Changeset 2511

Timestamp:

Jan 19, 2006, 3:44:53 AM (19 years ago)

Author:

bird

Message:

Fixes #30: Fixed two locale bugs. First, setlocale called a locale
dependent function for getting ctype flags. Second, all the
is<>() macros / functions was busted for non-ascii chars (-1..-128).

Location:

branches/libc-0.6/src/emx

Files:

• ChangeLog.LIBC (modified) (1 diff)
• include/InnoTekLIBC/locale.h (modified) (1 diff)
• include/_ctype.h (modified) (4 diffs)
• src/lib/locale/locale_ctype.c (modified) (4 diffs)
• src/lib/locale/setlocale.c (modified) (2 diffs)

branches/libc-0.6/src/emx/ChangeLog.LIBC ¶

@@ -2 +2 @@
 
 TODO: open replace on RAMFS fails with error 32!
+
+2006-01-18: knut st. osmundsen <bird-gccos2-spam@anduin.net>
+    - libc:
+        o #30: Fixed two locale bugs. First, setlocale called a locale
+          dependent function for getting ctype flags. Second, all the
+          is<>() macros / functions was busted for non-ascii chars (-1..-128).
 
 2005-12-14: knut st. osmundsen <bird-gccos2-spam@anduin.net>

branches/libc-0.6/src/emx/include/InnoTekLIBC/locale.h ¶

@@ -316 +316 @@
 /** @} */
 
+#include <ctype.h>
+
+/**
+ * Convert the type info we get from the unicode lib to libc talk.
+ * ASSUMES that none of the locals differs from the unicode spec
+ *
+ * @returns libc ctype flags.
+ * @param   pUniType    The unicode type info to translate.
+ * @param   wc          The unicode code point.
+ */
+static inline unsigned ___wctype_uni(const UNICTYPE *pUniType, wchar_t wc)
+{
+    unsigned    ufType = 0;
+    /* ASSUMES CT_* << 8 == __* ! */
+    ufType = ((unsigned)pUniType->itype << 8)
+           & (__CT_UPPER  | __CT_LOWER  | __CT_DIGIT | __CT_SPACE |
+              __CT_PUNCT  | __CT_CNTRL  | __CT_BLANK | __CT_XDIGIT |
+              __CT_ALPHA  | __CT_ALNUM  | __CT_GRAPH | __CT_PRINT |
+              __CT_NUMBER | __CT_SYMBOL | __CT_ASCII);
+    if (pUniType->extend & C3_IDEOGRAPH)
+        ufType |= __CT_IDEOGRAM;
+    if (ufType & (__CT_XDIGIT | __CT_DIGIT))
+    {
+        if (     (unsigned)wc - 0x30U <= (0x39 - 0x30))
+            ufType |= (unsigned)wc - 0x30;
+        else if ((unsigned)wc - 0x41U <= (0x46 - 0x41))
+            ufType |= (unsigned)wc - 0x41 + 0xa;
+        else
+        {
+            unsigned uVal = UniQueryNumericValue(wc);
+            if (!(uVal & ~0xffU))
+                ufType |= uVal;
+        }
+    }
+    ufType |= (pUniType->bidi & 0xf << 24);
+
+    /** @todo screen width. */
+    return ufType;
+}
+
 __END_DECLS
 

branches/libc-0.6/src/emx/include/_ctype.h ¶

@@ -166 +166 @@
 static __inline__ unsigned __ctype(__ct_rune_t __ch, unsigned __f)
 {
-    return !(__ch & ~0xffU)
-        ? __libc_GLocaleCtype.aufType[__ch] & __f
+    return __ch < 256 && __ch >= -128
+        ? __libc_GLocaleCtype.aufType[__ch & 0xff] & __f
         : 0;
 }
@@ -178 +178 @@
 static __inline__ int __isctype(__ct_rune_t __ch, unsigned __f)
 {
-    return !(__ch & ~0xffU)
-        ? !!(__libc_GLocaleCtypeDefault.aufType[__ch] & __f)
+    return __ch <= 255 && __ch >= -128
+        ? !!(__libc_GLocaleCtypeDefault.aufType[__ch & 0xff] & __f)
         : 0;
 }
@@ -185 +185 @@
 static __inline__ __ct_rune_t __toupper(__ct_rune_t __ch)
 {
-    return !(__ch & ~0xffU)
-        ? __libc_GLocaleCtype.auchUpper[__ch]
+    return __ch <= 255 && __ch >= -128
+        ? __libc_GLocaleCtype.auchUpper[__ch & 0xff]
         : (__ch);
 }
@@ -192 +192 @@
 static __inline__ __ct_rune_t __tolower(__ct_rune_t __ch)
 {
-    return !(__ch & ~0xffU)
-        ? __libc_GLocaleCtype.auchLower[__ch]
+    return __ch <= 255 && __ch >= -128
+        ? __libc_GLocaleCtype.auchLower[__ch & 0xff]
         : __ch;
 }

branches/libc-0.6/src/emx/src/lib/locale/locale_ctype.c ¶

@@ -220 +220 @@
 unsigned __ctype(__ct_rune_t __ch, unsigned __f)
 {
-    return !((__ch) & ~0xffU)
-        ? __libc_GLocaleCtype.aufType[(__ch)/* & 0xff*/] & (__f)
+    return __ch <= 255 && __ch >= -128
+        ? __libc_GLocaleCtype.aufType[__ch & 0xff] & __f
         : 0;
 }
@@ -227 +227 @@
 int __istype(__ct_rune_t __ch, unsigned __f)
 {
-    return !!__ctype((__ch), (__f));
+    return !!__ctype(__ch, __f);
 }
 
 __ct_rune_t __toupper(__ct_rune_t __ch)
 {
-    return !((__ch) & ~0xffU)
-        ? __libc_GLocaleCtype.auchUpper[(__ch)/* & 0xff*/]
-        : (__ch);
+    return __ch <= 255 && __ch >= -128
+        ? __libc_GLocaleCtype.auchUpper[__ch & 0xff]
+        : __ch;
 }
 
 __ct_rune_t __tolower(__ct_rune_t __ch)
 {
-    return !((__ch) & ~0xffU)
-        ? __libc_GLocaleCtype.auchLower[(__ch)/* & 0xff*/]
-        : (__ch);
+    return __ch <= 255 && __ch >= -128
+        ? __libc_GLocaleCtype.auchLower[__ch & 0xff]
+        : __ch;
 }
 
@@ -397 +397 @@
 }
 
-
 /**
  * Get unicode type.
@@ -404 +403 @@
 unsigned ___wctype(__wchar_t wc)
 {
-    unsigned    ufType = 0;
     if (    __libc_GLocaleWCtype.uMask != ~0x7f
         ||  wc <= 127)
     {
-        UNICTYPE   *pUniType = UniQueryCharType(wc);
+        UNICTYPE *pUniType = UniQueryCharType(wc);
         if (pUniType)
-        {
-            /* ASSUMES CT_* << 8 == __* ! */
-            ufType = ((unsigned)pUniType->itype << 8)
-                   & (__CT_UPPER  | __CT_LOWER  | __CT_DIGIT | __CT_SPACE |
-                      __CT_PUNCT  | __CT_CNTRL  | __CT_BLANK | __CT_XDIGIT |
-                      __CT_ALPHA  | __CT_ALNUM  | __CT_GRAPH | __CT_PRINT |
-                      __CT_NUMBER | __CT_SYMBOL | __CT_ASCII);
-            if (pUniType->extend & C3_IDEOGRAPH)
-                ufType |= __CT_IDEOGRAM;
-            if (ufType & (__CT_XDIGIT | __CT_DIGIT))
-            {
-                if (     (unsigned)wc - 0x30U <= (0x39 - 0x30))
-                    ufType |= (unsigned)wc - 0x30;
-                else if ((unsigned)wc - 0x41U <= (0x46 - 0x41))
-                    ufType |= (unsigned)wc - 0x41 + 0xa;
-                else
-                {
-                    unsigned uVal = UniQueryNumericValue(wc);
-                    if (!(uVal & ~0xffU))
-                        ufType |= uVal;
-                }
-            }
-            ufType |= (pUniType->bidi & 0xf << 24);
-
-            /** @todo screen width. */
-        }
-    }
-    return ufType;
+            return ___wctype_uni(pUniType, wc);
+    }
+    return 0;
 }
 

branches/libc-0.6/src/emx/src/lib/locale/setlocale.c ¶

@@ -599 +599 @@
         UniChar         uc = 0xffff;
 
-        /* isxxx() do not support MBCS characters at all. */
+        /* isxxx() does not support MBCS characters at all. */
         if (!pCtype->mbcs || !IS_MBCS_PREFIX(pCtype, i))
         {
@@ -609 +609 @@
                 if (pUniType)
                 {
-                    ufType = ___wctype(uc);
+                    ufType = ___wctype_uni(pUniType, uc);
                     if (ufType & __CT_LOWER)
                         uchUpper = Transform(lobj, uobj, UniTransUpper, uc, i);