• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  *
3  * (C) Copyright IBM Corp. 1998-2010 - All Rights Reserved
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 
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 #define _al  (CC_AL_LAKUNA)
52 
53 // split matras
54 #define _s1  (_dv | _x1)
55 #define _s2  (_dv | _x2)
56 #define _s3  (_dv | _x3)
57 #define _s4  (_dv | _x4)
58 #define _s5  (_dv | _x5)
59 #define _s6  (_dv | _x6)
60 #define _s7  (_dv | _x7)
61 #define _s8  (_dv | _x8)
62 #define _s9  (_dv | _x9)
63 
64 // consonants with special forms
65 // NOTE: this assumes that no consonants with nukta have
66 // special forms... (Bengali RA?)
67 #define _bb  (_ct | CF_BELOW_BASE)
68 #define _pb  (_ct | CF_POST_BASE)
69 #define _fb  (_ct | CF_PRE_BASE)
70 #define _vt  (_bb | CF_VATTU)
71 #define _rv  (_vt | CF_REPH)
72 #define _rp  (_pb | CF_REPH)
73 #define _rb  (_bb | CF_REPH)
74 
75 //
76 // Character class tables
77 //
78 static const IndicClassTable::CharClass devaCharClasses[] =
79 {
80     _xx, _ma, _ma, _mp, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0900 - 090F
81     _iv, _iv, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0910 - 091F
82     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _ct, // 0920 - 092F
83     _rv, _cn, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0930 - 093F
84     _dr, _db, _db, _db, _db, _da, _da, _da, _da, _dr, _dr, _dr, _dr, _vr, _xx, _xx, // 0940 - 094F
85     _xx, _sa, _sb, _sa, _sa, _xx, _xx, _xx, _cn, _cn, _cn, _cn, _cn, _cn, _cn, _cn, // 0950 - 095F
86     _iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0960 - 096F
87     _xx                                                                             // 0970
88 };
89 
90 static const IndicClassTable::CharClass bengCharClasses[] =
91 {
92     _xx, _ma, _mp, _mp, _xx, _i2, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _i2, // 0980 - 098F
93     _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0990 - 099F
94     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _bb, _ct, _ct, _pb, // 09A0 - 09AF
95     _rv, _xx, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _dl, // 09B0 - 09BF
96     _dr, _db, _db, _db, _db, _xx, _xx, _l1, _dl, _xx, _xx, _s1, _s2, _vr, _xx, _xx, // 09C0 - 09CF
97     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 09D0 - 09DF
98     _iv, _iv, _dv, _dv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 09E0 - 09EF
99     _rv, _ct, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx                           // 09F0 - 09FA
100 };
101 
102 static const IndicClassTable::CharClass punjCharClasses[] =
103 {
104     _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _xx, _iv, // 0A00 - 0A0F
105     _iv, _xx, _xx, _i3, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A10 - 0A1F
106     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _bb, // 0A20 - 0A2F
107     _vt, _xx, _ct, _cn, _xx, _bb, _cn, _xx, _ct, _bb, _xx, _xx, _nu, _xx, _dr, _dl, // 0A30 - 0A3F
108     _dr, _b2, _db, _xx, _xx, _xx, _xx, _da, _da, _xx, _xx, _a1, _da, _vr, _xx, _xx, // 0A40 - 0A4F
109     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _cn, _cn, _cn, _ct, _xx, _cn, _xx, // 0A50 - 0A5F
110     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0A60 - 0A6F
111     _ma, _ma, _xx, _xx, _xx                                                         // 0A70 - 0A74
112 };
113 
114 static const IndicClassTable::CharClass gujrCharClasses[] =
115 {
116     _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _xx, _iv, // 0A80 - 0A8F
117     _iv, _iv, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A90 - 0A9F
118     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0AA0 - 0AAF
119     _rv, _xx, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0AB0 - 0ABF
120     _dr, _db, _db, _db, _db, _da, _xx, _da, _da, _dr, _xx, _dr, _dr, _vr, _xx, _xx, // 0AC0 - 0ACF
121     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0AD0 - 0ADF
122     _iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0AE0 - 0AEF
123 };
124 
125 #if 1
126 static const IndicClassTable::CharClass oryaCharClasses[] =
127 {
128     _xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, /* 0B00 - 0B0F */
129     _iv, _xx, _xx, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _ct, _bb, /* 0B10 - 0B1F */
130     _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _pb, /* 0B20 - 0B2F */
131     _rb, _xx, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _nu, _xx, _dr, _da, /* 0B30 - 0B3F */
132     _dr, _db, _db, _db, _xx, _xx, _xx, _dl, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, /* 0B40 - 0B4F */
133     _xx, _xx, _xx, _xx, _xx, _xx, _da, _dr, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _pb, /* 0B50 - 0B5F */
134     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, /* 0B60 - 0B6F */
135     _xx, _bb                                                                        /* 0B70 - 0B71 */
136 };
137 #else
138 static const IndicClassTable::CharClass oryaCharClasses[] =
139 {
140     _xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, // 0B00 - 0B0F
141     _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0B10 - 0B1F
142     _ct, _ct, _ct, _ct, _bb, _ct, _ct, _ct, _bb, _xx, _ct, _ct, _bb, _bb, _bb, _pb, // 0B20 - 0B2F
143     _rb, _xx, _bb, _bb, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _da, // 0B30 - 0B3F
144     _dr, _db, _db, _db, _xx, _xx, _xx, _l1, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, // 0B40 - 0B4F
145     _xx, _xx, _xx, _xx, _xx, _xx, _m2, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 0B50 - 0B5F
146     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0B60 - 0B6F
147     _xx, _ct                                                                        // 0B70 - 0B71
148 };
149 #endif
150 
151 static const IndicClassTable::CharClass tamlCharClasses[] =
152 {
153     _xx, _xx, _ma, _xx, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _iv, _iv, // 0B80 - 0B8F
154     _iv, _xx, _iv, _iv, _iv, _ct, _xx, _xx, _xx, _ct, _ct, _xx, _ct, _xx, _ct, _ct, // 0B90 - 0B9F
155     _xx, _xx, _xx, _ct, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _xx, _xx, _xx, _ct, _ct, // 0BA0 - 0BAF
156     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0BB0 - 0BBF
157     _da, _dr, _dr, _xx, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0BC0 - 0BCF
158     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BD0 - 0BDF
159     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BE0 - 0BEF
160     _xx, _xx, _xx                                                                   // 0BF0 - 0BF2
161 };
162 
163 // FIXME: Should some of the bb's be pb's? (KA, NA, MA, YA, VA, etc. (approx 13))
164 static const IndicClassTable::CharClass teluCharClasses[] =
165 {
166     _xx, _mp, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0C00 - 0C0F
167     _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C10 - 0C1F
168     _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0C20 - 0C2F
169     _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _da, _da, // 0C30 - 0C3F
170     _da, _dr, _dr, _dr, _dr, _xx, _a1, _da, _s1, _xx, _da, _da, _da, _vr, _xx, _xx, // 0C40 - 0C4F
171     _xx, _xx, _xx, _xx, _xx, _da, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0C50 - 0C5F
172     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0C60 - 0C6F
173 };
174 
175 // U+CC3 and U+CC4 are _lm here not _dr since the Kannada rendering
176 // rules want them below and to the right of the entire cluster
177 //
178 // There's some information about this in:
179 //
180 //  http://brahmi.sourceforge.net/docs/KannadaComputing.html
181 static const IndicClassTable::CharClass kndaCharClasses[] =
182 {
183     _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0C80 - 0C8F
184     _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C90 - 0C9F
185     _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0CA0 - 0CAF
186     _rb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _dr, _da, // 0CB0 - 0CBF
187     _s1, _dr, _r2, _lm, _lm, _xx, _a1, _s2, _s3, _xx, _s4, _s5, _da, _vr, _xx, _xx, // 0CC0 - 0CCF
188     _xx, _xx, _xx, _xx, _xx, _m3, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _ct, _xx, // 0CD0 - 0CDF
189     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0CE0 - 0CEF
190 };
191 
192 // FIXME: this is correct for old-style Malayalam (MAL) but not for reformed Malayalam (MLR)
193 // FIXME: should there be a REPH for old-style Malayalam?
194 static const IndicClassTable::CharClass mlymCharClasses[] =
195 {
196     _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0D00 - 0D0F
197     _iv, _xx, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0D10 - 0D1F
198     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _pb, // 0D20 - 0D2F
199     _fb, _fb, _bb, _ct, _ct, _pb, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0D30 - 0D3F
200     _dr, _dr, _dr, _dr, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0D40 - 0D4F
201     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0D50 - 0D5F
202     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0D60 - 0D6F
203 };
204 
205 static const IndicClassTable::CharClass sinhCharClasses[] =
206 {
207     _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0D80 - 0D8F
208     _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0D90 - 0D9F
209     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0DA0 - 0DAF
210     _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _xx, _xx, // 0DB0 - 0DBF
211     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _al, _xx, _xx, _xx, _xx, _dr, // 0DC0 - 0DCF
212     _dr, _dr, _da, _da, _db, _xx, _db, _xx, _dr, _dl, _s1, _dl, _s2, _s3, _s4, _dr, // 0DD0 - 0DDF
213     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0DE0 - 0DEF
214     _xx, _xx, _dr, _dr, _xx                                                         // 0DF0 - 0DF4
215 };
216 
217 //
218 // Split matra tables
219 //
220 static const SplitMatra bengSplitTable[] = {{0x09C7, 0x09BE}, {0x09C7, 0x09D7}};
221 
222 static const SplitMatra oryaSplitTable[] = {{0x0B47, 0x0B56}, {0x0B47, 0x0B3E}, {0x0B47, 0x0B57}};
223 
224 static const SplitMatra tamlSplitTable[] = {{0x0BC6, 0x0BBE}, {0x0BC7, 0x0BBE}, {0x0BC6, 0x0BD7}};
225 
226 static const SplitMatra teluSplitTable[] = {{0x0C46, 0x0C56}};
227 
228 static const SplitMatra kndaSplitTable[] = {{0x0CBF, 0x0CD5}, {0x0CC6, 0x0CD5}, {0x0CC6, 0x0CD6}, {0x0CC6, 0x0CC2},
229                                             {0x0CC6, 0x0CC2, 0x0CD5}};
230 
231 static const SplitMatra mlymSplitTable[] = {{0x0D46, 0x0D3E}, {0x0D47, 0x0D3E}, {0x0D46, 0x0D57}};
232 
233 
234 static const SplitMatra sinhSplitTable[] = {{0x0DD9, 0x0DCA}, {0x0DD9, 0x0DCF}, {0x0DD9, 0x0DCF, 0x0DCA},
235                                             {0x0DD9, 0x0DDF}};
236 //
237 // Script Flags
238 //
239 
240 // FIXME: post 'GSUB' reordering of MATRA_PRE's for Malayalam and Tamil
241 // FIXME: reformed Malayalam needs to reorder VATTU to before base glyph...
242 // FIXME: not sure passing ZWJ/ZWNJ is best way to render Malayalam Cillu...
243 // FIXME: eyelash RA only for Devanagari??
244 #define DEVA_SCRIPT_FLAGS (SF_EYELASH_RA | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
245 #define BENG_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
246 #define PUNJ_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
247 #define GUJR_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
248 #define ORYA_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
249 #define TAML_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
250 #define TELU_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3)
251 #define KNDA_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3)
252 #define MLYM_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT /*| SF_FILTER_ZERO_WIDTH*/)
253 #define SINH_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT)
254 
255 //
256 // Indic Class Tables
257 //
258 static const IndicClassTable devaClassTable = {0x0900, 0x0970, 2, DEVA_SCRIPT_FLAGS, devaCharClasses, NULL};
259 
260 static const IndicClassTable bengClassTable = {0x0980, 0x09FA, 3, BENG_SCRIPT_FLAGS, bengCharClasses, bengSplitTable};
261 
262 static const IndicClassTable punjClassTable = {0x0A00, 0x0A74, 2, PUNJ_SCRIPT_FLAGS, punjCharClasses, NULL};
263 
264 static const IndicClassTable gujrClassTable = {0x0A80, 0x0AEF, 2, GUJR_SCRIPT_FLAGS, gujrCharClasses, NULL};
265 
266 static const IndicClassTable oryaClassTable = {0x0B00, 0x0B71, 3, ORYA_SCRIPT_FLAGS, oryaCharClasses, oryaSplitTable};
267 
268 static const IndicClassTable tamlClassTable = {0x0B80, 0x0BF2, 3, TAML_SCRIPT_FLAGS, tamlCharClasses, tamlSplitTable};
269 
270 static const IndicClassTable teluClassTable = {0x0C00, 0x0C6F, 3, TELU_SCRIPT_FLAGS, teluCharClasses, teluSplitTable};
271 
272 static const IndicClassTable kndaClassTable = {0x0C80, 0x0CEF, 4, KNDA_SCRIPT_FLAGS, kndaCharClasses, kndaSplitTable};
273 
274 static const IndicClassTable mlymClassTable = {0x0D00, 0x0D6F, 3, MLYM_SCRIPT_FLAGS, mlymCharClasses, mlymSplitTable};
275 
276 static const IndicClassTable sinhClassTable = {0x0D80, 0x0DF4, 4, SINH_SCRIPT_FLAGS, sinhCharClasses, sinhSplitTable};
277 
278 //
279 // IndicClassTable addresses
280 //
281 static const IndicClassTable * const indicClassTables[scriptCodeCount] = {
282     NULL,            /* 'zyyy' (COMMON) */
283     NULL,            /* 'qaai' (INHERITED) */
284     NULL,            /* 'arab' (ARABIC) */
285     NULL,            /* 'armn' (ARMENIAN) */
286     &bengClassTable, /* 'beng' (BENGALI) */
287     NULL,            /* 'bopo' (BOPOMOFO) */
288     NULL,            /* 'cher' (CHEROKEE) */
289     NULL,            /* 'copt' (COPTIC) */
290     NULL,            /* 'cyrl' (CYRILLIC) */
291     NULL,            /* 'dsrt' (DESERET) */
292     &devaClassTable, /* 'deva' (DEVANAGARI) */
293     NULL,            /* 'ethi' (ETHIOPIC) */
294     NULL,            /* 'geor' (GEORGIAN) */
295     NULL,            /* 'goth' (GOTHIC) */
296     NULL,            /* 'grek' (GREEK) */
297     &gujrClassTable, /* 'gujr' (GUJARATI) */
298     &punjClassTable, /* 'guru' (GURMUKHI) */
299     NULL,            /* 'hani' (HAN) */
300     NULL,            /* 'hang' (HANGUL) */
301     NULL,            /* 'hebr' (HEBREW) */
302     NULL,            /* 'hira' (HIRAGANA) */
303     &kndaClassTable, /* 'knda' (KANNADA) */
304     NULL,            /* 'kata' (KATAKANA) */
305     NULL,            /* 'khmr' (KHMER) */
306     NULL,            /* 'laoo' (LAO) */
307     NULL,            /* 'latn' (LATIN) */
308     &mlymClassTable, /* 'mlym' (MALAYALAM) */
309     NULL,            /* 'mong' (MONGOLIAN) */
310     NULL,            /* 'mymr' (MYANMAR) */
311     NULL,            /* 'ogam' (OGHAM) */
312     NULL,            /* 'ital' (OLD-ITALIC) */
313     &oryaClassTable, /* 'orya' (ORIYA) */
314     NULL,            /* 'runr' (RUNIC) */
315     &sinhClassTable, /* 'sinh' (SINHALA) */
316     NULL,            /* 'syrc' (SYRIAC) */
317     &tamlClassTable, /* 'taml' (TAMIL) */
318     &teluClassTable, /* 'telu' (TELUGU) */
319     NULL,            /* 'thaa' (THAANA) */
320     NULL,            /* 'thai' (THAI) */
321     NULL,            /* 'tibt' (TIBETAN) */
322     NULL,            /* 'cans' (CANADIAN-ABORIGINAL) */
323     NULL,            /* 'yiii' (YI) */
324     NULL,            /* 'tglg' (TAGALOG) */
325     NULL,            /* 'hano' (HANUNOO) */
326     NULL,            /* 'buhd' (BUHID) */
327     NULL,            /* 'tagb' (TAGBANWA) */
328     NULL,            /* 'brai' (BRAILLE) */
329     NULL,            /* 'cprt' (CYPRIOT) */
330     NULL,            /* 'limb' (LIMBU) */
331     NULL,            /* 'linb' (LINEAR_B) */
332     NULL,            /* 'osma' (OSMANYA) */
333     NULL,            /* 'shaw' (SHAVIAN) */
334     NULL,            /* 'tale' (TAI_LE) */
335     NULL,            /* 'ugar' (UGARITIC) */
336     NULL,            /* 'hrkt' (KATAKANA_OR_HIRAGANA) */
337     NULL,            /* 'bugi' (BUGINESE) */
338     NULL,            /* 'glag' (GLAGOLITIC) */
339     NULL,            /* 'khar' (KHAROSHTHI) */
340     NULL,            /* 'sylo' (SYLOTI_NAGRI) */
341     NULL,            /* 'talu' (NEW_TAI_LUE) */
342     NULL,            /* 'tfng' (TIFINAGH) */
343     NULL,            /* 'xpeo' (OLD_PERSIAN) */
344     NULL,            /* 'bali' (BALINESE) */
345     NULL,            /* 'batk' (BATK) */
346     NULL,            /* 'blis' (BLIS) */
347     NULL,            /* 'brah' (BRAH) */
348     NULL,            /* 'cham' (CHAM) */
349     NULL,            /* 'cirt' (CIRT) */
350     NULL,            /* 'cyrs' (CYRS) */
351     NULL,            /* 'egyd' (EGYD) */
352     NULL,            /* 'egyh' (EGYH) */
353     NULL,            /* 'egyp' (EGYP) */
354     NULL,            /* 'geok' (GEOK) */
355     NULL,            /* 'hans' (HANS) */
356     NULL,            /* 'hant' (HANT) */
357     NULL,            /* 'hmng' (HMNG) */
358     NULL,            /* 'hung' (HUNG) */
359     NULL,            /* 'inds' (INDS) */
360     NULL,            /* 'java' (JAVA) */
361     NULL,            /* 'kali' (KALI) */
362     NULL,            /* 'latf' (LATF) */
363     NULL,            /* 'latg' (LATG) */
364     NULL,            /* 'lepc' (LEPC) */
365     NULL,            /* 'lina' (LINA) */
366     NULL,            /* 'mand' (MAND) */
367     NULL,            /* 'maya' (MAYA) */
368     NULL,            /* 'mero' (MERO) */
369     NULL,            /* 'nko ' (NKO) */
370     NULL,            /* 'orkh' (ORKH) */
371     NULL,            /* 'perm' (PERM) */
372     NULL,            /* 'phag' (PHAGS_PA) */
373     NULL,            /* 'phnx' (PHOENICIAN) */
374     NULL,            /* 'plrd' (PLRD) */
375     NULL,            /* 'roro' (RORO) */
376     NULL,            /* 'sara' (SARA) */
377     NULL,            /* 'syre' (SYRE) */
378     NULL,            /* 'syrj' (SYRJ) */
379     NULL,            /* 'syrn' (SYRN) */
380     NULL,            /* 'teng' (TENG) */
381     NULL,            /* 'vai ' (VAII) */
382     NULL,            /* 'visp' (VISP) */
383     NULL,            /* 'xsux' (CUNEIFORM) */
384     NULL,            /* 'zxxx' (ZXXX) */
385     NULL,            /* 'zzzz' (UNKNOWN) */
386     NULL,            /* 'cari' (CARI) */
387     NULL,            /* 'jpan' (JPAN) */
388     NULL,            /* 'lana' (LANA) */
389     NULL,            /* 'lyci' (LYCI) */
390     NULL,            /* 'lydi' (LYDI) */
391     NULL,            /* 'olck' (OLCK) */
392     NULL,            /* 'rjng' (RJNG) */
393     NULL,            /* 'saur' (SAUR) */
394     NULL,            /* 'sgnw' (SGNW) */
395     NULL,            /* 'sund' (SUND) */
396     NULL,            /* 'moon' (MOON) */
397     NULL,            /* 'mtei' (MTEI) */
398     NULL,            /* 'armi' (ARMI) */
399     NULL,            /* 'avst' (AVST) */
400     NULL,            /* 'cakm' (CAKM) */
401     NULL,            /* 'kore' (KORE) */
402     NULL,            /* 'kthi' (KTHI) */
403     NULL,            /* 'mani' (MANI) */
404     NULL,            /* 'phli' (PHLI) */
405     NULL,            /* 'phlp' (PHLP) */
406     NULL,            /* 'phlv' (PHLV) */
407     NULL,            /* 'prti' (PRTI) */
408     NULL,            /* 'samr' (SAMR) */
409     NULL,            /* 'tavt' (TAVT) */
410     NULL,            /* 'zmth' (ZMTH) */
411     NULL,            /* 'zsym' (ZSYM) */
412     NULL,            /* 'bamu' (BAMUM) */
413     NULL,            /* 'lisu' (LISU) */
414     NULL,            /* 'nkgb' (NKGB) */
415     NULL             /* 'sarb' (OLD_SOUTH_ARABIAN) */
416 };
417 
getCharClass(LEUnicode ch) const418 IndicClassTable::CharClass IndicClassTable::getCharClass(LEUnicode ch) const
419 {
420     if (ch == C_SIGN_ZWJ) {
421         return CF_CONSONANT | CC_ZERO_WIDTH_MARK;
422     }
423 
424     if (ch == C_SIGN_ZWNJ) {
425         return CC_ZERO_WIDTH_MARK;
426     }
427 
428     if (ch < firstChar || ch > lastChar) {
429         return CC_RESERVED;
430     }
431 
432     return classTable[ch - firstChar];
433 }
434 
getScriptClassTable(le_int32 scriptCode)435 const IndicClassTable *IndicClassTable::getScriptClassTable(le_int32 scriptCode)
436 {
437     if (scriptCode < 0 || scriptCode >= scriptCodeCount) {
438         return NULL;
439     }
440 
441     return indicClassTables[scriptCode];
442 }
443 
getWorstCaseExpansion(le_int32 scriptCode)444 le_int32 IndicReordering::getWorstCaseExpansion(le_int32 scriptCode)
445 {
446     const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);
447 
448     if (classTable == NULL) {
449         return 1;
450     }
451 
452     return classTable->getWorstCaseExpansion();
453 }
454 
getFilterZeroWidth(le_int32 scriptCode)455 le_bool IndicReordering::getFilterZeroWidth(le_int32 scriptCode)
456 {
457     const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);
458 
459     if (classTable == NULL) {
460         return TRUE;
461     }
462 
463     return classTable->getFilterZeroWidth();
464 }
465 
466 U_NAMESPACE_END
467