Replace the kludge for the \z escape sequence by an actual

implementation.  As a side effect, minus ten lines of code.

As another side effect, this also fixes the assertion failure that
used to be triggered by "\z\o'ab'c" at the beginning of an output
line, found by jsg@ with afl (test case 022/Apr27).

3 files changed, 39 insertions, 57 deletions

diff --git a/roff.7 b/roff.7
index dc39b69b..02b7212e 100644
--- a/roff.7
+++ b/roff.7
@@ -1,4 +1,4 @@
-.\"	$Id: roff.7,v 1.70 2015/02/17 17:16:52 schwarze Exp $
+.\"	$Id: roff.7,v 1.71 2015/04/29 18:35:00 schwarze Exp $
 .\"
 .\" Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
 .\" Copyright (c) 2010, 2011, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
@@ -15,7 +15,7 @@
 .\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 .\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 .\"
-.Dd $Mdocdate: February 17 2015 $
+.Dd $Mdocdate: April 29 2015 $
 .Dt ROFF 7
 .Os
 .Sh NAME
@@ -2028,10 +2028,7 @@ Print
 with zero width and height; ignored by
 .Xr mandoc 1 .
 .Ss \ez
-Output the next character without advancing the cursor position;
-approximated in
-.Xr mandoc 1
-by simply skipping the next character.
+Output the next character without advancing the cursor position.
 .Sh COMPATIBILITY
 The
 .Xr mandoc 1
diff --git a/term.c b/term.c
index 68f3ea04..79d116a3 100644
--- a/term.c
+++ b/term.c
@@ -1,4 +1,4 @@
-/*	$Id: term.c,v 1.247 2015/04/04 17:47:18 schwarze Exp $ */
+/*	$Id: term.c,v 1.248 2015/04/29 18:35:00 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
@@ -265,6 +265,7 @@ term_flushln(struct termp *p)
 
 	p->col = 0;
 	p->overstep = 0;
+	p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE);
 
 	if ( ! (TERMP_NOBREAK & p->flags)) {
 		p->viscol = 0;
@@ -417,11 +418,6 @@ term_word(struct termp *p, const char *word)
 
 	while ('\0' != *word) {
 		if ('\\' != *word) {
-			if (TERMP_SKIPCHAR & p->flags) {
-				p->flags &= ~TERMP_SKIPCHAR;
-				word++;
-				continue;
-			}
 			if (TERMP_NBRWORD & p->flags) {
 				if (' ' == *word) {
 					encode(p, nbrsp, 1);
@@ -480,13 +476,13 @@ term_word(struct termp *p, const char *word)
 			term_fontlast(p);
 			continue;
 		case ESCAPE_NOSPACE:
-			if (TERMP_SKIPCHAR & p->flags)
-				p->flags &= ~TERMP_SKIPCHAR;
-			else if ('\0' == *word)
+			if (p->flags & TERMP_BACKAFTER)
+				p->flags &= ~TERMP_BACKAFTER;
+			else if (*word == '\0')
 				p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
 			continue;
 		case ESCAPE_SKIPCHAR:
-			p->flags |= TERMP_SKIPCHAR;
+			p->flags |= TERMP_BACKAFTER;
 			continue;
 		case ESCAPE_OVERSTRIKE:
 			cp = seq + sz;
@@ -496,9 +492,14 @@ term_word(struct termp *p, const char *word)
 					continue;
 				}
 				encode1(p, *seq++);
-				if (seq < cp)
-					encode(p, "\b", 1);
+				if (seq < cp) {
+					if (p->flags & TERMP_BACKBEFORE)
+						p->flags |= TERMP_BACKAFTER;
+					else
+						p->flags |= TERMP_BACKBEFORE;
+				}
 			}
+			continue;
 		default:
 			continue;
 		}
@@ -553,16 +554,16 @@ encode1(struct termp *p, int c)
 {
 	enum termfont	  f;
 
-	if (TERMP_SKIPCHAR & p->flags) {
-		p->flags &= ~TERMP_SKIPCHAR;
-		return;
-	}
+	if (p->col + 7 >= p->maxcols)
+		adjbuf(p, p->col + 7);
 
-	if (p->col + 6 >= p->maxcols)
-		adjbuf(p, p->col + 6);
-
-	f = p->fontq[p->fonti];
+	f = (c == ASCII_HYPH || isgraph(c)) ?
+	    p->fontq[p->fonti] : TERMFONT_NONE;
 
+	if (p->flags & TERMP_BACKBEFORE) {
+		p->buf[p->col++] = 8;
+		p->flags &= ~TERMP_BACKBEFORE;
+	}
 	if (TERMFONT_UNDER == f || TERMFONT_BI == f) {
 		p->buf[p->col++] = '_';
 		p->buf[p->col++] = 8;
@@ -575,6 +576,10 @@ encode1(struct termp *p, int c)
 		p->buf[p->col++] = 8;
 	}
 	p->buf[p->col++] = c;
+	if (p->flags & TERMP_BACKAFTER) {
+		p->flags |= TERMP_BACKBEFORE;
+		p->flags &= ~TERMP_BACKAFTER;
+	}
 }
 
 static void
@@ -582,29 +587,8 @@ encode(struct termp *p, const char *word, size_t sz)
 {
 	size_t		  i;
 
-	if (TERMP_SKIPCHAR & p->flags) {
-		p->flags &= ~TERMP_SKIPCHAR;
-		return;
-	}
-
-	/*
-	 * Encode and buffer a string of characters.  If the current
-	 * font mode is unset, buffer directly, else encode then buffer
-	 * character by character.
-	 */
-
-	if (p->fontq[p->fonti] == TERMFONT_NONE) {
-		if (p->col + sz >= p->maxcols)
-			adjbuf(p, p->col + sz);
-		for (i = 0; i < sz; i++)
-			p->buf[p->col++] = word[i];
-		return;
-	}
-
-	/* Pre-buffer, assuming worst-case. */
-
-	if (p->col + 1 + (sz * 5) >= p->maxcols)
-		adjbuf(p, p->col + 1 + (sz * 5));
+	if (p->col + 2 + (sz * 5) >= p->maxcols)
+		adjbuf(p, p->col + 2 + (sz * 5));
 
 	for (i = 0; i < sz; i++) {
 		if (ASCII_HYPH == word[i] ||
diff --git a/term.h b/term.h
index bda7f7da..0aa467b4 100644
--- a/term.h
+++ b/term.h
@@ -1,4 +1,4 @@
-/*	$Id: term.h,v 1.113 2015/04/04 17:47:18 schwarze Exp $ */
+/*	$Id: term.h,v 1.114 2015/04/29 18:35:00 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2011-2015 Ingo Schwarze <schwarze@openbsd.org>
@@ -67,13 +67,14 @@ struct	termp {
 	int		  overstep;	/* See termp_flushln(). */
 	int		  skipvsp;	/* Vertical space to skip. */
 	int		  flags;
-#define	TERMP_SENTENCE	 (1 << 1)	/* Space before a sentence. */
-#define	TERMP_NOSPACE	 (1 << 2)	/* No space before words. */
-#define	TERMP_NONOSPACE	 (1 << 3)	/* No space (no autounset). */
-#define	TERMP_NBRWORD	 (1 << 4)	/* Make next word nonbreaking. */
-#define	TERMP_KEEP	 (1 << 5)	/* Keep words together. */
-#define	TERMP_PREKEEP	 (1 << 6)	/* ...starting with the next one. */
-#define	TERMP_SKIPCHAR	 (1 << 7)	/* Skip the next character. */
+#define	TERMP_SENTENCE	 (1 << 0)	/* Space before a sentence. */
+#define	TERMP_NOSPACE	 (1 << 1)	/* No space before words. */
+#define	TERMP_NONOSPACE	 (1 << 2)	/* No space (no autounset). */
+#define	TERMP_NBRWORD	 (1 << 3)	/* Make next word nonbreaking. */
+#define	TERMP_KEEP	 (1 << 4)	/* Keep words together. */
+#define	TERMP_PREKEEP	 (1 << 5)	/* ...starting with the next one. */
+#define	TERMP_BACKAFTER	 (1 << 6)	/* Back up after next character. */
+#define	TERMP_BACKBEFORE (1 << 7)	/* Back up before next character. */
 #define	TERMP_NOBREAK	 (1 << 8)	/* See term_flushln(). */
 #define	TERMP_BRIND	 (1 << 9)	/* See term_flushln(). */
 #define	TERMP_DANGLE	 (1 << 10)	/* See term_flushln(). */