1 ///////////////////////////////////////////////////////////////////////////////
4 /// \brief Argument parsing
6 /// \note Filter-specific options parsing is in options.c.
8 // Copyright (C) 2007 Lasse Collin
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.
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.
20 ///////////////////////////////////////////////////////////////////////////////
28 bool opt_stdout = false;
29 bool opt_force = false;
30 bool opt_keep_original = false;
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)";
38 parse_real(args_info *args, int argc, char **argv)
41 OPT_SUBBLOCK = INT_MIN,
56 static const char short_opts[] = "cC:defF:hHlkM:qrS:tT:vVz0123456789";
58 static const struct option long_opts[] = {
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' },
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 },
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' },
83 { "extreme", no_argument, NULL, 'e' },
84 { "fast", no_argument, NULL, '0' },
85 { "best", no_argument, NULL, '9' },
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 },
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' },
114 while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
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');
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));
138 hardware_threadlimit_set(str_to_uint64(
139 "threads", optarg, 1, SIZE_MAX));
144 // This doesn't return.
154 opt_mode = MODE_DECOMPRESS;
169 // This doesn't return.
174 // This doesn't return.
179 opt_mode = MODE_LIST;
184 opt_keep_original = true;
189 message_verbosity_decrease();
193 opt_mode = MODE_TEST;
198 message_verbosity_increase();
202 opt_mode = MODE_COMPRESS;
208 coder_add_filter(LZMA_FILTER_SUBBLOCK,
209 options_subblock(optarg));
213 coder_add_filter(LZMA_FILTER_X86, NULL);
217 coder_add_filter(LZMA_FILTER_POWERPC, NULL);
221 coder_add_filter(LZMA_FILTER_IA64, NULL);
225 coder_add_filter(LZMA_FILTER_ARM, NULL);
229 coder_add_filter(LZMA_FILTER_ARMTHUMB, NULL);
233 coder_add_filter(LZMA_FILTER_SPARC, NULL);
237 coder_add_filter(LZMA_FILTER_DELTA,
238 options_delta(optarg));
242 coder_add_filter(LZMA_FILTER_LZMA1,
243 options_lzma(optarg));
247 coder_add_filter(LZMA_FILTER_LZMA2,
248 options_lzma(optarg));
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 {
260 enum format_type format;
262 { "auto", FORMAT_AUTO },
264 { "lzma", FORMAT_LZMA },
265 { "alone", FORMAT_LZMA },
266 // { "gzip", FORMAT_GZIP },
267 // { "gz", FORMAT_GZIP },
268 { "raw", FORMAT_RAW },
272 while (strcmp(types[i].str, optarg) != 0)
273 if (++i == ARRAY_SIZE(types))
274 message_fatal(_("%s: Unknown file "
278 opt_format = types[i].format;
284 static const struct {
288 { "none", LZMA_CHECK_NONE },
289 { "crc32", LZMA_CHECK_CRC32 },
290 { "crc64", LZMA_CHECK_CRC64 },
291 { "sha256", LZMA_CHECK_SHA256 },
295 while (strcmp(types[i].str, optarg) != 0) {
296 if (++i == ARRAY_SIZE(types))
297 message_fatal(_("%s: Unsupported "
299 "check type"), optarg);
302 // Use a separate check in case we are using different
303 // liblzma than what was used to compile us.
304 if (!lzma_check_is_supported(types[i].check))
305 message_fatal(_("%s: Unsupported integrity "
306 "check type"), optarg);
308 coder_set_check(types[i].check);
313 args->files_delim = '\n';
318 if (args->files_name != NULL)
319 message_fatal(_("Only one file can be "
320 "specified with `--files'"
323 if (optarg == NULL) {
324 args->files_name = (char *)stdin_filename;
325 args->files_file = stdin;
327 args->files_name = optarg;
328 args->files_file = fopen(optarg,
329 c == OPT_FILES ? "r" : "rb");
330 if (args->files_file == NULL)
331 message_fatal("%s: %s", optarg,
348 parse_environment(args_info *args, char *argv0)
350 char *env = getenv("XZ_OPT");
354 // We modify the string, so make a copy of it.
357 // Calculate the number of arguments in env. argc stats at one
358 // to include space for the program name.
360 bool prev_was_space = true;
361 for (size_t i = 0; env[i] != '\0'; ++i) {
362 if (isspace(env[i])) {
363 prev_was_space = true;
364 } else if (prev_was_space) {
365 prev_was_space = false;
367 // Keep argc small enough to fit into a singed int
368 // and to keep it usable for memory allocation.
369 if (++argc == MIN(INT_MAX, SIZE_MAX / sizeof(char *)))
370 message_fatal(_("The environment variable "
371 "XZ_OPT contains too many "
376 // Allocate memory to hold pointers to the arguments. Add one to get
377 // space for the terminating NULL (if some systems happen to need it).
378 char **argv = xmalloc(((size_t)(argc) + 1) * sizeof(char *));
382 // Go through the string again. Split the arguments using '\0'
383 // characters and add pointers to the resulting strings to argv.
385 prev_was_space = true;
386 for (size_t i = 0; env[i] != '\0'; ++i) {
387 if (isspace(env[i])) {
388 prev_was_space = true;
390 } else if (prev_was_space) {
391 prev_was_space = false;
392 argv[argc++] = env + i;
396 // Parse the argument list we got from the environment. All non-option
397 // arguments i.e. filenames are ignored.
398 parse_real(args, argc, argv);
400 // Reset the state of the getopt_long() so that we can parse the
401 // command line options too. There are two incompatible ways to
412 // We don't need the argument list from environment anymore.
421 args_parse(args_info *args, int argc, char **argv)
423 // Initialize those parts of *args that we need later.
424 args->files_name = NULL;
425 args->files_file = NULL;
426 args->files_delim = '\0';
428 // Type of the file format to use when --format=auto or no --format
430 enum format_type format_compress_auto = FORMAT_XZ;
432 // Check how we were called.
435 // We adjusted argv[0] in the beginning of main() so we don't
436 // need to do anything here.
437 const char *name = argv[0];
439 // Remove the leading path name, if any.
440 const char *name = strrchr(argv[0], '/');
447 // NOTE: It's possible that name[0] is now '\0' if argv[0]
448 // is weird, but it doesn't matter here.
450 // The default file format is .lzma if the command name
452 if (strstr(name, "lz") != NULL)
453 format_compress_auto = FORMAT_LZMA;
456 if (strstr(name, "cat") != NULL) {
457 // Imply --decompress --stdout
458 opt_mode = MODE_DECOMPRESS;
460 } else if (strstr(name, "un") != NULL) {
461 // Imply --decompress
462 opt_mode = MODE_DECOMPRESS;
466 // First the flags from environment
467 parse_environment(args, argv[0]);
469 // Then from the command line
471 parse_real(args, argc, argv);
473 // Never remove the source file when the destination is not on disk.
474 // In test mode the data is written nowhere, but setting opt_stdout
475 // will make the rest of the code behave well.
476 if (opt_stdout || opt_mode == MODE_TEST) {
477 opt_keep_original = true;
481 // If no --format flag was used, or it was --format=auto, we need to
482 // decide what is the target file format we are going to use. This
483 // depends on how we were called (checked earlier in this function).
484 if (opt_mode == MODE_COMPRESS && opt_format == FORMAT_AUTO)
485 opt_format = format_compress_auto;
487 // Compression settings need to be validated (options themselves and
488 // their memory usage) when compressing to any file format. It has to
489 // be done also when uncompressing raw data, since for raw decoding
490 // the options given on the command line are used to know what kind
491 // of raw data we are supposed to decode.
492 if (opt_mode == MODE_COMPRESS || opt_format == FORMAT_RAW)
493 coder_set_compression_settings();
495 // If no filenames are given, use stdin.
496 if (argv[optind] == NULL && args->files_name == NULL) {
497 // We don't modify or free() the "-" constant. The caller
498 // modifies this so don't make the struct itself const.
499 static char *names_stdin[2] = { (char *)"-", NULL };
500 args->arg_names = names_stdin;
503 // We got at least one filename from the command line, or
504 // --files or --files0 was specified.
505 args->arg_names = argv + optind;
506 args->arg_count = argc - optind;