* samtools-0.1.9-1 (r786)

[samtools.git] / bcftools / bcf.c

diff --git a/bcftools/bcf.c b/bcftools/bcf.c
index 2fd1696eff466f4ad9490c041cdd0f6e09448e68..494dcccd3641b3c1b0856a8e88c0d065399ca8a7 100644 (file)
--- a/bcftools/bcf.c
+++ b/bcftools/bcf.c
@@ -4,12 +4,6 @@
 #include "kstring.h"
 #include "bcf.h"
 
-void bcf_hdr_clear(bcf_hdr_t *b)
-{
-       free(b->name); free(b->sname); free(b->txt); free(b->ns); free(b->sns);
-       memset(b, 0, sizeof(bcf_hdr_t));
-}
-
 bcf_t *bcf_open(const char *fn, const char *mode)
 {
        bcf_t *b;
@@ -110,7 +104,7 @@ int bcf_sync(bcf1_t *b)
        for (p = b->str, n = 0; p < b->str + b->l_str; ++p)
                if (*p == 0 && p+1 != b->str + b->l_str) tmp[n++] = p + 1;
        if (n != 5) {
-               fprintf(stderr, "[bcf_sync] incorrect number of fields (%d != 5)\n", n);
+               fprintf(stderr, "[%s] incorrect number of fields (%d != 5). Corrupted file?\n", __func__, n);
                return -1;
        }
        b->ref = tmp[0]; b->alt = tmp[1]; b->flt = tmp[2]; b->info = tmp[3]; b->fmt = tmp[4];
@@ -140,7 +134,9 @@ int bcf_sync(bcf1_t *b)
                        b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2;
                } else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("HQ", 2)) {
                        b->gi[i].len = 2;
-               } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("GT", 2)) {
+               } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("GT", 2)
+                                  || b->gi[i].fmt == bcf_str2int("SP", 2))
+               {
                        b->gi[i].len = 1;
                } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
                        b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2 * 4;
@@ -196,7 +192,7 @@ int bcf_destroy(bcf1_t *b)
        int i;
        if (b == 0) return -1;
        free(b->str);
-       for (i = 0; i < b->n_gi; ++i)
+       for (i = 0; i < b->m_gi; ++i)
                free(b->gi[i].data);
        free(b->gi);
        free(b);
@@ -242,7 +238,7 @@ void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s)
                                }
                        } else if (b->gi[i].fmt == bcf_str2int("DP", 2)) {
                                kputw(((uint16_t*)b->gi[i].data)[j], s);
-                       } else if (b->gi[i].fmt == bcf_str2int("GQ", 2)) {
+                       } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("SP", 2)) {
                                kputw(((uint8_t*)b->gi[i].data)[j], s);
                        } else if (b->gi[i].fmt == bcf_str2int("GT", 2)) {
                                int y = ((uint8_t*)b->gi[i].data)[j];
@@ -273,3 +269,40 @@ char *bcf_fmt(const bcf_hdr_t *h, bcf1_t *b)
        bcf_fmt_core(h, b, &s);
        return s.s;
 }
+
+int bcf_append_info(bcf1_t *b, const char *info, int l)
+{
+       int shift = b->fmt - b->str;
+       int l_fmt = b->l_str - shift;
+       char *ori = b->str;
+       if (b->l_str + l > b->m_str) { // enlarge if necessary
+               b->m_str = b->l_str + l;
+               kroundup32(b->m_str);
+               b->str = realloc(b->str, b->m_str);
+       }
+       memmove(b->str + shift + l, b->str + shift, l_fmt); // move the FORMAT field
+       memcpy(b->str + shift - 1, info, l); // append to the INFO field
+       b->str[shift + l - 1] = '\0';
+       b->fmt = b->str + shift + l;
+       b->l_str += l;
+       if (ori != b->str) bcf_sync(b); // synchronize when realloc changes the pointer
+       return 0;
+}
+
+int bcf_cpy(bcf1_t *r, const bcf1_t *b)
+{
+       char *t1 = r->str;
+       bcf_ginfo_t *t2 = r->gi;
+       int i, t3 = r->m_str, t4 = r->m_gi;
+       *r = *b;
+       r->str = t1; r->gi = t2; r->m_str = t3; r->m_gi = t4;
+       if (r->m_str < b->m_str) {
+               r->m_str = b->m_str;
+               r->str = realloc(r->str, r->m_str);
+       }
+       memcpy(r->str, b->str, r->m_str);
+       bcf_sync(r); // calling bcf_sync() is simple but inefficient
+       for (i = 0; i < r->n_gi; ++i)
+               memcpy(r->gi[i].data, b->gi[i].data, r->n_smpl * r->gi[i].len);
+       return 0;
+}