Ensure that the encoding and nencoding values are respected when the OCI

environment is created; ensure that these values are also used when LOBs are
read and written.

Environment.c

@@ -11,9 +11,11 @@ typedef struct {
     OCIEnv *handle;
     OCIError *errorHandle;
     int maxBytesPerCharacter;
-    int fixedWidth;
+    int nmaxBytesPerCharacter;
     char *encoding;
     char *nencoding;
+    ub2 charsetId;
+    ub2 ncharsetId;
     PyObject *cloneEnv;
     udt_Buffer numberToStringFormatBuffer;
     udt_Buffer numberFromStringFormatBuffer;
@@ -73,8 +75,8 @@ static udt_Environment *Environment_New(
     if (!env)
         return NULL;
     env->handle = handle;
-    env->fixedWidth = 1;
     env->maxBytesPerCharacter = 1;
+    env->nmaxBytesPerCharacter = 4;
     cxBuffer_Init(&env->numberToStringFormatBuffer);
     cxBuffer_Init(&env->numberFromStringFormatBuffer);
     cxBuffer_Init(&env->nlsNumericCharactersBuffer);
@@ -107,12 +109,19 @@ static int Environment_GetCharacterSetName(
     udt_Environment *self,              // environment object
     ub2 attribute,                      // attribute to fetch
     const char *overrideValue,          // override value, if specified
-    char **result)                      // place to store result
+    char **result,                      // place to store result
+    ub2 *charsetId)                     // character set ID (OUT)
 {
     char charsetName[OCI_NLS_MAXBUFSZ], ianaCharsetName[OCI_NLS_MAXBUFSZ];
-    ub2 charsetId;
     sword status;
 
+    // get character set id
+    status = OCIAttrGet(self->handle, OCI_HTYPE_ENV, charsetId, NULL,
+            attribute, self->errorHandle);
+    if (Environment_CheckForError(self, status,
+            "Environment_GetCharacterSetName(): get charset id") < 0)
+        return -1;
+
     // if override value specified, use it
     if (overrideValue) {
         *result = PyMem_Malloc(strlen(overrideValue) + 1);
@@ -122,16 +131,9 @@ static int Environment_GetCharacterSetName(
         return 0;
     }
 
-    // get character set id
-    status = OCIAttrGet(self->handle, OCI_HTYPE_ENV, &charsetId, NULL,
-            attribute, self->errorHandle);
-    if (Environment_CheckForError(self, status,
-            "Environment_GetCharacterSetName(): get charset id") < 0)
-        return -1;
-
     // get character set name
     status = OCINlsCharSetIdToName(self->handle, (text*) charsetName,
-            OCI_NLS_MAXBUFSZ, charsetId);
+            OCI_NLS_MAXBUFSZ, *charsetId);
     if (Environment_CheckForError(self, status,
             "Environment_GetCharacterSetName(): get Oracle charset name") < 0)
         return -1;
@@ -292,23 +294,20 @@ static udt_Environment *Environment_NewFromScratch(
         return NULL;
     }
 
-    // acquire whether character set is fixed width
-    status = OCINlsNumericInfoGet(env->handle, env->errorHandle,
-            &env->fixedWidth, OCI_NLS_CHARSET_FIXEDWIDTH);
-    if (Environment_CheckForError(env, status,
-            "Environment_New(): determine if charset fixed width") < 0) {
-        Py_DECREF(env);
-        return NULL;
-    }
-
     // determine encodings to use for Unicode values
     if (Environment_GetCharacterSetName(env, OCI_ATTR_ENV_CHARSET_ID,
-            encoding, &env->encoding) < 0)
+            encoding, &env->encoding, &env->charsetId) < 0)
         return NULL;
     if (Environment_GetCharacterSetName(env, OCI_ATTR_ENV_NCHARSET_ID,
-            nencoding, &env->nencoding) < 0)
+            nencoding, &env->nencoding, &env->ncharsetId) < 0)
         return NULL;
 
+    // max bytes per character for NCHAR can be assigned only if it matches the
+    // character set used for CHAR data; OCI does not provide a way of
+    // determining it otherwise
+    if (env->ncharsetId == env->charsetId)
+        env->nmaxBytesPerCharacter = env->maxBytesPerCharacter;
+
     // fill buffers for number formats
     if (Environment_SetBuffer(&env->numberToStringFormatBuffer, "TM9",
             env->encoding) < 0)
@@ -339,11 +338,12 @@ static udt_Environment *Environment_Clone(
     if (!env)
         return NULL;
     env->maxBytesPerCharacter = cloneEnv->maxBytesPerCharacter;
-    env->fixedWidth = cloneEnv->fixedWidth;
     Py_INCREF(cloneEnv);
     env->cloneEnv = (PyObject*) cloneEnv;
     env->encoding = cloneEnv->encoding;
     env->nencoding = cloneEnv->nencoding;
+    env->charsetId = cloneEnv->charsetId;
+    env->ncharsetId = cloneEnv->ncharsetId;
     cxBuffer_Copy(&env->numberToStringFormatBuffer,
             &cloneEnv->numberToStringFormatBuffer);
     cxBuffer_Copy(&env->numberFromStringFormatBuffer,

ExternalLobVar.c

@@ -169,17 +169,13 @@ static int ExternalLobVar_InternalRead(
     oraub8 *length,                     // length of data (IN/OUT)
     oraub8 offset)                      // offset
 {
-    oraub8 lengthInBytes, lengthInChars;
+    oraub8 lengthInBytes = 0, lengthInChars = 0;
     ub2 charsetId;
     sword status;
 
-    if (var->lobVar->type == &vt_NCLOB || var->lobVar->type == &vt_CLOB) {
+    if (var->lobVar->type == &vt_NCLOB || var->lobVar->type == &vt_CLOB)
-        lengthInBytes = 0;
         lengthInChars = *length;
-    } else {
+    else lengthInBytes = *length;
-        lengthInChars = 0;
-        lengthInBytes = *length;
-    }
 
     if (var->lobVar->isFile) {
         Py_BEGIN_ALLOW_THREADS
@@ -193,9 +189,9 @@ static int ExternalLobVar_InternalRead(
     }
 
     Py_BEGIN_ALLOW_THREADS
-    if (var->lobVar->type == &vt_NCLOB)
+    charsetId = (var->lobVar->type->charsetForm == SQLCS_NCHAR) ?
-        charsetId = OCI_UTF16ID;
+            var->lobVar->environment->ncharsetId :
-    else charsetId = 0;
+            var->lobVar->environment->charsetId;
     status = OCILobRead2(var->lobVar->connection->handle,
             var->lobVar->environment->errorHandle, var->lobVar->data[var->pos],
             &lengthInBytes, &lengthInChars, offset, buffer, bufferSize,
@@ -275,10 +271,11 @@ static PyObject *ExternalLobVar_Value(
         else amount = 1;
     }
     length = amount;
+
     if (var->lobVar->type == &vt_CLOB)
         bufferSize = amount * var->lobVar->environment->maxBytesPerCharacter;
     else if (var->lobVar->type == &vt_NCLOB)
-        bufferSize = amount * 2;
+        bufferSize = amount * var->lobVar->environment->nmaxBytesPerCharacter;
     else bufferSize = amount;
 
     // create a string for retrieving the value
@@ -296,7 +293,8 @@ static PyObject *ExternalLobVar_Value(
         result = cxString_FromEncodedString(buffer, length,
                 var->lobVar->environment->encoding);
     } else if (var->lobVar->type == &vt_NCLOB) {
-        result = PyUnicode_DecodeUTF16(buffer, length, NULL, NULL);
+        result = PyUnicode_Decode(buffer, length,
+                var->lobVar->environment->encoding, NULL);
     } else {
         result = PyBytes_FromStringAndSize(buffer, length);
     }
@@ -413,7 +411,7 @@ static PyObject *ExternalLobVar_Str(
 
 //-----------------------------------------------------------------------------
 // ExternalLobVar_Write()
-//   Write a value to the LOB variable; return the number of bytes written.
+//   Write a value to the LOB at the specified offset.
 //-----------------------------------------------------------------------------
 static PyObject *ExternalLobVar_Write(
     udt_ExternalLobVar *var,            // variable to perform write against
@@ -421,8 +419,8 @@ static PyObject *ExternalLobVar_Write(
     PyObject *keywordArgs)              // keyword arguments
 {
     static char *keywordList[] = { "data", "offset", NULL };
-    oraub8 amount, offset;
     PyObject *dataObj;
+    oraub8 offset;
 
     // buffer is expected, offset is optional
     offset = 1;
@@ -433,11 +431,10 @@ static PyObject *ExternalLobVar_Write(
     // perform the write, if possible
     if (ExternalLobVar_Verify(var) < 0)
         return NULL;
-    if (LobVar_Write(var->lobVar, var->pos, dataObj, offset, &amount) < 0)
+    if (LobVar_Write(var->lobVar, var->pos, dataObj, offset) < 0)
         return NULL;
 
-    // return the result
+    Py_RETURN_NONE;
-    return PyLong_FromUnsignedLong(amount);
 }
 
 

LobVar.c

@@ -21,7 +21,7 @@ static int LobVar_PreFetch(udt_LobVar*);
 static void LobVar_Finalize(udt_LobVar*);
 static PyObject *LobVar_GetValue(udt_LobVar*, unsigned);
 static int LobVar_SetValue(udt_LobVar*, unsigned, PyObject*);
-static int LobVar_Write(udt_LobVar*, unsigned, PyObject*, oraub8, oraub8*);
+static int LobVar_Write(udt_LobVar*, unsigned, PyObject*, oraub8);
 
 //-----------------------------------------------------------------------------
 // Python type declarations
@@ -318,49 +318,44 @@ static int LobVar_Write(
     udt_LobVar *var,                    // variable to perform write against
     unsigned position,                  // position to perform write against
     PyObject *dataObj,                  // data object to write into LOB
-    oraub8 offset,                      // offset into variable
+    oraub8 offset)                      // offset into variable
-    oraub8 *amount)                     // amount to write
 {
+    oraub8 lengthInBytes, lengthInChars = 0;
+    const char *encoding;
     udt_Buffer buffer;
+    ub2 charsetId;
     sword status;
 
     // verify the data type
     if (var->type == &vt_BFILE) {
         PyErr_SetString(PyExc_TypeError, "BFILEs are read only");
         return -1;
-    } else if (var->type == &vt_BLOB) {
-        if (cxBuffer_FromObject(&buffer, dataObj,
-                var->environment->encoding) < 0)
-            return -1;
-        *amount = buffer.size;
-#if PY_MAJOR_VERSION < 3
-    } else if (var->type == &vt_NCLOB) {
-        if (cxBuffer_FromObject(&buffer, dataObj,
-                var->environment->nencoding) < 0)
-            return -1;
-        *amount = buffer.size;
-#endif
-    } else {
-        if (cxBuffer_FromObject(&buffer, dataObj,
-                var->environment->encoding) < 0)
-            return -1;
-        if (var->environment->fixedWidth
-                && var->environment->maxBytesPerCharacter > 1)
-            *amount = buffer.size / var->environment->maxBytesPerCharacter;
-        else *amount = buffer.size;
     }
 
+    // determine the buffer to write
+    if (var->type->charsetForm == SQLCS_NCHAR) {
+        charsetId = var->environment->ncharsetId;
+        encoding = var->environment->nencoding;
+    } else {
+        charsetId = var->environment->charsetId;
+        encoding = var->environment->encoding;
+    }
+    if (cxBuffer_FromObject(&buffer, dataObj, encoding) < 0)
+        return -1;
+    lengthInBytes = buffer.size;
+
     // nothing to do if no data to write
-    if (*amount == 0) {
+    if (lengthInBytes == 0) {
         cxBuffer_Clear(&buffer);
         return 0;
     }
 
+    // write the data with the correct character set
     Py_BEGIN_ALLOW_THREADS
     status = OCILobWrite2(var->connection->handle,
-            var->environment->errorHandle, var->data[position], amount, 0,
+            var->environment->errorHandle, var->data[position], &lengthInBytes,
-            offset, (void*) buffer.ptr, buffer.size, OCI_ONE_PIECE, NULL, NULL,
+            &lengthInChars, offset, (void*) buffer.ptr, buffer.size,
-            0, var->type->charsetForm);
+            OCI_ONE_PIECE, NULL, NULL, charsetId, var->type->charsetForm);
     Py_END_ALLOW_THREADS
     cxBuffer_Clear(&buffer);
     if (Environment_CheckForError(var->environment, status,
@@ -393,7 +388,6 @@ static int LobVar_SetValue(
     PyObject *value)                    // value to set
 {
     boolean isTemporary;
-    oraub8 amount;
     sword status;
     ub1 lobType;
 
@@ -428,6 +422,6 @@ static int LobVar_SetValue(
         return -1;
 
     // set the current value
-    return LobVar_Write(var, position, value, 1, &amount);
+    return LobVar_Write(var, position, value, 1);
 }