Implement the roff \z escape sequence, intended to output the next

character without advancing the cursor position; implement it to
simply skip the next character, as it will usually be overwritten.

With this change, the pod2man(1) preamble user-defined string \*:,
intended to render as a diaeresis or umlaut diacritic above the
preceding character, is rendered in a slightly less ugly way,
though still not correctly.  It was rendered as "z.." and is now
rendered as ".".

Given that the definition of \*: uses elaborate manual \h positioning,
there is little chance for mandoc(1) to ever render it correctly,
but at least we can refrain from printing out a spurious "z", and
we can make the \z do something semi-reasonable for easier cases.

"just commit" kristaps@

1 files changed, 54 insertions, 14 deletions

diff --git a/term.c b/term.c
index 3dadb7e4..a10fce7a 100644
--- a/term.c
+++ b/term.c
@@ -1,4 +1,4 @@
-/*	$Id: term.c,v 1.202 2012/05/27 18:02:49 schwarze Exp $ */
+/*	$Id: term.c,v 1.203 2012/05/31 22:29:13 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011, 2012 Ingo Schwarze <schwarze@openbsd.org>
@@ -33,6 +33,7 @@
 #include "term.h"
 #include "main.h"
 
+static	size_t		 cond_width(const struct termp *, int, int *);
 static	void		 adjbuf(struct termp *p, int);
 static	void		 bufferc(struct termp *, char);
 static	void		 encode(struct termp *, const char *, size_t);
@@ -419,12 +420,17 @@ term_word(struct termp *p, const char *word)
 	p->flags &= ~(TERMP_SENTENCE | TERMP_IGNDELIM);
 
 	while ('\0' != *word) {
-		if ((ssz = strcspn(word, "\\")) > 0)
+		if ('\\' != *word) {
+			if (TERMP_SKIPCHAR & p->flags) {
+				p->flags &= ~TERMP_SKIPCHAR;
+				word++;
+				continue;
+			}
+			ssz = strcspn(word, "\\");
 			encode(p, word, ssz);
-
-		word += (int)ssz;
-		if ('\\' != *word)
+			word += (int)ssz;
 			continue;
+		}
 
 		word++;
 		esc = mandoc_escape(&word, &seq, &sz);
@@ -480,9 +486,14 @@ term_word(struct termp *p, const char *word)
 			term_fontlast(p);
 			break;
 		case (ESCAPE_NOSPACE):
-			if ('\0' == *word)
+			if (TERMP_SKIPCHAR & p->flags)
+				p->flags &= ~TERMP_SKIPCHAR;
+			else if ('\0' == *word)
 				p->flags |= TERMP_NOSPACE;
 			break;
+		case (ESCAPE_SKIPCHAR):
+			p->flags |= TERMP_SKIPCHAR;
+			break;
 		default:
 			break;
 		}
@@ -522,6 +533,11 @@ encode1(struct termp *p, int c)
 {
 	enum termfont	  f;
 
+	if (TERMP_SKIPCHAR & p->flags) {
+		p->flags &= ~TERMP_SKIPCHAR;
+		return;
+	}
+
 	if (p->col + 4 >= p->maxcols)
 		adjbuf(p, p->col + 4);
 
@@ -545,6 +561,11 @@ encode(struct termp *p, const char *word, size_t sz)
 	enum termfont	  f;
 	int		  i, len;
 
+	if (TERMP_SKIPCHAR & p->flags) {
+		p->flags &= ~TERMP_SKIPCHAR;
+		return;
+	}
+
 	/* LINTED */
 	len = sz;
 
@@ -593,12 +614,22 @@ term_len(const struct termp *p, size_t sz)
 	return((*p->width)(p, ' ') * sz);
 }
 
+static size_t
+cond_width(const struct termp *p, int c, int *skip)
+{
+
+	if (*skip) {
+		(*skip) = 0;
+		return(0);
+	} else
+		return((*p->width)(p, c));
+}
 
 size_t
 term_strlen(const struct termp *p, const char *cp)
 {
 	size_t		 sz, rsz, i;
-	int		 ssz, c;
+	int		 ssz, skip, c;
 	const char	*seq, *rhs;
 	enum mandoc_esc	 esc;
 	static const char rej[] = { '\\', ASCII_HYPH, ASCII_NBRSP, '\0' };
@@ -610,10 +641,11 @@ term_strlen(const struct termp *p, const char *cp)
 	 */
 
 	sz = 0;
+	skip = 0;
 	while ('\0' != *cp) {
 		rsz = strcspn(cp, rej);
 		for (i = 0; i < rsz; i++)
-			sz += (*p->width)(p, *cp++);
+			sz += cond_width(p, *cp++, &skip);
 
 		c = 0;
 		switch (*cp) {
@@ -630,14 +662,14 @@ term_strlen(const struct termp *p, const char *cp)
 						(seq + 1, ssz - 1);
 					if ('\0' == c)
 						break;
-					sz += (*p->width)(p, c);
+					sz += cond_width(p, c, &skip);
 					continue;
 				case (ESCAPE_SPECIAL):
 					c = mchars_spec2cp
 						(p->symtab, seq, ssz);
 					if (c <= 0)
 						break;
-					sz += (*p->width)(p, c);
+					sz += cond_width(p, c, &skip);
 					continue;
 				default:
 					break;
@@ -647,12 +679,12 @@ term_strlen(const struct termp *p, const char *cp)
 
 			switch (esc) {
 			case (ESCAPE_UNICODE):
-				sz += (*p->width)(p, '?');
+				sz += cond_width(p, '?', &skip);
 				break;
 			case (ESCAPE_NUMBERED):
 				c = mchars_num2char(seq, ssz);
 				if ('\0' != c)
-					sz += (*p->width)(p, c);
+					sz += cond_width(p, c, &skip);
 				break;
 			case (ESCAPE_SPECIAL):
 				rhs = mchars_spec2str
@@ -664,6 +696,9 @@ term_strlen(const struct termp *p, const char *cp)
 				rhs = seq;
 				rsz = ssz;
 				break;
+			case (ESCAPE_SKIPCHAR):
+				skip = 1;
+				break;
 			default:
 				break;
 			}
@@ -671,15 +706,20 @@ term_strlen(const struct termp *p, const char *cp)
 			if (NULL == rhs)
 				break;
 
+			if (skip) {
+				skip = 0;
+				break;
+			}
+
 			for (i = 0; i < rsz; i++)
 				sz += (*p->width)(p, *rhs++);
 			break;
 		case (ASCII_NBRSP):
-			sz += (*p->width)(p, ' ');
+			sz += cond_width(p, ' ', &skip);
 			cp++;
 			break;
 		case (ASCII_HYPH):
-			sz += (*p->width)(p, '-');
+			sz += cond_width(p, '-', &skip);
 			cp++;
 			break;
 		default: