Пример #1
0
bam_header_t *sam_header_read2(const char *fn)
{
	bam_header_t *header;
	int c, dret, ret;
	gzFile fp;
	kstream_t *ks;
	kstring_t *str;
	kh_ref_t *hash;
	khiter_t k;
	hash = kh_init(ref);
	fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
	assert(fp);
	ks = ks_init(fp);
	str = (kstring_t*)calloc(1, sizeof(kstring_t));
	while (ks_getuntil(ks, 0, str, &dret) > 0) {
		char *s = strdup(str->s);
		int len, i;
		i = kh_size(hash);
		ks_getuntil(ks, 0, str, &dret);
		len = atoi(str->s);
		k = kh_put(ref, hash, s, &ret);
		kh_value(hash, k) = (uint64_t)len<<32 | i;
		if (dret != '\n')
			while ((c = ks_getc(ks)) != '\n' && c != -1);
	}
	ks_destroy(ks);
	gzclose(fp);
	free(str->s); free(str);
	fprintf(stderr, "[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
	header = hash2header(hash);
	kh_destroy(ref, hash);
	return header;
}
Пример #2
0
bam_header_t *sam_header_read2(const char *fn)
{
	bam_header_t *header;
	int c, dret, ret, error = 0;
	gzFile fp;
	kstream_t *ks;
	kstring_t *str;
	kh_ref_t *hash;
	khiter_t k;
	if (fn == 0) return 0;
	fp = gzopen(fn, "r");
	if (fp == 0) return 0;
	hash = kh_init(ref);
	ks = ks_init(fp);
	str = (kstring_t*)calloc(1, sizeof(kstring_t));
	while (ks_getuntil(ks, 0, str, &dret) > 0) {
		char *s = malloc(strlen(str->s) + 1);
		strcpy(s,str->s);

		int len, i;
		i = kh_size(hash);
		ks_getuntil(ks, 0, str, &dret);
		len = atoi(str->s);
		k = kh_put(ref, hash, s, &ret);
		if (ret == 0) {
			Rprintf("[sam_header_read2] duplicated sequence name: %s\n", s);
			error = 1;
		}
		kh_value(hash, k) = (uint64_t)len<<32 | i;
		if (dret != '\n')
			while ((c = ks_getc(ks)) != '\n' && c != -1);
	}
	ks_destroy(ks);
	gzclose(fp);
	free(str->s); free(str);
	Rprintf("[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
	if (error) return 0;
	header = hash2header(hash);
	kh_destroy(ref, hash);
	return header;
}