X-Git-Url: https://git.donarmstrong.com/?p=samtools.git;a=blobdiff_plain;f=bcftools%2Fvcf.c;h=e8526a38706357c4aa2d067fcf9c64dadd8978f6;hp=50a73566a6d6e45dca212e909ca36d10580dcd0e;hb=2fa4cde9b8806faceffb8677440b54ca4c90b6f7;hpb=86329087668959dbb63af0e0ca97597c23c9678e

diff --git a/bcftools/vcf.c b/bcftools/vcf.c
index 50a7356..e8526a3 100644
--- a/bcftools/vcf.c
+++ b/bcftools/vcf.c
@@ -30,7 +30,12 @@ bcf_hdr_t *vcf_hdr_read(bcf_t *bp)
 	memset(&smpl, 0, sizeof(kstring_t));
 	while (ks_getuntil(v->ks, '\n', &v->line, &dret) >= 0) {
 		if (v->line.l < 2) continue;
-		if (v->line.s[0] != '#') return 0; // no sample line
+		if (v->line.s[0] != '#') {
+            free(meta.s);
+            free(smpl.s);
+            free(h);
+            return 0; // no sample line
+        }
 		if (v->line.s[0] == '#' && v->line.s[1] == '#') {
 			kputsn(v->line.s, v->line.l, &meta); kputc('\n', &meta);
 		} else if (v->line.s[0] == '#') {
@@ -72,6 +77,33 @@ bcf_t *vcf_open(const char *fn, const char *mode)
 	return bp;
 }
 
+int vcf_dictread(bcf_t *bp, bcf_hdr_t *h, const char *fn)
+{
+	vcf_t *v;
+	gzFile fp;
+	kstream_t *ks;
+	kstring_t s, rn;
+	int dret;
+	if (bp == 0) return -1;
+	if (!bp->is_vcf) return 0;
+	s.l = s.m = 0; s.s = 0;
+	rn.m = rn.l = h->l_nm; rn.s = h->name;
+	v = (vcf_t*)bp->v;
+	fp = gzopen(fn, "r");
+	ks = ks_init(fp);
+	while (ks_getuntil(ks, 0, &s, &dret) >= 0) {
+		bcf_str2id_add(v->refhash, strdup(s.s));
+		kputs(s.s, &rn); kputc('\0', &rn);
+		if (dret != '\n') ks_getuntil(ks, '\n', &s, &dret);
+	}
+	ks_destroy(ks);
+	gzclose(fp);
+	h->l_nm = rn.l; h->name = rn.s;
+	bcf_hdr_sync(h);
+	free(s.s);
+	return 0;
+}
+
 int vcf_close(bcf_t *bp)
 {
 	vcf_t *v;
@@ -84,7 +116,7 @@ int vcf_close(bcf_t *bp)
 	}
 	if (v->fpout) fclose(v->fpout);
 	free(v->line.s);
-	bcf_str2id_destroy(v->refhash);
+	bcf_str2id_thorough_destroy(v->refhash);
 	free(v);
 	free(bp);
 	return 0;
@@ -93,9 +125,14 @@ int vcf_close(bcf_t *bp)
 int vcf_hdr_write(bcf_t *bp, const bcf_hdr_t *h)
 {
 	vcf_t *v = (vcf_t*)bp->v;
-	int i;
+	int i, has_ver = 0;
 	if (!bp->is_vcf) return bcf_hdr_write(bp, h);
-	if (h->l_txt > 0) fwrite(h->txt, 1, h->l_txt - 1, v->fpout);
+	if (h->l_txt > 0) {
+		if (strstr(h->txt, "##fileformat=")) has_ver = 1;
+		if (has_ver == 0) fprintf(v->fpout, "##fileformat=VCFv4.1\n");
+		fwrite(h->txt, 1, h->l_txt - 1, v->fpout);
+	}
+	if (h->l_txt == 0) fprintf(v->fpout, "##fileformat=VCFv4.1\n");
 	fprintf(v->fpout, "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT");
 	for (i = 0; i < h->n_smpl; ++i)
 		fprintf(v->fpout, "\t%s", h->sns[i]);
@@ -126,38 +163,84 @@ int vcf_read(bcf_t *bp, bcf_hdr_t *h, bcf1_t *b)
 	str.l = 0; str.m = b->m_str; str.s = b->str;
 	rn.l = rn.m = h->l_nm; rn.s = h->name;
 	if (ks_getuntil(v->ks, '\n', &v->line, &dret) < 0) return -1;
+	b->n_smpl = h->n_smpl;
 	for (p = kstrtok(v->line.s, "\t", &aux), k = 0; p; p = kstrtok(0, 0, &aux), ++k) {
 		*(char*)aux.p = 0;
 		if (k == 0) { // ref
 			int tid = bcf_str2id(v->refhash, p);
 			if (tid < 0) {
-				tid = bcf_str2id_add(v->refhash, p);
-				kputs(p, &rn);
+				tid = bcf_str2id_add(v->refhash, strdup(p));
+				kputs(p, &rn); kputc('\0', &rn);
 				sync = 1;
 			}
 			b->tid = tid;
 		} else if (k == 1) { // pos
-			b->pos = atoi(p);
+			b->pos = atoi(p) - 1;
 		} else if (k == 5) { // qual
-			b->qual = (p[0] >= '0' && p[0] <= '9')? atoi(p) : 0;
+			b->qual = (p[0] >= '0' && p[0] <= '9')? atof(p) : 0;
 		} else if (k <= 8) { // variable length strings
 			kputs(p, &str); kputc('\0', &str);
 			b->l_str = str.l; b->m_str = str.m; b->str = str.s;
-			if (k == 8) bcf_sync(h->n_smpl, b);
-		} else {
+			if (k == 8) bcf_sync(b);
+		} else { // k > 9
+			if (strncmp(p, "./.", 3) == 0) {
+				for (i = 0; i < b->n_gi; ++i) {
+					if (b->gi[i].fmt == bcf_str2int("GT", 2)) {
+						((uint8_t*)b->gi[i].data)[k-9] = 1<<7;
+					} else if (b->gi[i].fmt == bcf_str2int("GQ", 2)) {
+						((uint8_t*)b->gi[i].data)[k-9] = 0;
+					} else if (b->gi[i].fmt == bcf_str2int("SP", 2)) {
+						((int32_t*)b->gi[i].data)[k-9] = 0;
+					} else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("DV", 2)) {
+						((uint16_t*)b->gi[i].data)[k-9] = 0;
+					} else if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
+						int y = b->n_alleles * (b->n_alleles + 1) / 2;
+						memset((uint8_t*)b->gi[i].data + (k - 9) * y, 0, y);
+					} else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
+						int y = b->n_alleles * (b->n_alleles + 1) / 2;
+						memset((float*)b->gi[i].data + (k - 9) * y, 0, y * 4);
+					}
+				}
+				goto endblock;
+			}
 			for (q = kstrtok(p, ":", &a2), i = 0; q && i < b->n_gi; q = kstrtok(0, 0, &a2), ++i) {
 				if (b->gi[i].fmt == bcf_str2int("GT", 2)) {
 					((uint8_t*)b->gi[i].data)[k-9] = (q[0] - '0')<<3 | (q[2] - '0') | (q[1] == '/'? 0 : 1) << 6;
 				} else if (b->gi[i].fmt == bcf_str2int("GQ", 2)) {
-					int x = strtol(q, &q, 10);
+					double _x = strtod(q, &q);
+					int x = (int)(_x + .499);
 					if (x > 255) x = 255;
 					((uint8_t*)b->gi[i].data)[k-9] = x;
-				} else if (b->gi[i].fmt == bcf_str2int("DP", 2)) {
+				} else if (b->gi[i].fmt == bcf_str2int("SP", 2)) {
+					int x = strtol(q, &q, 10);
+					if (x > 0xffff) x = 0xffff;
+					((uint32_t*)b->gi[i].data)[k-9] = x;
+				} else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("DV", 2)) {
 					int x = strtol(q, &q, 10);
 					if (x > 0xffff) x = 0xffff;
 					((uint16_t*)b->gi[i].data)[k-9] = x;
+				} else if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
+					int x, y, j;
+					uint8_t *data = (uint8_t*)b->gi[i].data;
+					y = b->n_alleles * (b->n_alleles + 1) / 2;
+					for (j = 0; j < y; ++j) {
+						x = strtol(q, &q, 10);
+						if (x > 255) x = 255;
+						data[(k-9) * y + j] = x;
+						++q;
+					}
+				} else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
+					int j, y;
+					float x, *data = (float*)b->gi[i].data;
+					y = b->n_alleles * (b->n_alleles + 1) / 2;
+					for (j = 0; j < y; ++j) {
+						x = strtod(q, &q);
+						data[(k-9) * y + j] = x > 0? -x/10. : x;
+						++q;
+					}
 				}
 			}
+		endblock: i = i;
 		}
 	}
 	h->l_nm = rn.l; h->name = rn.s;