From fb40e7adf35e6b21b0e64e170b71e1a4d2cf5351 Mon Sep 17 00:00:00 2001
From: Kristaps Dzonsons <kristaps@bsd.lv>
Date: Thu, 4 Dec 2008 16:34:59 +0000
Subject: Character-set validation fixes.

---
 index.7   | 16 +++++++++++++---
 mlg.c     | 13 ++++++++-----
 private.h |  3 ++-
 roff.c    | 10 +++++++++-
 tokens.c  |  9 ++++++++-
 5 files changed, 40 insertions(+), 11 deletions(-)

diff --git a/index.7 b/index.7
index ff3f0d59..b88246bf 100644
--- a/index.7
+++ b/index.7
@@ -21,8 +21,9 @@ and
 .Nm
 acts directly on source documents, validating its input and producing a
 variety of outputs.
-.Ss Features
-In order to operate properly, 
+.\"
+.Ss Validation
+In order to operate sanely,
 .Nm
 fully validates its input.  This includes, but is not limited to, the
 following checks:
@@ -40,10 +41,19 @@ macros clobbering a pending
 .Sq \&.Bl
 scope),
 .It 
-predefined characters (such as \\*(>= and \\*q),
+predefined characters (such as \\*(>= and \\*q, rendering as \*(>= and \*q,
+respectively),
 .It
 correctly-ordered document prelude,
+.It
+sane argument values (such as those for 
+.Sq \& Dt
+or
+.Sq \& Sm ) ,
+.It
+and so on.
 .El
+.Pp
 .\" 
 .Sh ENVIRONMENT
 The
diff --git a/mlg.c b/mlg.c
index ac413e3a..52bbb822 100644
--- a/mlg.c
+++ b/mlg.c
@@ -594,10 +594,6 @@ mlg_rofftoken(void *arg, int space, int value)
 		seq = "&nbsp;";
 		sz = 6;
 		break;
-	case (ROFFTok_Null):
-		seq = "";
-		sz = 0;
-		break;
 	case (ROFFTok_Hyphen):
 		seq = "&#8208;";
 		sz = 7;
@@ -670,13 +666,20 @@ mlg_rofftoken(void *arg, int space, int value)
 		seq = "Nan";
 		sz = 3;
 		break;
+	case (ROFFTok_Quote):
+		seq = "&quot;";
+		sz = 6;
+		break;
+	default:
+		/* TODO: print error. */
+		return(0);
 	}
 
 	if (space && ! ml_nputs(p->mbuf, " ", 1, &res))
 		return(0);
 	p->pos += res;
 
-	if (0 != sz && ! ml_nputs(p->mbuf, seq, sz, &res))
+	if ( ! ml_nputs(p->mbuf, seq, sz, &res))
 		return(0);
 	p->pos += res;
 
diff --git a/private.h b/private.h
index e5a55431..b60686a8 100644
--- a/private.h
+++ b/private.h
@@ -64,7 +64,8 @@ struct	md_mbuf {
 #define	ROFFTok_Infty	 24
 #define	ROFFTok_Bar	 25
 #define	ROFFTok_Nan	 26
-#define	ROFFTok_MAX	 27
+#define	ROFFTok_Quote	 27
+#define	ROFFTok_MAX	 28
 
 #define	ROFF___	 	 0
 #define	ROFF_Dd		 1
diff --git a/roff.c b/roff.c
index b30add22..08d10ffb 100644
--- a/roff.c
+++ b/roff.c
@@ -971,12 +971,20 @@ roffdata(struct rofftree *tree, int space, char *buf)
 {
 	int		 tok;
 
+	if (0 == *buf)
+		return(1);
+
 	if (-1 == (tok = rofftok_scan(buf))) {
 		roff_err(tree, buf, "invalid character sequence");
 		return(0);
-	} else if (ROFFTok_MAX != tok) 
+	} else if (ROFFTok_MAX != tok) {
+		if (ROFFTok_Null == tok) { /* FIXME */
+			buf += 2;
+			return(roffdata(tree, space, buf));
+		}
 		return((*tree->cb.rofftoken)
 				(tree->arg, space != 0, tok));
+	}
 
 	return((*tree->cb.roffdata)(tree->arg, 
 				space != 0, tree->cur, buf));
diff --git a/tokens.c b/tokens.c
index 07cdb671..10d35153 100644
--- a/tokens.c
+++ b/tokens.c
@@ -88,7 +88,12 @@ rofftok_predef(const char *buf)
 	if ('(' == *buf)
 		return(rofftok_defined(++buf));
 
-	/* TODO */
+	switch (*buf) {
+	case ('q'):
+		return(ROFFTok_Quote);
+	default:
+		break;
+	}
 
 	return(-1);
 }
@@ -173,6 +178,8 @@ rofftok_scan(const char *buf)
 			return(ROFFTok_Hyphen);
 		case ('*'):
 			return(rofftok_predef(++buf));
+		case ('\\'):
+			return(ROFFTok_MAX);
 		default:
 			break;
 		}
-- 
cgit