1 ///////////////////////////////////////////////////////////////////////////////
4 /// \brief Argument parsing
6 /// \note Filter-specific options parsing is in options.c.
8 // Author: Lasse Collin
10 // This file has been put into the public domain.
11 // You can do whatever you want with this file.
13 ///////////////////////////////////////////////////////////////////////////////
21 bool opt_stdout = false;
22 bool opt_force = false;
23 bool opt_keep_original = false;
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)";
31 parse_real(args_info *args, int argc, char **argv)
34 OPT_SUBBLOCK = INT_MIN,
49 static const char short_opts[] = "cC:defF:hHlkM:qrS:tT:vVz0123456789";
51 static const struct option long_opts[] = {
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' },
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 },
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' },
76 { "extreme", no_argument, NULL, 'e' },
77 { "fast", no_argument, NULL, '0' },
78 { "best", no_argument, NULL, '9' },
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 },
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' },
107 while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
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');
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));
131 hardware_threadlimit_set(str_to_uint64(
132 "threads", optarg, 1, SIZE_MAX));
137 // This doesn't return.
147 opt_mode = MODE_DECOMPRESS;
162 // This doesn't return.
167 // This doesn't return.
172 opt_mode = MODE_LIST;
177 opt_keep_original = true;
182 message_verbosity_decrease();
186 opt_mode = MODE_TEST;
191 message_verbosity_increase();
195 opt_mode = MODE_COMPRESS;
201 coder_add_filter(LZMA_FILTER_SUBBLOCK,
202 options_subblock(optarg));
206 coder_add_filter(LZMA_FILTER_X86, NULL);
210 coder_add_filter(LZMA_FILTER_POWERPC, NULL);
214 coder_add_filter(LZMA_FILTER_IA64, NULL);
218 coder_add_filter(LZMA_FILTER_ARM, NULL);
222 coder_add_filter(LZMA_FILTER_ARMTHUMB, NULL);
226 coder_add_filter(LZMA_FILTER_SPARC, NULL);
230 coder_add_filter(LZMA_FILTER_DELTA,
231 options_delta(optarg));
235 coder_add_filter(LZMA_FILTER_LZMA1,
236 options_lzma(optarg));
240 coder_add_filter(LZMA_FILTER_LZMA2,
241 options_lzma(optarg));
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 {
253 enum format_type format;
255 { "auto", FORMAT_AUTO },
257 { "lzma", FORMAT_LZMA },
258 { "alone", FORMAT_LZMA },
259 // { "gzip", FORMAT_GZIP },
260 // { "gz", FORMAT_GZIP },
261 { "raw", FORMAT_RAW },
265 while (strcmp(types[i].str, optarg) != 0)
266 if (++i == ARRAY_SIZE(types))
267 message_fatal(_("%s: Unknown file "
271 opt_format = types[i].format;
277 static const struct {
281 { "none", LZMA_CHECK_NONE },
282 { "crc32", LZMA_CHECK_CRC32 },
283 { "crc64", LZMA_CHECK_CRC64 },
284 { "sha256", LZMA_CHECK_SHA256 },
288 while (strcmp(types[i].str, optarg) != 0) {
289 if (++i == ARRAY_SIZE(types))
290 message_fatal(_("%s: Unsupported "
292 "check type"), optarg);
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);
301 coder_set_check(types[i].check);
306 args->files_delim = '\n';
311 if (args->files_name != NULL)
312 message_fatal(_("Only one file can be "
313 "specified with `--files'"
316 if (optarg == NULL) {
317 args->files_name = (char *)stdin_filename;
318 args->files_file = stdin;
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,
341 parse_environment(args_info *args, char *argv0)
343 char *env = getenv("XZ_OPT");
347 // We modify the string, so make a copy of it.
350 // Calculate the number of arguments in env. argc stats at one
351 // to include space for the program name.
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;
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 "
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 *));
375 // Go through the string again. Split the arguments using '\0'
376 // characters and add pointers to the resulting strings to argv.
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;
383 } else if (prev_was_space) {
384 prev_was_space = false;
385 argv[argc++] = env + i;
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);
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
405 // We don't need the argument list from environment anymore.
414 args_parse(args_info *args, int argc, char **argv)
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';
421 // Type of the file format to use when --format=auto or no --format
423 enum format_type format_compress_auto = FORMAT_XZ;
425 // Check how we were called.
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];
432 // Remove the leading path name, if any.
433 const char *name = strrchr(argv[0], '/');
440 // NOTE: It's possible that name[0] is now '\0' if argv[0]
441 // is weird, but it doesn't matter here.
443 // The default file format is .lzma if the command name
445 if (strstr(name, "lz") != NULL)
446 format_compress_auto = FORMAT_LZMA;
449 if (strstr(name, "cat") != NULL) {
450 // Imply --decompress --stdout
451 opt_mode = MODE_DECOMPRESS;
453 } else if (strstr(name, "un") != NULL) {
454 // Imply --decompress
455 opt_mode = MODE_DECOMPRESS;
459 // First the flags from environment
460 parse_environment(args, argv[0]);
462 // Then from the command line
464 parse_real(args, argc, argv);
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;
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;
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();
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;
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;