648db22b |
1 | /* |
2 | * Copyright (c) Meta Platforms, Inc. and affiliates. |
3 | * All rights reserved. |
4 | * |
5 | * This source code is licensed under both the BSD-style license (found in the |
6 | * LICENSE file in the root directory of this source tree) and the GPLv2 (found |
7 | * in the COPYING file in the root directory of this source tree). |
8 | * You may select, at your option, one of the above-listed licenses. |
9 | */ |
10 | |
11 | |
12 | /*-************************************ |
13 | * Tuning parameters |
14 | **************************************/ |
15 | #ifndef ZSTDCLI_CLEVEL_DEFAULT |
16 | # define ZSTDCLI_CLEVEL_DEFAULT 3 |
17 | #endif |
18 | |
19 | #ifndef ZSTDCLI_CLEVEL_MAX |
20 | # define ZSTDCLI_CLEVEL_MAX 19 /* without using --ultra */ |
21 | #endif |
22 | |
23 | #ifndef ZSTDCLI_NBTHREADS_DEFAULT |
24 | # define ZSTDCLI_NBTHREADS_DEFAULT 1 |
25 | #endif |
26 | |
27 | /*-************************************ |
28 | * Dependencies |
29 | **************************************/ |
30 | #include "platform.h" /* PLATFORM_POSIX_VERSION */ |
31 | #include "util.h" /* UTIL_HAS_CREATEFILELIST, UTIL_createFileList, UTIL_isConsole */ |
32 | #include <stdlib.h> /* getenv */ |
33 | #include <string.h> /* strcmp, strlen */ |
34 | #include <stdio.h> /* fprintf(), stdin, stdout, stderr */ |
35 | #include <errno.h> /* errno */ |
36 | #include <assert.h> /* assert */ |
37 | |
38 | #include "fileio.h" /* stdinmark, stdoutmark, ZSTD_EXTENSION */ |
39 | #ifndef ZSTD_NOBENCH |
40 | # include "benchzstd.h" /* BMK_benchFilesAdvanced */ |
41 | #endif |
42 | #ifndef ZSTD_NODICT |
43 | # include "dibio.h" /* ZDICT_cover_params_t, DiB_trainFromFiles() */ |
44 | #endif |
45 | #ifndef ZSTD_NOTRACE |
46 | # include "zstdcli_trace.h" |
47 | #endif |
48 | #include "../lib/zstd.h" /* ZSTD_VERSION_STRING, ZSTD_minCLevel, ZSTD_maxCLevel */ |
49 | #include "fileio_asyncio.h" |
50 | |
51 | |
52 | /*-************************************ |
53 | * Constants |
54 | **************************************/ |
55 | #define COMPRESSOR_NAME "Zstandard CLI" |
56 | #ifndef ZSTD_VERSION |
57 | # define ZSTD_VERSION "v" ZSTD_VERSION_STRING |
58 | #endif |
59 | #define AUTHOR "Yann Collet" |
60 | #define WELCOME_MESSAGE "*** %s (%i-bit) %s, by %s ***\n", COMPRESSOR_NAME, (int)(sizeof(size_t)*8), ZSTD_VERSION, AUTHOR |
61 | |
62 | #define ZSTD_ZSTDMT "zstdmt" |
63 | #define ZSTD_UNZSTD "unzstd" |
64 | #define ZSTD_CAT "zstdcat" |
65 | #define ZSTD_ZCAT "zcat" |
66 | #define ZSTD_GZ "gzip" |
67 | #define ZSTD_GUNZIP "gunzip" |
68 | #define ZSTD_GZCAT "gzcat" |
69 | #define ZSTD_LZMA "lzma" |
70 | #define ZSTD_UNLZMA "unlzma" |
71 | #define ZSTD_XZ "xz" |
72 | #define ZSTD_UNXZ "unxz" |
73 | #define ZSTD_LZ4 "lz4" |
74 | #define ZSTD_UNLZ4 "unlz4" |
75 | |
76 | #define KB *(1 <<10) |
77 | #define MB *(1 <<20) |
78 | #define GB *(1U<<30) |
79 | |
80 | #define DISPLAY_LEVEL_DEFAULT 2 |
81 | |
82 | static const char* g_defaultDictName = "dictionary"; |
83 | static const unsigned g_defaultMaxDictSize = 110 KB; |
84 | static const int g_defaultDictCLevel = 3; |
85 | static const unsigned g_defaultSelectivityLevel = 9; |
86 | static const unsigned g_defaultMaxWindowLog = 27; |
87 | #define OVERLAP_LOG_DEFAULT 9999 |
88 | #define LDM_PARAM_DEFAULT 9999 /* Default for parameters where 0 is valid */ |
89 | static U32 g_overlapLog = OVERLAP_LOG_DEFAULT; |
90 | static U32 g_ldmHashLog = 0; |
91 | static U32 g_ldmMinMatch = 0; |
92 | static U32 g_ldmHashRateLog = LDM_PARAM_DEFAULT; |
93 | static U32 g_ldmBucketSizeLog = LDM_PARAM_DEFAULT; |
94 | |
95 | |
96 | #define DEFAULT_ACCEL 1 |
97 | |
98 | typedef enum { cover, fastCover, legacy } dictType; |
99 | |
100 | /*-************************************ |
101 | * Display Macros |
102 | **************************************/ |
103 | #define DISPLAY_F(f, ...) fprintf((f), __VA_ARGS__) |
104 | #define DISPLAYOUT(...) DISPLAY_F(stdout, __VA_ARGS__) |
105 | #define DISPLAY(...) DISPLAY_F(stderr, __VA_ARGS__) |
106 | #define DISPLAYLEVEL(l, ...) { if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } } |
107 | static int g_displayLevel = DISPLAY_LEVEL_DEFAULT; /* 0 : no display, 1: errors, 2 : + result + interaction + warnings, 3 : + progression, 4 : + information */ |
108 | |
109 | |
110 | /*-************************************ |
111 | * Check Version (when CLI linked to dynamic library) |
112 | **************************************/ |
113 | |
114 | /* Due to usage of experimental symbols and capabilities by the CLI, |
115 | * the CLI must be linked against a dynamic library of same version */ |
116 | static void checkLibVersion(void) |
117 | { |
118 | if (strcmp(ZSTD_VERSION_STRING, ZSTD_versionString())) { |
119 | DISPLAYLEVEL(1, "Error : incorrect library version (expecting : %s ; actual : %s ) \n", |
120 | ZSTD_VERSION_STRING, ZSTD_versionString()); |
121 | DISPLAYLEVEL(1, "Please update library to version %s, or use stand-alone zstd binary \n", |
122 | ZSTD_VERSION_STRING); |
123 | exit(1); |
124 | } |
125 | } |
126 | |
127 | |
128 | /*! exeNameMatch() : |
129 | @return : a non-zero value if exeName matches test, excluding the extension |
130 | */ |
131 | static int exeNameMatch(const char* exeName, const char* test) |
132 | { |
133 | return !strncmp(exeName, test, strlen(test)) && |
134 | (exeName[strlen(test)] == '\0' || exeName[strlen(test)] == '.'); |
135 | } |
136 | |
137 | /*-************************************ |
138 | * Command Line |
139 | **************************************/ |
140 | /* print help either in `stderr` or `stdout` depending on originating request |
f535537f |
141 | * error (badUsage) => stderr |
142 | * help (usageAdvanced) => stdout |
648db22b |
143 | */ |
144 | static void usage(FILE* f, const char* programName) |
145 | { |
146 | DISPLAY_F(f, "Compress or decompress the INPUT file(s); reads from STDIN if INPUT is `-` or not provided.\n\n"); |
147 | DISPLAY_F(f, "Usage: %s [OPTIONS...] [INPUT... | -] [-o OUTPUT]\n\n", programName); |
148 | DISPLAY_F(f, "Options:\n"); |
149 | DISPLAY_F(f, " -o OUTPUT Write output to a single file, OUTPUT.\n"); |
150 | DISPLAY_F(f, " -k, --keep Preserve INPUT file(s). [Default] \n"); |
151 | DISPLAY_F(f, " --rm Remove INPUT file(s) after successful (de)compression.\n"); |
152 | #ifdef ZSTD_GZCOMPRESS |
153 | if (exeNameMatch(programName, ZSTD_GZ)) { /* behave like gzip */ |
154 | DISPLAY_F(f, " -n, --no-name Do not store original filename when compressing.\n\n"); |
155 | } |
156 | #endif |
157 | DISPLAY_F(f, "\n"); |
158 | #ifndef ZSTD_NOCOMPRESS |
159 | DISPLAY_F(f, " -# Desired compression level, where `#` is a number between 1 and %d;\n", ZSTDCLI_CLEVEL_MAX); |
160 | DISPLAY_F(f, " lower numbers provide faster compression, higher numbers yield\n"); |
161 | DISPLAY_F(f, " better compression ratios. [Default: %d]\n\n", ZSTDCLI_CLEVEL_DEFAULT); |
162 | #endif |
163 | #ifndef ZSTD_NODECOMPRESS |
164 | DISPLAY_F(f, " -d, --decompress Perform decompression.\n"); |
165 | #endif |
166 | DISPLAY_F(f, " -D DICT Use DICT as the dictionary for compression or decompression.\n\n"); |
167 | DISPLAY_F(f, " -f, --force Disable input and output checks. Allows overwriting existing files,\n"); |
168 | DISPLAY_F(f, " receiving input from the console, printing output to STDOUT, and\n"); |
169 | DISPLAY_F(f, " operating on links, block devices, etc. Unrecognized formats will be\n"); |
170 | DISPLAY_F(f, " passed-through through as-is.\n\n"); |
171 | |
172 | DISPLAY_F(f, " -h Display short usage and exit.\n"); |
173 | DISPLAY_F(f, " -H, --help Display full help and exit.\n"); |
174 | DISPLAY_F(f, " -V, --version Display the program version and exit.\n"); |
175 | DISPLAY_F(f, "\n"); |
176 | } |
177 | |
f535537f |
178 | static void usageAdvanced(const char* programName) |
648db22b |
179 | { |
180 | DISPLAYOUT(WELCOME_MESSAGE); |
181 | DISPLAYOUT("\n"); |
182 | usage(stdout, programName); |
183 | DISPLAYOUT("Advanced options:\n"); |
184 | DISPLAYOUT(" -c, --stdout Write to STDOUT (even if it is a console) and keep the INPUT file(s).\n\n"); |
185 | |
186 | DISPLAYOUT(" -v, --verbose Enable verbose output; pass multiple times to increase verbosity.\n"); |
187 | DISPLAYOUT(" -q, --quiet Suppress warnings; pass twice to suppress errors.\n"); |
188 | #ifndef ZSTD_NOTRACE |
189 | DISPLAYOUT(" --trace LOG Log tracing information to LOG.\n"); |
190 | #endif |
191 | DISPLAYOUT("\n"); |
192 | DISPLAYOUT(" --[no-]progress Forcibly show/hide the progress counter. NOTE: Any (de)compressed\n"); |
193 | DISPLAYOUT(" output to terminal will mix with progress counter text.\n\n"); |
194 | |
195 | #ifdef UTIL_HAS_CREATEFILELIST |
196 | DISPLAYOUT(" -r Operate recursively on directories.\n"); |
197 | DISPLAYOUT(" --filelist LIST Read a list of files to operate on from LIST.\n"); |
198 | DISPLAYOUT(" --output-dir-flat DIR Store processed files in DIR.\n"); |
199 | #endif |
200 | |
201 | #ifdef UTIL_HAS_MIRRORFILELIST |
202 | DISPLAYOUT(" --output-dir-mirror DIR Store processed files in DIR, respecting original directory structure.\n"); |
203 | #endif |
204 | if (AIO_supported()) |
205 | DISPLAYOUT(" --[no-]asyncio Use asynchronous IO. [Default: Enabled]\n"); |
206 | |
207 | DISPLAYOUT("\n"); |
208 | #ifndef ZSTD_NOCOMPRESS |
209 | DISPLAYOUT(" --[no-]check Add XXH64 integrity checksums during compression. [Default: Add, Validate]\n"); |
210 | #ifndef ZSTD_NODECOMPRESS |
211 | DISPLAYOUT(" If `-d` is present, ignore/validate checksums during decompression.\n"); |
212 | #endif |
213 | #else |
214 | #ifdef ZSTD_NOCOMPRESS |
215 | DISPLAYOUT(" --[no-]check Ignore/validate checksums during decompression. [Default: Validate]"); |
216 | #endif |
217 | #endif /* ZSTD_NOCOMPRESS */ |
218 | |
219 | DISPLAYOUT("\n"); |
220 | DISPLAYOUT(" -- Treat remaining arguments after `--` as files.\n"); |
221 | |
222 | #ifndef ZSTD_NOCOMPRESS |
223 | DISPLAYOUT("\n"); |
224 | DISPLAYOUT("Advanced compression options:\n"); |
225 | DISPLAYOUT(" --ultra Enable levels beyond %i, up to %i; requires more memory.\n", ZSTDCLI_CLEVEL_MAX, ZSTD_maxCLevel()); |
226 | DISPLAYOUT(" --fast[=#] Use to very fast compression levels. [Default: %u]\n", 1); |
227 | #ifdef ZSTD_GZCOMPRESS |
228 | if (exeNameMatch(programName, ZSTD_GZ)) { /* behave like gzip */ |
229 | DISPLAYOUT(" --best Compatibility alias for `-9`.\n"); |
230 | } |
231 | #endif |
232 | DISPLAYOUT(" --adapt Dynamically adapt compression level to I/O conditions.\n"); |
233 | DISPLAYOUT(" --long[=#] Enable long distance matching with window log #. [Default: %u]\n", g_defaultMaxWindowLog); |
234 | DISPLAYOUT(" --patch-from=REF Use REF as the reference point for Zstandard's diff engine. \n\n"); |
235 | # ifdef ZSTD_MULTITHREAD |
236 | DISPLAYOUT(" -T# Spawn # compression threads. [Default: 1; pass 0 for core count.]\n"); |
237 | DISPLAYOUT(" --single-thread Share a single thread for I/O and compression (slightly different than `-T1`).\n"); |
238 | DISPLAYOUT(" --auto-threads={physical|logical}\n"); |
239 | DISPLAYOUT(" Use physical/logical cores when using `-T0`. [Default: Physical]\n\n"); |
240 | DISPLAYOUT(" -B# Set job size to #. [Default: 0 (automatic)]\n"); |
241 | DISPLAYOUT(" --rsyncable Compress using a rsync-friendly method (`-B` sets block size). \n"); |
242 | DISPLAYOUT("\n"); |
243 | # endif |
244 | DISPLAYOUT(" --exclude-compressed Only compress files that are not already compressed.\n\n"); |
245 | |
246 | DISPLAYOUT(" --stream-size=# Specify size of streaming input from STDIN.\n"); |
247 | DISPLAYOUT(" --size-hint=# Optimize compression parameters for streaming input of approximately size #.\n"); |
248 | DISPLAYOUT(" --target-compressed-block-size=#\n"); |
249 | DISPLAYOUT(" Generate compressed blocks of approximately # size.\n\n"); |
250 | DISPLAYOUT(" --no-dictID Don't write `dictID` into the header (dictionary compression only).\n"); |
251 | DISPLAYOUT(" --[no-]compress-literals Force (un)compressed literals.\n"); |
252 | DISPLAYOUT(" --[no-]row-match-finder Explicitly enable/disable the fast, row-based matchfinder for\n"); |
253 | DISPLAYOUT(" the 'greedy', 'lazy', and 'lazy2' strategies.\n"); |
254 | |
255 | DISPLAYOUT("\n"); |
256 | DISPLAYOUT(" --format=zstd Compress files to the `.zst` format. [Default]\n"); |
f535537f |
257 | DISPLAYOUT(" --[no-]mmap-dict Memory-map dictionary file rather than mallocing and loading all at once\n"); |
648db22b |
258 | #ifdef ZSTD_GZCOMPRESS |
259 | DISPLAYOUT(" --format=gzip Compress files to the `.gz` format.\n"); |
260 | #endif |
261 | #ifdef ZSTD_LZMACOMPRESS |
262 | DISPLAYOUT(" --format=xz Compress files to the `.xz` format.\n"); |
263 | DISPLAYOUT(" --format=lzma Compress files to the `.lzma` format.\n"); |
264 | #endif |
265 | #ifdef ZSTD_LZ4COMPRESS |
266 | DISPLAYOUT( " --format=lz4 Compress files to the `.lz4` format.\n"); |
267 | #endif |
268 | #endif /* !ZSTD_NOCOMPRESS */ |
269 | |
270 | #ifndef ZSTD_NODECOMPRESS |
271 | DISPLAYOUT("\n"); |
272 | DISPLAYOUT("Advanced decompression options:\n"); |
273 | DISPLAYOUT(" -l Print information about Zstandard-compressed files.\n"); |
274 | DISPLAYOUT(" --test Test compressed file integrity.\n"); |
275 | DISPLAYOUT(" -M# Set the memory usage limit to # megabytes.\n"); |
276 | # if ZSTD_SPARSE_DEFAULT |
277 | DISPLAYOUT(" --[no-]sparse Enable sparse mode. [Default: Enabled for files, disabled for STDOUT.]\n"); |
278 | # else |
279 | DISPLAYOUT(" --[no-]sparse Enable sparse mode. [Default: Disabled]\n"); |
280 | # endif |
281 | { |
282 | char const* passThroughDefault = "Disabled"; |
283 | if (exeNameMatch(programName, ZSTD_CAT) || |
284 | exeNameMatch(programName, ZSTD_ZCAT) || |
285 | exeNameMatch(programName, ZSTD_GZCAT)) { |
286 | passThroughDefault = "Enabled"; |
287 | } |
288 | DISPLAYOUT(" --[no-]pass-through Pass through uncompressed files as-is. [Default: %s]\n", passThroughDefault); |
289 | } |
290 | #endif /* ZSTD_NODECOMPRESS */ |
291 | |
292 | #ifndef ZSTD_NODICT |
293 | DISPLAYOUT("\n"); |
294 | DISPLAYOUT("Dictionary builder:\n"); |
295 | DISPLAYOUT(" --train Create a dictionary from a training set of files.\n\n"); |
296 | DISPLAYOUT(" --train-cover[=k=#,d=#,steps=#,split=#,shrink[=#]]\n"); |
297 | DISPLAYOUT(" Use the cover algorithm (with optional arguments).\n"); |
298 | DISPLAYOUT(" --train-fastcover[=k=#,d=#,f=#,steps=#,split=#,accel=#,shrink[=#]]\n"); |
299 | DISPLAYOUT(" Use the fast cover algorithm (with optional arguments).\n\n"); |
300 | DISPLAYOUT(" --train-legacy[=s=#] Use the legacy algorithm with selectivity #. [Default: %u]\n", g_defaultSelectivityLevel); |
301 | DISPLAYOUT(" -o NAME Use NAME as dictionary name. [Default: %s]\n", g_defaultDictName); |
302 | DISPLAYOUT(" --maxdict=# Limit dictionary to specified size #. [Default: %u]\n", g_defaultMaxDictSize); |
303 | DISPLAYOUT(" --dictID=# Force dictionary ID to #. [Default: Random]\n"); |
304 | #endif |
305 | |
306 | #ifndef ZSTD_NOBENCH |
307 | DISPLAYOUT("\n"); |
308 | DISPLAYOUT("Benchmark options:\n"); |
309 | DISPLAYOUT(" -b# Perform benchmarking with compression level #. [Default: %d]\n", ZSTDCLI_CLEVEL_DEFAULT); |
310 | DISPLAYOUT(" -e# Test all compression levels up to #; starting level is `-b#`. [Default: 1]\n"); |
311 | DISPLAYOUT(" -i# Set the minimum evaluation to time # seconds. [Default: 3]\n"); |
312 | DISPLAYOUT(" -B# Cut file into independent chunks of size #. [Default: No chunking]\n"); |
313 | DISPLAYOUT(" -S Output one benchmark result per input file. [Default: Consolidated result]\n"); |
314 | DISPLAYOUT(" --priority=rt Set process priority to real-time.\n"); |
315 | #endif |
316 | |
317 | } |
318 | |
f535537f |
319 | static void badUsage(const char* programName, const char* parameter) |
648db22b |
320 | { |
f535537f |
321 | DISPLAYLEVEL(1, "Incorrect parameter: %s \n", parameter); |
648db22b |
322 | if (g_displayLevel >= 2) usage(stderr, programName); |
323 | } |
324 | |
325 | static void waitEnter(void) |
326 | { |
327 | int unused; |
328 | DISPLAY("Press enter to continue... \n"); |
329 | unused = getchar(); |
330 | (void)unused; |
331 | } |
332 | |
333 | static const char* lastNameFromPath(const char* path) |
334 | { |
335 | const char* name = path; |
336 | if (strrchr(name, '/')) name = strrchr(name, '/') + 1; |
337 | if (strrchr(name, '\\')) name = strrchr(name, '\\') + 1; /* windows */ |
338 | return name; |
339 | } |
340 | |
341 | static void errorOut(const char* msg) |
342 | { |
343 | DISPLAYLEVEL(1, "%s \n", msg); exit(1); |
344 | } |
345 | |
346 | /*! readU32FromCharChecked() : |
347 | * @return 0 if success, and store the result in *value. |
348 | * allows and interprets K, KB, KiB, M, MB and MiB suffix. |
349 | * Will also modify `*stringPtr`, advancing it to position where it stopped reading. |
350 | * @return 1 if an overflow error occurs */ |
351 | static int readU32FromCharChecked(const char** stringPtr, unsigned* value) |
352 | { |
353 | unsigned result = 0; |
354 | while ((**stringPtr >='0') && (**stringPtr <='9')) { |
355 | unsigned const max = ((unsigned)(-1)) / 10; |
356 | unsigned last = result; |
357 | if (result > max) return 1; /* overflow error */ |
358 | result *= 10; |
359 | result += (unsigned)(**stringPtr - '0'); |
360 | if (result < last) return 1; /* overflow error */ |
361 | (*stringPtr)++ ; |
362 | } |
363 | if ((**stringPtr=='K') || (**stringPtr=='M')) { |
364 | unsigned const maxK = ((unsigned)(-1)) >> 10; |
365 | if (result > maxK) return 1; /* overflow error */ |
366 | result <<= 10; |
367 | if (**stringPtr=='M') { |
368 | if (result > maxK) return 1; /* overflow error */ |
369 | result <<= 10; |
370 | } |
371 | (*stringPtr)++; /* skip `K` or `M` */ |
372 | if (**stringPtr=='i') (*stringPtr)++; |
373 | if (**stringPtr=='B') (*stringPtr)++; |
374 | } |
375 | *value = result; |
376 | return 0; |
377 | } |
378 | |
379 | /*! readU32FromChar() : |
380 | * @return : unsigned integer value read from input in `char` format. |
381 | * allows and interprets K, KB, KiB, M, MB and MiB suffix. |
382 | * Will also modify `*stringPtr`, advancing it to position where it stopped reading. |
383 | * Note : function will exit() program if digit sequence overflows */ |
384 | static unsigned readU32FromChar(const char** stringPtr) { |
385 | static const char errorMsg[] = "error: numeric value overflows 32-bit unsigned int"; |
386 | unsigned result; |
387 | if (readU32FromCharChecked(stringPtr, &result)) { errorOut(errorMsg); } |
388 | return result; |
389 | } |
390 | |
391 | /*! readIntFromChar() : |
392 | * @return : signed integer value read from input in `char` format. |
393 | * allows and interprets K, KB, KiB, M, MB and MiB suffix. |
394 | * Will also modify `*stringPtr`, advancing it to position where it stopped reading. |
395 | * Note : function will exit() program if digit sequence overflows */ |
396 | static int readIntFromChar(const char** stringPtr) { |
397 | static const char errorMsg[] = "error: numeric value overflows 32-bit int"; |
398 | int sign = 1; |
399 | unsigned result; |
400 | if (**stringPtr=='-') { |
401 | (*stringPtr)++; |
402 | sign = -1; |
403 | } |
404 | if (readU32FromCharChecked(stringPtr, &result)) { errorOut(errorMsg); } |
405 | return (int) result * sign; |
406 | } |
407 | |
408 | /*! readSizeTFromCharChecked() : |
409 | * @return 0 if success, and store the result in *value. |
410 | * allows and interprets K, KB, KiB, M, MB and MiB suffix. |
411 | * Will also modify `*stringPtr`, advancing it to position where it stopped reading. |
412 | * @return 1 if an overflow error occurs */ |
413 | static int readSizeTFromCharChecked(const char** stringPtr, size_t* value) |
414 | { |
415 | size_t result = 0; |
416 | while ((**stringPtr >='0') && (**stringPtr <='9')) { |
417 | size_t const max = ((size_t)(-1)) / 10; |
418 | size_t last = result; |
419 | if (result > max) return 1; /* overflow error */ |
420 | result *= 10; |
421 | result += (size_t)(**stringPtr - '0'); |
422 | if (result < last) return 1; /* overflow error */ |
423 | (*stringPtr)++ ; |
424 | } |
425 | if ((**stringPtr=='K') || (**stringPtr=='M')) { |
426 | size_t const maxK = ((size_t)(-1)) >> 10; |
427 | if (result > maxK) return 1; /* overflow error */ |
428 | result <<= 10; |
429 | if (**stringPtr=='M') { |
430 | if (result > maxK) return 1; /* overflow error */ |
431 | result <<= 10; |
432 | } |
433 | (*stringPtr)++; /* skip `K` or `M` */ |
434 | if (**stringPtr=='i') (*stringPtr)++; |
435 | if (**stringPtr=='B') (*stringPtr)++; |
436 | } |
437 | *value = result; |
438 | return 0; |
439 | } |
440 | |
441 | /*! readSizeTFromChar() : |
442 | * @return : size_t value read from input in `char` format. |
443 | * allows and interprets K, KB, KiB, M, MB and MiB suffix. |
444 | * Will also modify `*stringPtr`, advancing it to position where it stopped reading. |
445 | * Note : function will exit() program if digit sequence overflows */ |
446 | static size_t readSizeTFromChar(const char** stringPtr) { |
447 | static const char errorMsg[] = "error: numeric value overflows size_t"; |
448 | size_t result; |
449 | if (readSizeTFromCharChecked(stringPtr, &result)) { errorOut(errorMsg); } |
450 | return result; |
451 | } |
452 | |
453 | /** longCommandWArg() : |
454 | * check if *stringPtr is the same as longCommand. |
455 | * If yes, @return 1 and advances *stringPtr to the position which immediately follows longCommand. |
456 | * @return 0 and doesn't modify *stringPtr otherwise. |
457 | */ |
458 | static int longCommandWArg(const char** stringPtr, const char* longCommand) |
459 | { |
460 | size_t const comSize = strlen(longCommand); |
461 | int const result = !strncmp(*stringPtr, longCommand, comSize); |
462 | if (result) *stringPtr += comSize; |
463 | return result; |
464 | } |
465 | |
466 | |
467 | #ifndef ZSTD_NODICT |
468 | |
469 | static const unsigned kDefaultRegression = 1; |
470 | /** |
471 | * parseCoverParameters() : |
472 | * reads cover parameters from *stringPtr (e.g. "--train-cover=k=48,d=8,steps=32") into *params |
473 | * @return 1 means that cover parameters were correct |
474 | * @return 0 in case of malformed parameters |
475 | */ |
476 | static unsigned parseCoverParameters(const char* stringPtr, ZDICT_cover_params_t* params) |
477 | { |
478 | memset(params, 0, sizeof(*params)); |
479 | for (; ;) { |
480 | if (longCommandWArg(&stringPtr, "k=")) { params->k = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
481 | if (longCommandWArg(&stringPtr, "d=")) { params->d = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
482 | if (longCommandWArg(&stringPtr, "steps=")) { params->steps = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
483 | if (longCommandWArg(&stringPtr, "split=")) { |
484 | unsigned splitPercentage = readU32FromChar(&stringPtr); |
485 | params->splitPoint = (double)splitPercentage / 100.0; |
486 | if (stringPtr[0]==',') { stringPtr++; continue; } else break; |
487 | } |
488 | if (longCommandWArg(&stringPtr, "shrink")) { |
489 | params->shrinkDictMaxRegression = kDefaultRegression; |
490 | params->shrinkDict = 1; |
491 | if (stringPtr[0]=='=') { |
492 | stringPtr++; |
493 | params->shrinkDictMaxRegression = readU32FromChar(&stringPtr); |
494 | } |
495 | if (stringPtr[0]==',') { |
496 | stringPtr++; |
497 | continue; |
498 | } |
499 | else break; |
500 | } |
501 | return 0; |
502 | } |
503 | if (stringPtr[0] != 0) return 0; |
504 | DISPLAYLEVEL(4, "cover: k=%u\nd=%u\nsteps=%u\nsplit=%u\nshrink%u\n", params->k, params->d, params->steps, (unsigned)(params->splitPoint * 100), params->shrinkDictMaxRegression); |
505 | return 1; |
506 | } |
507 | |
508 | /** |
509 | * parseFastCoverParameters() : |
510 | * reads fastcover parameters from *stringPtr (e.g. "--train-fastcover=k=48,d=8,f=20,steps=32,accel=2") into *params |
511 | * @return 1 means that fastcover parameters were correct |
512 | * @return 0 in case of malformed parameters |
513 | */ |
514 | static unsigned parseFastCoverParameters(const char* stringPtr, ZDICT_fastCover_params_t* params) |
515 | { |
516 | memset(params, 0, sizeof(*params)); |
517 | for (; ;) { |
518 | if (longCommandWArg(&stringPtr, "k=")) { params->k = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
519 | if (longCommandWArg(&stringPtr, "d=")) { params->d = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
520 | if (longCommandWArg(&stringPtr, "f=")) { params->f = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
521 | if (longCommandWArg(&stringPtr, "steps=")) { params->steps = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
522 | if (longCommandWArg(&stringPtr, "accel=")) { params->accel = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
523 | if (longCommandWArg(&stringPtr, "split=")) { |
524 | unsigned splitPercentage = readU32FromChar(&stringPtr); |
525 | params->splitPoint = (double)splitPercentage / 100.0; |
526 | if (stringPtr[0]==',') { stringPtr++; continue; } else break; |
527 | } |
528 | if (longCommandWArg(&stringPtr, "shrink")) { |
529 | params->shrinkDictMaxRegression = kDefaultRegression; |
530 | params->shrinkDict = 1; |
531 | if (stringPtr[0]=='=') { |
532 | stringPtr++; |
533 | params->shrinkDictMaxRegression = readU32FromChar(&stringPtr); |
534 | } |
535 | if (stringPtr[0]==',') { |
536 | stringPtr++; |
537 | continue; |
538 | } |
539 | else break; |
540 | } |
541 | return 0; |
542 | } |
543 | if (stringPtr[0] != 0) return 0; |
544 | DISPLAYLEVEL(4, "cover: k=%u\nd=%u\nf=%u\nsteps=%u\nsplit=%u\naccel=%u\nshrink=%u\n", params->k, params->d, params->f, params->steps, (unsigned)(params->splitPoint * 100), params->accel, params->shrinkDictMaxRegression); |
545 | return 1; |
546 | } |
547 | |
548 | /** |
549 | * parseLegacyParameters() : |
550 | * reads legacy dictionary builder parameters from *stringPtr (e.g. "--train-legacy=selectivity=8") into *selectivity |
551 | * @return 1 means that legacy dictionary builder parameters were correct |
552 | * @return 0 in case of malformed parameters |
553 | */ |
554 | static unsigned parseLegacyParameters(const char* stringPtr, unsigned* selectivity) |
555 | { |
556 | if (!longCommandWArg(&stringPtr, "s=") && !longCommandWArg(&stringPtr, "selectivity=")) { return 0; } |
557 | *selectivity = readU32FromChar(&stringPtr); |
558 | if (stringPtr[0] != 0) return 0; |
559 | DISPLAYLEVEL(4, "legacy: selectivity=%u\n", *selectivity); |
560 | return 1; |
561 | } |
562 | |
563 | static ZDICT_cover_params_t defaultCoverParams(void) |
564 | { |
565 | ZDICT_cover_params_t params; |
566 | memset(¶ms, 0, sizeof(params)); |
567 | params.d = 8; |
568 | params.steps = 4; |
569 | params.splitPoint = 1.0; |
570 | params.shrinkDict = 0; |
571 | params.shrinkDictMaxRegression = kDefaultRegression; |
572 | return params; |
573 | } |
574 | |
575 | static ZDICT_fastCover_params_t defaultFastCoverParams(void) |
576 | { |
577 | ZDICT_fastCover_params_t params; |
578 | memset(¶ms, 0, sizeof(params)); |
579 | params.d = 8; |
580 | params.f = 20; |
581 | params.steps = 4; |
582 | params.splitPoint = 0.75; /* different from default splitPoint of cover */ |
583 | params.accel = DEFAULT_ACCEL; |
584 | params.shrinkDict = 0; |
585 | params.shrinkDictMaxRegression = kDefaultRegression; |
586 | return params; |
587 | } |
588 | #endif |
589 | |
590 | |
591 | /** parseAdaptParameters() : |
f535537f |
592 | * reads adapt parameters from *stringPtr (e.g. "--adapt=min=1,max=19) and store them into adaptMinPtr and adaptMaxPtr. |
648db22b |
593 | * Both adaptMinPtr and adaptMaxPtr must be already allocated and correctly initialized. |
594 | * There is no guarantee that any of these values will be updated. |
595 | * @return 1 means that parsing was successful, |
596 | * @return 0 in case of malformed parameters |
597 | */ |
598 | static unsigned parseAdaptParameters(const char* stringPtr, int* adaptMinPtr, int* adaptMaxPtr) |
599 | { |
600 | for ( ; ;) { |
601 | if (longCommandWArg(&stringPtr, "min=")) { *adaptMinPtr = readIntFromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
602 | if (longCommandWArg(&stringPtr, "max=")) { *adaptMaxPtr = readIntFromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
603 | DISPLAYLEVEL(4, "invalid compression parameter \n"); |
604 | return 0; |
605 | } |
606 | if (stringPtr[0] != 0) return 0; /* check the end of string */ |
607 | if (*adaptMinPtr > *adaptMaxPtr) { |
608 | DISPLAYLEVEL(4, "incoherent adaptation limits \n"); |
609 | return 0; |
610 | } |
611 | return 1; |
612 | } |
613 | |
614 | |
615 | /** parseCompressionParameters() : |
616 | * reads compression parameters from *stringPtr (e.g. "--zstd=wlog=23,clog=23,hlog=22,slog=6,mml=3,tlen=48,strat=6") into *params |
617 | * @return 1 means that compression parameters were correct |
618 | * @return 0 in case of malformed parameters |
619 | */ |
620 | static unsigned parseCompressionParameters(const char* stringPtr, ZSTD_compressionParameters* params) |
621 | { |
622 | for ( ; ;) { |
623 | if (longCommandWArg(&stringPtr, "windowLog=") || longCommandWArg(&stringPtr, "wlog=")) { params->windowLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
624 | if (longCommandWArg(&stringPtr, "chainLog=") || longCommandWArg(&stringPtr, "clog=")) { params->chainLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
625 | if (longCommandWArg(&stringPtr, "hashLog=") || longCommandWArg(&stringPtr, "hlog=")) { params->hashLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
626 | if (longCommandWArg(&stringPtr, "searchLog=") || longCommandWArg(&stringPtr, "slog=")) { params->searchLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
627 | if (longCommandWArg(&stringPtr, "minMatch=") || longCommandWArg(&stringPtr, "mml=")) { params->minMatch = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
628 | if (longCommandWArg(&stringPtr, "targetLength=") || longCommandWArg(&stringPtr, "tlen=")) { params->targetLength = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
629 | if (longCommandWArg(&stringPtr, "strategy=") || longCommandWArg(&stringPtr, "strat=")) { params->strategy = (ZSTD_strategy)(readU32FromChar(&stringPtr)); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
630 | if (longCommandWArg(&stringPtr, "overlapLog=") || longCommandWArg(&stringPtr, "ovlog=")) { g_overlapLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
631 | if (longCommandWArg(&stringPtr, "ldmHashLog=") || longCommandWArg(&stringPtr, "lhlog=")) { g_ldmHashLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
632 | if (longCommandWArg(&stringPtr, "ldmMinMatch=") || longCommandWArg(&stringPtr, "lmml=")) { g_ldmMinMatch = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
633 | if (longCommandWArg(&stringPtr, "ldmBucketSizeLog=") || longCommandWArg(&stringPtr, "lblog=")) { g_ldmBucketSizeLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
634 | if (longCommandWArg(&stringPtr, "ldmHashRateLog=") || longCommandWArg(&stringPtr, "lhrlog=")) { g_ldmHashRateLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } |
635 | DISPLAYLEVEL(4, "invalid compression parameter \n"); |
636 | return 0; |
637 | } |
638 | |
639 | DISPLAYLEVEL(4, "windowLog=%d, chainLog=%d, hashLog=%d, searchLog=%d \n", params->windowLog, params->chainLog, params->hashLog, params->searchLog); |
640 | DISPLAYLEVEL(4, "minMatch=%d, targetLength=%d, strategy=%d \n", params->minMatch, params->targetLength, params->strategy); |
641 | if (stringPtr[0] != 0) return 0; /* check the end of string */ |
642 | return 1; |
643 | } |
644 | |
645 | static void printVersion(void) |
646 | { |
647 | if (g_displayLevel < DISPLAY_LEVEL_DEFAULT) { |
648 | DISPLAYOUT("%s\n", ZSTD_VERSION_STRING); |
649 | return; |
650 | } |
651 | |
652 | DISPLAYOUT(WELCOME_MESSAGE); |
653 | if (g_displayLevel >= 3) { |
654 | /* format support */ |
655 | DISPLAYOUT("*** supports: zstd"); |
656 | #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>0) && (ZSTD_LEGACY_SUPPORT<8) |
657 | DISPLAYOUT(", zstd legacy v0.%d+", ZSTD_LEGACY_SUPPORT); |
658 | #endif |
659 | #ifdef ZSTD_GZCOMPRESS |
660 | DISPLAYOUT(", gzip"); |
661 | #endif |
662 | #ifdef ZSTD_LZ4COMPRESS |
663 | DISPLAYOUT(", lz4"); |
664 | #endif |
665 | #ifdef ZSTD_LZMACOMPRESS |
666 | DISPLAYOUT(", lzma, xz "); |
667 | #endif |
668 | DISPLAYOUT("\n"); |
669 | if (g_displayLevel >= 4) { |
670 | /* library versions */ |
671 | DISPLAYOUT("zlib version %s\n", FIO_zlibVersion()); |
672 | DISPLAYOUT("lz4 version %s\n", FIO_lz4Version()); |
673 | DISPLAYOUT("lzma version %s\n", FIO_lzmaVersion()); |
674 | |
675 | /* posix support */ |
676 | #ifdef _POSIX_C_SOURCE |
677 | DISPLAYOUT("_POSIX_C_SOURCE defined: %ldL\n", (long) _POSIX_C_SOURCE); |
678 | #endif |
679 | #ifdef _POSIX_VERSION |
680 | DISPLAYOUT("_POSIX_VERSION defined: %ldL \n", (long) _POSIX_VERSION); |
681 | #endif |
682 | #ifdef PLATFORM_POSIX_VERSION |
683 | DISPLAYOUT("PLATFORM_POSIX_VERSION defined: %ldL\n", (long) PLATFORM_POSIX_VERSION); |
684 | #endif |
685 | } } |
686 | } |
687 | |
688 | #define ZSTD_NB_STRATEGIES 9 |
689 | static const char* ZSTD_strategyMap[ZSTD_NB_STRATEGIES + 1] = { "", "ZSTD_fast", |
690 | "ZSTD_dfast", "ZSTD_greedy", "ZSTD_lazy", "ZSTD_lazy2", "ZSTD_btlazy2", |
691 | "ZSTD_btopt", "ZSTD_btultra", "ZSTD_btultra2"}; |
692 | |
693 | #ifndef ZSTD_NOCOMPRESS |
694 | |
695 | static void printDefaultCParams(const char* filename, const char* dictFileName, int cLevel) { |
696 | unsigned long long fileSize = UTIL_getFileSize(filename); |
697 | const size_t dictSize = dictFileName != NULL ? (size_t)UTIL_getFileSize(dictFileName) : 0; |
698 | const ZSTD_compressionParameters cParams = ZSTD_getCParams(cLevel, fileSize, dictSize); |
699 | if (fileSize != UTIL_FILESIZE_UNKNOWN) DISPLAY("%s (%u bytes)\n", filename, (unsigned)fileSize); |
700 | else DISPLAY("%s (src size unknown)\n", filename); |
701 | DISPLAY(" - windowLog : %u\n", cParams.windowLog); |
702 | DISPLAY(" - chainLog : %u\n", cParams.chainLog); |
703 | DISPLAY(" - hashLog : %u\n", cParams.hashLog); |
704 | DISPLAY(" - searchLog : %u\n", cParams.searchLog); |
705 | DISPLAY(" - minMatch : %u\n", cParams.minMatch); |
706 | DISPLAY(" - targetLength : %u\n", cParams.targetLength); |
707 | assert(cParams.strategy < ZSTD_NB_STRATEGIES + 1); |
708 | DISPLAY(" - strategy : %s (%u)\n", ZSTD_strategyMap[(int)cParams.strategy], (unsigned)cParams.strategy); |
709 | } |
710 | |
711 | static void printActualCParams(const char* filename, const char* dictFileName, int cLevel, const ZSTD_compressionParameters* cParams) { |
712 | unsigned long long fileSize = UTIL_getFileSize(filename); |
713 | const size_t dictSize = dictFileName != NULL ? (size_t)UTIL_getFileSize(dictFileName) : 0; |
714 | ZSTD_compressionParameters actualCParams = ZSTD_getCParams(cLevel, fileSize, dictSize); |
715 | assert(g_displayLevel >= 4); |
716 | actualCParams.windowLog = cParams->windowLog == 0 ? actualCParams.windowLog : cParams->windowLog; |
717 | actualCParams.chainLog = cParams->chainLog == 0 ? actualCParams.chainLog : cParams->chainLog; |
718 | actualCParams.hashLog = cParams->hashLog == 0 ? actualCParams.hashLog : cParams->hashLog; |
719 | actualCParams.searchLog = cParams->searchLog == 0 ? actualCParams.searchLog : cParams->searchLog; |
720 | actualCParams.minMatch = cParams->minMatch == 0 ? actualCParams.minMatch : cParams->minMatch; |
721 | actualCParams.targetLength = cParams->targetLength == 0 ? actualCParams.targetLength : cParams->targetLength; |
722 | actualCParams.strategy = cParams->strategy == 0 ? actualCParams.strategy : cParams->strategy; |
723 | DISPLAY("--zstd=wlog=%d,clog=%d,hlog=%d,slog=%d,mml=%d,tlen=%d,strat=%d\n", |
724 | actualCParams.windowLog, actualCParams.chainLog, actualCParams.hashLog, actualCParams.searchLog, |
725 | actualCParams.minMatch, actualCParams.targetLength, actualCParams.strategy); |
726 | } |
727 | |
728 | #endif |
729 | |
730 | /* Environment variables for parameter setting */ |
731 | #define ENV_CLEVEL "ZSTD_CLEVEL" |
732 | #define ENV_NBTHREADS "ZSTD_NBTHREADS" /* takes lower precedence than directly specifying -T# in the CLI */ |
733 | |
734 | /* pick up environment variable */ |
735 | static int init_cLevel(void) { |
736 | const char* const env = getenv(ENV_CLEVEL); |
737 | if (env != NULL) { |
738 | const char* ptr = env; |
739 | int sign = 1; |
740 | if (*ptr == '-') { |
741 | sign = -1; |
742 | ptr++; |
743 | } else if (*ptr == '+') { |
744 | ptr++; |
745 | } |
746 | |
747 | if ((*ptr>='0') && (*ptr<='9')) { |
748 | unsigned absLevel; |
749 | if (readU32FromCharChecked(&ptr, &absLevel)) { |
750 | DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: numeric value too large \n", ENV_CLEVEL, env); |
751 | return ZSTDCLI_CLEVEL_DEFAULT; |
752 | } else if (*ptr == 0) { |
753 | return sign * (int)absLevel; |
754 | } } |
755 | |
756 | DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: not a valid integer value \n", ENV_CLEVEL, env); |
757 | } |
758 | |
759 | return ZSTDCLI_CLEVEL_DEFAULT; |
760 | } |
761 | |
762 | #ifdef ZSTD_MULTITHREAD |
763 | static unsigned init_nbThreads(void) { |
764 | const char* const env = getenv(ENV_NBTHREADS); |
765 | if (env != NULL) { |
766 | const char* ptr = env; |
767 | if ((*ptr>='0') && (*ptr<='9')) { |
768 | unsigned nbThreads; |
769 | if (readU32FromCharChecked(&ptr, &nbThreads)) { |
770 | DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: numeric value too large \n", ENV_NBTHREADS, env); |
771 | return ZSTDCLI_NBTHREADS_DEFAULT; |
772 | } else if (*ptr == 0) { |
773 | return nbThreads; |
774 | } |
775 | } |
776 | DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: not a valid unsigned value \n", ENV_NBTHREADS, env); |
777 | } |
778 | |
779 | return ZSTDCLI_NBTHREADS_DEFAULT; |
780 | } |
781 | #endif |
782 | |
783 | #define NEXT_FIELD(ptr) { \ |
784 | if (*argument == '=') { \ |
785 | ptr = ++argument; \ |
786 | argument += strlen(ptr); \ |
787 | } else { \ |
788 | argNb++; \ |
789 | if (argNb >= argCount) { \ |
790 | DISPLAYLEVEL(1, "error: missing command argument \n"); \ |
791 | CLEAN_RETURN(1); \ |
792 | } \ |
793 | ptr = argv[argNb]; \ |
794 | assert(ptr != NULL); \ |
795 | if (ptr[0]=='-') { \ |
796 | DISPLAYLEVEL(1, "error: command cannot be separated from its argument by another command \n"); \ |
797 | CLEAN_RETURN(1); \ |
798 | } } } |
799 | |
800 | #define NEXT_UINT32(val32) { \ |
801 | const char* __nb; \ |
802 | NEXT_FIELD(__nb); \ |
803 | val32 = readU32FromChar(&__nb); \ |
804 | if(*__nb != 0) { \ |
805 | errorOut("error: only numeric values with optional suffixes K, KB, KiB, M, MB, MiB are allowed"); \ |
806 | } \ |
807 | } |
808 | |
809 | #define NEXT_TSIZE(valTsize) { \ |
810 | const char* __nb; \ |
811 | NEXT_FIELD(__nb); \ |
812 | valTsize = readSizeTFromChar(&__nb); \ |
813 | if(*__nb != 0) { \ |
814 | errorOut("error: only numeric values with optional suffixes K, KB, KiB, M, MB, MiB are allowed"); \ |
815 | } \ |
816 | } |
817 | |
818 | typedef enum { zom_compress, zom_decompress, zom_test, zom_bench, zom_train, zom_list } zstd_operation_mode; |
819 | |
820 | #define CLEAN_RETURN(i) { operationResult = (i); goto _end; } |
821 | |
822 | #ifdef ZSTD_NOCOMPRESS |
823 | /* symbols from compression library are not defined and should not be invoked */ |
824 | # define MINCLEVEL -99 |
825 | # define MAXCLEVEL 22 |
826 | #else |
827 | # define MINCLEVEL ZSTD_minCLevel() |
828 | # define MAXCLEVEL ZSTD_maxCLevel() |
829 | #endif |
830 | |
831 | int main(int argCount, const char* argv[]) |
832 | { |
833 | int argNb, |
834 | followLinks = 0, |
835 | allowBlockDevices = 0, |
836 | forceStdin = 0, |
837 | forceStdout = 0, |
838 | hasStdout = 0, |
839 | ldmFlag = 0, |
840 | main_pause = 0, |
841 | adapt = 0, |
842 | adaptMin = MINCLEVEL, |
843 | adaptMax = MAXCLEVEL, |
844 | rsyncable = 0, |
845 | nextArgumentsAreFiles = 0, |
846 | operationResult = 0, |
847 | separateFiles = 0, |
848 | setRealTimePrio = 0, |
849 | singleThread = 0, |
850 | defaultLogicalCores = 0, |
851 | showDefaultCParams = 0, |
852 | ultra=0, |
853 | contentSize=1, |
854 | removeSrcFile=0; |
855 | ZSTD_paramSwitch_e mmapDict=ZSTD_ps_auto; |
856 | ZSTD_paramSwitch_e useRowMatchFinder = ZSTD_ps_auto; |
857 | FIO_compressionType_t cType = FIO_zstdCompression; |
858 | unsigned nbWorkers = 0; |
f535537f |
859 | double compressibility = -1.0; /* lorem ipsum generator */ |
648db22b |
860 | unsigned bench_nbSeconds = 3; /* would be better if this value was synchronized from bench */ |
861 | size_t blockSize = 0; |
862 | |
863 | FIO_prefs_t* const prefs = FIO_createPreferences(); |
864 | FIO_ctx_t* const fCtx = FIO_createContext(); |
865 | FIO_progressSetting_e progress = FIO_ps_auto; |
866 | zstd_operation_mode operation = zom_compress; |
867 | ZSTD_compressionParameters compressionParams; |
868 | int cLevel = init_cLevel(); |
869 | int cLevelLast = MINCLEVEL - 1; /* lower than minimum */ |
870 | unsigned recursive = 0; |
871 | unsigned memLimit = 0; |
872 | FileNamesTable* filenames = UTIL_allocateFileNamesTable((size_t)argCount); /* argCount >= 1 */ |
873 | FileNamesTable* file_of_names = UTIL_allocateFileNamesTable((size_t)argCount); /* argCount >= 1 */ |
874 | const char* programName = argv[0]; |
875 | const char* outFileName = NULL; |
876 | const char* outDirName = NULL; |
877 | const char* outMirroredDirName = NULL; |
878 | const char* dictFileName = NULL; |
879 | const char* patchFromDictFileName = NULL; |
880 | const char* suffix = ZSTD_EXTENSION; |
881 | unsigned maxDictSize = g_defaultMaxDictSize; |
882 | unsigned dictID = 0; |
883 | size_t streamSrcSize = 0; |
884 | size_t targetCBlockSize = 0; |
885 | size_t srcSizeHint = 0; |
886 | size_t nbInputFileNames = 0; |
887 | int dictCLevel = g_defaultDictCLevel; |
888 | unsigned dictSelect = g_defaultSelectivityLevel; |
889 | #ifndef ZSTD_NODICT |
890 | ZDICT_cover_params_t coverParams = defaultCoverParams(); |
891 | ZDICT_fastCover_params_t fastCoverParams = defaultFastCoverParams(); |
892 | dictType dict = fastCover; |
893 | #endif |
894 | #ifndef ZSTD_NOBENCH |
895 | BMK_advancedParams_t benchParams = BMK_initAdvancedParams(); |
896 | #endif |
897 | ZSTD_paramSwitch_e literalCompressionMode = ZSTD_ps_auto; |
898 | |
899 | |
900 | /* init */ |
901 | checkLibVersion(); |
902 | (void)recursive; (void)cLevelLast; /* not used when ZSTD_NOBENCH set */ |
903 | (void)memLimit; |
904 | assert(argCount >= 1); |
905 | if ((filenames==NULL) || (file_of_names==NULL)) { DISPLAYLEVEL(1, "zstd: allocation error \n"); exit(1); } |
906 | programName = lastNameFromPath(programName); |
907 | #ifdef ZSTD_MULTITHREAD |
908 | nbWorkers = init_nbThreads(); |
909 | #endif |
910 | |
911 | /* preset behaviors */ |
912 | if (exeNameMatch(programName, ZSTD_ZSTDMT)) nbWorkers=0, singleThread=0; |
913 | if (exeNameMatch(programName, ZSTD_UNZSTD)) operation=zom_decompress; |
914 | if (exeNameMatch(programName, ZSTD_CAT)) { operation=zom_decompress; FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; FIO_setPassThroughFlag(prefs, 1); outFileName=stdoutmark; g_displayLevel=1; } /* supports multiple formats */ |
915 | if (exeNameMatch(programName, ZSTD_ZCAT)) { operation=zom_decompress; FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; FIO_setPassThroughFlag(prefs, 1); outFileName=stdoutmark; g_displayLevel=1; } /* behave like zcat, also supports multiple formats */ |
916 | if (exeNameMatch(programName, ZSTD_GZ)) { /* behave like gzip */ |
917 | suffix = GZ_EXTENSION; cType = FIO_gzipCompression; removeSrcFile=1; |
918 | dictCLevel = cLevel = 6; /* gzip default is -6 */ |
919 | } |
920 | if (exeNameMatch(programName, ZSTD_GUNZIP)) { operation=zom_decompress; removeSrcFile=1; } /* behave like gunzip, also supports multiple formats */ |
921 | if (exeNameMatch(programName, ZSTD_GZCAT)) { operation=zom_decompress; FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; FIO_setPassThroughFlag(prefs, 1); outFileName=stdoutmark; g_displayLevel=1; } /* behave like gzcat, also supports multiple formats */ |
922 | if (exeNameMatch(programName, ZSTD_LZMA)) { suffix = LZMA_EXTENSION; cType = FIO_lzmaCompression; removeSrcFile=1; } /* behave like lzma */ |
923 | if (exeNameMatch(programName, ZSTD_UNLZMA)) { operation=zom_decompress; cType = FIO_lzmaCompression; removeSrcFile=1; } /* behave like unlzma, also supports multiple formats */ |
924 | if (exeNameMatch(programName, ZSTD_XZ)) { suffix = XZ_EXTENSION; cType = FIO_xzCompression; removeSrcFile=1; } /* behave like xz */ |
925 | if (exeNameMatch(programName, ZSTD_UNXZ)) { operation=zom_decompress; cType = FIO_xzCompression; removeSrcFile=1; } /* behave like unxz, also supports multiple formats */ |
926 | if (exeNameMatch(programName, ZSTD_LZ4)) { suffix = LZ4_EXTENSION; cType = FIO_lz4Compression; } /* behave like lz4 */ |
927 | if (exeNameMatch(programName, ZSTD_UNLZ4)) { operation=zom_decompress; cType = FIO_lz4Compression; } /* behave like unlz4, also supports multiple formats */ |
928 | memset(&compressionParams, 0, sizeof(compressionParams)); |
929 | |
930 | /* init crash handler */ |
931 | FIO_addAbortHandler(); |
932 | |
933 | /* command switches */ |
934 | for (argNb=1; argNb<argCount; argNb++) { |
935 | const char* argument = argv[argNb]; |
f535537f |
936 | const char* const originalArgument = argument; |
648db22b |
937 | if (!argument) continue; /* Protection if argument empty */ |
938 | |
939 | if (nextArgumentsAreFiles) { |
940 | UTIL_refFilename(filenames, argument); |
941 | continue; |
942 | } |
943 | |
944 | /* "-" means stdin/stdout */ |
945 | if (!strcmp(argument, "-")){ |
946 | UTIL_refFilename(filenames, stdinmark); |
947 | continue; |
948 | } |
949 | |
950 | /* Decode commands (note : aggregated commands are allowed) */ |
951 | if (argument[0]=='-') { |
952 | |
953 | if (argument[1]=='-') { |
954 | /* long commands (--long-word) */ |
955 | if (!strcmp(argument, "--")) { nextArgumentsAreFiles=1; continue; } /* only file names allowed from now on */ |
956 | if (!strcmp(argument, "--list")) { operation=zom_list; continue; } |
957 | if (!strcmp(argument, "--compress")) { operation=zom_compress; continue; } |
958 | if (!strcmp(argument, "--decompress")) { operation=zom_decompress; continue; } |
959 | if (!strcmp(argument, "--uncompress")) { operation=zom_decompress; continue; } |
960 | if (!strcmp(argument, "--force")) { FIO_overwriteMode(prefs); forceStdin=1; forceStdout=1; followLinks=1; allowBlockDevices=1; continue; } |
961 | if (!strcmp(argument, "--version")) { printVersion(); CLEAN_RETURN(0); } |
f535537f |
962 | if (!strcmp(argument, "--help")) { usageAdvanced(programName); CLEAN_RETURN(0); } |
648db22b |
963 | if (!strcmp(argument, "--verbose")) { g_displayLevel++; continue; } |
964 | if (!strcmp(argument, "--quiet")) { g_displayLevel--; continue; } |
f535537f |
965 | if (!strcmp(argument, "--stdout")) { forceStdout=1; outFileName=stdoutmark; continue; } |
648db22b |
966 | if (!strcmp(argument, "--ultra")) { ultra=1; continue; } |
967 | if (!strcmp(argument, "--check")) { FIO_setChecksumFlag(prefs, 2); continue; } |
968 | if (!strcmp(argument, "--no-check")) { FIO_setChecksumFlag(prefs, 0); continue; } |
969 | if (!strcmp(argument, "--sparse")) { FIO_setSparseWrite(prefs, 2); continue; } |
970 | if (!strcmp(argument, "--no-sparse")) { FIO_setSparseWrite(prefs, 0); continue; } |
971 | if (!strcmp(argument, "--pass-through")) { FIO_setPassThroughFlag(prefs, 1); continue; } |
972 | if (!strcmp(argument, "--no-pass-through")) { FIO_setPassThroughFlag(prefs, 0); continue; } |
973 | if (!strcmp(argument, "--test")) { operation=zom_test; continue; } |
974 | if (!strcmp(argument, "--asyncio")) { FIO_setAsyncIOFlag(prefs, 1); continue;} |
975 | if (!strcmp(argument, "--no-asyncio")) { FIO_setAsyncIOFlag(prefs, 0); continue;} |
976 | if (!strcmp(argument, "--train")) { operation=zom_train; if (outFileName==NULL) outFileName=g_defaultDictName; continue; } |
977 | if (!strcmp(argument, "--no-dictID")) { FIO_setDictIDFlag(prefs, 0); continue; } |
978 | if (!strcmp(argument, "--keep")) { removeSrcFile=0; continue; } |
979 | if (!strcmp(argument, "--rm")) { removeSrcFile=1; continue; } |
980 | if (!strcmp(argument, "--priority=rt")) { setRealTimePrio = 1; continue; } |
981 | if (!strcmp(argument, "--show-default-cparams")) { showDefaultCParams = 1; continue; } |
982 | if (!strcmp(argument, "--content-size")) { contentSize = 1; continue; } |
983 | if (!strcmp(argument, "--no-content-size")) { contentSize = 0; continue; } |
984 | if (!strcmp(argument, "--adapt")) { adapt = 1; continue; } |
985 | if (!strcmp(argument, "--no-row-match-finder")) { useRowMatchFinder = ZSTD_ps_disable; continue; } |
986 | if (!strcmp(argument, "--row-match-finder")) { useRowMatchFinder = ZSTD_ps_enable; continue; } |
f535537f |
987 | if (longCommandWArg(&argument, "--adapt=")) { adapt = 1; if (!parseAdaptParameters(argument, &adaptMin, &adaptMax)) { badUsage(programName, originalArgument); CLEAN_RETURN(1); } continue; } |
648db22b |
988 | if (!strcmp(argument, "--single-thread")) { nbWorkers = 0; singleThread = 1; continue; } |
989 | if (!strcmp(argument, "--format=zstd")) { suffix = ZSTD_EXTENSION; cType = FIO_zstdCompression; continue; } |
990 | if (!strcmp(argument, "--mmap-dict")) { mmapDict = ZSTD_ps_enable; continue; } |
991 | if (!strcmp(argument, "--no-mmap-dict")) { mmapDict = ZSTD_ps_disable; continue; } |
992 | #ifdef ZSTD_GZCOMPRESS |
993 | if (!strcmp(argument, "--format=gzip")) { suffix = GZ_EXTENSION; cType = FIO_gzipCompression; continue; } |
994 | if (exeNameMatch(programName, ZSTD_GZ)) { /* behave like gzip */ |
995 | if (!strcmp(argument, "--best")) { dictCLevel = cLevel = 9; continue; } |
996 | if (!strcmp(argument, "--no-name")) { /* ignore for now */; continue; } |
997 | } |
998 | #endif |
999 | #ifdef ZSTD_LZMACOMPRESS |
1000 | if (!strcmp(argument, "--format=lzma")) { suffix = LZMA_EXTENSION; cType = FIO_lzmaCompression; continue; } |
1001 | if (!strcmp(argument, "--format=xz")) { suffix = XZ_EXTENSION; cType = FIO_xzCompression; continue; } |
1002 | #endif |
1003 | #ifdef ZSTD_LZ4COMPRESS |
1004 | if (!strcmp(argument, "--format=lz4")) { suffix = LZ4_EXTENSION; cType = FIO_lz4Compression; continue; } |
1005 | #endif |
1006 | if (!strcmp(argument, "--rsyncable")) { rsyncable = 1; continue; } |
1007 | if (!strcmp(argument, "--compress-literals")) { literalCompressionMode = ZSTD_ps_enable; continue; } |
1008 | if (!strcmp(argument, "--no-compress-literals")) { literalCompressionMode = ZSTD_ps_disable; continue; } |
1009 | if (!strcmp(argument, "--no-progress")) { progress = FIO_ps_never; continue; } |
1010 | if (!strcmp(argument, "--progress")) { progress = FIO_ps_always; continue; } |
1011 | if (!strcmp(argument, "--exclude-compressed")) { FIO_setExcludeCompressedFile(prefs, 1); continue; } |
1012 | if (!strcmp(argument, "--fake-stdin-is-console")) { UTIL_fakeStdinIsConsole(); continue; } |
1013 | if (!strcmp(argument, "--fake-stdout-is-console")) { UTIL_fakeStdoutIsConsole(); continue; } |
1014 | if (!strcmp(argument, "--fake-stderr-is-console")) { UTIL_fakeStderrIsConsole(); continue; } |
1015 | if (!strcmp(argument, "--trace-file-stat")) { UTIL_traceFileStat(); continue; } |
1016 | |
1017 | /* long commands with arguments */ |
1018 | #ifndef ZSTD_NODICT |
1019 | if (longCommandWArg(&argument, "--train-cover")) { |
1020 | operation = zom_train; |
1021 | if (outFileName == NULL) |
1022 | outFileName = g_defaultDictName; |
1023 | dict = cover; |
1024 | /* Allow optional arguments following an = */ |
1025 | if (*argument == 0) { memset(&coverParams, 0, sizeof(coverParams)); } |
f535537f |
1026 | else if (*argument++ != '=') { badUsage(programName, originalArgument); CLEAN_RETURN(1); } |
1027 | else if (!parseCoverParameters(argument, &coverParams)) { badUsage(programName, originalArgument); CLEAN_RETURN(1); } |
648db22b |
1028 | continue; |
1029 | } |
1030 | if (longCommandWArg(&argument, "--train-fastcover")) { |
1031 | operation = zom_train; |
1032 | if (outFileName == NULL) |
1033 | outFileName = g_defaultDictName; |
1034 | dict = fastCover; |
1035 | /* Allow optional arguments following an = */ |
1036 | if (*argument == 0) { memset(&fastCoverParams, 0, sizeof(fastCoverParams)); } |
f535537f |
1037 | else if (*argument++ != '=') { badUsage(programName, originalArgument); CLEAN_RETURN(1); } |
1038 | else if (!parseFastCoverParameters(argument, &fastCoverParams)) { badUsage(programName, originalArgument); CLEAN_RETURN(1); } |
648db22b |
1039 | continue; |
1040 | } |
1041 | if (longCommandWArg(&argument, "--train-legacy")) { |
1042 | operation = zom_train; |
1043 | if (outFileName == NULL) |
1044 | outFileName = g_defaultDictName; |
1045 | dict = legacy; |
1046 | /* Allow optional arguments following an = */ |
1047 | if (*argument == 0) { continue; } |
f535537f |
1048 | else if (*argument++ != '=') { badUsage(programName, originalArgument); CLEAN_RETURN(1); } |
1049 | else if (!parseLegacyParameters(argument, &dictSelect)) { badUsage(programName, originalArgument); CLEAN_RETURN(1); } |
648db22b |
1050 | continue; |
1051 | } |
1052 | #endif |
1053 | if (longCommandWArg(&argument, "--threads")) { NEXT_UINT32(nbWorkers); continue; } |
1054 | if (longCommandWArg(&argument, "--memlimit")) { NEXT_UINT32(memLimit); continue; } |
1055 | if (longCommandWArg(&argument, "--memory")) { NEXT_UINT32(memLimit); continue; } |
1056 | if (longCommandWArg(&argument, "--memlimit-decompress")) { NEXT_UINT32(memLimit); continue; } |
1057 | if (longCommandWArg(&argument, "--block-size")) { NEXT_TSIZE(blockSize); continue; } |
1058 | if (longCommandWArg(&argument, "--maxdict")) { NEXT_UINT32(maxDictSize); continue; } |
1059 | if (longCommandWArg(&argument, "--dictID")) { NEXT_UINT32(dictID); continue; } |
f535537f |
1060 | if (longCommandWArg(&argument, "--zstd=")) { if (!parseCompressionParameters(argument, &compressionParams)) { badUsage(programName, originalArgument); CLEAN_RETURN(1); } ; cType = FIO_zstdCompression; continue; } |
648db22b |
1061 | if (longCommandWArg(&argument, "--stream-size")) { NEXT_TSIZE(streamSrcSize); continue; } |
1062 | if (longCommandWArg(&argument, "--target-compressed-block-size")) { NEXT_TSIZE(targetCBlockSize); continue; } |
1063 | if (longCommandWArg(&argument, "--size-hint")) { NEXT_TSIZE(srcSizeHint); continue; } |
1064 | if (longCommandWArg(&argument, "--output-dir-flat")) { |
1065 | NEXT_FIELD(outDirName); |
1066 | if (strlen(outDirName) == 0) { |
1067 | DISPLAYLEVEL(1, "error: output dir cannot be empty string (did you mean to pass '.' instead?)\n"); |
1068 | CLEAN_RETURN(1); |
1069 | } |
1070 | continue; |
1071 | } |
1072 | if (longCommandWArg(&argument, "--auto-threads")) { |
1073 | const char* threadDefault = NULL; |
1074 | NEXT_FIELD(threadDefault); |
1075 | if (strcmp(threadDefault, "logical") == 0) |
1076 | defaultLogicalCores = 1; |
1077 | continue; |
1078 | } |
1079 | #ifdef UTIL_HAS_MIRRORFILELIST |
1080 | if (longCommandWArg(&argument, "--output-dir-mirror")) { |
1081 | NEXT_FIELD(outMirroredDirName); |
1082 | if (strlen(outMirroredDirName) == 0) { |
1083 | DISPLAYLEVEL(1, "error: output dir cannot be empty string (did you mean to pass '.' instead?)\n"); |
1084 | CLEAN_RETURN(1); |
1085 | } |
1086 | continue; |
1087 | } |
1088 | #endif |
1089 | #ifndef ZSTD_NOTRACE |
1090 | if (longCommandWArg(&argument, "--trace")) { char const* traceFile; NEXT_FIELD(traceFile); TRACE_enable(traceFile); continue; } |
1091 | #endif |
1092 | if (longCommandWArg(&argument, "--patch-from")) { NEXT_FIELD(patchFromDictFileName); continue; } |
1093 | if (longCommandWArg(&argument, "--long")) { |
1094 | unsigned ldmWindowLog = 0; |
1095 | ldmFlag = 1; |
1096 | /* Parse optional window log */ |
1097 | if (*argument == '=') { |
1098 | ++argument; |
1099 | ldmWindowLog = readU32FromChar(&argument); |
1100 | } else if (*argument != 0) { |
1101 | /* Invalid character following --long */ |
f535537f |
1102 | badUsage(programName, originalArgument); |
648db22b |
1103 | CLEAN_RETURN(1); |
1104 | } else { |
1105 | ldmWindowLog = g_defaultMaxWindowLog; |
1106 | } |
1107 | /* Only set windowLog if not already set by --zstd */ |
1108 | if (compressionParams.windowLog == 0) |
1109 | compressionParams.windowLog = ldmWindowLog; |
1110 | continue; |
1111 | } |
1112 | #ifndef ZSTD_NOCOMPRESS /* linking ZSTD_minCLevel() requires compression support */ |
1113 | if (longCommandWArg(&argument, "--fast")) { |
1114 | /* Parse optional acceleration factor */ |
1115 | if (*argument == '=') { |
1116 | U32 const maxFast = (U32)-ZSTD_minCLevel(); |
1117 | U32 fastLevel; |
1118 | ++argument; |
1119 | fastLevel = readU32FromChar(&argument); |
1120 | if (fastLevel > maxFast) fastLevel = maxFast; |
1121 | if (fastLevel) { |
1122 | dictCLevel = cLevel = -(int)fastLevel; |
1123 | } else { |
f535537f |
1124 | badUsage(programName, originalArgument); |
648db22b |
1125 | CLEAN_RETURN(1); |
1126 | } |
1127 | } else if (*argument != 0) { |
1128 | /* Invalid character following --fast */ |
f535537f |
1129 | badUsage(programName, originalArgument); |
648db22b |
1130 | CLEAN_RETURN(1); |
1131 | } else { |
1132 | cLevel = -1; /* default for --fast */ |
1133 | } |
1134 | continue; |
1135 | } |
1136 | #endif |
1137 | |
1138 | if (longCommandWArg(&argument, "--filelist")) { |
1139 | const char* listName; |
1140 | NEXT_FIELD(listName); |
1141 | UTIL_refFilename(file_of_names, listName); |
1142 | continue; |
1143 | } |
1144 | |
f535537f |
1145 | badUsage(programName, originalArgument); |
1146 | CLEAN_RETURN(1); |
648db22b |
1147 | } |
1148 | |
1149 | argument++; |
1150 | while (argument[0]!=0) { |
1151 | |
1152 | #ifndef ZSTD_NOCOMPRESS |
1153 | /* compression Level */ |
1154 | if ((*argument>='0') && (*argument<='9')) { |
1155 | dictCLevel = cLevel = (int)readU32FromChar(&argument); |
1156 | continue; |
1157 | } |
1158 | #endif |
1159 | |
1160 | switch(argument[0]) |
1161 | { |
1162 | /* Display help */ |
1163 | case 'V': printVersion(); CLEAN_RETURN(0); /* Version Only */ |
f535537f |
1164 | case 'H': usageAdvanced(programName); CLEAN_RETURN(0); |
648db22b |
1165 | case 'h': usage(stdout, programName); CLEAN_RETURN(0); |
1166 | |
1167 | /* Compress */ |
1168 | case 'z': operation=zom_compress; argument++; break; |
1169 | |
1170 | /* Decoding */ |
1171 | case 'd': |
1172 | #ifndef ZSTD_NOBENCH |
1173 | benchParams.mode = BMK_decodeOnly; |
1174 | if (operation==zom_bench) { argument++; break; } /* benchmark decode (hidden option) */ |
1175 | #endif |
1176 | operation=zom_decompress; argument++; break; |
1177 | |
1178 | /* Force stdout, even if stdout==console */ |
f535537f |
1179 | case 'c': forceStdout=1; outFileName=stdoutmark; argument++; break; |
1180 | |
1181 | /* destination file name */ |
1182 | case 'o': argument++; NEXT_FIELD(outFileName); break; |
648db22b |
1183 | |
1184 | /* do not store filename - gzip compatibility - nothing to do */ |
1185 | case 'n': argument++; break; |
1186 | |
1187 | /* Use file content as dictionary */ |
1188 | case 'D': argument++; NEXT_FIELD(dictFileName); break; |
1189 | |
1190 | /* Overwrite */ |
1191 | case 'f': FIO_overwriteMode(prefs); forceStdin=1; forceStdout=1; followLinks=1; allowBlockDevices=1; argument++; break; |
1192 | |
1193 | /* Verbose mode */ |
1194 | case 'v': g_displayLevel++; argument++; break; |
1195 | |
1196 | /* Quiet mode */ |
1197 | case 'q': g_displayLevel--; argument++; break; |
1198 | |
1199 | /* keep source file (default) */ |
1200 | case 'k': removeSrcFile=0; argument++; break; |
1201 | |
1202 | /* Checksum */ |
1203 | case 'C': FIO_setChecksumFlag(prefs, 2); argument++; break; |
1204 | |
1205 | /* test compressed file */ |
1206 | case 't': operation=zom_test; argument++; break; |
1207 | |
648db22b |
1208 | /* limit memory */ |
1209 | case 'M': |
1210 | argument++; |
1211 | memLimit = readU32FromChar(&argument); |
1212 | break; |
1213 | case 'l': operation=zom_list; argument++; break; |
1214 | #ifdef UTIL_HAS_CREATEFILELIST |
1215 | /* recursive */ |
1216 | case 'r': recursive=1; argument++; break; |
1217 | #endif |
1218 | |
1219 | #ifndef ZSTD_NOBENCH |
1220 | /* Benchmark */ |
1221 | case 'b': |
1222 | operation=zom_bench; |
1223 | argument++; |
1224 | break; |
1225 | |
1226 | /* range bench (benchmark only) */ |
1227 | case 'e': |
1228 | /* compression Level */ |
1229 | argument++; |
1230 | cLevelLast = (int)readU32FromChar(&argument); |
1231 | break; |
1232 | |
1233 | /* Modify Nb Iterations (benchmark only) */ |
1234 | case 'i': |
1235 | argument++; |
1236 | bench_nbSeconds = readU32FromChar(&argument); |
1237 | break; |
1238 | |
1239 | /* cut input into blocks (benchmark only) */ |
1240 | case 'B': |
1241 | argument++; |
1242 | blockSize = readU32FromChar(&argument); |
1243 | break; |
1244 | |
1245 | /* benchmark files separately (hidden option) */ |
1246 | case 'S': |
1247 | argument++; |
1248 | separateFiles = 1; |
1249 | break; |
1250 | |
1251 | #endif /* ZSTD_NOBENCH */ |
1252 | |
1253 | /* nb of threads (hidden option) */ |
1254 | case 'T': |
1255 | argument++; |
1256 | nbWorkers = readU32FromChar(&argument); |
1257 | break; |
1258 | |
1259 | /* Dictionary Selection level */ |
1260 | case 's': |
1261 | argument++; |
1262 | dictSelect = readU32FromChar(&argument); |
1263 | break; |
1264 | |
1265 | /* Pause at the end (-p) or set an additional param (-p#) (hidden option) */ |
1266 | case 'p': argument++; |
1267 | #ifndef ZSTD_NOBENCH |
1268 | if ((*argument>='0') && (*argument<='9')) { |
1269 | benchParams.additionalParam = (int)readU32FromChar(&argument); |
1270 | } else |
1271 | #endif |
1272 | main_pause=1; |
1273 | break; |
1274 | |
1275 | /* Select compressibility of synthetic sample */ |
1276 | case 'P': |
1277 | argument++; |
1278 | compressibility = (double)readU32FromChar(&argument) / 100; |
1279 | break; |
1280 | |
1281 | /* unknown command */ |
f535537f |
1282 | default : |
1283 | { char shortArgument[3] = {'-', 0, 0}; |
1284 | shortArgument[1] = argument[0]; |
1285 | badUsage(programName, shortArgument); |
1286 | CLEAN_RETURN(1); |
1287 | } |
648db22b |
1288 | } |
1289 | } |
1290 | continue; |
1291 | } /* if (argument[0]=='-') */ |
1292 | |
1293 | /* none of the above : add filename to list */ |
1294 | UTIL_refFilename(filenames, argument); |
1295 | } |
1296 | |
1297 | /* Welcome message (if verbose) */ |
1298 | DISPLAYLEVEL(3, WELCOME_MESSAGE); |
1299 | |
1300 | #ifdef ZSTD_MULTITHREAD |
1301 | if ((operation==zom_decompress) && (!singleThread) && (nbWorkers > 1)) { |
1302 | DISPLAYLEVEL(2, "Warning : decompression does not support multi-threading\n"); |
1303 | } |
1304 | if ((nbWorkers==0) && (!singleThread)) { |
1305 | /* automatically set # workers based on # of reported cpus */ |
1306 | if (defaultLogicalCores) { |
1307 | nbWorkers = (unsigned)UTIL_countLogicalCores(); |
1308 | DISPLAYLEVEL(3, "Note: %d logical core(s) detected \n", nbWorkers); |
1309 | } else { |
1310 | nbWorkers = (unsigned)UTIL_countPhysicalCores(); |
1311 | DISPLAYLEVEL(3, "Note: %d physical core(s) detected \n", nbWorkers); |
1312 | } |
1313 | } |
1314 | #else |
1315 | (void)singleThread; (void)nbWorkers; (void)defaultLogicalCores; |
1316 | #endif |
1317 | |
1318 | g_utilDisplayLevel = g_displayLevel; |
1319 | |
1320 | #ifdef UTIL_HAS_CREATEFILELIST |
1321 | if (!followLinks) { |
1322 | unsigned u, fileNamesNb; |
1323 | unsigned const nbFilenames = (unsigned)filenames->tableSize; |
1324 | for (u=0, fileNamesNb=0; u<nbFilenames; u++) { |
1325 | if ( UTIL_isLink(filenames->fileNames[u]) |
1326 | && !UTIL_isFIFO(filenames->fileNames[u]) |
1327 | ) { |
1328 | DISPLAYLEVEL(2, "Warning : %s is a symbolic link, ignoring \n", filenames->fileNames[u]); |
1329 | } else { |
1330 | filenames->fileNames[fileNamesNb++] = filenames->fileNames[u]; |
1331 | } } |
1332 | if (fileNamesNb == 0 && nbFilenames > 0) /* all names are eliminated */ |
1333 | CLEAN_RETURN(1); |
1334 | filenames->tableSize = fileNamesNb; |
1335 | } /* if (!followLinks) */ |
1336 | |
1337 | /* read names from a file */ |
1338 | if (file_of_names->tableSize) { |
1339 | size_t const nbFileLists = file_of_names->tableSize; |
1340 | size_t flNb; |
1341 | for (flNb=0; flNb < nbFileLists; flNb++) { |
1342 | FileNamesTable* const fnt = UTIL_createFileNamesTable_fromFileName(file_of_names->fileNames[flNb]); |
1343 | if (fnt==NULL) { |
1344 | DISPLAYLEVEL(1, "zstd: error reading %s \n", file_of_names->fileNames[flNb]); |
1345 | CLEAN_RETURN(1); |
1346 | } |
1347 | filenames = UTIL_mergeFileNamesTable(filenames, fnt); |
1348 | } |
1349 | } |
1350 | |
1351 | nbInputFileNames = filenames->tableSize; /* saving number of input files */ |
1352 | |
1353 | if (recursive) { /* at this stage, filenameTable is a list of paths, which can contain both files and directories */ |
1354 | UTIL_expandFNT(&filenames, followLinks); |
1355 | } |
1356 | #else |
1357 | (void)followLinks; |
1358 | #endif |
1359 | |
1360 | if (operation == zom_list) { |
1361 | #ifndef ZSTD_NODECOMPRESS |
1362 | int const ret = FIO_listMultipleFiles((unsigned)filenames->tableSize, filenames->fileNames, g_displayLevel); |
1363 | CLEAN_RETURN(ret); |
1364 | #else |
1365 | DISPLAYLEVEL(1, "file information is not supported \n"); |
1366 | CLEAN_RETURN(1); |
1367 | #endif |
1368 | } |
1369 | |
1370 | /* Check if benchmark is selected */ |
1371 | if (operation==zom_bench) { |
1372 | #ifndef ZSTD_NOBENCH |
1373 | if (cType != FIO_zstdCompression) { |
1374 | DISPLAYLEVEL(1, "benchmark mode is only compatible with zstd format \n"); |
1375 | CLEAN_RETURN(1); |
1376 | } |
1377 | benchParams.blockSize = blockSize; |
f535537f |
1378 | benchParams.targetCBlockSize = targetCBlockSize; |
648db22b |
1379 | benchParams.nbWorkers = (int)nbWorkers; |
1380 | benchParams.realTime = (unsigned)setRealTimePrio; |
1381 | benchParams.nbSeconds = bench_nbSeconds; |
1382 | benchParams.ldmFlag = ldmFlag; |
1383 | benchParams.ldmMinMatch = (int)g_ldmMinMatch; |
1384 | benchParams.ldmHashLog = (int)g_ldmHashLog; |
1385 | benchParams.useRowMatchFinder = (int)useRowMatchFinder; |
1386 | if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) { |
1387 | benchParams.ldmBucketSizeLog = (int)g_ldmBucketSizeLog; |
1388 | } |
1389 | if (g_ldmHashRateLog != LDM_PARAM_DEFAULT) { |
1390 | benchParams.ldmHashRateLog = (int)g_ldmHashRateLog; |
1391 | } |
1392 | benchParams.literalCompressionMode = literalCompressionMode; |
1393 | |
1394 | if (cLevel > ZSTD_maxCLevel()) cLevel = ZSTD_maxCLevel(); |
1395 | if (cLevelLast > ZSTD_maxCLevel()) cLevelLast = ZSTD_maxCLevel(); |
1396 | if (cLevelLast < cLevel) cLevelLast = cLevel; |
1397 | if (cLevelLast > cLevel) |
1398 | DISPLAYLEVEL(3, "Benchmarking levels from %d to %d\n", cLevel, cLevelLast); |
1399 | if (filenames->tableSize > 0) { |
1400 | if(separateFiles) { |
1401 | unsigned i; |
1402 | for(i = 0; i < filenames->tableSize; i++) { |
1403 | int c; |
1404 | DISPLAYLEVEL(3, "Benchmarking %s \n", filenames->fileNames[i]); |
1405 | for(c = cLevel; c <= cLevelLast; c++) { |
1406 | operationResult = BMK_benchFilesAdvanced(&filenames->fileNames[i], 1, dictFileName, c, &compressionParams, g_displayLevel, &benchParams); |
1407 | } } |
1408 | } else { |
1409 | for(; cLevel <= cLevelLast; cLevel++) { |
1410 | operationResult = BMK_benchFilesAdvanced(filenames->fileNames, (unsigned)filenames->tableSize, dictFileName, cLevel, &compressionParams, g_displayLevel, &benchParams); |
1411 | } } |
1412 | } else { |
1413 | for(; cLevel <= cLevelLast; cLevel++) { |
1414 | operationResult = BMK_syntheticTest(cLevel, compressibility, &compressionParams, g_displayLevel, &benchParams); |
1415 | } } |
1416 | |
1417 | #else |
1418 | (void)bench_nbSeconds; (void)blockSize; (void)setRealTimePrio; (void)separateFiles; (void)compressibility; |
1419 | #endif |
1420 | goto _end; |
1421 | } |
1422 | |
1423 | /* Check if dictionary builder is selected */ |
1424 | if (operation==zom_train) { |
1425 | #ifndef ZSTD_NODICT |
1426 | ZDICT_params_t zParams; |
1427 | zParams.compressionLevel = dictCLevel; |
1428 | zParams.notificationLevel = (unsigned)g_displayLevel; |
1429 | zParams.dictID = dictID; |
1430 | if (dict == cover) { |
1431 | int const optimize = !coverParams.k || !coverParams.d; |
1432 | coverParams.nbThreads = (unsigned)nbWorkers; |
1433 | coverParams.zParams = zParams; |
1434 | operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (int)filenames->tableSize, blockSize, NULL, &coverParams, NULL, optimize, memLimit); |
1435 | } else if (dict == fastCover) { |
1436 | int const optimize = !fastCoverParams.k || !fastCoverParams.d; |
1437 | fastCoverParams.nbThreads = (unsigned)nbWorkers; |
1438 | fastCoverParams.zParams = zParams; |
1439 | operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (int)filenames->tableSize, blockSize, NULL, NULL, &fastCoverParams, optimize, memLimit); |
1440 | } else { |
1441 | ZDICT_legacy_params_t dictParams; |
1442 | memset(&dictParams, 0, sizeof(dictParams)); |
1443 | dictParams.selectivityLevel = dictSelect; |
1444 | dictParams.zParams = zParams; |
1445 | operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (int)filenames->tableSize, blockSize, &dictParams, NULL, NULL, 0, memLimit); |
1446 | } |
1447 | #else |
1448 | (void)dictCLevel; (void)dictSelect; (void)dictID; (void)maxDictSize; /* not used when ZSTD_NODICT set */ |
1449 | DISPLAYLEVEL(1, "training mode not available \n"); |
1450 | operationResult = 1; |
1451 | #endif |
1452 | goto _end; |
1453 | } |
1454 | |
1455 | #ifndef ZSTD_NODECOMPRESS |
1456 | if (operation==zom_test) { FIO_setTestMode(prefs, 1); outFileName=nulmark; removeSrcFile=0; } /* test mode */ |
1457 | #endif |
1458 | |
1459 | /* No input filename ==> use stdin and stdout */ |
1460 | if (filenames->tableSize == 0) { |
1461 | /* It is possible that the input |
1462 | was a number of empty directories. In this case |
1463 | stdin and stdout should not be used */ |
1464 | if (nbInputFileNames > 0 ){ |
1465 | DISPLAYLEVEL(1, "please provide correct input file(s) or non-empty directories -- ignored \n"); |
1466 | CLEAN_RETURN(0); |
1467 | } |
1468 | UTIL_refFilename(filenames, stdinmark); |
1469 | } |
1470 | |
1471 | if (filenames->tableSize == 1 && !strcmp(filenames->fileNames[0], stdinmark) && !outFileName) |
1472 | outFileName = stdoutmark; /* when input is stdin, default output is stdout */ |
1473 | |
1474 | /* Check if input/output defined as console; trigger an error in this case */ |
1475 | if (!forceStdin |
1476 | && (UTIL_searchFileNamesTable(filenames, stdinmark) != -1) |
1477 | && UTIL_isConsole(stdin) ) { |
1478 | DISPLAYLEVEL(1, "stdin is a console, aborting\n"); |
1479 | CLEAN_RETURN(1); |
1480 | } |
1481 | if ( (!outFileName || !strcmp(outFileName, stdoutmark)) |
1482 | && UTIL_isConsole(stdout) |
1483 | && (UTIL_searchFileNamesTable(filenames, stdinmark) != -1) |
1484 | && !forceStdout |
1485 | && operation!=zom_decompress ) { |
1486 | DISPLAYLEVEL(1, "stdout is a console, aborting\n"); |
1487 | CLEAN_RETURN(1); |
1488 | } |
1489 | |
1490 | #ifndef ZSTD_NOCOMPRESS |
1491 | /* check compression level limits */ |
1492 | { int const maxCLevel = ultra ? ZSTD_maxCLevel() : ZSTDCLI_CLEVEL_MAX; |
1493 | if (cLevel > maxCLevel) { |
1494 | DISPLAYLEVEL(2, "Warning : compression level higher than max, reduced to %i \n", maxCLevel); |
1495 | cLevel = maxCLevel; |
1496 | } } |
1497 | #endif |
1498 | |
1499 | if (showDefaultCParams) { |
1500 | if (operation == zom_decompress) { |
1501 | DISPLAYLEVEL(1, "error : can't use --show-default-cparams in decompression mode \n"); |
1502 | CLEAN_RETURN(1); |
1503 | } |
1504 | } |
1505 | |
1506 | if (dictFileName != NULL && patchFromDictFileName != NULL) { |
1507 | DISPLAYLEVEL(1, "error : can't use -D and --patch-from=# at the same time \n"); |
1508 | CLEAN_RETURN(1); |
1509 | } |
1510 | |
1511 | if (patchFromDictFileName != NULL && filenames->tableSize > 1) { |
1512 | DISPLAYLEVEL(1, "error : can't use --patch-from=# on multiple files \n"); |
1513 | CLEAN_RETURN(1); |
1514 | } |
1515 | |
1516 | /* No status message by default when output is stdout */ |
1517 | hasStdout = outFileName && !strcmp(outFileName,stdoutmark); |
1518 | if (hasStdout && (g_displayLevel==2)) g_displayLevel=1; |
1519 | |
1520 | /* when stderr is not the console, do not pollute it with progress updates (unless requested) */ |
1521 | if (!UTIL_isConsole(stderr) && (progress!=FIO_ps_always)) progress=FIO_ps_never; |
1522 | FIO_setProgressSetting(progress); |
1523 | |
1524 | /* don't remove source files when output is stdout */; |
1525 | if (hasStdout && removeSrcFile) { |
1526 | DISPLAYLEVEL(3, "Note: src files are not removed when output is stdout \n"); |
1527 | removeSrcFile = 0; |
1528 | } |
1529 | FIO_setRemoveSrcFile(prefs, removeSrcFile); |
1530 | |
1531 | /* IO Stream/File */ |
1532 | FIO_setHasStdoutOutput(fCtx, hasStdout); |
1533 | FIO_setNbFilesTotal(fCtx, (int)filenames->tableSize); |
1534 | FIO_determineHasStdinInput(fCtx, filenames); |
1535 | FIO_setNotificationLevel(g_displayLevel); |
1536 | FIO_setAllowBlockDevices(prefs, allowBlockDevices); |
1537 | FIO_setPatchFromMode(prefs, patchFromDictFileName != NULL); |
1538 | FIO_setMMapDict(prefs, mmapDict); |
1539 | if (memLimit == 0) { |
1540 | if (compressionParams.windowLog == 0) { |
1541 | memLimit = (U32)1 << g_defaultMaxWindowLog; |
1542 | } else { |
1543 | memLimit = (U32)1 << (compressionParams.windowLog & 31); |
1544 | } } |
1545 | if (patchFromDictFileName != NULL) |
1546 | dictFileName = patchFromDictFileName; |
1547 | FIO_setMemLimit(prefs, memLimit); |
1548 | if (operation==zom_compress) { |
1549 | #ifndef ZSTD_NOCOMPRESS |
1550 | FIO_setCompressionType(prefs, cType); |
1551 | FIO_setContentSize(prefs, contentSize); |
1552 | FIO_setNbWorkers(prefs, (int)nbWorkers); |
1553 | FIO_setBlockSize(prefs, (int)blockSize); |
1554 | if (g_overlapLog!=OVERLAP_LOG_DEFAULT) FIO_setOverlapLog(prefs, (int)g_overlapLog); |
1555 | FIO_setLdmFlag(prefs, (unsigned)ldmFlag); |
1556 | FIO_setLdmHashLog(prefs, (int)g_ldmHashLog); |
1557 | FIO_setLdmMinMatch(prefs, (int)g_ldmMinMatch); |
1558 | if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) FIO_setLdmBucketSizeLog(prefs, (int)g_ldmBucketSizeLog); |
1559 | if (g_ldmHashRateLog != LDM_PARAM_DEFAULT) FIO_setLdmHashRateLog(prefs, (int)g_ldmHashRateLog); |
1560 | FIO_setAdaptiveMode(prefs, adapt); |
1561 | FIO_setUseRowMatchFinder(prefs, (int)useRowMatchFinder); |
1562 | FIO_setAdaptMin(prefs, adaptMin); |
1563 | FIO_setAdaptMax(prefs, adaptMax); |
1564 | FIO_setRsyncable(prefs, rsyncable); |
1565 | FIO_setStreamSrcSize(prefs, streamSrcSize); |
1566 | FIO_setTargetCBlockSize(prefs, targetCBlockSize); |
1567 | FIO_setSrcSizeHint(prefs, srcSizeHint); |
1568 | FIO_setLiteralCompressionMode(prefs, literalCompressionMode); |
1569 | FIO_setSparseWrite(prefs, 0); |
1570 | if (adaptMin > cLevel) cLevel = adaptMin; |
1571 | if (adaptMax < cLevel) cLevel = adaptMax; |
1572 | |
1573 | /* Compare strategies constant with the ground truth */ |
1574 | { ZSTD_bounds strategyBounds = ZSTD_cParam_getBounds(ZSTD_c_strategy); |
1575 | assert(ZSTD_NB_STRATEGIES == strategyBounds.upperBound); |
1576 | (void)strategyBounds; } |
1577 | |
1578 | if (showDefaultCParams || g_displayLevel >= 4) { |
1579 | size_t fileNb; |
1580 | for (fileNb = 0; fileNb < (size_t)filenames->tableSize; fileNb++) { |
1581 | if (showDefaultCParams) |
1582 | printDefaultCParams(filenames->fileNames[fileNb], dictFileName, cLevel); |
1583 | if (g_displayLevel >= 4) |
1584 | printActualCParams(filenames->fileNames[fileNb], dictFileName, cLevel, &compressionParams); |
1585 | } |
1586 | } |
1587 | |
1588 | if (g_displayLevel >= 4) |
1589 | FIO_displayCompressionParameters(prefs); |
1590 | if ((filenames->tableSize==1) && outFileName) |
1591 | operationResult = FIO_compressFilename(fCtx, prefs, outFileName, filenames->fileNames[0], dictFileName, cLevel, compressionParams); |
1592 | else |
1593 | operationResult = FIO_compressMultipleFilenames(fCtx, prefs, filenames->fileNames, outMirroredDirName, outDirName, outFileName, suffix, dictFileName, cLevel, compressionParams); |
1594 | #else |
1595 | /* these variables are only used when compression mode is enabled */ |
1596 | (void)contentSize; (void)suffix; (void)adapt; (void)rsyncable; |
1597 | (void)ultra; (void)cLevel; (void)ldmFlag; (void)literalCompressionMode; |
1598 | (void)targetCBlockSize; (void)streamSrcSize; (void)srcSizeHint; |
1599 | (void)ZSTD_strategyMap; (void)useRowMatchFinder; (void)cType; |
1600 | DISPLAYLEVEL(1, "Compression not supported \n"); |
1601 | #endif |
1602 | } else { /* decompression or test */ |
1603 | #ifndef ZSTD_NODECOMPRESS |
1604 | if (filenames->tableSize == 1 && outFileName) { |
1605 | operationResult = FIO_decompressFilename(fCtx, prefs, outFileName, filenames->fileNames[0], dictFileName); |
1606 | } else { |
1607 | operationResult = FIO_decompressMultipleFilenames(fCtx, prefs, filenames->fileNames, outMirroredDirName, outDirName, outFileName, dictFileName); |
1608 | } |
1609 | #else |
1610 | DISPLAYLEVEL(1, "Decompression not supported \n"); |
1611 | #endif |
1612 | } |
1613 | |
1614 | _end: |
1615 | FIO_freePreferences(prefs); |
1616 | FIO_freeContext(fCtx); |
1617 | if (main_pause) waitEnter(); |
1618 | UTIL_freeFileNamesTable(filenames); |
1619 | UTIL_freeFileNamesTable(file_of_names); |
1620 | #ifndef ZSTD_NOTRACE |
1621 | TRACE_finish(); |
1622 | #endif |
1623 | |
1624 | return operationResult; |
1625 | } |