xref: /third_party/lz4/examples/frameCompress.c (revision 27b27ec6)
1/* LZ4frame API example : compress a file
2 * Modified from an example code by Zbigniew Jędrzejewski-Szmek
3 *
4 * This example streams an input file into an output file
5 * using a bounded memory budget.
6 * Input is read in chunks of IN_CHUNK_SIZE */
7
8#include <stdio.h>
9#include <stdlib.h>
10#include <string.h>
11#include <errno.h>
12#include <assert.h>
13
14#include <getopt.h>
15#include <lz4frame.h>
16#include <lz4frame_static.h>
17
18#define IN_CHUNK_SIZE  (16*1024)
19
20static const LZ4F_preferences_t kPrefs = {
21    { LZ4F_max256KB, LZ4F_blockLinked, LZ4F_noContentChecksum, LZ4F_frame,
22      0 /* unknown content size */, 0 /* no dictID */ , LZ4F_noBlockChecksum },
23    0,   /* compression level; 0 == default */
24    0,   /* autoflush */
25    0,   /* favor decompression speed */
26    { 0, 0, 0 },  /* reserved, must be set to 0 */
27};
28
29
30/* safe_fwrite() :
31 * performs fwrite(), ensure operation success, or immediately exit() */
32static void safe_fwrite(void* buf, size_t eltSize, size_t nbElt, FILE* f)
33{
34    size_t const writtenSize = fwrite(buf, eltSize, nbElt, f);
35    size_t const expectedSize = eltSize * nbElt;
36    if (nbElt>0) assert(expectedSize / nbElt == eltSize);  /* check overflow */
37    if (writtenSize < expectedSize) {
38        if (ferror(f))  /* note : ferror() must follow fwrite */
39            fprintf(stderr, "Write failed \n");
40        else
41            fprintf(stderr, "Write too short \n");
42        exit(1);
43    }
44}
45
46
47/* ================================================= */
48/*     Streaming Compression example               */
49/* ================================================= */
50
51typedef struct {
52    int error;
53    unsigned long long size_in;
54    unsigned long long size_out;
55} compressResult_t;
56
57static compressResult_t
58compress_file_internal(FILE* f_in, FILE* f_out,
59                       LZ4F_compressionContext_t ctx,
60                       void* inBuff,  size_t inChunkSize,
61                       void* outBuff, size_t outCapacity,
62                       FILE* f_unc, long uncOffset)
63{
64    compressResult_t result = { 1, 0, 0 };  /* result for an error */
65    long long count_in = 0, count_out, bytesToOffset = -1;
66
67    assert(f_in != NULL); assert(f_out != NULL);
68    assert(ctx != NULL);
69    assert(outCapacity >= LZ4F_HEADER_SIZE_MAX);
70    assert(outCapacity >= LZ4F_compressBound(inChunkSize, &kPrefs));
71
72    /* write frame header */
73    {   size_t const headerSize = LZ4F_compressBegin(ctx, outBuff, outCapacity, &kPrefs);
74        if (LZ4F_isError(headerSize)) {
75            printf("Failed to start compression: error %u \n", (unsigned)headerSize);
76            return result;
77        }
78        count_out = headerSize;
79        printf("Buffer size is %u bytes, header size %u bytes \n",
80                (unsigned)outCapacity, (unsigned)headerSize);
81        safe_fwrite(outBuff, 1, headerSize, f_out);
82    }
83
84    /* stream file */
85    for (;;) {
86      size_t compressedSize;
87      long long inSize = IN_CHUNK_SIZE;
88      if (uncOffset >= 0) {
89        bytesToOffset = uncOffset - count_in;
90
91        /* read only remaining bytes to offset position */
92        if (bytesToOffset < IN_CHUNK_SIZE && bytesToOffset > 0) {
93          inSize = bytesToOffset;
94        }
95      }
96
97      /* input data is at uncompressed data offset */
98      if (bytesToOffset <= 0 && uncOffset >= 0 && f_unc) {
99        size_t const readSize = fread(inBuff, 1, inSize, f_unc);
100        if (readSize == 0) {
101          uncOffset = -1;
102          continue;
103        }
104        count_in += readSize;
105        compressedSize = LZ4F_uncompressedUpdate(ctx,
106                                             outBuff, outCapacity,
107                                             inBuff, readSize,
108                                             NULL);
109      } else {
110        size_t const readSize = fread(inBuff, 1, inSize, f_in);
111        if (readSize == 0) break; /* nothing left to read from input file */
112        count_in += readSize;
113        compressedSize = LZ4F_compressUpdate(ctx,
114                                                outBuff, outCapacity,
115                                                inBuff, readSize,
116                                                NULL);
117
118      }
119
120      if (LZ4F_isError(compressedSize)) {
121        printf("Compression failed: error %u \n", (unsigned)compressedSize);
122        return result;
123      }
124
125      printf("Writing %u bytes\n", (unsigned)compressedSize);
126      safe_fwrite(outBuff, 1, compressedSize, f_out);
127      count_out += compressedSize;
128    }
129
130    /* flush whatever remains within internal buffers */
131    {   size_t const compressedSize = LZ4F_compressEnd(ctx,
132                                                outBuff, outCapacity,
133                                                NULL);
134        if (LZ4F_isError(compressedSize)) {
135            printf("Failed to end compression: error %u \n", (unsigned)compressedSize);
136            return result;
137        }
138
139        printf("Writing %u bytes \n", (unsigned)compressedSize);
140        safe_fwrite(outBuff, 1, compressedSize, f_out);
141        count_out += compressedSize;
142    }
143
144    result.size_in = count_in;
145    result.size_out = count_out;
146    result.error = 0;
147    return result;
148}
149
150static compressResult_t
151compress_file(FILE* f_in, FILE* f_out,
152              FILE* f_unc, int uncOffset)
153{
154    assert(f_in != NULL);
155    assert(f_out != NULL);
156
157    /* resource allocation */
158    LZ4F_compressionContext_t ctx;
159    size_t const ctxCreation = LZ4F_createCompressionContext(&ctx, LZ4F_VERSION);
160    void* const src = malloc(IN_CHUNK_SIZE);
161    size_t const outbufCapacity = LZ4F_compressBound(IN_CHUNK_SIZE, &kPrefs);   /* large enough for any input <= IN_CHUNK_SIZE */
162    void* const outbuff = malloc(outbufCapacity);
163
164    compressResult_t result = { 1, 0, 0 };  /* == error (default) */
165    if (!LZ4F_isError(ctxCreation) && src && outbuff) {
166        result = compress_file_internal(f_in, f_out,
167                                        ctx,
168                                        src, IN_CHUNK_SIZE,
169                                        outbuff, outbufCapacity,
170                                        f_unc, uncOffset);
171    } else {
172        printf("error : resource allocation failed \n");
173    }
174
175    LZ4F_freeCompressionContext(ctx);   /* supports free on NULL */
176    free(src);
177    free(outbuff);
178    return result;
179}
180
181
182/* ================================================= */
183/*     Streaming decompression example               */
184/* ================================================= */
185
186static size_t get_block_size(const LZ4F_frameInfo_t* info) {
187    switch (info->blockSizeID) {
188        case LZ4F_default:
189        case LZ4F_max64KB:  return 1 << 16;
190        case LZ4F_max256KB: return 1 << 18;
191        case LZ4F_max1MB:   return 1 << 20;
192        case LZ4F_max4MB:   return 1 << 22;
193        default:
194            printf("Impossible with expected frame specification (<=v1.6.1)\n");
195            exit(1);
196    }
197}
198
199/* @return : 1==error, 0==success */
200static int
201decompress_file_internal(FILE* f_in, FILE* f_out,
202                         LZ4F_dctx* dctx,
203                         void* src, size_t srcCapacity, size_t filled, size_t alreadyConsumed,
204                         void* dst, size_t dstCapacity)
205{
206    int firstChunk = 1;
207    size_t ret = 1;
208
209    assert(f_in != NULL); assert(f_out != NULL);
210    assert(dctx != NULL);
211    assert(src != NULL); assert(srcCapacity > 0); assert(filled <= srcCapacity); assert(alreadyConsumed <= filled);
212    assert(dst != NULL); assert(dstCapacity > 0);
213
214    /* Decompression */
215    while (ret != 0) {
216        /* Load more input */
217        size_t readSize = firstChunk ? filled : fread(src, 1, srcCapacity, f_in); firstChunk=0;
218        const void* srcPtr = (const char*)src + alreadyConsumed; alreadyConsumed=0;
219        const void* const srcEnd = (const char*)srcPtr + readSize;
220        if (readSize == 0 || ferror(f_in)) {
221            printf("Decompress: not enough input or error reading file\n");
222            return 1;
223        }
224
225        /* Decompress:
226         * Continue while there is more input to read (srcPtr != srcEnd)
227         * and the frame isn't over (ret != 0)
228         */
229        while (srcPtr < srcEnd && ret != 0) {
230            /* Any data within dst has been flushed at this stage */
231            size_t dstSize = dstCapacity;
232            size_t srcSize = (const char*)srcEnd - (const char*)srcPtr;
233            ret = LZ4F_decompress(dctx, dst, &dstSize, srcPtr, &srcSize, /* LZ4F_decompressOptions_t */ NULL);
234            if (LZ4F_isError(ret)) {
235                printf("Decompression error: %s\n", LZ4F_getErrorName(ret));
236                return 1;
237            }
238            /* Flush output */
239            if (dstSize != 0) safe_fwrite(dst, 1, dstSize, f_out);
240            /* Update input */
241            srcPtr = (const char*)srcPtr + srcSize;
242        }
243
244        assert(srcPtr <= srcEnd);
245
246        /* Ensure all input data has been consumed.
247         * It is valid to have multiple frames in the same file,
248         * but this example only supports one frame.
249         */
250        if (srcPtr < srcEnd) {
251            printf("Decompress: Trailing data left in file after frame\n");
252            return 1;
253        }
254    }
255
256    /* Check that there isn't trailing data in the file after the frame.
257     * It is valid to have multiple frames in the same file,
258     * but this example only supports one frame.
259     */
260    {   size_t const readSize = fread(src, 1, 1, f_in);
261        if (readSize != 0 || !feof(f_in)) {
262            printf("Decompress: Trailing data left in file after frame\n");
263            return 1;
264    }   }
265
266    return 0;
267}
268
269
270/* @return : 1==error, 0==completed */
271static int
272decompress_file_allocDst(FILE* f_in, FILE* f_out,
273                        LZ4F_dctx* dctx,
274                        void* src, size_t srcCapacity)
275{
276    assert(f_in != NULL); assert(f_out != NULL);
277    assert(dctx != NULL);
278    assert(src != NULL);
279    assert(srcCapacity >= LZ4F_HEADER_SIZE_MAX);  /* ensure LZ4F_getFrameInfo() can read enough data */
280
281    /* Read Frame header */
282    size_t const readSize = fread(src, 1, srcCapacity, f_in);
283    if (readSize == 0 || ferror(f_in)) {
284        printf("Decompress: not enough input or error reading file\n");
285        return 1;
286    }
287
288    LZ4F_frameInfo_t info;
289    size_t consumedSize = readSize;
290    {   size_t const fires = LZ4F_getFrameInfo(dctx, &info, src, &consumedSize);
291        if (LZ4F_isError(fires)) {
292            printf("LZ4F_getFrameInfo error: %s\n", LZ4F_getErrorName(fires));
293            return 1;
294    }   }
295
296    /* Allocating enough space for an entire block isn't necessary for
297     * correctness, but it allows some memcpy's to be elided.
298     */
299    size_t const dstCapacity = get_block_size(&info);
300    void* const dst = malloc(dstCapacity);
301    if (!dst) { perror("decompress_file(dst)"); return 1; }
302
303    int const decompressionResult = decompress_file_internal(
304                        f_in, f_out,
305                        dctx,
306                        src, srcCapacity, readSize-consumedSize, consumedSize,
307                        dst, dstCapacity);
308
309    free(dst);
310    return decompressionResult;
311}
312
313
314/* @result : 1==error, 0==success */
315static int decompress_file(FILE* f_in, FILE* f_out)
316{
317    assert(f_in != NULL); assert(f_out != NULL);
318
319    /* Resource allocation */
320    void* const src = malloc(IN_CHUNK_SIZE);
321    if (!src) { perror("decompress_file(src)"); return 1; }
322
323    LZ4F_dctx* dctx;
324    {   size_t const dctxStatus = LZ4F_createDecompressionContext(&dctx, LZ4F_VERSION);
325        if (LZ4F_isError(dctxStatus)) {
326            printf("LZ4F_dctx creation error: %s\n", LZ4F_getErrorName(dctxStatus));
327    }   }
328
329    int const result = !dctx ? 1 /* error */ :
330                       decompress_file_allocDst(f_in, f_out, dctx, src, IN_CHUNK_SIZE);
331
332    free(src);
333    LZ4F_freeDecompressionContext(dctx);   /* note : free works on NULL */
334    return result;
335}
336
337
338int compareFiles(FILE* fp0, FILE* fp1, FILE* fpUnc, long uncOffset)
339{
340    int result = 0;
341    long bytesRead = 0;
342    long bytesToOffset = -1;
343    long b1Size = 1024;
344
345    while (result==0) {
346        char b1[b1Size];
347        size_t r1;
348        size_t bytesToRead = sizeof b1;
349        if (uncOffset >= 0) {
350          bytesToOffset = uncOffset - bytesRead;
351
352          /* read remainder to offset */
353          if (bytesToOffset < b1Size) {
354            bytesToRead = bytesToOffset;
355          }
356        }
357
358        char b0[1024];
359        size_t r0;
360        if (bytesToOffset <= 0 && fpUnc) {
361          bytesToRead = sizeof b1;
362          r0 = fread(b0, 1,bytesToRead, fpUnc);
363        } else {
364          r0 = fread(b0, 1, bytesToRead, fp0);
365        }
366
367        r1 = fread(b1, 1, r0, fp1);
368
369        result = (r0 != r1);
370        if (!r0 || !r1) break;
371        if (!result) result = memcmp(b0, b1, r0);
372
373        bytesRead += r1;
374    }
375
376    return result;
377}
378
379
380int main(int argc, char **argv) {
381    char inpFilename[256] = { 0 };
382    char lz4Filename[256] = { 0 };
383    char decFilename[256] = { 0 };
384
385    int uncOffset = -1;
386    char uncFilename[256] = { 0 };
387    int opt;
388
389    if (argc < 2) {
390        printf("Please specify input filename\n");
391        return EXIT_FAILURE;
392    }
393
394    snprintf(inpFilename, 256, "%s", argv[1]);
395    snprintf(lz4Filename, 256, "%s.lz4", argv[1]);
396    snprintf(decFilename, 256, "%s.lz4.dec", argv[1]);
397
398    while ((opt = getopt(argc, argv, "o:d:")) != -1) {
399      switch (opt) {
400      case 'd':
401        snprintf(uncFilename, 256, "%s", optarg);
402        break;
403      case 'o':
404        uncOffset = atoi(optarg);
405        break;
406      default:
407        printf("usage: %s <input file> [-o <offset> -d <file>]\n", argv[0]);
408        printf("-o uncompressed data offset\n");
409        printf("   inject uncompressed data at this offset into the lz4 file\n");
410        printf("-d uncompressed file\n");
411        printf("   file to inject without compression into the lz4 file\n");
412        return EXIT_FAILURE;
413      }
414    }
415
416    printf("inp = [%s]\n", inpFilename);
417    printf("lz4 = [%s]\n", lz4Filename);
418    printf("dec = [%s]\n", decFilename);
419    if (uncOffset > 0) {
420      printf("unc = [%s]\n", uncFilename);
421      printf("ofs = [%i]\n", uncOffset);
422    }
423
424    /* compress */
425    {   FILE* const inpFp = fopen(inpFilename, "rb");
426        FILE* const outFp = fopen(lz4Filename, "wb");
427        FILE* const uncFp = fopen(uncFilename, "rb");
428
429        printf("compress : %s -> %s\n", inpFilename, lz4Filename);
430        compressResult_t const ret = compress_file(
431            inpFp, outFp,
432            uncFp, uncOffset);
433
434        fclose(outFp);
435        fclose(inpFp);
436        if (uncFp)
437          fclose(uncFp);
438
439        if (ret.error) {
440            printf("compress : failed with code %i\n", ret.error);
441            return ret.error;
442        }
443        printf("%s: %zu → %zu bytes, %.1f%%\n",
444            inpFilename,
445            (size_t)ret.size_in, (size_t)ret.size_out,  /* might overflow is size_t is 32 bits and size_{in,out} > 4 GB */
446            (double)ret.size_out / ret.size_in * 100);
447        printf("compress : done\n");
448    }
449
450    /* decompress */
451    {   FILE* const inpFp = fopen(lz4Filename, "rb");
452        FILE* const outFp = fopen(decFilename, "wb");
453
454        printf("decompress : %s -> %s\n", lz4Filename, decFilename);
455        int const ret = decompress_file(inpFp, outFp);
456
457        fclose(outFp);
458        fclose(inpFp);
459
460        if (ret) {
461            printf("decompress : failed with code %i\n", ret);
462            return ret;
463        }
464        printf("decompress : done\n");
465    }
466
467    /* verify */
468    {   FILE* const inpFp = fopen(inpFilename, "rb");
469        FILE* const decFp = fopen(decFilename, "rb");
470        FILE* const uncFp = fopen(uncFilename, "rb");
471
472        printf("verify : %s <-> %s\n", inpFilename, decFilename);
473        int const cmp = compareFiles(inpFp, decFp,
474                                     uncFp, uncOffset);
475
476        fclose(decFp);
477        fclose(inpFp);
478        if (uncFp)
479          fclose(uncFp);
480
481        if (cmp) {
482            printf("corruption detected : decompressed file differs from original\n");
483            return cmp;
484        }
485        printf("verify : OK\n");
486    }
487
488    return 0;
489}
490