| 
0
 | 
     1 #include <ctype.h>
 | 
| 
 | 
     2 #include "bam.h"
 | 
| 
 | 
     3 #include "khash.h"
 | 
| 
 | 
     4 typedef char *str_p;
 | 
| 
 | 
     5 KHASH_MAP_INIT_STR(s, int)
 | 
| 
 | 
     6 KHASH_MAP_INIT_STR(r2l, str_p)
 | 
| 
 | 
     7 
 | 
| 
 | 
     8 void bam_aux_append(bam1_t *b, const char tag[2], char type, int len, uint8_t *data)
 | 
| 
 | 
     9 {
 | 
| 
 | 
    10 	int ori_len = b->data_len;
 | 
| 
 | 
    11 	b->data_len += 3 + len;
 | 
| 
 | 
    12 	b->l_aux += 3 + len;
 | 
| 
 | 
    13 	if (b->m_data < b->data_len) {
 | 
| 
 | 
    14 		b->m_data = b->data_len;
 | 
| 
 | 
    15 		kroundup32(b->m_data);
 | 
| 
 | 
    16 		b->data = (uint8_t*)realloc(b->data, b->m_data);
 | 
| 
 | 
    17 	}
 | 
| 
 | 
    18 	b->data[ori_len] = tag[0]; b->data[ori_len + 1] = tag[1];
 | 
| 
 | 
    19 	b->data[ori_len + 2] = type;
 | 
| 
 | 
    20 	memcpy(b->data + ori_len + 3, data, len);
 | 
| 
 | 
    21 }
 | 
| 
 | 
    22 
 | 
| 
 | 
    23 uint8_t *bam_aux_get_core(bam1_t *b, const char tag[2])
 | 
| 
 | 
    24 {
 | 
| 
 | 
    25 	return bam_aux_get(b, tag);
 | 
| 
 | 
    26 }
 | 
| 
 | 
    27 
 | 
| 
 | 
    28 #define __skip_tag(s) do { \
 | 
| 
 | 
    29 		int type = toupper(*(s)); \
 | 
| 
 | 
    30 		++(s); \
 | 
| 
 | 
    31 		if (type == 'Z' || type == 'H') { while (*(s)) ++(s); ++(s); } \
 | 
| 
 | 
    32 		else if (type == 'B') (s) += 5 + bam_aux_type2size(*(s)) * (*(int32_t*)((s)+1)); \
 | 
| 
 | 
    33 		else (s) += bam_aux_type2size(type); \
 | 
| 
 | 
    34 	} while(0)
 | 
| 
 | 
    35 
 | 
| 
 | 
    36 uint8_t *bam_aux_get(const bam1_t *b, const char tag[2])
 | 
| 
 | 
    37 {
 | 
| 
 | 
    38 	uint8_t *s;
 | 
| 
 | 
    39 	int y = tag[0]<<8 | tag[1];
 | 
| 
 | 
    40 	s = bam1_aux(b);
 | 
| 
 | 
    41 	while (s < b->data + b->data_len) {
 | 
| 
 | 
    42 		int x = (int)s[0]<<8 | s[1];
 | 
| 
 | 
    43 		s += 2;
 | 
| 
 | 
    44 		if (x == y) return s;
 | 
| 
 | 
    45 		__skip_tag(s);
 | 
| 
 | 
    46 	}
 | 
| 
 | 
    47 	return 0;
 | 
| 
 | 
    48 }
 | 
| 
 | 
    49 // s MUST BE returned by bam_aux_get()
 | 
| 
 | 
    50 int bam_aux_del(bam1_t *b, uint8_t *s)
 | 
| 
 | 
    51 {
 | 
| 
 | 
    52 	uint8_t *p, *aux;
 | 
| 
 | 
    53 	aux = bam1_aux(b);
 | 
| 
 | 
    54 	p = s - 2;
 | 
| 
 | 
    55 	__skip_tag(s);
 | 
| 
 | 
    56 	memmove(p, s, b->l_aux - (s - aux));
 | 
| 
 | 
    57 	b->data_len -= s - p;
 | 
| 
 | 
    58 	b->l_aux -= s - p;
 | 
| 
 | 
    59 	return 0;
 | 
| 
 | 
    60 }
 | 
| 
 | 
    61 
 | 
| 
 | 
    62 void bam_init_header_hash(bam_header_t *header)
 | 
| 
 | 
    63 {
 | 
| 
 | 
    64 	if (header->hash == 0) {
 | 
| 
 | 
    65 		int ret, i;
 | 
| 
 | 
    66 		khiter_t iter;
 | 
| 
 | 
    67 		khash_t(s) *h;
 | 
| 
 | 
    68 		header->hash = h = kh_init(s);
 | 
| 
 | 
    69 		for (i = 0; i < header->n_targets; ++i) {
 | 
| 
 | 
    70 			iter = kh_put(s, h, header->target_name[i], &ret);
 | 
| 
 | 
    71 			kh_value(h, iter) = i;
 | 
| 
 | 
    72 		}
 | 
| 
 | 
    73 	}
 | 
| 
 | 
    74 }
 | 
| 
 | 
    75 
 | 
| 
 | 
    76 void bam_destroy_header_hash(bam_header_t *header)
 | 
| 
 | 
    77 {
 | 
| 
 | 
    78 	if (header->hash)
 | 
| 
 | 
    79 		kh_destroy(s, (khash_t(s)*)header->hash);
 | 
| 
 | 
    80 }
 | 
| 
 | 
    81 
 | 
| 
 | 
    82 int32_t bam_get_tid(const bam_header_t *header, const char *seq_name)
 | 
| 
 | 
    83 {
 | 
| 
 | 
    84 	khint_t k;
 | 
| 
 | 
    85 	khash_t(s) *h = (khash_t(s)*)header->hash;
 | 
| 
 | 
    86 	k = kh_get(s, h, seq_name);
 | 
| 
 | 
    87 	return k == kh_end(h)? -1 : kh_value(h, k);
 | 
| 
 | 
    88 }
 | 
| 
 | 
    89 
 | 
| 
 | 
    90 int bam_parse_region(bam_header_t *header, const char *str, int *ref_id, int *begin, int *end)
 | 
| 
 | 
    91 {
 | 
| 
 | 
    92 	char *s, *p;
 | 
| 
 | 
    93 	int i, l, k;
 | 
| 
 | 
    94 	khiter_t iter;
 | 
| 
 | 
    95 	khash_t(s) *h;
 | 
| 
 | 
    96 
 | 
| 
 | 
    97 	bam_init_header_hash(header);
 | 
| 
 | 
    98 	h = (khash_t(s)*)header->hash;
 | 
| 
 | 
    99 
 | 
| 
 | 
   100 	l = strlen(str);
 | 
| 
 | 
   101 	p = s = (char*)malloc(l+1);
 | 
| 
 | 
   102 	/* squeeze out "," */
 | 
| 
 | 
   103 	for (i = k = 0; i != l; ++i)
 | 
| 
 | 
   104 		if (str[i] != ',' && !isspace(str[i])) s[k++] = str[i];
 | 
| 
 | 
   105 	s[k] = 0;
 | 
| 
 | 
   106 	for (i = 0; i != k; ++i) if (s[i] == ':') break;
 | 
| 
 | 
   107 	s[i] = 0;
 | 
| 
 | 
   108 	iter = kh_get(s, h, s); /* get the ref_id */
 | 
| 
 | 
   109 	if (iter == kh_end(h)) { // name not found
 | 
| 
 | 
   110 		*ref_id = -1; free(s);
 | 
| 
 | 
   111 		return -1;
 | 
| 
 | 
   112 	}
 | 
| 
 | 
   113 	*ref_id = kh_value(h, iter);
 | 
| 
 | 
   114 	if (i == k) { /* dump the whole sequence */
 | 
| 
 | 
   115 		*begin = 0; *end = 1<<29; free(s);
 | 
| 
 | 
   116 		return 0;
 | 
| 
 | 
   117 	}
 | 
| 
 | 
   118 	for (p = s + i + 1; i != k; ++i) if (s[i] == '-') break;
 | 
| 
 | 
   119 	*begin = atoi(p);
 | 
| 
 | 
   120 	if (i < k) {
 | 
| 
 | 
   121 		p = s + i + 1;
 | 
| 
 | 
   122 		*end = atoi(p);
 | 
| 
 | 
   123 	} else *end = 1<<29;
 | 
| 
 | 
   124 	if (*begin > 0) --*begin;
 | 
| 
 | 
   125 	free(s);
 | 
| 
 | 
   126 	if (*begin > *end) {
 | 
| 
 | 
   127 		fprintf(stderr, "[bam_parse_region] invalid region.\n");
 | 
| 
 | 
   128 		return -1;
 | 
| 
 | 
   129 	}
 | 
| 
 | 
   130 	return 0;
 | 
| 
 | 
   131 }
 | 
| 
 | 
   132 
 | 
| 
 | 
   133 int32_t bam_aux2i(const uint8_t *s)
 | 
| 
 | 
   134 {
 | 
| 
 | 
   135 	int type;
 | 
| 
 | 
   136 	if (s == 0) return 0;
 | 
| 
 | 
   137 	type = *s++;
 | 
| 
 | 
   138 	if (type == 'c') return (int32_t)*(int8_t*)s;
 | 
| 
 | 
   139 	else if (type == 'C') return (int32_t)*(uint8_t*)s;
 | 
| 
 | 
   140 	else if (type == 's') return (int32_t)*(int16_t*)s;
 | 
| 
 | 
   141 	else if (type == 'S') return (int32_t)*(uint16_t*)s;
 | 
| 
 | 
   142 	else if (type == 'i' || type == 'I') return *(int32_t*)s;
 | 
| 
 | 
   143 	else return 0;
 | 
| 
 | 
   144 }
 | 
| 
 | 
   145 
 | 
| 
 | 
   146 float bam_aux2f(const uint8_t *s)
 | 
| 
 | 
   147 {
 | 
| 
 | 
   148 	int type;
 | 
| 
 | 
   149 	type = *s++;
 | 
| 
 | 
   150 	if (s == 0) return 0.0;
 | 
| 
 | 
   151 	if (type == 'f') return *(float*)s;
 | 
| 
 | 
   152 	else return 0.0;
 | 
| 
 | 
   153 }
 | 
| 
 | 
   154 
 | 
| 
 | 
   155 double bam_aux2d(const uint8_t *s)
 | 
| 
 | 
   156 {
 | 
| 
 | 
   157 	int type;
 | 
| 
 | 
   158 	type = *s++;
 | 
| 
 | 
   159 	if (s == 0) return 0.0;
 | 
| 
 | 
   160 	if (type == 'd') return *(double*)s;
 | 
| 
 | 
   161 	else return 0.0;
 | 
| 
 | 
   162 }
 | 
| 
 | 
   163 
 | 
| 
 | 
   164 char bam_aux2A(const uint8_t *s)
 | 
| 
 | 
   165 {
 | 
| 
 | 
   166 	int type;
 | 
| 
 | 
   167 	type = *s++;
 | 
| 
 | 
   168 	if (s == 0) return 0;
 | 
| 
 | 
   169 	if (type == 'A') return *(char*)s;
 | 
| 
 | 
   170 	else return 0;
 | 
| 
 | 
   171 }
 | 
| 
 | 
   172 
 | 
| 
 | 
   173 char *bam_aux2Z(const uint8_t *s)
 | 
| 
 | 
   174 {
 | 
| 
 | 
   175 	int type;
 | 
| 
 | 
   176 	type = *s++;
 | 
| 
 | 
   177 	if (s == 0) return 0;
 | 
| 
 | 
   178 	if (type == 'Z' || type == 'H') return (char*)s;
 | 
| 
 | 
   179 	else return 0;
 | 
| 
 | 
   180 }
 | 
| 
 | 
   181 
 | 
| 
 | 
   182 #ifdef _WIN32
 | 
| 
 | 
   183 double drand48()
 | 
| 
 | 
   184 {
 | 
| 
 | 
   185 	return (double)rand() / RAND_MAX;
 | 
| 
 | 
   186 }
 | 
| 
 | 
   187 #endif
 |