7 void bcf_hdr_clear(bcf_hdr_t *b)
9 free(b->name); free(b->sname); free(b->txt); free(b->ns); free(b->sns);
10 memset(b, 0, sizeof(bcf_hdr_t));
13 bcf_t *bcf_open(const char *fn, const char *mode)
16 b = calloc(1, sizeof(bcf_t));
17 if (strchr(mode, 'w')) {
18 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), mode);
20 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode);
22 b->fp->owned_file = 1;
26 int bcf_close(bcf_t *b)
29 return bgzf_close(b->fp);
32 int bcf_hdr_write(bcf_t *b, const bcf_hdr_t *h)
34 if (b == 0 || h == 0) return -1;
35 bgzf_write(b->fp, "BCF\4", 4);
36 bgzf_write(b->fp, &h->l_nm, 4);
37 bgzf_write(b->fp, h->name, h->l_nm);
38 bgzf_write(b->fp, &h->l_smpl, 4);
39 bgzf_write(b->fp, h->sname, h->l_smpl);
40 bgzf_write(b->fp, &h->l_txt, 4);
41 bgzf_write(b->fp, h->txt, h->l_txt);
43 return 16 + h->l_nm + h->l_smpl + h->l_txt;
46 bcf_hdr_t *bcf_hdr_read(bcf_t *b)
51 h = calloc(1, sizeof(bcf_hdr_t));
52 bgzf_read(b->fp, magic, 4);
53 bgzf_read(b->fp, &h->l_nm, 4);
54 h->name = malloc(h->l_nm);
55 bgzf_read(b->fp, h->name, h->l_nm);
56 bgzf_read(b->fp, &h->l_smpl, 4);
57 h->sname = malloc(h->l_smpl);
58 bgzf_read(b->fp, h->sname, h->l_smpl);
59 bgzf_read(b->fp, &h->l_txt, 4);
60 h->txt = malloc(h->l_txt);
61 bgzf_read(b->fp, h->txt, h->l_txt);
66 void bcf_hdr_destroy(bcf_hdr_t *h)
69 free(h->name); free(h->sname); free(h->txt); free(h->ns); free(h->sns);
73 static inline char **cnt_null(int l, char *str, int *_n)
78 if (l == 0 || str == 0) return 0;
79 for (p = str; p != str + l; ++p)
82 list = calloc(n, sizeof(void*));
84 for (p = str, n = 1; p < str + l - 1; ++p)
85 if (*p == 0) list[n++] = p + 1;
89 int bcf_hdr_sync(bcf_hdr_t *b)
91 if (b == 0) return -1;
92 if (b->ns) free(b->ns);
93 if (b->sns) free(b->sns);
94 if (b->l_nm) b->ns = cnt_null(b->l_nm, b->name, &b->n_ref);
95 else b->ns = 0, b->n_ref = 0;
96 b->sns = cnt_null(b->l_smpl, b->sname, &b->n_smpl);
100 #define char2int(s) (((int)s[0])<<8|s[1])
102 int bcf_sync(int n_smpl, bcf1_t *b)
104 char *p, *tmp[5], *s;
106 // set ref, alt, flt, info, fmt
107 b->ref = b->alt = b->flt = b->info = b->fmt = 0;
108 for (p = b->str, n = 0; p < b->str + b->l_str; ++p)
109 if (*p == 0 && p+1 != b->str + b->l_str) tmp[n++] = p + 1;
110 if (n != 5) return -1;
111 b->ref = tmp[0]; b->alt = tmp[1]; b->flt = tmp[2]; b->info = tmp[3]; b->fmt = tmp[4];
113 for (p = b->alt, n = 1; *p; ++p)
115 b->n_alleles = n + 1;
116 // set n_gi and gi[i].fmt
117 for (p = b->fmt, n = 1; *p; ++p)
123 b->gi = realloc(b->gi, b->m_gi * sizeof(bcf_ginfo_t));
124 memset(b->gi + old_m, 0, (b->m_gi - old_m) * sizeof(bcf_ginfo_t));
127 for (p = s = b->fmt, n = 0; *p; ++p) {
128 if (*p == ':' || *(p+1) == 0) {
129 char *q = *p == ':'? p : p + 1;
130 if ((q - s) != 2) return -2;
131 b->gi[n].fmt = char2int(s);
136 for (i = 0; i < b->n_gi; ++i) {
137 if (b->gi[i].fmt == char2int("PL")) {
138 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2;
139 } else if (b->gi[i].fmt == char2int("DP") || b->gi[i].fmt == char2int("HQ")) {
141 } else if (b->gi[i].fmt == char2int("GQ") || b->gi[i].fmt == char2int("GT")) {
143 } else if (b->gi[i].fmt == char2int("GL")) {
146 b->gi[i].data = realloc(b->gi[i].data, n_smpl * b->gi[i].len);
151 int bcf_write(bcf_t *bp, const bcf_hdr_t *h, const bcf1_t *b)
155 if (b == 0) return -1;
156 bgzf_write(bp->fp, &b->tid, 4);
157 bgzf_write(bp->fp, &b->pos, 4);
158 x = b->qual<<24 | b->l_str;
159 bgzf_write(bp->fp, &x, 4);
160 bgzf_write(bp->fp, b->str, b->l_str);
162 for (i = 0; i < b->n_gi; ++i) {
163 bgzf_write(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
164 l += b->gi[i].len * h->n_smpl;
169 int bcf_read(bcf_t *bp, const bcf_hdr_t *h, bcf1_t *b)
173 if (b == 0) return -1;
174 if (bgzf_read(bp->fp, &b->tid, 4) == 0) return -1;
175 bgzf_read(bp->fp, &b->pos, 4);
176 bgzf_read(bp->fp, &x, 4);
177 b->qual = x >> 24; b->l_str = x << 8 >> 8;
178 if (b->l_str > b->m_str) {
180 kroundup32(b->m_str);
181 b->str = realloc(b->str, b->m_str);
183 bgzf_read(bp->fp, b->str, b->l_str);
185 bcf_sync(h->n_smpl, b);
186 for (i = 0; i < b->n_gi; ++i) {
187 bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
188 l += b->gi[i].len * h->n_smpl;
193 int bcf_destroy(bcf1_t *b)
196 if (b == 0) return -1;
198 for (i = 0; i < b->n_gi; ++i)
205 static inline void fmt_str(const char *p, kstring_t *s)
207 if (*p == 0) kputc('.', s);
211 char *bcf_fmt(const bcf_hdr_t *h, bcf1_t *b)
215 memset(&s, 0, sizeof(kstring_t));
216 kputs(h->ns[b->tid], &s); kputc('\t', &s);
217 kputw(b->pos + 1, &s); kputc('\t', &s);
218 fmt_str(b->str, &s); kputc('\t', &s);
219 fmt_str(b->ref, &s); kputc('\t', &s);
220 fmt_str(b->alt, &s); kputc('\t', &s);
221 kputw(b->qual, &s); kputc('\t', &s);
222 fmt_str(b->flt, &s); kputc('\t', &s);
223 fmt_str(b->info, &s);
228 x = b->n_alleles * (b->n_alleles + 1) / 2;
229 for (j = 0; j < h->n_smpl; ++j) {
231 for (i = 0; i < b->n_gi; ++i) {
232 if (i) kputc(':', &s);
233 if (b->gi[i].fmt == char2int("PL")) {
234 uint8_t *d = (uint8_t*)b->gi[i].data + j * x;
236 for (k = 0; k < x; ++k) {
237 if (k > 0) kputc(',', &s);
240 } else if (b->gi[i].fmt == char2int("DP")) {
241 kputw(((uint16_t*)b->gi[i].data)[j], &s);
242 } else if (b->gi[i].fmt == char2int("GQ")) {
243 kputw(((uint8_t*)b->gi[i].data)[j], &s);