Added ability to configure string converters

10 files changed, 419 insertions, 37 deletions

diff --git a/cpp/include/Ice/BasicStream.h b/cpp/include/Ice/BasicStream.h
index e87d85b0082..fbcb0d467bb 100644
--- a/cpp/include/Ice/BasicStream.h
+++ b/cpp/include/Ice/BasicStream.h
@@ -16,6 +16,7 @@
 #include <Ice/ObjectFactoryF.h>
 #include <Ice/Buffer.h>
 #include <Ice/Protocol.h>
+#include <Ice/StringConverter.h>
 #include <IceUtil/AutoArray.h>
 #include <IceUtil/Unicode.h>
 
@@ -33,6 +34,47 @@ class ICE_API BasicStream : public Buffer
 {
 public:
 
+    class UTF8BufferI : public Ice::UTF8Buffer
+    {
+    public:
+
+        UTF8BufferI(BasicStream& stream) : 
+	    _stream(stream)
+	{
+	}
+
+        Ice::Byte*
+        getMoreBytes(size_t howMany, Ice::Byte* firstUnused)
+        {
+	    assert(howMany > 0);
+
+            if(firstUnused != 0)
+            {
+                //
+                // Return unused bytes
+                //
+                _stream.b.resize(firstUnused - _stream.b.begin());
+            }
+
+            //
+            // Index of first unused byte
+            //
+            Container::size_type pos = _stream.b.size();
+
+            //
+            // Since resize may reallocate the buffer, when firstUnused != 0, the
+            // return value can be != firstUnused
+            //
+            _stream.resize(pos + howMany);
+
+            return &_stream.b[pos];
+        }
+
+    private:
+
+	BasicStream& _stream;
+    };
+
     typedef void (*PatchFunc)(void*, Ice::ObjectPtr&);
 
     BasicStream(Instance*);
@@ -241,6 +283,31 @@ public:
             write(static_cast<Ice::Byte>(v));
         }
     }
+    void rewriteSize(Ice::Int v, Container::iterator dest)
+    {
+        assert(v >= 0);
+        if(v > 254)
+        {
+            *dest++ = Ice::Byte(255);
+#ifdef ICE_BIG_ENDIAN
+            const Ice::Byte* src = reinterpret_cast<const Ice::Byte*>(&v) + sizeof(Ice::Int) - 1;
+            *dest++ = *src--;
+            *dest++ = *src--;
+            *dest++ = *src--;
+            *dest = *src;
+#else
+            const Ice::Byte* src = reinterpret_cast<const Ice::Byte*>(&v);
+            *dest++ = *src++;
+            *dest++ = *src++;
+            *dest++ = *src++;
+            *dest = *src;
+#endif
+        }
+        else
+        {
+            *dest = static_cast<Ice::Byte>(v);
+        }
+    }
     void readSize(Ice::Int& v) // Inlined for performance reasons.
     {
         Ice::Byte byte;
@@ -405,16 +472,24 @@ public:
     //
     void write(const char*);
 
+    void writeConverted(const std::string& v);
     void write(const std::string& v)
     {
         Ice::Int sz = static_cast<Ice::Int>(v.size());
-        writeSize(sz);
-        if(sz > 0)
-        {
-            Container::size_type pos = b.size();
-            resize(pos + sz);
-            memcpy(&b[pos], v.c_str(), sz);
-        }
+        if(sz > 0 && _stringConverter)
+	{
+	    writeConverted(v);
+	}
+	else
+	{
+            writeSize(sz);
+            if(sz > 0)
+            {
+                Container::size_type pos = b.size();
+                resize(pos + sz);
+                memcpy(&b[pos], v.data(), sz);
+            }
+	}
     }
     void write(const std::string*, const std::string*);
     void read(std::string& v)
@@ -427,8 +502,15 @@ public:
             {
                 throwUnmarshalOutOfBoundsException(__FILE__, __LINE__);
             }
-            std::string(reinterpret_cast<const char*>(&*i), reinterpret_cast<const char*>(&*i) + sz).swap(v);
-//          v.assign(reinterpret_cast<const char*>(&(*i)), sz);
+	    if(_stringConverter)
+	    {
+		_stringConverter->fromUTF8(i, i + sz, v);
+	    }
+	    else
+	    {
+                std::string(reinterpret_cast<const char*>(&*i), reinterpret_cast<const char*>(&*i) + sz).swap(v);
+//              v.assign(reinterpret_cast<const char*>(&(*i)), sz);
+	    }
             i += sz;
         }
         else
@@ -438,17 +520,25 @@ public:
     }
     void read(std::vector<std::string>&);
 
+    void writeConverted(const std::wstring& v);
     void write(const std::wstring& v)
     {
-	std::string s = IceUtil::wstringToString(v);
-        Ice::Int sz = static_cast<Ice::Int>(s.size());
-        writeSize(sz);
-        if(sz > 0)
-        {
-            Container::size_type pos = b.size();
-            resize(pos + sz);
-            memcpy(&b[pos], s.c_str(), sz);
-        }
+        if(v.size() > 0 && _wstringConverter)
+	{
+	    writeConverted(v);
+	}
+	else
+	{
+	    std::string s = IceUtil::wstringToString(v);
+            Ice::Int sz = static_cast<Ice::Int>(s.size());
+            writeSize(sz);
+            if(sz > 0)
+            {
+                Container::size_type pos = b.size();
+                resize(pos + sz);
+                memcpy(&b[pos], s.data(), sz);
+            }
+	}
     }
     void write(const std::wstring*, const std::wstring*);
     void read(std::wstring& v)
@@ -461,8 +551,15 @@ public:
             {
                 throwUnmarshalOutOfBoundsException(__FILE__, __LINE__);
             }
-            std::string s(reinterpret_cast<const char*>(&*i), reinterpret_cast<const char*>(&*i) + sz);
-	    IceUtil::stringToWstring(s).swap(v);
+	    if(_wstringConverter)
+	    {
+		_wstringConverter->fromUTF8(i, i + sz, v);
+	    }
+	    else
+	    {
+                std::string s(reinterpret_cast<const char*>(&*i), reinterpret_cast<const char*>(&*i) + sz);
+	        IceUtil::stringToWstring(s).swap(v);
+	    }
             i += sz;
         }
         else
@@ -625,6 +722,8 @@ private:
     bool _sliceObjects;
 
     const Container::size_type _messageSizeMax;
+    const Ice::StringConverterPtr _stringConverter;
+    const Ice::WstringConverterPtr _wstringConverter;
 
     struct SeqData
     {
diff --git a/cpp/include/Ice/Initialize.h b/cpp/include/Ice/Initialize.h
index f76eadb8fce..824d3f5f985 100644
--- a/cpp/include/Ice/Initialize.h
+++ b/cpp/include/Ice/Initialize.h
@@ -16,6 +16,7 @@
 #include <Ice/LoggerF.h>
 #include <Ice/StreamF.h>
 #include <Ice/StatsF.h>
+#include <Ice/StringConverter.h>
 #include <Ice/BuiltinSequences.h>
 
 namespace Ice
@@ -42,6 +43,8 @@ struct InitializationData
     LoggerPtr logger;
     StatsPtr stats;
     Context defaultContext;
+    StringConverterPtr stringConverter;
+    WstringConverterPtr wstringConverter;
 };
 
 ICE_API CommunicatorPtr initialize(int&, char*[], InitializationData = InitializationData(),
diff --git a/cpp/src/Ice/BasicStream.cpp b/cpp/src/Ice/BasicStream.cpp
index bcef19cc5ee..77a48fe5eac 100644
--- a/cpp/src/Ice/BasicStream.cpp
+++ b/cpp/src/Ice/BasicStream.cpp
@@ -29,6 +29,8 @@ using namespace IceInternal;
 
 IceInternal::BasicStream::BasicStream(Instance* instance) :
     _instance(instance),
+    _stringConverter(instance->initializationData().stringConverter),
+    _wstringConverter(instance->initializationData().wstringConverter),
     _currentReadEncaps(0),
     _currentWriteEncaps(0),
     _traceSlicing(-1),
@@ -1318,6 +1320,64 @@ IceInternal::BasicStream::write(const char*)
 */
 
 void
+IceInternal::BasicStream::writeConverted(const string& v)
+{
+    //
+    // What is the size of the resulting UTF-8 encoded string?
+    // Impossible to tell, so we guess. If we don't guess correctly,
+    // we'll have to fix the mistake afterwards
+    //
+	
+    Int guessedSize = static_cast<Int>(v.size());
+    writeSize(guessedSize); // writeSize() only writes the size; it does not reserve any buffer space.
+	
+    size_t firstIndex = b.size();
+    UTF8BufferI buffer(*this);
+	    
+    Byte* lastByte = _stringConverter->toUTF8(v.data(), v.data() + v.size(), buffer);
+    if(lastByte != b.end())
+    {
+        b.resize(lastByte - b.begin());
+    }
+    size_t lastIndex = b.size();
+	
+    Int actualSize = static_cast<Int>(lastIndex - firstIndex);
+	
+    //
+    // Check against the guess
+    //
+    if(guessedSize != actualSize)
+    {
+        if(guessedSize <= 254 && actualSize > 254)
+        {
+    	    //
+    	    // Move the UTF-8 sequence 4 bytes further
+    	    // Use memmove instead of memcpy since the source and destination typically overlap.
+    	    //
+    	    resize(b.size() + 4);
+    	    std::memmove(b.begin() + firstIndex + 4, b.begin() + firstIndex, actualSize);
+        }
+        else if(guessedSize > 254 && actualSize <= 254)
+        {
+    	    //
+    	    // Move the UTF-8 sequence 4 bytes back
+    	    //
+    	    std::memmove(b.begin() + firstIndex - 4, b.begin() + firstIndex, actualSize);
+    	    resize(b.size() - 4);
+        }
+    
+        if(guessedSize <= 254)
+        {
+    	    rewriteSize(actualSize, b.begin() + firstIndex - 1);
+        }
+        else
+        {
+    	    rewriteSize(actualSize, b.begin() + firstIndex - 1 - 4);
+        }
+    }
+}
+
+void
 IceInternal::BasicStream::write(const string* begin, const string* end)
 {
     Int sz = static_cast<Int>(end - begin);
@@ -1355,6 +1415,64 @@ IceInternal::BasicStream::read(vector<string>& v)
 }
 
 void
+IceInternal::BasicStream::writeConverted(const wstring& v)
+{
+    //
+    // What is the size of the resulting UTF-8 encoded string?
+    // Impossible to tell, so we guess. If we don't guess correctly,
+    // we'll have to fix the mistake afterwards
+    //
+	
+    Int guessedSize = static_cast<Int>(v.size());
+    writeSize(guessedSize); // writeSize() only writes the size; it does not reserve any buffer space.
+	
+    size_t firstIndex = b.size();
+    UTF8BufferI buffer(*this);
+	    
+    Byte* lastByte = _wstringConverter->toUTF8(v.data(), v.data() + v.size(), buffer);
+    if(lastByte != b.end())
+    {
+        b.resize(lastByte - b.begin());
+    }
+    size_t lastIndex = b.size();
+	
+    Int actualSize = static_cast<Int>(lastIndex - firstIndex);
+	
+    //
+    // Check against the guess
+    //
+    if(guessedSize != actualSize)
+    {
+        if(guessedSize <= 254 && actualSize > 254)
+        {
+    	    //
+    	    // Move the UTF-8 sequence 4 bytes further
+    	    // Use memmove instead of memcpy since the source and destination typically overlap.
+    	    //
+    	    resize(b.size() + 4);
+    	    std::memmove(b.begin() + firstIndex + 4, b.begin() + firstIndex, actualSize);
+        }
+        else if(guessedSize > 254 && actualSize <= 254)
+        {
+    	    //
+    	    // Move the UTF-8 sequence 4 bytes back
+    	    //
+    	    std::memmove(b.begin() + firstIndex - 4, b.begin() + firstIndex, actualSize);
+    	    resize(b.size() - 4);
+        }
+    
+        if(guessedSize <= 254)
+        {
+    	    rewriteSize(actualSize, b.begin() + firstIndex - 1);
+        }
+        else
+        {
+    	    rewriteSize(actualSize, b.begin() + firstIndex - 1 - 4);
+        }
+    }
+}
+
+void
 IceInternal::BasicStream::write(const wstring* begin, const wstring* end)
 {
     Int sz = static_cast<Int>(end - begin);
diff --git a/cpp/test/Ice/custom/Client.cpp b/cpp/test/Ice/custom/Client.cpp
index 510eaa37d10..6bd9210dc90 100644
--- a/cpp/test/Ice/custom/Client.cpp
+++ b/cpp/test/Ice/custom/Client.cpp
@@ -10,6 +10,7 @@
 #include <Ice/Ice.h>
 #include <TestCommon.h>
 #include <Test.h>
+#include <StringConverterI.h>
 
 using namespace std;
 
@@ -31,7 +32,10 @@ main(int argc, char** argv)
 
     try
     {
-        communicator = Ice::initialize(argc, argv);
+        Ice::InitializationData initData;
+	initData.stringConverter = new Test::StringConverterI();
+	initData.wstringConverter = new Test::WstringConverterI();
+        communicator = Ice::initialize(argc, argv, initData);
         status = run(argc, argv, communicator);
     }
     catch(const Ice::Exception& ex)
diff --git a/cpp/test/Ice/custom/Collocated.cpp b/cpp/test/Ice/custom/Collocated.cpp
index fc60bbaeae7..084a7999ec6 100644
--- a/cpp/test/Ice/custom/Collocated.cpp
+++ b/cpp/test/Ice/custom/Collocated.cpp
@@ -11,6 +11,7 @@
 #include <TestCommon.h>
 #include <TestI.h>
 #include <WstringI.h>
+#include <StringConverterI.h>
 
 using namespace std;
 
@@ -37,10 +38,12 @@ main(int argc, char** argv)
 
     try
     {
-        Ice::PropertiesPtr properties = Ice::getDefaultProperties(argc, argv);
-	properties->setProperty("TestAdapter.Endpoints", "default -p 12010 -t 10000");
-
-        communicator = Ice::initialize(argc, argv);
+        Ice::InitializationData initData;
+	initData.properties = Ice::getDefaultProperties(argc, argv);
+	initData.properties->setProperty("TestAdapter.Endpoints", "default -p 12010 -t 10000");
+	initData.stringConverter = new Test::StringConverterI();
+	initData.wstringConverter = new Test::WstringConverterI();
+        communicator = Ice::initialize(argc, argv, initData);
         status = run(argc, argv, communicator);
     }
     catch(const Ice::Exception& ex)
diff --git a/cpp/test/Ice/custom/Makefile b/cpp/test/Ice/custom/Makefile
index 72e5202de8e..dada88e4294 100644
--- a/cpp/test/Ice/custom/Makefile
+++ b/cpp/test/Ice/custom/Makefile
@@ -20,21 +20,24 @@ COBJS		= Test.o \
 		  Wstring.o \
 		  Client.o \
 		  AllTests.o \
-		  MyByteSeq.o
+		  MyByteSeq.o \
+		  StringConverterI.o
 
 SOBJS		= Test.o \
 		  Wstring.o \
 		  TestI.o \
 		  WstringI.o \
 		  Server.o \
-		  MyByteSeq.o
+		  MyByteSeq.o \
+		  StringConverterI.o
 
 SAMDOBJS	= TestAMD.o \
 		  TestAMDI.o \
 		  WstringAMD.o \
 		  WstringAMDI.o \
 		  ServerAMD.o \
-		  MyByteSeq.o
+		  MyByteSeq.o \
+		  StringConverterI.o
 
 COLOBJS		= Test.o \
 		  Wstring.o \
@@ -42,7 +45,8 @@ COLOBJS		= Test.o \
 		  WstringI.o \
 		  Collocated.o \
 		  AllTests.o \
-		  MyByteSeq.o
+		  MyByteSeq.o \
+		  StringConverterI.o
 
 SRCS		= $(COBJS:.o=.cpp) \
 		  $(SOBJS:.o=.cpp) \
diff --git a/cpp/test/Ice/custom/Server.cpp b/cpp/test/Ice/custom/Server.cpp
index 5bb712444b6..92188346034 100644
--- a/cpp/test/Ice/custom/Server.cpp
+++ b/cpp/test/Ice/custom/Server.cpp
@@ -11,6 +11,7 @@
 #include <TestCommon.h>
 #include <TestI.h>
 #include <WstringI.h>
+#include <StringConverterI.h>
 
 using namespace std;
 
@@ -36,10 +37,12 @@ main(int argc, char** argv)
 
     try
     {
-        Ice::PropertiesPtr properties = Ice::getDefaultProperties(argc, argv);
-	properties->setProperty("TestAdapter.Endpoints", "default -p 12010 -t 10000");
-
-        communicator = Ice::initialize(argc, argv);
+        Ice::InitializationData initData;
+	initData.properties = Ice::getDefaultProperties(argc, argv);
+	initData.properties->setProperty("TestAdapter.Endpoints", "default -p 12010 -t 10000");
+	initData.stringConverter = new Test::StringConverterI();
+	initData.wstringConverter = new Test::WstringConverterI();
+        communicator = Ice::initialize(argc, argv, initData);
         status = run(argc, argv, communicator);
     }
     catch(const Ice::Exception& ex)
diff --git a/cpp/test/Ice/custom/ServerAMD.cpp b/cpp/test/Ice/custom/ServerAMD.cpp
index 21608202898..b3561cf14aa 100644
--- a/cpp/test/Ice/custom/ServerAMD.cpp
+++ b/cpp/test/Ice/custom/ServerAMD.cpp
@@ -11,6 +11,7 @@
 #include <TestCommon.h>
 #include <TestAMDI.h>
 #include <WstringAMDI.h>
+#include <StringConverterI.h>
 
 using namespace std;
 
@@ -36,10 +37,12 @@ main(int argc, char** argv)
 
     try
     {
-        Ice::PropertiesPtr properties = Ice::getDefaultProperties(argc, argv);
-	properties->setProperty("TestAdapter.Endpoints", "default -p 12010 -t 10000");
-
-        communicator = Ice::initialize(argc, argv);
+        Ice::InitializationData initData;
+	initData.properties = Ice::getDefaultProperties(argc, argv);
+	initData.properties->setProperty("TestAdapter.Endpoints", "default -p 12010 -t 10000");
+	initData.stringConverter = new Test::StringConverterI();
+	initData.wstringConverter = new Test::WstringConverterI();
+        communicator = Ice::initialize(argc, argv, initData);
         status = run(argc, argv, communicator);
     }
     catch(const Ice::Exception& ex)
diff --git a/cpp/test/Ice/custom/StringConverterI.cpp b/cpp/test/Ice/custom/StringConverterI.cpp
new file mode 100644
index 00000000000..7187194249b
--- /dev/null
+++ b/cpp/test/Ice/custom/StringConverterI.cpp
@@ -0,0 +1,102 @@
+// **********************************************************************
+//
+// Copyright (c) 2003-2006 ZeroC, Inc. All rights reserved.
+//
+// This copy of Ice is licensed to you under the terms described in the
+// ICE_LICENSE file included in this distribution.
+//
+// **********************************************************************
+
+#include <StringConverterI.h>
+#include <IceUtil/Unicode.h>
+
+Ice::Byte*
+Test::StringConverterI::toUTF8(const char* sourceStart, const char* sourceEnd, Ice::UTF8Buffer& buffer) const
+{
+    size_t size = static_cast<size_t>(sourceEnd - sourceStart);
+    Ice::Byte* targetStart = buffer.getMoreBytes(size, 0);
+    Ice::Byte* targetEnd = targetStart + size;
+
+    char* p = const_cast<char*>(sourceEnd);
+    for(unsigned int i = 0; i < size; ++i)
+    {
+        targetStart[i] = *(--p);
+    }
+    return targetEnd;
+}
+
+void
+Test::StringConverterI::fromUTF8(const Ice::Byte* sourceStart, const Ice::Byte* sourceEnd, 
+				 const char*& targetStart, const char*& targetEnd) const
+{
+    size_t size = static_cast<size_t>(sourceEnd - sourceStart);
+    char* buf = new char[size];
+
+    Ice::Byte* p = const_cast<Ice::Byte*>(sourceEnd);
+    for(unsigned int i = 0; i < size; ++i)
+    {
+        buf[i] = *(--p);
+    }
+
+    targetStart = buf;
+    targetEnd = targetStart + size;
+}
+
+void 
+Test::StringConverterI::freeTarget(const char* target) const
+{
+    delete[] target;
+}
+
+Ice::Byte*
+Test::WstringConverterI::toUTF8(const wchar_t* sourceStart, const wchar_t* sourceEnd, Ice::UTF8Buffer& buffer) const
+{
+    std::wstring ws(sourceStart, sourceEnd);
+    std::string s = IceUtil::wstringToString(ws);
+
+    size_t size = s.size();
+    Ice::Byte* targetStart = buffer.getMoreBytes(size, 0);
+    Ice::Byte* targetEnd = targetStart + size;
+
+    char* p = const_cast<char*>(s.c_str() + size);
+    for(unsigned int i = 0; i < size; ++i)
+    {
+        targetStart[i] = static_cast<Ice::Byte>(*(--p));
+    }
+    return targetEnd;
+}
+
+void
+Test::WstringConverterI::fromUTF8(const Ice::Byte* sourceStart, const Ice::Byte* sourceEnd, 
+				  const wchar_t*& targetStart, const wchar_t*& targetEnd) const
+{
+    size_t size = static_cast<size_t>(sourceEnd - sourceStart);
+    std::string s(sourceStart, sourceEnd);
+
+    Ice::Byte* p = const_cast<Ice::Byte*>(sourceEnd);
+    for(unsigned int i = 0; i < size; ++i)
+    {
+        s[i] = *(--p);
+    }
+
+    std::wstring ws = IceUtil::stringToWstring(s);
+    size = ws.size();
+    wchar_t* buf = new wchar_t[size];
+    for(unsigned int i = 0; i < size; ++i)
+    {
+        buf[i] = ws[i];
+    }
+
+    targetStart = buf;
+    targetEnd = targetStart + size;
+}
+
+void 
+Test::WstringConverterI::freeTarget(const wchar_t* target) const
+{
+#if defined(_MSC_VER) && _MSC_VER < 1300
+    delete[] const_cast<wchar_t*>(target);
+#else
+    delete[] target;
+#endif
+}
diff --git a/cpp/test/Ice/custom/StringConverterI.h b/cpp/test/Ice/custom/StringConverterI.h
new file mode 100644
index 00000000000..48eae68b3be
--- /dev/null
+++ b/cpp/test/Ice/custom/StringConverterI.h
@@ -0,0 +1,43 @@
+// **********************************************************************
+//
+// Copyright (c) 2003-2006 ZeroC, Inc. All rights reserved.
+//
+// This copy of Ice is licensed to you under the terms described in the
+// ICE_LICENSE file included in this distribution.
+//
+// **********************************************************************
+
+#ifndef STRING_CONVERTER_I_H
+#define STRING_CONVERTER_I_H
+
+#include <Ice/StringConverter.h>
+
+//
+// Simple contrived string converters which simply reverse the order of the
+// characters being sent.
+//
+
+namespace Test
+{
+
+class StringConverterI : public Ice::StringConverter
+{
+public:
+
+    virtual Ice::Byte* toUTF8(const char*, const char*, Ice::UTF8Buffer&) const;
+    virtual void fromUTF8(const Ice::Byte*, const Ice::Byte*, const char*&, const char*&) const;
+    virtual void freeTarget(const char*) const;
+};
+
+class WstringConverterI : public Ice::WstringConverter
+{
+public:
+
+    virtual Ice::Byte* toUTF8(const wchar_t*, const wchar_t*, Ice::UTF8Buffer&) const;
+    virtual void fromUTF8(const Ice::Byte*, const Ice::Byte*, const wchar_t*&, const wchar_t*&) const;
+    virtual void freeTarget(const wchar_t*) const;
+};
+
+}
+
+#endif