ICU-6606 add #define option for >>default charset is UTF-8<<

X-SVN-Rev: 25544

icu4c/source/common/putil.c

@@ -1572,6 +1572,7 @@ The leftmost codepage (.xxx) wins.
 
 }
 
+#if !U_CHARSET_IS_UTF8
 #if U_POSIX_LOCALE
 /*
 Due to various platform differences, one platform may specify a charset,
@@ -1806,6 +1807,7 @@ uprv_getDefaultCodepage()
     umtx_unlock(NULL);
     return name;
 }
+#endif  /* !U_CHARSET_IS_UTF8 */
 
 
 /* end of platform-specific implementation -------------- */

icu4c/source/common/ucnv_bld.c

@@ -1,7 +1,7 @@
 /*
  ********************************************************************
  * COPYRIGHT:
- * Copyright (c) 1996-2008, International Business Machines Corporation and
+ * Copyright (c) 1996-2009, International Business Machines Corporation and
  * others. All Rights Reserved.
  ********************************************************************
  *
@@ -158,6 +158,8 @@ static UMTX        cnvCacheMutex = NULL;  /* Mutex for synchronizing cnv cache a
 static const char **gAvailableConverters = NULL;
 static uint16_t gAvailableConverterCount = 0;
 
+#if !U_CHARSET_IS_UTF8
+
 /* This contains the resolved converter name. So no further alias lookup is needed again. */
 static char gDefaultConverterNameBuffer[UCNV_MAX_CONVERTER_NAME_LENGTH + 1]; /* +1 for NULL */
 static const char *gDefaultConverterName = NULL;
@@ -173,6 +175,7 @@ static const UConverterSharedData *gDefaultAlgorithmicSharedData = NULL;
 /* Does gDefaultConverterName have a converter option and require extra parsing? */
 static UBool gDefaultConverterContainsOption;
 
+#endif  /* !U_CHARSET_IS_UTF8 */
 
 static const char DATA_TYPE[] = "cnv";
 
@@ -201,10 +204,12 @@ static UBool U_CALLCONV ucnv_cleanup(void) {
     /* Isn't called from flushCache because other threads may have preexisting references to the table. */
     ucnv_flushAvailableConverterCache();
 
+#if !U_CHARSET_IS_UTF8
     gDefaultConverterName = NULL;
     gDefaultConverterNameBuffer[0] = 0;
     gDefaultConverterContainsOption = FALSE;
     gDefaultAlgorithmicSharedData = NULL;
+#endif
 
     umtx_destroy(&cnvCacheMutex);    /* Don't worry about destroying the mutex even  */
                                      /*  if the hash table still exists.  The mutex  */
@@ -707,6 +712,9 @@ ucnv_loadSharedData(const char *converterName, UConverterLookupData *lookup, UEr
 
     /* In case "name" is NULL we want to open the default converter. */
     if (converterName == NULL) {
+#if U_CHARSET_IS_UTF8
+        return (UConverterSharedData *)converterData[UCNV_UTF8];
+#else
         /* Call ucnv_getDefaultName first to query the name from the OS. */
         lookup->realName = ucnv_getDefaultName();
         if (lookup->realName == NULL) {
@@ -717,6 +725,7 @@ ucnv_loadSharedData(const char *converterName, UConverterLookupData *lookup, UEr
         checkForAlgorithmic = FALSE;
         mayContainOption = gDefaultConverterContainsOption;
         /* the default converter name is already canonical */
+#endif
     }
     else if((converterName[0] == 'U' ?
             (                           converterName[1] == 'T' && converterName[2] == 'F') :
@@ -1113,6 +1122,7 @@ ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode) {
 
 /* default converter name --------------------------------------------------- */
 
+#if !U_CHARSET_IS_UTF8
 /*
 Copy the canonical converter name.
 ucnv_getDefaultName must be thread safe, which can call this function.
@@ -1147,7 +1157,7 @@ internalSetName(const char *name, UErrorCode *status) {
     gDefaultConverterContainsOption = containsOption;
     uprv_memcpy(gDefaultConverterNameBuffer, name, length);
     gDefaultConverterNameBuffer[length]=0;
-    
+
     /* gDefaultConverterName MUST be the last global var set by this function.  */
     /*    It is the variable checked in ucnv_getDefaultName() to see if initialization is required. */
     gDefaultConverterName = gDefaultConverterNameBuffer;
@@ -1156,6 +1166,7 @@ internalSetName(const char *name, UErrorCode *status) {
 
     umtx_unlock(&cnvCacheMutex);
 }
+#endif
 
 /*
  * In order to be really thread-safe, the get function would have to take
@@ -1167,6 +1178,9 @@ internalSetName(const char *name, UErrorCode *status) {
 
 U_CAPI const char*  U_EXPORT2
 ucnv_getDefaultName() {
+#if U_CHARSET_IS_UTF8
+    return "UTF-8";
+#else
     /* local variable to be thread-safe */
     const char *name;
 
@@ -1211,6 +1225,7 @@ ucnv_getDefaultName() {
     }
 
     return name;
+#endif
 }
 
 /*
@@ -1219,6 +1234,7 @@ See internalSetName or the API reference for details.
 */
 U_CAPI void U_EXPORT2
 ucnv_setDefaultName(const char *converterName) {
+#if !U_CHARSET_IS_UTF8
     if(converterName==NULL) {
         /* reset to the default codepage */
         gDefaultConverterName=NULL;
@@ -1241,6 +1257,7 @@ ucnv_setDefaultName(const char *converterName) {
         /* The close may make the current name go away. */
         ucnv_close(cnv);
     }
+#endif
 }
 
 /* data swapping ------------------------------------------------------------ */

icu4c/source/common/unicode/putil.h

@@ -1,7 +1,7 @@
 /*
 ******************************************************************************
 *
-*   Copyright (C) 1997-2008, International Business Machines
+*   Copyright (C) 1997-2009, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 ******************************************************************************
@@ -29,7 +29,7 @@
   * \file
   * \brief C API: Platform Utilities
   */
-  
+
 /** Define this to 1 if your platform supports IEEE 754 floating point,
    to 0 if it does not. */
 #ifndef IEEE_754
@@ -93,6 +93,7 @@ U_STABLE const char* U_EXPORT2 u_getDataDirectory(void);
  */
 U_STABLE void U_EXPORT2 u_setDataDirectory(const char *directory);
 
+#if !U_CHARSET_IS_UTF8
 /**
  * Please use ucnv_getDefaultName() instead.
  * Return the default codepage for this platform and locale.
@@ -102,6 +103,7 @@ U_STABLE void U_EXPORT2 u_setDataDirectory(const char *directory);
  * @internal
  */
 U_INTERNAL const char*  U_EXPORT2 uprv_getDefaultCodepage(void);
+#endif
 
 /**
  * Please use uloc_getDefault() instead.

icu4c/source/common/unicode/ucnv.h

@@ -1,6 +1,6 @@
 /*
 **********************************************************************
-*   Copyright (C) 1999-2008, International Business Machines
+*   Copyright (C) 1999-2009, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 **********************************************************************
  *  ucnv.h:
@@ -1790,6 +1790,9 @@ ucnv_getCanonicalName(const char *alias, const char *standard, UErrorCode *pErro
  * It is faster if you pass a NULL argument to ucnv_open the
  * default converter.
  *
+ * If U_CHARSET_IS_UTF8 is defined to 1 in utypes.h then this function
+ * always returns "UTF-8".
+ *
  * @return returns the current default converter name.
  *         Storage owned by the library
  * @see ucnv_setDefaultName
@@ -1805,6 +1808,10 @@ ucnv_getDefaultName(void);
  * should be called during application initialization. Most of the time, the
  * results from ucnv_getDefaultName() or ucnv_open with a NULL string argument
  * is sufficient for your application.
+ *
+ * If U_CHARSET_IS_UTF8 is defined to 1 in utypes.h then this function
+ * does nothing.
+ *
  * @param name the converter name to be the default (must be known by ICU).
  * @see ucnv_getDefaultName
  * @system

icu4c/source/common/unicode/uconfig.h

@@ -1,6 +1,6 @@
 /*  
 **********************************************************************
-*   Copyright (C) 2002-2008, International Business Machines
+*   Copyright (C) 2002-2009, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 **********************************************************************
 *   file name:  uconfig.h
@@ -104,7 +104,11 @@
  * ICU will not completely build with this switch turned on.
  * This switch turns off all converters.
  *
+ * You may want to use this together with U_CHARSET_IS_UTF8 defined to 1
+ * in utypes.h if char* strings in your environment are always in UTF-8.
+ *
  * @stable ICU 3.2
+ * @see U_CHARSET_IS_UTF8
  */
 #ifndef UCONFIG_NO_CONVERSION
 #   define UCONFIG_NO_CONVERSION 0

icu4c/source/common/unicode/utypes.h

@@ -1,6 +1,6 @@
 /*
 **********************************************************************
-*   Copyright (C) 1996-2008, International Business Machines
+*   Copyright (C) 1996-2009, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 **********************************************************************
 *
@@ -139,6 +139,30 @@
 #   define U_CHARSET_FAMILY 0
 #endif
 
+/**
+ * \def U_CHARSET_IS_UTF8
+ *
+ * Hardcode the default charset to UTF-8.
+ *
+ * If this is set to 1, then
+ * - ICU will assume that all non-invariant char*, StringPiece, std::string etc.
+ *   contain UTF-8 text, regardless of what the system API uses
+ * - some ICU code will use fast functions like u_strFromUTF8()
+ *   rather than the more general and more heavy-weight conversion API (ucnv.h)
+ * - ucnv_getDefaultName() always returns "UTF-8"
+ * - ucnv_setDefaultName() is disabled and will not change the default charset
+ * - static builds of ICU are smaller
+ * - more functionality is available with the UCONFIG_NO_CONVERSION build-time
+ *   configuration option (see unicode/uconfig.h)
+ * - the UCONFIG_NO_CONVERSION build option in uconfig.h is more usable
+ *
+ * @draft ICU 4.2
+ * @see UCONFIG_NO_CONVERSION
+ */
+#ifndef U_CHARSET_IS_UTF8
+#   define U_CHARSET_IS_UTF8 0
+#endif
+
 /*===========================================================================*/
 /* ICUDATA naming scheme                                                     */
 /*===========================================================================*/

icu4c/source/test/cintltst/ccapitst.c

@@ -1,6 +1,6 @@
 /********************************************************************
  * COPYRIGHT: 
- * Copyright (c) 1997-2008, International Business Machines Corporation and
+ * Copyright (c) 1997-2009, International Business Machines Corporation and
  * others. All Rights Reserved.
  ********************************************************************/
 /*****************************************************************************
@@ -3335,11 +3335,11 @@ TestToUCountPending(){
 #endif
 }
 
-static void TestOneDefaultNameChange(const char *name) {
+static void TestOneDefaultNameChange(const char *name, const char *expected) {
     UErrorCode status = U_ZERO_ERROR;
     UConverter *cnv;
     ucnv_setDefaultName(name);
-    if(strcmp(ucnv_getDefaultName(), name)==0)
+    if(strcmp(ucnv_getDefaultName(), expected)==0)
         log_verbose("setDefaultName of %s works.\n", name);
     else
         log_err("setDefaultName of %s failed\n", name);
@@ -3348,7 +3348,7 @@ static void TestOneDefaultNameChange(const char *name) {
         log_err("opening the default converter of %s failed\n", name);
         return;
     }
-    if(strcmp(ucnv_getName(cnv, &status), name)==0)
+    if(strcmp(ucnv_getName(cnv, &status), expected)==0)
         log_verbose("ucnv_getName of %s works.\n", name);
     else
         log_err("ucnv_getName of %s failed\n", name);
@@ -3363,12 +3363,18 @@ static void TestDefaultName(void) {
     log_verbose("getDefaultName returned %s\n", defaultName);
 
     /*change the default name by setting it */
-    TestOneDefaultNameChange("UTF-8");
+    TestOneDefaultNameChange("UTF-8", "UTF-8");
-#if !UCONFIG_NO_LEGACY_CONVERSION
+#if U_CHARSET_IS_UTF8
-    TestOneDefaultNameChange("ISCII,version=1");
+    TestOneDefaultNameChange("ISCII,version=1", "UTF-8");
-    TestOneDefaultNameChange("ISCII,version=2");
+    TestOneDefaultNameChange("ISCII,version=2", "UTF-8");
+    TestOneDefaultNameChange("ISO-8859-1", "UTF-8");
+#else
+# if !UCONFIG_NO_LEGACY_CONVERSION
+    TestOneDefaultNameChange("ISCII,version=1", "ISCII,version=1");
+    TestOneDefaultNameChange("ISCII,version=2", "ISCII,version=2");
+# endif
+    TestOneDefaultNameChange("ISO-8859-1", "ISO-8859-1");
 #endif
-    TestOneDefaultNameChange("ISO-8859-1");
 
     /*set the default name back*/
     ucnv_setDefaultName(defaultName);