add utf-16 to utf-8 conversion functions to vwdll api for use from c# .net

  the existing functions were modified and a new FcnA form added that
  retains the current ansi/utf8 access

1 files changed, 52 insertions, 4 deletions

diff --git a/vowpalwabbit/vwdll.cpp b/vowpalwabbit/vwdll.cpp
index 1616ffcf..7fb6eed1 100644
--- a/vowpalwabbit/vwdll.cpp
+++ b/vowpalwabbit/vwdll.cpp
@@ -1,12 +1,34 @@
 #include <memory>
+#include <codecvt>
+#include <locale>
+#include <string>
 
 #include "vwdll.h"
 #include "parser.h"
 
+// This interface now provides "wide" functions for compatibility with .NET interop
+// The default functions assume a wide (16 bit char pointer) that is converted to a utf8-string and passed to
+// a function which takes a narrow (8 bit char pointer) function. Both are exposed in the c/c++ API 
+// so that programs using 8 bit wide characters can use the direct call without conversion and 
+//  programs using 16 bit characters can use the default wide versions of the functions.
+// "Ansi versions  (FcnA instead of Fcn) have only been written for functions which handle strings.
+
+// a future optimization would be to write an inner version of hash feature which either hashed the
+// wide string directly (and live with the different hash values) or incorporate the UTF-16 to UTF-8 conversion
+// in the hashing to avoid allocating an intermediate string.
+
 extern "C"
 {
+
+	VW_DLL_MEMBER VW_HANDLE VW_CALLING_CONV VW_Initialize(const char16_t * pstrArgs)
+	{
+		std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> convert;
+		std::string sa(convert.to_bytes(pstrArgs));
+		return VW_InitializeA(sa.c_str());
+	}
+
 	
-	VW_DLL_MEMBER VW_HANDLE VW_CALLING_CONV VW_Initialize(const char * pstrArgs)
+	VW_DLL_MEMBER VW_HANDLE VW_CALLING_CONV VW_InitializeA(const char * pstrArgs)
 	{
 		std::auto_ptr<vw> inst(new vw);
 		try
@@ -59,7 +81,17 @@ extern "C"
 		}
 	}
 	
-	VW_DLL_MEMBER VW_EXAMPLE VW_CALLING_CONV VW_ReadExample(VW_HANDLE handle, const char * line)
+
+	VW_DLL_MEMBER VW_EXAMPLE VW_CALLING_CONV VW_ReadExample(VW_HANDLE handle, const char16_t * line)
+	{
+		std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> convert;
+		std::string sa(convert.to_bytes(line));
+		return VW_ReadExampleA(handle, sa.c_str());
+	}
+
+
+
+	VW_DLL_MEMBER VW_EXAMPLE VW_CALLING_CONV VW_ReadExampleA(VW_HANDLE handle, const char * line)
 	{
 		try
 		{
@@ -87,7 +119,14 @@ extern "C"
 		}
 	}
 
-	VW_DLL_MEMBER uint32_t VW_CALLING_CONV VW_HashSpace(VW_HANDLE handle, const char * s)
+	VW_DLL_MEMBER uint32_t VW_CALLING_CONV VW_HashSpace(VW_HANDLE handle, const char16_t * s)
+	{
+		std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> convert;
+		std::string sa(convert.to_bytes(s));
+		return VW_HashSpaceA(handle,sa.c_str());
+	}
+
+	VW_DLL_MEMBER uint32_t VW_CALLING_CONV VW_HashSpaceA(VW_HANDLE handle, const char * s)
 	{
 		try
 		{
@@ -102,7 +141,16 @@ extern "C"
 		}
 	}
 
-	VW_DLL_MEMBER uint32_t VW_CALLING_CONV VW_HashFeature(VW_HANDLE handle, const char * s, unsigned long u)
+
+	VW_DLL_MEMBER uint32_t VW_CALLING_CONV VW_HashFeature(VW_HANDLE handle, const char16_t * s, unsigned long u)
+	{
+		std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> convert;
+		std::string sa(convert.to_bytes(s));
+		return VW_HashFeatureA(handle,sa.c_str(),u);
+	}
+
+
+	VW_DLL_MEMBER uint32_t VW_CALLING_CONV VW_HashFeatureA(VW_HANDLE handle, const char * s, unsigned long u)
 	{
 		try
 		{