xref: /freebsd/sys/contrib/zstd/examples/dictionary_decompression.c (revision cfd6422a5217410fbd66f7a7a8a64d9d85e61229)
1 /*
2  * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
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 #include <stdio.h>     // printf
13 #include <stdlib.h>    // free
14 #include <zstd.h>      // presumes zstd library is installed
15 #include "common.h"    // Helper functions, CHECK(), and CHECK_ZSTD()
16 
17 /* createDict() :
18    `dictFileName` is supposed to have been created using `zstd --train` */
19 static ZSTD_DDict* createDict_orDie(const char* dictFileName)
20 {
21     size_t dictSize;
22     printf("loading dictionary %s \n", dictFileName);
23     void* const dictBuffer = mallocAndLoadFile_orDie(dictFileName, &dictSize);
24     ZSTD_DDict* const ddict = ZSTD_createDDict(dictBuffer, dictSize);
25     CHECK(ddict != NULL, "ZSTD_createDDict() failed!");
26     free(dictBuffer);
27     return ddict;
28 }
29 
30 static void decompress(const char* fname, const ZSTD_DDict* ddict)
31 {
32     size_t cSize;
33     void* const cBuff = mallocAndLoadFile_orDie(fname, &cSize);
34     /* Read the content size from the frame header. For simplicity we require
35      * that it is always present. By default, zstd will write the content size
36      * in the header when it is known. If you can't guarantee that the frame
37      * content size is always written into the header, either use streaming
38      * decompression, or ZSTD_decompressBound().
39      */
40     unsigned long long const rSize = ZSTD_getFrameContentSize(cBuff, cSize);
41     CHECK(rSize != ZSTD_CONTENTSIZE_ERROR, "%s: not compressed by zstd!", fname);
42     CHECK(rSize != ZSTD_CONTENTSIZE_UNKNOWN, "%s: original size unknown!", fname);
43     void* const rBuff = malloc_orDie((size_t)rSize);
44 
45     /* Check that the dictionary ID matches.
46      * If a non-zstd dictionary is used, then both will be zero.
47      * By default zstd always writes the dictionary ID into the frame.
48      * Zstd will check if there is a dictionary ID mismatch as well.
49      */
50     unsigned const expectedDictID = ZSTD_getDictID_fromDDict(ddict);
51     unsigned const actualDictID = ZSTD_getDictID_fromFrame(cBuff, cSize);
52     CHECK(actualDictID == expectedDictID,
53           "DictID mismatch: expected %u got %u",
54           expectedDictID,
55           actualDictID);
56 
57     /* Decompress using the dictionary.
58      * If you need to control the decompression parameters, then use the
59      * advanced API: ZSTD_DCtx_setParameter(), ZSTD_DCtx_refDDict(), and
60      * ZSTD_decompressDCtx().
61      */
62     ZSTD_DCtx* const dctx = ZSTD_createDCtx();
63     CHECK(dctx != NULL, "ZSTD_createDCtx() failed!");
64     size_t const dSize = ZSTD_decompress_usingDDict(dctx, rBuff, rSize, cBuff, cSize, ddict);
65     CHECK_ZSTD(dSize);
66     /* When zstd knows the content size, it will error if it doesn't match. */
67     CHECK(dSize == rSize, "Impossible because zstd will check this condition!");
68 
69     /* success */
70     printf("%25s : %6u -> %7u \n", fname, (unsigned)cSize, (unsigned)rSize);
71 
72     ZSTD_freeDCtx(dctx);
73     free(rBuff);
74     free(cBuff);
75 }
76 
77 
78 int main(int argc, const char** argv)
79 {
80     const char* const exeName = argv[0];
81 
82     if (argc<3) {
83         printf("wrong arguments\n");
84         printf("usage:\n");
85         printf("%s [FILES] dictionary\n", exeName);
86         return 1;
87     }
88 
89     /* load dictionary only once */
90     const char* const dictName = argv[argc-1];
91     ZSTD_DDict* const dictPtr = createDict_orDie(dictName);
92 
93     int u;
94     for (u=1; u<argc-1; u++) decompress(argv[u], dictPtr);
95 
96     ZSTD_freeDDict(dictPtr);
97     printf("All %u files correctly decoded (in memory) \n", argc-2);
98     return 0;
99 }
100