prob_realn() seems working!

[samtools.git] / bcftools / bcf.c

diff --git a/bcftools/bcf.c b/bcftools/bcf.c
index c3313881ef90fc2930c72d218c090b35e5cf9466..feeda578304473a4022cb87674666790939a72dd 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;
@@ -100,21 +94,27 @@ int bcf_hdr_sync(bcf_hdr_t *b)
        return 0;
 }
 
-int bcf_sync(int n_smpl, bcf1_t *b)
+int bcf_sync(bcf1_t *b)
 {
        char *p, *tmp[5];
-       int i, n;
+       int i, n, n_smpl = b->n_smpl;
        ks_tokaux_t aux;
        // set ref, alt, flt, info, fmt
        b->ref = b->alt = b->flt = b->info = b->fmt = 0;
        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) return -1;
+       if (n != 5) {
+               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];
        // set n_alleles
-       for (p = b->alt, n = 1; *p; ++p)
-               if (*p == ',') ++n;
-       b->n_alleles = n + 1;
+       if (*b->alt == 0) b->n_alleles = 1;
+       else {
+               for (p = b->alt, n = 1; *p; ++p)
+                       if (*p == ',') ++n;
+               b->n_alleles = n + 1;
+       }
        // set n_gi and gi[i].fmt
        for (p = b->fmt, n = 1; *p; ++p)
                if (*p == ':') ++n;
@@ -137,7 +137,7 @@ int bcf_sync(int n_smpl, bcf1_t *b)
                } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("GT", 2)) {
                        b->gi[i].len = 1;
                } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
-                       b->gi[i].len = 4;
+                       b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2 * 4;
                }
                b->gi[i].data = realloc(b->gi[i].data, n_smpl * b->gi[i].len);
        }
@@ -146,13 +146,12 @@ int bcf_sync(int n_smpl, bcf1_t *b)
 
 int bcf_write(bcf_t *bp, const bcf_hdr_t *h, const bcf1_t *b)
 {
-       uint32_t x;
        int i, l = 0;
        if (b == 0) return -1;
        bgzf_write(bp->fp, &b->tid, 4);
        bgzf_write(bp->fp, &b->pos, 4);
-       x = b->qual<<24 | b->l_str;
-       bgzf_write(bp->fp, &x, 4);
+       bgzf_write(bp->fp, &b->qual, 4);
+       bgzf_write(bp->fp, &b->l_str, 4);
        bgzf_write(bp->fp, b->str, b->l_str);
        l = 12 + b->l_str;
        for (i = 0; i < b->n_gi; ++i) {
@@ -165,12 +164,12 @@ int bcf_write(bcf_t *bp, const bcf_hdr_t *h, const bcf1_t *b)
 int bcf_read(bcf_t *bp, const bcf_hdr_t *h, bcf1_t *b)
 {
        int i, l = 0;
-       uint32_t x;
        if (b == 0) return -1;
        if (bgzf_read(bp->fp, &b->tid, 4) == 0) return -1;
+       b->n_smpl = h->n_smpl;
        bgzf_read(bp->fp, &b->pos, 4);
-       bgzf_read(bp->fp, &x, 4);
-       b->qual = x >> 24; b->l_str = x << 8 >> 8;
+       bgzf_read(bp->fp, &b->qual, 4);
+       bgzf_read(bp->fp, &b->l_str, 4);
        if (b->l_str > b->m_str) {
                b->m_str = b->l_str;
                kroundup32(b->m_str);
@@ -178,7 +177,7 @@ 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(h->n_smpl, b);
+       bcf_sync(b);
        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;
@@ -215,7 +214,7 @@ void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s)
        fmt_str(b->str, s); kputc('\t', s);
        fmt_str(b->ref, s); kputc('\t', s);
        fmt_str(b->alt, s); kputc('\t', s);
-       kputw(b->qual, s); kputc('\t', s);
+       ksprintf(s, "%.3g", b->qual); kputc('\t', s);
        fmt_str(b->flt, s); kputc('\t', s);
        fmt_str(b->info, s);
        if (b->fmt[0]) {
@@ -241,9 +240,21 @@ void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s)
                                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];
-                               kputc('0' + (y>>3&7), s);
-                               kputc("/|"[y>>6&1], s);
-                               kputc('0' + (y&7), s);
+                               if (y>>7&1) {
+                                       kputsn("./.", 3, s);
+                               } else {
+                                       kputc('0' + (y>>3&7), s);
+                                       kputc("/|"[y>>6&1], s);
+                                       kputc('0' + (y&7), s);
+                               }
+                       } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
+                               float *d = (float*)b->gi[i].data + j * x;
+                               int k;
+                               //printf("- %lx\n", d);
+                               for (k = 0; k < x; ++k) {
+                                       if (k > 0) kputc(',', s);
+                                       ksprintf(s, "%.2f", d[k]);
+                               }
                        }
                }
        }