| 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 | #include <stdio.h> // printf |
| 13 | #include <stdlib.h> // free |
| 14 | #include <string.h> // memset, strcat, strlen |
| 15 | #include <zstd.h> // presumes zstd library is installed |
| 16 | #include "common.h" // Helper functions, CHECK(), and CHECK_ZSTD() |
| 17 | |
| 18 | static void compressFile_orDie(const char* fname, const char* outName, int cLevel, |
| 19 | int nbThreads) |
| 20 | { |
| 21 | fprintf (stderr, "Starting compression of %s with level %d, using %d threads\n", |
| 22 | fname, cLevel, nbThreads); |
| 23 | |
| 24 | /* Open the input and output files. */ |
| 25 | FILE* const fin = fopen_orDie(fname, "rb"); |
| 26 | FILE* const fout = fopen_orDie(outName, "wb"); |
| 27 | /* Create the input and output buffers. |
| 28 | * They may be any size, but we recommend using these functions to size them. |
| 29 | * Performance will only suffer significantly for very tiny buffers. |
| 30 | */ |
| 31 | size_t const buffInSize = ZSTD_CStreamInSize(); |
| 32 | void* const buffIn = malloc_orDie(buffInSize); |
| 33 | size_t const buffOutSize = ZSTD_CStreamOutSize(); |
| 34 | void* const buffOut = malloc_orDie(buffOutSize); |
| 35 | |
| 36 | /* Create the context. */ |
| 37 | ZSTD_CCtx* const cctx = ZSTD_createCCtx(); |
| 38 | CHECK(cctx != NULL, "ZSTD_createCCtx() failed!"); |
| 39 | |
| 40 | /* Set any parameters you want. |
| 41 | * Here we set the compression level, and enable the checksum. |
| 42 | */ |
| 43 | CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_compressionLevel, cLevel) ); |
| 44 | CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_checksumFlag, 1) ); |
| 45 | ZSTD_CCtx_setParameter(cctx, ZSTD_c_nbWorkers, nbThreads); |
| 46 | |
| 47 | /* This loop read from the input file, compresses that entire chunk, |
| 48 | * and writes all output produced to the output file. |
| 49 | */ |
| 50 | size_t const toRead = buffInSize; |
| 51 | for (;;) { |
| 52 | size_t read = fread_orDie(buffIn, toRead, fin); |
| 53 | /* Select the flush mode. |
| 54 | * If the read may not be finished (read == toRead) we use |
| 55 | * ZSTD_e_continue. If this is the last chunk, we use ZSTD_e_end. |
| 56 | * Zstd optimizes the case where the first flush mode is ZSTD_e_end, |
| 57 | * since it knows it is compressing the entire source in one pass. |
| 58 | */ |
| 59 | int const lastChunk = (read < toRead); |
| 60 | ZSTD_EndDirective const mode = lastChunk ? ZSTD_e_end : ZSTD_e_continue; |
| 61 | /* Set the input buffer to what we just read. |
| 62 | * We compress until the input buffer is empty, each time flushing the |
| 63 | * output. |
| 64 | */ |
| 65 | ZSTD_inBuffer input = { buffIn, read, 0 }; |
| 66 | int finished; |
| 67 | do { |
| 68 | /* Compress into the output buffer and write all of the output to |
| 69 | * the file so we can reuse the buffer next iteration. |
| 70 | */ |
| 71 | ZSTD_outBuffer output = { buffOut, buffOutSize, 0 }; |
| 72 | size_t const remaining = ZSTD_compressStream2(cctx, &output , &input, mode); |
| 73 | CHECK_ZSTD(remaining); |
| 74 | fwrite_orDie(buffOut, output.pos, fout); |
| 75 | /* If we're on the last chunk we're finished when zstd returns 0, |
| 76 | * which means its consumed all the input AND finished the frame. |
| 77 | * Otherwise, we're finished when we've consumed all the input. |
| 78 | */ |
| 79 | finished = lastChunk ? (remaining == 0) : (input.pos == input.size); |
| 80 | } while (!finished); |
| 81 | CHECK(input.pos == input.size, |
| 82 | "Impossible: zstd only returns 0 when the input is completely consumed!"); |
| 83 | |
| 84 | if (lastChunk) { |
| 85 | break; |
| 86 | } |
| 87 | } |
| 88 | |
| 89 | ZSTD_freeCCtx(cctx); |
| 90 | fclose_orDie(fout); |
| 91 | fclose_orDie(fin); |
| 92 | free(buffIn); |
| 93 | free(buffOut); |
| 94 | } |
| 95 | |
| 96 | |
| 97 | static char* createOutFilename_orDie(const char* filename) |
| 98 | { |
| 99 | size_t const inL = strlen(filename); |
| 100 | size_t const outL = inL + 5; |
| 101 | void* const outSpace = malloc_orDie(outL); |
| 102 | memset(outSpace, 0, outL); |
| 103 | strcat(outSpace, filename); |
| 104 | strcat(outSpace, ".zst"); |
| 105 | return (char*)outSpace; |
| 106 | } |
| 107 | |
| 108 | int main(int argc, const char** argv) |
| 109 | { |
| 110 | const char* const exeName = argv[0]; |
| 111 | |
| 112 | if (argc < 2) { |
| 113 | printf("wrong arguments\n"); |
| 114 | printf("usage:\n"); |
| 115 | printf("%s FILE [LEVEL] [THREADS]\n", exeName); |
| 116 | return 1; |
| 117 | } |
| 118 | |
| 119 | int cLevel = 1; |
| 120 | int nbThreads = 4; |
| 121 | |
| 122 | if (argc >= 3) { |
| 123 | cLevel = atoi (argv[2]); |
| 124 | CHECK(cLevel != 0, "can't parse LEVEL!"); |
| 125 | } |
| 126 | |
| 127 | if (argc >= 4) { |
| 128 | nbThreads = atoi (argv[3]); |
| 129 | CHECK(nbThreads != 0, "can't parse THREADS!"); |
| 130 | } |
| 131 | |
| 132 | const char* const inFilename = argv[1]; |
| 133 | |
| 134 | char* const outFilename = createOutFilename_orDie(inFilename); |
| 135 | compressFile_orDie(inFilename, outFilename, cLevel, nbThreads); |
| 136 | |
| 137 | free(outFilename); /* not strictly required, since program execution stops there, |
| 138 | * but some static analyzer may complain otherwise */ |
| 139 | return 0; |
| 140 | } |