[apple/icu.git] / icuSources / layout / IndicClassTables.cpp

/*
 *
 * (C) Copyright IBM Corp. 1998-2006 - All Rights Reserved
 *
 */

#include "LETypes.h"
#include "LEScripts.h"
#include "OpenTypeTables.h"
#include "OpenTypeUtilities.h"
#include "IndicReordering.h"

U_NAMESPACE_BEGIN

// Split matra table indices
#define _x1  (1 << CF_INDEX_SHIFT)
#define _x2  (2 << CF_INDEX_SHIFT)
#define _x3  (3 << CF_INDEX_SHIFT)
#define _x4  (4 << CF_INDEX_SHIFT)
#define _x5  (5 << CF_INDEX_SHIFT)
#define _x6  (6 << CF_INDEX_SHIFT)
#define _x7  (7 << CF_INDEX_SHIFT)
#define _x8  (8 << CF_INDEX_SHIFT)
#define _x9  (9 << CF_INDEX_SHIFT)

// simple classes
#define _xx  (CC_RESERVED)
#define _ma  (CC_VOWEL_MODIFIER | CF_POS_ABOVE)
#define _mp  (CC_VOWEL_MODIFIER | CF_POS_AFTER)
#define _sa  (CC_STRESS_MARK | CF_POS_ABOVE)
#define _sb  (CC_STRESS_MARK | CF_POS_BELOW)
#define _iv  (CC_INDEPENDENT_VOWEL)
#define _i2  (CC_INDEPENDENT_VOWEL_2)
#define _i3  (CC_INDEPENDENT_VOWEL_3)
#define _ct  (CC_CONSONANT | CF_CONSONANT)
#define _cn  (CC_CONSONANT_WITH_NUKTA | CF_CONSONANT)
#define _nu  (CC_NUKTA)
#define _dv  (CC_DEPENDENT_VOWEL)
#define _dl  (_dv | CF_POS_BEFORE)
#define _db  (_dv | CF_POS_BELOW)
#define _da  (_dv | CF_POS_ABOVE)
#define _dr  (_dv | CF_POS_AFTER)
#define _lm  (_dv | CF_LENGTH_MARK)
#define _l1  (CC_SPLIT_VOWEL_PIECE_1 | CF_POS_BEFORE)
#define _a1  (CC_SPLIT_VOWEL_PIECE_1 | CF_POS_ABOVE)
#define _b2  (CC_SPLIT_VOWEL_PIECE_2 | CF_POS_BELOW)
#define _r2  (CC_SPLIT_VOWEL_PIECE_2 | CF_POS_AFTER)
#define _m2  (CC_SPLIT_VOWEL_PIECE_2 | CF_LENGTH_MARK)
#define _m3  (CC_SPLIT_VOWEL_PIECE_3 | CF_LENGTH_MARK)
#define _vr  (CC_VIRAMA)

// split matras
#define _s1  (_dv | _x1)
#define _s2  (_dv | _x2)
#define _s3  (_dv | _x3)
#define _s4  (_dv | _x4)
#define _s5  (_dv | _x5)
#define _s6  (_dv | _x6)
#define _s7  (_dv | _x7)
#define _s8  (_dv | _x8)
#define _s9  (_dv | _x9)

// consonants with special forms
// NOTE: this assumes that no consonants with nukta have
// special forms... (Bengali RA?)
#define _bb  (_ct | CF_BELOW_BASE)
#define _pb  (_ct | CF_POST_BASE)
#define _vt  (_bb | CF_VATTU)
#define _rv  (_vt | CF_REPH)
#define _rp  (_pb | CF_REPH)
#define _rb  (_bb | CF_REPH)

//
// Character class tables
//
static const IndicClassTable::CharClass devaCharClasses[] =
{
    _xx, _ma, _ma, _mp, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0900 - 090F
    _iv, _iv, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0910 - 091F
    _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _ct, // 0920 - 092F
    _rv, _cn, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0930 - 093F
    _dr, _db, _db, _db, _db, _da, _da, _da, _da, _dr, _dr, _dr, _dr, _vr, _xx, _xx, // 0940 - 094F
    _xx, _sa, _sb, _sa, _sa, _xx, _xx, _xx, _cn, _cn, _cn, _cn, _cn, _cn, _cn, _cn, // 0950 - 095F
    _iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0960 - 096F
    _xx                                                                             // 0970
};

static const IndicClassTable::CharClass bengCharClasses[] =
{
    _xx, _ma, _mp, _mp, _xx, _i2, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _i2, // 0980 - 098F
    _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0990 - 099F
    _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _bb, _ct, _ct, _pb, // 09A0 - 09AF
    _rv, _xx, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _dl, // 09B0 - 09BF
    _dr, _db, _db, _db, _db, _xx, _xx, _l1, _dl, _xx, _xx, _s1, _s2, _vr, _xx, _xx, // 09C0 - 09CF
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 09D0 - 09DF
    _iv, _iv, _dv, _dv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 09E0 - 09EF
    _ct, _ct, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx                           // 09F0 - 09FA
};

static const IndicClassTable::CharClass punjCharClasses[] =
{
    _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _xx, _iv, // 0A00 - 0A0F
    _iv, _xx, _xx, _i3, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A10 - 0A1F
    _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _bb, // 0A20 - 0A2F
    _vt, _xx, _ct, _cn, _xx, _bb, _cn, _xx, _ct, _bb, _xx, _xx, _nu, _xx, _dr, _dl, // 0A30 - 0A3F
    _dr, _b2, _db, _xx, _xx, _xx, _xx, _da, _da, _xx, _xx, _a1, _da, _vr, _xx, _xx, // 0A40 - 0A4F
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _cn, _cn, _cn, _ct, _xx, _cn, _xx, // 0A50 - 0A5F
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0A60 - 0A6F
    _ma, _ma, _xx, _xx, _xx                                                         // 0A70 - 0A74
};

static const IndicClassTable::CharClass gujrCharClasses[] =
{
    _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _xx, _iv, // 0A80 - 0A8F
    _iv, _iv, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A90 - 0A9F
    _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0AA0 - 0AAF
    _rv, _xx, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0AB0 - 0ABF
    _dr, _db, _db, _db, _db, _da, _xx, _da, _da, _dr, _xx, _dr, _dr, _vr, _xx, _xx, // 0AC0 - 0ACF
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0AD0 - 0ADF
    _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0AE0 - 0AEF
};

static const IndicClassTable::CharClass oryaCharClasses[] =
{
    _xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, // 0B00 - 0B0F
    _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0B10 - 0B1F
    _ct, _ct, _ct, _ct, _bb, _ct, _ct, _ct, _bb, _xx, _ct, _ct, _bb, _bb, _bb, _pb, // 0B20 - 0B2F
    _rb, _xx, _bb, _bb, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _da, // 0B30 - 0B3F
    _dr, _db, _db, _db, _xx, _xx, _xx, _l1, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, // 0B40 - 0B4F
    _xx, _xx, _xx, _xx, _xx, _xx, _m2, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 0B50 - 0B5F
    _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0B60 - 0B6F
    _xx, _ct                                                                        // 0B70 - 0B71
};

static const IndicClassTable::CharClass tamlCharClasses[] =
{
    _xx, _xx, _ma, _xx, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _iv, _iv, // 0B80 - 0B8F
    _iv, _xx, _iv, _iv, _iv, _ct, _xx, _xx, _xx, _ct, _ct, _xx, _ct, _xx, _ct, _ct, // 0B90 - 0B9F
    _xx, _xx, _xx, _ct, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _xx, _xx, _xx, _ct, _ct, // 0BA0 - 0BAF
    _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0BB0 - 0BBF
    _da, _dr, _dr, _xx, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0BC0 - 0BCF
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BD0 - 0BDF
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BE0 - 0BEF
    _xx, _xx, _xx                                                                   // 0BF0 - 0BF2
};

// FIXME: Should some of the bb's be pb's? (KA, NA, MA, YA, VA, etc. (approx 13))
static const IndicClassTable::CharClass teluCharClasses[] =
{
    _xx, _mp, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0C00 - 0C0F
    _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C10 - 0C1F
    _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0C20 - 0C2F
    _bb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _da, _da, // 0C30 - 0C3F
    _da, _dr, _dr, _dr, _dr, _xx, _a1, _da, _s1, _xx, _da, _da, _da, _vr, _xx, _xx, // 0C40 - 0C4F
    _xx, _xx, _xx, _xx, _xx, _da, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0C50 - 0C5F
    _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0C60 - 0C6F
};

// U+CC3 and U+CC4 are _lm here not _dr since the Kannada rendering
// rules want them below and to the right of the entire cluster
//
// There's some information about this in:
//
//  http://brahmi.sourceforge.net/docs/KannadaComputing.html
static const IndicClassTable::CharClass kndaCharClasses[] =
{
    _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, // 0C80 - 0C8F
    _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C90 - 0C9F
    _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0CA0 - 0CAF
    _rb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _dr, _da, // 0CB0 - 0CBF
    _s1, _dr, _r2, _lm, _lm, _xx, _a1, _s2, _s3, _xx, _s4, _s5, _da, _vr, _xx, _xx, // 0CC0 - 0CCF
    _xx, _xx, _xx, _xx, _xx, _m3, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _ct, _xx, // 0CD0 - 0CDF
    _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0CE0 - 0CEF
};

// FIXME: this is correct for old-style Malayalam (MAL) but not for reformed Malayalam (MLR)
// FIXME: should there be a REPH for old-style Malayalam?
static const IndicClassTable::CharClass mlymCharClasses[] =
{
    _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0D00 - 0D0F
    _iv, _xx, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _bb, // 0D10 - 0D1F
    _ct, _ct, _ct, _bb, _ct, _bb, _bb, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _pb, // 0D20 - 0D2F
    _pb, _cn, _bb, _ct, _ct, _pb, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0D30 - 0D3F
    _dr, _dr, _dr, _dr, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0D40 - 0D4F
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0D50 - 0D5F
    _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0D60 - 0D6F
};
 
static const IndicClassTable::CharClass sinhCharClasses[] =
{
    _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0D80 - 0D8F
    _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0D90 - 0D9F
    _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0DA0 - 0DAF
    _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _xx, _xx, // 0DB0 - 0DBF
    _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _vr, _xx, _xx, _xx, _xx, _dr, // 0DC0 - 0DCF
    _dr, _dr, _da, _da, _db, _xx, _db, _xx, _dr, _dl, _s1, _dl, _s2, _s3, _s4, _dr, // 0DD0 - 0DDF
    _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0DE0 - 0DEF
    _xx, _xx, _dr, _dr, _xx                                                         // 0DF0 - 0DF4
};

//
// Split matra tables
//
static const SplitMatra bengSplitTable[] = {{0x09C7, 0x09BE}, {0x09C7, 0x09D7}};

static const SplitMatra oryaSplitTable[] = {{0x0B47, 0x0B56}, {0x0B47, 0x0B3E}, {0x0B47, 0x0B57}};

static const SplitMatra tamlSplitTable[] = {{0x0BC6, 0x0BBE}, {0x0BC7, 0x0BBE}, {0x0BC6, 0x0BD7}};

static const SplitMatra teluSplitTable[] = {{0x0C46, 0x0C56}};

static const SplitMatra kndaSplitTable[] = {{0x0CBF, 0x0CD5}, {0x0CC6, 0x0CD5}, {0x0CC6, 0x0CD6}, {0x0CC6, 0x0CC2},
                                            {0x0CC6, 0x0CC2, 0x0CD5}};

static const SplitMatra mlymSplitTable[] = {{0x0D46, 0x0D3E}, {0x0D47, 0x0D3E}, {0x0D46, 0x0D57}};

 
static const SplitMatra sinhSplitTable[] = {{0x0DD9, 0x0DCA}, {0x0DD9, 0x0DCF}, {0x0DD9, 0x0DCF,0x0DCA},
                                            {0x0DD9, 0x0DDF}};
//
// Script Flags
//

// FIXME: post 'GSUB' reordering of MATRA_PRE's for Malayalam and Tamil
// FIXME: reformed Malayalam needs to reorder VATTU to before base glyph...
// FIXME: eyelash RA only for Devanagari??
#define DEVA_SCRIPT_FLAGS (SF_EYELASH_RA | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
#define BENG_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
#define PUNJ_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
#define GUJR_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
#define ORYA_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
#define TAML_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
#define TELU_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3)
#define KNDA_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3)
#define MLYM_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
#define SINH_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT)

//
// Indic Class Tables
//
static const IndicClassTable devaClassTable = {0x0900, 0x0970, 2, DEVA_SCRIPT_FLAGS, devaCharClasses, NULL};

static const IndicClassTable bengClassTable = {0x0980, 0x09FA, 3, BENG_SCRIPT_FLAGS, bengCharClasses, bengSplitTable};

static const IndicClassTable punjClassTable = {0x0A00, 0x0A74, 2, PUNJ_SCRIPT_FLAGS, punjCharClasses, NULL};

static const IndicClassTable gujrClassTable = {0x0A80, 0x0AEF, 2, GUJR_SCRIPT_FLAGS, gujrCharClasses, NULL};

static const IndicClassTable oryaClassTable = {0x0B00, 0x0B71, 3, ORYA_SCRIPT_FLAGS, oryaCharClasses, oryaSplitTable};

static const IndicClassTable tamlClassTable = {0x0B80, 0x0BF2, 3, TAML_SCRIPT_FLAGS, tamlCharClasses, tamlSplitTable};

static const IndicClassTable teluClassTable = {0x0C00, 0x0C6F, 3, TELU_SCRIPT_FLAGS, teluCharClasses, teluSplitTable};

static const IndicClassTable kndaClassTable = {0x0C80, 0x0CEF, 4, KNDA_SCRIPT_FLAGS, kndaCharClasses, kndaSplitTable};

static const IndicClassTable mlymClassTable = {0x0D00, 0x0D6F, 3, MLYM_SCRIPT_FLAGS, mlymCharClasses, mlymSplitTable};

static const IndicClassTable sinhClassTable = {0x0D80, 0x0DF4, 3, SINH_SCRIPT_FLAGS, sinhCharClasses, sinhSplitTable};

//
// IndicClassTable addresses
//
static const IndicClassTable * const indicClassTables[scriptCodeCount] = {
    NULL,            /* 'zyyy' (COMMON) */
    NULL,            /* 'qaai' (INHERITED) */
    NULL,            /* 'arab' (ARABIC) */
    NULL,            /* 'armn' (ARMENIAN) */
    &bengClassTable, /* 'beng' (BENGALI) */
    NULL,            /* 'bopo' (BOPOMOFO) */
    NULL,            /* 'cher' (CHEROKEE) */
    NULL,            /* 'copt' (COPTIC) */
    NULL,            /* 'cyrl' (CYRILLIC) */
    NULL,            /* 'dsrt' (DESERET) */
    &devaClassTable, /* 'deva' (DEVANAGARI) */
    NULL,            /* 'ethi' (ETHIOPIC) */
    NULL,            /* 'geor' (GEORGIAN) */
    NULL,            /* 'goth' (GOTHIC) */
    NULL,            /* 'grek' (GREEK) */
    &gujrClassTable, /* 'gujr' (GUJARATI) */
    &punjClassTable, /* 'guru' (GURMUKHI) */
    NULL,            /* 'hani' (HAN) */
    NULL,            /* 'hang' (HANGUL) */
    NULL,            /* 'hebr' (HEBREW) */
    NULL,            /* 'hira' (HIRAGANA) */
    &kndaClassTable, /* 'knda' (KANNADA) */
    NULL,            /* 'kata' (KATAKANA) */
    NULL,            /* 'khmr' (KHMER) */
    NULL,            /* 'laoo' (LAO) */
    NULL,            /* 'latn' (LATIN) */
    &mlymClassTable, /* 'mlym' (MALAYALAM) */
    NULL,            /* 'mong' (MONGOLIAN) */
    NULL,            /* 'mymr' (MYANMAR) */
    NULL,            /* 'ogam' (OGHAM) */
    NULL,            /* 'ital' (OLD-ITALIC) */
    &oryaClassTable, /* 'orya' (ORIYA) */
    NULL,            /* 'runr' (RUNIC) */
    &sinhClassTable, /* 'sinh' (SINHALA) */
    NULL,            /* 'syrc' (SYRIAC) */
    &tamlClassTable, /* 'taml' (TAMIL) */
    &teluClassTable, /* 'telu' (TELUGU) */
    NULL,            /* 'thaa' (THAANA) */
    NULL,            /* 'thai' (THAI) */
    NULL,            /* 'tibt' (TIBETAN) */
    NULL,            /* 'cans' (CANADIAN-ABORIGINAL) */
    NULL,            /* 'yiii' (YI) */
    NULL,            /* 'tglg' (TAGALOG) */
    NULL,            /* 'hano' (HANUNOO) */
    NULL,            /* 'buhd' (BUHID) */
    NULL,            /* 'tagb' (TAGBANWA) */
    NULL,            /* 'brai' (BRAILLE) */
    NULL,            /* 'cprt' (CYPRIOT) */
    NULL,            /* 'limb' (LIMBU) */
    NULL,            /* 'linb' (LINEAR_B) */
    NULL,            /* 'osma' (OSMANYA) */
    NULL,            /* 'shaw' (SHAVIAN) */
    NULL,            /* 'tale' (TAI_LE) */
    NULL,            /* 'ugar' (UGARITIC) */
    NULL,            /* 'hrkt' (KATAKANA_OR_HIRAGANA) */
    NULL,            /* 'bugi' (BUGINESE) */
    NULL,            /* 'glag' (GLAGOLITIC) */
    NULL,            /* 'khar' (KHAROSHTHI) */
    NULL,            /* 'sylo' (SYLOTI_NAGRI) */
    NULL,            /* 'talu' (NEW_TAI_LUE) */
    NULL,            /* 'tfng' (TIFINAGH) */
    NULL,            /* 'xpeo' (OLD_PERSIAN) */
    NULL,            /* 'bali' (BALINESE) */
    NULL,            /* 'batk' (BATK) */
    NULL,            /* 'blis' (BLIS) */
    NULL,            /* 'brah' (BRAH) */
    NULL,            /* 'cham' (CHAM) */
    NULL,            /* 'cirt' (CIRT) */
    NULL,            /* 'cyrs' (CYRS) */
    NULL,            /* 'egyd' (EGYD) */
    NULL,            /* 'egyh' (EGYH) */
    NULL,            /* 'egyp' (EGYP) */
    NULL,            /* 'geok' (GEOK) */
    NULL,            /* 'hans' (HANS) */
    NULL,            /* 'hant' (HANT) */
    NULL,            /* 'hmng' (HMNG) */
    NULL,            /* 'hung' (HUNG) */
    NULL,            /* 'inds' (INDS) */
    NULL,            /* 'java' (JAVA) */
    NULL,            /* 'kali' (KALI) */
    NULL,            /* 'latf' (LATF) */
    NULL,            /* 'latg' (LATG) */
    NULL,            /* 'lepc' (LEPC) */
    NULL,            /* 'lina' (LINA) */
    NULL,            /* 'mand' (MAND) */
    NULL,            /* 'maya' (MAYA) */
    NULL,            /* 'mero' (MERO) */
    NULL,            /* 'nko ' (NKO) */
    NULL,            /* 'orkh' (ORKH) */
    NULL,            /* 'perm' (PERM) */
    NULL,            /* 'phag' (PHAGS_PA) */
    NULL,            /* 'phnx' (PHOENICIAN) */
    NULL,            /* 'plrd' (PLRD) */
    NULL,            /* 'roro' (RORO) */
    NULL,            /* 'sara' (SARA) */
    NULL,            /* 'syre' (SYRE) */
    NULL,            /* 'syrj' (SYRJ) */
    NULL,            /* 'syrn' (SYRN) */
    NULL,            /* 'teng' (TENG) */
    NULL,            /* 'vai ' (VAII) */
    NULL,            /* 'visp' (VISP) */
    NULL,            /* 'xsux' (CUNEIFORM) */
    NULL,            /* 'zxxx' (ZXXX) */
    NULL             /* 'zzzz' (UNKNOWN) */
};

IndicClassTable::CharClass IndicClassTable::getCharClass(LEUnicode ch) const
{
    if (ch == C_SIGN_ZWJ) {
        return CF_CONSONANT | CC_ZERO_WIDTH_MARK;
    }

    if (ch == C_SIGN_ZWNJ) {
        return CC_ZERO_WIDTH_MARK;
    }

    if (ch < firstChar || ch > lastChar) {
        return CC_RESERVED;
    }

    return classTable[ch - firstChar];
}

const IndicClassTable *IndicClassTable::getScriptClassTable(le_int32 scriptCode)
{
    if (scriptCode < 0 || scriptCode >= scriptCodeCount) {
        return NULL;
    }

    return indicClassTables[scriptCode];
}

le_int32 IndicReordering::getWorstCaseExpansion(le_int32 scriptCode)
{
    const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);

    if (classTable == NULL) {
        return 1;
    }

    return classTable->getWorstCaseExpansion();
}

le_bool IndicReordering::getFilterZeroWidth(le_int32 scriptCode)
{
    const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);

    if (classTable == NULL) {
        return TRUE;
    }

    return classTable->getFilterZeroWidth();
}

U_NAMESPACE_END
Commit	Line	Data
b75a7d8f	1	/*
b75a7d8f	2	*
73c04bcf	3	* (C) Copyright IBM Corp. 1998-2006 - All Rights Reserved
b75a7d8f A	4	*
	5	*/
	6
	7	#include "LETypes.h"
	8	#include "LEScripts.h"
	9	#include "OpenTypeTables.h"
	10	#include "OpenTypeUtilities.h"
	11	#include "IndicReordering.h"
	12
	13	U_NAMESPACE_BEGIN
	14
73c04bcf A	15	// Split matra table indices
	16	#define _x1 (1 << CF_INDEX_SHIFT)
	17	#define _x2 (2 << CF_INDEX_SHIFT)
	18	#define _x3 (3 << CF_INDEX_SHIFT)
	19	#define _x4 (4 << CF_INDEX_SHIFT)
	20	#define _x5 (5 << CF_INDEX_SHIFT)
	21	#define _x6 (6 << CF_INDEX_SHIFT)
	22	#define _x7 (7 << CF_INDEX_SHIFT)
	23	#define _x8 (8 << CF_INDEX_SHIFT)
	24	#define _x9 (9 << CF_INDEX_SHIFT)
	25
	26	// simple classes
	27	#define _xx (CC_RESERVED)
	28	#define _ma (CC_VOWEL_MODIFIER \| CF_POS_ABOVE)
	29	#define _mp (CC_VOWEL_MODIFIER \| CF_POS_AFTER)
	30	#define _sa (CC_STRESS_MARK \| CF_POS_ABOVE)
	31	#define _sb (CC_STRESS_MARK \| CF_POS_BELOW)
	32	#define _iv (CC_INDEPENDENT_VOWEL)
	33	#define _i2 (CC_INDEPENDENT_VOWEL_2)
	34	#define _i3 (CC_INDEPENDENT_VOWEL_3)
	35	#define _ct (CC_CONSONANT \| CF_CONSONANT)
	36	#define _cn (CC_CONSONANT_WITH_NUKTA \| CF_CONSONANT)
	37	#define _nu (CC_NUKTA)
	38	#define _dv (CC_DEPENDENT_VOWEL)
	39	#define _dl (_dv \| CF_POS_BEFORE)
	40	#define _db (_dv \| CF_POS_BELOW)
	41	#define _da (_dv \| CF_POS_ABOVE)
	42	#define _dr (_dv \| CF_POS_AFTER)
	43	#define _lm (_dv \| CF_LENGTH_MARK)
	44	#define _l1 (CC_SPLIT_VOWEL_PIECE_1 \| CF_POS_BEFORE)
	45	#define _a1 (CC_SPLIT_VOWEL_PIECE_1 \| CF_POS_ABOVE)
	46	#define _b2 (CC_SPLIT_VOWEL_PIECE_2 \| CF_POS_BELOW)
	47	#define _r2 (CC_SPLIT_VOWEL_PIECE_2 \| CF_POS_AFTER)
	48	#define _m2 (CC_SPLIT_VOWEL_PIECE_2 \| CF_LENGTH_MARK)
	49	#define _m3 (CC_SPLIT_VOWEL_PIECE_3 \| CF_LENGTH_MARK)
	50	#define _vr (CC_VIRAMA)
	51
	52	// split matras
	53	#define _s1 (_dv \| _x1)
	54	#define _s2 (_dv \| _x2)
	55	#define _s3 (_dv \| _x3)
	56	#define _s4 (_dv \| _x4)
	57	#define _s5 (_dv \| _x5)
	58	#define _s6 (_dv \| _x6)
	59	#define _s7 (_dv \| _x7)
	60	#define _s8 (_dv \| _x8)
	61	#define _s9 (_dv \| _x9)
	62
	63	// consonants with special forms
	64	// NOTE: this assumes that no consonants with nukta have
	65	// special forms... (Bengali RA?)
	66	#define _bb (_ct \| CF_BELOW_BASE)
	67	#define _pb (_ct \| CF_POST_BASE)
	68	#define _vt (_bb \| CF_VATTU)
	69	#define _rv (_vt \| CF_REPH)
	70	#define _rp (_pb \| CF_REPH)
	71	#define _rb (_bb \| CF_REPH)
b75a7d8f A	72
	73	//
	74	// Character class tables
	75	//
	76	static const IndicClassTable::CharClass devaCharClasses[] =
	77	{
374ca955	78	_xx, _ma, _ma, _mp, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0900 - 090F
b75a7d8f A	79	_iv, _iv, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0910 - 091F
b75a7d8f A	80	_ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _ct, // 0920 - 092F
374ca955	81	_rv, _cn, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0930 - 093F
b75a7d8f	82	_dr, _db, _db, _db, _db, _da, _da, _da, _da, _dr, _dr, _dr, _dr, _vr, _xx, _xx, // 0940 - 094F
374ca955	83	_xx, _sa, _sb, _sa, _sa, _xx, _xx, _xx, _cn, _cn, _cn, _cn, _cn, _cn, _cn, _cn, // 0950 - 095F
b75a7d8f A	84	_iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0960 - 096F
	85	_xx // 0970
	86	};
	87
	88	static const IndicClassTable::CharClass bengCharClasses[] =
	89	{
374ca955	90	_xx, _ma, _mp, _mp, _xx, _i2, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _i2, // 0980 - 098F
b75a7d8f A	91	_iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0990 - 099F
b75a7d8f A	92	_ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _bb, _ct, _ct, _pb, // 09A0 - 09AF
374ca955 A	93	_rv, _xx, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _dl, // 09B0 - 09BF
	94	_dr, _db, _db, _db, _db, _xx, _xx, _l1, _dl, _xx, _xx, _s1, _s2, _vr, _xx, _xx, // 09C0 - 09CF
	95	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 09D0 - 09DF
b75a7d8f A	96	_iv, _iv, _dv, _dv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 09E0 - 09EF
	97	_ct, _ct, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 09F0 - 09FA
	98	};
	99
	100	static const IndicClassTable::CharClass punjCharClasses[] =
	101	{
374ca955	102	_xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _xx, _iv, // 0A00 - 0A0F
73c04bcf	103	_iv, _xx, _xx, _i3, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A10 - 0A1F
b75a7d8f	104	_ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _bb, // 0A20 - 0A2F
73c04bcf A	105	_vt, _xx, _ct, _cn, _xx, _bb, _cn, _xx, _ct, _bb, _xx, _xx, _nu, _xx, _dr, _dl, // 0A30 - 0A3F
	106	_dr, _b2, _db, _xx, _xx, _xx, _xx, _da, _da, _xx, _xx, _a1, _da, _vr, _xx, _xx, // 0A40 - 0A4F
	107	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _cn, _cn, _cn, _ct, _xx, _cn, _xx, // 0A50 - 0A5F
b75a7d8f	108	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0A60 - 0A6F
73c04bcf	109	_ma, _ma, _xx, _xx, _xx // 0A70 - 0A74
b75a7d8f A	110	};
	111
	112	static const IndicClassTable::CharClass gujrCharClasses[] =
	113	{
	114	_xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _xx, _iv, // 0A80 - 0A8F
	115	_iv, _iv, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A90 - 0A9F
	116	_ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0AA0 - 0AAF
	117	_rv, _xx, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0AB0 - 0ABF
	118	_dr, _db, _db, _db, _db, _da, _xx, _da, _da, _dr, _xx, _dr, _dr, _vr, _xx, _xx, // 0AC0 - 0ACF
	119	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0AD0 - 0ADF
	120	_iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0AE0 - 0AEF
	121	};
	122
	123	static const IndicClassTable::CharClass oryaCharClasses[] =
	124	{
73c04bcf	125	_xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, // 0B00 - 0B0F
b75a7d8f A	126	_iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0B10 - 0B1F
b75a7d8f A	127	_ct, _ct, _ct, _ct, _bb, _ct, _ct, _ct, _bb, _xx, _ct, _ct, _bb, _bb, _bb, _pb, // 0B20 - 0B2F
73c04bcf	128	_rb, _xx, _bb, _bb, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _da, // 0B30 - 0B3F
374ca955 A	129	_dr, _db, _db, _db, _xx, _xx, _xx, _l1, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, // 0B40 - 0B4F
374ca955 A	130	_xx, _xx, _xx, _xx, _xx, _xx, _m2, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 0B50 - 0B5F
b75a7d8f	131	_iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0B60 - 0B6F
73c04bcf	132	_xx, _ct // 0B70 - 0B71
b75a7d8f A	133	};
	134
	135	static const IndicClassTable::CharClass tamlCharClasses[] =
	136	{
73c04bcf	137	_xx, _xx, _ma, _xx, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _iv, _iv, // 0B80 - 0B8F
b75a7d8f A	138	_iv, _xx, _iv, _iv, _iv, _ct, _xx, _xx, _xx, _ct, _ct, _xx, _ct, _xx, _ct, _ct, // 0B90 - 0B9F
b75a7d8f A	139	_xx, _xx, _xx, _ct, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _xx, _xx, _xx, _ct, _ct, // 0BA0 - 0BAF
374ca955 A	140	_ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0BB0 - 0BBF
	141	_da, _dr, _dr, _xx, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0BC0 - 0BCF
	142	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BD0 - 0BDF
b75a7d8f A	143	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BE0 - 0BEF
	144	_xx, _xx, _xx // 0BF0 - 0BF2
	145	};
	146
	147	// FIXME: Should some of the bb's be pb's? (KA, NA, MA, YA, VA, etc. (approx 13))
	148	static const IndicClassTable::CharClass teluCharClasses[] =
	149	{
	150	_xx, _mp, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0C00 - 0C0F
	151	_iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C10 - 0C1F
	152	_bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0C20 - 0C2F
	153	_bb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _da, _da, // 0C30 - 0C3F
374ca955 A	154	_da, _dr, _dr, _dr, _dr, _xx, _a1, _da, _s1, _xx, _da, _da, _da, _vr, _xx, _xx, // 0C40 - 0C4F
374ca955 A	155	_xx, _xx, _xx, _xx, _xx, _da, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0C50 - 0C5F
b75a7d8f A	156	_iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0C60 - 0C6F
	157	};
	158
374ca955 A	159	// U+CC3 and U+CC4 are _lm here not _dr since the Kannada rendering
	160	// rules want them below and to the right of the entire cluster
	161	//
	162	// There's some information about this in:
	163	//
	164	// http://brahmi.sourceforge.net/docs/KannadaComputing.html
b75a7d8f A	165	static const IndicClassTable::CharClass kndaCharClasses[] =
	166	{
	167	_xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, // 0C80 - 0C8F
	168	_iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C90 - 0C9F
	169	_bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0CA0 - 0CAF
	170	_rb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _dr, _da, // 0CB0 - 0CBF
374ca955 A	171	_s1, _dr, _r2, _lm, _lm, _xx, _a1, _s2, _s3, _xx, _s4, _s5, _da, _vr, _xx, _xx, // 0CC0 - 0CCF
374ca955 A	172	_xx, _xx, _xx, _xx, _xx, _m3, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _ct, _xx, // 0CD0 - 0CDF
b75a7d8f A	173	_iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0CE0 - 0CEF
	174	};
	175
	176	// FIXME: this is correct for old-style Malayalam (MAL) but not for reformed Malayalam (MLR)
	177	// FIXME: should there be a REPH for old-style Malayalam?
	178	static const IndicClassTable::CharClass mlymCharClasses[] =
	179	{
	180	_xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0D00 - 0D0F
	181	_iv, _xx, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _bb, // 0D10 - 0D1F
374ca955 A	182	_ct, _ct, _ct, _bb, _ct, _bb, _bb, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _pb, // 0D20 - 0D2F
	183	_pb, _cn, _bb, _ct, _ct, _pb, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0D30 - 0D3F
	184	_dr, _dr, _dr, _dr, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0D40 - 0D4F
	185	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0D50 - 0D5F
b75a7d8f A	186	_iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0D60 - 0D6F
b75a7d8f A	187	};
73c04bcf A	188
	189	static const IndicClassTable::CharClass sinhCharClasses[] =
	190	{
	191	_xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0D80 - 0D8F
	192	_iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0D90 - 0D9F
	193	_ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0DA0 - 0DAF
	194	_ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _xx, _xx, // 0DB0 - 0DBF
	195	_ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _vr, _xx, _xx, _xx, _xx, _dr, // 0DC0 - 0DCF
	196	_dr, _dr, _da, _da, _db, _xx, _db, _xx, _dr, _dl, _s1, _dl, _s2, _s3, _s4, _dr, // 0DD0 - 0DDF
	197	_xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0DE0 - 0DEF
	198	_xx, _xx, _dr, _dr, _xx // 0DF0 - 0DF4
	199	};
b75a7d8f A	200
	201	//
	202	// Split matra tables
	203	//
	204	static const SplitMatra bengSplitTable[] = {{0x09C7, 0x09BE}, {0x09C7, 0x09D7}};
	205
	206	static const SplitMatra oryaSplitTable[] = {{0x0B47, 0x0B56}, {0x0B47, 0x0B3E}, {0x0B47, 0x0B57}};
	207
	208	static const SplitMatra tamlSplitTable[] = {{0x0BC6, 0x0BBE}, {0x0BC7, 0x0BBE}, {0x0BC6, 0x0BD7}};
	209
	210	static const SplitMatra teluSplitTable[] = {{0x0C46, 0x0C56}};
	211
	212	static const SplitMatra kndaSplitTable[] = {{0x0CBF, 0x0CD5}, {0x0CC6, 0x0CD5}, {0x0CC6, 0x0CD6}, {0x0CC6, 0x0CC2},
73c04bcf	213	{0x0CC6, 0x0CC2, 0x0CD5}};
b75a7d8f A	214
	215	static const SplitMatra mlymSplitTable[] = {{0x0D46, 0x0D3E}, {0x0D47, 0x0D3E}, {0x0D46, 0x0D57}};
	216
73c04bcf A	217
	218	static const SplitMatra sinhSplitTable[] = {{0x0DD9, 0x0DCA}, {0x0DD9, 0x0DCF}, {0x0DD9, 0x0DCF,0x0DCA},
	219	{0x0DD9, 0x0DDF}};
b75a7d8f A	220	//
	221	// Script Flags
	222	//
	223
	224	// FIXME: post 'GSUB' reordering of MATRA_PRE's for Malayalam and Tamil
	225	// FIXME: reformed Malayalam needs to reorder VATTU to before base glyph...
	226	// FIXME: eyelash RA only for Devanagari??
73c04bcf A	227	#define DEVA_SCRIPT_FLAGS (SF_EYELASH_RA \| SF_NO_POST_BASE_LIMIT \| SF_FILTER_ZERO_WIDTH)
	228	#define BENG_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW \| SF_NO_POST_BASE_LIMIT \| SF_FILTER_ZERO_WIDTH)
	229	#define PUNJ_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT \| SF_FILTER_ZERO_WIDTH)
	230	#define GUJR_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT \| SF_FILTER_ZERO_WIDTH)
	231	#define ORYA_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW \| SF_NO_POST_BASE_LIMIT \| SF_FILTER_ZERO_WIDTH)
	232	#define TAML_SCRIPT_FLAGS (SF_MPRE_FIXUP \| SF_NO_POST_BASE_LIMIT \| SF_FILTER_ZERO_WIDTH)
	233	#define TELU_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE \| SF_FILTER_ZERO_WIDTH \| 3)
	234	#define KNDA_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE \| SF_FILTER_ZERO_WIDTH \| 3)
	235	#define MLYM_SCRIPT_FLAGS (SF_MPRE_FIXUP \| SF_NO_POST_BASE_LIMIT \| SF_FILTER_ZERO_WIDTH)
	236	#define SINH_SCRIPT_FLAGS (SF_MPRE_FIXUP \| SF_NO_POST_BASE_LIMIT)
b75a7d8f A	237
	238	//
	239	// Indic Class Tables
	240	//
	241	static const IndicClassTable devaClassTable = {0x0900, 0x0970, 2, DEVA_SCRIPT_FLAGS, devaCharClasses, NULL};
	242
	243	static const IndicClassTable bengClassTable = {0x0980, 0x09FA, 3, BENG_SCRIPT_FLAGS, bengCharClasses, bengSplitTable};
	244
	245	static const IndicClassTable punjClassTable = {0x0A00, 0x0A74, 2, PUNJ_SCRIPT_FLAGS, punjCharClasses, NULL};
	246
	247	static const IndicClassTable gujrClassTable = {0x0A80, 0x0AEF, 2, GUJR_SCRIPT_FLAGS, gujrCharClasses, NULL};
	248
73c04bcf	249	static const IndicClassTable oryaClassTable = {0x0B00, 0x0B71, 3, ORYA_SCRIPT_FLAGS, oryaCharClasses, oryaSplitTable};
b75a7d8f A	250
	251	static const IndicClassTable tamlClassTable = {0x0B80, 0x0BF2, 3, TAML_SCRIPT_FLAGS, tamlCharClasses, tamlSplitTable};
	252
	253	static const IndicClassTable teluClassTable = {0x0C00, 0x0C6F, 3, TELU_SCRIPT_FLAGS, teluCharClasses, teluSplitTable};
	254
	255	static const IndicClassTable kndaClassTable = {0x0C80, 0x0CEF, 4, KNDA_SCRIPT_FLAGS, kndaCharClasses, kndaSplitTable};
	256
	257	static const IndicClassTable mlymClassTable = {0x0D00, 0x0D6F, 3, MLYM_SCRIPT_FLAGS, mlymCharClasses, mlymSplitTable};
	258
73c04bcf A	259	static const IndicClassTable sinhClassTable = {0x0D80, 0x0DF4, 3, SINH_SCRIPT_FLAGS, sinhCharClasses, sinhSplitTable};
73c04bcf A	260
b75a7d8f A	261	//
	262	// IndicClassTable addresses
	263	//
73c04bcf	264	static const IndicClassTable * const indicClassTables[scriptCodeCount] = {
b75a7d8f A	265	NULL, /* 'zyyy' (COMMON) */
	266	NULL, /* 'qaai' (INHERITED) */
	267	NULL, /* 'arab' (ARABIC) */
	268	NULL, /* 'armn' (ARMENIAN) */
	269	&bengClassTable, /* 'beng' (BENGALI) */
	270	NULL, /* 'bopo' (BOPOMOFO) */
	271	NULL, /* 'cher' (CHEROKEE) */
73c04bcf	272	NULL, /* 'copt' (COPTIC) */
b75a7d8f A	273	NULL, /* 'cyrl' (CYRILLIC) */
	274	NULL, /* 'dsrt' (DESERET) */
	275	&devaClassTable, /* 'deva' (DEVANAGARI) */
	276	NULL, /* 'ethi' (ETHIOPIC) */
	277	NULL, /* 'geor' (GEORGIAN) */
	278	NULL, /* 'goth' (GOTHIC) */
	279	NULL, /* 'grek' (GREEK) */
	280	&gujrClassTable, /* 'gujr' (GUJARATI) */
	281	&punjClassTable, /* 'guru' (GURMUKHI) */
	282	NULL, /* 'hani' (HAN) */
	283	NULL, /* 'hang' (HANGUL) */
	284	NULL, /* 'hebr' (HEBREW) */
	285	NULL, /* 'hira' (HIRAGANA) */
	286	&kndaClassTable, /* 'knda' (KANNADA) */
	287	NULL, /* 'kata' (KATAKANA) */
	288	NULL, /* 'khmr' (KHMER) */
	289	NULL, /* 'laoo' (LAO) */
	290	NULL, /* 'latn' (LATIN) */
	291	&mlymClassTable, /* 'mlym' (MALAYALAM) */
	292	NULL, /* 'mong' (MONGOLIAN) */
	293	NULL, /* 'mymr' (MYANMAR) */
	294	NULL, /* 'ogam' (OGHAM) */
	295	NULL, /* 'ital' (OLD-ITALIC) */
	296	&oryaClassTable, /* 'orya' (ORIYA) */
	297	NULL, /* 'runr' (RUNIC) */
73c04bcf	298	&sinhClassTable, /* 'sinh' (SINHALA) */
b75a7d8f A	299	NULL, /* 'syrc' (SYRIAC) */
	300	&tamlClassTable, /* 'taml' (TAMIL) */
	301	&teluClassTable, /* 'telu' (TELUGU) */
	302	NULL, /* 'thaa' (THAANA) */
	303	NULL, /* 'thai' (THAI) */
	304	NULL, /* 'tibt' (TIBETAN) */
	305	NULL, /* 'cans' (CANADIAN-ABORIGINAL) */
	306	NULL, /* 'yiii' (YI) */
	307	NULL, /* 'tglg' (TAGALOG) */
	308	NULL, /* 'hano' (HANUNOO) */
	309	NULL, /* 'buhd' (BUHID) */
	310	NULL, /* 'tagb' (TAGBANWA) */
	311	NULL, /* 'brai' (BRAILLE) */
	312	NULL, /* 'cprt' (CYPRIOT) */
	313	NULL, /* 'limb' (LIMBU) */
	314	NULL, /* 'linb' (LINEAR_B) */
	315	NULL, /* 'osma' (OSMANYA) */
	316	NULL, /* 'shaw' (SHAVIAN) */
	317	NULL, /* 'tale' (TAI_LE) */
374ca955	318	NULL, /* 'ugar' (UGARITIC) */
73c04bcf A	319	NULL, /* 'hrkt' (KATAKANA_OR_HIRAGANA) */
	320	NULL, /* 'bugi' (BUGINESE) */
	321	NULL, /* 'glag' (GLAGOLITIC) */
	322	NULL, /* 'khar' (KHAROSHTHI) */
	323	NULL, /* 'sylo' (SYLOTI_NAGRI) */
	324	NULL, /* 'talu' (NEW_TAI_LUE) */
	325	NULL, /* 'tfng' (TIFINAGH) */
	326	NULL, /* 'xpeo' (OLD_PERSIAN) */
	327	NULL, /* 'bali' (BALINESE) */
	328	NULL, /* 'batk' (BATK) */
	329	NULL, /* 'blis' (BLIS) */
	330	NULL, /* 'brah' (BRAH) */
	331	NULL, /* 'cham' (CHAM) */
	332	NULL, /* 'cirt' (CIRT) */
	333	NULL, /* 'cyrs' (CYRS) */
	334	NULL, /* 'egyd' (EGYD) */
	335	NULL, /* 'egyh' (EGYH) */
	336	NULL, /* 'egyp' (EGYP) */
	337	NULL, /* 'geok' (GEOK) */
	338	NULL, /* 'hans' (HANS) */
	339	NULL, /* 'hant' (HANT) */
	340	NULL, /* 'hmng' (HMNG) */
	341	NULL, /* 'hung' (HUNG) */
	342	NULL, /* 'inds' (INDS) */
	343	NULL, /* 'java' (JAVA) */
	344	NULL, /* 'kali' (KALI) */
	345	NULL, /* 'latf' (LATF) */
	346	NULL, /* 'latg' (LATG) */
	347	NULL, /* 'lepc' (LEPC) */
	348	NULL, /* 'lina' (LINA) */
	349	NULL, /* 'mand' (MAND) */
	350	NULL, /* 'maya' (MAYA) */
	351	NULL, /* 'mero' (MERO) */
	352	NULL, /* 'nko ' (NKO) */
	353	NULL, /* 'orkh' (ORKH) */
	354	NULL, /* 'perm' (PERM) */
	355	NULL, /* 'phag' (PHAGS_PA) */
	356	NULL, /* 'phnx' (PHOENICIAN) */
	357	NULL, /* 'plrd' (PLRD) */
	358	NULL, /* 'roro' (RORO) */
	359	NULL, /* 'sara' (SARA) */
	360	NULL, /* 'syre' (SYRE) */
	361	NULL, /* 'syrj' (SYRJ) */
	362	NULL, /* 'syrn' (SYRN) */
	363	NULL, /* 'teng' (TENG) */
	364	NULL, /* 'vai ' (VAII) */
	365	NULL, /* 'visp' (VISP) */
	366	NULL, /* 'xsux' (CUNEIFORM) */
	367	NULL, /* 'zxxx' (ZXXX) */
	368	NULL /* 'zzzz' (UNKNOWN) */
b75a7d8f A	369	};
	370
	371	IndicClassTable::CharClass IndicClassTable::getCharClass(LEUnicode ch) const
	372	{
	373	if (ch == C_SIGN_ZWJ) {
	374	return CF_CONSONANT \| CC_ZERO_WIDTH_MARK;
	375	}
	376
	377	if (ch == C_SIGN_ZWNJ) {
	378	return CC_ZERO_WIDTH_MARK;
	379	}
	380
	381	if (ch < firstChar \|\| ch > lastChar) {
	382	return CC_RESERVED;
	383	}
	384
	385	return classTable[ch - firstChar];
	386	}
	387
	388	const IndicClassTable *IndicClassTable::getScriptClassTable(le_int32 scriptCode)
	389	{
	390	if (scriptCode < 0 \|\| scriptCode >= scriptCodeCount) {
	391	return NULL;
	392	}
	393
	394	return indicClassTables[scriptCode];
	395	}
	396
	397	le_int32 IndicReordering::getWorstCaseExpansion(le_int32 scriptCode)
	398	{
	399	const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);
	400
	401	if (classTable == NULL) {
	402	return 1;
	403	}
	404
	405	return classTable->getWorstCaseExpansion();
	406	}
	407
73c04bcf A	408	le_bool IndicReordering::getFilterZeroWidth(le_int32 scriptCode)
	409	{
	410	const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);
	411
	412	if (classTable == NULL) {
	413	return TRUE;
	414	}
	415
	416	return classTable->getFilterZeroWidth();
	417	}
	418
b75a7d8f	419	U_NAMESPACE_END