网络数据同步开发库libsync

 

"数据同步算法研究"一文提出了一种改进的数据同步算法，我在实现的原型系统基础上，将文件切分、差异编码、文件同步等关键算法抽取出来封装成动态开发库libsync，方便自己的开发应用。在本人开发的deduputil，WSIO, wsync等软件中，均使用了libsync动态函数库，现已将libsync发布至google code。libsync函数库由提供三个API，原型描述如下：

1、int file_chunk(char *src_filename, char *chunk_filename, int chunk_algo)
功能：对件进行切分生成分块描述文件。
参数：src_filename为切分文件，chunk_filename为生成的块信息描述文件，chunk_algo为文件切分算法，目前支持定长FSP切分(Ffixed-size partition)、CDC切分(content-defined chunking)和滑动块(sliding block)切分三种。

2、int file_delta(char *src_filename, char *chunk_filename, char *delta_filename, int chunk_algo)
功能：使用生成的块描述信息对文件进行差异编码。
参数：src_filename为待编码文件，chunk_filename为通过函数file_chunk生成的块描述文件，chunk_algo为文件切分算法。

3、int file_sync(char *src_filename, char *delta_filename)
功能：使用差异编码文件将源文件同步至目标文件。
参数：src_filename为基本文件，delta_filename为通过函数file_delta生成的差异编码文件。

/* Copyright (C) 2010, Aigui Liu Filename: sync.h Author: Aigui.Liu@gmail.com */ #ifndef _SYNC_H #define _SYNC_H #include <stdint.h> #define BLOCK_SZ 2048 #define BLOCK_MIN_SZ 1024 #define BLOCK_MAX_SZ 4096 #define BLOCK_WIN_SZ 48 #define NAME_MAX_SZ 256 #define BUF_MAX_SZ 32768 #define HASHTABLE_BUCKET_SZ 1024 #define CHUNK_CDC_D BLOCK_SZ #define CHUNK_CDC_R 13 enum chunk_algo { CHUNK_FSP = 0, /* fixed-size partition */ CHUNK_CDC, /* content-define chunking */ CHUNK_SBC /* slide block chunking */ }; /* define chunk file header and block entry */ typedef struct _chunk_file_header { uint32_t block_sz; uint32_t block_nr; } chunk_file_header; #define CHUNK_FILE_HEADER_SZ (sizeof(chunk_file_header)) typedef struct _chunk_block_entry { uint64_t offset; uint32_t len; uint8_t md5[16 + 1]; uint8_t csum[10 + 1]; } chunk_block_entry; #define CHUNK_BLOCK_ENTRY_SZ (sizeof(chunk_block_entry)) /* define delta file header and block entry */ typedef struct _delta_file_header { uint32_t block_nr; uint32_t last_block_sz; uint64_t last_block_offset; /* offset in delta file */ } delta_file_header; #define DELTA_FILE_HEADER_SZ (sizeof(delta_file_header)) typedef struct _delta_block_entry { uint64_t offset; uint32_t len; uint8_t embeded; /* 1, block in delta file; 0, block in source file. */ } delta_block_entry; #define DELTA_BLOCK_ENTRY_SZ (sizeof(delta_block_entry)) int file_chunk(char *src_filename, char *chunk_filename, int chunk_algo); int file_delta(char *src_filename, char *chunk_filename, char *delta_filename, int chunk_algo); int file_sync(char *src_filename, char *delta_filename); #endif

数据同步有PULL和PUSH两种应用模式，PULL是将远程数据同步到本地，而PUSH是将本地数据同步到远程。对应到同步算法，主要区别在于数据分块和差异编码位置不同。PULL和PUSH同步模式步骤分别如下所述。

PULL同步模式流程：
1、本地对文件A进行数据切分，生成数据块描述文件chunk；
2、上传chunk文件至远程服务器；
3、远程服务器对文件B进行差异编码，生成差异编码文件delta；
4、下载delta文件至本地；
5、本地同步文件A至文件B，相当于下载文件B到本地文件A。

例子程序：
/* Copyright (c) 2010, Aigui Liu All rights reserved. Filename: wspull.c Author: Aigui.Liu@gmail.com */ #include <stdio.h> #include <stdlib.h> #include <string.h> #include <fcntl.h> #include <unistd.h> #include "wsioclient.h" #include "wsio_api.h" #include "sync.h" /* main function */ int main(int argc, char **argv) { char fap[32] = {0}; char server[256] = {0}; char *local_filename = NULL; char remote_filename[256] = {0}; char local_chunkname[256] = {0}; char remote_chunkname[256] = {0}; char local_deltaname[256] = {0}; char remote_deltaname[256] = {0}; int ret = 0; if (argc < 3) { fprintf(stderr, "Usage: wspull remote_filename local_filename/n"); exit(0); } if(argv[1]) { if(parseURL(argv[1], fap, server, remote_filename) == -1) { fprintf(stderr, "wsio address is illegal/n"); exit(1); } } wsio_init(fap, server); local_filename = argv[2]; if (0 == access(local_filename, F_OK)) { sprintf(local_chunkname, "%s_%d", tmpnam(NULL), getpid()); sprintf(remote_chunkname, "%s_%d", tmpnam(NULL), getpid()); sprintf(local_deltaname, "%s_%d", tmpnam(NULL), getpid()); sprintf(remote_deltaname, "%s_%d", tmpnam(NULL), getpid()); ret = file_chunk(local_filename, local_chunkname, CHUNK_CDC); if (ret == -1) { fprintf(stderr, "local file chunking failed/n"); return -1; } ret = wsio_putfile(local_chunkname, remote_chunkname); if (ret == -1) { fprintf(stderr, "upload chunking file failed/n"); return -1; } ret = wsio_deltafile(remote_filename, remote_chunkname, remote_deltaname, CHUNK_CDC); if (ret == -1) { fprintf(stderr, "remote file delta coding failed/n"); return -1; } ret = wsio_getfile(remote_deltaname, local_deltaname); if (ret == -1) { fprintf(stderr, "download delta file failed/n"); } ret = file_sync(local_filename, local_deltaname); if (ret == -1) { fprintf(stderr, "local file sync failed/n"); return -1; } } else { ret = wsio_getfile(remote_filename, local_filename); if (ret == -1) { fprintf(stderr, "download file failed/n"); return -1; } } wsio_unlink(remote_chunkname); wsio_unlink(remote_deltaname); unlink(local_chunkname); unlink(local_deltaname); wsio_destroy(); fprintf(stderr, "pull %s to %s %s/n", argv[1], local_filename, (ret == 0)? "sucessfully" : "failed"); return 0; }  

PUSH同步模式流程：
1、远程服务器对文件B进行数据切分，生成数据块描述文件chunk；
2、下载chunk文件至本地；
3、本地对文件A进行差异编码，生成差异编码文件delta；
4、上传delta文件至远程服务器；
5、远程同步文件B到A，相当于上传文件A到远程文件B。

例子程序：
/* Copyright (c) 2010, Aigui Liu All rights reserved. Filename: wspush.c Author: Aigui.Liu@gmail.com */ #include <stdio.h> #include <stdlib.h> #include <string.h> #include <fcntl.h> #include <unistd.h> #include "wsioclient.h" #include "wsio_api.h" #include "sync.h" /* main function */ int main(int argc, char **argv) { char fap[32] = {0}; char server[256] = {0}; char remote_filename[256] = {0}; char *local_filename = NULL; char local_chunkname[256] = {0}; char remote_chunkname[256] = {0}; char local_deltaname[256] = {0}; char remote_deltaname[256] = {0}; int ret = 0; if (argc < 3) { fprintf(stderr, "Usage: wspush local_filename remote_filename/n"); exit(0); } if(argv[2]) { if(parseURL(argv[2], fap, server, remote_filename) == -1) { fprintf(stderr, "wsio address is illegal/n"); exit(1); } } wsio_init(fap, server); local_filename = argv[1]; if (0 == wsio_access(remote_filename, F_OK)) { sprintf(local_chunkname, "%s_%d", tmpnam(NULL), getpid()); sprintf(remote_chunkname, "%s_%d", tmpnam(NULL), getpid()); sprintf(local_deltaname, "%s_%d", tmpnam(NULL), getpid()); sprintf(remote_deltaname, "%s_%d", tmpnam(NULL), getpid()); ret = wsio_chunkfile(remote_filename, remote_chunkname, CHUNK_CDC); if (ret == -1) { fprintf(stderr, "remote file chunking failed/n"); return -1; } ret = wsio_getfile(remote_chunkname, local_chunkname); if (ret == -1) { fprintf(stderr, "download chunking file failed/n"); return -1; } ret = file_delta(local_filename, local_chunkname, local_deltaname, CHUNK_CDC); if (ret == -1) { fprintf(stderr, "local file delta coding failed/n"); return -1; } ret = wsio_putfile(local_deltaname, remote_deltaname); if (ret == -1) { fprintf(stderr, "upload delta file failed/n"); } ret = wsio_syncfile(remote_filename, remote_deltaname); if (ret == -1) { fprintf(stderr, "remote file sync failed/n"); return -1; } } else { ret = wsio_putfile(local_filename, remote_filename); if (ret == -1) { fprintf(stderr, "upload file failed/n"); return -1; } } wsio_unlink(remote_chunkname); wsio_unlink(remote_deltaname); unlink(local_chunkname); unlink(local_deltaname); wsio_destroy(); fprintf(stderr, "push %s to %s %s/n", local_filename, remote_filename, (ret == 0)? "sucessfully" : "failed"); return 0; }