Implement VK_HANJA for Korean on Windows.

6 files changed, 294 insertions, 7 deletions

diff --git a/win32/HanjaDic.cxx b/win32/HanjaDic.cxx
new file mode 100644
index 000000000..25105ee6d
--- /dev/null
+++ b/win32/HanjaDic.cxx
@@ -0,0 +1,122 @@
+// Scintilla source code edit control
+/** @file HanjaDic.cxx
+ ** Korean Hanja Dictionary
+ ** Convert between Korean Hanja and Hangul by COM interface.
+ **/
+// Copyright 2015 by Neil Hodgson <neilh@scintilla.org>
+// The License.txt file describes the conditions under which this software may be distributed.
+
+#include <windows.h>
+
+#include "UniConversion.h"
+#include "HanjaDic.h"
+
+#ifdef SCI_NAMESPACE
+using namespace Scintilla;
+#endif
+
+namespace HanjaDict {
+
+interface IRadical;
+interface IHanja;
+interface IStrokes;
+
+typedef enum { HANJA_UNKNOWN = 0, HANJA_K0 = 1, HANJA_K1 = 2, HANJA_OTHER = 3 } HANJA_TYPE;
+
+interface IHanjaDic : IUnknown {
+	STDMETHOD(OpenMainDic)();
+	STDMETHOD(CloseMainDic)();
+	STDMETHOD(GetHanjaWords)(BSTR bstrHangul, SAFEARRAY* ppsaHanja, VARIANT_BOOL* pfFound);
+	STDMETHOD(GetHanjaChars)(unsigned short wchHangul, BSTR* pbstrHanjaChars, VARIANT_BOOL* pfFound);
+	STDMETHOD(HanjaToHangul)(BSTR bstrHanja, BSTR* pbstrHangul);
+	STDMETHOD(GetHanjaType)(unsigned short wchHanja, HANJA_TYPE* pHanjaType);
+	STDMETHOD(GetHanjaSense)(unsigned short wchHanja, BSTR* pbstrSense);
+	STDMETHOD(GetRadicalID)(short SeqNumOfRadical, short* pRadicalID, unsigned short* pwchRadical);
+	STDMETHOD(GetRadical)(short nRadicalID, IRadical** ppIRadical);
+	STDMETHOD(RadicalIDToHanja)(short nRadicalID, unsigned short* pwchRadical);
+	STDMETHOD(GetHanja)(unsigned short wchHanja, IHanja** ppIHanja);
+	STDMETHOD(GetStrokes)(short nStrokes, IStrokes** ppIStrokes);
+	STDMETHOD(OpenDefaultCustomDic)();
+	STDMETHOD(OpenCustomDic)(BSTR bstrPath, long* plUdr);
+	STDMETHOD(CloseDefaultCustomDic)();
+	STDMETHOD(CloseCustomDic)(long lUdr);
+	STDMETHOD(CloseAllCustomDics)();
+	STDMETHOD(GetDefaultCustomHanjaWords)(BSTR bstrHangul, SAFEARRAY** ppsaHanja, VARIANT_BOOL* pfFound);
+	STDMETHOD(GetCustomHanjaWords)(long lUdr, BSTR bstrHangul, SAFEARRAY** ppsaHanja, VARIANT_BOOL* pfFound);
+	STDMETHOD(PutDefaultCustomHanjaWord)(BSTR bstrHangul, BSTR bstrHanja);
+	STDMETHOD(PutCustomHanjaWord)(long lUdr, BSTR bstrHangul, BSTR bstrHanja);
+	STDMETHOD(MaxNumOfRadicals)(short* pVal);
+	STDMETHOD(MaxNumOfStrokes)(short* pVal);
+	STDMETHOD(DefaultCustomDic)(long* pVal);
+	STDMETHOD(DefaultCustomDic)(long pVal);
+	STDMETHOD(MaxHanjaType)(HANJA_TYPE* pHanjaType);
+	STDMETHOD(MaxHanjaType)(HANJA_TYPE pHanjaType);
+};
+
+extern "C" const GUID __declspec(selectany) IID_IHanjaDic =
+{ 0xad75f3ac, 0x18cd, 0x48c6, { 0xa2, 0x7d, 0xf1, 0xe9, 0xa7, 0xdc, 0xe4, 0x32 } };
+
+class HanjaDic {
+private:
+	HRESULT hr;
+	CLSID CLSID_HanjaDic;
+
+public:
+	IHanjaDic *HJinterface;
+
+	HanjaDic() {
+		hr = CLSIDFromProgID(OLESTR("mshjdic.hanjadic"), &CLSID_HanjaDic);
+		if (SUCCEEDED(hr)) {
+			hr = CoCreateInstance(CLSID_HanjaDic, NULL,
+					CLSCTX_INPROC_SERVER, IID_IHanjaDic,
+					(LPVOID *)& HJinterface);
+			if (SUCCEEDED(hr)) {
+				hr = HJinterface->OpenMainDic();
+			}
+		}
+	}
+
+	~HanjaDic() {
+		if (SUCCEEDED(hr)) {
+			hr = HJinterface->CloseMainDic();
+			HJinterface->Release();
+		}
+	}
+
+	bool HJdictAvailable() {
+		return SUCCEEDED(hr);
+	}
+
+	bool IsHanja(int hanja) {
+		HANJA_TYPE hanjaType;
+		hr = HJinterface->GetHanjaType(static_cast<unsigned short>(hanja), &hanjaType);
+		if (SUCCEEDED(hr)) {
+			return (hanjaType > 0);
+		}
+		return false;
+	}
+};
+
+int GetHangulOfHanja(int hanjaChar) {
+	// Convert hanja character to hangul one.
+	int hangulChar = -1; // If fails, return -1.
+	HanjaDic dict;
+	if (!dict.HJdictAvailable() || !dict.IsHanja(hanjaChar)) {
+		return hangulChar;
+	}
+	wchar_t convHnja[UTF8MaxBytes] = {0};
+	convHnja[0] = static_cast<wchar_t>(hanjaChar);
+
+	BSTR bstrHangul = SysAllocString(NULL);
+	BSTR bstrHanja = SysAllocString(convHnja);
+	HRESULT hr = dict.HJinterface->HanjaToHangul(bstrHanja, &bstrHangul);
+	if (SUCCEEDED(hr)) {
+		wchar_t wHangul = static_cast<wchar_t *>(bstrHangul)[0];
+		hangulChar = static_cast<int>(wHangul);
+	}
+	SysFreeString(bstrHangul);
+	SysFreeString(bstrHanja);
+	return hangulChar;
+}
+
+}
diff --git a/win32/HanjaDic.h b/win32/HanjaDic.h
new file mode 100644
index 000000000..5ba61f661
--- /dev/null
+++ b/win32/HanjaDic.h
@@ -0,0 +1,26 @@
+// Scintilla source code edit control
+/** @file HanjaDic.h
+ ** Korean Hanja Dictionary
+ ** Convert between Korean Hanja and Hangul by COM interface.
+ **/
+// Copyright 2015 by Neil Hodgson <neilh@scintilla.org>
+// The License.txt file describes the conditions under which this software may be distributed.
+
+#ifndef HANJADIC_H
+#define HANJADIC_H
+
+#ifdef SCI_NAMESPACE
+namespace Scintilla {
+#endif
+
+namespace HanjaDict {
+
+int GetHangulOfHanja(int character);
+
+}
+
+#ifdef SCI_NAMESPACE
+}
+#endif
+
+#endif
diff --git a/win32/SciLexer.vcxproj b/win32/SciLexer.vcxproj
index 50b2e5f7f..2e821b39a 100644
--- a/win32/SciLexer.vcxproj
+++ b/win32/SciLexer.vcxproj
@@ -111,6 +111,7 @@
     <ClCompile Include="..\lexers\*.cxx" />

     <ClCompile Include="..\lexlib\*.cxx" />

     <ClCompile Include="..\src\*.cxx" />

+    <ClCompile Include="..\win32\HanjaDic.cxx" />

     <ClCompile Include="..\win32\PlatWin.cxx" />

     <ClCompile Include="..\win32\ScintillaWin.cxx" />

   </ItemGroup>

diff --git a/win32/ScintillaWin.cxx b/win32/ScintillaWin.cxx
index 100361a1b..f4e43bb54 100644
--- a/win32/ScintillaWin.cxx
+++ b/win32/ScintillaWin.cxx
@@ -84,6 +84,7 @@
 #endif
 
 #include "PlatWin.h"
+#include "HanjaDic.h"
 
 #ifndef SPI_GETWHEELSCROLLLINES
 #define SPI_GETWHEELSCROLLLINES   104
@@ -251,6 +252,11 @@ class ScintillaWin :
 	void MoveImeCarets(int offset);
 	void DrawImeIndicator(int indicator, int len);
 	void SetCandidateWindowPos();
+	void BytesToUniChar(const char *bytes, const int bytesLen, wchar_t *character, int &charsLen);
+	void UniCharToBytes(const wchar_t *character, const int charsLen, char *bytes, int &bytesLen);
+	void RangeToHangul(int uniStrLen);
+	void EscapeHanja();
+	void ToggleHanja();
 
 	UINT CodePageOfDocument();
 	virtual bool ValidCodePage(int codePage) const;
@@ -802,6 +808,127 @@ void ScintillaWin::SetCandidateWindowPos() {
 	}
 }
 
+void ScintillaWin::BytesToUniChar(const char *bytes, const int bytesLen, wchar_t *characters, int &charsLen) {
+	// Return results over characters and charsLen.
+	if (IsUnicodeMode()) {
+		charsLen = ::MultiByteToWideChar(SC_CP_UTF8, 0, bytes, bytesLen, NULL, 0);
+		::MultiByteToWideChar(SC_CP_UTF8, 0, bytes, bytesLen, characters, charsLen);
+	} else {
+		charsLen = ::MultiByteToWideChar(CodePageOfDocument(), 0, bytes, bytesLen, NULL, 0);
+		::MultiByteToWideChar(CodePageOfDocument(), 0, bytes, bytesLen, characters, charsLen);
+	}
+}
+
+void ScintillaWin::UniCharToBytes(const wchar_t *characters, const int charsLen, char *bytes, int &bytesLen) {
+	// Return results over bytes and bytesLen.
+	if (IsUnicodeMode()) {
+		bytesLen = UTF8Length(characters, charsLen);
+		UTF8FromUTF16(characters, charsLen, bytes, bytesLen);
+		bytes[bytesLen] = '\0';
+	} else {
+		bytesLen = ::WideCharToMultiByte(CodePageOfDocument(), 0,
+			characters, charsLen, bytes, bytesLen, 0, 0);
+		bytes[bytesLen] = '\0';
+	}
+}
+
+void ScintillaWin::RangeToHangul(int uniStrLen) {
+	// Convert every hanja to hangul from current position to the length uniStrLen.
+	// Even if not coverted, the caret should advance by 1 character.
+	pdoc->BeginUndoAction();
+	for (int i=0; i<uniStrLen; i++) {
+		unsigned int const safeLength = UTF8MaxBytes+1;
+
+		int currentPos = CurrentPosition();
+		int oneCharLen = pdoc->LenChar(currentPos);
+
+		if (oneCharLen > 1) {
+			wchar_t uniChar[safeLength] = { 0 };
+			int uniCharLen = 1;
+			char oneChar[safeLength] = "\0\0\0\0";
+			pdoc->GetCharRange(oneChar, currentPos, oneCharLen);
+
+			BytesToUniChar(oneChar, oneCharLen, uniChar, uniCharLen);
+
+			int hangul = HanjaDict::GetHangulOfHanja(uniChar[0]);
+			if (hangul > 0) {
+				uniChar[0] = static_cast<wchar_t>(hangul);
+
+				UniCharToBytes(uniChar, uniCharLen, oneChar, oneCharLen);
+
+				pdoc->DelChar(currentPos);
+				InsertPaste(oneChar, oneCharLen);
+			} else {
+				MoveImeCarets(oneCharLen);
+			}
+		} else {
+			MoveImeCarets(oneCharLen);
+		}
+	}
+	pdoc->EndUndoAction();
+}
+
+void ScintillaWin::EscapeHanja() {
+	// The candidate box pops up to user to select a hanja.
+	// It comes into WM_IME_COMPOSITION with GCS_RESULTSTR.
+	// The existing hangul or hanja is replaced with it.
+	if (sel.Count() > 1) {
+		return; // Do not allow multi carets.
+	}
+	int currentPos = CurrentPosition();
+	int oneCharLen = pdoc->LenChar(currentPos);
+
+	if (oneCharLen < 2) {
+		return; // No need to handle SBCS.
+	}
+
+	// ImmEscapeW() may overwrite uniChar[] with a null terminated string.
+	// So enlarge it enough to Maximum 4 as in UTF-8.
+	unsigned int const safeLength = UTF8MaxBytes+1;
+	wchar_t uniChar[safeLength] = {0};
+	int uniCharLen = 1;
+	char oneChar[safeLength] = "\0\0\0\0";
+
+	pdoc->GetCharRange(oneChar, currentPos, oneCharLen);
+
+	BytesToUniChar(oneChar, oneCharLen, uniChar, uniCharLen);
+
+	// Set the candidate box position since IME may show it.
+	SetCandidateWindowPos();
+
+	// IME_ESC_HANJA_MODE appears to receive the first character only.
+	HIMC hIMC=ImmGetContext(MainHWND());
+	if (hIMC) {
+		if (ImmEscapeW(GetKeyboardLayout(0), hIMC, IME_ESC_HANJA_MODE, &uniChar)) { 
+			SetCandidateWindowPos(); // Force it again for sure.
+			SetSelection (currentPos, currentPos + oneCharLen);
+		}
+		::ImmReleaseContext(MainHWND(), hIMC);
+	}
+}
+
+void ScintillaWin::ToggleHanja() {
+	// If selection, convert every hanja to hangul within the main range. 
+	// If no selection, commit to IME.
+	if (sel.Count() > 1) {
+		return; // Do not allow multi carets.
+	}
+
+	int selStart = sel.RangeMain().Start().Position();
+	int documentStrLen = sel.RangeMain().Length();
+	int selEnd = selStart + documentStrLen;
+	int uniStrLen = pdoc->CountCharacters(selStart, selEnd);
+
+	// Convert one by one from the start of main range.
+	SetSelection(selStart, selStart);
+
+	if (uniStrLen > 0) {
+		RangeToHangul(uniStrLen);
+	} else {
+		EscapeHanja();
+	}
+}
+
 sptr_t ScintillaWin::HandleCompositionInline(uptr_t, sptr_t lParam) {
 	// Copy & paste by johnsonj with a lot of helps of Neil.
 	// Great thanks for my foreruners, jiniya and BLUEnLIVE.
@@ -1327,8 +1454,12 @@ sptr_t ScintillaWin::WndProc(unsigned int iMessage, uptr_t wParam, sptr_t lParam
 				break;
 			}
 
-		case WM_IME_KEYDOWN:
-			return ::DefWindowProc(MainHWND(), iMessage, wParam, lParam);
+		case WM_IME_KEYDOWN: {
+				if (wParam == VK_HANJA) {
+					ToggleHanja();
+				}
+				return ::DefWindowProc(MainHWND(), iMessage, wParam, lParam);
+			}
 
 		case WM_KEYUP:
 			//Platform::DebugPrintf("S keyup %d %x %x\n",iMessage, wParam, lParam);
diff --git a/win32/makefile b/win32/makefile
index 6012905b4..6aa0d6139 100644
--- a/win32/makefile
+++ b/win32/makefile
@@ -91,7 +91,8 @@ BASEOBJS = \
 	Style.o \
 	UniConversion.o \
 	ViewStyle.o \
-	XPM.o
+	XPM.o \
+	HanjaDic.o
 
 SOBJS = ScintillaWin.o ScintillaBase.o $(BASEOBJS)
 
diff --git a/win32/scintilla.mak b/win32/scintilla.mak
index 17fb3e2df..d200597e5 100644
--- a/win32/scintilla.mak
+++ b/win32/scintilla.mak
@@ -88,7 +88,8 @@ SHAREDOBJS=\
 	$(DIR_O)\Style.obj \
 	$(DIR_O)\UniConversion.obj \
 	$(DIR_O)\ViewStyle.obj \
-	$(DIR_O)\XPM.obj
+	$(DIR_O)\XPM.obj \
+	$(DIR_O)\HanjaDic.obj \
 
 SOBJS=\
 	$(SHAREDOBJS) \
@@ -877,7 +878,8 @@ $(DIR_O)\ScintillaWin.obj: \
 	../src/Editor.h \
 	../src/AutoComplete.h \
 	../src/ScintillaBase.h \
-	PlatWin.h
+	PlatWin.h \
+	HanjaDic.h
 $(DIR_O)\ScintillaWinL.obj: \
 	ScintillaWin.cxx \
 	../include/Platform.h \
@@ -910,7 +912,8 @@ $(DIR_O)\ScintillaWinL.obj: \
 	../src/Editor.h \
 	../src/AutoComplete.h \
 	../src/ScintillaBase.h \
-	PlatWin.h
+	PlatWin.h \
+	HanjaDic.h
 $(DIR_O)\ScintillaWinS.obj: \
 	ScintillaWin.cxx \
 	../include/Platform.h \
@@ -943,7 +946,8 @@ $(DIR_O)\ScintillaWinS.obj: \
 	../src/Editor.h \
 	../src/AutoComplete.h \
 	../src/ScintillaBase.h \
-	PlatWin.h
+	PlatWin.h \
+	HanjaDic.h
 $(DIR_O)\Selection.obj: \
 	../src/Selection.cxx \
 	../include/Platform.h \
@@ -983,3 +987,5 @@ $(DIR_O)\XPM.obj: \
 	../src/XPM.cxx \
 	../include/Platform.h \
 	../src/XPM.h
+$(DIR_O)\HanjaDic: \
+	./HanjaDic.h