===================================================================
RCS file: /cvs/mandoc/chars.c,v
retrieving revision 1.42
retrieving revision 1.45
diff -u -p -r1.42 -r1.45
--- mandoc/chars.c	2011/05/15 15:47:46	1.42
+++ mandoc/chars.c	2011/05/24 21:24:16	1.45
@@ -1,4 +1,4 @@
-/*	$Id: chars.c,v 1.42 2011/05/15 15:47:46 kristaps Exp $ */
+/*	$Id: chars.c,v 1.45 2011/05/24 21:24:16 kristaps Exp $ */
 /*
  * Copyright (c) 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -26,6 +26,7 @@
 #include <string.h>
 
 #include "mandoc.h"
+#include "libmandoc.h"
 
 #define	PRINT_HI	 126
 #define	PRINT_LO	 32
@@ -35,20 +36,12 @@ struct	ln {
 	const char	 *code;
 	const char	 *ascii;
 	int		  unicode;
-	int		  type;
-#define	CHARS_CHAR	 (1 << 0)
-#define	CHARS_STRING	 (1 << 1)
-#define CHARS_BOTH	 (CHARS_CHAR | CHARS_STRING)
 };
 
-#define	LINES_MAX	  353
+#define	LINES_MAX	  325
 
 #define CHAR(in, ch, code) \
-	{ NULL, (in), (ch), (code), CHARS_CHAR },
-#define STRING(in, ch, code) \
-	{ NULL, (in), (ch), (code), CHARS_STRING },
-#define BOTH(in, ch, code) \
-	{ NULL, (in), (ch), (code), CHARS_BOTH },
+	{ NULL, (in), (ch), (code) },
 
 #define	CHAR_TBL_START	  static struct ln lines[LINES_MAX] = {
 #define	CHAR_TBL_END	  };
@@ -59,9 +52,8 @@ struct	mchars {
 	struct ln	**htab;
 };
 
-static	inline int	  match(const struct ln *,
-				const char *, size_t, int);
-static	const struct ln	 *find(struct mchars *, const char *, size_t, int);
+static	inline int	  match(const struct ln *, const char *, size_t);
+static	const struct ln	 *find(struct mchars *, const char *, size_t);
 
 void
 mchars_free(struct mchars *arg)
@@ -115,7 +107,7 @@ mchars_spec2cp(struct mchars *arg, const char *p, size
 {
 	const struct ln	*ln;
 
-	ln = find(arg, p, sz, CHARS_CHAR);
+	ln = find(arg, p, sz);
 	if (NULL == ln)
 		return(-1);
 	return(ln->unicode);
@@ -130,14 +122,14 @@ mchars_res2cp(struct mchars *arg, const char *p, size_
 {
 	const struct ln	*ln;
 
-	ln = find(arg, p, sz, CHARS_STRING);
+	ln = find(arg, p, sz);
 	if (NULL == ln)
 		return(-1);
 	return(ln->unicode);
 }
 
 /*
- * Numbered character to literal character.
+ * Numbered character string to ASCII codepoint.
  * This can only be a printable character (i.e., alnum, punct, space) so
  * prevent the character from ruining our state (backspace, newline, and
  * so on).
@@ -148,14 +140,26 @@ mchars_num2char(const char *p, size_t sz)
 {
 	int		  i;
 
-	if (sz > 3)
+	if ((i = mandoc_strntou(p, sz, 10)) < 0)
 		return('\0');
-
-	i = atoi(p);
-	/* LINTED */
 	return(isprint(i) ? i : '\0');
 }
 
+/*
+ * Hex character string to Unicode codepoint.
+ * If the character is illegal, returns '\0'.
+ */
+int
+mchars_num2uc(const char *p, size_t sz)
+{
+	int               i;
+
+	if ((i = mandoc_strntou(p, sz, 16)) < 0)
+		return('\0');
+	/* FIXME: make sure we're not in a bogus range. */
+	return(i > 0x80 && i <= 0x10FFFF ? i : '\0');
+}
+
 /* 
  * Special character to string array.
  */
@@ -164,7 +168,7 @@ mchars_spec2str(struct mchars *arg, const char *p, siz
 {
 	const struct ln	*ln;
 
-	ln = find(arg, p, sz, CHARS_CHAR);
+	ln = find(arg, p, sz);
 	if (NULL == ln)
 		return(NULL);
 
@@ -180,7 +184,7 @@ mchars_res2str(struct mchars *arg, const char *p, size
 {
 	const struct ln	*ln;
 
-	ln = find(arg, p, sz, CHARS_STRING);
+	ln = find(arg, p, sz);
 	if (NULL == ln)
 		return(NULL);
 
@@ -189,7 +193,7 @@ mchars_res2str(struct mchars *arg, const char *p, size
 }
 
 static const struct ln *
-find(struct mchars *tab, const char *p, size_t sz, int type)
+find(struct mchars *tab, const char *p, size_t sz)
 {
 	struct ln	 *pp, *prev;
 	struct ln	**htab;
@@ -215,7 +219,7 @@ find(struct mchars *tab, const char *p, size_t sz, int
 		return(NULL);
 
 	for (prev = NULL; pp; pp = pp->next) {
-		if ( ! match(pp, p, sz, type)) {
+		if ( ! match(pp, p, sz)) {
 			prev = pp;
 			continue;
 		}
@@ -233,11 +237,9 @@ find(struct mchars *tab, const char *p, size_t sz, int
 }
 
 static inline int
-match(const struct ln *ln, const char *p, size_t sz, int type)
+match(const struct ln *ln, const char *p, size_t sz)
 {
 
-	if ( ! (ln->type & type))
-		return(0);
 	if (strncmp(ln->code, p, sz))
 		return(0);
 	return('\0' == ln->code[(int)sz]);