* samtools-0.1.11-1 (r858)

[samtools.git] / bcftools / bcf.c

diff --git a/bcftools/bcf.c b/bcftools/bcf.c
index feeda578304473a4022cb87674666790939a72dd..6e45695cfb8c35a22dc4699728e445129293a699 100644 (file)
--- a/bcftools/bcf.c
+++ b/bcftools/bcf.c
@@ -13,7 +13,9 @@ bcf_t *bcf_open(const char *fn, const char *mode)
        } else {
                b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode);
        }
+#ifndef BCF_LITE
        b->fp->owned_file = 1;
+#endif
        return b;
 }
 
@@ -134,7 +136,9 @@ 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;
@@ -177,7 +181,7 @@ int bcf_read(bcf_t *bp, const bcf_hdr_t *h, bcf1_t *b)
        }
        bgzf_read(bp->fp, b->str, b->l_str);
        l = 12 + b->l_str;
-       bcf_sync(b);
+       if (bcf_sync(b) < 0) return -2;
        for (i = 0; i < b->n_gi; ++i) {
                bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
                l += b->gi[i].len * h->n_smpl;
@@ -190,7 +194,7 @@ 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);
@@ -236,7 +240,7 @@ 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];
@@ -267,3 +271,50 @@ 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;
+}
+
+int bcf_is_indel(const bcf1_t *b)
+{
+       char *p;
+       if (strlen(b->ref) > 1) return 1;
+       for (p = b->alt; *p; ++p)
+               if (*p != ',' && p[1] != ',' && p[1] != '\0')
+                       return 1;
+       return 0;
+}