X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=src%2Fread_dna.c;h=01a4db32c5d2b1de517d2f659496a73c1d1b9c13;hb=a0436318d70829a2d16134be7ca1d6d454613a20;hp=6d44c6665603d19c0bcd0f6c8aea2409c94e671b;hpb=be6a044200152fd83d0b72f348012a0adc2593c5;p=ape.git

diff --git a/src/read_dna.c b/src/read_dna.c
index 6d44c66..01a4db3 100644
--- a/src/read_dna.c
+++ b/src/read_dna.c
@@ -1,6 +1,6 @@
-/* read_dna.c       2012-12-27 */
+/* read_dna.c       2013-01-04 */
 
-/* Copyright 2008-2012 Emmanuel Paradis */
+/* Copyright 2013 Emmanuel Paradis */
 
 /* This file is part of the R-package `ape'. */
 /* See the file ../COPYING for licensing issues. */
@@ -81,7 +81,7 @@ static const unsigned char lineFeed = 0x0a;
 
 SEXP rawStreamToDNAbin(SEXP x)
 {
-	int N, i, j, k, n;
+	int N, i, j, k, n, startOfSeq;
 	unsigned char *xr, *rseq, *buffer, tmp;
 	SEXP obj, nms, seq;
 
@@ -89,15 +89,25 @@ SEXP rawStreamToDNAbin(SEXP x)
 	N = LENGTH(x);
 	xr = RAW(x);
 
-/* do a 1st pass to find the number of sequences */
+/* do a 1st pass to find the number of sequences
 
-	n = 0;
-	j = 0; /* use j as a flag */
-	for (i = 0; i < N; i++) {
+   this code should be robust to '>' present inside
+   a label or in the header text before the sequences */
+
+	n = j = 0; /* use j as a flag */
+	if (xr[0] == hook) {
+		j = 1;
+		startOfSeq = 0;
+	}
+	i = 1;
+	for (i = 1; i < N; i++) {
 		if (j && xr[i] == lineFeed) {
 			n++;
 			j = 0;
-		} else if (xr[i] == hook) j = 1;
+		} else if (xr[i] == hook) {
+			if (!n) startOfSeq = i;
+			j = 1;
+		}
 	}
 
 	PROTECT(obj = allocVector(VECSXP, n));
@@ -106,16 +116,15 @@ SEXP rawStreamToDNAbin(SEXP x)
 /* Refine the way the size of the buffer is set? */
 	buffer = (unsigned char *)R_alloc(N, sizeof(unsigned char *));
 
-	i = j = 0;
+	i = startOfSeq;
+	j = 0; /* gives the index of the sequence */
 	while (i < N) {
 		/* 1st read the label... */
-		if (xr[i] == hook) {
-			i++;
-			k = 0;
-			while (xr[i] != lineFeed) buffer[k++] = xr[i++];
-			buffer[k] = '\0';
-			SET_STRING_ELT(nms, j, mkChar(buffer));
-		}
+		i++;
+		k = 0;
+		while (xr[i] != lineFeed) buffer[k++] = xr[i++];
+		buffer[k] = '\0';
+		SET_STRING_ELT(nms, j, mkChar(buffer));
 		/* ... then read the sequence */
 		n = 0;
 		while (xr[i] != hook && i < N) {
@@ -125,7 +134,7 @@ SEXP rawStreamToDNAbin(SEXP x)
    then the following check would not be needed; additionally the size
    of tab_trans could be restriced to 0-121. This check has the
    advantage that all invalid characters are simply ignored without
-   causing error. */
+   causing error -- except if '>' occurs in the middle of a sequence. */
 			if (tmp) buffer[n++] = tmp;
 		}
 		PROTECT(seq = allocVector(RAWSXP, n));