Merge commit 'upstream/0.1.18'

[samtools.git] / bam_aux.c

diff --git a/bam_aux.c b/bam_aux.c
index fbcd9822b233dab64f6c861fb332846acb951ce7..28b22e3059273da57dd8395fba5e29287dc3228b 100644 (file)
--- a/bam_aux.c
+++ b/bam_aux.c
@@ -26,14 +26,12 @@ uint8_t *bam_aux_get_core(bam1_t *b, const char tag[2])
 }
 
 #define __skip_tag(s) do { \
-               int type = toupper(*(s));                                                                               \
-               ++(s);                                                                                                                  \
-               if (type == 'C' || type == 'A') ++(s);                                                  \
-               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)) ++(s); ++(s); } \
-       } while (0)
+               int type = toupper(*(s)); \
+               ++(s); \
+               if (type == 'Z' || type == 'H') { while (*(s)) ++(s); ++(s); } \
+               else if (type == 'B') (s) += 5 + bam_aux_type2size(*(s)) * (*(int32_t*)((s)+1)); \
+               else (s) += bam_aux_type2size(type); \
+       } while(0)
 
 uint8_t *bam_aux_get(const bam1_t *b, const char tag[2])
 {
@@ -61,6 +59,23 @@ int bam_aux_del(bam1_t *b, uint8_t *s)
        return 0;
 }
 
+int bam_aux_drop_other(bam1_t *b, uint8_t *s)
+{
+       if (s) {
+               uint8_t *p, *aux;
+               aux = bam1_aux(b);
+               p = s - 2;
+               __skip_tag(s);
+               memmove(aux, p, s - p);
+               b->data_len -= b->l_aux - (s - p);
+               b->l_aux = s - p;
+       } else {
+               b->data_len -= b->l_aux;
+               b->l_aux = 0;
+       }
+       return 0;
+}
+
 void bam_init_header_hash(bam_header_t *header)
 {
        if (header->hash == 0) {
@@ -89,47 +104,56 @@ int32_t bam_get_tid(const bam_header_t *header, const char *seq_name)
        return k == kh_end(h)? -1 : kh_value(h, k);
 }
 
-int 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 *beg, int *end)
 {
-       char *s, *p;
-       int i, l, k;
+       char *s;
+       int i, l, k, name_end;
        khiter_t iter;
        khash_t(s) *h;
 
        bam_init_header_hash(header);
        h = (khash_t(s)*)header->hash;
 
-       l = strlen(str);
-       p = s = (char*)malloc(l+1);
-       /* squeeze out "," */
-       for (i = k = 0; i != l; ++i)
-               if (str[i] != ',' && !isspace(str[i])) s[k++] = str[i];
-       s[k] = 0;
-       for (i = 0; i != k; ++i) if (s[i] == ':') break;
-       s[i] = 0;
-       iter = kh_get(s, h, s); /* get the ref_id */
-       if (iter == kh_end(h)) { // name not found
-               *ref_id = -1; free(s);
-               return -1;
-       }
-       *ref_id = kh_value(h, iter);
-       if (i == k) { /* dump the whole sequence */
-               *begin = 0; *end = 1<<29; free(s);
-               return 0;
-       }
-       for (p = s + i + 1; i != k; ++i) if (s[i] == '-') break;
-       *begin = atoi(p);
-       if (i < k) {
-               p = s + i + 1;
-               *end = atoi(p);
-       } else *end = 1<<29;
-       if (*begin > 0) --*begin;
+       *ref_id = *beg = *end = -1;
+       name_end = l = strlen(str);
+       s = (char*)malloc(l+1);
+       // remove space
+       for (i = k = 0; i < l; ++i)
+               if (!isspace(str[i])) s[k++] = str[i];
+       s[k] = 0; l = k;
+       // determine the sequence name
+       for (i = l - 1; i >= 0; --i) if (s[i] == ':') break; // look for colon from the end
+       if (i >= 0) name_end = i;
+       if (name_end < l) { // check if this is really the end
+               int n_hyphen = 0;
+               for (i = name_end + 1; i < l; ++i) {
+                       if (s[i] == '-') ++n_hyphen;
+                       else if (!isdigit(s[i]) && s[i] != ',') break;
+               }
+               if (i < l || n_hyphen > 1) name_end = l; // malformated region string; then take str as the name
+               s[name_end] = 0;
+               iter = kh_get(s, h, s);
+               if (iter == kh_end(h)) { // cannot find the sequence name
+                       iter = kh_get(s, h, str); // try str as the name
+                       if (iter == kh_end(h)) {
+                               if (bam_verbose >= 2) fprintf(stderr, "[%s] fail to determine the sequence name.\n", __func__);
+                               free(s); return -1;
+                       } else s[name_end] = ':', name_end = l;
+               }
+       } else iter = kh_get(s, h, str);
+       *ref_id = kh_val(h, iter);
+       // parse the interval
+       if (name_end < l) {
+               for (i = k = name_end + 1; i < l; ++i)
+                       if (s[i] != ',') s[k++] = s[i];
+               s[k] = 0;
+               *beg = atoi(s + name_end + 1);
+               for (i = name_end + 1; i != k; ++i) if (s[i] == '-') break;
+               *end = i < k? atoi(s + i + 1) : 1<<29;
+               if (*beg > 0) --*beg;
+       } else *beg = 0, *end = 1<<29;
        free(s);
-       if (*begin > *end) {
-               fprintf(stderr, "[bam_parse_region] invalid region.\n");
-               return -1;
-       }
-       return 0;
+       return *beg <= *end? 0 : -1;
 }
 
 int32_t bam_aux2i(const uint8_t *s)
@@ -180,3 +204,10 @@ char *bam_aux2Z(const uint8_t *s)
        if (type == 'Z' || type == 'H') return (char*)s;
        else return 0;
 }
+
+#ifdef _WIN32
+double drand48()
+{
+       return (double)rand() / RAND_MAX;
+}
+#endif