Several improvements to escape sequence handling.

* Add the missing special character \_ (underscore).
* Partial implementations of \a (leader character)
and \E (uninterpreted escape character).
* Parse and ignore \r (reverse line feed).
* Add a WARNING message about undefined escape sequences.
* Add an UNSUPP message about unsupported escape sequences.
* Mark \! and \? (transparent throughput)
and \O (suppress output) as unsupported.
* Treat the various variants of zero-width spaces as one-byte escape
sequences rather than as special characters, to avoid defining bogus
forms with square brackets.
* For special characters with one-byte names, do not define bogus
forms with square brackets, except for \[-], which is valid.
* In the form with square brackets, undefined special characters do not
fall back to printing the name verbatim, not even for one-byte names.
* Starting a special character name with a blank is an error.
* Undefined escape sequences never abort formatting of the input
string, not even in HTML output mode.
* Document the newly handled escapes, and a few that were missing.
* Regression tests for most of the above.

1 files changed, 6 insertions, 16 deletions

diff --git a/chars.c b/chars.c
index 3ef273c1..24166dbd 100644
--- a/chars.c
+++ b/chars.c
@@ -1,7 +1,7 @@
-/*	$Id: chars.c,v 1.77 2018/12/14 01:18:25 schwarze Exp $ */
+/*	$Id: chars.c,v 1.78 2018/12/15 19:30:26 schwarze Exp $ */
 /*
  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2011, 2014, 2015, 2017 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2011,2014,2015,2017,2018 Ingo Schwarze <schwarze@openbsd.org>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * purpose with or without fee is hereby granted, provided that the above
@@ -48,21 +48,13 @@ static struct ln lines[] = {
 	{ " ",			ascii_nbrsp,	0x00a0	},
 	{ "~",			ascii_nbrsp,	0x00a0	},
 	{ "0",			" ",		0x2002	},
-	{ "|",			"",		0	},
-	{ "^",			"",		0	},
-	{ "&",			"",		0	},
-	{ ")",			"",		0	},
-	{ "%",			"",		0	},
 	{ ":",			ascii_break,	0	},
-	/* XXX The following three do not really belong here. */
-	{ "t",			"",		0	},
-	{ "c",			"",		0	},
-	{ "}",			"",		0	},
 
 	/* Lines. */
 	{ "ba",			"|",		0x007c	},
 	{ "br",			"|",		0x2502	},
 	{ "ul",			"_",		0x005f	},
+	{ "_",			"_",		0x005f	},
 	{ "ru",			"_",		0x005f	},
 	{ "rn",			"-",		0x203e	},
 	{ "bb",			"|",		0x00a6	},
@@ -465,7 +457,7 @@ mchars_spec2cp(const char *p, size_t sz)
 
 	end = p + sz;
 	ln = ohash_find(&mchars, ohash_qlookupi(&mchars, p, &end));
-	return ln != NULL ? ln->unicode : sz == 1 ? (unsigned char)*p : -1;
+	return ln != NULL ? ln->unicode : -1;
 }
 
 int
@@ -495,10 +487,8 @@ mchars_spec2str(const char *p, size_t sz, size_t *rsz)
 
 	end = p + sz;
 	ln = ohash_find(&mchars, ohash_qlookupi(&mchars, p, &end));
-	if (ln == NULL) {
-		*rsz = 1;
-		return sz == 1 ? p : NULL;
-	}
+	if (ln == NULL)
+		return NULL;
 
 	*rsz = strlen(ln->ascii);
 	return ln->ascii;