struct scan_ctx *scan_init(void) {
struct scan_ctx *scan;
scan = malloc(sizeof(struct scan_ctx));
if (!scan)
goto unwind0;
scan->buffer_size = 3 * (1<<24);
scan->buffer[0] = malloc(scan->buffer_size);
if (!scan->buffer[0])
goto unwind1;
scan->buffer[1] = scan->buffer[0] + scan->buffer_size/3;
scan->buffer[2] = scan->buffer[1] + scan->buffer_size/3;
scan->buffer_end = scan->buffer[0] + scan->buffer_size;
scan->p = scan->buffer[0];
scan->eof = 0;
scan->bytes_left = 0;
scan->source_offset = 0;
scan->window_size = 48;
scan->target_chunk_size = 1 << 18;
scan->minimum_chunk_size = 1 << 16;
scan->maximum_chunk_size = 1 << 24;
scan->rabin_ctx = rabin_init(1103515245, scan->window_size);
if (!scan->rabin_ctx)
goto unwind2;
scan->start_io = start_sync_io;
scan->finish_io = finish_sync_io;
return scan;
unwind2:
rabin_free(scan->rabin_ctx);
unwind1:
free(scan->buffer[0]);
unwind0:
return 0;
}
int main(int argc, const char *argv[])
{
int window_size = 48 ;
int avg_seg_size = 4096;
int min_seg_size = 2048;
int max_seg_size = 8192;
char fname[PATH_MAX] = {0};
int c;
while ((c = getopt(argc, (char * const*)argv, "f:w:a:i:x:")) != -1) {
switch (c) {
case 'f':
strncpy(fname, optarg, sizeof fname);
break;
case 'w':
window_size = atoi(optarg);
break;
case 'a':
avg_seg_size = atoi(optarg);
break;
case 'i':
min_seg_size = atoi(optarg);
break;
case 'x':
max_seg_size = atoi(optarg);
break;
default:
usage(argv[0]);
exit(1);
}
}
int fd;
if (fname[0]) {
fd = open(fname, O_RDONLY);
if (fd == -1) {
perror("open failed:");
exit(2);
}
} else {
/* Use stdin if filename was not specified */
fd = STDIN_FILENO;
}
/*
printf("Window size : %d\n", window_size);
printf("Avg segment size requested : %d\n", avg_seg_size);
printf("Min segment size requested : %d\n", min_seg_size);
printf("Max segment size requested : %d\n", max_seg_size);
printf("Reading file %s\n", fname);
*/
rabinpoly_t *rp = rabin_init( window_size, avg_seg_size,
min_seg_size, max_seg_size);
if (!rp) {
fprintf(stderr, "Failed to init rabinhash algorithm\n");
exit(1);
}
MD5_CTX ctx;
unsigned char md5[MD5_DIGEST_LENGTH];
int new_segment = 0;
int len, segment_len = 0, b;
char buf[1024];
int bytes;
MD5_Init(&ctx);
while( (bytes = read(fd, buf, sizeof buf)) > 0 ) {
char *buftoread = (char *)&buf[0];
while ((len = rabin_segment_next(rp, buftoread, bytes,
&new_segment)) > 0) {
MD5_Update(&ctx, buftoread, len);
segment_len += len;
if (new_segment) {
MD5_Final(md5, &ctx);
printf("%u ", segment_len);
for(b = 0; b < MD5_DIGEST_LENGTH; b++)
printf("%02x", md5[b]);
printf("\n");
MD5_Init(&ctx);
segment_len = 0;
}
buftoread += len;
bytes -= len;
if (!bytes) {
break;
}
}
if (len == -1) {
fprintf(stderr, "Failed to process the segment\n");
exit(2);
}
}
MD5_Final(md5, &ctx);
printf("%u ", segment_len);
for(b = 0; b < MD5_DIGEST_LENGTH; b++) {
printf("%02x", md5[b]);
}
printf("\n");
rabin_free(&rp);
return 0;
}
void cdc_init ()
{
rabin_init (BLOCK_WIN_SZ);
}
