• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // LZ4 API example : Dictionary Random Access
2 
3 #ifdef _MSC_VER    /* Visual Studio */
4 #  define _CRT_SECURE_NO_WARNINGS
5 #  define snprintf sprintf_s
6 #endif
7 #include "lz4.h"
8 
9 #include <stdio.h>
10 #include <stdint.h>
11 #include <stdlib.h>
12 #include <string.h>
13 
14 #define MIN(x, y) (x) < (y) ? (x) : (y)
15 
16 enum {
17     BLOCK_BYTES = 1024,  /* 1 KiB of uncompressed data in a block */
18     DICTIONARY_BYTES = 1024, /* Load a 1 KiB dictionary */
19     MAX_BLOCKS = 1024 /* For simplicity of implementation */
20 };
21 
22 /**
23  * Magic bytes for this test case.
24  * This is not a great magic number because it is a common word in ASCII.
25  * However, it is important to have some versioning system in your format.
26  */
27 const char kTestMagic[] = { 'T', 'E', 'S', 'T' };
28 
29 
write_int(FILE * fp,int i)30 void write_int(FILE* fp, int i) {
31     size_t written = fwrite(&i, sizeof(i), 1, fp);
32     if (written != 1) { exit(10); }
33 }
34 
write_bin(FILE * fp,const void * array,size_t arrayBytes)35 void write_bin(FILE* fp, const void* array, size_t arrayBytes) {
36     size_t written = fwrite(array, 1, arrayBytes, fp);
37     if (written != arrayBytes) { exit(11); }
38 }
39 
read_int(FILE * fp,int * i)40 void read_int(FILE* fp, int* i) {
41     size_t read = fread(i, sizeof(*i), 1, fp);
42     if (read != 1) { exit(12); }
43 }
44 
read_bin(FILE * fp,void * array,size_t arrayBytes)45 size_t read_bin(FILE* fp, void* array, size_t arrayBytes) {
46     size_t read = fread(array, 1, arrayBytes, fp);
47     if (ferror(fp)) { exit(12); }
48     return read;
49 }
50 
seek_bin(FILE * fp,long offset,int origin)51 void seek_bin(FILE* fp, long offset, int origin) {
52     if (fseek(fp, offset, origin)) { exit(14); }
53 }
54 
55 
test_compress(FILE * outFp,FILE * inpFp,void * dict,int dictSize)56 void test_compress(FILE* outFp, FILE* inpFp, void *dict, int dictSize)
57 {
58     LZ4_stream_t lz4Stream_body;
59     LZ4_stream_t* lz4Stream = &lz4Stream_body;
60 
61     char inpBuf[BLOCK_BYTES];
62     int offsets[MAX_BLOCKS];
63     int *offsetsEnd = offsets;
64 
65 
66     LZ4_resetStream(lz4Stream);
67 
68     /* Write header magic */
69     write_bin(outFp, kTestMagic, sizeof(kTestMagic));
70 
71     *offsetsEnd++ = sizeof(kTestMagic);
72     /* Write compressed data blocks.  Each block contains BLOCK_BYTES of plain
73        data except possibly the last. */
74     for(;;) {
75         const int inpBytes = (int) read_bin(inpFp, inpBuf, BLOCK_BYTES);
76         if(0 == inpBytes) {
77             break;
78         }
79 
80         /* Forget previously compressed data and load the dictionary */
81         LZ4_loadDict(lz4Stream, dict, dictSize);
82         {
83             char cmpBuf[LZ4_COMPRESSBOUND(BLOCK_BYTES)];
84             const int cmpBytes = LZ4_compress_fast_continue(
85                 lz4Stream, inpBuf, cmpBuf, inpBytes, sizeof(cmpBuf), 1);
86             if(cmpBytes <= 0) { exit(1); }
87             write_bin(outFp, cmpBuf, (size_t)cmpBytes);
88             /* Keep track of the offsets */
89             *offsetsEnd = *(offsetsEnd - 1) + cmpBytes;
90             ++offsetsEnd;
91         }
92         if (offsetsEnd - offsets > MAX_BLOCKS) { exit(2); }
93     }
94     /* Write the tailing jump table */
95     {
96         int *ptr = offsets;
97         while (ptr != offsetsEnd) {
98             write_int(outFp, *ptr++);
99         }
100         write_int(outFp, offsetsEnd - offsets);
101     }
102 }
103 
104 
test_decompress(FILE * outFp,FILE * inpFp,void * dict,int dictSize,int offset,int length)105 void test_decompress(FILE* outFp, FILE* inpFp, void *dict, int dictSize, int offset, int length)
106 {
107     LZ4_streamDecode_t lz4StreamDecode_body;
108     LZ4_streamDecode_t* lz4StreamDecode = &lz4StreamDecode_body;
109 
110     /* The blocks [currentBlock, endBlock) contain the data we want */
111     int currentBlock = offset / BLOCK_BYTES;
112     int endBlock = ((offset + length - 1) / BLOCK_BYTES) + 1;
113 
114     char decBuf[BLOCK_BYTES];
115     int offsets[MAX_BLOCKS];
116 
117     /* Special cases */
118     if (length == 0) { return; }
119 
120     /* Read the magic bytes */
121     {
122         char magic[sizeof(kTestMagic)];
123         size_t read = read_bin(inpFp, magic, sizeof(magic));
124         if (read != sizeof(magic)) { exit(1); }
125         if (memcmp(kTestMagic, magic, sizeof(magic))) { exit(2); }
126     }
127 
128     /* Read the offsets tail */
129     {
130         int numOffsets;
131         int block;
132         int *offsetsPtr = offsets;
133         seek_bin(inpFp, -4, SEEK_END);
134         read_int(inpFp, &numOffsets);
135         if (numOffsets <= endBlock) { exit(3); }
136         seek_bin(inpFp, -4 * (numOffsets + 1), SEEK_END);
137         for (block = 0; block <= endBlock; ++block) {
138             read_int(inpFp, offsetsPtr++);
139         }
140     }
141     /* Seek to the first block to read */
142     seek_bin(inpFp, offsets[currentBlock], SEEK_SET);
143     offset = offset % BLOCK_BYTES;
144 
145     /* Start decoding */
146     for(; currentBlock < endBlock; ++currentBlock) {
147         char cmpBuf[LZ4_COMPRESSBOUND(BLOCK_BYTES)];
148         /* The difference in offsets is the size of the block */
149         int  cmpBytes = offsets[currentBlock + 1] - offsets[currentBlock];
150         {
151             const size_t read = read_bin(inpFp, cmpBuf, (size_t)cmpBytes);
152             if(read != (size_t)cmpBytes) { exit(4); }
153         }
154 
155         /* Load the dictionary */
156         LZ4_setStreamDecode(lz4StreamDecode, dict, dictSize);
157         {
158             const int decBytes = LZ4_decompress_safe_continue(
159                 lz4StreamDecode, cmpBuf, decBuf, cmpBytes, BLOCK_BYTES);
160             if(decBytes <= 0) { exit(5); }
161             {
162                 /* Write out the part of the data we care about */
163                 int blockLength = MIN(length, (decBytes - offset));
164                 write_bin(outFp, decBuf + offset, (size_t)blockLength);
165                 offset = 0;
166                 length -= blockLength;
167             }
168         }
169     }
170 }
171 
172 
compare(FILE * fp0,FILE * fp1,int length)173 int compare(FILE* fp0, FILE* fp1, int length)
174 {
175     int result = 0;
176 
177     while(0 == result) {
178         char b0[4096];
179         char b1[4096];
180         const size_t r0 = read_bin(fp0, b0, MIN(length, (int)sizeof(b0)));
181         const size_t r1 = read_bin(fp1, b1, MIN(length, (int)sizeof(b1)));
182 
183         result = (int) r0 - (int) r1;
184 
185         if(0 == r0 || 0 == r1) {
186             break;
187         }
188         if(0 == result) {
189             result = memcmp(b0, b1, r0);
190         }
191         length -= r0;
192     }
193 
194     return result;
195 }
196 
197 
main(int argc,char * argv[])198 int main(int argc, char* argv[])
199 {
200     char inpFilename[256] = { 0 };
201     char lz4Filename[256] = { 0 };
202     char decFilename[256] = { 0 };
203     char dictFilename[256] = { 0 };
204     int offset;
205     int length;
206     char dict[DICTIONARY_BYTES];
207     int dictSize;
208 
209     if(argc < 5) {
210         printf("Usage: %s input dictionary offset length", argv[0]);
211         return 0;
212     }
213 
214     snprintf(inpFilename, 256, "%s", argv[1]);
215     snprintf(lz4Filename, 256, "%s.lz4s-%d", argv[1], BLOCK_BYTES);
216     snprintf(decFilename, 256, "%s.lz4s-%d.dec", argv[1], BLOCK_BYTES);
217     snprintf(dictFilename, 256, "%s", argv[2]);
218     offset = atoi(argv[3]);
219     length = atoi(argv[4]);
220 
221     printf("inp    = [%s]\n", inpFilename);
222     printf("lz4    = [%s]\n", lz4Filename);
223     printf("dec    = [%s]\n", decFilename);
224     printf("dict   = [%s]\n", dictFilename);
225     printf("offset = [%d]\n", offset);
226     printf("length = [%d]\n", length);
227 
228     /* Load dictionary */
229     {
230         FILE* dictFp = fopen(dictFilename, "rb");
231         dictSize = (int)read_bin(dictFp, dict, DICTIONARY_BYTES);
232         fclose(dictFp);
233     }
234 
235     /* compress */
236     {
237         FILE* inpFp = fopen(inpFilename, "rb");
238         FILE* outFp = fopen(lz4Filename, "wb");
239 
240         printf("compress : %s -> %s\n", inpFilename, lz4Filename);
241         test_compress(outFp, inpFp, dict, dictSize);
242         printf("compress : done\n");
243 
244         fclose(outFp);
245         fclose(inpFp);
246     }
247 
248     /* decompress */
249     {
250         FILE* inpFp = fopen(lz4Filename, "rb");
251         FILE* outFp = fopen(decFilename, "wb");
252 
253         printf("decompress : %s -> %s\n", lz4Filename, decFilename);
254         test_decompress(outFp, inpFp, dict, DICTIONARY_BYTES, offset, length);
255         printf("decompress : done\n");
256 
257         fclose(outFp);
258         fclose(inpFp);
259     }
260 
261     /* verify */
262     {
263         FILE* inpFp = fopen(inpFilename, "rb");
264         FILE* decFp = fopen(decFilename, "rb");
265         seek_bin(inpFp, offset, SEEK_SET);
266 
267         printf("verify : %s <-> %s\n", inpFilename, decFilename);
268         const int cmp = compare(inpFp, decFp, length);
269         if(0 == cmp) {
270             printf("verify : OK\n");
271         } else {
272             printf("verify : NG\n");
273         }
274 
275         fclose(decFp);
276         fclose(inpFp);
277     }
278 
279     return 0;
280 }
281