Support some escape sequences, in particular character escape sequences,

inside \w arguments, and skip most other escape sequences when measuring
the output length in this way because most escape sequences contribute
little or nothing to text width: for example, consider font escapes in
terminal output.

This implementation is very rudimentary.  In particular, it assumes that
every character has the same width.  No attempt is made to detect
double-width or zero-width Unicode characters or to take dependencies on
output devices or fonts into account.  These limitations are hard to
avoid because mandoc has to interpolate \w at the parsing stage when the
output device is not yet known.  I really do not want the content of the
syntax tree to depend on the output device.

Feature requested by Paul <Eggert at cs dot ucla dot edu>, who also
submitted a patch, but i chose to commit this very different patch
with almost the same functionality.
His input was still very valuable because complete support for \w is
out of the question, and consequently, the main task is identifying
subsets of the feature that are needed for real-world manual pages
and can be supported without uprooting the whole forest.

5 files changed, 61 insertions, 11 deletions

diff --git a/regress/roff/esc/w.in b/regress/roff/esc/w.in
index c30a31ac..59a9bdb5 100644
--- a/regress/roff/esc/w.in
+++ b/regress/roff/esc/w.in
@@ -1,4 +1,4 @@
-.\" $OpenBSD: w.in,v 1.4 2022/06/08 13:08:00 schwarze Exp $
+.\" $OpenBSD: w.in,v 1.5 2023/10/23 20:07:19 schwarze Exp $
 .Dd $Mdocdate$
 .Dt ESC-W 1
 .Os
@@ -13,6 +13,20 @@ character: \w'n'
 blank: \w' '
 .br
 text: \w'text'
+.br
+special: \w'\(bu'
+.br
+numbered: \w'\N'100''
+.br
+Unicode: \w'\[u2013]'
+.br
+overstrike: \w'\o'ab''
+.br
+undefined: \w'\G'
+.br
+zero-width: \w'\fB\&\fP'
+.br
+skipchar: \w'a\zb\z\(buc'
 .Ss Argument delimiters
 unsupported \er: \w\rM\ru
 .br
diff --git a/regress/roff/esc/w.out_ascii b/regress/roff/esc/w.out_ascii
index 7ed32ece..a8fdce54 100644
--- a/regress/roff/esc/w.out_ascii
+++ b/regress/roff/esc/w.out_ascii
@@ -8,6 +8,13 @@ DDEESSCCRRIIPPTTIIOONN
      character: 24
      blank: 24
      text: 96
+     special: 24
+     numbered: 24
+     Unicode: 24
+     overstrike: 24
+     undefined: 24
+     zero-width: 0
+     skipchar: 48
 
    AArrgguummeenntt ddeelliimmiitteerrss
      unsupported \r: 24u
@@ -27,4 +34,4 @@ DDEESSCCRRIIPPTTIIOONN
      overstrike: 24u
      unterminated: 72
 
-OpenBSD                          June 8, 2022                          OpenBSD
+OpenBSD                        October 23, 2023                        OpenBSD
diff --git a/regress/roff/esc/w.out_lint b/regress/roff/esc/w.out_lint
index fd2a0948..9c6417c5 100644
--- a/regress/roff/esc/w.out_lint
+++ b/regress/roff/esc/w.out_lint
@@ -1,4 +1,5 @@
-mandoc: w.in:17:20: UNSUPP: unsupported escape sequence: \r
-mandoc: w.in:17:23: UNSUPP: unsupported escape sequence: \r
-mandoc: w.in:23:16: WARNING: undefined escape, printing literally: \G
-mandoc: w.in:51:15: ERROR: incomplete escape sequence: \w'foo
+mandoc: w.in:25:15: WARNING: undefined escape, printing literally: \G
+mandoc: w.in:31:20: UNSUPP: unsupported escape sequence: \r
+mandoc: w.in:31:23: UNSUPP: unsupported escape sequence: \r
+mandoc: w.in:37:16: WARNING: undefined escape, printing literally: \G
+mandoc: w.in:65:15: ERROR: incomplete escape sequence: \w'foo
diff --git a/roff.7 b/roff.7
index 597d4aa7..22459a9c 100644
--- a/roff.7
+++ b/roff.7
@@ -1,6 +1,6 @@
 .\" $Id$
 .\"
-.\" Copyright (c) 2010-2019, 2022 Ingo Schwarze <schwarze@openbsd.org>
+.\" Copyright (c) 2010-2019, 2022-2023 Ingo Schwarze <schwarze@openbsd.org>
 .\" Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
 .\"
 .\" Permission to use, copy, modify, and distribute this software for any
@@ -2224,7 +2224,8 @@ The
 .Xr mandoc 1
 implementation assumes that after expansion of user-defined strings, the
 .Ar string
-only contains normal characters, no escape sequences, and that each
+only contains normal characters, characters expressed as escape sequences,
+and zero-width escape sequences, and that each
 character has a width of 24 basic units.
 .It Ic \eX\(aq Ns Ar string Ns Ic \(aq
 Output
diff --git a/roff.c b/roff.c
index 0005b999..53f9df46 100644
--- a/roff.c
+++ b/roff.c
@@ -1,6 +1,6 @@
 /* $Id$ */
 /*
- * Copyright (c) 2010-2015, 2017-2022 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2010-2015, 2017-2023 Ingo Schwarze <schwarze@openbsd.org>
  * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  *
  * Permission to use, copy, modify, and distribute this software for any
@@ -1362,6 +1362,7 @@ roff_expand(struct roff *r, struct buf *buf, int ln, int pos, char ec)
 	const char	*res;		/* the string to be pasted */
 	const char	*src;		/* source for copying */
 	char		*dst;		/* destination for copying */
+	enum mandoc_esc	 subtype;	/* return value from roff_escape */
 	int		 iesc;		/* index of leading escape char */
 	int		 inam;		/* index of the escape name */
 	int		 iarg;		/* index beginning the argument */
@@ -1551,8 +1552,34 @@ roff_expand(struct roff *r, struct buf *buf, int ln, int pos, char ec)
 			res = ubuf;
 			break;
 		case 'w':
-			(void)snprintf(ubuf, sizeof(ubuf),
-			    "%d", (iendarg - iarg) * 24);
+			rsz = 0;
+			subtype = ESCAPE_UNDEF;
+			while (iarg < iendarg) {
+				asz = subtype == ESCAPE_SKIPCHAR ? 0 : 1;
+				if (buf->buf[iarg] != '\\') {
+					rsz += asz;
+					iarg++;
+					continue;
+				}
+				switch ((subtype = roff_escape(buf->buf, 0,
+				    iarg, NULL, NULL, NULL, NULL, &iarg))) {
+				case ESCAPE_SPECIAL:
+				case ESCAPE_NUMBERED:
+				case ESCAPE_UNICODE:
+				case ESCAPE_OVERSTRIKE:
+				case ESCAPE_UNDEF:
+					break;
+				case ESCAPE_DEVICE:
+					asz *= 8;
+					break;
+				case ESCAPE_EXPAND:
+					abort();
+				default:
+					continue;
+				}
+				rsz += asz;
+			}
+			(void)snprintf(ubuf, sizeof(ubuf), "%d", rsz * 24);
 			res = ubuf;
 			break;
 		default: