371
 372         // Allocate memory to hold pointers to the arguments. Add one to get
 373         // space for the terminating NULL (if some systems happen to need it).
 374         char **argv = xmalloc(((size_t)(argc) + 1) * sizeof(char *));
 375         argv[0] = argv0;
 376         argv[argc] = NULL;
 377
 378         // Go through the string again. Split the arguments using '\0'
 379         // characters and add pointers to the resulting strings to argv.
 380         argc = 1;
 381         prev_was_space = true;
 382         for (size_t i = 0; env[i] != '\0'; ++i) {
 383                 if (isspace(env[i])) {
 384                         prev_was_space = true;
 385                         env[i] = '\0';
 386                 } else if (prev_was_space) {
 387                         prev_was_space = false;
 388                         argv[argc++] = env + i;
 389                 }
 390         }
 391
 392         // Parse the argument list we got from the environment. All non-option
 393         // arguments i.e. filenames are ignored.
 394         parse_real(args, argc, argv);
 395
 396         // Reset the state of the getopt_long() so that we can parse the
 397         // command line options too. There are two incompatible ways to
 398         // do it.
 399 #ifdef HAVE_OPTRESET
 400         // BSD
 401         optind = 1;
 402         optreset = 1;
 403 #else
 404         // GNU, Solaris
 405         optind = 0;
 406 #endif
 407
 408         // We don't need the argument list from environment anymore.
 409         free(argv);
 410         free(env);
 411
 412         return;
 413 }
 414
 415
 416 extern void
 417 args_parse(args_info *args, int argc, char **argv)
 418 {
 419         // Initialize those parts of *args that we need later.
 420         args->files_name = NULL;
 421         args->files_file = NULL;
 422         args->files_delim = '\0';
 423
 424         // Type of the file format to use when --format=auto or no --format
 425         // was specified.
 426         enum format_type format_compress_auto = FORMAT_XZ;
 427
 428         // Check how we were called.
 429         {
 430                 // Remove the leading path name, if any.
 431                 const char *name = strrchr(argv[0], '/');
 432                 if (name == NULL)
 433                         name = argv[0];
 434                 else
 435                         ++name;
 436
 437                 // NOTE: It's possible that name[0] is now '\0' if argv[0]
 438                 // is weird, but it doesn't matter here.
 439
 440                 // The default file format is .lzma if the command name
 441                 // contains "lz".
 442                 if (strstr(name, "lz") != NULL)
 443                         format_compress_auto = FORMAT_LZMA;
 444
 445                 // Operation mode
 446                 if (strstr(name, "cat") != NULL) {
 447                         // Imply --decompress --stdout
 448                         opt_mode = MODE_DECOMPRESS;
 449                         opt_stdout = true;
 450                 } else if (strstr(name, "un") != NULL) {
 451                         // Imply --decompress
 452                         opt_mode = MODE_DECOMPRESS;
 453                 }
 454         }
 455
 456         // First the flags from environment
 457         parse_environment(args, argv[0]);
 458
 459         // Then from the command line
 460         optind = 1;
 461         parse_real(args, argc, argv);
 462
 463         // Never remove the source file when the destination is not on disk.
 464         // In test mode the data is written nowhere, but setting opt_stdout
 465         // will make the rest of the code behave well.
 466         if (opt_stdout || opt_mode == MODE_TEST) {
 467                 opt_keep_original = true;
 468                 opt_stdout = true;
 469         }
 470
 471         // If no --format flag was used, or it was --format=auto, we need to
 472         // decide what is the target file format we are going to use. This
 473         // depends on how we were called (checked earlier in this function).
 474         if (opt_mode == MODE_COMPRESS && opt_format == FORMAT_AUTO)
 475                 opt_format = format_compress_auto;
 476
 477         // Compression settings need to be validated (options themselves and
 478         // their memory usage) when compressing to any file format. It has to
 479         // be done also when uncompressing raw data, since for raw decoding
 480         // the options given on the command line are used to know what kind
 481         // of raw data we are supposed to decode.
 482         if (opt_mode == MODE_COMPRESS || opt_format == FORMAT_RAW)
 483                 coder_set_compression_settings();
 484
 485         // If no filenames are given, use stdin.
 486         if (argv[optind] == NULL && args->files_name == NULL) {
 487                 // We don't modify or free() the "-" constant. The caller
 488                 // modifies this so don't make the struct itself const.
 489                 static char *names_stdin[2] = { (char *)"-", NULL };
 490                 args->arg_names = names_stdin;
 491                 args->arg_count = 1;
 492         } else {
 493                 // We got at least one filename from the command line, or
 494                 // --files or --files0 was specified.
 495                 args->arg_names = argv + optind;
 496                 args->arg_count = argc - optind;
 497         }
 498
 499         return;
 500 }