Ctype patch for review

Andrey Chernov <[email protected]>
Newsgroups gmane.os.freebsd.devel.internationalization
Message-ID <[email protected]>
The problem is: currently our single byte ctype functions are broken for 
wide characters locales in the argument range >= 0x80 - they may return 
false positives.

For example, for UTF-8 locale we currently have:
iswspace(0xA0)==1 and isspace(0xA0)==1
(because iswspace() and isspace() are the same code)
but must have
isspace(0xA0)==0
(because there is no such character and all others in the range 
0x80..0xff for the wide locales, they keep ASCII only in the single byte 
range because our internal wchar_t representation is UCS-4).

Attached patch address this issue and also fix iswascii()
(currently iswascii() is broken for arguments > 0xFF).
This patch is 100% binary compatible with old binaries, their (broken) 
behaviour is not changed.

I want to hear some comments.

-- 
http://ache.pp.ru/

_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-i18n
To unsubscribe, send any mail to "[email protected]"
ctype.patch (text/plain, 9.3 KB)
--- _ctype.h.old	2007-09-16 21:13:59.000000000 +0400
+++ _ctype.h	2007-09-16 23:00:38.000000000 +0400
@@ -63,6 +63,7 @@
 #define	_CTYPE_I	0x00080000L		/* Ideogram */
 #define	_CTYPE_T	0x00100000L		/* Special */
 #define	_CTYPE_Q	0x00200000L		/* Phonogram */
+#define	_CTYPE_WID	0x10000000L		/* wide character function */
 #define	_CTYPE_SW0	0x20000000L		/* 0 width character */
 #define	_CTYPE_SW1	0x40000000L		/* 1 width character */
 #define	_CTYPE_SW2	0x80000000L		/* 2 width character */
@@ -87,6 +88,8 @@
 #define	__inline
 #endif
 
+extern int __mb_cur_max;
+
 /*
  * Use inline functions if we are allowed to and the compiler supports them.
  */
@@ -98,8 +101,11 @@
 static __inline int
 __maskrune(__ct_rune_t _c, unsigned long _f)
 {
-	return ((_c < 0 || _c >= _CACHED_RUNES) ? ___runetype(_c) :
+	return __mb_cur_max > 1 && !(_f & _CTYPE_WID) && (_c >= 0x80) ? 0 :
+	       ((_c < 0 || _c >= _CACHED_RUNES) ? ___runetype(_c) :
 		_CurrentRuneLocale->__runetype[_c]) & _f;
+		/* We never set _CTYPE_WID in the locale data, */
+		/* so can skip ... & (_f & ~_CTYPE_WID).       */
 }
 
 static __inline int
@@ -111,8 +117,11 @@
 static __inline int
 __isctype(__ct_rune_t _c, unsigned long _f)
 {
-	return (_c < 0 || _c >= _CACHED_RUNES) ? 0 :
+	return  __mb_cur_max > 1 && !(_f & _CTYPE_WID) && (_c >= 0x80) ? 0 :
+	       (_c < 0 || _c >= _CACHED_RUNES) ? 0 :
 	       !!(_DefaultRuneLocale.__runetype[_c] & _f);
+		  /* We never set _CTYPE_WID in the locale data, */
+		  /* so can skip ... & (_f & ~_CTYPE_WID).	 */
 }
 
 static __inline __ct_rune_t
@@ -129,6 +138,22 @@
 	       _CurrentRuneLocale->__maplower[_c];
 }
 
+static __inline __ct_rune_t
+__tosupper(__ct_rune_t _c)
+{
+	return  __mb_cur_max > 1 && (_c >= 0x80) ? _c :
+	       (_c < 0 || _c >= _CACHED_RUNES) ? ___toupper(_c) :
+	       _CurrentRuneLocale->__mapupper[_c];
+}
+
+static __inline __ct_rune_t
+__toslower(__ct_rune_t _c)
+{
+	return  __mb_cur_max > 1 && (_c >= 0x80) ? _c :
+	       (_c < 0 || _c >= _CACHED_RUNES) ? ___tolower(_c) :
+	       _CurrentRuneLocale->__maplower[_c];
+}
+
 static __inline int
 __wcwidth(__ct_rune_t _c)
 {
@@ -150,6 +175,8 @@
 int		__isctype(__ct_rune_t, unsigned long);
 __ct_rune_t	__toupper(__ct_rune_t);
 __ct_rune_t	__tolower(__ct_rune_t);
+__ct_rune_t	__tosupper(__ct_rune_t);
+__ct_rune_t	__toslower(__ct_rune_t);
 int		__wcwidth(__ct_rune_t);
 __END_DECLS
 #endif /* using inlines */
--- ctype.h.old	2007-09-16 22:03:55.000000000 +0400
+++ ctype.h	2007-09-16 22:56:10.000000000 +0400
@@ -97,8 +97,8 @@
 #define	isspace(c)	__istype((c), _CTYPE_S)
 #define	isupper(c)	__istype((c), _CTYPE_U)
 #define	isxdigit(c)	__isctype((c), _CTYPE_X) /* ANSI -- locale independent */
-#define	tolower(c)	__tolower(c)
-#define	toupper(c)	__toupper(c)
+#define	tolower(c)	__toslower(c)
+#define	toupper(c)	__tosupper(c)
 
 #if __XSI_VISIBLE
 /*
@@ -112,8 +112,8 @@
  *
  * XXX isascii() and toascii() should similarly be undocumented.
  */
-#define	_tolower(c)	__tolower(c)
-#define	_toupper(c)	__toupper(c)
+#define	_tolower(c)	__toslower(c)
+#define	_toupper(c)	__tosupper(c)
 #define	isascii(c)	(((c) & ~0x7F) == 0)
 #define	toascii(c)	((c) & 0x7F)
 #endif
@@ -128,7 +128,7 @@
 #define	isideogram(c)	__istype((c), _CTYPE_I)
 #define	isnumber(c)	__istype((c), _CTYPE_D)
 #define	isphonogram(c)	__istype((c), _CTYPE_Q)
-#define	isrune(c)	__istype((c), 0xFFFFFF00L)
+#define	isrune(c)	__istype((c), 0xFFFFFF00L & ~_CTYPE_WID)
 #define	isspecial(c)	__istype((c), _CTYPE_T)
 #endif
 
--- wctype.h.old	2007-09-16 21:59:37.000000000 +0400
+++ wctype.h	2007-09-16 22:56:44.000000000 +0400
@@ -89,30 +89,30 @@
 #endif
 __END_DECLS
 
-#define	iswalnum(wc)		__istype((wc), _CTYPE_A|_CTYPE_D)
-#define	iswalpha(wc)		__istype((wc), _CTYPE_A)
-#define	iswblank(wc)		__istype((wc), _CTYPE_B)
-#define	iswcntrl(wc)		__istype((wc), _CTYPE_C)
-#define	iswctype(wc, charclass)	__istype((wc), (charclass))
-#define	iswdigit(wc)		__isctype((wc), _CTYPE_D)
-#define	iswgraph(wc)		__istype((wc), _CTYPE_G)
-#define	iswlower(wc)		__istype((wc), _CTYPE_L)
-#define	iswprint(wc)		__istype((wc), _CTYPE_R)
-#define	iswpunct(wc)		__istype((wc), _CTYPE_P)
-#define	iswspace(wc)		__istype((wc), _CTYPE_S)
-#define	iswupper(wc)		__istype((wc), _CTYPE_U)
-#define	iswxdigit(wc)		__isctype((wc), _CTYPE_X)
+#define	iswalnum(wc)		__istype((wc), _CTYPE_A|_CTYPE_D|_CTYPE_WID)
+#define	iswalpha(wc)		__istype((wc), _CTYPE_A|_CTYPE_WID)
+#define	iswblank(wc)		__istype((wc), _CTYPE_B|_CTYPE_WID)
+#define	iswcntrl(wc)		__istype((wc), _CTYPE_C|_CTYPE_WID)
+#define	iswctype(wc, charclass)	__istype((wc), (charclass)|_CTYPE_WID)
+#define	iswdigit(wc)		__isctype((wc), _CTYPE_D|_CTYPE_WID)
+#define	iswgraph(wc)		__istype((wc), _CTYPE_G|_CTYPE_WID)
+#define	iswlower(wc)		__istype((wc), _CTYPE_L|_CTYPE_WID)
+#define	iswprint(wc)		__istype((wc), _CTYPE_R|_CTYPE_WID)
+#define	iswpunct(wc)		__istype((wc), _CTYPE_P|_CTYPE_WID)
+#define	iswspace(wc)		__istype((wc), _CTYPE_S|_CTYPE_WID)
+#define	iswupper(wc)		__istype((wc), _CTYPE_U|_CTYPE_WID)
+#define	iswxdigit(wc)		__isctype((wc), _CTYPE_X|_CTYPE_WID)
 #define	towlower(wc)		__tolower(wc)
 #define	towupper(wc)		__toupper(wc)
 
 #if __BSD_VISIBLE
-#define	iswascii(wc)		(((wc) & ~0x7F) == 0)
-#define	iswhexnumber(wc)	__istype((wc), _CTYPE_X)
-#define	iswideogram(wc)		__istype((wc), _CTYPE_I)
-#define	iswnumber(wc)		__istype((wc), _CTYPE_D)
-#define	iswphonogram(wc)	__istype((wc), _CTYPE_Q)
-#define	iswrune(wc)		__istype((wc), 0xFFFFFF00L)
-#define	iswspecial(wc)		__istype((wc), _CTYPE_T)
+#define	iswascii(wc)		((wc) < 0x80)
+#define	iswhexnumber(wc)	__istype((wc), _CTYPE_X|_CTYPE_WID)
+#define	iswideogram(wc)		__istype((wc), _CTYPE_I|_CTYPE_WID)
+#define	iswnumber(wc)		__istype((wc), _CTYPE_D|_CTYPE_WID)
+#define	iswphonogram(wc)	__istype((wc), _CTYPE_Q|_CTYPE_WID)
+#define	iswrune(wc)		__istype((wc), 0xFFFFFF00L) /* already have _CTYPE_WID */
+#define	iswspecial(wc)		__istype((wc), _CTYPE_T|_CTYPE_WID)
 #endif
 
 #endif		/* _WCTYPE_H_ */
--- isctype.c.old	2007-09-16 22:31:26.000000000 +0400
+++ isctype.c	2007-09-16 22:37:54.000000000 +0400
@@ -168,7 +168,7 @@
 isrune(c)
 	int c;
 {
-	return (__istype(c, 0xFFFFFF00L));
+	return (__istype(c, 0xFFFFFF00L & ~_CTYPE_WID));
 }
 
 #undef isspace
@@ -216,7 +216,7 @@
 tolower(c)
 	int c;
 {
-        return (__tolower(c));
+	return (__toslower(c));
 }
 
 #undef toupper
@@ -224,6 +224,6 @@
 toupper(c)
 	int c;
 {
-        return (__toupper(c));
+	return (__tosupper(c));
 }
 
--- iswctype.c.old	2007-09-16 22:31:30.000000000 +0400
+++ iswctype.c	2007-09-16 22:41:39.000000000 +0400
@@ -45,7 +45,7 @@
 iswalnum(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_A|_CTYPE_D));
+	return (__istype(wc, _CTYPE_A|_CTYPE_D|_CTYPE_WID));
 }
 
 #undef iswalpha
@@ -53,7 +53,7 @@
 iswalpha(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_A));
+	return (__istype(wc, _CTYPE_A|_CTYPE_WID)));
 }
 
 #undef iswascii
@@ -61,7 +61,7 @@
 iswascii(wc)
 	wint_t wc;
 {
-	return ((wc & ~0x7F) == 0);
+	return (wc < 0x80);
 }
 
 #undef iswblank
@@ -69,7 +69,7 @@
 iswblank(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_B));
+	return (__istype(wc, _CTYPE_B|_CTYPE_WID)));
 }
 
 #undef iswcntrl
@@ -77,7 +77,7 @@
 iswcntrl(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_C));
+	return (__istype(wc, _CTYPE_C|_CTYPE_WID)));
 }
 
 #undef iswdigit
@@ -85,7 +85,7 @@
 iswdigit(wc)
 	wint_t wc;
 {
-	return (__isctype(wc, _CTYPE_D));
+	return (__isctype(wc, _CTYPE_D|_CTYPE_WID)));
 }
 
 #undef iswgraph
@@ -93,7 +93,7 @@
 iswgraph(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_G));
+	return (__istype(wc, _CTYPE_G|_CTYPE_WID)));
 }
 
 #undef iswhexnumber 
@@ -101,7 +101,7 @@
 iswhexnumber(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_X));
+	return (__istype(wc, _CTYPE_X|_CTYPE_WID)));
 }
 
 #undef iswideogram
@@ -109,7 +109,7 @@
 iswideogram(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_I));
+	return (__istype(wc, _CTYPE_I|_CTYPE_WID)));
 }
 
 #undef iswlower
@@ -117,7 +117,7 @@
 iswlower(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_L));
+	return (__istype(wc, _CTYPE_L|_CTYPE_WID)));
 }
 
 #undef iswnumber
@@ -125,7 +125,7 @@
 iswnumber(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_D));
+	return (__istype(wc, _CTYPE_D|_CTYPE_WID)));
 }
 
 #undef iswphonogram	
@@ -133,7 +133,7 @@
 iswphonogram(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_Q));
+	return (__istype(wc, _CTYPE_Q|_CTYPE_WID)));
 }
 
 #undef iswprint
@@ -141,7 +141,7 @@
 iswprint(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_R));
+	return (__istype(wc, _CTYPE_R|_CTYPE_WID)));
 }
 
 #undef iswpunct
@@ -149,7 +149,7 @@
 iswpunct(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_P));
+	return (__istype(wc, _CTYPE_P|_CTYPE_WID)));
 }
 
 #undef iswrune
@@ -157,7 +157,7 @@
 iswrune(wc)
 	wint_t wc;
 {
-	return (__istype(wc, 0xFFFFFF00L));
+	return (__istype(wc, 0xFFFFFF00L)); /* already have _CTYPE_WID */
 }
 
 #undef iswspace
@@ -165,7 +165,7 @@
 iswspace(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_S));
+	return (__istype(wc, _CTYPE_S|_CTYPE_WID)));
 }
 
 #undef iswspecial
@@ -173,7 +173,7 @@
 iswspecial(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_T));
+	return (__istype(wc, _CTYPE_T|_CTYPE_WID)));
 }
 
 #undef iswupper
@@ -181,7 +181,7 @@
 iswupper(wc)
 	wint_t wc;
 {
-	return (__istype(wc, _CTYPE_U));
+	return (__istype(wc, _CTYPE_U|_CTYPE_WID)));
 }
 
 #undef iswxdigit
@@ -189,7 +189,7 @@
 iswxdigit(wc)
 	wint_t wc;
 {
-	return (__isctype(wc, _CTYPE_X));
+	return (__isctype(wc, _CTYPE_X|_CTYPE_WID)));
 }
 
 #undef towlower
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.