Rewrote codepage support to use the new codepage tables.

dlls/kernel/kernel.spec

@@ -209,7 +209,7 @@ file	krnl386.exe
 204 stub SwapRecording
 205 stub CVWBreak
 206 pascal16 AllocSelectorArray(word) AllocSelectorArray16
-207 pascal16 IsDBCSLeadByte(word) IsDBCSLeadByte16
+207 pascal16 IsDBCSLeadByte(word) IsDBCSLeadByte
 
 
 # 208-237 are Win95 extensions; a few of those are also present in WinNT

memory/Makefile.in

@@ -7,6 +7,7 @@ MODULE    = memory
 
 C_SRCS = \
 	atom.c \
+	codepage.c \
 	environ.c \
 	global.c \
 	heap.c \

memory/codepage.c

+/*
+ * Code page functions
+ *
+ * Copyright 2000 Alexandre Julliard
+ */
+
+#include <assert.h>
+#include <stdio.h>
+
+#include "winbase.h"
+#include "winerror.h"
+#include "winnls.h"
+#include "wine/unicode.h"
+#include "debugtools.h"
+
+DEFAULT_DEBUG_CHANNEL(string);
+
+/* current code pages */
+static unsigned int ansi_cp = 1252;  /* Windows 3.1 ISO Latin */
+static unsigned int oem_cp = 437;    /* MS-DOS United States */
+static unsigned int mac_cp = 10000;  /* Mac Roman */
+
+static const union cptable *ansi_cptable;
+static const union cptable *oem_cptable;
+static const union cptable *mac_cptable;
+
+
+/* find the table for a given codepage, handling CP_ACP etc. pseudo-codepages */
+static const union cptable *get_codepage_table( unsigned int codepage )
+{
+    const union cptable *ret = NULL;
+
+    if (!ansi_cptable)  /* initialize them */
+    {
+        /* FIXME: should load from the registry */
+        ansi_cptable = cp_get_table( ansi_cp );
+        oem_cptable = cp_get_table( oem_cp );
+        mac_cptable = cp_get_table( mac_cp );
+        assert( ansi_cptable );
+        assert( oem_cptable );
+        assert( mac_cptable );
+    }
+
+    switch(codepage)
+    {
+    case CP_SYMBOL:
+    case CP_ACP:        return ansi_cptable;
+    case CP_OEMCP:      return oem_cptable;
+    case CP_MACCP:      return mac_cptable;
+    case CP_THREAD_ACP: return ansi_cptable; /* FIXME */
+    case CP_UTF7:
+    case CP_UTF8:
+        break;
+    default:
+        if (codepage == ansi_cp) return ansi_cptable;
+        if (codepage == oem_cp) return oem_cptable;
+        if (codepage == mac_cp) return mac_cptable;
+        ret = cp_get_table( codepage );
+        break;
+    }
+    return ret;
+}
+
+/******************************************************************************
+ *              GetACP   (KERNEL32)
+ *
+ * RETURNS
+ *    Current ANSI code-page identifier, default if no current defined
+ */
+UINT WINAPI GetACP(void)
+{
+    return ansi_cp;
+}
+
+
+/***********************************************************************
+ *              GetOEMCP   (KERNEL32)
+ */
+UINT WINAPI GetOEMCP(void)
+{
+    return oem_cp;
+}
+
+
+/***********************************************************************
+ *           IsValidCodePage   (KERNEL32)
+ */
+BOOL WINAPI IsValidCodePage( UINT codepage )
+{
+    return cp_get_table( codepage ) != NULL;
+}
+
+
+/***********************************************************************
+ *           IsDBCSLeadByteEx   (KERNEL32)
+ */
+BOOL WINAPI IsDBCSLeadByteEx( UINT codepage, BYTE testchar )
+{
+    const union cptable *table = get_codepage_table( codepage );
+    return table && is_dbcs_leadbyte( table, testchar );
+}
+
+
+/***********************************************************************
+ *           IsDBCSLeadByte   (KERNEL32)
+ */
+BOOL WINAPI IsDBCSLeadByte( BYTE testchar )
+{
+    return is_dbcs_leadbyte( ansi_cptable, testchar );
+}
+
+
+/***********************************************************************
+ *           GetCPInfo   (KERNEL32)
+ */
+BOOL WINAPI GetCPInfo( UINT codepage, LPCPINFO cpinfo )
+{
+    const union cptable *table = get_codepage_table( codepage );
+
+    if (!table) 
+    {
+        SetLastError( ERROR_INVALID_PARAMETER );
+        return FALSE;
+    }
+    if (table->info.def_char & 0xff00)
+    {
+        cpinfo->DefaultChar[0] = table->info.def_char & 0xff00;
+        cpinfo->DefaultChar[1] = table->info.def_char & 0x00ff;
+    }
+    else
+    {
+        cpinfo->DefaultChar[0] = table->info.def_char & 0xff;
+        cpinfo->DefaultChar[1] = 0;
+    }
+    if ((cpinfo->MaxCharSize = table->info.char_size) == 2)
+        memcpy( cpinfo->LeadByte, table->dbcs.lead_bytes, sizeof(cpinfo->LeadByte) );
+    else
+        cpinfo->LeadByte[0] = cpinfo->LeadByte[1] = 0;
+
+    return TRUE;
+}
+
+
+/***********************************************************************
+ *              EnumSystemCodePagesA   (KERNEL32)
+ */
+BOOL WINAPI EnumSystemCodePagesA( CODEPAGE_ENUMPROCA lpfnCodePageEnum, DWORD flags )
+{
+    const union cptable *table;
+    char buffer[10];
+    int index = 0;
+
+    for (;;)
+    {
+        if (!(table = cp_enum_table( index++ ))) break;
+        sprintf( buffer, "%d", table->info.codepage );
+        if (!lpfnCodePageEnum( buffer )) break;
+    }
+    return TRUE;
+}
+
+
+/***********************************************************************
+ *              EnumSystemCodePagesW   (KERNEL32)
+ */
+BOOL WINAPI EnumSystemCodePagesW( CODEPAGE_ENUMPROCW lpfnCodePageEnum, DWORD flags )
+{
+    const union cptable *table;
+    WCHAR buffer[10], *p;
+    int page, index = 0;
+
+    for (;;)
+    {
+        if (!(table = cp_enum_table( index++ ))) break;
+        p = buffer + sizeof(buffer)/sizeof(WCHAR);
+        *--p = 0;
+        page = table->info.codepage;
+        do
+        {
+            *--p = '0' + (page % 10);
+            page /= 10;
+        } while( page );
+        if (!lpfnCodePageEnum( p )) break;
+    }
+    return TRUE;
+}
+
+
+/***********************************************************************
+ *              MultiByteToWideChar   (KERNEL32)
+ *
+ * PARAMS
+ *   page [in]    Codepage character set to convert from
+ *   flags [in]   Character mapping flags
+ *   src [in]     Source string buffer
+ *   srclen [in]  Length of source string buffer
+ *   dst [in]     Destination buffer
+ *   dstlen [in]  Length of destination buffer
+ *
+ * NOTES
+ *   The returned length includes the null terminator character.
+ *
+ * RETURNS
+ *   Success: If dstlen > 0, number of characters written to destination
+ *            buffer.  If dstlen == 0, number of characters needed to do
+ *            conversion.
+ *   Failure: 0. Occurs if not enough space is available.
+ *
+ * ERRORS
+ *   ERROR_INSUFFICIENT_BUFFER
+ *   ERROR_INVALID_PARAMETER
+ *   ERROR_NO_UNICODE_TRANSLATION
+ *
+ */
+INT WINAPI MultiByteToWideChar( UINT page, DWORD flags, LPCSTR src, INT srclen,
+                                LPWSTR dst, INT dstlen )
+{
+    const union cptable *table;
+    int ret;
+
+    if (srclen == -1) srclen = strlen(src) + 1;
+
+    if (page >= CP_UTF7)
+    {
+        FIXME("UTF not supported\n");
+        SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
+        return 0;
+    }
+
+    if (!(table = get_codepage_table( page )))
+    {
+        SetLastError( ERROR_INVALID_PARAMETER );
+        return 0;
+    }
+
+    if (flags & MB_COMPOSITE) FIXME("MB_COMPOSITE not supported\n");
+    if (flags & MB_USEGLYPHCHARS) FIXME("MB_USEGLYPHCHARS not supported\n");
+
+    ret = cp_mbstowcs( table, flags, src, srclen, dst, dstlen );
+
+    if (ret < 0)
+    {
+        switch(ret)
+        {
+        case -1: SetLastError( ERROR_INSUFFICIENT_BUFFER ); break;
+        case -2: SetLastError( ERROR_NO_UNICODE_TRANSLATION ); break;
+        }
+        ret = 0;
+    }
+    return ret;
+}
+
+
+/***********************************************************************
+ *              WideCharToMultiByte   (KERNEL32)
+ *
+ * PARAMS
+ *   page [in]    Codepage character set to convert to
+ *   flags [in]   Character mapping flags
+ *   src [in]     Source string buffer
+ *   srclen [in]  Length of source string buffer
+ *   dst [in]     Destination buffer
+ *   dstlen [in]  Length of destination buffer
+ *   defchar [in] Default character to use for conversion if no exact
+ *		    conversion can be made
+ *   used [out]   Set if default character was used in the conversion
+ *
+ * NOTES
+ *   The returned length includes the null terminator character.
+ *
+ * RETURNS
+ *   Success: If dstlen > 0, number of characters written to destination
+ *            buffer.  If dstlen == 0, number of characters needed to do
+ *            conversion.
+ *   Failure: 0. Occurs if not enough space is available.
+ *
+ * ERRORS
+ *   ERROR_INSUFFICIENT_BUFFER
+ *   ERROR_INVALID_PARAMETER
+ */
+INT WINAPI WideCharToMultiByte( UINT page, DWORD flags, LPCWSTR src, INT srclen,
+                                LPSTR dst, INT dstlen, LPCSTR defchar, BOOL *used )
+{
+    const union cptable *table;
+    int ret, used_tmp;
+
+    if (srclen == -1) srclen = strlenW(src) + 1;
+
+    if (page >= CP_UTF7)
+    {
+        FIXME("UTF not supported\n");
+        SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
+        return 0;
+    }
+
+    if (!(table = get_codepage_table( page )))
+    {
+        SetLastError( ERROR_INVALID_PARAMETER );
+        return 0;
+    }
+
+/*    if (flags & WC_COMPOSITECHECK) FIXME( "WC_COMPOSITECHECK (%lx) not supported\n", flags );*/
+
+    ret = cp_wcstombs( table, flags, src, srclen, dst, dstlen, defchar, used ? &used_tmp : NULL );
+    if (used) *used = used_tmp;
+
+    if (ret == -1)
+    {
+        SetLastError( ERROR_INSUFFICIENT_BUFFER );
+        ret = 0;
+    }
+    return ret;
+}

memory/string.c

@@ -394,11 +394,7 @@ INT WINAPI lstrlenW( LPCWSTR str )
  */
 LPWSTR WINAPI lstrcpyAtoW( LPWSTR dst, LPCSTR src )
 {
-    register LPWSTR p = dst;
-
-    TRACE("(%p, %s)\n", dst, debugstr_a(src));
-
-    while ((*p++ = (WCHAR)(unsigned char)*src++));
+    MultiByteToWideChar( CP_ACP, 0, src, -1, dst, 0x7fffffff );
     return dst;
 }
 
@@ -408,11 +404,7 @@ LPWSTR WINAPI lstrcpyAtoW( LPWSTR dst, LPCSTR src )
  */
 LPSTR WINAPI lstrcpyWtoA( LPSTR dst, LPCWSTR src )
 {
-    register LPSTR p = dst;
-
-    TRACE("(%p, %s)\n", dst, debugstr_w(src));
-
-    while ((*p++ = (CHAR)*src++));
+    WideCharToMultiByte( CP_ACP, 0, src, -1, dst, 0x7fffffff, NULL, NULL );
     return dst;
 }
 
@@ -424,12 +416,7 @@ LPSTR WINAPI lstrcpyWtoA( LPSTR dst, LPCWSTR src )
  */
 LPWSTR WINAPI lstrcpynAtoW( LPWSTR dst, LPCSTR src, INT n )
 {
-    LPWSTR p = dst;
-
-    TRACE("(%p, %s, %i)\n", dst, debugstr_an(src,n), n);
-
-    while ((n-- > 1) && *src) *p++ = (WCHAR)(unsigned char)*src++;
-    if (n >= 0) *p = 0;
+    if (n > 0 && !MultiByteToWideChar( CP_ACP, 0, src, -1, dst, n )) dst[n-1] = 0;
     return dst;
 }
 
@@ -445,14 +432,7 @@ LPWSTR WINAPI lstrcpynAtoW( LPWSTR dst, LPCSTR src, INT n )
  */
 LPSTR WINAPI lstrcpynWtoA( LPSTR dst, LPCWSTR src, INT n )
 {
-    if (--n >= 0)
-    {
-        TRACE("(%p, %s, %i)\n", dst, debugstr_wn(src,n), n);
-        n = CRTDLL_wcstombs( dst, src, n );
-        if(n<0)
-                 n=0;
-        dst[n] = 0;
-    }
+    if (n > 0 && !WideCharToMultiByte( CP_ACP, 0, src, -1, dst, n, NULL, NULL )) dst[n-1] = 0;
     return dst;
 }
 

win32/Makefile.in

@@ -6,7 +6,6 @@ VPATH     = @srcdir@
 MODULE    = win32
 
 C_SRCS = \
-	code_page.c \
 	console.c \
 	device.c \
 	error.c \

win32/code_page.c

-/*
- * Win32 kernel functions
- *
- * Copyright 1995 Martin von Loewis and Cameron Heide
- */
-
-#include <stdlib.h>
-#include "winerror.h"
-#include "winnls.h"
-#include "heap.h"
-#include "debugtools.h"
-
-DEFAULT_DEBUG_CHANNEL(win32)
-
-
-/******************************************************************************
- * GetACP [KERNEL32.276]  Gets current ANSI code-page identifier.
- *
- * RETURNS
- *    Current ANSI code-page identifier, default if no current defined
- */
-UINT WINAPI GetACP(void)
-{
-    /* This introduces too many messages */
-/*    FIXME(win32, "(void): stub\n"); */
-    return 1252;    /* Windows 3.1 ISO Latin */
-}
-
-
-/***********************************************************************
- *           GetCPInfo            (KERNEL32.154)
- */
-BOOL WINAPI GetCPInfo( UINT codepage, LPCPINFO cpinfo )
-{
-    cpinfo->DefaultChar[0] = '?';
-    switch (codepage)
-    {
-    case 932 : /* Shift JIS (japan) */
-        cpinfo->MaxCharSize = 2;
-        cpinfo->LeadByte[0]= 0x81; cpinfo->LeadByte[1] = 0x9F;
-        cpinfo->LeadByte[2]= 0xE0; cpinfo->LeadByte[3] = 0xFC;
-        cpinfo->LeadByte[4]= 0x00; cpinfo->LeadByte[5] = 0x00;
-        break;
-    case 936 : /* GB2312 (Chinese) */
-    case 949 : /* KSC5601-1987 (Korean) */
-    case 950 : /* BIG5 (Chinese) */
-        cpinfo->MaxCharSize = 2;
-        cpinfo->LeadByte[0]= 0x81; cpinfo->LeadByte[1] = 0xFE;
-        cpinfo->LeadByte[2]= 0x00; cpinfo->LeadByte[3] = 0x00;
-        break;
-    case 1361 : /* Johab (Korean) */
-        cpinfo->MaxCharSize = 2;
-        cpinfo->LeadByte[0]= 0x84; cpinfo->LeadByte[1] = 0xD3;
-        cpinfo->LeadByte[2]= 0xD8; cpinfo->LeadByte[3] = 0xDE;
-        cpinfo->LeadByte[4]= 0xE0; cpinfo->LeadByte[5] = 0xF9;
-        cpinfo->LeadByte[6]= 0x00; cpinfo->LeadByte[7] = 0x00;
-        break;
-    default :
-    	cpinfo->MaxCharSize = 1;
-        cpinfo->LeadByte[0]= 0x00; cpinfo->LeadByte[1] = 0x00;
-    	break;
-    }
-    return 1;
-}
-
-/***********************************************************************
- *              GetOEMCP                (KERNEL32.248)
- */
-UINT WINAPI GetOEMCP(void)
-{
-    return 437;    /* MS-DOS United States */
-}
-
-/***********************************************************************
- *           IsValidCodePage   (KERNEL32.360)
- */
-BOOL WINAPI IsValidCodePage(UINT CodePage)
-{
-    switch ( CodePage )
-    {
-    case 1252 :
-    case 437 :
-        return TRUE;
-    default :
-        return FALSE;
-    }
-}
-
-
-/***********************************************************************
- *              MultiByteToWideChar                (KERNEL32.534)
- *
- * PARAMS
- *   page [in]   Codepage character set to convert from
- *   flags [in]  Character mapping flags
- *   src [in]     Source string buffer
- *   srclen [in]  Length of source string buffer
- *   dst [in]     Destination buffer
- *   dstlen [in]  Length of destination buffer
- *
- * NOTES
- *   The returned length includes the null terminator character.
- *
- * RETURNS
- *   Success: If dstlen > 0, number of characters written to destination
- *            buffer.  If dstlen == 0, number of characters needed to do
- *            conversion.
- *   Failure: 0. Occurs if not enough space is available.
- *
- * ERRORS
- *   ERROR_INSUFFICIENT_BUFFER
- *   ERROR_INVALID_FLAGS (not yet implemented)
- *   ERROR_INVALID_PARAMETER (not yet implemented)
- *
- * BUGS
- *   Does not properly handle codepage conversions.
- *   Does not properly handle flags.
- *
- */
-INT WINAPI MultiByteToWideChar(UINT page, DWORD flags,
-			         LPCSTR src, INT srclen,
-                                 LPWSTR dst, INT dstlen)
-{
-    int ret;
-
-    if (srclen == -1)
-	srclen = lstrlenA(src)+1;
-    if (!dstlen || !dst)
-	return srclen;
-
-    ret = srclen;
-    while (srclen && dstlen) {
-	*dst = (WCHAR)(unsigned char)*src;
-	dst++;    src++;
-	dstlen--; srclen--;
-    }
-    if (!dstlen && srclen) {
-	SetLastError(ERROR_INSUFFICIENT_BUFFER);
-	return 0;
-    }
-    return ret;
-}
-
-/***********************************************************************
- *              WideCharToMultiByte                (KERNEL32.727)
- *
- * PARAMS
- *   page [in]    Codepage character set to convert to
- *   flags [in]   Character mapping flags
- *   src [in]     Source string buffer
- *   srclen [in]  Length of source string buffer
- *   dst [in]     Destination buffer
- *   dstlen [in]  Length of destination buffer
- *   defchar [in] Default character to use for conversion if no exact
- *		    conversion can be made
- *   used [out]   Set if default character was used in the conversion
- *
- * NOTES
- *   The returned length includes the null terminator character.
- *
- * RETURNS
- *   Success: If dstlen > 0, number of characters written to destination
- *            buffer.  If dstlen == 0, number of characters needed to do
- *            conversion.
- *   Failure: 0. Occurs if not enough space is available.
- *
- * ERRORS
- *   ERROR_INSUFFICIENT_BUFFER
- *   ERROR_INVALID_FLAGS (not yet implemented)
- *
- * BUGS
- *   Does not properly handle codepage conversions.
- *   Does not properly handle flags.
- *
- */
-INT WINAPI WideCharToMultiByte(UINT page, DWORD flags, LPCWSTR src,
-				 INT srclen,LPSTR dst, INT dstlen,
-				 LPCSTR defchar, BOOL *used)
-{
-    int count = 0;
-    int eos = 0;
-    int care_for_eos=0;
-    int dont_copy= (dstlen==0);
-
-    if ((!src) || ((!dst) && (!dont_copy)) )
-    {	SetLastError(ERROR_INVALID_PARAMETER);
-	return 0;
-    }
-    
-    if (page!=GetACP() && page!=CP_OEMCP && page!=CP_ACP)
-	FIXME("Conversion in CP %d not supported\n",page);
-#if 0
-    if (flags)
-	FIXME("flags %lx not supported\n",flags);
-#endif
-    if(used)
-	*used=0;
-    if (srclen == -1)
-      {
-	srclen = lstrlenW(src)+1;
-	 care_for_eos=1;
-      }
-    while(srclen && (dont_copy || dstlen))
-    {
-	if(!dont_copy){
-	    if(*src<256)
-		*dst = *src;
-	    else
-	    {
-		/* ??? The WC_DEFAULTCHAR flag only gets used in
-		 * combination with the WC_COMPOSITECHECK flag or at
-		 * least this is what it seems from using the function
-		 * on NT4.0 in combination with reading the documentation.
-		 */
-		*dst = defchar ? *defchar : '?';
-		if(used)*used=1;
-	    }
-	    dstlen--;
-	    dst++;
-	}
-	count++;
-	srclen--;
-	if((!*src) && care_for_eos) {
-	    eos = 1;
-	    break;
-	}
-	src++;
-    }
-    if (dont_copy)
-	return count;
-
-    if (!eos && srclen > 0) {
-	SetLastError(ERROR_INSUFFICIENT_BUFFER);
-	return 0;
-    }
-    return count;
-}
-
-
-/***********************************************************************
- *           IsDBCSLeadByteEx   (KERNEL32.359)
- */
-BOOL WINAPI IsDBCSLeadByteEx( UINT codepage, BYTE testchar )
-{
-    CPINFO cpinfo;
-    int i;
-
-    GetCPInfo(codepage, &cpinfo);
-    for (i = 0 ; i < sizeof(cpinfo.LeadByte)/sizeof(cpinfo.LeadByte[0]); i+=2)
-    {
-	if (cpinfo.LeadByte[i] == 0)
-            return FALSE;
-	if (cpinfo.LeadByte[i] <= testchar && testchar <= cpinfo.LeadByte[i+1])
-            return TRUE;
-    }
-    return FALSE;
-}
-
-
-/***********************************************************************
- *           IsDBCSLeadByte16   (KERNEL.207)
- */
-BOOL16 WINAPI IsDBCSLeadByte16( BYTE testchar )
-{
-    return IsDBCSLeadByteEx(GetACP(), testchar);
-}
-
-
-/***********************************************************************
- *           IsDBCSLeadByte   (KERNEL32.358)
- */
-BOOL WINAPI IsDBCSLeadByte( BYTE testchar )
-{
-    return IsDBCSLeadByteEx(GetACP(), testchar);
-}
-
-
-/***********************************************************************
- *              EnumSystemCodePagesA                (KERNEL32.92)
- */
-BOOL WINAPI EnumSystemCodePagesA(CODEPAGE_ENUMPROCA lpfnCodePageEnum,DWORD flags)
-{
-	TRACE("(%p,%08lx)\n",lpfnCodePageEnum,flags);
-	lpfnCodePageEnum("437");
-	return TRUE;
-}
-
-/***********************************************************************
- *              EnumSystemCodePagesW                (KERNEL32.93)
- */
-BOOL WINAPI EnumSystemCodePagesW( CODEPAGE_ENUMPROCW lpfnCodePageEnum,
-                                      DWORD flags)
-{
-    WCHAR	*cp;
-    TRACE("(%p,%08lx)\n",lpfnCodePageEnum,flags );
-
-    cp = HEAP_strdupAtoW( GetProcessHeap(), 0, "437" );
-    lpfnCodePageEnum(cp);
-    HeapFree( GetProcessHeap(), 0, cp );
-    return TRUE;
-}