Code:
/ Dotnetfx_Vista_SP2 / Dotnetfx_Vista_SP2 / 8.0.50727.4016 / DEVDIV / depot / DevDiv / releases / Orcas / QFE / wpf / src / Core / CSharp / MS / Internal / Shaping / IndicCharClassifier.cs / 1 / IndicCharClassifier.cs
//----------------------------------------------------------------------
//
// Microsoft Windows Client Platform
// Copyright (C) Microsoft Corporation, 2003
//
// File: IndicCharClassifier.cs
//
// Contents: Implementation of Indic shaping engine and its factory
//
// Created: 03-04-2007 Nick Beal (nbeal)
//
//-----------------------------------------------------------------------
// #define VALIDATE_CLUSTER_PARAMETERS
using System;
using System.Security;
using System.Security.Permissions;
using System.Diagnostics;
using System.Collections;
using System.Globalization;
using System.Windows;
using System.Windows.Media;
using System.Windows.Media.TextFormatting;
using MS.Internal.FontCache;
using MS.Internal.FontFace;
using MS.Internal.PresentationCore;
namespace MS.Internal.Shaping
{
internal enum IndicScriptIndices : byte
{
Devanagari,
Bengali,
Gurmukhi,
Gujarati,
Oriya,
Tamil,
Telugu,
Kannada,
Malayalam
};
///
/// IndicCharClassifier -
///
internal class IndicCharClassifier : ShaperCharacterClassifier
{
internal const byte CharsPerIndicScript = 0x80;
internal const ushort RaOffset = 0x30;
internal const ushort HalantOffset = 0x4d;
private IndicScriptIndices _scriptIx;
private DependentCharacterList[] _scriptMatraInvalidVowels;
private DependentCharacterList[] _compositeChars;
private byte[] _charFormsInfo;
bool _isFormsDiscoveryDone = false;
public IndicCharClassifier( ScriptTags scriptTag,
GlyphTypeface fontFace ) : base(scriptTag, fontFace)
{
_scriptIx = GetScriptIx(scriptTag);
_unknownClass = (byte)IndicCharClass.Reserved;
_spaceClass = (byte)IndicCharClass.Neutral;
_zwControlClass = (byte)IndicCharClass.Neutral;
_zwjClass = (byte)IndicCharClass.ZWJ;
_zwnjClass = (byte)IndicCharClass.ZWNJ;
_shyClass = (byte)IndicCharClass.Reserved;
_firstChar = '\u0900'; // this should be set to the script's first Unicode char
_lastChar = '\u0d7f'; // this should be set to the script's last Unicode char
_xorRange = 0x0080; // this range is used in GetCharClass
_xorMask = (ushort) ((ushort)_firstChar + ((ushort)_scriptIx << 7)); // this mask is used in GetCharClass (scriptIx * 0x80)
_charClassTable = _indicCharClasses;
_compositeChars = GetCompositeCharactersList();
_scriptMatraInvalidVowels = GetInvalidVowelsList();
_charFormsInfo = null;
}
///
///IndicCharClassifier.ToShapeInfo - returns CharShapeInfo for the Unicode character
///
public override CharShapeInfo ToShapeInfo (char unicodeChar )
{
CharShapeInfo charShape = base.ToShapeInfo(unicodeChar);
// if the character is returned as "unknown", check for vedic characters
if (charShape == (CharShapeInfo.IsStartOfCluster | (CharShapeInfo)_unknownClass))
{
if (((ushort)unicodeChar >= 0x0951 && (ushort)unicodeChar <= 0x0965) &&
((ushort)unicodeChar <= 0x00955 || (ushort)unicodeChar >= 0x00964))
{
// all Indic text can include Devanagari Anudatta or Udatta...
charShape = (CharShapeInfo) _charClassTable[(int)unicodeChar - _firstChar];
}
}
return charShape;
}
///
/// componentRepoInfo - returns the repositioning class info.
///
public IndicRepositioningClass ToRepositioningClass ( char unicodeChar )
{
CharShapeInfo charShape = ToShapeInfo(unicodeChar);
Debug.Assert((charShape & (CharShapeInfo) IndicCharClass.IncludesPositioningInfo) != 0,"\"Includes Positioning Info\" flag not set.");
charShape = (CharShapeInfo)((ushort)charShape >> IndicShape.RepositioningInfoShift);
return (IndicRepositioningClass)charShape;
}
///
///IndicCharClassifier.ToFormInfo - returns IndicFormFlags for the Unicode character
///
public IndicFormFlags ToFormInfo (char unicodeChar )
{
IndicFormFlags charFormInfo = 0;
if (((ushort)unicodeChar ^ _xorMask) < _xorRange)
{
if (_charFormsInfo == null)
{
charFormInfo = (IndicFormFlags)_indicCharFormsInfo[unicodeChar - _firstChar];
}
else
{
charFormInfo = (IndicFormFlags)_charFormsInfo[unicodeChar - FirstCharacterInScriptRange];
}
}
return charFormInfo;
}
///
/// ToDecompositionList - returns the decomposition character sequence for the given character
/// (if one exists)
///
public char[] ToDecompositionList ( char unicodeChar )
{
return GetDependentCharacterList(unicodeChar, _compositeChars);
}
///
/// GetInvalidVowelList - returns the decomposition character sequence for the given character
/// (if one exists)
///
public char[] GetMatraInvalidVowelList ( char unicodeChar )
{
return GetDependentCharacterList(unicodeChar, _scriptMatraInvalidVowels);
}
public IndicRepositioningClass HalantRepositioningClass
{
get { return ToRepositioningClass( (char)(_xorMask + HalantOffset) ); }
}
public IndicScriptIndices ScriptIx
{
get { return _scriptIx; }
}
public char FirstCharacterInScriptRange
{
get { return (char)_xorMask; }
}
public char[][] TamilAkhands
{
get { return _tamilAkhands; }
}
public char HalantCharacter
{
get { return (char)(_xorMask + HalantOffset); }
}
///
/// Critical - calls critical code
///
[SecurityCritical]
public bool GetFontCharacterFormsInfo (IndicFontClient fontClient)
{
if (!_isFormsDiscoveryDone)
{
byte[] fontCharacterForms = new byte[IndicCharClassifier.CharsPerIndicScript];
for (int i = 0; i < IndicCharClassifier.CharsPerIndicScript; ++i)
{
fontCharacterForms[i] = (byte)ToFormInfo((char)(_xorMask + i));
}
// see if font has anything to add. (If so, keep the new info)
if (fontClient.GetCharacterFormsInfo(fontCharacterForms))
{
_charFormsInfo = fontCharacterForms;
}
_isFormsDiscoveryDone = true;
}
return true;
}
private IndicScriptIndices GetScriptIx (ScriptTags scriptTag)
{
switch (scriptTag)
{
case ScriptTags.Devanagari:
return IndicScriptIndices.Devanagari;
case ScriptTags.Bengali:
return IndicScriptIndices.Bengali;
case ScriptTags.Gurmukhi:
return IndicScriptIndices.Gurmukhi;
case ScriptTags.Gujarati:
return IndicScriptIndices.Gujarati;
case ScriptTags.Oriya:
return IndicScriptIndices.Oriya;
case ScriptTags.Tamil:
return IndicScriptIndices.Tamil;
case ScriptTags.Telugu:
return IndicScriptIndices.Telugu;
case ScriptTags.Kannada:
return IndicScriptIndices.Kannada;
case ScriptTags.Malayalam:
return IndicScriptIndices.Malayalam;
case ScriptTags.Devanagari_v2:
return IndicScriptIndices.Devanagari;
case ScriptTags.Bengali_v2:
return IndicScriptIndices.Bengali;
case ScriptTags.Gurmukhi_v2:
return IndicScriptIndices.Gurmukhi;
case ScriptTags.Gujarati_v2:
return IndicScriptIndices.Gujarati;
case ScriptTags.Oriya_v2:
return IndicScriptIndices.Oriya;
case ScriptTags.Tamil_v2:
return IndicScriptIndices.Tamil;
case ScriptTags.Telugu_v2:
return IndicScriptIndices.Telugu;
case ScriptTags.Kannada_v2:
return IndicScriptIndices.Kannada;
case ScriptTags.Malayalam_v2:
return IndicScriptIndices.Malayalam;
default:
throw new NotSupportedException();
}
}
private DependentCharacterList[] GetCompositeCharactersList ()
{
switch (_scriptIx)
{
case IndicScriptIndices.Devanagari:
return null;
case IndicScriptIndices.Bengali:
return _bengaliSplitMatras;
case IndicScriptIndices.Gurmukhi:
return null;
case IndicScriptIndices.Gujarati:
return null;
case IndicScriptIndices.Oriya:
return _oriyaSplitMatras;
case IndicScriptIndices.Tamil:
return _tamilSplitMatras;
case IndicScriptIndices.Telugu:
return _teluguSplitMatras;
case IndicScriptIndices.Kannada:
return _kannadaSplitMatras;
case IndicScriptIndices.Malayalam:
return _malayalamSplitMatras;
default:
throw new NotSupportedException();
}
}
private DependentCharacterList[] GetInvalidVowelsList ()
{
switch (_scriptIx)
{
case IndicScriptIndices.Devanagari:
return _hindiMatraInvalidVowels;
case IndicScriptIndices.Bengali:
return _bengaliMatraInvalidVowels;
case IndicScriptIndices.Gurmukhi:
return _gurmukhiMatraInvalidVowels;
case IndicScriptIndices.Gujarati:
return _gujaratiMatraInvalidVowels;
case IndicScriptIndices.Oriya:
return _oriyaMatraInvalidVowels;
case IndicScriptIndices.Tamil:
return null;
case IndicScriptIndices.Telugu:
return _teluguMatraInvalidVowels;
case IndicScriptIndices.Kannada:
return _kannadaMatraInvalidVowels;
case IndicScriptIndices.Malayalam:
return _malayalamMatraInvalidVowels;
default:
throw new NotSupportedException();
}
}
// Bengali Script split matras
private static readonly DependentCharacterList[] _bengaliSplitMatras =
{
new DependentCharacterList( '\u09cb', new char[] {'\u09c7','\u09be'} ),
new DependentCharacterList( '\u09cc', new char[] {'\u09c7','\u09d7'} )
};
// Bengali invalid vowel matras
private static readonly DependentCharacterList[] _bengaliMatraInvalidVowels =
{
new DependentCharacterList( '\u09be', new char[] {'\u0985'} ),
new DependentCharacterList( '\u09c3', new char[] {'\u098b'} ),
new DependentCharacterList( '\u09e2', new char[] {'\u098c'} ),
};
// Gujarati invalid vowel matras
private static readonly DependentCharacterList[] _gujaratiMatraInvalidVowels =
{
new DependentCharacterList( '\u0abe', new char[] {'\u0a85'} ),
new DependentCharacterList( '\u0ac5', new char[] {'\u0a85','\u0a86'} ),
new DependentCharacterList( '\u0ac7', new char[] {'\u0a85','\u0a86'} ),
new DependentCharacterList( '\u0ac8', new char[] {'\u0a85','\u0a86'} ),
new DependentCharacterList( '\u0ac9', new char[] {'\u0a85'} ),
new DependentCharacterList( '\u0acb', new char[] {'\u0a85'} ),
new DependentCharacterList( '\u0acc', new char[] {'\u0a85'} ),
};
// Gurmukhi invalid vowel matras
private static readonly DependentCharacterList[] _gurmukhiMatraInvalidVowels =
{
new DependentCharacterList( '\u0a3e', new char[] {'\u0a05'} ),
new DependentCharacterList( '\u0a3f', new char[] {'\u0a72'} ),
new DependentCharacterList( '\u0a40', new char[] {'\u0a72'} ),
new DependentCharacterList( '\u0a41', new char[] {'\u0a73'} ),
new DependentCharacterList( '\u0a42', new char[] {'\u0a73'} ),
new DependentCharacterList( '\u0a47', new char[] {'\u0a72'} ),
new DependentCharacterList( '\u0a48', new char[] {'\u0a05'} ),
new DependentCharacterList( '\u0a4c', new char[] {'\u0a05'} )
};
// Hindi (Devanagari) invalid vowel matras
private static readonly DependentCharacterList[] _hindiMatraInvalidVowels =
{
new DependentCharacterList( '\u093e', new char[] {'\u0905'} ),
new DependentCharacterList( '\u0945', new char[] {'\u0906','\u090f'} ),
new DependentCharacterList( '\u0946', new char[] {'\u0905','\u0906','\u090f'} ),
new DependentCharacterList( '\u0947', new char[] {'\u0906','\u090f'} ),
new DependentCharacterList( '\u0948', new char[] {'\u0906'} ),
new DependentCharacterList( '\u0949', new char[] {'\u0905'} ),
new DependentCharacterList( '\u094a', new char[] {'\u0905'} ),
new DependentCharacterList( '\u094b', new char[] {'\u0905'} ),
new DependentCharacterList( '\u094c', new char[] {'\u0905'} )
};
// Kannada Script split matras
private static readonly DependentCharacterList[] _kannadaSplitMatras =
{
new DependentCharacterList('\u0cc0',new char[] {'\u0cbf','\u0cd5'}),
new DependentCharacterList('\u0cc7',new char[] {'\u0cc6','\u0cd5'}),
new DependentCharacterList('\u0cc8',new char[] {'\u0cc6','\u0cd6'}),
new DependentCharacterList('\u0cca',new char[] {'\u0cc6','\u0cc2'}),
new DependentCharacterList('\u0ccb',new char[] {'\u0cc6','\u0cc2','\u0cd5'}),
};
// Kannada invalid vowel matras
private static readonly DependentCharacterList[] _kannadaMatraInvalidVowels =
{
new DependentCharacterList( '\u0ccc', new char[] {'\u0c92'} )
};
// Malayalam Script split matras
private static readonly DependentCharacterList[] _malayalamSplitMatras =
{
new DependentCharacterList('\u0d4a', new char[] {'\u0d46','\u0d3e'}),
new DependentCharacterList('\u0d4b', new char[] {'\u0d47','\u0d3e'}),
new DependentCharacterList('\u0d4c', new char[] {'\u0d46','\u0d57'}),
};
// Malayalam invalid vowel matras
private static readonly DependentCharacterList[] _malayalamMatraInvalidVowels =
{
new DependentCharacterList( '\u0d3e', new char[] {'\u0d12'} ),
new DependentCharacterList( '\u0d46', new char[] {'\u0d0e'} ),
new DependentCharacterList( '\u0d57', new char[] {'\u0d07','\u0d09','\u0d12'} ),
};
// Oriya Script split matras
private static readonly DependentCharacterList[] _oriyaSplitMatras =
{
new DependentCharacterList('\u0b48', new char[] {'\u0b47','\u0b56'}),
new DependentCharacterList('\u0b4b', new char[] {'\u0b47','\u0b3e'}),
new DependentCharacterList('\u0b4c', new char[] {'\u0b47','\u0b57'}),
};
// Oriya invalid vowel matras
private static readonly DependentCharacterList[] _oriyaMatraInvalidVowels =
{
new DependentCharacterList( '\u0b3e', new char[] {'\u0b05'} )
};
private static readonly DependentCharacterList[] _tamilSplitMatras =
{
// Tamil Script split matras
new DependentCharacterList('\u0b94', new char[] {'\u0b92','\u0bd7'}),
new DependentCharacterList('\u0bca', new char[] {'\u0bc6','\u0bbe'}),
new DependentCharacterList('\u0bcb', new char[] {'\u0bc7','\u0bbe'}),
new DependentCharacterList('\u0bcc', new char[] {'\u0bc6','\u0bd7'}),
};
private static readonly char[][] _tamilAkhands =
{
// Tamil Script akhands
new char[] {'\u0b95','\u0bcd','\u0bb7'},
new char[] {'\u0bb6','\u0bcd','\u0bb0','\u0bc0'},
new char[] {'\u0bb8','\u0bcd','\u0bb0','\u0bc0'},
};
// Telugu Script split matras
private static readonly DependentCharacterList[] _teluguSplitMatras =
{
new DependentCharacterList('\u0c48', new char[] {'\u0c46','\u0c56'}),
};
// Telugu invalid vowel matras
private static readonly DependentCharacterList[] _teluguMatraInvalidVowels =
{
new DependentCharacterList( '\u0c3e', new char[] {'\u0c0b','\u0c0c'} ),
new DependentCharacterList( '\u0c4c', new char[] {'\u0c12'} ),
new DependentCharacterList( '\u0c55', new char[] {'\u0c12'} )
};
#region Classification Tables
private const byte RepositioningClassShift = (byte)IndicRepositioningClass.RepositioningClassShift;
private const byte UndefinedChar = (byte)IndicCharClass.Reserved;
private const byte Neutral = (byte)IndicCharClass.Neutral;
private const byte Consonant = (byte)IndicCharClass.Consonant;
private const byte NuktaConsonant = (byte)IndicCharClass.NuktaConsonant;
private const byte Ra = (byte)(IndicCharClass.Ra | IndicCharClass.IncludesPositioningInfo);
private const byte Vowel = (byte)IndicCharClass.Vowel;
private const byte Nukta = (byte)IndicCharClass.Nukta;
private const byte Halant = (byte)(IndicCharClass.Halant | IndicCharClass.IncludesPositioningInfo);
private const byte Vedic = (byte)(IndicCharClass.Vedic | IndicCharClass.IncludesPositioningInfo);
private const byte VowelSign = (byte)(IndicCharClass.VowelSign | IndicCharClass.IncludesPositioningInfo);
private const byte Matra = (byte)(IndicCharClass.Matra | IndicCharClass.IncludesPositioningInfo);
private const byte SplitMatra = (byte)IndicCharClass.Matra;
private const byte RepositionAtStart = ((byte)IndicRepositioningClass.AtStart << RepositioningClassShift);
private const byte RepositionBeforeMain = ((byte)IndicRepositioningClass.BeforeMain << RepositioningClassShift);
private const byte RepositionAfterMain = ((byte)IndicRepositioningClass.AfterMain << RepositioningClassShift);
private const byte RepositionBeforeSub = ((byte)IndicRepositioningClass.BeforeSub << RepositioningClassShift);
private const byte RepositionAfterSub = ((byte)IndicRepositioningClass.AfterSub << RepositioningClassShift);
private const byte RepositionBeforePost = ((byte)IndicRepositioningClass.BeforePost << RepositioningClassShift);
private const byte RepositionAfterPost = ((byte)IndicRepositioningClass.AfterPost << RepositioningClassShift);
private const byte RepositionBeforeEnd = ((byte)IndicRepositioningClass.BeforeEnd << RepositioningClassShift);
private static readonly byte[] _indicCharClasses =
{
// All the Devanagari Unicode chars (U+900 - U+97F) classified
//090 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
Vowel, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
//091 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//092 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, NuktaConsonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//093 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionBeforePost,
NuktaConsonant, Consonant, Consonant,
NuktaConsonant, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterSub,
Matra|RepositionAtStart,
//094 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Halant|RepositionAfterSub,
UndefinedChar, UndefinedChar,
//095 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, Vedic|RepositionBeforeEnd,
Vedic|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, UndefinedChar, UndefinedChar,
NuktaConsonant, NuktaConsonant, NuktaConsonant, NuktaConsonant,
NuktaConsonant, NuktaConsonant, NuktaConsonant, NuktaConsonant,
//096 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//097 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Consonant,
Consonant, Neutral, Consonant, Consonant,
// All the Bengali Unicode chars (U+980 - U+9FF) classified
//098 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionAfterPost,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, UndefinedChar, Vowel,
//099 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//09A 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//09B 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterSub,
UndefinedChar, Consonant, UndefinedChar,
UndefinedChar, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterPost,
Matra|RepositionAtStart,
//09C 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, Matra|RepositionAtStart,
Matra|RepositionAtStart,
UndefinedChar, UndefinedChar, SplitMatra,
SplitMatra, Halant|RepositionAfterPost,
Consonant, UndefinedChar,
//09D 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
NuktaConsonant, NuktaConsonant, UndefinedChar, NuktaConsonant,
//09E 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//09F 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterSub,
Consonant, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Gurmukhi Unicode chars (U+a00 - U+a7F) classified
//0A0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionAfterPost,
VowelSign|RepositionAfterPost,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Vowel,
//0A1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0A2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0A3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionBeforePost,
UndefinedChar, Consonant, NuktaConsonant,
UndefinedChar, Consonant, NuktaConsonant, UndefinedChar,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAtStart,
//0A4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Halant|RepositionAfterSub,
UndefinedChar, UndefinedChar,
//0A5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, NuktaConsonant, NuktaConsonant, NuktaConsonant,
Consonant, UndefinedChar, NuktaConsonant, UndefinedChar,
//0A6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0A7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
VowelSign|RepositionAfterPost,
VowelSign|RepositionAfterPost,
Vowel, Vowel,
Neutral, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Gujarati Unicode chars (U+a80 - U+afF) classified
//0A8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, Vowel, UndefinedChar, Vowel,
//0A9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0AA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0AB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionBeforePost,
UndefinedChar, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterSub,
Matra|RepositionAtStart,
//0AC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Halant|RepositionAfterSub,
UndefinedChar, UndefinedChar,
//0AD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0AE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0AF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, Neutral, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Oriya Unicode chars (U+B00 - U+B7F) classified
//0B0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeSub,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, UndefinedChar, Vowel,
//0B1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0B2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0B3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterMain,
UndefinedChar, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterPost,
Matra|RepositionAfterMain,
//0B4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAtStart,
SplitMatra,
UndefinedChar, UndefinedChar, SplitMatra,
SplitMatra,
Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0B5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionAfterMain,
Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
NuktaConsonant, NuktaConsonant, UndefinedChar, Consonant,
//0B6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0B7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Tamil Unicode chars (U+B80 - U+BFF) classified
//0B8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, VowelSign|RepositionBeforeEnd,
Neutral, //Vowel,
UndefinedChar, Vowel, Consonant, Vowel,
Vowel, Vowel, Vowel, UndefinedChar,
UndefinedChar, UndefinedChar, Vowel, Vowel,
//0B9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, Consonant, Consonant, UndefinedChar,
Consonant, UndefinedChar, Consonant, Consonant,
//0BA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, Consonant,
Consonant, UndefinedChar, UndefinedChar, UndefinedChar,
Consonant, Consonant, Consonant, UndefinedChar,
UndefinedChar, UndefinedChar, Consonant, Consonant,
//0BB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterPost,
Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
//0BC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterSub,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionBeforeMain,
Matra|RepositionBeforeMain,
Matra|RepositionBeforeMain,
UndefinedChar, SplitMatra, SplitMatra,
SplitMatra, Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0BD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0BE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0BF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Telugu Unicode chars (U+c00 - U+c7F) classified
//0C0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, Vowel, Vowel,
//0C1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0C2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0C3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterPost,
Consonant, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
//0C4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
SplitMatra, UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0C5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0C6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0C7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Kannada Unicode chars (U+C80 - U+CFF) classified
//0C8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, Vowel, Vowel,
//0C9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0CA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0CB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterPost,
Consonant, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
//0CC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SplitMatra, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionBeforeSub,
SplitMatra,
SplitMatra, UndefinedChar, SplitMatra, SplitMatra,
Matra|RepositionBeforeSub,
Halant|RepositionBeforeSub,
UndefinedChar, UndefinedChar,
//0CD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Consonant, UndefinedChar,
//0CE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0CF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Malayalam Unicode chars (U+d00 - U+d7F) classified
//0D0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, Vowel, Vowel,
//0D1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0D2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0D3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterMain,
Ra|RepositionAfterMain,
Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
//0D4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, Matra|RepositionAtStart,
Matra|RepositionAtStart,
Matra|RepositionAtStart,
UndefinedChar, SplitMatra, SplitMatra,
SplitMatra, Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0D5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0D6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0D7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
};
#endregion // end of Classification Tables
#region Character Form Tables
private const byte PreBaseComponent = (byte)IndicFormFlags.PreBaseMatra;
private const byte AboveBaseComponent = (byte)IndicFormFlags.AboveBaseMatra;
private const byte BelowBaseComponent = (byte)IndicFormFlags.BelowBaseMatra;
private const byte PostBaseComponent = (byte)IndicFormFlags.PostBaseMatra;
private const byte PreBaseMatra = (byte)IndicFormFlags.PreBaseMatra;
private const byte AboveBaseMatra = (byte)IndicFormFlags.AboveBaseMatra;
private const byte BelowBaseMatra = (byte)IndicFormFlags.BelowBaseMatra;
private const byte PostBaseMatra = (byte)IndicFormFlags.PostBaseMatra;
private const byte Reph = (byte)IndicFormFlags.RephForm;
private const byte Vattu = (byte)IndicFormFlags.VattuForm;
private const byte InitialChar = (byte)IndicFormFlags.InitForm;
private const byte PostBaseReph = (byte)IndicFormFlags.PostbaseReph;
private const byte DravidianBase = (byte)IndicFormFlags.DravidianBase;
private const byte SubForm = (byte)IndicFormFlags.SubForm;
private const byte PostForm = (byte)IndicFormFlags.PostForm;
private static readonly byte[] _indicCharFormsInfo =
{
// All the Devanagari Unicode chars (U+900 - U+97F) forms info
//090 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//091 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//092 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//093 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | Vattu, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//094 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
BelowBaseMatra, AboveBaseMatra, AboveBaseMatra, AboveBaseMatra,
AboveBaseMatra, PostBaseMatra, PostBaseMatra, PostBaseMatra,
PostBaseMatra, 0, 0, 0,
//095 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//096 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, BelowBaseMatra, BelowBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//097 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Bengali Unicode chars (U+980 - U+9FF) consonant/matra forms (the consonant
// forms will be updated from the font)
//098 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//099 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//09A 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
SubForm, 0, 0, PostForm,
//09B 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
InitialChar | Reph | SubForm,
0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//09C 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
BelowBaseMatra, 0, 0, PreBaseMatra|InitialChar,
PreBaseMatra|InitialChar,
0, 0, InitialChar|PreBaseComponent|PostBaseComponent,
InitialChar|PreBaseComponent|PostBaseComponent,
0, 0, 0,
//09D 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, PostBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
//09E 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, BelowBaseMatra,
BelowBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//09F 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | SubForm, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Gurmukhi Unicode chars (U+a00 - U+a7F) classified
//0A0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, PostForm,
//0A3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, 0, 0, 0,
0, SubForm, 0, 0,
0, SubForm, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//0A4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, 0,
0, 0, 0, AboveBaseMatra,
AboveBaseMatra, 0, 0, AboveBaseMatra,
AboveBaseMatra, 0, 0, 0,
//0A5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Gujarati Unicode chars (U+a80 - U+afF) classified
//0A8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | Vattu, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//0AC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
BelowBaseMatra, AboveBaseMatra, 0, AboveBaseMatra,
AboveBaseMatra, PostBaseMatra, 0, PostBaseMatra,
PostBaseMatra, 0, 0, 0,
//0AD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, BelowBaseMatra, BelowBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Oriya Unicode chars (U+B00 - U+B7F) classified
//0B0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
SubForm, 0, 0, 0,
SubForm, 0, 0, 0,
SubForm, SubForm, SubForm, PostForm,
//0B3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | SubForm, 0, SubForm, SubForm,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, AboveBaseMatra,
//0B4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
0, 0, 0, PreBaseMatra,
PreBaseComponent|AboveBaseComponent,
0, 0, PreBaseComponent|PostBaseComponent,
PreBaseComponent|AboveBaseComponent|PostBaseComponent,
0, 0, 0,
//0B5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, AboveBaseMatra, PostBaseMatra|AboveBaseComponent,
0, 0, 0, 0,
0, 0, 0, PostForm,
//0B6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Tamil Unicode chars (U+B80 - U+BFF) classified
//0B8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0BA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0BB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PostBaseMatra,
//0BC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
AboveBaseMatra, BelowBaseMatra, BelowBaseMatra, 0,
0, 0, PreBaseMatra, PreBaseMatra,
PreBaseMatra, 0, PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
0, 0, 0,
//0BD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, PostBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
//0BE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0BF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Telugu Unicode chars (U+c00 - U+c7F) classified
//0C0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0C2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, 0, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0C3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, SubForm, SubForm, SubForm,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, 0, 0,
0, 0, AboveBaseMatra, AboveBaseMatra,
//0C4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
AboveBaseMatra, PostBaseMatra, PostBaseMatra, PostBaseMatra ,
PostBaseMatra, 0, AboveBaseMatra, AboveBaseMatra,
AboveBaseComponent|BelowBaseComponent,
0, AboveBaseMatra, AboveBaseMatra,
AboveBaseMatra, 0, 0, 0,
//0C5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, AboveBaseMatra, BelowBaseMatra, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Kannada Unicode chars (U+C80 - U+CFF) classified
//0C8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0CA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, 0, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0CB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | SubForm, SubForm, SubForm, SubForm,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, 0, 0,
0, 0, PostBaseMatra, AboveBaseMatra,
//0CC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
AboveBaseComponent|PostBaseComponent,
PostBaseMatra, PostBaseMatra, PostBaseMatra,
PostBaseMatra, 0, AboveBaseMatra, AboveBaseComponent|PostBaseComponent,
AboveBaseComponent|PostBaseComponent,
0, AboveBaseComponent|PostBaseComponent,
AboveBaseComponent|PostBaseComponent,
AboveBaseMatra, 0, 0, 0,
//0CD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, PostBaseMatra, PostBaseMatra, 0,
0, 0, 0, 0,
0, 0, SubForm, 0,
//0CE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0CF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Malayalam Unicode chars (U+d00 - U+d7F) classified
//0D0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0D1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0D2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, PostForm,
//0D3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostForm|PostBaseReph,
DravidianBase, SubForm, DravidianBase,
DravidianBase, PostForm, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PostBaseMatra,
//0D4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, PostBaseMatra, PostBaseMatra, PostBaseMatra,
0, 0, PreBaseMatra, PreBaseMatra,
PreBaseMatra, 0, PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
0, 0, 0,
//0D5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, PostBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
//0D6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0D7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
};
#endregion // Character Form Tables
}
}
// File provided for Reference Use Only by Microsoft Corporation (c) 2007.
// Copyright (c) Microsoft Corporation. All rights reserved.
//----------------------------------------------------------------------
//
// Microsoft Windows Client Platform
// Copyright (C) Microsoft Corporation, 2003
//
// File: IndicCharClassifier.cs
//
// Contents: Implementation of Indic shaping engine and its factory
//
// Created: 03-04-2007 Nick Beal (nbeal)
//
//-----------------------------------------------------------------------
// #define VALIDATE_CLUSTER_PARAMETERS
using System;
using System.Security;
using System.Security.Permissions;
using System.Diagnostics;
using System.Collections;
using System.Globalization;
using System.Windows;
using System.Windows.Media;
using System.Windows.Media.TextFormatting;
using MS.Internal.FontCache;
using MS.Internal.FontFace;
using MS.Internal.PresentationCore;
namespace MS.Internal.Shaping
{
internal enum IndicScriptIndices : byte
{
Devanagari,
Bengali,
Gurmukhi,
Gujarati,
Oriya,
Tamil,
Telugu,
Kannada,
Malayalam
};
///
/// IndicCharClassifier -
///
internal class IndicCharClassifier : ShaperCharacterClassifier
{
internal const byte CharsPerIndicScript = 0x80;
internal const ushort RaOffset = 0x30;
internal const ushort HalantOffset = 0x4d;
private IndicScriptIndices _scriptIx;
private DependentCharacterList[] _scriptMatraInvalidVowels;
private DependentCharacterList[] _compositeChars;
private byte[] _charFormsInfo;
bool _isFormsDiscoveryDone = false;
public IndicCharClassifier( ScriptTags scriptTag,
GlyphTypeface fontFace ) : base(scriptTag, fontFace)
{
_scriptIx = GetScriptIx(scriptTag);
_unknownClass = (byte)IndicCharClass.Reserved;
_spaceClass = (byte)IndicCharClass.Neutral;
_zwControlClass = (byte)IndicCharClass.Neutral;
_zwjClass = (byte)IndicCharClass.ZWJ;
_zwnjClass = (byte)IndicCharClass.ZWNJ;
_shyClass = (byte)IndicCharClass.Reserved;
_firstChar = '\u0900'; // this should be set to the script's first Unicode char
_lastChar = '\u0d7f'; // this should be set to the script's last Unicode char
_xorRange = 0x0080; // this range is used in GetCharClass
_xorMask = (ushort) ((ushort)_firstChar + ((ushort)_scriptIx << 7)); // this mask is used in GetCharClass (scriptIx * 0x80)
_charClassTable = _indicCharClasses;
_compositeChars = GetCompositeCharactersList();
_scriptMatraInvalidVowels = GetInvalidVowelsList();
_charFormsInfo = null;
}
///
///IndicCharClassifier.ToShapeInfo - returns CharShapeInfo for the Unicode character
///
public override CharShapeInfo ToShapeInfo (char unicodeChar )
{
CharShapeInfo charShape = base.ToShapeInfo(unicodeChar);
// if the character is returned as "unknown", check for vedic characters
if (charShape == (CharShapeInfo.IsStartOfCluster | (CharShapeInfo)_unknownClass))
{
if (((ushort)unicodeChar >= 0x0951 && (ushort)unicodeChar <= 0x0965) &&
((ushort)unicodeChar <= 0x00955 || (ushort)unicodeChar >= 0x00964))
{
// all Indic text can include Devanagari Anudatta or Udatta...
charShape = (CharShapeInfo) _charClassTable[(int)unicodeChar - _firstChar];
}
}
return charShape;
}
///
/// componentRepoInfo - returns the repositioning class info.
///
public IndicRepositioningClass ToRepositioningClass ( char unicodeChar )
{
CharShapeInfo charShape = ToShapeInfo(unicodeChar);
Debug.Assert((charShape & (CharShapeInfo) IndicCharClass.IncludesPositioningInfo) != 0,"\"Includes Positioning Info\" flag not set.");
charShape = (CharShapeInfo)((ushort)charShape >> IndicShape.RepositioningInfoShift);
return (IndicRepositioningClass)charShape;
}
///
///IndicCharClassifier.ToFormInfo - returns IndicFormFlags for the Unicode character
///
public IndicFormFlags ToFormInfo (char unicodeChar )
{
IndicFormFlags charFormInfo = 0;
if (((ushort)unicodeChar ^ _xorMask) < _xorRange)
{
if (_charFormsInfo == null)
{
charFormInfo = (IndicFormFlags)_indicCharFormsInfo[unicodeChar - _firstChar];
}
else
{
charFormInfo = (IndicFormFlags)_charFormsInfo[unicodeChar - FirstCharacterInScriptRange];
}
}
return charFormInfo;
}
///
/// ToDecompositionList - returns the decomposition character sequence for the given character
/// (if one exists)
///
public char[] ToDecompositionList ( char unicodeChar )
{
return GetDependentCharacterList(unicodeChar, _compositeChars);
}
///
/// GetInvalidVowelList - returns the decomposition character sequence for the given character
/// (if one exists)
///
public char[] GetMatraInvalidVowelList ( char unicodeChar )
{
return GetDependentCharacterList(unicodeChar, _scriptMatraInvalidVowels);
}
public IndicRepositioningClass HalantRepositioningClass
{
get { return ToRepositioningClass( (char)(_xorMask + HalantOffset) ); }
}
public IndicScriptIndices ScriptIx
{
get { return _scriptIx; }
}
public char FirstCharacterInScriptRange
{
get { return (char)_xorMask; }
}
public char[][] TamilAkhands
{
get { return _tamilAkhands; }
}
public char HalantCharacter
{
get { return (char)(_xorMask + HalantOffset); }
}
///
/// Critical - calls critical code
///
[SecurityCritical]
public bool GetFontCharacterFormsInfo (IndicFontClient fontClient)
{
if (!_isFormsDiscoveryDone)
{
byte[] fontCharacterForms = new byte[IndicCharClassifier.CharsPerIndicScript];
for (int i = 0; i < IndicCharClassifier.CharsPerIndicScript; ++i)
{
fontCharacterForms[i] = (byte)ToFormInfo((char)(_xorMask + i));
}
// see if font has anything to add. (If so, keep the new info)
if (fontClient.GetCharacterFormsInfo(fontCharacterForms))
{
_charFormsInfo = fontCharacterForms;
}
_isFormsDiscoveryDone = true;
}
return true;
}
private IndicScriptIndices GetScriptIx (ScriptTags scriptTag)
{
switch (scriptTag)
{
case ScriptTags.Devanagari:
return IndicScriptIndices.Devanagari;
case ScriptTags.Bengali:
return IndicScriptIndices.Bengali;
case ScriptTags.Gurmukhi:
return IndicScriptIndices.Gurmukhi;
case ScriptTags.Gujarati:
return IndicScriptIndices.Gujarati;
case ScriptTags.Oriya:
return IndicScriptIndices.Oriya;
case ScriptTags.Tamil:
return IndicScriptIndices.Tamil;
case ScriptTags.Telugu:
return IndicScriptIndices.Telugu;
case ScriptTags.Kannada:
return IndicScriptIndices.Kannada;
case ScriptTags.Malayalam:
return IndicScriptIndices.Malayalam;
case ScriptTags.Devanagari_v2:
return IndicScriptIndices.Devanagari;
case ScriptTags.Bengali_v2:
return IndicScriptIndices.Bengali;
case ScriptTags.Gurmukhi_v2:
return IndicScriptIndices.Gurmukhi;
case ScriptTags.Gujarati_v2:
return IndicScriptIndices.Gujarati;
case ScriptTags.Oriya_v2:
return IndicScriptIndices.Oriya;
case ScriptTags.Tamil_v2:
return IndicScriptIndices.Tamil;
case ScriptTags.Telugu_v2:
return IndicScriptIndices.Telugu;
case ScriptTags.Kannada_v2:
return IndicScriptIndices.Kannada;
case ScriptTags.Malayalam_v2:
return IndicScriptIndices.Malayalam;
default:
throw new NotSupportedException();
}
}
private DependentCharacterList[] GetCompositeCharactersList ()
{
switch (_scriptIx)
{
case IndicScriptIndices.Devanagari:
return null;
case IndicScriptIndices.Bengali:
return _bengaliSplitMatras;
case IndicScriptIndices.Gurmukhi:
return null;
case IndicScriptIndices.Gujarati:
return null;
case IndicScriptIndices.Oriya:
return _oriyaSplitMatras;
case IndicScriptIndices.Tamil:
return _tamilSplitMatras;
case IndicScriptIndices.Telugu:
return _teluguSplitMatras;
case IndicScriptIndices.Kannada:
return _kannadaSplitMatras;
case IndicScriptIndices.Malayalam:
return _malayalamSplitMatras;
default:
throw new NotSupportedException();
}
}
private DependentCharacterList[] GetInvalidVowelsList ()
{
switch (_scriptIx)
{
case IndicScriptIndices.Devanagari:
return _hindiMatraInvalidVowels;
case IndicScriptIndices.Bengali:
return _bengaliMatraInvalidVowels;
case IndicScriptIndices.Gurmukhi:
return _gurmukhiMatraInvalidVowels;
case IndicScriptIndices.Gujarati:
return _gujaratiMatraInvalidVowels;
case IndicScriptIndices.Oriya:
return _oriyaMatraInvalidVowels;
case IndicScriptIndices.Tamil:
return null;
case IndicScriptIndices.Telugu:
return _teluguMatraInvalidVowels;
case IndicScriptIndices.Kannada:
return _kannadaMatraInvalidVowels;
case IndicScriptIndices.Malayalam:
return _malayalamMatraInvalidVowels;
default:
throw new NotSupportedException();
}
}
// Bengali Script split matras
private static readonly DependentCharacterList[] _bengaliSplitMatras =
{
new DependentCharacterList( '\u09cb', new char[] {'\u09c7','\u09be'} ),
new DependentCharacterList( '\u09cc', new char[] {'\u09c7','\u09d7'} )
};
// Bengali invalid vowel matras
private static readonly DependentCharacterList[] _bengaliMatraInvalidVowels =
{
new DependentCharacterList( '\u09be', new char[] {'\u0985'} ),
new DependentCharacterList( '\u09c3', new char[] {'\u098b'} ),
new DependentCharacterList( '\u09e2', new char[] {'\u098c'} ),
};
// Gujarati invalid vowel matras
private static readonly DependentCharacterList[] _gujaratiMatraInvalidVowels =
{
new DependentCharacterList( '\u0abe', new char[] {'\u0a85'} ),
new DependentCharacterList( '\u0ac5', new char[] {'\u0a85','\u0a86'} ),
new DependentCharacterList( '\u0ac7', new char[] {'\u0a85','\u0a86'} ),
new DependentCharacterList( '\u0ac8', new char[] {'\u0a85','\u0a86'} ),
new DependentCharacterList( '\u0ac9', new char[] {'\u0a85'} ),
new DependentCharacterList( '\u0acb', new char[] {'\u0a85'} ),
new DependentCharacterList( '\u0acc', new char[] {'\u0a85'} ),
};
// Gurmukhi invalid vowel matras
private static readonly DependentCharacterList[] _gurmukhiMatraInvalidVowels =
{
new DependentCharacterList( '\u0a3e', new char[] {'\u0a05'} ),
new DependentCharacterList( '\u0a3f', new char[] {'\u0a72'} ),
new DependentCharacterList( '\u0a40', new char[] {'\u0a72'} ),
new DependentCharacterList( '\u0a41', new char[] {'\u0a73'} ),
new DependentCharacterList( '\u0a42', new char[] {'\u0a73'} ),
new DependentCharacterList( '\u0a47', new char[] {'\u0a72'} ),
new DependentCharacterList( '\u0a48', new char[] {'\u0a05'} ),
new DependentCharacterList( '\u0a4c', new char[] {'\u0a05'} )
};
// Hindi (Devanagari) invalid vowel matras
private static readonly DependentCharacterList[] _hindiMatraInvalidVowels =
{
new DependentCharacterList( '\u093e', new char[] {'\u0905'} ),
new DependentCharacterList( '\u0945', new char[] {'\u0906','\u090f'} ),
new DependentCharacterList( '\u0946', new char[] {'\u0905','\u0906','\u090f'} ),
new DependentCharacterList( '\u0947', new char[] {'\u0906','\u090f'} ),
new DependentCharacterList( '\u0948', new char[] {'\u0906'} ),
new DependentCharacterList( '\u0949', new char[] {'\u0905'} ),
new DependentCharacterList( '\u094a', new char[] {'\u0905'} ),
new DependentCharacterList( '\u094b', new char[] {'\u0905'} ),
new DependentCharacterList( '\u094c', new char[] {'\u0905'} )
};
// Kannada Script split matras
private static readonly DependentCharacterList[] _kannadaSplitMatras =
{
new DependentCharacterList('\u0cc0',new char[] {'\u0cbf','\u0cd5'}),
new DependentCharacterList('\u0cc7',new char[] {'\u0cc6','\u0cd5'}),
new DependentCharacterList('\u0cc8',new char[] {'\u0cc6','\u0cd6'}),
new DependentCharacterList('\u0cca',new char[] {'\u0cc6','\u0cc2'}),
new DependentCharacterList('\u0ccb',new char[] {'\u0cc6','\u0cc2','\u0cd5'}),
};
// Kannada invalid vowel matras
private static readonly DependentCharacterList[] _kannadaMatraInvalidVowels =
{
new DependentCharacterList( '\u0ccc', new char[] {'\u0c92'} )
};
// Malayalam Script split matras
private static readonly DependentCharacterList[] _malayalamSplitMatras =
{
new DependentCharacterList('\u0d4a', new char[] {'\u0d46','\u0d3e'}),
new DependentCharacterList('\u0d4b', new char[] {'\u0d47','\u0d3e'}),
new DependentCharacterList('\u0d4c', new char[] {'\u0d46','\u0d57'}),
};
// Malayalam invalid vowel matras
private static readonly DependentCharacterList[] _malayalamMatraInvalidVowels =
{
new DependentCharacterList( '\u0d3e', new char[] {'\u0d12'} ),
new DependentCharacterList( '\u0d46', new char[] {'\u0d0e'} ),
new DependentCharacterList( '\u0d57', new char[] {'\u0d07','\u0d09','\u0d12'} ),
};
// Oriya Script split matras
private static readonly DependentCharacterList[] _oriyaSplitMatras =
{
new DependentCharacterList('\u0b48', new char[] {'\u0b47','\u0b56'}),
new DependentCharacterList('\u0b4b', new char[] {'\u0b47','\u0b3e'}),
new DependentCharacterList('\u0b4c', new char[] {'\u0b47','\u0b57'}),
};
// Oriya invalid vowel matras
private static readonly DependentCharacterList[] _oriyaMatraInvalidVowels =
{
new DependentCharacterList( '\u0b3e', new char[] {'\u0b05'} )
};
private static readonly DependentCharacterList[] _tamilSplitMatras =
{
// Tamil Script split matras
new DependentCharacterList('\u0b94', new char[] {'\u0b92','\u0bd7'}),
new DependentCharacterList('\u0bca', new char[] {'\u0bc6','\u0bbe'}),
new DependentCharacterList('\u0bcb', new char[] {'\u0bc7','\u0bbe'}),
new DependentCharacterList('\u0bcc', new char[] {'\u0bc6','\u0bd7'}),
};
private static readonly char[][] _tamilAkhands =
{
// Tamil Script akhands
new char[] {'\u0b95','\u0bcd','\u0bb7'},
new char[] {'\u0bb6','\u0bcd','\u0bb0','\u0bc0'},
new char[] {'\u0bb8','\u0bcd','\u0bb0','\u0bc0'},
};
// Telugu Script split matras
private static readonly DependentCharacterList[] _teluguSplitMatras =
{
new DependentCharacterList('\u0c48', new char[] {'\u0c46','\u0c56'}),
};
// Telugu invalid vowel matras
private static readonly DependentCharacterList[] _teluguMatraInvalidVowels =
{
new DependentCharacterList( '\u0c3e', new char[] {'\u0c0b','\u0c0c'} ),
new DependentCharacterList( '\u0c4c', new char[] {'\u0c12'} ),
new DependentCharacterList( '\u0c55', new char[] {'\u0c12'} )
};
#region Classification Tables
private const byte RepositioningClassShift = (byte)IndicRepositioningClass.RepositioningClassShift;
private const byte UndefinedChar = (byte)IndicCharClass.Reserved;
private const byte Neutral = (byte)IndicCharClass.Neutral;
private const byte Consonant = (byte)IndicCharClass.Consonant;
private const byte NuktaConsonant = (byte)IndicCharClass.NuktaConsonant;
private const byte Ra = (byte)(IndicCharClass.Ra | IndicCharClass.IncludesPositioningInfo);
private const byte Vowel = (byte)IndicCharClass.Vowel;
private const byte Nukta = (byte)IndicCharClass.Nukta;
private const byte Halant = (byte)(IndicCharClass.Halant | IndicCharClass.IncludesPositioningInfo);
private const byte Vedic = (byte)(IndicCharClass.Vedic | IndicCharClass.IncludesPositioningInfo);
private const byte VowelSign = (byte)(IndicCharClass.VowelSign | IndicCharClass.IncludesPositioningInfo);
private const byte Matra = (byte)(IndicCharClass.Matra | IndicCharClass.IncludesPositioningInfo);
private const byte SplitMatra = (byte)IndicCharClass.Matra;
private const byte RepositionAtStart = ((byte)IndicRepositioningClass.AtStart << RepositioningClassShift);
private const byte RepositionBeforeMain = ((byte)IndicRepositioningClass.BeforeMain << RepositioningClassShift);
private const byte RepositionAfterMain = ((byte)IndicRepositioningClass.AfterMain << RepositioningClassShift);
private const byte RepositionBeforeSub = ((byte)IndicRepositioningClass.BeforeSub << RepositioningClassShift);
private const byte RepositionAfterSub = ((byte)IndicRepositioningClass.AfterSub << RepositioningClassShift);
private const byte RepositionBeforePost = ((byte)IndicRepositioningClass.BeforePost << RepositioningClassShift);
private const byte RepositionAfterPost = ((byte)IndicRepositioningClass.AfterPost << RepositioningClassShift);
private const byte RepositionBeforeEnd = ((byte)IndicRepositioningClass.BeforeEnd << RepositioningClassShift);
private static readonly byte[] _indicCharClasses =
{
// All the Devanagari Unicode chars (U+900 - U+97F) classified
//090 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
Vowel, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
//091 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//092 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, NuktaConsonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//093 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionBeforePost,
NuktaConsonant, Consonant, Consonant,
NuktaConsonant, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterSub,
Matra|RepositionAtStart,
//094 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Halant|RepositionAfterSub,
UndefinedChar, UndefinedChar,
//095 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, Vedic|RepositionBeforeEnd,
Vedic|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, UndefinedChar, UndefinedChar,
NuktaConsonant, NuktaConsonant, NuktaConsonant, NuktaConsonant,
NuktaConsonant, NuktaConsonant, NuktaConsonant, NuktaConsonant,
//096 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//097 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Consonant,
Consonant, Neutral, Consonant, Consonant,
// All the Bengali Unicode chars (U+980 - U+9FF) classified
//098 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionAfterPost,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, UndefinedChar, Vowel,
//099 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//09A 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//09B 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterSub,
UndefinedChar, Consonant, UndefinedChar,
UndefinedChar, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterPost,
Matra|RepositionAtStart,
//09C 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, Matra|RepositionAtStart,
Matra|RepositionAtStart,
UndefinedChar, UndefinedChar, SplitMatra,
SplitMatra, Halant|RepositionAfterPost,
Consonant, UndefinedChar,
//09D 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
NuktaConsonant, NuktaConsonant, UndefinedChar, NuktaConsonant,
//09E 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//09F 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterSub,
Consonant, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Gurmukhi Unicode chars (U+a00 - U+a7F) classified
//0A0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionAfterPost,
VowelSign|RepositionAfterPost,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Vowel,
//0A1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0A2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0A3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionBeforePost,
UndefinedChar, Consonant, NuktaConsonant,
UndefinedChar, Consonant, NuktaConsonant, UndefinedChar,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAtStart,
//0A4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Halant|RepositionAfterSub,
UndefinedChar, UndefinedChar,
//0A5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, NuktaConsonant, NuktaConsonant, NuktaConsonant,
Consonant, UndefinedChar, NuktaConsonant, UndefinedChar,
//0A6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0A7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
VowelSign|RepositionAfterPost,
VowelSign|RepositionAfterPost,
Vowel, Vowel,
Neutral, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Gujarati Unicode chars (U+a80 - U+afF) classified
//0A8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, Vowel, UndefinedChar, Vowel,
//0A9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0AA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0AB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionBeforePost,
UndefinedChar, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterSub,
Matra|RepositionAtStart,
//0AC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Halant|RepositionAfterSub,
UndefinedChar, UndefinedChar,
//0AD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0AE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0AF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, Neutral, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Oriya Unicode chars (U+B00 - U+B7F) classified
//0B0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeSub,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, UndefinedChar, Vowel,
//0B1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, UndefinedChar, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0B2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0B3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterMain,
UndefinedChar, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionAfterPost,
Matra|RepositionAfterMain,
//0B4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAtStart,
SplitMatra,
UndefinedChar, UndefinedChar, SplitMatra,
SplitMatra,
Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0B5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionAfterMain,
Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
NuktaConsonant, NuktaConsonant, UndefinedChar, Consonant,
//0B6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0B7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Tamil Unicode chars (U+B80 - U+BFF) classified
//0B8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, VowelSign|RepositionBeforeEnd,
Neutral, //Vowel,
UndefinedChar, Vowel, Consonant, Vowel,
Vowel, Vowel, Vowel, UndefinedChar,
UndefinedChar, UndefinedChar, Vowel, Vowel,
//0B9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, Consonant, Consonant, UndefinedChar,
Consonant, UndefinedChar, Consonant, Consonant,
//0BA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, Consonant,
Consonant, UndefinedChar, UndefinedChar, UndefinedChar,
Consonant, Consonant, Consonant, UndefinedChar,
UndefinedChar, UndefinedChar, Consonant, Consonant,
//0BB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterPost,
Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
//0BC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterSub,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionBeforeMain,
Matra|RepositionBeforeMain,
Matra|RepositionBeforeMain,
UndefinedChar, SplitMatra, SplitMatra,
SplitMatra, Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0BD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0BE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0BF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Telugu Unicode chars (U+c00 - U+c7F) classified
//0C0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, Vowel, Vowel,
//0C1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0C2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0C3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterPost,
Consonant, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
//0C4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
SplitMatra, UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0C5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0C6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0C7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Kannada Unicode chars (U+C80 - U+CFF) classified
//0C8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, Vowel, Vowel,
//0C9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0CA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0CB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterPost,
Consonant, Consonant, Consonant,
UndefinedChar, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
Nukta, Neutral, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
//0CC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SplitMatra, Matra|RepositionBeforeSub,
Matra|RepositionBeforeSub,
Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar, Matra|RepositionBeforeSub,
SplitMatra,
SplitMatra, UndefinedChar, SplitMatra, SplitMatra,
Matra|RepositionBeforeSub,
Halant|RepositionBeforeSub,
UndefinedChar, UndefinedChar,
//0CD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, Matra|RepositionAfterSub,
Matra|RepositionAfterSub,
UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Consonant, UndefinedChar,
//0CE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0CF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
// All the Malayalam Unicode chars (U+d00 - U+d7F) classified
//0D0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, VowelSign|RepositionBeforeEnd,
VowelSign|RepositionBeforeEnd,
UndefinedChar, Vowel, Vowel, Vowel,
Vowel, Vowel, Vowel, Vowel,
Vowel, UndefinedChar, Vowel, Vowel,
//0D1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, UndefinedChar, Vowel, Vowel,
Vowel, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0D2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, UndefinedChar, Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
//0D3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Ra|RepositionAfterMain,
Ra|RepositionAfterMain,
Consonant, Consonant,
Consonant, Consonant, Consonant, Consonant,
Consonant, Consonant, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
//0D4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, Matra|RepositionAtStart,
Matra|RepositionAtStart,
Matra|RepositionAtStart,
UndefinedChar, SplitMatra, SplitMatra,
SplitMatra, Halant|RepositionAfterPost,
UndefinedChar, UndefinedChar,
//0D5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, Matra|RepositionAfterPost,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
//0D6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Vowel, Vowel, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
Neutral, Neutral, Neutral, Neutral,
//0D7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
UndefinedChar, UndefinedChar, UndefinedChar, UndefinedChar,
};
#endregion // end of Classification Tables
#region Character Form Tables
private const byte PreBaseComponent = (byte)IndicFormFlags.PreBaseMatra;
private const byte AboveBaseComponent = (byte)IndicFormFlags.AboveBaseMatra;
private const byte BelowBaseComponent = (byte)IndicFormFlags.BelowBaseMatra;
private const byte PostBaseComponent = (byte)IndicFormFlags.PostBaseMatra;
private const byte PreBaseMatra = (byte)IndicFormFlags.PreBaseMatra;
private const byte AboveBaseMatra = (byte)IndicFormFlags.AboveBaseMatra;
private const byte BelowBaseMatra = (byte)IndicFormFlags.BelowBaseMatra;
private const byte PostBaseMatra = (byte)IndicFormFlags.PostBaseMatra;
private const byte Reph = (byte)IndicFormFlags.RephForm;
private const byte Vattu = (byte)IndicFormFlags.VattuForm;
private const byte InitialChar = (byte)IndicFormFlags.InitForm;
private const byte PostBaseReph = (byte)IndicFormFlags.PostbaseReph;
private const byte DravidianBase = (byte)IndicFormFlags.DravidianBase;
private const byte SubForm = (byte)IndicFormFlags.SubForm;
private const byte PostForm = (byte)IndicFormFlags.PostForm;
private static readonly byte[] _indicCharFormsInfo =
{
// All the Devanagari Unicode chars (U+900 - U+97F) forms info
//090 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//091 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//092 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//093 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | Vattu, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//094 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
BelowBaseMatra, AboveBaseMatra, AboveBaseMatra, AboveBaseMatra,
AboveBaseMatra, PostBaseMatra, PostBaseMatra, PostBaseMatra,
PostBaseMatra, 0, 0, 0,
//095 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//096 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, BelowBaseMatra, BelowBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//097 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Bengali Unicode chars (U+980 - U+9FF) consonant/matra forms (the consonant
// forms will be updated from the font)
//098 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//099 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//09A 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
SubForm, 0, 0, PostForm,
//09B 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
InitialChar | Reph | SubForm,
0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//09C 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
BelowBaseMatra, 0, 0, PreBaseMatra|InitialChar,
PreBaseMatra|InitialChar,
0, 0, InitialChar|PreBaseComponent|PostBaseComponent,
InitialChar|PreBaseComponent|PostBaseComponent,
0, 0, 0,
//09D 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, PostBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
//09E 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, BelowBaseMatra,
BelowBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//09F 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | SubForm, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Gurmukhi Unicode chars (U+a00 - U+a7F) classified
//0A0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, PostForm,
//0A3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, 0, 0, 0,
0, SubForm, 0, 0,
0, SubForm, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//0A4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, 0,
0, 0, 0, AboveBaseMatra,
AboveBaseMatra, 0, 0, AboveBaseMatra,
AboveBaseMatra, 0, 0, 0,
//0A5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Gujarati Unicode chars (U+a80 - U+afF) classified
//0A8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0A9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | Vattu, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PreBaseMatra,
//0AC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
BelowBaseMatra, AboveBaseMatra, 0, AboveBaseMatra,
AboveBaseMatra, PostBaseMatra, 0, PostBaseMatra,
PostBaseMatra, 0, 0, 0,
//0AD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, BelowBaseMatra, BelowBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0AF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Oriya Unicode chars (U+B00 - U+B7F) classified
//0B0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
SubForm, 0, 0, 0,
SubForm, 0, 0, 0,
SubForm, SubForm, SubForm, PostForm,
//0B3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | SubForm, 0, SubForm, SubForm,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, AboveBaseMatra,
//0B4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, BelowBaseMatra, BelowBaseMatra, BelowBaseMatra,
0, 0, 0, PreBaseMatra,
PreBaseComponent|AboveBaseComponent,
0, 0, PreBaseComponent|PostBaseComponent,
PreBaseComponent|AboveBaseComponent|PostBaseComponent,
0, 0, 0,
//0B5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, AboveBaseMatra, PostBaseMatra|AboveBaseComponent,
0, 0, 0, 0,
0, 0, 0, PostForm,
//0B6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Tamil Unicode chars (U+B80 - U+BFF) classified
//0B8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0B9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0BA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0BB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PostBaseMatra,
//0BC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
AboveBaseMatra, BelowBaseMatra, BelowBaseMatra, 0,
0, 0, PreBaseMatra, PreBaseMatra,
PreBaseMatra, 0, PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
0, 0, 0,
//0BD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, PostBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
//0BE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0BF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Telugu Unicode chars (U+c00 - U+c7F) classified
//0C0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0C2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, 0, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0C3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, SubForm, SubForm, SubForm,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, 0, 0,
0, 0, AboveBaseMatra, AboveBaseMatra,
//0C4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
AboveBaseMatra, PostBaseMatra, PostBaseMatra, PostBaseMatra ,
PostBaseMatra, 0, AboveBaseMatra, AboveBaseMatra,
AboveBaseComponent|BelowBaseComponent,
0, AboveBaseMatra, AboveBaseMatra,
AboveBaseMatra, 0, 0, 0,
//0C5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, AboveBaseMatra, BelowBaseMatra, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Kannada Unicode chars (U+C80 - U+CFF) classified
//0C8 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0C9 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0CA 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
SubForm, SubForm, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
SubForm, 0, SubForm, SubForm,
SubForm, SubForm, SubForm, SubForm,
//0CB 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
Reph | SubForm, SubForm, SubForm, SubForm,
0, SubForm, SubForm, SubForm,
SubForm, SubForm, 0, 0,
0, 0, PostBaseMatra, AboveBaseMatra,
//0CC 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
AboveBaseComponent|PostBaseComponent,
PostBaseMatra, PostBaseMatra, PostBaseMatra,
PostBaseMatra, 0, AboveBaseMatra, AboveBaseComponent|PostBaseComponent,
AboveBaseComponent|PostBaseComponent,
0, AboveBaseComponent|PostBaseComponent,
AboveBaseComponent|PostBaseComponent,
AboveBaseMatra, 0, 0, 0,
//0CD 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, PostBaseMatra, PostBaseMatra, 0,
0, 0, 0, 0,
0, 0, SubForm, 0,
//0CE 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0CF 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
// All the Malayalam Unicode chars (U+d00 - U+d7F) classified
//0D0 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0D1 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0D2 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, PostForm,
//0D3 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostForm|PostBaseReph,
DravidianBase, SubForm, DravidianBase,
DravidianBase, PostForm, 0, 0,
0, 0, 0, 0,
0, 0, PostBaseMatra, PostBaseMatra,
//0D4 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
PostBaseMatra, PostBaseMatra, PostBaseMatra, PostBaseMatra,
0, 0, PreBaseMatra, PreBaseMatra,
PreBaseMatra, 0, PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
PreBaseComponent|PostBaseComponent,
0, 0, 0,
//0D5 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, PostBaseMatra,
0, 0, 0, 0,
0, 0, 0, 0,
//0D6 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
//0D7 0, 4, 8, C 1, 5, 9, D 2, 6, A, E 3, 7, B, F
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
};
#endregion // Character Form Tables
}
}
// File provided for Reference Use Only by Microsoft Corporation (c) 2007.
// Copyright (c) Microsoft Corporation. All rights reserved.
Link Menu
This book is available now!
Buy at Amazon US or
Buy at Amazon UK
- UIElement3D.cs
- TextMarkerSource.cs
- FormViewUpdateEventArgs.cs
- SqlConnectionString.cs
- PasswordBoxAutomationPeer.cs
- SqlFacetAttribute.cs
- RepeaterCommandEventArgs.cs
- ClientFormsAuthenticationMembershipProvider.cs
- SoundPlayer.cs
- ObjectDataSourceFilteringEventArgs.cs
- RtfFormatStack.cs
- DecoderExceptionFallback.cs
- ListManagerBindingsCollection.cs
- ArithmeticException.cs
- CompareInfo.cs
- Comparer.cs
- XmlAttributeProperties.cs
- RuntimeWrappedException.cs
- EnlistmentTraceIdentifier.cs
- ExpressionBuilderCollection.cs
- ApplyImportsAction.cs
- DuplicateDetector.cs
- TimeSpanParse.cs
- Point3DConverter.cs
- WpfXamlLoader.cs
- DataGridViewRowsAddedEventArgs.cs
- ChtmlTextWriter.cs
- ToolboxItemFilterAttribute.cs
- LoginCancelEventArgs.cs
- XslVisitor.cs
- CompatibleComparer.cs
- WebPartDescriptionCollection.cs
- SystemIcons.cs
- UnsafeNativeMethodsCLR.cs
- InlineObject.cs
- Menu.cs
- TextDpi.cs
- DefaultMergeHelper.cs
- OleDbParameterCollection.cs
- SizeValueSerializer.cs
- TrackingParameters.cs
- TransactionInformation.cs
- TextSelection.cs
- TableLayoutPanel.cs
- CollectionChangedEventManager.cs
- TraceInternal.cs
- AttachInfo.cs
- IdentityManager.cs
- Point3D.cs
- AndCondition.cs
- sqlpipe.cs
- GraphicsPath.cs
- DataGridColumnCollectionEditor.cs
- SessionStateItemCollection.cs
- ServiceModelConfigurationSectionCollection.cs
- ProcessHostFactoryHelper.cs
- LocalizabilityAttribute.cs
- Rotation3DAnimationBase.cs
- LinqDataSource.cs
- SuppressMessageAttribute.cs
- DbFunctionCommandTree.cs
- AutoScrollExpandMessageFilter.cs
- SafeNativeMethods.cs
- ExpressionBuilder.cs
- TypeDescriptionProviderAttribute.cs
- DataControlPagerLinkButton.cs
- SessionParameter.cs
- ComplexTypeEmitter.cs
- Subset.cs
- TextBox.cs
- UTF7Encoding.cs
- ColorContext.cs
- XmlQueryStaticData.cs
- HostedTransportConfigurationManager.cs
- MemberProjectedSlot.cs
- EmitterCache.cs
- _WebProxyDataBuilder.cs
- PersonalizationState.cs
- ComponentCollection.cs
- BufferBuilder.cs
- xml.cs
- AutomationPatternInfo.cs
- FileDataSourceCache.cs
- TreeView.cs
- WorkflowDispatchContext.cs
- KeyValuePair.cs
- TreeViewDesigner.cs
- HttpDictionary.cs
- Stroke2.cs
- IdentitySection.cs
- TextCompositionManager.cs
- MaxMessageSizeStream.cs
- GroupItem.cs
- MatchSingleFxEngineOpcode.cs
- WebPartDisplayMode.cs
- UnconditionalPolicy.cs
- XmlEncodedRawTextWriter.cs
- BinaryFormatterWriter.cs
- TreeBuilder.cs
- Module.cs