X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=padding.c;h=88945e2ca5c54c1a592a45899358996ddd810ece;hb=0b3418cf166ce4a58cedf0d9a2df5ec3dd4cc5fa;hp=23d1b23c9ac68dd8dd6a5d1c7847d714e3e837ed;hpb=2b4d9b14c3cdd96b9b1df2e9f2c83ceef4a76593;p=samtools.git

diff --git a/padding.c b/padding.c
index 23d1b23..88945e2 100644
--- a/padding.c
+++ b/padding.c
@@ -170,7 +170,7 @@ int bam_pad2unpad(samfile_t *in, samfile_t *out, faidx_t *fai)
 			r_tid = b->core.tid;
 			unpad_seq(b, &r);
 			if (h->target_len[r_tid] != r.l) {
-				fprintf(stderr, "[depad] ERROR: (Padded) length of '%s' is %i in BAM header, but %id in embedded reference\n", bam1_qname(b), h->target_len[r_tid], r.l);
+				fprintf(stderr, "[depad] ERROR: (Padded) length of '%s' is %d in BAM header, but %ld in embedded reference\n", bam1_qname(b), h->target_len[r_tid], r.l);
 				return -1;
 			}
 			if (fai) {
@@ -186,8 +186,8 @@ int bam_pad2unpad(samfile_t *in, samfile_t *out, faidx_t *fai)
 						// Show gaps as ASCII 45
 						fprintf(stderr, "[depad] ERROR: Embedded sequence and reference FASTA don't match for %s base %i, '%c' vs '%c'\n",
 							h->target_name[b->core.tid], i+1,
-							r.s[i] ? bam_nt16_rev_table[r.s[i]] : 45,
-							q.s[i] ? bam_nt16_rev_table[q.s[i]] : 45);
+							r.s[i] ? bam_nt16_rev_table[(int)r.s[i]] : 45,
+							q.s[i] ? bam_nt16_rev_table[(int)q.s[i]] : 45);
 						return -1;
 					}
 				}
@@ -295,10 +295,48 @@ bam_header_t * fix_header(bam_header_t *old, faidx_t *fai)
 			fprintf(stderr, "[depad] ERROR getting unpadded length of '%s', padded length %i\n", old->target_name[i], old->target_len[i]);
 		} else {
 			header->target_len[i] = unpadded_len;
-			fprintf(stderr, "[depad] Recalculating '%s' length %i -> %i\n", old->target_name[i], old->target_len[i], header->target_len[i]);
+			//fprintf(stderr, "[depad] Recalculating '%s' length %i -> %i\n", old->target_name[i], old->target_len[i], header->target_len[i]);
 		}
 	}
-	/* TODO - Correct the @SQ reference lengths from padded to unpadded */
+	/* Duplicating the header allocated new buffer for header string */
+	/* After modifying the @SQ lines it will only get smaller, since */
+	/* the LN entries will be the same or shorter, and we'll remove */
+	/* any MD entries (MD5 checksums). */
+	assert(strlen(old->text) == strlen(header->text));
+	assert (0==strcmp(old->text, header->text));
+	const char *text;
+	text = old->text;
+	header->text[0] = '\0'; /* Resuse the allocated buffer */
+	char * newtext = header->text;
+	char * end=NULL;
+	while (text[0]=='@') {
+		end = strchr(text, '\n');
+		assert(end != 0);
+		if (text[1]=='S' && text[2]=='Q' && text[3]=='\t') {
+			/* TODO - edit the @SQ line here to remove MD and fix LN. */
+			/* For now just remove the @SQ line, and samtools will */
+			/* automatically generate a minimal replacement with LN. */
+			/* However, that discards any other tags like AS, SP, UR. */
+			//fprintf(stderr, "[depad] Removing @SQ line\n");
+		} else {
+			/* Copy this line to the new header */
+			strncat(newtext, text, end - text + 1);
+		}
+		text = end + 1;
+	}
+	assert (text[0]=='\0');
+	/* Check we didn't overflow the buffer */
+	assert (strlen(header->text) <= strlen(old->text));
+	if (strlen(header->text) < header->l_text) {
+		//fprintf(stderr, "[depad] Reallocating header buffer\n");
+		assert (newtext == header->text);
+		newtext = calloc(strlen(header->text) + 1, 1);
+		strcpy(newtext, header->text);
+		free(header->text);
+		header->text = newtext;
+		header->l_text = strlen(newtext);
+	}
+	//fprintf(stderr, "[depad] Here is the new header (pending @SQ lines),\n\n%s\n(end)\n", header->text);
 	return header;
 }