Adding native number support for Cyrillic numerals

Change-Id: I054786a97b6f0e6ba3fe3b824ff9ff9748d8be4a
Reviewed-on: https://gerrit.libreoffice.org/20013
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>

3 files changed, 284 insertions, 4 deletions

diff --git a/i18npool/source/localedata/data/cu_RU.xml b/i18npool/source/localedata/data/cu_RU.xml
index d4dde4665dc9..fd28ddfeeeab 100644
--- a/i18npool/source/localedata/data/cu_RU.xml
+++ b/i18npool/source/localedata/data/cu_RU.xml
@@ -52,7 +52,155 @@
     <TimePM>по по́лꙋд.</TimePM>
     <MeasurementSystem>metric</MeasurementSystem>
   </LC_CTYPE>
-  <LC_FORMAT ref="ru_RU" replaceFrom="[CURRENCY]" replaceTo="[$р.-694]" />
+  <LC_FORMAT replaceFrom="[CURRENCY]" replaceTo="[$р.-694]">
+    <DateAcceptancePattern>D.M.</DateAcceptancePattern>
+    <DateAcceptancePattern>D/M/</DateAcceptancePattern>
+    <FormatElement msgid="FixedFormatskey1" default="true" type="medium" usage="FIXED_NUMBER" formatindex="0">
+      <FormatCode>General</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="FixedFormatskey2" default="true" type="short" usage="FIXED_NUMBER" formatindex="1">
+      <FormatCode>0</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="FixedFormatskey3" default="false" type="medium" usage="FIXED_NUMBER" formatindex="2">
+      <FormatCode>0,00</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="FixedFormatskey4" default="false" type="short" usage="FIXED_NUMBER" formatindex="3">
+      <FormatCode># ##0</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="FixedFormatskey5" default="false" type="medium" usage="FIXED_NUMBER" formatindex="4">
+      <FormatCode># ##0,00</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="FixedFormatskey6" default="false" type="medium" usage="FIXED_NUMBER" formatindex="5">
+      <FormatCode># ###,00</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="FixedFormatskey7" default="false" type="medium" usage="FIXED_NUMBER" formatindex="59">
+      <FormatCode>[NatNum1]0;-0</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="ScientificFormatskey1" default="false" type="medium" usage="SCIENTIFIC_NUMBER" formatindex="6">
+      <FormatCode>0,00E+000</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="ScientificFormatskey2" default="true" type="medium" usage="SCIENTIFIC_NUMBER" formatindex="7">
+      <FormatCode>0,00E+00</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="ScientificFormatskey3" default="false" type="medium" usage="SCIENTIFIC_NUMBER" formatindex="78">
+      <FormatCode>##0,00E+00</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="PercentFormatskey1" default="true" type="short" usage="PERCENT_NUMBER" formatindex="8">
+      <FormatCode>0%</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="PercentFormatskey2" default="true" type="long" usage="PERCENT_NUMBER" formatindex="9">
+      <FormatCode>0,00%</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="CurrencyFormatskey1" default="true" type="short" usage="CURRENCY" formatindex="12">
+      <FormatCode># ##0 [CURRENCY];-# ##0 [CURRENCY]</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="CurrencyFormatskey2" default="false" type="medium" usage="CURRENCY" formatindex="13">
+      <FormatCode># ##0,00 [CURRENCY];-# ##0,00 [CURRENCY]</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="CurrencyFormatskey3" default="false" type="medium" usage="CURRENCY" formatindex="14">
+      <FormatCode># ##0 [CURRENCY];[RED]-# ##0 [CURRENCY]</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="CurrencyFormatskey4" default="true" type="medium" usage="CURRENCY" formatindex="15">
+      <FormatCode># ##0,00 [CURRENCY];[RED]-# ##0,00 [CURRENCY]</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="CurrencyFormatskey5" default="false" type="medium" usage="CURRENCY" formatindex="16">
+      <FormatCode># ##0,00 CCC</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="CurrencyFormatskey6" default="false" type="medium" usage="CURRENCY" formatindex="17">
+      <FormatCode># ##0,-- [CURRENCY];[RED]-# ##0,-- [CURRENCY]</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey1" default="true" type="short" usage="DATE" formatindex="18">
+      <FormatCode>D.M.YY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey9" default="true" type="long" usage="DATE" formatindex="19">
+      <FormatCode>NNNNDD MMMM, YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey8" default="true" type="medium" usage="DATE" formatindex="20">
+      <FormatCode>[NatNum1]DD.MM.YY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey7" default="false" type="medium" usage="DATE" formatindex="21">
+      <FormatCode>[NatNum1]DD.MM.YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey10" default="false" type="long" usage="DATE" formatindex="22">
+      <FormatCode>[NatNum1]D MMM, YY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey11" default="false" type="long" usage="DATE" formatindex="23">
+      <FormatCode>[NatNum1]D MMM, YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey16" default="false" type="long" usage="DATE" formatindex="24">
+      <FormatCode>[NatNum1]D. MMM. YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey12" default="false" type="long" usage="DATE" formatindex="25">
+      <FormatCode>[NatNum1]D MMMM, YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey17" default="false" type="long" usage="DATE" formatindex="26">
+      <FormatCode>[NatNum1]D. MMMM YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey13" default="false" type="long" usage="DATE" formatindex="27">
+      <FormatCode>[NatNum1]NN, D MMM, YY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey2" default="false" type="medium" usage="DATE" formatindex="28">
+      <FormatCode>[NatNum1]NN DD.MMM YY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey14" default="false" type="long" usage="DATE" formatindex="29">
+      <FormatCode>[NatNum1]NN, D MMMM, YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey15" default="false" type="long" usage="DATE" formatindex="30">
+      <FormatCode>[NatNum1]NNNND MMMM, YYYY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey18" default="false" type="short" usage="DATE" formatindex="31">
+      <FormatCode>MM-DD</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey19" default="false" type="medium" usage="DATE" formatindex="32">
+      <FormatCode>YY-MM-DD</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey20" default="false" type="medium" usage="DATE" formatindex="33">
+      <FormatCode>YYYY-MM-DD</FormatCode>
+      <DefaultName>ISO 8601</DefaultName>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey3" default="false" type="medium" usage="DATE" formatindex="34">
+      <FormatCode>MM.YY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey4" default="false" type="medium" usage="DATE" formatindex="35">
+      <FormatCode>DD.MMM</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey5" default="false" type="medium" usage="DATE" formatindex="36">
+      <FormatCode>MMMM</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey6" default="false" type="medium" usage="DATE" formatindex="37">
+      <FormatCode>QQ YY</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateFormatskey21" default="false" type="medium" usage="DATE" formatindex="38">
+      <FormatCode>WW</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="TimeFormatskey1" default="true" type="short" usage="TIME" formatindex="39">
+      <FormatCode>HH:MM</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="TimeFormatskey2" default="true" type="medium" usage="TIME" formatindex="40">
+      <FormatCode>HH:MM:SS</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="TimeFormatskey3" default="false" type="short" usage="TIME" formatindex="41">
+      <FormatCode>HH:MM AM/PM</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="TimeFormatskey4" default="false" type="medium" usage="TIME" formatindex="42">
+      <FormatCode>HH:MM:SS AM/PM</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="TimeFormatskey5" default="false" type="medium" usage="TIME" formatindex="43">
+      <FormatCode>[HH]:MM:SS</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="TimeFormatskey6" default="false" type="short" usage="TIME" formatindex="44">
+      <FormatCode>MM:SS,00</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="TimeFormatskey7" default="false" type="medium" usage="TIME" formatindex="45">
+      <FormatCode>[HH]:MM:SS,00</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateTimeFormatskey1" default="true" type="medium" usage="DATE_TIME" formatindex="46">
+      <FormatCode>DD.MM.YY HH:MM</FormatCode>
+    </FormatElement>
+    <FormatElement msgid="DateTimeFormatskey2" default="false" type="medium" usage="DATE_TIME" formatindex="47">
+      <FormatCode>DD.MM.YYYY HH:MM:SS</FormatCode>
+    </FormatElement>
+  </LC_FORMAT>
   <LC_COLLATION>
     <Collator unoid="charset" default="true"/>
     <CollationOptions>
diff --git a/i18npool/source/nativenumber/data/numberchar.h b/i18npool/source/nativenumber/data/numberchar.h
index cd22f12fd2bd..62e490fd71e1 100644
--- a/i18npool/source/nativenumber/data/numberchar.h
+++ b/i18npool/source/nativenumber/data/numberchar.h
@@ -53,7 +53,8 @@ static const sal_Int16 NumberChar_mn        = 27;
 static const sal_Int16 NumberChar_he        = 28;
 static const sal_Int16 NumberChar_ne        = 29;
 static const sal_Int16 NumberChar_dz        = 30;
-static const sal_Int16 NumberChar_Count     = 31;
+static const sal_Int16 NumberChar_cu        = 31;
+static const sal_Int16 NumberChar_Count     = 32;
 
 static const sal_Unicode NumberChar[][10] = {
 //  0   1   2   3   4   5   6   7   8   9
@@ -88,6 +89,7 @@ static const sal_Unicode NumberChar[][10] = {
     { 0x0020, 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7, 0x05D8 }, // Hebrew
     { 0x0966, 0x0967, 0x0968, 0x0969, 0x096A, 0x096B, 0x096C, 0x096D, 0x096E, 0x096F }, // Nepali
     { 0x0F20, 0x0F21, 0x0F22, 0x0F23, 0x0F24, 0x0F25, 0x0F26, 0x0F27, 0x0F28, 0x0F29 }, // Dzongkha
+    { 0x0030, 0x0430, 0x0432, 0x0433, 0x0434, 0x0454, 0x0455, 0x0437, 0x0438, 0x0473 }, // Church Slavic
 };
 
 static sal_Unicode DecimalChar[] = {
@@ -122,6 +124,7 @@ static sal_Unicode DecimalChar[] = {
     0x0000, // Hebrew
     0x0000, // Nepali
     0x0000, // Dzongkha
+    0x0000, // Church Slavic
 };
 
 static const sal_Unicode MinusChar[] = {
@@ -156,6 +159,7 @@ static const sal_Unicode MinusChar[] = {
     0x0000, // Hebrew
     0x0000, // Nepali
     0x0000, // Dzongkha
+    0x0000, // Church Slavic
 };
 
 static sal_uInt16 SeparatorChar[] = {
@@ -190,6 +194,7 @@ static sal_uInt16 SeparatorChar[] = {
     0x0000, // Hebrew
     0x0000, // Nepali
     0x0000, // Dzongkha
+    0x0000, // Church Slavic
 };
 
 #define NUMBER_ZERO     NumberChar[NumberChar_HalfWidth][0] // 0x0030
diff --git a/i18npool/source/nativenumber/nativenumbersupplier.cxx b/i18npool/source/nativenumber/nativenumbersupplier.cxx
index 76f421433ecd..94b5a78c3f42 100644
--- a/i18npool/source/nativenumber/nativenumbersupplier.cxx
+++ b/i18npool/source/nativenumber/nativenumbersupplier.cxx
@@ -59,6 +59,8 @@ namespace com { namespace sun { namespace star { namespace i18n {
 
 OUString SAL_CALL getHebrewNativeNumberString(const OUString& aNumberString, bool useGeresh);
 
+OUString SAL_CALL getCyrillicNativeNumberString(const OUString& aNumberString);
+
 OUString SAL_CALL AsciiToNativeChar( const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount,
         Sequence< sal_Int32 >& offset, bool useOffset, sal_Int16 number ) throw(RuntimeException)
 {
@@ -458,7 +460,8 @@ static const sal_Char *natnum1Locales[] = {
     "mn",
     "ne",
     "dz",
-    "fa"
+    "fa",
+    "cu"
 };
 static sal_Int16 nbOfLocale = SAL_N_ELEMENTS(natnum1Locales);
 
@@ -489,7 +492,8 @@ static const sal_Int16 natnum1[] = {
     NumberChar_mn,
     NumberChar_ne,
     NumberChar_dz,
-    NumberChar_EastIndic_ar
+    NumberChar_EastIndic_ar,
+    NumberChar_cu
 };
 static const sal_Int16 sizeof_natnum1 = SAL_N_ELEMENTS(natnum1);
 
@@ -592,6 +596,8 @@ OUString SAL_CALL NativeNumberSupplierService::getNativeNumberString(const OUStr
         else if (num == NumberChar_he)
             return getHebrewNativeNumberString(aNumberString,
                     nNativeNumberMode == NativeNumberMode::NATNUM2);
+        else if (num == NumberChar_cu)
+            return getCyrillicNativeNumberString(aNumberString);
         else
             return AsciiToNativeChar(aNumberString, 0, aNumberString.getLength(), offset, useOffset, num);
     }
@@ -902,6 +908,127 @@ OUString SAL_CALL getHebrewNativeNumberString(const OUString& aNumberString, boo
         return aNumberString;
 }
 
+// Support for Cyrillic Numerals
+// See UTN 41 for implementation information
+// http://www.unicode.org/notes/tn41/
+
+static sal_Unicode cyrillicThousandsMark = 0x0482;
+static sal_Unicode cyrillicTitlo = 0x0483;
+static sal_Unicode cyrillicTen = 0x0456;
+
+struct CyrillicNumberChar {
+    sal_Unicode code;
+    sal_Int16 value;
+} CyrillicNumberCharArray[] = {
+    { 0x0446, 900 },
+    { 0x047f, 800 },
+    { 0x0471, 700 },
+    { 0x0445, 600 },
+    { 0x0444, 500 },
+    { 0x0443, 400 },
+    { 0x0442, 300 },
+    { 0x0441, 200 },
+    { 0x0440, 100 },
+    { 0x0447, 90 },
+    { 0x043f, 80 },
+    { 0x047b, 70 },
+    { 0x046f, 60 },
+    { 0x043d, 50 },
+    { 0x043c, 40 },
+    { 0x043b, 30 },
+    { 0x043a, 20 },
+    { 0x0456, 10 },
+    { 0x0473, 9 },
+    { 0x0438, 8 },
+    { 0x0437, 7 },
+    { 0x0455, 6 },
+    { 0x0454, 5 },
+    { 0x0434, 4 },
+    { 0x0433, 3 },
+    { 0x0432, 2 },
+    { 0x0430, 1 }
+};
+
+static sal_Int16 nbOfCyrillicNumberChar = sizeof(CyrillicNumberCharArray)/sizeof(CyrillicNumberChar);
+
+void makeCyrillicNumber(sal_Int64 value, OUStringBuffer& output, bool addTitlo)
+{
+    sal_Int16 num = sal::static_int_cast<sal_Int16>(value % 1000);
+    if (value >= 1000) {
+        output.append(cyrillicThousandsMark);
+        makeCyrillicNumber(value / 1000, output, false);
+        if (value >= 10000 && (value - 10000) % 1000 != 0) {
+            output.append(" ");
+        }
+        if (value % 1000 == 0)
+            addTitlo = false;
+    }
+
+    for (sal_Int32 j = 0; num > 0 && j < nbOfCyrillicNumberChar; j++) {
+        if (num < 20 && num > 10) {
+            num -= 10;
+            makeCyrillicNumber(num, output, false);
+            output.append(cyrillicTen);
+            break;
+        }
+
+        if (CyrillicNumberCharArray[j].value <= num) {
+            output.append(CyrillicNumberCharArray[j].code);
+            num = sal::static_int_cast<sal_Int16>( num - CyrillicNumberCharArray[j].value );
+        }
+    }
+
+    if (addTitlo) {
+        if (output.getLength() == 1) {
+            output.append(cyrillicTitlo);
+        } else if (output.getLength() == 2) {
+            if (value > 800 && value < 900) {
+                output.append(cyrillicTitlo);
+            } else {
+                output.insert(1, cyrillicTitlo);
+            }
+        } else if (output.getLength() > 2) {
+            if (output.indexOf(" ") == output.getLength() - 2) {
+                output.append(cyrillicTitlo);
+            } else {
+                output.insert(output.getLength() - 1, cyrillicTitlo);
+            }
+        }
+    }
+}
+
+OUString SAL_CALL getCyrillicNativeNumberString(const OUString& aNumberString)
+{
+    sal_Int64 value = 0;
+    sal_Int32 i, count = 0, len = aNumberString.getLength();
+    const sal_Unicode *src = aNumberString.getStr();
+
+    for (i = 0; i < len; i++) {
+        sal_Unicode ch = src[i];
+        if (isNumber(ch)) {
+            if (++count >= 8) // Number is too long, could not be handled.
+                return aNumberString;
+            value = value * 10 + (ch - NUMBER_ZERO);
+        }
+        else if (isSeparator(ch) && count > 0) continue;
+        else if (isMinus(ch) && count == 0) continue;
+        else break;
+    }
+
+    if (value > 0) {
+        OUStringBuffer output(count*2 + 2 + len - i);
+
+        makeCyrillicNumber(value, output, true);
+
+        if (i < len)
+            output.append(aNumberString.copy(i));
+
+        return output.makeStringAndClear();
+    }
+    else
+        return aNumberString;
+}
+
 static const sal_Char* implementationName = "com.sun.star.i18n.NativeNumberSupplier";
 
 OUString SAL_CALL NativeNumberSupplierService::getImplementationName() throw( RuntimeException, std::exception )