static void md5_one(const char *fn)
{
hts_md5_context *md5_one, *md5_all;
int l, i, k;
gzFile fp;
kseq_t *seq;
unsigned char unordered[16], digest[16];
char hex[33];
for (l = 0; l < 16; ++l) unordered[l] = 0;
fp = strcmp(fn, "-")? gzopen(fn, "r") : gzdopen(fileno(stdin), "r");
if (fp == 0) {
fprintf(stderr, "md5fa: %s: No such file or directory\n", fn);
exit(1);
}
if (!(md5_all = hts_md5_init()))
exit(1);
if (!(md5_one = hts_md5_init())) {
hts_md5_destroy(md5_all);
exit(1);
}
seq = kseq_init(fp);
while ((l = kseq_read(seq)) >= 0) {
for (i = k = 0; i < seq->seq.l; ++i) {
if (islower(seq->seq.s[i])) seq->seq.s[k++] = toupper(seq->seq.s[i]);
else if (isupper(seq->seq.s[i])) seq->seq.s[k++] = seq->seq.s[i];
}
hts_md5_reset(md5_one);
hts_md5_update(md5_one, (unsigned char*)seq->seq.s, k);
hts_md5_final(digest, md5_one);
hts_md5_hex(hex, digest);
for (l = 0; l < 16; ++l)
unordered[l] ^= digest[l];
printf("%s %s %s\n", hex, fn, seq->name.s);
hts_md5_update(md5_all, (unsigned char*)seq->seq.s, k);
}
hts_md5_final(digest, md5_all);
kseq_destroy(seq);
hts_md5_hex(hex, digest);
printf("%s %s >ordered\n", hex, fn);
hts_md5_hex(hex, unordered);
printf("%s %s >unordered\n", hex, fn);
hts_md5_destroy(md5_all);
hts_md5_destroy(md5_one);
}
static void md5_one(const char *fn)
{
unsigned char buf[4096], digest[16];
char hex[33];
hts_md5_context *md5;
int l;
FILE *fp;
fp = strcmp(fn, "-")? fopen(fn, "r") : stdin;
if (fp == 0) {
fprintf(stderr, "md5sum: %s: No such file or directory\n", fn);
exit(1);
}
if (!(md5 = hts_md5_init())) {
fprintf(stderr, "md5sum: %s: Failed to allocate md5 buffer\n", fn);
exit(1);
}
while ((l = fread(buf, 1, 4096, fp)) > 0)
hts_md5_update(md5, buf, l);
hts_md5_final(digest, md5);
if (fp != stdin) fclose(fp);
hts_md5_hex(hex, digest);
printf("%s %s\n", hex, fn);
hts_md5_destroy(md5);
}
static void write_dict(const char *fn, args_t *args)
{
hts_md5_context *md5;
int l, i, k;
gzFile fp;
kseq_t *seq;
unsigned char digest[16];
char hex[33];
fp = strcmp(fn, "-") ? gzopen(fn, "r") : gzdopen(fileno(stdin), "r");
if (fp == 0) {
fprintf(stderr, "dict: %s: No such file or directory\n", fn);
exit(1);
}
FILE *out = stdout;
if (args->output_fname) {
out = fopen(args->output_fname, "w");
if (out == NULL) {
fprintf(stderr, "dict: %s: Cannot open file for writing\n", args->output_fname);
exit(1);
}
}
if (!(md5 = hts_md5_init()))
exit(1);
seq = kseq_init(fp);
if (args->header) fprintf(out, "@HD\tVN:1.0\tSO:unsorted\n");
while ((l = kseq_read(seq)) >= 0) {
for (i = k = 0; i < seq->seq.l; ++i) {
if (seq->seq.s[i] >= '!' && seq->seq.s[i] <= '~')
seq->seq.s[k++] = toupper(seq->seq.s[i]);
}
hts_md5_reset(md5);
hts_md5_update(md5, (unsigned char*)seq->seq.s, k);
hts_md5_final(digest, md5);
hts_md5_hex(hex, digest);
fprintf(out, "@SQ\tSN:%s\tLN:%d\tM5:%s", seq->name.s, k, hex);
if (args->uri)
fprintf(out, "\tUR:%s", args->uri);
else if (strcmp(fn, "-") != 0) {
#ifdef _WIN32
char *real_path = _fullpath(NULL, fn, PATH_MAX);
#else
char *real_path = realpath(fn, NULL);
#endif
fprintf(out, "\tUR:file://%s", real_path);
free(real_path);
}
if (args->assembly) fprintf(out, "\tAS:%s", args->assembly);
if (args->species) fprintf(out, "\tSP:%s", args->species);
fprintf(out, "\n");
}
kseq_destroy(seq);
hts_md5_destroy(md5);
if (args->output_fname) fclose(out);
}