X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_import.c;h=9d84328bdbdafb5f4e76a19e7edf176121aca5b4;hb=205b696ea778f5b7cfe70e5531bfe6583b5af333;hp=427140db9cbe9e3c80d173b0d69e31b1774c3037;hpb=d0e30eec1158752010659982342a611fc91ae8e3;p=samtools.git

diff --git a/bam_import.c b/bam_import.c
index 427140d..9d84328 100644
--- a/bam_import.c
+++ b/bam_import.c
@@ -116,7 +116,7 @@ static bam_header_t *hash2header(const kh_ref_t *hash)
 bam_header_t *sam_header_read2(const char *fn)
 {
 	bam_header_t *header;
-	int c, dret, ret;
+	int c, dret, ret, error = 0;
 	gzFile fp;
 	kstream_t *ks;
 	kstring_t *str;
@@ -135,6 +135,10 @@ bam_header_t *sam_header_read2(const char *fn)
 		ks_getuntil(ks, 0, str, &dret);
 		len = atoi(str->s);
 		k = kh_put(ref, hash, s, &ret);
+		if (ret == 0) {
+			fprintf(stderr, "[sam_header_read2] duplicated sequence name: %s\n", s);
+			error = 1;
+		}
 		kh_value(hash, k) = (uint64_t)len<<32 | i;
 		if (dret != '\n')
 			while ((c = ks_getc(ks)) != '\n' && c != -1);
@@ -143,6 +147,7 @@ bam_header_t *sam_header_read2(const char *fn)
 	gzclose(fp);
 	free(str->s); free(str);
 	fprintf(stderr, "[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
+	if (error) return 0;
 	header = hash2header(hash);
 	kh_destroy(ref, hash);
 	return header;
@@ -163,44 +168,48 @@ static inline void parse_error(int64_t n_lines, const char * __restrict msg)
 }
 static inline void append_text(bam_header_t *header, kstring_t *str)
 {
-	int x = header->l_text, y = header->l_text + str->l + 2; // 2 = 1 byte dret + 1 byte null
+	size_t x = header->l_text, y = header->l_text + str->l + 2; // 2 = 1 byte dret + 1 byte null
 	kroundup32(x); kroundup32(y);
-	if (x < y) header->text = (char*)realloc(header->text, y);
+	if (x < y) 
+    {
+        header->n_text = y;
+        header->text = (char*)realloc(header->text, y);
+        if ( !header->text ) 
+        {
+            fprintf(stderr,"realloc failed to alloc %ld bytes\n", y);
+            abort();
+        }
+    }
+    // Sanity check
+    if ( header->l_text+str->l+1 >= header->n_text )
+    {
+        fprintf(stderr,"append_text FIXME: %ld>=%ld, x=%ld,y=%ld\n",  header->l_text+str->l+1,header->n_text,x,y);
+        abort();
+    }
 	strncpy(header->text + header->l_text, str->s, str->l+1); // we cannot use strcpy() here.
 	header->l_text += str->l + 1;
 	header->text[header->l_text] = 0;
 }
 
-int sam_header_parse_rg(bam_header_t *h)
-{
-	if (h->dict == 0) h->dict = sam_header_parse2(h->text);
-	if (h->rg2lib) h->rg2lib = sam_header2tbl(h->dict, "RG", "ID", "LB");
-	return sam_tbl_size(h->rg2lib);
-}
-
 int sam_header_parse(bam_header_t *h)
 {
-	void *tbl;
 	char **tmp;
 	int i;
 	free(h->target_len); free(h->target_name);
 	h->n_targets = 0; h->target_len = 0; h->target_name = 0;
 	if (h->l_text < 3) return 0;
 	if (h->dict == 0) h->dict = sam_header_parse2(h->text);
-	tbl = sam_header2tbl(h->dict, "SQ", "SN", "LN");
-	h->n_targets = sam_tbl_size(tbl);
-	if (h->n_targets == 0) {
-		sam_tbl_destroy(tbl);
-		return 0;
-	}
-	h->target_len = (uint32_t*)calloc(h->n_targets, 4);
-	h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
-	tmp = (char**)calloc(h->n_targets, sizeof(void*));
-	sam_tbl_pair(tbl, h->target_name, tmp);
+	tmp = sam_header2list(h->dict, "SQ", "SN", &h->n_targets);
+	if (h->n_targets == 0) return 0;
+	h->target_name = calloc(h->n_targets, sizeof(void*));
+	for (i = 0; i < h->n_targets; ++i)
+		h->target_name[i] = strdup(tmp[i]);
+	free(tmp);
+	tmp = sam_header2list(h->dict, "SQ", "LN", &h->n_targets);
+	h->target_len = calloc(h->n_targets, 4);
 	for (i = 0; i < h->n_targets; ++i)
 		h->target_len[i] = atoi(tmp[i]);
 	free(tmp);
-	sam_tbl_destroy(tbl);
 	return h->n_targets;
 }