handle the non-portable GNU-style \[charNN], \[charNNN] character

escape sequences, used for example in the groff_char(7) manual page

3 files changed, 31 insertions, 10 deletions

diff --git a/TODO b/TODO
index 4d0c2190..c986d3c2 100644
--- a/TODO
+++ b/TODO
@@ -40,9 +40,10 @@ are mere guesses, and some may be wrong.
 
 - \*(.T prints the device being used,
   see groff_char(7) for an example
-
-- \[charNN], \[charNNN] prints a single-byte codepoint
-  see groff_char(7) for examples
+  This is slightly hard because -Tlocale only decides to use ascii or
+  utf8 when initializing the formatter, so the information is not
+  yet available to the preprocessor at the parsing stage.
+  loc **  exist **  algo *  size *  imp *
 
 - .ad (adjust margins)
   .ad l -- adjust left margin only (flush left)
diff --git a/mandoc.c b/mandoc.c
index fd0ad7e7..7e785312 100644
--- a/mandoc.c
+++ b/mandoc.c
@@ -41,7 +41,7 @@ enum mandoc_esc
 mandoc_escape(const char **end, const char **start, int *sz)
 {
 	const char	*local_start;
-	int		 local_sz;
+	int		 local_sz, c, i;
 	char		 term;
 	enum mandoc_esc	 gly;
 
@@ -330,8 +330,26 @@ mandoc_escape(const char **end, const char **start, int *sz)
 		}
 		break;
 	case ESCAPE_SPECIAL:
-		if (1 == *sz && 'c' == **start)
-			gly = ESCAPE_NOSPACE;
+		if (**start == 'c') {
+			if (*sz == 1) {
+				gly = ESCAPE_NOSPACE;
+				break;
+			}
+			if (*sz < 6 || *sz > 7 ||
+			    strncmp(*start, "char", 4) != 0 ||
+			    (int)strspn(*start + 4, "0123456789") + 4 < *sz)
+				break;
+			c = 0;
+			for (i = 4; i < *sz; i++)
+				c = 10 * c + ((*start)[i] - '0');
+			if (c < 0x21 || (c > 0x7e && c < 0xa0) || c > 0xff)
+				break;
+			*start += 4;
+			*sz -= 4;
+			gly = ESCAPE_NUMBERED;
+			break;
+		}
+
 		/*
 		 * Unicode escapes are defined in groff as \[u0000]
 		 * to \[u10FFFF], where the contained value must be
diff --git a/mandoc_char.7 b/mandoc_char.7
index c919f783..84100951 100644
--- a/mandoc_char.7
+++ b/mandoc_char.7
@@ -761,14 +761,16 @@ For backward compatibility with existing manuals,
 .Xr mandoc 1
 also supports the
 .Pp
-.Dl \eN\(aq Ns Ar number Ns \(aq
+.Dl \eN\(aq Ns Ar number Ns \(aq and \e[ Ns Cm char Ns Ar number ]
 .Pp
-escape sequence, inserting the character
+escape sequences, inserting the character
 .Ar number
 from the current character set into the output.
 Of course, this is inherently non-portable and is already marked
-as deprecated in the Heirloom roff manual.
-For example, do not use \eN\(aq34\(aq, use \e(dq, or even the plain
+as deprecated in the Heirloom roff manual;
+on top of that, the second form is a GNU extension.
+For example, do not use \eN\(aq34\(aq or \e[char34], use \e(dq,
+or even the plain
 .Sq \(dq
 character where possible.
 .Sh COMPATIBILITY