#include <ctype.h>
#include "bam.h"
#include "khash.h"
+typedef char *str_p;
KHASH_MAP_INIT_STR(s, int)
+KHASH_MAP_INIT_STR(r2l, str_p)
void bam_aux_append(bam1_t *b, const char tag[2], char type, int len, uint8_t *data)
{
memcpy(b->data + ori_len + 3, data, len);
}
-uint8_t *bam_aux_get(bam1_t *b, const char tag[2])
+uint8_t *bam_aux_get_core(bam1_t *b, const char tag[2])
+{
+ return bam_aux_get(b, tag);
+}
+
+uint8_t *bam_aux_get(const bam1_t *b, const char tag[2])
{
uint8_t *s;
int y = tag[0]<<8 | tag[1];
else if (type == 'S') s += 2;
else if (type == 'I' || type == 'F') s += 4;
else if (type == 'D') s += 8;
- else if (type == 'Z' || type == 'H') { while (*s) putchar(*s++); ++s; }
+ else if (type == 'Z' || type == 'H') { while (*s) ++s; ++s; }
}
return 0;
}
return k == kh_end(h)? -1 : kh_value(h, k);
}
-void bam_parse_region(bam_header_t *header, const char *str, int *ref_id, int *begin, int *end)
+int bam_parse_region(bam_header_t *header, const char *str, int *ref_id, int *begin, int *end)
{
char *s, *p;
int i, l, k;
iter = kh_get(s, h, s); /* get the ref_id */
if (iter == kh_end(h)) { // name not found
*ref_id = -1; free(s);
- return;
+ return -1;
}
*ref_id = kh_value(h, iter);
if (i == k) { /* dump the whole sequence */
*begin = 0; *end = 1<<29; free(s);
- return;
+ return -1;
}
for (p = s + i + 1; i != k; ++i) if (s[i] == '-') break;
*begin = atoi(p);
*end = atoi(p);
} else *end = 1<<29;
if (*begin > 0) --*begin;
- assert(*begin <= *end);
free(s);
+ if (*begin > *end) {
+ fprintf(stderr, "[bam_parse_region] invalid region.\n");
+ return -1;
+ }
+ return 0;
}
int32_t bam_aux2i(const uint8_t *s)
else return 0;
}
-char bam_aux_getCSi(bam1_t *b, int i)
-{
- uint8_t *c = bam_aux_get(b, "CS");
- char *cs = NULL;
-
- // return the base if the tag was not found
- if(0 == c) return 0;
+/******************
+ * rg2lib related *
+ ******************/
- cs = bam_aux2Z(c);
- // adjust for strandedness and leading adaptor
- if(bam1_strand(b)) i = strlen(cs) - 1 - i;
- else i++;
- return cs[i];
+int bam_strmap_put(void *rg2lib, const char *rg, const char *lib)
+{
+ int ret;
+ khint_t k;
+ khash_t(r2l) *h = (khash_t(r2l)*)rg2lib;
+ char *key;
+ if (h == 0) return 1;
+ key = strdup(rg);
+ k = kh_put(r2l, h, key, &ret);
+ if (ret) kh_val(h, k) = strdup(lib);
+ else {
+ fprintf(stderr, "[bam_rg2lib_put] duplicated @RG ID: %s\n", rg);
+ free(key);
+ }
+ return 0;
}
-char bam_aux_getCQi(bam1_t *b, int i)
+const char *bam_strmap_get(const void *rg2lib, const char *rg)
{
- uint8_t *c = bam_aux_get(b, "CQ");
- char *cq = NULL;
-
- // return the base if the tag was not found
- if(0 == c) return 0;
-
- cq = bam_aux2Z(c);
- // adjust for strandedness
- if(bam1_strand(b)) i = strlen(cq) - 1 - i;
- return cq[i];
+ const khash_t(r2l) *h = (const khash_t(r2l)*)rg2lib;
+ khint_t k;
+ if (h == 0) return 0;
+ k = kh_get(r2l, h, rg);
+ if (k != kh_end(h)) return (const char*)kh_val(h, k);
+ else return 0;
}
-char bam_aux_nt2int(char a)
+void *bam_strmap_dup(const void *rg2lib)
{
- switch(toupper(a)) {
- case 'A':
- return 0;
- break;
- case 'C':
- return 1;
- break;
- case 'G':
- return 2;
- break;
- case 'T':
- return 3;
- break;
- default:
- return 4;
- break;
+ const khash_t(r2l) *h = (const khash_t(r2l)*)rg2lib;
+ khash_t(r2l) *g;
+ khint_t k, l;
+ int ret;
+ if (h == 0) return 0;
+ g = kh_init(r2l);
+ for (k = kh_begin(h); k < kh_end(h); ++k) {
+ if (kh_exist(h, k)) {
+ char *key = strdup(kh_key(h, k));
+ l = kh_put(r2l, g, key, &ret);
+ kh_val(g, l) = strdup(kh_val(h, k));
+ }
}
+ return g;
}
-char bam_aux_ntnt2cs(char a, char b)
+void *bam_strmap_init()
{
- a = bam_aux_nt2int(a);
- b = bam_aux_nt2int(b);
- if(4 == a || 4 == b) return '4';
- return "0123"[(int)(a ^ b)];
+ return (void*)kh_init(r2l);
}
-char bam_aux_getCEi(bam1_t *b, int i)
+void bam_strmap_destroy(void *rg2lib)
{
- int cs_i;
- uint8_t *c = bam_aux_get(b, "CS");
- char *cs = NULL;
- char prev_b, cur_b;
- char cur_color, cor_color;
-
- // return the base if the tag was not found
- if(0 == c) return 0;
-
- cs = bam_aux2Z(c);
-
- // adjust for strandedness and leading adaptor
- if(bam1_strand(b)) { //reverse strand
- cs_i = strlen(cs) - 1 - i;
- // get current color
- cur_color = cs[cs_i];
- // get previous base
- prev_b = (0 == cs_i) ? cs[0] : bam_nt16_rev_table[bam1_seqi(bam1_seq(b), i+1)];
- // get current base
- cur_b = bam_nt16_rev_table[bam1_seqi(bam1_seq(b), i)];
- }
- else {
- cs_i=i+1;
- // get current color
- cur_color = cs[cs_i];
- // get previous base
- prev_b = (0 == i) ? cs[0] : bam_nt16_rev_table[bam1_seqi(bam1_seq(b), i-1)];
- // get current base
- cur_b = bam_nt16_rev_table[bam1_seqi(bam1_seq(b), i)];
- }
-
- // corrected color
- cor_color = bam_aux_ntnt2cs(prev_b, cur_b);
-
- if(cur_color == cor_color) {
- return '-';
- }
- else {
- return cur_color;
+ khash_t(r2l) *h = (khash_t(r2l)*)rg2lib;
+ khint_t k;
+ if (h == 0) return;
+ for (k = kh_begin(h); k < kh_end(h); ++k) {
+ if (kh_exist(h, k)) {
+ free((char*)kh_key(h, k)); free(kh_val(h, k));
+ }
}
+ kh_destroy(r2l, h);
}
projects / samtools.git / blobdiff