Split J9STR_CODE_PLATFORM into _PLATFORM_RAW and _PLATFORM_OMR_INTERNAL

Issue: eclipse-omr#41
Signed-off-by: Peter Bain <[email protected]>

fvtest/porttest/omrstrTest.cpp

@@ -878,7 +878,7 @@ TEST(PortStrTest, str_convPlatTo8)
 
 	reportTestEntry(OMRPORTLIB, testName);
 	memset(outBuff, 0, sizeof(outBuff));
-	convertedStringLength = omrstr_convert(J9STR_CODE_PLATFORM, J9STR_CODE_MUTF8,
+	convertedStringLength = omrstr_convert(J9STR_CODE_PLATFORM_RAW, J9STR_CODE_MUTF8,
 										   platformString, originalStringLength,  outBuff, sizeof(outBuff));
 	if (convertedStringLength != expectedStringLength) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "buffer length wrong.  Expected %d actual %d\n", expectedStringLength, convertedStringLength);
@@ -887,7 +887,7 @@ TEST(PortStrTest, str_convPlatTo8)
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "Converted string wrong.");
 	}
 	/* test string length */
-	convertedStringLength = omrstr_convert(J9STR_CODE_PLATFORM, J9STR_CODE_MUTF8,
+	convertedStringLength = omrstr_convert(J9STR_CODE_PLATFORM_RAW, J9STR_CODE_MUTF8,
 										   platformString, originalStringLength, NULL, 0);
 	if (convertedStringLength != expectedStringLength) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "buffer length wrong.  Expected %d actual %d\n", expectedStringLength, convertedStringLength);
@@ -915,27 +915,27 @@ TEST(PortStrTest, str_convLongString)
 		longPlatformString[i] = (i % 127) + 1; /* stick to ASCII */
 	}
 	/* test string length */
-	mutf8StringLength = omrstr_convert(J9STR_CODE_PLATFORM, J9STR_CODE_MUTF8,
+	mutf8StringLength = omrstr_convert(J9STR_CODE_PLATFORM_RAW, J9STR_CODE_MUTF8,
 									   longPlatformString, originalStringLength,  NULL, 0);
 	if (mutf8StringLength < 0) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "Length calculation for modified UTF8 failed, error code %d\n", mutf8StringLength);
 	}
 	/* do the actual conversion */
-	mutf8StringLength = omrstr_convert(J9STR_CODE_PLATFORM, J9STR_CODE_MUTF8,
+	mutf8StringLength = omrstr_convert(J9STR_CODE_PLATFORM_RAW, J9STR_CODE_MUTF8,
 									   longPlatformString, originalStringLength,  mutf8Buff, sizeof(mutf8Buff));
 	if (mutf8StringLength < 0) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "Conversion to modified UTF8 failed, error code %d\n", mutf8StringLength);
 	}
 
 	/* test string length in the other direction */
-	returnStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+	returnStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 										mutf8Buff, mutf8StringLength, NULL, 0);
 	if (returnStringLength < 0) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "Length calculation for platform failed, error code %d\n", mutf8StringLength);
 	}
 
 	/* convert back and verify that it matches the original */
-	returnStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+	returnStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 										mutf8Buff, mutf8StringLength, returnBuff, sizeof(returnBuff));
 	if (returnStringLength < 0) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "Conversion to platform failed, error code %d\n", mutf8StringLength);
@@ -963,7 +963,7 @@ TEST(PortStrTest, str_convU8ToPlat)
 
 	reportTestEntry(OMRPORTLIB, testName);
 	memset(outBuff, 0, sizeof(outBuff));
-	convertedStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+	convertedStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 										   utf8String, originalStringLength,  outBuff, sizeof(outBuff));
 	if (convertedStringLength != expectedStringLength) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "buffer length wrong.  Expected %d actual %d\n", originalStringLength, convertedStringLength);
@@ -972,7 +972,7 @@ TEST(PortStrTest, str_convU8ToPlat)
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "Converted string wrong.");
 	}
 	/* test string length */
-	convertedStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+	convertedStringLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 										   utf8String, originalStringLength, NULL, 0);
 	if (convertedStringLength != expectedStringLength) {
 		outputErrorMessage(PORTTEST_ERROR_ARGS, "buffer length wrong.  Expected %d actual %d\n", originalStringLength, convertedStringLength);
@@ -1115,7 +1115,7 @@ TEST(PortStrTest, str_convU8ToPlat_Null)
 
 	(void)memset(outBuff, '^', sizeof(outBuff)); /* initialize to non-zero */
 	convertedStringLength =
-		omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+		omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 					   inputString, inputStringLength,  outBuff, sizeof(outBuff));
 
 	convertedStringTerminator = (uint32_t *)(outBuff + convertedStringLength);

fvtest/porttest/testProcessHelpers.cpp

@@ -361,7 +361,7 @@ translateModifiedUtf8ToPlatform(OMRPortLibrary *portLibrary, const char *inBuffe
 	int32_t resultLength = 0;
 
 	*outBuffer = NULL;
-	bufferLength = 	omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+	bufferLength = 	omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 								   inBuffer, inBufferSize, NULL, 0);
 	/* get the size of the platform string */
 
@@ -376,7 +376,7 @@ translateModifiedUtf8ToPlatform(OMRPortLibrary *portLibrary, const char *inBuffe
 		return OMRPORT_ERROR_STRING_MEM_ALLOCATE_FAILED;
 	}
 
-	resultLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+	resultLength = omrstr_convert(J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 								  inBuffer, inBufferSize, result, bufferLength);
 	/* do the conversion */
 

include_core/omrport.h

@@ -777,7 +777,7 @@ typedef struct J9ProcessorInfos {
 
 /* omrstr_convert encodings. */
 /* character set currently in effect */
-#define J9STR_CODE_PLATFORM 1
+#define J9STR_CODE_PLATFORM_RAW 1
 /* modified UTF-8 */
 #define J9STR_CODE_MUTF8 2
 /* UTF-16 */
@@ -793,6 +793,22 @@ typedef struct J9ProcessorInfos {
 /* Windows current thread ANSI code page */
 #define J9STR_CODE_WINTHREADACP 8
 
+#if defined(J9ZOS390)
+/*
+ * OMR on z/OS translates the output of certain system calls such as getenv to ASCII using functions in atoe.c; see stdlib.h for a list.
+ * Use J9STR_CODE_PLATFORM_OMR_INTERNAL to when processing the output of these calls.  Otherwise, use J9STR_CODE_PLATFORM_RAW.
+ */
+#define J9STR_CODE_PLATFORM_OMR_INTERNAL J9STR_CODE_LATIN1
+#elif defined(WIN32)
+/*
+ * Most system calls on Windows use the "wide" versions which return UTF-16, which OMR then converts to UTF-8.
+ */
+#define J9STR_CODE_PLATFORM_OMR_INTERNAL J9STR_CODE_UTF8
+#else /* defined(WIN32) */
+/* on other platforms the internal encoding is the actual operating system encoding */
+#define J9STR_CODE_PLATFORM_OMR_INTERNAL J9STR_CODE_PLATFORM_RAW
+#endif /* defined(J9ZOS390) */
+
 #define UNICODE_REPLACEMENT_CHARACTER 0xFFFD
 #define MAX_STRING_TERMINATOR_LENGTH 4
 /* worst case terminating sequence: wchar or UTF-32 */

port/common/omrstr.c

@@ -209,7 +209,10 @@ omrstr_printf(struct OMRPortLibrary *portLibrary, char *buf, uintptr_t bufLen, c
  * @param[in] fromCode      Input string encoding.  Only the following encodings are allowed:
  *                              J9STR_CODE_MUTF8 (Modified UTF-8)
  *                              J9STR_CODE_WIDE (UTF-16)
- *                              J9STR_CODE_PLATFORM (encoding used by the operating system)
+ *                              J9STR_CODE_LATIN1
+ *                              J9STR_CODE_PLATFORM_RAW (encoding used by the operating system)
+ *                              J9STR_CODE_PLATFORM_OMR_INTERNAL (encoding used by certain operating system calls)
+ *                              J9STR_CODE_WINTHREADACP, J9STR_CODE_WINDEFAULTACP (thread and default ANSI code page, Windows only)
  * @param[in] toCode        Output string encoding.  Only the encodings listed above are allowed.
  * @param[in] inBuffer      Input text to be converted.  May contain embedded null characters.
  * @param[in] inBufferSize  input string size in bytes, not including the terminating null.
@@ -227,11 +230,15 @@ omrstr_printf(struct OMRPortLibrary *portLibrary, char *buf, uintptr_t bufLen, c
  *  	- OMRPORT_ERROR_STRING_UNSUPPORTED_ENCODING if the input or output encoding is not supported
  *  	- OMRPORT_ERROR_STRING_MEM_ALLOCATE_FAILED if the port library could not allocate a working buffer
  *  	The following translations are supported:
- *  	platform to [wide, modified UTF-8]
- *  	modified UTF-8 to [platform, wide]
- *  	wide to [modified UTF-8, platform]
+ *  	ANSI code page to modified UTF-8 (Windows only)
+ *  	platform raw to [wide, modified UTF-8]
+ *  	modified UTF-8 to [platform raw, wide]
+ *  	[Latin-1, UTF-8] to modified UTF-8
+ *  	wide to [modified UTF-8, platform raw]
  *  	[ISO Latin-1 (8859-1), UTF-8, Windows ANSI default and current code pages] to modified UTF-8
- *
+ * @note J9STR_CODE_PLATFORM_OMR_INTERNAL is an alias for other encodings depending on the platform.
+ * @note J9STR_CODE_PLATFORM is deprecated. Use J9STR_CODE_PLATFORM_OMR_INTERNAL for results of system calls such as getenv (see stdlib.h) and
+ * @note J9STR_CODE_PLATFORM_RAW where the port library does not do implicit translation.
  */
 #ifndef OS_ENCODING_CODE_PAGE
 /* placeholder on non-Windows systems */
@@ -244,7 +251,7 @@ omrstr_convert(struct OMRPortLibrary *portLibrary, int32_t fromCode, int32_t toC
 	int32_t result = OMRPORT_ERROR_STRING_UNSUPPORTED_ENCODING;
 
 	switch (fromCode) {
-	case J9STR_CODE_PLATFORM: {
+	case J9STR_CODE_PLATFORM_RAW: {
 		switch (toCode) {
 		case J9STR_CODE_MUTF8:
 			result = convertPlatformToMutf8(portLibrary, OS_ENCODING_CODE_PAGE, inBuffer, inBufferSize, outBuffer, outBufferSize);
@@ -274,7 +281,7 @@ omrstr_convert(struct OMRPortLibrary *portLibrary, int32_t fromCode, int32_t toC
 #endif
 	case J9STR_CODE_MUTF8: {
 		switch (toCode) {
-		case J9STR_CODE_PLATFORM:
+		case J9STR_CODE_PLATFORM_RAW:
 			result = convertMutf8ToPlatform(portLibrary, inBuffer, inBufferSize, outBuffer, outBufferSize);
 			break;
 		case J9STR_CODE_WIDE: {
@@ -319,7 +326,7 @@ omrstr_convert(struct OMRPortLibrary *portLibrary, int32_t fromCode, int32_t toC
 	break;
 	case J9STR_CODE_WIDE: {
 		switch (toCode) {
-		case J9STR_CODE_PLATFORM:
+		case J9STR_CODE_PLATFORM_RAW:
 			result = OMRPORT_ERROR_STRING_UNSUPPORTED_ENCODING;
 			break;
 		case J9STR_CODE_MUTF8: {

port/unix/omrsyslog.c

@@ -128,7 +128,7 @@ get_conv_txt_num_bytes(struct OMRPortLibrary *portLibrary, const char *message)
 
 	const int32_t convRes =
 		portLibrary->str_convert(portLibrary, J9STR_CODE_MUTF8,
-								 J9STR_CODE_PLATFORM, message, strlen(message),
+								 J9STR_CODE_PLATFORM_RAW, message, strlen(message),
 								 NULL, 0U);
 
 	if (0 < convRes) {
@@ -171,7 +171,7 @@ syslog_write(struct OMRPortLibrary *portLibrary, const char *buf, int priority)
 			 */
 			const int32_t convRes =
 				portLibrary->str_convert(portLibrary,
-										 J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM,
+										 J9STR_CODE_MUTF8, J9STR_CODE_PLATFORM_RAW,
 										 buf, strlen(buf), lclMsg, lclMsgLen);
 
 			lclConvSuccess = (BOOLEAN)(convRes >= 0);

port/zos390/omrgetjobid.c

@@ -64,11 +64,11 @@ omrget_jobid(struct OMRPortLibrary *portLibrary, char *jobid, uintptr_t length)
 
 			if (NULL != jsab_ptr) {
 				/* Job ID is JSABJBID field in JSAB control block. Convert from EBCDIC to internal UTF8 */
-				result = omrstr_convert(portLibrary, J9STR_CODE_PLATFORM, J9STR_CODE_MUTF8,
+				result = omrstr_convert(portLibrary, J9STR_CODE_PLATFORM_RAW, J9STR_CODE_MUTF8,
 									   jsab_ptr->jsabjbid, sizeof(jsab_ptr->jsabjbid), jobid, length);
 				if (OMRPORT_ERROR_STRING_BUFFER_TOO_SMALL == result) {
 					/* return the number of bytes required to hold the converted text */
-					return omrstr_convert(portLibrary, J9STR_CODE_PLATFORM, J9STR_CODE_MUTF8,
+					return omrstr_convert(portLibrary, J9STR_CODE_PLATFORM_RAW, J9STR_CODE_MUTF8,
 										 jsab_ptr->jsabjbid, sizeof(jsab_ptr->jsabjbid), NULL, 0);
 				}
 			}