ICU-199 add Jamo-Hangul and Hangul-Jamo

X-SVN-Rev: 618

icu4c/source/i18n/hangjamo.cpp

@@ -0,0 +1,135 @@
+/*
+**********************************************************************
+*   Copyright (c) 2000, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   01/17/2000  aliu        Ported from Java.
+**********************************************************************
+*/
+#include "unicode/hangjamo.h"
+#include "unicode/rep.h"
+#include "unicode/unifilt.h"
+#include "unicode/unicode.h"
+
+// These constants are from the Unicode book's algorithm.
+
+#define SBase   (0xAC00)
+#define LBase   (0x1100)
+#define VBase   (0x1161)
+#define TBase   (0x11A7)
+#define LCount  (19)
+#define VCount  (21)
+#define TCount  (28)
+#define NCount  (VCount * TCount)   // 588
+#define SCount  (LCount * NCount)   // 11172
+
+/**
+ * ID for this transliterator.
+ */
+const char* HangulJamoTransliterator::_ID = "Hangul-Jamo";
+
+/**
+ * Constructs a transliterator.
+ */
+HangulJamoTransliterator::HangulJamoTransliterator(UnicodeFilter* adoptedFilter) :
+    Transliterator(_ID, adoptedFilter) {
+}
+
+/**
+ * Copy constructor.
+ */
+HangulJamoTransliterator::HangulJamoTransliterator(const HangulJamoTransliterator& o) :
+    Transliterator(o) {
+}
+
+/**
+ * Assignment operator.
+ */
+HangulJamoTransliterator& HangulJamoTransliterator::operator=(
+                                             const HangulJamoTransliterator& o) {
+    Transliterator::operator=(o);
+    return *this;
+}
+
+/**
+ * Transliterator API.
+ */
+Transliterator* HangulJamoTransliterator::clone(void) const {
+    return new HangulJamoTransliterator(*this);
+}
+
+/**
+ * Transliterates a segment of a string.  <code>Transliterator</code> API.
+ * @param text the string to be transliterated
+ * @param start the beginning index, inclusive; <code>0 <= start
+ * <= limit</code>.
+ * @param limit the ending index, exclusive; <code>start <= limit
+ * <= text.length()</code>.
+ * @return the new limit index
+ */
+int32_t HangulJamoTransliterator::transliterate(Replaceable& text,
+                                                int32_t start, int32_t limit) const {
+    int32_t offsets[3] = { start, limit, start };
+    handleKeyboardTransliterate(text, offsets);
+    return offsets[LIMIT];
+}
+
+/**
+ * Implements {@link Transliterator#handleKeyboardTransliterate}.
+ */
+void HangulJamoTransliterator::handleKeyboardTransliterate(Replaceable& text,
+                                                           int32_t offsets[3]) const {
+    int32_t cursor = offsets[CURSOR];
+    int32_t limit = offsets[LIMIT];
+
+    UnicodeString replacement;
+    while (cursor < limit) {
+        UChar c = filteredCharAt(text, cursor);
+        if (decomposeHangul(c, replacement)) {
+            text.handleReplaceBetween(cursor, cursor+1, replacement);
+            cursor += replacement.length(); // skip over replacement
+            limit += replacement.length() - 1; // fix up limit
+        } else {
+            ++cursor;
+        }
+    }
+
+    offsets[LIMIT] = limit;
+    offsets[CURSOR] = cursor;
+}
+
+bool_t HangulJamoTransliterator::decomposeHangul(UChar s, UnicodeString& result) {
+    int32_t SIndex = s - SBase;
+    if (0 > SIndex || SIndex >= SCount) {
+        return FALSE;
+    }
+    int32_t L = LBase + SIndex / NCount;
+    int32_t V = VBase + (SIndex % NCount) / TCount;
+    int32_t T = TBase + SIndex % TCount;
+    result.truncate(0);
+    result.append((UChar)L);
+    result.append((UChar)V);
+    if (T != TBase) {
+        result.append((UChar)T);
+    }
+    return TRUE;
+}
+
+UChar HangulJamoTransliterator::filteredCharAt(Replaceable& text, int32_t i) const {
+    UChar c;
+    const UnicodeFilter* filter = getFilter();
+    return (filter == 0) ? text.charAt(i) :
+        (filter->isIn(c = text.charAt(i)) ? c : (UChar)0xFFFF);
+}
+
+/**
+ * Return the length of the longest context required by this transliterator.
+ * This is <em>preceding</em> context.
+ * @param direction either <code>FORWARD</code> or <code>REVERSE</code>
+ * @return maximum number of preceding context characters this
+ * transliterator needs to examine
+ */
+int32_t HangulJamoTransliterator::getMaximumContextLength(void) const {
+    return 0;
+}

icu4c/source/i18n/i18n.dsp

@@ -165,10 +165,18 @@ SOURCE=.\gregocal.cpp
 # End Source File
 # Begin Source File
 
+SOURCE=.\hangjamo.cpp
+# End Source File
+# Begin Source File
+
 SOURCE=.\hextouni.cpp
 # End Source File
 # Begin Source File
 
+SOURCE=.\jamohang.cpp
+# End Source File
+# Begin Source File
+
 SOURCE=.\mergecol.cpp
 # End Source File
 # Begin Source File
@@ -718,6 +726,25 @@ InputPath=.\unicode\gregocal.h
 # End Source File
 # Begin Source File
 
+SOURCE=.\unicode\hangjamo.h
+
+!IF  "$(CFG)" == "i18n - Win32 Release"
+
+!ELSEIF  "$(CFG)" == "i18n - Win32 Debug"
+
+# Begin Custom Build
+InputPath=.\unicode\hangjamo.h
+
+"..\..\include\unicode\hangjamo.h" : $(SOURCE) "$(INTDIR)" "$(OUTDIR)"
+	copy   unicode\hangjamo.h    ..\..\include\unicode
+
+# End Custom Build
+
+!ENDIF 
+
+# End Source File
+# Begin Source File
+
 SOURCE=.\unicode\hextouni.h
 
 !IF  "$(CFG)" == "i18n - Win32 Release"
@@ -745,6 +772,25 @@ InputPath=.\unicode\hextouni.h
 # End Source File
 # Begin Source File
 
+SOURCE=.\unicode\jamohang.h
+
+!IF  "$(CFG)" == "i18n - Win32 Release"
+
+!ELSEIF  "$(CFG)" == "i18n - Win32 Debug"
+
+# Begin Custom Build
+InputPath=.\unicode\jamohang.h
+
+"..\..\include\unicode\jamohang.h" : $(SOURCE) "$(INTDIR)" "$(OUTDIR)"
+	copy   unicode\jamohang.h    ..\..\include\unicode
+
+# End Custom Build
+
+!ENDIF 
+
+# End Source File
+# Begin Source File
+
 SOURCE=.\mergecol.h
 # End Source File
 # Begin Source File

icu4c/source/i18n/jamohang.cpp

@@ -0,0 +1,158 @@
+/*
+**********************************************************************
+*   Copyright (c) 2000, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   01/17/2000  aliu        Ported from Java.
+**********************************************************************
+*/
+#include "unicode/jamohang.h"
+#include "unicode/rep.h"
+#include "unicode/unifilt.h"
+#include "unicode/unicode.h"
+
+/**
+ * ID for this transliterator.
+ */
+const char* JamoHangulTransliterator::_ID = "Jamo-Hangul";
+
+/**
+ * Constructs a transliterator.
+ */
+JamoHangulTransliterator::JamoHangulTransliterator(UnicodeFilter* adoptedFilter) :
+    Transliterator(_ID, adoptedFilter) {
+}
+
+/**
+ * Copy constructor.
+ */
+JamoHangulTransliterator::JamoHangulTransliterator(const JamoHangulTransliterator& o) :
+    Transliterator(o) {
+}
+
+/**
+ * Assignment operator.
+ */
+JamoHangulTransliterator& JamoHangulTransliterator::operator=(
+                                             const JamoHangulTransliterator& o) {
+    Transliterator::operator=(o);
+    return *this;
+}
+
+/**
+ * Transliterator API.
+ */
+Transliterator* JamoHangulTransliterator::clone(void) const {
+    return new JamoHangulTransliterator(*this);
+}
+
+/**
+ * Transliterates a segment of a string.  <code>Transliterator</code> API.
+ * @param text the string to be transliterated
+ * @param start the beginning index, inclusive; <code>0 <= start
+ * <= limit</code>.
+ * @param limit the ending index, exclusive; <code>start <= limit
+ * <= text.length()</code>.
+ * @return the new limit index
+ */
+int32_t JamoHangulTransliterator::transliterate(Replaceable& text,
+                                                int32_t start, int32_t limit) const {
+    int32_t offsets[3] = { start, limit, start };
+    handleKeyboardTransliterate(text, offsets);
+    return offsets[LIMIT];
+}
+
+/**
+ * Implements {@link Transliterator#handleKeyboardTransliterate}.
+ */
+void JamoHangulTransliterator::handleKeyboardTransliterate(Replaceable& text,
+                                                           int32_t offsets[3]) const {
+    /**
+     * Performs transliteration changing Jamo to Hangul 
+     */
+    int32_t cursor = offsets[CURSOR];
+    int32_t limit = offsets[LIMIT];
+    if (cursor >= limit) {
+        return;
+    }
+    
+    // get last character
+    UChar last = filteredCharAt(text, cursor++);
+    UnicodeString str = UNICODE_STRING("a", 1);
+    while (cursor < limit) {
+        UChar c = filteredCharAt(text, cursor);
+        UChar replacement = composeHangul(last, c);
+        if (replacement != 0) {
+            str[0] = replacement;
+            text.handleReplaceBetween(cursor-1, cursor+1, str);
+            last = replacement;
+            // leave cursor where it is
+            --limit; // fix up limit
+        } else {
+            ++cursor;
+        }
+    }
+
+    offsets[LIMIT] = limit + 1;
+    offsets[CURSOR] = cursor;
+}
+
+// These constants are from the Unicode book's algorithm.
+// There's no need to make them class members, since they
+// are only used here.
+
+#define SBase   (0xAC00)
+#define LBase   (0x1100)
+#define VBase   (0x1161)
+#define TBase   (0x11A7)
+#define LCount  (19)
+#define VCount  (21)
+#define TCount  (28)
+#define NCount  (VCount * TCount)   // 588
+#define SCount  (LCount * NCount)   // 11172
+
+/**
+ * Return composed character (if it composes)
+ * 0 otherwise
+ */
+UChar JamoHangulTransliterator::composeHangul(UChar last, UChar ch) {
+  // check to see if two current characters are L and V
+  int32_t LIndex = last - LBase;
+  if (0 <= LIndex && LIndex < LCount) {
+      int32_t VIndex = ch - VBase;
+      if (0 <= VIndex && VIndex < VCount) {
+          // make syllable of form LV
+          return (UChar)(SBase + (LIndex * VCount + VIndex) * TCount);
+      }
+  }
+  // check to see if two current characters are LV and T
+  int32_t SIndex = last - SBase;
+  if (0 <= SIndex && SIndex < SCount && (SIndex % TCount) == 0) {
+      int32_t TIndex = ch - TBase;
+      if (0 <= TIndex && TIndex <= TCount) {
+          // make syllable of form LVT
+          return (UChar)(last + TIndex);
+      }
+  }
+  // if neither case was true, skip
+  return 0x0000;
+}    
+
+UChar JamoHangulTransliterator::filteredCharAt(Replaceable& text, int32_t i) const {
+    UChar c;
+    const UnicodeFilter* filter = getFilter();
+    return (filter == 0) ? text.charAt(i) :
+        (filter->isIn(c = text.charAt(i)) ? c : (UChar)0xFFFF);
+}
+
+/**
+ * Return the length of the longest context required by this transliterator.
+ * This is <em>preceding</em> context.
+ * @param direction either <code>FORWARD</code> or <code>REVERSE</code>
+ * @return maximum number of preceding context characters this
+ * transliterator needs to examine
+ */
+int32_t JamoHangulTransliterator::getMaximumContextLength(void) const {
+    return 3;
+}

icu4c/source/i18n/translit.cpp

@@ -24,6 +24,8 @@
 #include "unicode/nultrans.h"
 #include "unicode/putil.h"
 #include "unicode/cpdtrans.h"
+#include "unicode/jamohang.h"
+#include "unicode/hangjamo.h"
 
 /**
  * Dictionary of known transliterators.  Keys are <code>String</code>
@@ -893,6 +895,8 @@ void Transliterator::initializeCache(void) {
     status = U_ZERO_ERROR; // Reset status for following calls
     _registerInstance(new HexToUnicodeTransliterator(), status);
     _registerInstance(new UnicodeToHexTransliterator(), status);
+    _registerInstance(new JamoHangulTransliterator(), status);
+    _registerInstance(new HangulJamoTransliterator(), status);
     _registerInstance(new NullTransliterator(), status);
 
     cacheInitialized = TRUE;

icu4c/source/i18n/unicode/hangjamo.h

@@ -0,0 +1,91 @@
+/*
+**********************************************************************
+*   Copyright (c) 2000, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   01/17/2000  aliu        Ported from Java.
+**********************************************************************
+*/
+#ifndef HANGJAMO_H
+#define HANGJAMO_H
+
+#include "unicode/translit.h"
+
+/**
+ * A transliterator that converts Hangul to Jamo.
+ *
+ * @author Mark Davis
+ * @version $RCSfile: hangjamo.h,v $ $Revision: 1.1 $ $Date: 2000/01/18 01:55:52 $
+ */
+class U_I18N_API HangulJamoTransliterator : public Transliterator {
+
+    /**
+     * ID for this transliterator.
+     */
+    static const char* _ID;
+
+public:
+
+    /**
+     * Constructs a transliterator.
+     */
+    HangulJamoTransliterator(UnicodeFilter* adoptedFilter = 0);
+
+    /**
+     * Destructor.
+     */
+    virtual ~HangulJamoTransliterator();
+
+    /**
+     * Copy constructor.
+     */
+    HangulJamoTransliterator(const HangulJamoTransliterator&);
+
+    /**
+     * Assignment operator.
+     */
+    HangulJamoTransliterator& operator=(const HangulJamoTransliterator&);
+
+    /**
+     * Transliterator API.
+     */
+    Transliterator* clone(void) const;
+
+    /**
+     * Transliterates a segment of a string.  <code>Transliterator</code> API.
+     * @param text the string to be transliterated
+     * @param start the beginning index, inclusive; <code>0 <= start
+     * <= limit</code>.
+     * @param limit the ending index, exclusive; <code>start <= limit
+     * <= text.length()</code>.
+     * @return the new limit index
+     */
+    virtual int32_t transliterate(Replaceable &text,
+                                  int32_t start, int32_t limit) const;
+
+    /**
+     * Implements {@link Transliterator#handleKeyboardTransliterate}.
+     */
+    virtual void handleKeyboardTransliterate(Replaceable& text,
+                                             int32_t offsets[3]) const;
+
+    /**
+     * Return the length of the longest context required by this transliterator.
+     * This is <em>preceding</em> context.
+     * @param direction either <code>FORWARD</code> or <code>REVERSE</code>
+     * @return maximum number of preceding context characters this
+     * transliterator needs to examine
+     */
+    virtual int32_t getMaximumContextLength(void) const;
+
+private:
+
+    static bool_t decomposeHangul(UChar s, UnicodeString& result);
+
+    UChar filteredCharAt(Replaceable& text, int32_t i) const;
+};
+
+inline HangulJamoTransliterator::~HangulJamoTransliterator() {}
+
+#endif

icu4c/source/i18n/unicode/jamohang.h

@@ -0,0 +1,95 @@
+/*
+**********************************************************************
+*   Copyright (c) 2000, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   01/17/2000  aliu        Ported from Java.
+**********************************************************************
+*/
+#ifndef JAMOHANG_H
+#define JAMOHANG_H
+
+#include "unicode/translit.h"
+
+/**
+ * A transliterator that converts Jamo to Hangul.
+ *
+ * @author Mark Davis
+ * @version $RCSfile: jamohang.h,v $ $Revision: 1.1 $ $Date: 2000/01/18 01:55:52 $
+ */
+class U_I18N_API JamoHangulTransliterator : public Transliterator {
+
+    /**
+     * ID for this transliterator.
+     */
+    static const char* _ID;
+
+public:
+
+    /**
+     * Constructs a transliterator.
+     */
+    JamoHangulTransliterator(UnicodeFilter* adoptedFilter = 0);
+
+    /**
+     * Destructor.
+     */
+    virtual ~JamoHangulTransliterator();
+
+    /**
+     * Copy constructor.
+     */
+    JamoHangulTransliterator(const JamoHangulTransliterator&);
+
+    /**
+     * Assignment operator.
+     */
+    JamoHangulTransliterator& operator=(const JamoHangulTransliterator&);
+
+    /**
+     * Transliterator API.
+     */
+    Transliterator* clone(void) const;
+
+    /**
+     * Transliterates a segment of a string.  <code>Transliterator</code> API.
+     * @param text the string to be transliterated
+     * @param start the beginning index, inclusive; <code>0 <= start
+     * <= limit</code>.
+     * @param limit the ending index, exclusive; <code>start <= limit
+     * <= text.length()</code>.
+     * @return the new limit index
+     */
+    virtual int32_t transliterate(Replaceable &text,
+                                  int32_t start, int32_t limit) const;
+
+    /**
+     * Implements {@link Transliterator#handleKeyboardTransliterate}.
+     */
+    virtual void handleKeyboardTransliterate(Replaceable& text,
+                                             int32_t offsets[3]) const;
+
+    /**
+     * Return the length of the longest context required by this transliterator.
+     * This is <em>preceding</em> context.
+     * @param direction either <code>FORWARD</code> or <code>REVERSE</code>
+     * @return maximum number of preceding context characters this
+     * transliterator needs to examine
+     */
+    virtual int32_t getMaximumContextLength(void) const;
+
+private:
+
+    /**
+     * Return composed character (if it composes)
+     * 0 otherwise
+     */
+    static UChar composeHangul(UChar last, UChar ch);
+
+    UChar filteredCharAt(Replaceable& text, int32_t i) const;
+};
+
+inline JamoHangulTransliterator::~JamoHangulTransliterator() {}
+
+#endif