#include "kstring.h"
#include "bcf.h"
-int bcf_hdr_read(bcf_t *b);
-
void bcf_hdr_clear(bcf_hdr_t *b)
{
free(b->name); free(b->sname); free(b->txt); free(b->ns); free(b->sns);
bcf_t *b;
b = calloc(1, sizeof(bcf_t));
if (strchr(mode, 'w')) {
- b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), "w");
+ b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), mode);
} else {
- b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), "r");
+ b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode);
}
b->fp->owned_file = 1;
- if (strchr(mode, 'r')) bcf_hdr_read(b);
return b;
}
int ret;
if (b == 0) return 0;
ret = bgzf_close(b->fp);
- free(b->h.name); free(b->h.sname); free(b->h.txt); free(b->h.ns); free(b->h.sns);
free(b);
return ret;
}
-int bcf_hdr_write(bcf_t *b)
+int bcf_hdr_write(bcf_t *b, const bcf_hdr_t *h)
{
- bcf_hdr_t *h;
- if (b == 0) return -1;
- h = &b->h;
+ if (b == 0 || h == 0) return -1;
bgzf_write(b->fp, "BCF\4", 4);
bgzf_write(b->fp, &h->l_nm, 4);
bgzf_write(b->fp, h->name, h->l_nm);
return 16 + h->l_nm + h->l_smpl + h->l_txt;
}
-int bcf_hdr_cpy(bcf_hdr_t *h, const bcf_hdr_t *h0)
-{
- *h = *h0;
- h->name = malloc(h->l_nm);
- h->sname = malloc(h->l_smpl);
- h->txt = malloc(h->l_txt);
- memcpy(h->name, h0->name, h->l_nm);
- memcpy(h->sname, h0->sname, h->l_smpl);
- memcpy(h->txt, h0->txt, h->l_txt);
- bcf_hdr_sync(h);
- return 0;
-}
-
-int bcf_hdr_read(bcf_t *b)
+bcf_hdr_t *bcf_hdr_read(bcf_t *b)
{
uint8_t magic[4];
bcf_hdr_t *h;
- if (b == 0) return -1;
- bcf_hdr_clear(&b->h);
- h = &b->h;
+ if (b == 0) return 0;
+ h = calloc(1, sizeof(bcf_hdr_t));
bgzf_read(b->fp, magic, 4);
bgzf_read(b->fp, &h->l_nm, 4);
h->name = malloc(h->l_nm);
bgzf_read(b->fp, &h->l_txt, 4);
h->txt = malloc(h->l_txt);
bgzf_read(b->fp, h->txt, h->l_txt);
- bcf_hdr_sync(&b->h);
- return 16 + h->l_nm + h->l_smpl + h->l_txt;
+ bcf_hdr_sync(h);
+ return h;
+}
+
+void bcf_hdr_destroy(bcf_hdr_t *h)
+{
+ if (h == 0) return;
+ free(h->name); free(h->sname); free(h->txt); free(h->ns); free(h->sns);
+ free(h);
}
static inline char **cnt_null(int l, char *str, int *_n)
int bcf_hdr_sync(bcf_hdr_t *b)
{
if (b == 0) return -1;
- b->ns = cnt_null(b->l_nm, b->name, &b->n_ref);
+ if (b->ns) free(b->ns);
+ if (b->sns) free(b->sns);
+ if (b->l_nm) b->ns = cnt_null(b->l_nm, b->name, &b->n_ref);
+ else b->ns = 0, b->n_ref = 0;
b->sns = cnt_null(b->l_smpl, b->sname, &b->n_smpl);
return 0;
}
-#define char2int(s) (((int)s[0])<<8|s[1])
-
int bcf_sync(int n_smpl, bcf1_t *b)
{
- char *p, *tmp[5], *s;
+ char *p, *tmp[5];
int i, n;
+ 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)
memset(b->gi + old_m, 0, (b->m_gi - old_m) * sizeof(bcf_ginfo_t));
}
b->n_gi = n;
- for (p = s = b->fmt, n = 0; *p; ++p) {
- if (*p == ':' || *(p+1) == 0) {
- char *q = *p == ':'? p : p + 1;
- if ((q - s) != 2) return -2;
- b->gi[n].fmt = char2int(s);
- s = q;
- }
- }
+ for (p = kstrtok(b->fmt, ":", &aux), n = 0; p; p = kstrtok(0, 0, &aux))
+ b->gi[n++].fmt = bcf_str2int(p, aux.p - p);
// set gi[i].len
for (i = 0; i < b->n_gi; ++i) {
- if (b->gi[i].fmt == char2int("PL")) {
+ if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2;
- } else if (b->gi[i].fmt == char2int("DP") || b->gi[i].fmt == char2int("HQ")) {
+ } 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 == char2int("GQ") || b->gi[i].fmt == char2int("GT")) {
+ } 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 == char2int("GL")) {
+ } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
b->gi[i].len = 4;
}
b->gi[i].data = realloc(b->gi[i].data, n_smpl * b->gi[i].len);
return 0;
}
-int bcf_write(bcf_t *bp, const 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;
bgzf_write(bp->fp, b->str, b->l_str);
l = 12 + b->l_str;
for (i = 0; i < b->n_gi; ++i) {
- bgzf_write(bp->fp, b->gi[i].data, b->gi[i].len * bp->h.n_smpl);
- l += b->gi[i].len * bp->h.n_smpl;
+ bgzf_write(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
+ l += b->gi[i].len * h->n_smpl;
}
return l;
}
-int bcf_read(bcf_t *bp, 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 0;
+ if (bgzf_read(bp->fp, &b->tid, 4) == 0) return -1;
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->str, b->l_str);
l = 12 + b->l_str;
- bcf_sync(bp->h.n_smpl, b);
+ bcf_sync(h->n_smpl, b);
for (i = 0; i < b->n_gi; ++i) {
- bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * bp->h.n_smpl);
- l += b->gi[i].len * bp->h.n_smpl;
+ bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
+ l += b->gi[i].len * h->n_smpl;
}
return l;
}
else kputs(p, s);
}
-char *bcf_fmt(bcf_t *bp, bcf1_t *b)
+void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s)
{
- kstring_t s;
int i, j, x;
- memset(&s, 0, sizeof(kstring_t));
- kputs(bp->h.ns[b->tid], &s); kputc('\t', &s);
- kputw(b->pos + 1, &s); kputc('\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);
- fmt_str(b->flt, &s); kputc('\t', &s);
- fmt_str(b->info, &s);
+ s->l = 0;
+ if (h->n_ref) kputs(h->ns[b->tid], s);
+ else kputw(b->tid, s);
+ kputc('\t', s);
+ kputw(b->pos + 1, s); kputc('\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);
+ fmt_str(b->flt, s); kputc('\t', s);
+ fmt_str(b->info, s);
if (b->fmt[0]) {
- kputc('\t', &s);
- fmt_str(b->fmt, &s);
+ kputc('\t', s);
+ fmt_str(b->fmt, s);
}
x = b->n_alleles * (b->n_alleles + 1) / 2;
- for (j = 0; j < bp->h.n_smpl; ++j) {
- kputc('\t', &s);
+ if (b->n_gi == 0) return;
+ for (j = 0; j < h->n_smpl; ++j) {
+ kputc('\t', s);
for (i = 0; i < b->n_gi; ++i) {
- if (i) kputc(':', &s);
- if (b->gi[i].fmt == char2int("PL")) {
+ if (i) kputc(':', s);
+ if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
uint8_t *d = (uint8_t*)b->gi[i].data + j * x;
int k;
for (k = 0; k < x; ++k) {
- if (k > 0) kputc(',', &s);
- kputw(d[k], &s);
+ if (k > 0) kputc(',', s);
+ kputw(d[k], s);
}
- } else if (b->gi[i].fmt == char2int("DP")) {
- kputw(((uint16_t*)b->gi[i].data)[j], &s);
- } else if (b->gi[i].fmt == char2int("GQ")) {
- kputw(((uint8_t*)b->gi[i].data)[j], &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)) {
+ 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);
}
}
}
+}
+
+char *bcf_fmt(const bcf_hdr_t *h, bcf1_t *b)
+{
+ kstring_t s;
+ s.l = s.m = 0; s.s = 0;
+ bcf_fmt_core(h, b, &s);
return s.s;
}
projects / samtools.git / blobdiff