#95565#add phonetic entry for CJK index modules

3 files changed, 164 insertions, 16 deletions

diff --git a/i18npool/source/indexentry/data/indexdata_ja_phonetic.h b/i18npool/source/indexentry/data/indexdata_ja_phonetic.h
index 8ec63352d5ba..4f56029215cd 100644
--- a/i18npool/source/indexentry/data/indexdata_ja_phonetic.h
+++ b/i18npool/source/indexentry/data/indexdata_ja_phonetic.h
@@ -2,9 +2,9 @@
  *
  *  $RCSfile: indexdata_ja_phonetic.h,v $
  *
- *  $Revision: 1.2 $
+ *  $Revision: 1.3 $
  *
- *  last change: $Author: khong $ $Date: 2002-05-31 04:48:43 $
+ *  last change: $Author: khong $ $Date: 2002-07-25 04:34:24 $
  *
  *  The Contents of this file are made available subject to the terms of
  *  either of the following licenses
@@ -39,7 +39,7 @@
  ************************************************************************/
 #include <sal/types.h>
 
-static sal_uInt16 idx1[] = {
+static sal_uInt16 idx[] = {
     0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF,  // 0000
     0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF,  // 1000
     0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF,  // 2000
@@ -58,7 +58,7 @@ static sal_uInt16 idx1[] = {
     0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0x0100,  // f000
 };
 
-static sal_Unicode idx2[] = {
+static sal_Unicode syllable[] = {
     0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // 3000
     0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // 3010
     0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // 3020
@@ -94,3 +94,40 @@ static sal_Unicode idx2[] = {
     0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000   // FFF0
 };
 
+
+static sal_Unicode consonant[] = {
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // 3000
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // 3010
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // 3020
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // 3030
+    0x3040, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x304B, 0x304B, 0x304B, 0x304B, 0x304B,  // 3040
+    0x304B, 0x304B, 0x304B, 0x304B, 0x304B, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x305F,  // 3050
+    0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x306A, 0x306A, 0x306A, 0x306A, 0x306A, 0x306F,  // 3060
+    0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x307E, 0x307E,  // 3070
+    0x307E, 0x307E, 0x307E, 0x3084, 0x3084, 0x3084, 0x3084, 0x3084, 0x3084, 0x3089, 0x3089, 0x3089, 0x3089, 0x3089, 0x308F, 0x308F,  // 3080
+    0x308F, 0x308F, 0x308F, 0x308F, 0x3042, 0x304B, 0x304B, 0x3097, 0x3098, 0x3099, 0x309A, 0x309B, 0x309C, 0x309D, 0x309E, 0x309F,  // 3090
+    0x30a0, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x304B, 0x304B, 0x304B, 0x304B, 0x304B,  // 30A0
+    0x304B, 0x304B, 0x304B, 0x304B, 0x304B, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055, 0x305F,  // 30B0
+    0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x306A, 0x306A, 0x306A, 0x306A, 0x306A, 0x306F,  // 30C0
+    0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x307E, 0x307E,  // 30D0
+    0x307E, 0x307E, 0x307E, 0x3084, 0x3084, 0x3084, 0x3084, 0x3084, 0x3084, 0x3089, 0x3089, 0x3089, 0x3089, 0x3089, 0x308F, 0x308F,  // 30E0
+    0x308F, 0x308F, 0x308F, 0x308F, 0x3042, 0x304B, 0x304B, 0x308F, 0x308F, 0x308F, 0x308F, 0x30FB, 0x30FC, 0x30FD, 0x30FE, 0x30FF,  // 30F0
+
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FF00
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FF10
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FF20
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FF30
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FF40
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FF50
+    0xFF60, 0xFF61, 0xFF62, 0xFF63, 0xFF64, 0xFF65, 0x308F, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x3084, 0x3084, 0x3084, 0x305F,  // FF60
+    0xFF70, 0x3042, 0x3042, 0x3042, 0x3042, 0x3042, 0x304B, 0x304B, 0x304B, 0x304B, 0x304B, 0x3055, 0x3055, 0x3055, 0x3055, 0x3055,  // FF70
+    0x305F, 0x305F, 0x305F, 0x305F, 0x305F, 0x306A, 0x306A, 0x306A, 0x306A, 0x306A, 0x306F, 0x306F, 0x306F, 0x306F, 0x306F, 0x307E,  // FF80
+    0x307E, 0x307E, 0x307E, 0x307E, 0x3084, 0x3084, 0x3084, 0x3089, 0x3089, 0x3089, 0x3089, 0x3089, 0x308F, 0x308F, 0xFF9E, 0xFF9F,  // FF90
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FFA0
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FFB0
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FFC0
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FFD0
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,  // FFE0
+    0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000   // FFF0
+};
+
diff --git a/i18npool/source/indexentry/indexentrysupplier.cxx b/i18npool/source/indexentry/indexentrysupplier.cxx
index 5ac2737f1a7e..1952d7e83308 100644
--- a/i18npool/source/indexentry/indexentrysupplier.cxx
+++ b/i18npool/source/indexentry/indexentrysupplier.cxx
@@ -2,9 +2,9 @@
  *
  *  $RCSfile: indexentrysupplier.cxx,v $
  *
- *  $Revision: 1.8 $
+ *  $Revision: 1.9 $
  *
- *  last change: $Author: khong $ $Date: 2002-06-18 22:29:26 $
+ *  last change: $Author: khong $ $Date: 2002-07-25 04:38:12 $
  *
  *  The Contents of this file are made available subject to the terms of
  *  either of the following licenses
@@ -64,6 +64,9 @@
 #include <tools/intn.hxx>
 #include <tools/isolang.hxx>
 #include <indexentrysupplier.hxx>
+#include <data/zh_pinyin.h>
+#include <data/zh_zhuyin.h>
+#include <data/ko_phonetic.h>
 
 using namespace ::com::sun::star::uno;
 using namespace ::com::sun::star::lang;
@@ -92,7 +95,7 @@ static const struct {
     { "hu",  "alphanumeric", 1 },
     { "is",  "alphanumeric", 1 },
     { "it",  "alphanumeric", 1 },
-    { "ja",  "phonetic", 1 },
+    { "ja",  "phonetic (alphanumeric first) (grouped by syllable),phonetic (alphanumeric first) (grouped by consonant),phonetic (alphanumeric last) (grouped by syllable),phonetic (alphanumeric last) (grouped by consonant)", 4 },
     { "ko",  "dict", 1 },
     { "nb",  "alphanumeric", 1 },
     { "nl",  "alphanumeric", 1 },
@@ -104,8 +107,8 @@ static const struct {
     { "sv",  "alphanumeric", 1 },
     { "tr",  "alphanumeric", 1 },
     { "th",  "alphanumeric", 1 },
-    { "zh_CN",  "pinyin stroke radical", 3 },
-    { "zh_TW",  "stroke radical zhuyin pinyin", 4 },
+    { "zh_CN",  "pinyin,stroke,radical", 3 },
+    { "zh_TW",  "stroke,radical,zhuyin,pinyin", 4 },
 };
 
 static const sal_Int16 nbOfLocales = sizeof(aLocaleList) / sizeof(aLocaleList[0]);
@@ -167,7 +170,7 @@ Sequence < OUString > SAL_CALL IndexEntrySupplier::getAlgorithmList( const Local
         algorithmList.realloc(aLocaleList[i].pAlgorithmCount);
         index = 0;
         for (sal_Int16 j=0; j<aLocaleList[i].pAlgorithmCount; j++)
-            algorithmList[j] = algorithms.getToken(0, sal_Unicode(' '), index);
+            algorithmList[j] = algorithms.getToken(0, sal_Unicode(','), index);
 
         break;
         }
@@ -186,14 +189,57 @@ sal_Bool SAL_CALL IndexEntrySupplier::loadAlgorithm( const Locale& rLocale, cons
 sal_Bool SAL_CALL IndexEntrySupplier::usePhoneticEntry( const Locale& rLocale ) throw (RuntimeException)
 {
     // First implementation only turns the feature on for Japanese language.
-    return rLocale.Language.compareToAscii("ja") == 0;
+    return rLocale.Language.equalsAscii("zh")
+        || rLocale.Language.equalsAscii("ja")
+        || rLocale.Language.equalsAscii("ko");
 }
 
 OUString SAL_CALL IndexEntrySupplier::getPhoneticCandidate( const OUString& rIndexEntry,
     const Locale& rLocale ) throw (RuntimeException)
 {
+    static OUString space(OUString::createFromAscii(" "));
+    OUString candidate;
     // TODO: the phonetic candidate will be provided by language engine for CJK.
-    return OUString();
+    if (rLocale.Language.equalsAscii("zh")) {
+        sal_Unicode *Str;
+        sal_uInt16 *Index1, *Index2;
+        if (rLocale.Country.equalsAscii("TW") ||
+        rLocale.Country.equalsAscii("HK") ||
+        rLocale.Country.equalsAscii("MO")) {
+            Str = ZhuYinStr_zh;
+            Index1 = ZhuYinIndex1_zh;
+            Index2 = ZhuYinIndex2_zh;
+        } else {
+            Str = PinYinStr_zh;
+            Index1 = PinYinIndex1_zh;
+            Index2 = PinYinIndex2_zh;
+        }
+
+        for (sal_Int32 i=0; i < rIndexEntry.getLength(); i++) {
+        sal_Unicode ch = rIndexEntry[i];
+        sal_uInt16 address = Index1[ch>>8];
+        if (address != 0xFFFF)
+            address = Index2[address + (ch & 0xFF)];
+        if (i > 0)
+            candidate += space;
+        if (address != 0xFFFF)
+            candidate += OUString(&Str[address]);
+        }
+    } else if (rLocale.Language.equalsAscii("ja")) {
+        ; // TODO
+    } else if (rLocale.Language.equalsAscii("ko")) {
+        for (sal_Int32 i=0; i < rIndexEntry.getLength(); i++) {
+        sal_Unicode ch = rIndexEntry[i];
+        sal_uInt16 address = PhoneticIndex_ko[ch>>8];
+        if (address != 0xFFFF)
+            address = PhoneticCharacter_ko[address + (ch & 0xFF)];
+        if (address != 0xFFFF)
+            candidate += OUString(address);
+        else
+            candidate += space;
+        }
+    }
+    return candidate;
 }
 
 OUString SAL_CALL IndexEntrySupplier::getIndexKey( const OUString& rIndexEntry,
diff --git a/i18npool/source/indexentry/indexentrysupplier_ja_phonetic.cxx b/i18npool/source/indexentry/indexentrysupplier_ja_phonetic.cxx
index 82981a324ddd..5615aa81dba0 100644
--- a/i18npool/source/indexentry/indexentrysupplier_ja_phonetic.cxx
+++ b/i18npool/source/indexentry/indexentrysupplier_ja_phonetic.cxx
@@ -2,9 +2,9 @@
  *
  *  $RCSfile: indexentrysupplier_ja_phonetic.cxx,v $
  *
- *  $Revision: 1.3 $
+ *  $Revision: 1.4 $
  *
- *  last change: $Author: khong $Date: 2002/05/31 04:51:19 $
+ *  last change: $Author: khong $Date: 2002/06/18 22:29:26 $
  *
  *  The Contents of this file are made available subject to the terms of
  *  either of the following licenses
@@ -62,12 +62,77 @@
 #define INDEXENTRYSUPPLIER_ja_phonetic
 #include <indexentrysupplier_asian.hxx>
 #include <data/indexdata_ja_phonetic.h>
+#include <strings.h>
 
 namespace com { namespace sun { namespace star { namespace i18n {
 
 rtl::OUString SAL_CALL IndexEntrySupplier_ja_phonetic::getIndexCharacter( const rtl::OUString& rIndexEntry,
-    const lang::Locale& rLocale, const rtl::OUString& rSortAlgorithm ) throw (com::sun::star::uno::RuntimeException) {
-    return IndexEntrySupplier_CJK::getIndexString(rIndexEntry.toChar(), idx1, idx2);
+    const lang::Locale& rLocale, const rtl::OUString& rSortAlgorithm )
+    throw (com::sun::star::uno::RuntimeException)
+{
+    return IndexEntrySupplier_CJK::getIndexString(rIndexEntry.toChar(), idx,
+            strstr(implementationName, "syllable") ? syllable : consonant);
+}
+
+rtl::OUString SAL_CALL IndexEntrySupplier_ja_phonetic::getIndexKey( const rtl::OUString& IndexEntry,
+    const rtl::OUString& PhoneticEntry, const lang::Locale& rLocale )
+    throw (com::sun::star::uno::RuntimeException)
+{
+    return IndexEntrySupplier_CJK::getIndexString(
+        PhoneticEntry.getLength() > 0 ? PhoneticEntry.toChar() : IndexEntry.toChar(), idx,
+            strstr(implementationName, "syllable") ? syllable : consonant);
+}
+
+sal_Int16 SAL_CALL IndexEntrySupplier_ja_phonetic::compareIndexEntry(
+    const rtl::OUString& IndexEntry1, const rtl::OUString& PhoneticEntry1, const lang::Locale& rLocale1,
+    const rtl::OUString& IndexEntry2, const rtl::OUString& PhoneticEntry2, const lang::Locale& rLocale2 )
+    throw (com::sun::star::uno::RuntimeException)
+{
+    sal_Int16 result =  collator->compareString(
+        IndexEntrySupplier_ja_phonetic::getIndexKey(IndexEntry1, PhoneticEntry1, rLocale1),
+        IndexEntrySupplier_ja_phonetic::getIndexKey(IndexEntry2, PhoneticEntry2, rLocale2));
+
+    if (result == 0)
+        return collator->compareString(
+            PhoneticEntry1.getLength() > 0 ? PhoneticEntry1 : IndexEntry1,
+            PhoneticEntry2.getLength() > 0 ? PhoneticEntry2 : IndexEntry2);
+    else
+        return result;
+}
+
+static sal_Char first[] = "ja_phonetic_alphanumeric_first";
+sal_Bool SAL_CALL IndexEntrySupplier_ja_phonetic_alphanumeric_first_by_syllable::loadAlgorithm(
+    const com::sun::star::lang::Locale& rLocale, const rtl::OUString& SortAlgorithm,
+    sal_Int32 collatorOptions ) throw (com::sun::star::uno::RuntimeException)
+{
+    aSortAlgorithm = SortAlgorithm;
+    aLocale = rLocale;
+    return collator->loadCollatorAlgorithm(rtl::OUString::createFromAscii(first), rLocale, collatorOptions) == 0;
+}
+sal_Bool SAL_CALL IndexEntrySupplier_ja_phonetic_alphanumeric_first_by_consonant::loadAlgorithm(
+    const com::sun::star::lang::Locale& rLocale, const rtl::OUString& SortAlgorithm,
+    sal_Int32 collatorOptions ) throw (com::sun::star::uno::RuntimeException)
+{
+    aSortAlgorithm = SortAlgorithm;
+    aLocale = rLocale;
+    return collator->loadCollatorAlgorithm(rtl::OUString::createFromAscii(first), rLocale, collatorOptions) == 0;
+}
+static sal_Char last[] = "ja_phonetic_alphanumeric_last";
+sal_Bool SAL_CALL IndexEntrySupplier_ja_phonetic_alphanumeric_last_by_syllable::loadAlgorithm(
+    const com::sun::star::lang::Locale& rLocale, const rtl::OUString& SortAlgorithm,
+    sal_Int32 collatorOptions ) throw (com::sun::star::uno::RuntimeException)
+{
+    aSortAlgorithm = SortAlgorithm;
+    aLocale = rLocale;
+    return collator->loadCollatorAlgorithm(rtl::OUString::createFromAscii(last), rLocale, collatorOptions) == 0;
+}
+sal_Bool SAL_CALL IndexEntrySupplier_ja_phonetic_alphanumeric_last_by_consonant::loadAlgorithm(
+    const com::sun::star::lang::Locale& rLocale, const rtl::OUString& SortAlgorithm,
+    sal_Int32 collatorOptions ) throw (com::sun::star::uno::RuntimeException)
+{
+    aSortAlgorithm = SortAlgorithm;
+    aLocale = rLocale;
+    return collator->loadCollatorAlgorithm(rtl::OUString::createFromAscii(last), rLocale, collatorOptions) == 0;
 }
 
 } } } }