/**
* Hash, check or update files according to work mode.
*/
static void process_files(void)
{
timedelta_t timer;
struct rsh_stat_struct stat_buf;
int i;
rhash_timer_start(&timer);
rhash_data.processed = 0;
/* process filenames */
for(i = 0; i < opt.n_files; i++) {
int res = 0;
char* filepath = opt.files[i];
stat_buf.st_mode = 0;
if(!IS_DASH_STR(filepath) && rsh_stat(filepath, &stat_buf) < 0) {
log_file_error(filepath);
continue;
}
if(opt.flags & OPT_RECURSIVE) {
if(S_ISDIR(stat_buf.st_mode)) {
find_file(filepath, find_file_callback, 0, opt.find_max_depth, NULL);
continue;
}
} else {
if(S_ISDIR(stat_buf.st_mode)){
if(opt.flags & OPT_VERBOSE){
log_warning(_("%s: is a directory\n"), filepath);
}
continue;
}
}
if(opt.mode & (MODE_CHECK | MODE_CHECK_EMBEDDED)) {
res = check_hash_file(filepath, 0);
} else if(opt.mode & MODE_UPDATE) {
res = update_hash_file(filepath);
} else {
res = calculate_and_print_sums(rhash_data.out, filepath, filepath, &stat_buf);
rhash_data.processed++;
}
if(res < 0) rhash_data.error_flag = 1;
}
if((opt.mode & MODE_CHECK_EMBEDDED) && rhash_data.processed > 1) {
print_check_stats();
}
if((opt.flags & OPT_SPEED) && !(opt.mode&(MODE_CHECK | MODE_UPDATE)) &&
rhash_data.processed > 1) {
double time = rhash_timer_stop(&timer);
print_time_stats(time, rhash_data.total_size, 1);
}
}
/**
* Verify hash sums of the file.
*
* @param info structure file path to process
* @return zero on success, -1 on file error, -2 if hash sums are different
*/
static int verify_sums(struct file_info *info)
{
timedelta_t timer;
int res = 0;
errno = 0;
/* initialize percents output */
init_percents(info);
rhash_timer_start(&timer);
if(calc_sums(info) < 0) {
finish_percents(info, -1);
return -1;
}
info->time = rhash_timer_stop(&timer);
if(rhash_data.interrupted) {
report_interrupted();
return 0;
}
if((opt.flags & OPT_EMBED_CRC) && find_embedded_crc32(
info->print_path, &info->hc.embedded_crc32_be)) {
info->hc.flags |= HC_HAS_EMBCRC32;
assert(info->hc.hash_mask & RHASH_CRC32);
}
if(!hash_check_verify(&info->hc, info->rctx)) {
res = -2;
}
finish_percents(info, res);
if((opt.flags & OPT_SPEED) && info->sums_flags) {
print_file_time_stats(info);
}
return res;
}
/**
* Check hash sums in a hash file.
* Lines beginning with ';' and '#' are ignored.
*
* @param hash_file_path - the path of the file with hash sums to verify.
* @param chdir - true if function should emulate chdir to directory of filepath before checking it.
* @return zero on success, -1 on fail
*/
int check_hash_file(file_t* file, int chdir)
{
FILE *fd;
char buf[2048];
size_t pos;
const char *ralign;
timedelta_t timer;
struct file_info info;
const char* hash_file_path = file->path;
int res = 0, line_num = 0;
double time;
/* process --check-embedded option */
if(opt.mode & MODE_CHECK_EMBEDDED) {
unsigned crc32_be;
if(find_embedded_crc32(hash_file_path, &crc32_be)) {
/* initialize file_info structure */
memset(&info, 0, sizeof(info));
info.full_path = rsh_strdup(hash_file_path);
info.file = file;
file_info_set_print_path(&info, info.full_path);
info.sums_flags = info.hc.hash_mask = RHASH_CRC32;
info.hc.flags = HC_HAS_EMBCRC32;
info.hc.embedded_crc32_be = crc32_be;
res = verify_sums(&info);
fflush(rhash_data.out);
if(!rhash_data.interrupted) {
if(res == 0) rhash_data.ok++;
else if(res == -1 && errno == ENOENT) rhash_data.miss++;
rhash_data.processed++;
}
free(info.full_path);
file_info_destroy(&info);
} else {
log_warning(_("file name doesn't contain a CRC32: %s\n"), hash_file_path);
return -1;
}
return 0;
}
/* initialize statistics */
rhash_data.processed = rhash_data.ok = rhash_data.miss = 0;
rhash_data.total_size = 0;
if(file->mode & FILE_IFSTDIN) {
fd = stdin;
hash_file_path = "<stdin>";
} else if( !(fd = rsh_fopen_bin(hash_file_path, "rb") )) {
log_file_error(hash_file_path);
return -1;
}
pos = strlen(hash_file_path)+16;
ralign = str_set(buf, '-', (pos < 80 ? 80 - (int)pos : 2));
fprintf(rhash_data.out, _("\n--( Verifying %s )%s\n"), hash_file_path, ralign);
fflush(rhash_data.out);
rhash_timer_start(&timer);
/* mark the directory part of the path, by setting the pos index */
if(chdir) {
pos = strlen(hash_file_path);
for(; pos > 0 && !IS_PATH_SEPARATOR(hash_file_path[pos]); pos--);
if(IS_PATH_SEPARATOR(hash_file_path[pos])) pos++;
} else pos = 0;
/* read crc file line by line */
for(line_num = 0; fgets(buf, 2048, fd); line_num++)
{
char* line = buf;
char* path_without_ext = NULL;
/* skip unicode BOM */
if(line_num == 0 && buf[0] == (char)0xEF && buf[1] == (char)0xBB && buf[2] == (char)0xBF) line += 3;
if(*line == 0) continue; /* skip empty lines */
if(is_binary_string(line)) {
log_error(_("file is binary: %s\n"), hash_file_path);
if(fd != stdin) fclose(fd);
return -1;
}
/* skip comments and empty lines */
if(IS_COMMENT(*line) || *line == '\r' || *line == '\n') continue;
memset(&info, 0, sizeof(info));
if(!hash_check_parse_line(line, &info.hc, !feof(fd))) continue;
if(info.hc.hash_mask == 0) continue;
info.print_path = info.hc.file_path;
info.sums_flags = info.hc.hash_mask;
/* see if crc file contains a hash sum without a filename */
if(info.print_path == NULL) {
char* point;
path_without_ext = rsh_strdup(hash_file_path);
point = strrchr(path_without_ext, '.');
if(point) {
*point = '\0';
file_info_set_print_path(&info, path_without_ext);
}
}
if(info.print_path != NULL) {
file_t file_to_check;
int is_absolute = IS_PATH_SEPARATOR(info.print_path[0]);
IF_WINDOWS(is_absolute = is_absolute || (info.print_path[0] && info.print_path[1] == ':'));
/* if filename shall be prepended by a directory path */
if(pos && !is_absolute) {
size_t len = strlen(info.print_path);
info.full_path = (char*)rsh_malloc(pos + len + 1);
memcpy(info.full_path, hash_file_path, pos);
strcpy(info.full_path + pos, info.print_path);
} else {
info.full_path = rsh_strdup(info.print_path);
}
memset(&file_to_check, 0, sizeof(file_t));
file_to_check.path = info.full_path;
rsh_file_stat(&file_to_check);
info.file = &file_to_check;
/* verify hash sums of the file */
res = verify_sums(&info);
fflush(rhash_data.out);
rsh_file_cleanup(&file_to_check);
file_info_destroy(&info);
if(rhash_data.interrupted) {
free(path_without_ext);
break;
}
/* update statistics */
if(res == 0) rhash_data.ok++;
else if(res == -1 && errno == ENOENT) rhash_data.miss++;
rhash_data.processed++;
}
free(path_without_ext);
}
time = rhash_timer_stop(&timer);
fprintf(rhash_data.out, "%s\n", str_set(buf, '-', 80));
print_check_stats();
if(rhash_data.processed != rhash_data.ok) rhash_data.error_flag = 1;
if(opt.flags & OPT_SPEED && rhash_data.processed > 1) {
print_time_stats(time, rhash_data.total_size, 1);
}
rhash_data.processed = 0;
res = ferror(fd); /* check that crc file has been read without errors */
if(fd != stdin) fclose(fd);
return (res == 0 ? 0 : -1);
}
/**
* Calculate and print file hash sums using printf format.
*
* @param out a stream to print to
* @param file the file to calculate sums for
* @param print_path the path to print
* @return 0 on success, -1 on fail
*/
int calculate_and_print_sums(FILE* out, file_t* file, const char *print_path)
{
struct file_info info;
timedelta_t timer;
int res = 0;
memset(&info, 0, sizeof(info));
info.file = file;
info.full_path = rsh_strdup(file->path);
file_info_set_print_path(&info, print_path);
info.size = 0;
info.sums_flags = opt.sum_flags;
if(file->mode & FILE_IFSTDIN) {
print_path = "(stdin)";
} else {
if(file->mode & FILE_IFDIR) return 0; /* don't handle directories */
info.size = file->size; /* total size, in bytes */
}
/* initialize percents output */
init_percents(&info);
rhash_timer_start(&timer);
if(info.sums_flags) {
/* calculate sums */
if(calc_sums(&info) < 0) {
/* print error unless sharing access error occurred */
if(errno == EACCES) return 0;
log_file_error(file->path);
res = -1;
}
if(rhash_data.interrupted) {
report_interrupted();
return 0;
}
}
info.time = rhash_timer_stop(&timer);
finish_percents(&info, res);
if(opt.flags & OPT_EMBED_CRC) {
/* rename the file */
rename_file_by_embeding_crc32(&info);
}
if((opt.mode & MODE_TORRENT) && !opt.bt_batch_file) {
save_torrent(&info);
}
if((opt.mode & MODE_UPDATE) && opt.fmt == FMT_SFV) {
/* updating SFV file: print SFV header line */
print_sfv_header_line(rhash_data.upd_fd, file, 0);
if(opt.flags & OPT_VERBOSE) {
print_sfv_header_line(rhash_data.log, file, 0);
fflush(rhash_data.log);
}
rsh_file_cleanup(file);
}
if(rhash_data.print_list && res >= 0) {
if (!opt.bt_batch_file) {
print_line(out, rhash_data.print_list, &info);
fflush(out);
/* print calculated line to stderr or log-file if verbose */
if((opt.mode & MODE_UPDATE) && (opt.flags & OPT_VERBOSE)) {
print_line(rhash_data.log, rhash_data.print_list, &info);
fflush(rhash_data.log);
}
}
if((opt.flags & OPT_SPEED) && info.sums_flags) {
print_file_time_stats(&info);
}
}
free(info.full_path);
file_info_destroy(&info);
return res;
}
/**
* RHash program entry point.
*
* @param argc number of program arguments including the program path
* @param argv program arguments
* @return the program exit code, zero on success and 1 on error
*/
int main(int argc, char *argv[])
{
find_file_options search_opt;
timedelta_t timer;
int sfv;
i18n_initialize(); /* initialize locale and translation */
memset(&rhash_data, 0, sizeof(rhash_data));
rhash_data.out = stdout; /* set initial output streams */
rhash_data.log = stderr; /* can be altered by options later */
rhash_data.search_opt = &search_opt;
init_hash_info_table();
read_options(argc, argv); /* load config and parse command line options */
prev_sigint_handler = signal(SIGINT, ctrl_c_handler); /* install SIGINT handler */
rhash_library_init();
/* in benchmark mode just run benchmark and exit */
if(opt.mode & MODE_BENCHMARK) {
unsigned flags = (opt.flags & OPT_BENCH_RAW ? RHASH_BENCHMARK_CPB | RHASH_BENCHMARK_RAW : RHASH_BENCHMARK_CPB);
if((opt.flags & OPT_BENCH_RAW) == 0) {
fprintf(rhash_data.out, _("%s v%s benchmarking...\n"), PROGRAM_NAME, VERSION);
}
rhash_run_benchmark(opt.sum_flags, flags, rhash_data.out);
rsh_exit(0);
}
if(opt.n_files == 0) {
if(argc > 1) {
log_warning(_("no files/directories were specified at command line\n"));
}
/* print short usage help */
log_msg(_("Usage: %s [OPTION...] <FILE>...\n\n"
"Run `%s --help' for more help.\n"), CMD_FILENAME, CMD_FILENAME);
rsh_exit(0);
}
/* setup printf formating string */
rhash_data.printf_str = opt.printf_str;
if(opt.template_file) {
if(!load_printf_template()) rsh_exit(2);
} else if(!rhash_data.printf_str && !(opt.mode & (MODE_CHECK | MODE_CHECK_EMBEDDED))) {
/* initialize printf output format according to '--<hashname>' options */
init_printf_format( (rhash_data.template_text = rsh_str_new()) );
rhash_data.printf_str = rhash_data.template_text->str;
if(opt.flags & OPT_VERBOSE) {
char* str = rsh_strdup(rhash_data.printf_str);
log_msg(_("Format string is: %s\n"), str_trim(str));
free(str);
}
}
if(rhash_data.printf_str) {
rhash_data.print_list = parse_print_string(rhash_data.printf_str, &opt.sum_flags);
}
memset(&search_opt, 0, sizeof(search_opt));
search_opt.max_depth = (opt.flags & OPT_RECURSIVE ? opt.find_max_depth : 0);
search_opt.options = FIND_SKIP_DIRS;
search_opt.call_back = find_file_callback;
if((sfv = (opt.fmt == FMT_SFV && !opt.mode))) {
print_sfv_banner(rhash_data.out);
}
/* pre-process files */
if(sfv || opt.bt_batch_file) {
/* note: errors are not reported on pre-processing */
search_opt.call_back_data = (void*)1;
process_files((const char**)opt.files, opt.n_files, &search_opt);
fflush(rhash_data.out);
}
/* measure total processing time */
rhash_timer_start(&timer);
rhash_data.processed = 0;
/* process files */
search_opt.options |= FIND_LOG_ERRORS;
search_opt.call_back_data = (void*)0;
process_files((const char**)opt.files, opt.n_files, &search_opt);
if((opt.mode & MODE_CHECK_EMBEDDED) && rhash_data.processed > 1) {
print_check_stats();
}
if(!rhash_data.interrupted)
{
if(opt.bt_batch_file && rhash_data.rctx) {
rhash_final(rhash_data.rctx, 0);
save_torrent_to(opt.bt_batch_file, rhash_data.rctx);
}
if((opt.flags & OPT_SPEED) &&
!(opt.mode & (MODE_CHECK | MODE_UPDATE)) &&
rhash_data.processed > 1) {
double time = rhash_timer_stop(&timer);
print_time_stats(time, rhash_data.total_size, 1);
}
} else {
/* check if interruption was not reported yet */
if(rhash_data.interrupted == 1) report_interrupted();
}
options_destroy(&opt);
rhash_destroy(&rhash_data);
/* return non-zero error code if error occurred */
return (rhash_data.error_flag ? 1 :
search_opt.errors_count ? 2 :
rhash_data.interrupted ? 3 : 0);
}
/**
* Benchmark a hash algorithm.
*
* @deprecated This function shall be removed soon, since
* it is not related to the hashing library main functionality.
*
* @param hash_id hash algorithm identifier
* @param flags benchmark flags, can be RHASH_BENCHMARK_QUIET and RHASH_BENCHMARK_CPB
* @param output the stream to print results
*/
void rhash_run_benchmark(unsigned hash_id, unsigned flags, FILE* output)
{
unsigned char ALIGN_ATTR(16) message[8192]; /* 8 KiB */
timedelta_t timer;
int i, j;
size_t sz_mb, msg_size;
double time, total_time = 0;
const int rounds = 4;
const char* hash_name;
unsigned char out[130];
#ifdef HAVE_TSC
double cpb = 0;
#endif /* HAVE_TSC */
#ifdef _WIN32
benchmark_cpu_init(); /* set cpu affinity to improve test results */
#endif
/* set message size for fast and slow hash functions */
msg_size = 1073741824 / 2;
if (hash_id & (RHASH_WHIRLPOOL | RHASH_SNEFRU128 | RHASH_SNEFRU256 | RHASH_SHA3_224 | RHASH_SHA3_256 | RHASH_SHA3_384 | RHASH_SHA3_512)) {
msg_size /= 8;
} else if (hash_id & (RHASH_GOST | RHASH_GOST_CRYPTOPRO | RHASH_SHA384 | RHASH_SHA512)) {
msg_size /= 2;
}
sz_mb = msg_size / (1 << 20); /* size in MiB */
hash_name = rhash_get_name(hash_id);
if (!hash_name) hash_name = ""; /* benchmarking several hashes*/
for (i = 0; i < (int)sizeof(message); i++) message[i] = i & 0xff;
for (j = 0; j < rounds; j++) {
rhash_timer_start(&timer);
hash_in_loop(hash_id, message, sizeof(message), (int)(msg_size / sizeof(message)), out);
time = rhash_timer_stop(&timer);
total_time += time;
if ((flags & (RHASH_BENCHMARK_QUIET | RHASH_BENCHMARK_RAW)) == 0) {
fprintf(output, "%s %u MiB calculated in %.3f sec, %.3f MBps\n", hash_name, (unsigned)sz_mb, time, (double)sz_mb / time);
fflush(output);
}
}
#if defined(HAVE_TSC)
/* measure the CPU "clocks per byte" speed */
if (flags & RHASH_BENCHMARK_CPB) {
unsigned int c1 = -1, c2 = -1;
unsigned volatile long long cy0, cy1, cy2;
int msg_size = 128 * 1024;
/* make 200 tries */
for (i = 0; i < 200; i++) {
cy0 = read_tsc();
hash_in_loop(hash_id, message, sizeof(message), msg_size / sizeof(message), out);
cy1 = read_tsc();
hash_in_loop(hash_id, message, sizeof(message), msg_size / sizeof(message), out);
hash_in_loop(hash_id, message, sizeof(message), msg_size / sizeof(message), out);
cy2 = read_tsc();
cy2 -= cy1;
cy1 -= cy0;
c1 = (unsigned int)(c1 > cy1 ? cy1 : c1);
c2 = (unsigned int)(c2 > cy2 ? cy2 : c2);
}
cpb = ((c2 - c1) + 1) / (double)msg_size;
}
#endif /* HAVE_TSC */
if (flags & RHASH_BENCHMARK_RAW) {
/* output result in a "raw" machine-readable format */
fprintf(output, "%s\t%u\t%.3f\t%.3f", hash_name, ((unsigned)sz_mb * rounds), total_time, (double)(sz_mb * rounds) / total_time);
#if defined(HAVE_TSC)
if (flags & RHASH_BENCHMARK_CPB) fprintf(output, "\t%.2f", cpb);
#endif /* HAVE_TSC */
fprintf(output, "\n");
} else {
fprintf(output, "%s %u MiB total in %.3f sec, %.3f MBps", hash_name, ((unsigned)sz_mb * rounds), total_time, (double)(sz_mb * rounds) / total_time);
#if defined(HAVE_TSC)
if (flags & RHASH_BENCHMARK_CPB) fprintf(output, ", CPB=%.2f", cpb);
#endif /* HAVE_TSC */
fprintf(output, "\n");
}
}