Traditional roff(7) explicitly allows certain control characters

in the input stream (SOH, STX, ETX, ENQ, ACK, BEL, BS) for specific
purposes (leaders, backspace, delimiters, .tr), but making sure
these don't leak through to the output is tricky, so mark them as
unsupported for now.

2 files changed, 12 insertions, 9 deletions

diff --git a/mandoc.h b/mandoc.h
index 9cbc0ff0..8c0e9c8c 100644
--- a/mandoc.h
+++ b/mandoc.h
@@ -1,4 +1,4 @@
-/*	$Id: mandoc.h,v 1.184 2015/01/21 20:33:25 schwarze Exp $ */
+/*	$Id: mandoc.h,v 1.185 2015/01/22 21:38:16 schwarze Exp $ */
 /*
  * Copyright (c) 2010, 2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
@@ -145,7 +145,7 @@ enum	mandocerr {
 	/* related to document structure and macros */
 	MANDOCERR_FILE, /* cannot open file */
 	MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
-	MANDOCERR_BADCHAR, /* skipping bad character: number */
+	MANDOCERR_CHAR_BAD, /* skipping bad character: number */
 	MANDOCERR_MACRO, /* skipping unknown macro: macro */
 	MANDOCERR_REQ_INSEC, /* skipping insecure request: request */
 	MANDOCERR_IT_STRAY, /* skipping item outside list: It ... */
@@ -172,6 +172,7 @@ enum	mandocerr {
 	MANDOCERR_UNSUPP, /* ===== start of unsupported features ===== */
 
 	MANDOCERR_TOOLARGE, /* input too large */
+	MANDOCERR_CHAR_UNSUPP, /* unsupported control character: number */
 	MANDOCERR_REQ_UNSUPP, /* unsupported roff request: request */
 	MANDOCERR_TBL, /* unsupported table syntax */
 	MANDOCERR_TBLOPT, /* unsupported table option */
diff --git a/read.c b/read.c
index 8580f8de..0f79bed6 100644
--- a/read.c
+++ b/read.c
@@ -1,4 +1,4 @@
-/*	$Id: read.c,v 1.113 2015/01/22 19:26:50 schwarze Exp $ */
+/*	$Id: read.c,v 1.114 2015/01/22 21:38:16 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
@@ -215,6 +215,7 @@ static	const char * const	mandocerrs[MANDOCERR_MAX] = {
 
 	"unsupported feature",
 	"input too large",
+	"unsupported control character",
 	"unsupported roff request",
 	"unsupported table syntax",
 	"unsupported table option",
@@ -369,9 +370,8 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
 			if (c & 0x80) {
 				if ( ! (curp->filenc && preconv_encode(
 				    &blk, &i, &ln, &pos, &curp->filenc))) {
-					mandoc_vmsg(MANDOCERR_BADCHAR,
-					    curp, curp->line, pos,
-					    "0x%x", c);
+					mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
+					    curp->line, pos, "0x%x", c);
 					ln.buf[pos++] = '?';
 					i++;
 				}
@@ -383,8 +383,10 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
 			 */
 
 			if (c == 0x7f || (c < 0x20 && c != 0x09)) {
-				mandoc_vmsg(MANDOCERR_BADCHAR, curp,
-				    curp->line, pos, "0x%x", c);
+				mandoc_vmsg(c == 0x00 || c == 0x04 ||
+				    c > 0x0a ? MANDOCERR_CHAR_BAD :
+				    MANDOCERR_CHAR_UNSUPP,
+				    curp, curp->line, pos, "0x%x", c);
 				i++;
 				ln.buf[pos++] = '?';
 				continue;
@@ -440,7 +442,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
 
 			if ( ! (isascii(c) &&
 			    (isgraph(c) || isblank(c)))) {
-				mandoc_vmsg(MANDOCERR_BADCHAR, curp,
+				mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
 				    curp->line, pos, "0x%x", c);
 				i += 2;
 				ln.buf[pos++] = '?';