X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bcftools%2Fbcf.c;h=a538e6e14d7a0a78af7927b92073d05da0844640;hb=dc27682f70713a70d4f31bca652cf78e00757da2;hp=e40631975af4351436efdc938c01dca054567c85;hpb=c0490a669e9f1f9257e4b02a263e92e842bf5357;p=samtools.git

diff --git a/bcftools/bcf.c b/bcftools/bcf.c
index e406319..a538e6e 100644
--- a/bcftools/bcf.c
+++ b/bcftools/bcf.c
@@ -4,14 +4,6 @@
 #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);
-	memset(b, 0, sizeof(bcf_hdr_t));
-}
-
 bcf_t *bcf_open(const char *fn, const char *mode)
 {
 	bcf_t *b;
@@ -21,8 +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;
-	if (strchr(mode, 'r')) bcf_hdr_read(b);
+#endif
 	return b;
 }
 
@@ -31,16 +24,13 @@ int bcf_close(bcf_t *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);
@@ -52,26 +42,12 @@ int bcf_hdr_write(bcf_t *b)
 	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);
@@ -82,8 +58,15 @@ int bcf_hdr_read(bcf_t *b)
 	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)
@@ -105,27 +88,41 @@ 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)
+int bcf_sync(bcf1_t *b)
 {
-	char *p, *tmp[5], *s;
-	int i, n;
+	char *p, *tmp[5];
+	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;
+	for (p = b->str, n = 0; p < b->str + b->l_str; ++p) {
+		if (*p == 0 && p+1 != b->str + b->l_str) {
+			if (n == 5) {
+				++n;
+				break;
+			} else tmp[n++] = p + 1;
+		}
+	}
+	if (n != 5) {
+		fprintf(stderr, "[%s] incorrect number of fields (%d != 5) at %d:%d\n", __func__, n, b->tid, b->pos);
+		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;
@@ -137,57 +134,52 @@ int bcf_sync(int n_smpl, bcf1_t *b)
 		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].fmt == bcf_str2int("DV", 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("SP", 2)) {
 			b->gi[i].len = 4;
+		} else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
+			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);
 	}
 	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;
 	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) {
-		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;
+	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);
@@ -195,10 +187,10 @@ int bcf_read(bcf_t *bp, bcf1_t *b)
 	}
 	bgzf_read(bp->fp, b->str, b->l_str);
 	l = 12 + b->l_str;
-	bcf_sync(bp->h.n_smpl, 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 * 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;
 }
@@ -208,7 +200,7 @@ 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);
@@ -221,41 +213,156 @@ static inline void fmt_str(const char *p, kstring_t *s)
 	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);
+	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]) {
-		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;
+    int iPL = -1;
+    if ( b->n_alleles > 2 ) {
+        for (i=0; i<b->n_gi; i++) {
+            if ( b->gi[i].fmt == bcf_str2int("PL", 2) ) {
+                iPL = i;
+                break;
+            }
+        }
+    }
+	for (j = 0; j < h->n_smpl; ++j) {
+
+        // Determine GT with maximum PL (multiple ALT sites only)
+        int imax=-1;
+        if ( iPL!=-1 ) {
+            uint8_t *d = (uint8_t*)b->gi[iPL].data + j * x;
+            int k,identical=1;
+            imax=0;
+            for (k=1; k<x; k++)
+            {
+                if ( identical && d[k]!=d[k-1] ) identical = 0;
+                if ( d[k]<d[imax] ) imax = k;
+            }
+            // If all lks are identical, leave GT untouched
+            if ( identical ) imax = -1;
+        }
+		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) || b->gi[i].fmt == bcf_str2int("DV", 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("SP", 2)) {
+				kputw(((int32_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];
+                if ( y>>7&1 )
+                    kputsn("./.", 3, s);
+                else if ( imax==-1 )
+                {
+                    kputc('0' + (y>>3&7), s);
+                    kputc("/|"[y>>6&1], s);
+                    kputc('0' + (y&7), s);
+                }
+                else
+                {
+                    // Arguably, the while loop will be faster than two sqrts
+                    int n = 0;
+                    int row = 1;
+                    while ( n<imax )
+                    {
+                        row++;
+                        n += row;
+                    }
+                    row--;
+                    kputw(imax-n+row, s);
+                    kputc("/|"[y>>6&1], s);
+                    kputw(row, 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]);
+				}
+			} else kputc('.', s); // custom fields
 		}
 	}
+}
+
+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;
 }
+
+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;
+}