KhmerReordering.h revision 51cfa1a9a96cad34675a6415fe86dfdf3f525bb6
1/* 2 * (C) Copyright IBM Corp. 1998-2005 - All Rights Reserved 3 * 4 * This file is a modification of the ICU file IndicReordering.h 5 * by Jens Herden and Javier Sola for Khmer language 6 * 7 */ 8 9#ifndef __KHMERREORDERING_H 10#define __KHMERREORDERING_H 11 12/** 13 * \file 14 * \internal 15 */ 16 17#include "LETypes.h" 18#include "OpenTypeTables.h" 19 20U_NAMESPACE_BEGIN 21 22class LEGlyphStorage; 23 24// Vocabulary 25// Base -> A consonant or an independent vowel in its full (not subscript) form. It is the 26// center of the syllable, it can be souranded by coeng (subscript) consonants, vowels, 27// split vowels, signs... but there is only one base in a syllable, it has to be coded as 28// the first character of the syllable. 29// split vowel --> vowel that has two parts placed separately (e.g. Before and after the consonant). 30// Khmer language has five of them. Khmer split vowels either have one part before the 31// base and one after the base or they have a part before the base and a part above the base. 32// The first part of all Khmer split vowels is the same character, identical to 33// the glyph of Khmer dependent vowel SRA EI 34// coeng --> modifier used in Khmer to construct coeng (subscript) consonants 35// Differently than indian languages, the coeng modifies the consonant that follows it, 36// not the one preceding it Each consonant has two forms, the base form and the subscript form 37// the base form is the normal one (using the consonants code-point), the subscript form is 38// displayed when the combination coeng + consonant is encountered. 39// Consonant of type 1 -> A consonant which has subscript for that only occupies space under a base consonant 40// Consonant of type 2.-> Its subscript form occupies space under and before the base (only one, RO) 41// Consonant of Type 3 -> Its subscript form occupies space under and after the base (KHO, CHHO, THHO, BA, YO, SA) 42// Consonant shifter -> Khmer has to series of consonants. The same dependent vowel has different sounds 43// if it is attached to a consonant of the first series or a consonant of the second series 44// Most consonants have an equivalent in the other series, but some of theme exist only in 45// one series (for example SA). If we want to use the consonant SA with a vowel sound that 46// can only be done with a vowel sound that corresponds to a vowel accompanying a consonant 47// of the other series, then we need to use a consonant shifter: TRIISAP or MUSIKATOAN 48// x17C9 y x17CA. TRIISAP changes a first series consonant to second series sound and 49// MUSIKATOAN a second series consonant to have a first series vowel sound. 50// Consonant shifter are both normally supercript marks, but, when they are followed by a 51// superscript, they change shape and take the form of subscript dependent vowel SRA U. 52// If they are in the same syllable as a coeng consonant, Unicode 3.0 says that they 53// should be typed before the coeng. Unicode 4.0 breaks the standard and says that it should 54// be placed after the coeng consonant. 55// Dependent vowel -> In khmer dependent vowels can be placed above, below, before or after the base 56// Each vowel has its own position. Only one vowel per syllable is allowed. 57// Signs -> Khmer has above signs and post signs. Only one above sign and/or one post sign are 58// Allowed in a syllable. 59// 60// 61 62struct KhmerClassTable // This list must include all types of components that can be used inside a syllable 63{ 64 enum CharClassValues // order is important here! This order must be the same that is found in each horizontal 65 // line in the statetable for Khmer (file KhmerReordering.cpp). 66 { 67 CC_RESERVED = 0, 68 CC_CONSONANT = 1, // consonant of type 1 or independent vowel 69 CC_CONSONANT2 = 2, // Consonant of type 2 70 CC_CONSONANT3 = 3, // Consonant of type 3 71 CC_ZERO_WIDTH_NJ_MARK = 4, // Zero Width non joiner character (0x200C) 72 CC_CONSONANT_SHIFTER = 5, 73 CC_ROBAT = 6, // Khmer special diacritic accent -treated differently in state table 74 CC_COENG = 7, // Subscript consonant combining character 75 CC_DEPENDENT_VOWEL = 8, 76 CC_SIGN_ABOVE = 9, 77 CC_SIGN_AFTER = 10, 78 CC_ZERO_WIDTH_J_MARK = 11, // Zero width joiner character 79 CC_COUNT = 12 // This is the number of character classes 80 }; 81 82 enum CharClassFlags 83 { 84 CF_CLASS_MASK = 0x0000FFFF, 85 86 CF_CONSONANT = 0x01000000, // flag to speed up comparing 87 CF_SPLIT_VOWEL = 0x02000000, // flag for a split vowel -> the first part is added in front of the syllable 88 CF_DOTTED_CIRCLE = 0x04000000, // add a dotted circle if a character with this flag is the first in a syllable 89 CF_COENG = 0x08000000, // flag to speed up comparing 90 CF_SHIFTER = 0x10000000, // flag to speed up comparing 91 CF_ABOVE_VOWEL = 0x20000000, // flag to speed up comparing 92 93 // position flags 94 CF_POS_BEFORE = 0x00080000, 95 CF_POS_BELOW = 0x00040000, 96 CF_POS_ABOVE = 0x00020000, 97 CF_POS_AFTER = 0x00010000, 98 CF_POS_MASK = 0x000f0000 99 }; 100 101 typedef le_uint32 CharClass; 102 103 typedef le_int32 ScriptFlags; 104 105 LEUnicode firstChar; // for Khmer this will become x1780 106 LEUnicode lastChar; // and this x17DF 107 const CharClass *classTable; 108 109 CharClass getCharClass(LEUnicode ch) const; 110 111 static const KhmerClassTable *getKhmerClassTable(); 112}; 113 114 115class KhmerReordering /* not : public UObject because all methods are static */ { 116public: 117 static le_int32 reorder(const LEUnicode *theChars, le_int32 charCount, le_int32 scriptCode, 118 LEUnicode *outChars, LEGlyphStorage &glyphStorage); 119 120 static const FeatureMap *getFeatureMap(le_int32 &count); 121 122private: 123 // do not instantiate 124 KhmerReordering(); 125 126 static le_int32 findSyllable(const KhmerClassTable *classTable, const LEUnicode *chars, le_int32 prev, le_int32 charCount); 127 128}; 129 130 131U_NAMESPACE_END 132#endif 133