1 /*
2 *
3 * (C) Copyright IBM Corp. 1998-2009 - All Rights Reserved
4 *
5 */
6
7 #ifndef __INDICREORDERING_H
8 #define __INDICREORDERING_H
9
10 /**
11 * \file
12 * \internal
13 */
14
15 #include "LETypes.h"
16 #include "OpenTypeTables.h"
17
18 U_NAMESPACE_BEGIN
19
20 // Characters that get refered to by name...
21 #define C_SIGN_ZWNJ 0x200C
22 #define C_SIGN_ZWJ 0x200D
23
24 // Character class values
25 #define CC_RESERVED 0U
26 #define CC_VOWEL_MODIFIER 1U
27 #define CC_STRESS_MARK 2U
28 #define CC_INDEPENDENT_VOWEL 3U
29 #define CC_INDEPENDENT_VOWEL_2 4U
30 #define CC_INDEPENDENT_VOWEL_3 5U
31 #define CC_CONSONANT 6U
32 #define CC_CONSONANT_WITH_NUKTA 7U
33 #define CC_NUKTA 8U
34 #define CC_DEPENDENT_VOWEL 9U
35 #define CC_SPLIT_VOWEL_PIECE_1 10U
36 #define CC_SPLIT_VOWEL_PIECE_2 11U
37 #define CC_SPLIT_VOWEL_PIECE_3 12U
38 #define CC_VIRAMA 13U
39 #define CC_ZERO_WIDTH_MARK 14U
40 #define CC_AL_LAKUNA 15U
41 #define CC_COUNT 16U
42
43 // Character class flags
44 #define CF_CLASS_MASK 0x0000FFFFU
45
46 #define CF_CONSONANT 0x80000000U
47
48 #define CF_REPH 0x40000000U
49 #define CF_VATTU 0x20000000U
50 #define CF_BELOW_BASE 0x10000000U
51 #define CF_POST_BASE 0x08000000U
52 #define CF_LENGTH_MARK 0x04000000U
53 #define CF_PRE_BASE 0x02000000U
54
55 #define CF_POS_BEFORE 0x00300000U
56 #define CF_POS_BELOW 0x00200000U
57 #define CF_POS_ABOVE 0x00100000U
58 #define CF_POS_AFTER 0x00000000U
59 #define CF_POS_MASK 0x00300000U
60
61 #define CF_INDEX_MASK 0x000F0000U
62 #define CF_INDEX_SHIFT 16
63
64 // Script flag bits
65 #define SF_MATRAS_AFTER_BASE 0x80000000U
66 #define SF_REPH_AFTER_BELOW 0x40000000U
67 #define SF_EYELASH_RA 0x20000000U
68 #define SF_MPRE_FIXUP 0x10000000U
69 #define SF_FILTER_ZERO_WIDTH 0x08000000U
70
71 #define SF_POST_BASE_LIMIT_MASK 0x0000FFFFU
72 #define SF_NO_POST_BASE_LIMIT 0x00007FFFU
73
74 typedef LEUnicode SplitMatra[3];
75
76 class MPreFixups;
77 class LEGlyphStorage;
78
79 // Dynamic Properties ( v2 fonts only )
80 typedef le_uint32 DynamicProperties;
81
82 #define DP_REPH 0x80000000U
83 #define DP_HALF 0x40000000U
84 #define DP_PREF 0x20000000U
85 #define DP_BLWF 0x10000000U
86 #define DP_PSTF 0x08000000U
87
88 struct IndicClassTable
89 {
90 typedef le_uint32 CharClass;
91 typedef le_uint32 ScriptFlags;
92
93 LEUnicode firstChar;
94 LEUnicode lastChar;
95 le_int32 worstCaseExpansion;
96 ScriptFlags scriptFlags;
97 const CharClass *classTable;
98 const SplitMatra *splitMatraTable;
99
100 inline le_int32 getWorstCaseExpansion() const;
101 inline le_bool getFilterZeroWidth() const;
102
103 CharClass getCharClass(LEUnicode ch) const;
104
105 inline const SplitMatra *getSplitMatra(CharClass charClass) const;
106
107 inline le_bool isVowelModifier(LEUnicode ch) const;
108 inline le_bool isStressMark(LEUnicode ch) const;
109 inline le_bool isConsonant(LEUnicode ch) const;
110 inline le_bool isReph(LEUnicode ch) const;
111 inline le_bool isVirama(LEUnicode ch) const;
112 inline le_bool isAlLakuna(LEUnicode ch) const;
113 inline le_bool isNukta(LEUnicode ch) const;
114 inline le_bool isVattu(LEUnicode ch) const;
115 inline le_bool isMatra(LEUnicode ch) const;
116 inline le_bool isSplitMatra(LEUnicode ch) const;
117 inline le_bool isLengthMark(LEUnicode ch) const;
118 inline le_bool hasPostOrBelowBaseForm(LEUnicode ch) const;
119 inline le_bool hasPostBaseForm(LEUnicode ch) const;
120 inline le_bool hasBelowBaseForm(LEUnicode ch) const;
121 inline le_bool hasAboveBaseForm(LEUnicode ch) const;
122 inline le_bool hasPreBaseForm(LEUnicode ch) const;
123
124 inline static le_bool isVowelModifier(CharClass charClass);
125 inline static le_bool isStressMark(CharClass charClass);
126 inline static le_bool isConsonant(CharClass charClass);
127 inline static le_bool isReph(CharClass charClass);
128 inline static le_bool isVirama(CharClass charClass);
129 inline static le_bool isAlLakuna(CharClass charClass);
130 inline static le_bool isNukta(CharClass charClass);
131 inline static le_bool isVattu(CharClass charClass);
132 inline static le_bool isMatra(CharClass charClass);
133 inline static le_bool isSplitMatra(CharClass charClass);
134 inline static le_bool isLengthMark(CharClass charClass);
135 inline static le_bool hasPostOrBelowBaseForm(CharClass charClass);
136 inline static le_bool hasPostBaseForm(CharClass charClass);
137 inline static le_bool hasBelowBaseForm(CharClass charClass);
138 inline static le_bool hasAboveBaseForm(CharClass charClass);
139 inline static le_bool hasPreBaseForm(CharClass charClass);
140
141 static const IndicClassTable *getScriptClassTable(le_int32 scriptCode);
142 };
143
144 class IndicReordering /* not : public UObject because all methods are static */ {
145 public:
146 static le_int32 getWorstCaseExpansion(le_int32 scriptCode);
147
148 static le_bool getFilterZeroWidth(le_int32 scriptCode);
149
150 static le_int32 reorder(const LEUnicode *theChars, le_int32 charCount, le_int32 scriptCode,
151 LEUnicode *outChars, LEGlyphStorage &glyphStorage,
152 MPreFixups **outMPreFixups, LEErrorCode& success);
153
154 static void adjustMPres(MPreFixups *mpreFixups, LEGlyphStorage &glyphStorage, LEErrorCode& success);
155
156 static le_int32 v2process(const LEUnicode *theChars, le_int32 charCount, le_int32 scriptCode,
157 LEUnicode *outChars, LEGlyphStorage &glyphStorage);
158
159 static const FeatureMap *getFeatureMap(le_int32 &count);
160
161 static const FeatureMap *getv2FeatureMap(le_int32 &count);
162
163 static void applyPresentationForms(LEGlyphStorage &glyphStorage, le_int32 count);
164
165 static void finalReordering(LEGlyphStorage &glyphStorage, le_int32 count);
166
167 static void getDynamicProperties(DynamicProperties *dProps, const IndicClassTable *classTable);
168
169 private:
170 // do not instantiate
171 IndicReordering();
172
173 static le_int32 findSyllable(const IndicClassTable *classTable, const LEUnicode *chars, le_int32 prev, le_int32 charCount);
174
175 };
176
getWorstCaseExpansion()177 inline le_int32 IndicClassTable::getWorstCaseExpansion() const
178 {
179 return worstCaseExpansion;
180 }
181
getFilterZeroWidth()182 inline le_bool IndicClassTable::getFilterZeroWidth() const
183 {
184 return (scriptFlags & SF_FILTER_ZERO_WIDTH) != 0;
185 }
186
getSplitMatra(CharClass charClass)187 inline const SplitMatra *IndicClassTable::getSplitMatra(CharClass charClass) const
188 {
189 le_int32 index = (charClass & CF_INDEX_MASK) >> CF_INDEX_SHIFT;
190
191 return &splitMatraTable[index - 1];
192 }
193
isVowelModifier(CharClass charClass)194 inline le_bool IndicClassTable::isVowelModifier(CharClass charClass)
195 {
196 return (charClass & CF_CLASS_MASK) == CC_VOWEL_MODIFIER;
197 }
198
isStressMark(CharClass charClass)199 inline le_bool IndicClassTable::isStressMark(CharClass charClass)
200 {
201 return (charClass & CF_CLASS_MASK) == CC_STRESS_MARK;
202 }
203
isConsonant(CharClass charClass)204 inline le_bool IndicClassTable::isConsonant(CharClass charClass)
205 {
206 return (charClass & CF_CONSONANT) != 0;
207 }
208
isReph(CharClass charClass)209 inline le_bool IndicClassTable::isReph(CharClass charClass)
210 {
211 return (charClass & CF_REPH) != 0;
212 }
213
isNukta(CharClass charClass)214 inline le_bool IndicClassTable::isNukta(CharClass charClass)
215 {
216 return (charClass & CF_CLASS_MASK) == CC_NUKTA;
217 }
218
isVirama(CharClass charClass)219 inline le_bool IndicClassTable::isVirama(CharClass charClass)
220 {
221 return (charClass & CF_CLASS_MASK) == CC_VIRAMA;
222 }
223
isAlLakuna(CharClass charClass)224 inline le_bool IndicClassTable::isAlLakuna(CharClass charClass)
225 {
226 return (charClass & CF_CLASS_MASK) == CC_AL_LAKUNA;
227 }
228
isVattu(CharClass charClass)229 inline le_bool IndicClassTable::isVattu(CharClass charClass)
230 {
231 return (charClass & CF_VATTU) != 0;
232 }
233
isMatra(CharClass charClass)234 inline le_bool IndicClassTable::isMatra(CharClass charClass)
235 {
236 charClass &= CF_CLASS_MASK;
237
238 return charClass >= CC_DEPENDENT_VOWEL && charClass <= CC_SPLIT_VOWEL_PIECE_3;
239 }
240
isSplitMatra(CharClass charClass)241 inline le_bool IndicClassTable::isSplitMatra(CharClass charClass)
242 {
243 return (charClass & CF_INDEX_MASK) != 0;
244 }
245
isLengthMark(CharClass charClass)246 inline le_bool IndicClassTable::isLengthMark(CharClass charClass)
247 {
248 return (charClass & CF_LENGTH_MARK) != 0;
249 }
250
hasPostOrBelowBaseForm(CharClass charClass)251 inline le_bool IndicClassTable::hasPostOrBelowBaseForm(CharClass charClass)
252 {
253 return (charClass & (CF_POST_BASE | CF_BELOW_BASE)) != 0;
254 }
255
hasPostBaseForm(CharClass charClass)256 inline le_bool IndicClassTable::hasPostBaseForm(CharClass charClass)
257 {
258 return (charClass & CF_POST_BASE) != 0;
259 }
260
hasPreBaseForm(CharClass charClass)261 inline le_bool IndicClassTable::hasPreBaseForm(CharClass charClass)
262 {
263 return (charClass & CF_PRE_BASE) != 0;
264 }
265
hasBelowBaseForm(CharClass charClass)266 inline le_bool IndicClassTable::hasBelowBaseForm(CharClass charClass)
267 {
268 return (charClass & CF_BELOW_BASE) != 0;
269 }
270
hasAboveBaseForm(CharClass charClass)271 inline le_bool IndicClassTable::hasAboveBaseForm(CharClass charClass)
272 {
273 return ((charClass & CF_POS_MASK) == CF_POS_ABOVE);
274 }
275
isVowelModifier(LEUnicode ch)276 inline le_bool IndicClassTable::isVowelModifier(LEUnicode ch) const
277 {
278 return isVowelModifier(getCharClass(ch));
279 }
280
isStressMark(LEUnicode ch)281 inline le_bool IndicClassTable::isStressMark(LEUnicode ch) const
282 {
283 return isStressMark(getCharClass(ch));
284 }
285
isConsonant(LEUnicode ch)286 inline le_bool IndicClassTable::isConsonant(LEUnicode ch) const
287 {
288 return isConsonant(getCharClass(ch));
289 }
290
isReph(LEUnicode ch)291 inline le_bool IndicClassTable::isReph(LEUnicode ch) const
292 {
293 return isReph(getCharClass(ch));
294 }
295
isVirama(LEUnicode ch)296 inline le_bool IndicClassTable::isVirama(LEUnicode ch) const
297 {
298 return isVirama(getCharClass(ch));
299 }
300
isAlLakuna(LEUnicode ch)301 inline le_bool IndicClassTable::isAlLakuna(LEUnicode ch) const
302 {
303 return isAlLakuna(getCharClass(ch));
304 }
305
isNukta(LEUnicode ch)306 inline le_bool IndicClassTable::isNukta(LEUnicode ch) const
307 {
308 return isNukta(getCharClass(ch));
309 }
310
isVattu(LEUnicode ch)311 inline le_bool IndicClassTable::isVattu(LEUnicode ch) const
312 {
313 return isVattu(getCharClass(ch));
314 }
315
isMatra(LEUnicode ch)316 inline le_bool IndicClassTable::isMatra(LEUnicode ch) const
317 {
318 return isMatra(getCharClass(ch));
319 }
320
isSplitMatra(LEUnicode ch)321 inline le_bool IndicClassTable::isSplitMatra(LEUnicode ch) const
322 {
323 return isSplitMatra(getCharClass(ch));
324 }
325
isLengthMark(LEUnicode ch)326 inline le_bool IndicClassTable::isLengthMark(LEUnicode ch) const
327 {
328 return isLengthMark(getCharClass(ch));
329 }
330
hasPostOrBelowBaseForm(LEUnicode ch)331 inline le_bool IndicClassTable::hasPostOrBelowBaseForm(LEUnicode ch) const
332 {
333 return hasPostOrBelowBaseForm(getCharClass(ch));
334 }
335
hasPostBaseForm(LEUnicode ch)336 inline le_bool IndicClassTable::hasPostBaseForm(LEUnicode ch) const
337 {
338 return hasPostBaseForm(getCharClass(ch));
339 }
340
hasBelowBaseForm(LEUnicode ch)341 inline le_bool IndicClassTable::hasBelowBaseForm(LEUnicode ch) const
342 {
343 return hasBelowBaseForm(getCharClass(ch));
344 }
345
hasPreBaseForm(LEUnicode ch)346 inline le_bool IndicClassTable::hasPreBaseForm(LEUnicode ch) const
347 {
348 return hasPreBaseForm(getCharClass(ch));
349 }
350
hasAboveBaseForm(LEUnicode ch)351 inline le_bool IndicClassTable::hasAboveBaseForm(LEUnicode ch) const
352 {
353 return hasAboveBaseForm(getCharClass(ch));
354 }
355 U_NAMESPACE_END
356 #endif
357