===================================================================
RCS file: /cvs/mandoc/main.c,v
retrieving revision 1.67
retrieving revision 1.70
diff -u -p -r1.67 -r1.70
--- mandoc/main.c	2010/05/15 05:50:19	1.67
+++ mandoc/main.c	2010/05/15 16:20:12	1.70
@@ -1,4 +1,4 @@
-/*	$Id: main.c,v 1.67 2010/05/15 05:50:19 joerg Exp $ */
+/*	$Id: main.c,v 1.70 2010/05/15 16:20:12 joerg Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -268,11 +268,31 @@ ffile(const char *file, struct curparse *curp)
 
 
 static int
+resize_buf(struct buf *buf, size_t initial)
+{
+	void *tmp;
+	size_t sz;
+
+	if (buf->sz == 0)
+		sz = initial;
+	else
+		sz = 2 * buf->sz;
+	tmp = realloc(buf->buf, sz);
+	if (NULL == tmp) {
+		perror(NULL);
+		return(0);
+	}
+	buf->buf = tmp;
+	buf->sz = sz;
+	return(1);
+}
+
+
+static int
 read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
 {
 	struct stat	 st;
-	char		*buf;
-	size_t		 sz, off;
+	size_t		 off;
 	ssize_t		 ssz;
 
 	if (-1 == fstat(curp->fd, &st)) {
@@ -319,17 +339,8 @@ read_whole_file(struct curparse *curp, struct buf *fb,
 						curp->file);
 				break;
 			}
-			if (fb->sz == 0)
-				sz = 65536;
-			else
-				sz = 2 * fb->sz;
-			buf = realloc(fb->buf, sz);
-			if (NULL == buf) {
-				perror(NULL);
+			if (! resize_buf(fb, 65536))
 				break;
-			}
-			fb->buf = buf;
-			fb->sz = sz;
 		}
 		ssz = read(curp->fd, fb->buf + off, fb->sz - off);
 		if (ssz == 0) {
@@ -353,89 +364,85 @@ read_whole_file(struct curparse *curp, struct buf *fb,
 static void
 fdesc(struct curparse *curp)
 {
-	size_t		 sz;
 	struct buf	 ln, blk;
-	int		 j, i, pos, lnn, comment, with_mmap;
+	int		 i, pos, lnn, lnn_start, with_mmap;
 	struct man	*man;
 	struct mdoc	*mdoc;
 
-	sz = BUFSIZ;
 	man = NULL;
 	mdoc = NULL;
 	memset(&ln, 0, sizeof(struct buf));
 
 	/*
-	 * Two buffers: ln and buf.  buf is the input buffer optimised
-	 * here for each file's block size.  ln is a line buffer.  Both
-	 * growable, hence passed in by ptr-ptr.
+	 * Two buffers: ln and buf.  buf is the input file and may be
+	 * memory mapped.  ln is a line buffer and grows on-demand.
 	 */
 
 	if (!read_whole_file(curp, &blk, &with_mmap))
 		return;
 
-	/* Fill buf with file blocksize. */
-
-	for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) {
-		if (pos >= (int)ln.sz) {
-			ln.sz += 256; /* Step-size. */
-			ln.buf = realloc(ln.buf, ln.sz);
-			if (NULL == ln.buf) {
-				perror(NULL);
-				goto bailout;
+	for (i = 0, lnn = 1; i < (int)blk.sz;) {
+		pos = 0;
+		lnn_start = lnn;
+		while (i < (int)blk.sz) {
+			if ('\n' == blk.buf[i]) {
+				++i;
+				++lnn;
+				break;
 			}
-		}
-
-		if ('\n' != blk.buf[i]) {
-			if (comment)
+			/* Trailing backslash is like a plain character. */
+			if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
+				if (pos >= (int)ln.sz)
+					if (! resize_buf(&ln, 256))
+						goto bailout;
+				ln.buf[pos++] = blk.buf[i++];
 				continue;
-			ln.buf[pos++] = blk.buf[i];
-
-			/* Handle in-line `\"' comments. */
-
-			if (1 == pos || '\"' != ln.buf[pos - 1])
-				continue;
-
-			for (j = pos - 2; j >= 0; j--)
-				if ('\\' != ln.buf[j])
-					break;
-
-			if ( ! ((pos - 2 - j) % 2))
-				continue;
-
-			comment = 1;
-			pos -= 2;
-			for (; pos > 0; --pos) {
-				if (ln.buf[pos - 1] != ' ')
-					break;
-				if (pos > 2 && ln.buf[pos - 2] == '\\')
-					break;
 			}
-			continue;
-		} 
-
-		/* Handle escaped `\\n' newlines. */
-
-		if (pos > 0 && 0 == comment && '\\' == ln.buf[pos - 1]) {
-			for (j = pos - 1; j >= 0; j--)
-				if ('\\' != ln.buf[j])
-					break;
-			if ( ! ((pos - j) % 2)) {
-				pos--;
-				lnn++;
+			/* Found an escape and at least one other character. */
+			if ('\n' == blk.buf[i + 1]) {
+				/* Escaped newlines are skipped over */
+				i += 2;
+				++lnn;
 				continue;
 			}
+			if ('"' == blk.buf[i + 1]) {
+				i += 2;
+				/* Comment, skip to end of line */
+				for (; i < (int)blk.sz; ++i) {
+					if ('\n' == blk.buf[i]) {
+						++i;
+						++lnn;
+						break;
+					}
+				}
+				/* Backout trailing whitespaces */
+				for (; pos > 0; --pos) {
+					if (ln.buf[pos - 1] != ' ')
+						break;
+					if (pos > 2 && ln.buf[pos - 2] == '\\')
+						break;
+				}
+				break;
+			}
+			/* Some other escape sequence, copy and continue. */
+			if (pos + 1 >= (int)ln.sz)
+				if (! resize_buf(&ln, 256))
+					goto bailout;
+
+			ln.buf[pos++] = blk.buf[i++];
+			ln.buf[pos++] = blk.buf[i++];
 		}
 
+ 		if (pos >= (int)ln.sz)
+			if (! resize_buf(&ln, 256))
+				goto bailout;
 		ln.buf[pos] = 0;
-		lnn++;
 
 		/* If unset, assign parser in pset(). */
 
 		if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc))
 			goto bailout;
 
-		pos = comment = 0;
-
 		/* Pass down into parsers. */
 
 		if (man && ! man_parseln(man, lnn, ln.buf))
@@ -479,7 +486,7 @@ fdesc(struct curparse *curp)
 		case (OUTT_LINT):
 			break;
 		default:
-			curp->outdata = ascii_alloc();
+			curp->outdata = ascii_alloc(80);
 			curp->outman = terminal_man;
 			curp->outmdoc = terminal_mdoc;
 			curp->outfree = terminal_free;