X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_import.c;h=746dc5bbbde0415911794b28d666a4a53ef86be8;hb=8cc87acf088966330e253d44e67791696d74f35b;hp=6b3b4bc69352a92d0c9e74379fe8ddd2e4564c43;hpb=f93dae0d03856955f9424e8b2aaf261304ca647e;p=samtools.git

diff --git a/bam_import.c b/bam_import.c
index 6b3b4bc..746dc5b 100644
--- a/bam_import.c
+++ b/bam_import.c
@@ -44,29 +44,27 @@ struct __tamFile_t {
 	uint64_t n_lines;
 };
 
-char **bam_load_pos(const char *fn, int *_n)
+char **__bam_get_lines(const char *fn, int *_n) // for bam_plcmd.c only
 {
 	char **list = 0, *s;
-	int n = 0, dret, m = 0, c;
+	int n = 0, dret, m = 0;
 	gzFile fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
 	kstream_t *ks;
 	kstring_t *str;
 	str = (kstring_t*)calloc(1, sizeof(kstring_t));
 	ks = ks_init(fp);
-	while (ks_getuntil(ks, 0, str, &dret) > 0) {
+	while (ks_getuntil(ks, '\n', str, &dret) > 0) {
 		if (n == m) {
 			m = m? m << 1 : 16;
 			list = (char**)realloc(list, m * sizeof(char*));
 		}
-		s = list[n++] = (char*)calloc(str->l + 5, 1);
+		if (str->s[str->l-1] == '\r')
+			str->s[--str->l] = '\0';
+		s = list[n++] = (char*)calloc(str->l + 1, 1);
 		strcpy(s, str->s);
-		s += str->l + 1;
-		ks_getuntil(ks, 0, str, &dret);
-		*((uint32_t*)s) = atoi(str->s);
-		if (dret != '\n')
-			while ((c = ks_getc(fp)) >= 0 && c != '\n');
 	}
 	ks_destroy(ks);
+	gzclose(fp);
 	free(str->s); free(str);
 	*_n = n;
 	return list;
@@ -104,7 +102,7 @@ bam_header_t *sam_header_read2(const char *fn)
 	assert(fp);
 	ks = ks_init(fp);
 	str = (kstring_t*)calloc(1, sizeof(kstring_t));
-	while (ks_getuntil(ks, 0, str, &dret) >= 0) {
+	while (ks_getuntil(ks, 0, str, &dret) > 0) {
 		char *s = strdup(str->s);
 		int len, i;
 		i = kh_size(hash);
@@ -153,7 +151,7 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 	kstring_t *str = fp->str;
 	kstream_t *ks = fp->ks;
 
-	while ((ret = ks_getuntil(fp->ks, 0, str, &dret)) >= 0 && str->s[0] == '@') { // skip header
+	while ((ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret)) >= 0 && str->s[0] == '@') { // skip header
 		str->s[str->l] = dret; // note that str->s is NOT null terminated!!
 		append_text(header, str);
 		if (dret != '\n') {
@@ -163,7 +161,7 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 		}
 		++fp->n_lines;
 	}
-	while (ret == 0) ret = ks_getuntil(fp->ks, 0, str, &dret); // special consideration for "\r\n"
+	while (ret == 0) ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret); // special consideration for "\r\n"
 	if (ret < 0) return -1;
 	++fp->n_lines;
 	doff = 0;
@@ -174,10 +172,10 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 		doff += c->l_qname;
 	}
 	{ // flag, tid, pos, qual
-		ret = ks_getuntil(ks, 0, str, &dret); c->flag = atoi(str->s);
-		ret = ks_getuntil(ks, 0, str, &dret); c->tid = bam_get_tid(header, str->s);
-		ret = ks_getuntil(ks, 0, str, &dret); c->pos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
-		ret = ks_getuntil(ks, 0, str, &dret); c->qual = isdigit(str->s[0])? atoi(str->s) : 0;
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); c->flag = atoi(str->s);
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); c->tid = bam_get_tid(header, str->s);
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); c->pos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); c->qual = isdigit(str->s[0])? atoi(str->s) : 0;
 		if (ret < 0) return -2;
 	}
 	{ // cigar
@@ -185,7 +183,7 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 		int i, op;
 		long x;
 		c->n_cigar = 0;
-		if (ks_getuntil(ks, 0, str, &dret) < 0) return -3;
+		if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -3;
 		if (str->s[0] != '*') {
 			for (s = str->s; *s; ++s) {
 				if (isalpha(*s)) ++c->n_cigar;
@@ -209,18 +207,18 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 			if (*s) parse_error(fp->n_lines, "unmatched CIGAR operation");
 			c->bin = bam_reg2bin(c->pos, bam_calend(c, bam1_cigar(b)));
 			doff += c->n_cigar * 4;
-		}
+		} else c->bin = bam_reg2bin(c->pos, c->pos + 1);
 	}
 	{ // mtid, mpos, isize
-		ret = ks_getuntil(ks, 0, str, &dret); c->mtid = strcmp(str->s, "=")? bam_get_tid(header, str->s) : c->tid;
-		ret = ks_getuntil(ks, 0, str, &dret); c->mpos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
-		ret = ks_getuntil(ks, 0, str, &dret); c->isize = (str->s[0] == '-' || isdigit(str->s[0]))? atoi(str->s) : 0;
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); c->mtid = strcmp(str->s, "=")? bam_get_tid(header, str->s) : c->tid;
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); c->mpos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); c->isize = (str->s[0] == '-' || isdigit(str->s[0]))? atoi(str->s) : 0;
 		if (ret < 0) return -4;
 	}
 	{ // seq and qual
 		int i;
 		uint8_t *p;
-		if (ks_getuntil(ks, 0, str, &dret) < 0) return -5; // seq
+		if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -5; // seq
 		c->l_qseq = strlen(str->s);
 		if (c->n_cigar && c->l_qseq != (int32_t)bam_cigar2qlen(c, bam1_cigar(b)))
 			parse_error(fp->n_lines, "CIGAR and sequence length are inconsistent");
@@ -228,16 +226,17 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 		bzero(p, (c->l_qseq+1)/2);
 		for (i = 0; i < c->l_qseq; ++i)
 			p[i/2] |= bam_nt16_table[(int)str->s[i]] << 4*(1-i%2);
-		if (ks_getuntil(ks, 0, str, &dret) < 0) return -6; // qual
-		if (c->l_qseq != strlen(str->s))
+		if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -6; // qual
+		if (strcmp(str->s, "*") && c->l_qseq != strlen(str->s))
 			parse_error(fp->n_lines, "sequence and quality are inconsistent");
 		p += (c->l_qseq+1)/2;
-		for (i = 0; i < c->l_qseq; ++i) p[i] = str->s[i] - 33;
+		if (strcmp(str->s, "*") == 0) for (i = 0; i < c->l_qseq; ++i) p[i] = 0xff;
+		else for (i = 0; i < c->l_qseq; ++i) p[i] = str->s[i] - 33;
 		doff += c->l_qseq + (c->l_qseq+1)/2;
 	}
 	doff0 = doff;
 	if (dret != '\n' && dret != '\r') { // aux
-		while (ks_getuntil(ks, 0, str, &dret) >= 0) {
+		while (ks_getuntil(ks, KS_SEP_TAB, str, &dret) >= 0) {
 			uint8_t *s, type, key[2];
 			if (str->l < 6 || str->s[2] != ':' || str->s[4] != ':')
 				parse_error(fp->n_lines, "missing colon in auxiliary data");
@@ -245,9 +244,9 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 			type = str->s[3];
 			s = alloc_data(b, doff + 3) + doff;
 			s[0] = key[0]; s[1] = key[1]; s += 2; doff += 2;
-			if (type == 'A' || type == 'a') {
+			if (type == 'A' || type == 'a' || type == 'c' || type == 'C') { // c and C for backward compatibility
 				s = alloc_data(b, doff + 2) + doff;
-				*s++ = type; *s = str->s[5];
+				*s++ = 'A'; *s = str->s[5];
 				doff += 2;
 			} else if (type == 'I' || type == 'i') {
 				long long x;
@@ -287,6 +286,11 @@ int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
 				*s++ = 'f';
 				*(float*)s = (float)atof(str->s + 5);
 				s += 4; doff += 5;
+			} else if (type == 'd') {
+				s = alloc_data(b, doff + 9) + doff;
+				*s++ = 'd';
+				*(float*)s = (float)atof(str->s + 9);
+				s += 8; doff += 9;
 			} else if (type == 'Z' || type == 'H') {
 				int size = 1 + (str->l - 5) + 1;
 				if (type == 'H') { // check whether the hex string is valid
@@ -341,7 +345,7 @@ static void taf2baf_core(const char *fntaf, const char *fnbaf, bam_header_t *hea
 	int ret;
 
 	b = (bam1_t*)calloc(1, sizeof(bam1_t));
-	fpbaf = bam_open(fnbaf, "w");
+	fpbaf = (strcmp(fnbaf, "-") == 0)? bam_dopen(fileno(stdout), "w") : bam_open(fnbaf, "w");
 	fp = sam_open(fntaf);
 	ret = sam_read1(fp, header, b);
 	bam_header_write(fpbaf, header);