2 ******************************************************************************
4 * Copyright (C) 2000-2013, International Business Machines
5 * Corporation and others. All Rights Reserved.
7 ******************************************************************************
8 * file name: ushape.cpp
10 * tab size: 8 (not used)
13 * created on: 2000jun29
14 * created by: Markus W. Scherer
16 * Arabic letter shaping implemented by Ayman Roshdy
19 #include "unicode/utypes.h"
20 #include "unicode/uchar.h"
21 #include "unicode/ustring.h"
22 #include "unicode/ushape.h"
26 #include "ubidi_props.h"
29 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
32 * This implementation is designed for 16-bit Unicode strings.
33 * The main assumption is that the Arabic characters and their
34 * presentation forms each fit into a single UChar.
35 * With UTF-8, they occupy 2 or 3 bytes, and more than the ASCII
40 * ### TODO in general for letter shaping:
41 * - the letter shaping code is UTF-16-unaware; needs update
42 * + especially invertBuffer()?!
43 * - needs to handle the "Arabic Tail" that is used in some legacy codepages
44 * as a glyph fragment of wide-glyph letters
45 * + IBM Unicode conversion tables map it to U+200B (ZWSP)
46 * + IBM Egypt has proposed to encode the tail in Unicode among Arabic Presentation Forms
47 * + Unicode 3.2 added U+FE73 ARABIC TAIL FRAGMENT
50 /* definitions for Arabic letter shaping ------------------------------------ */
60 #define COMBINE (SHADDA+CSHADDA)
62 #define HAMZAFE_CHAR 0xfe80
63 #define HAMZA06_CHAR 0x0621
64 #define YEH_HAMZA_CHAR 0x0626
65 #define YEH_HAMZAFE_CHAR 0xFE89
66 #define LAMALEF_SPACE_SUB 0xFFFF
67 #define TASHKEEL_SPACE_SUB 0xFFFE
68 #define NEW_TAIL_CHAR 0xFE73
69 #define OLD_TAIL_CHAR 0x200B
70 #define LAM_CHAR 0x0644
71 #define SPACE_CHAR 0x0020
72 #define SHADDA_CHAR 0xFE7C
73 #define TATWEEL_CHAR 0x0640
74 #define SHADDA_TATWEEL_CHAR 0xFE7D
75 #define SHADDA06_CHAR 0x0651
78 #define DESHAPE_MODE 1
80 struct uShapeVariables {
82 uint32_t uShapeLamalefBegin;
83 uint32_t uShapeLamalefEnd;
84 uint32_t uShapeTashkeelBegin;
85 uint32_t uShapeTashkeelEnd;
86 int spacesRelativeToTextBeginEnd;
89 static const uint8_t tailFamilyIsolatedFinal[] = {
106 static const uint8_t tashkeelMedial[] = {
125 static const UChar yehHamzaToYeh[] =
127 /* isolated*/ 0xFEEF,
131 static const uint8_t IrrelevantPos[] = {
137 static const UChar convertLamAlef[] =
149 static const UChar araLink[178]=
151 1 + 32 + 256 * 0x11,/*0x0622*/
152 1 + 32 + 256 * 0x13,/*0x0623*/
153 1 + 256 * 0x15,/*0x0624*/
154 1 + 32 + 256 * 0x17,/*0x0625*/
155 1 + 2 + 256 * 0x19,/*0x0626*/
156 1 + 32 + 256 * 0x1D,/*0x0627*/
157 1 + 2 + 256 * 0x1F,/*0x0628*/
158 1 + 256 * 0x23,/*0x0629*/
159 1 + 2 + 256 * 0x25,/*0x062A*/
160 1 + 2 + 256 * 0x29,/*0x062B*/
161 1 + 2 + 256 * 0x2D,/*0x062C*/
162 1 + 2 + 256 * 0x31,/*0x062D*/
163 1 + 2 + 256 * 0x35,/*0x062E*/
164 1 + 256 * 0x39,/*0x062F*/
165 1 + 256 * 0x3B,/*0x0630*/
166 1 + 256 * 0x3D,/*0x0631*/
167 1 + 256 * 0x3F,/*0x0632*/
168 1 + 2 + 256 * 0x41,/*0x0633*/
169 1 + 2 + 256 * 0x45,/*0x0634*/
170 1 + 2 + 256 * 0x49,/*0x0635*/
171 1 + 2 + 256 * 0x4D,/*0x0636*/
172 1 + 2 + 256 * 0x51,/*0x0637*/
173 1 + 2 + 256 * 0x55,/*0x0638*/
174 1 + 2 + 256 * 0x59,/*0x0639*/
175 1 + 2 + 256 * 0x5D,/*0x063A*/
176 0, 0, 0, 0, 0, /*0x063B-0x063F*/
178 1 + 2 + 256 * 0x61,/*0x0641*/
179 1 + 2 + 256 * 0x65,/*0x0642*/
180 1 + 2 + 256 * 0x69,/*0x0643*/
181 1 + 2 + 16 + 256 * 0x6D,/*0x0644*/
182 1 + 2 + 256 * 0x71,/*0x0645*/
183 1 + 2 + 256 * 0x75,/*0x0646*/
184 1 + 2 + 256 * 0x79,/*0x0647*/
185 1 + 256 * 0x7D,/*0x0648*/
186 1 + 256 * 0x7F,/*0x0649*/
187 1 + 2 + 256 * 0x81,/*0x064A*/
188 4 + 256 * 1, /*0x064B*/
189 4 + 128 + 256 * 1, /*0x064C*/
190 4 + 128 + 256 * 1, /*0x064D*/
191 4 + 128 + 256 * 1, /*0x064E*/
192 4 + 128 + 256 * 1, /*0x064F*/
193 4 + 128 + 256 * 1, /*0x0650*/
194 4 + 64 + 256 * 3, /*0x0651*/
195 4 + 256 * 1, /*0x0652*/
196 4 + 256 * 7, /*0x0653*/
197 4 + 256 * 8, /*0x0654*/
198 4 + 256 * 8, /*0x0655*/
199 4 + 256 * 1, /*0x0656*/
200 0, 0, 0, 0, 0, /*0x0657-0x065B*/
201 1 + 256 * 0x85,/*0x065C*/
202 1 + 256 * 0x87,/*0x065D*/
203 1 + 256 * 0x89,/*0x065E*/
204 1 + 256 * 0x8B,/*0x065F*/
205 0, 0, 0, 0, 0, /*0x0660-0x0664*/
206 0, 0, 0, 0, 0, /*0x0665-0x0669*/
207 0, 0, 0, 0, 0, 0, /*0x066A-0x066F*/
208 4 + 256 * 6, /*0x0670*/
209 1 + 8 + 256 * 0x00,/*0x0671*/
214 1, 1, /*0x0676-0x0677*/
215 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x0678-0x067D*/
216 1+2+8+256 * 0x06, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x067E-0x0683*/
217 1+2, 1+2, 1+2+8+256 * 0x2A, 1+2, /*0x0684-0x0687*/
218 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /*0x0688-0x0691*/
219 1, 1, 1, 1, 1, 1, 1+8+256 * 0x3A, 1, /*0x0692-0x0699*/
220 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x069A-0x06A3*/
221 1+2, 1+2, 1+2, 1+2, /*0x069A-0x06A3*/
222 1+2, 1+2, 1+2, 1+2, 1+2, 1+2+8+256 * 0x3E, /*0x06A4-0x06AD*/
223 1+2, 1+2, 1+2, 1+2, /*0x06A4-0x06AD*/
224 1+2, 1+2+8+256 * 0x42, 1+2, 1+2, 1+2, 1+2, /*0x06AE-0x06B7*/
225 1+2, 1+2, 1+2, 1+2, /*0x06AE-0x06B7*/
226 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x06B8-0x06BF*/
227 1+2, 1+2, /*0x06B8-0x06BF*/
230 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /*0x06C2-0x06CB*/
231 1+2+8+256 * 0xAC, /*0x06CC*/
233 1+2, 1+2, 1+2, 1+2, /*0x06CE-0x06D1*/
234 1, 1 /*0x06D2-0x06D3*/
237 static const uint8_t presALink[] = {
238 /***********0*****1*****2*****3*****4*****5*****6*****7*****8*****9*****A*****B*****C*****D*****E*****F*/
239 /*FB5*/ 0, 1, 0, 0, 0, 0, 0, 1, 2,1 + 2, 0, 0, 0, 0, 0, 0,
240 /*FB6*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
241 /*FB7*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2,1 + 2, 0, 0,
242 /*FB8*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1,
243 /*FB9*/ 2,1 + 2, 0, 1, 2,1 + 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
244 /*FBA*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
245 /*FBB*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
246 /*FBC*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
247 /*FBD*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
248 /*FBE*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
249 /*FBF*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2,1 + 2,
250 /*FC0*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
251 /*FC1*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
252 /*FC2*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
253 /*FC3*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
254 /*FC4*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
255 /*FC5*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4, 4,
259 static const uint8_t presBLink[]=
261 /***********0*****1*****2*****3*****4*****5*****6*****7*****8*****9*****A*****B*****C*****D*****E*****F*/
262 /*FE7*/1 + 2,1 + 2,1 + 2, 0,1 + 2, 0,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,
263 /*FE8*/ 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 2,1 + 2, 0, 1, 0,
264 /*FE9*/ 1, 2,1 + 2, 0, 1, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
265 /*FEA*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 0, 1, 0, 1, 0,
266 /*FEB*/ 1, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
267 /*FEC*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
268 /*FED*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
269 /*FEE*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 0,
270 /*FEF*/ 1, 0, 1, 2,1 + 2, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0
273 static const UChar convertFBto06[] =
275 /***********0******1******2******3******4******5******6******7******8******9******A******B******C******D******E******F***/
276 /*FB5*/ 0x671, 0x671, 0, 0, 0, 0, 0x67E, 0x67E, 0x67E, 0x67E, 0, 0, 0, 0, 0, 0,
277 /*FB6*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
278 /*FB7*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x686, 0x686, 0x686, 0x686, 0, 0,
279 /*FB8*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x698, 0x698, 0, 0, 0x6A9, 0x6A9,
280 /*FB9*/ 0x6A9, 0x6A9, 0x6AF, 0x6AF, 0x6AF, 0x6AF, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
281 /*FBA*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
282 /*FBB*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
283 /*FBC*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
284 /*FBD*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
285 /*FBE*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
286 /*FBF*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x6CC, 0x6CC, 0x6CC, 0x6CC
289 static const UChar convertFEto06[] =
291 /***********0******1******2******3******4******5******6******7******8******9******A******B******C******D******E******F***/
292 /*FE7*/ 0x64B, 0x64B, 0x64C, 0x64C, 0x64D, 0x64D, 0x64E, 0x64E, 0x64F, 0x64F, 0x650, 0x650, 0x651, 0x651, 0x652, 0x652,
293 /*FE8*/ 0x621, 0x622, 0x622, 0x623, 0x623, 0x624, 0x624, 0x625, 0x625, 0x626, 0x626, 0x626, 0x626, 0x627, 0x627, 0x628,
294 /*FE9*/ 0x628, 0x628, 0x628, 0x629, 0x629, 0x62A, 0x62A, 0x62A, 0x62A, 0x62B, 0x62B, 0x62B, 0x62B, 0x62C, 0x62C, 0x62C,
295 /*FEA*/ 0x62C, 0x62D, 0x62D, 0x62D, 0x62D, 0x62E, 0x62E, 0x62E, 0x62E, 0x62F, 0x62F, 0x630, 0x630, 0x631, 0x631, 0x632,
296 /*FEB*/ 0x632, 0x633, 0x633, 0x633, 0x633, 0x634, 0x634, 0x634, 0x634, 0x635, 0x635, 0x635, 0x635, 0x636, 0x636, 0x636,
297 /*FEC*/ 0x636, 0x637, 0x637, 0x637, 0x637, 0x638, 0x638, 0x638, 0x638, 0x639, 0x639, 0x639, 0x639, 0x63A, 0x63A, 0x63A,
298 /*FED*/ 0x63A, 0x641, 0x641, 0x641, 0x641, 0x642, 0x642, 0x642, 0x642, 0x643, 0x643, 0x643, 0x643, 0x644, 0x644, 0x644,
299 /*FEE*/ 0x644, 0x645, 0x645, 0x645, 0x645, 0x646, 0x646, 0x646, 0x646, 0x647, 0x647, 0x647, 0x647, 0x648, 0x648, 0x649,
300 /*FEF*/ 0x649, 0x64A, 0x64A, 0x64A, 0x64A, 0x65C, 0x65C, 0x65D, 0x65D, 0x65E, 0x65E, 0x65F, 0x65F
303 static const uint8_t shapeTable[4][4][4]=
305 { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,1} },
306 { {0,0,2,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} },
307 { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,3} },
308 { {0,0,1,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} }
312 * This function shapes European digits to Arabic-Indic digits
313 * in-place, writing over the input characters.
314 * Since we know that we are only looking for BMP code points,
315 * we can safely just work with code units (again, at least UTF-16).
318 _shapeToArabicDigitsWithContext(UChar *s, int32_t length,
320 UBool isLogical, UBool lastStrongWasAL) {
321 const UBiDiProps *bdp;
325 bdp=ubidi_getSingleton();
328 /* the iteration direction depends on the type of input */
330 for(i=0; i<length; ++i) {
332 switch(ubidi_getClass(bdp, c)) {
333 case U_LEFT_TO_RIGHT: /* L */
334 case U_RIGHT_TO_LEFT: /* R */
335 lastStrongWasAL=FALSE;
337 case U_RIGHT_TO_LEFT_ARABIC: /* AL */
338 lastStrongWasAL=TRUE;
340 case U_EUROPEAN_NUMBER: /* EN */
341 if(lastStrongWasAL && (uint32_t)(c-0x30)<10) {
342 s[i]=(UChar)(digitBase+c); /* digitBase+(c-0x30) - digitBase was modified above */
350 for(i=length; i>0; /* pre-decrement in the body */) {
352 switch(ubidi_getClass(bdp, c)) {
353 case U_LEFT_TO_RIGHT: /* L */
354 case U_RIGHT_TO_LEFT: /* R */
355 lastStrongWasAL=FALSE;
357 case U_RIGHT_TO_LEFT_ARABIC: /* AL */
358 lastStrongWasAL=TRUE;
360 case U_EUROPEAN_NUMBER: /* EN */
361 if(lastStrongWasAL && (uint32_t)(c-0x30)<10) {
362 s[i]=(UChar)(digitBase+c); /* digitBase+(c-0x30) - digitBase was modified above */
374 *Function : This function inverts the buffer, it's used
375 * in case the user specifies the buffer to be
376 * U_SHAPE_TEXT_DIRECTION_LOGICAL
379 invertBuffer(UChar *buffer, int32_t size, uint32_t /*options*/, int32_t lowlimit, int32_t highlimit) {
382 for(i=lowlimit,j=size-highlimit-1;i<j;i++,j--) {
384 buffer[i] = buffer[j];
390 *Name : changeLamAlef
391 *Function : Converts the Alef characters into an equivalent
392 * LamAlef location in the 0x06xx Range, this is an
393 * intermediate stage in the operation of the program
394 * later it'll be converted into the 0xFExx LamAlefs
395 * in the shaping function.
398 changeLamAlef(UChar ch) {
414 *Function : Resolves the link between the characters as
415 * Arabic characters have four forms :
416 * Isolated, Initial, Middle and Final Form
420 if(ch >= 0x0622 && ch <= 0x06D3) {
421 return(araLink[ch-0x0622]);
422 } else if(ch == 0x200D) {
424 } else if(ch >= 0x206D && ch <= 0x206F) {
426 }else if(ch >= 0xFB50 && ch <= 0xFC62) {
427 return(presALink[ch-0xFB50]);
428 } else if(ch >= 0xFE70 && ch <= 0xFEFC) {
429 return(presBLink[ch-0xFE70]);
437 *Function : Counts the number of spaces
438 * at each end of the logical buffer
441 countSpaces(UChar *dest, int32_t size, uint32_t /*options*/, int32_t *spacesCountl, int32_t *spacesCountr) {
443 int32_t countl = 0,countr = 0;
444 while((dest[i] == SPACE_CHAR) && (countl < size)) {
448 if (countl < size) { /* the entire buffer is not all space */
449 while(dest[size-1] == SPACE_CHAR) {
454 *spacesCountl = countl;
455 *spacesCountr = countr;
459 *Name : isTashkeelChar
460 *Function : Returns 1 for Tashkeel characters in 06 range else return 0
462 static inline int32_t
463 isTashkeelChar(UChar ch) {
464 return (int32_t)( ch>=0x064B && ch<= 0x0652 );
468 *Name : isTashkeelCharFE
469 *Function : Returns 1 for Tashkeel characters in FE range else return 0
471 static inline int32_t
472 isTashkeelCharFE(UChar ch) {
473 return (int32_t)( ch>=0xFE70 && ch<= 0xFE7F );
478 *Function : Returns 1 for Alef characters else return 0
480 static inline int32_t
481 isAlefChar(UChar ch) {
482 return (int32_t)( (ch==0x0622)||(ch==0x0623)||(ch==0x0625)||(ch==0x0627) );
486 *Name : isLamAlefChar
487 *Function : Returns 1 for LamAlef characters else return 0
489 static inline int32_t
490 isLamAlefChar(UChar ch) {
491 return (int32_t)((ch>=0xFEF5)&&(ch<=0xFEFC) );
496 *Function : returns 1 if the character matches one of the tail characters (0xfe73 or 0x200b) otherwise returns 0
499 static inline int32_t
500 isTailChar(UChar ch) {
501 if(ch == OLD_TAIL_CHAR || ch == NEW_TAIL_CHAR){
509 *Name : isSeenTailFamilyChar
510 *Function : returns 1 if the character is a seen family isolated character
511 * in the FE range otherwise returns 0
514 static inline int32_t
515 isSeenTailFamilyChar(UChar ch) {
516 if(ch >= 0xfeb1 && ch < 0xfebf){
517 return tailFamilyIsolatedFinal [ch - 0xFEB1];
523 /* Name : isSeenFamilyChar
524 * Function : returns 1 if the character is a seen family character in the Unicode
525 * 06 range otherwise returns 0
528 static inline int32_t
529 isSeenFamilyChar(UChar ch){
530 if(ch >= 0x633 && ch <= 0x636){
539 *Name : isAlefMaksouraChar
540 *Function : returns 1 if the character is a Alef Maksoura Final or isolated
541 * otherwise returns 0
543 static inline int32_t
544 isAlefMaksouraChar(UChar ch) {
545 return (int32_t)( (ch == 0xFEEF) || ( ch == 0xFEF0) || (ch == 0x0649));
549 * Name : isYehHamzaChar
550 * Function : returns 1 if the character is a yehHamza isolated or yehhamza
551 * final is found otherwise returns 0
553 static inline int32_t
554 isYehHamzaChar(UChar ch) {
555 if((ch==0xFE89)||(ch==0xFE8A)){
563 * Name: isTashkeelOnTatweelChar
564 * Function: Checks if the Tashkeel Character is on Tatweel or not,if the
565 * Tashkeel on tatweel (FE range), it returns 1 else if the
566 * Tashkeel with shadda on tatweel (FC range)return 2 otherwise
569 static inline int32_t
570 isTashkeelOnTatweelChar(UChar ch){
571 if(ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75 && ch != SHADDA_TATWEEL_CHAR)
573 return tashkeelMedial [ch - 0xFE70];
574 }else if( (ch >= 0xfcf2 && ch <= 0xfcf4) || (ch == SHADDA_TATWEEL_CHAR)) {
582 * Name: isIsolatedTashkeelChar
583 * Function: Checks if the Tashkeel Character is in the isolated form
584 * (i.e. Unicode FE range) returns 1 else if the Tashkeel
585 * with shadda is in the isolated form (i.e. Unicode FC range)
586 * returns 2 otherwise returns 0
588 static inline int32_t
589 isIsolatedTashkeelChar(UChar ch){
590 if(ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75){
591 return (1 - tashkeelMedial [ch - 0xFE70]);
592 }else if(ch >= 0xfc5e && ch <= 0xfc63){
603 *Name : calculateSize
604 *Function : This function calculates the destSize to be used in preflighting
605 * when the destSize is equal to 0
606 * It is used also to calculate the new destsize in case the
607 * destination buffer will be resized.
611 calculateSize(const UChar *source, int32_t sourceLength,
612 int32_t destSize,uint32_t options) {
615 int lamAlefOption = 0;
616 int tashkeelOption = 0;
618 destSize = sourceLength;
620 if (((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE ||
621 ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED )) &&
622 ((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE )){
625 if((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE &&
626 ((options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_RESIZE ) ){
630 if(lamAlefOption || tashkeelOption){
631 if((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_VISUAL_LTR) {
632 for(i=0;i<sourceLength;i++) {
633 if( ((isAlefChar(source[i]))&& (i<(sourceLength-1)) &&(source[i+1] == LAM_CHAR)) || (isTashkeelCharFE(source[i])) ) {
637 }else if((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_LOGICAL) {
638 for(i=0;i<sourceLength;i++) {
639 if( ( (source[i] == LAM_CHAR) && (i<(sourceLength-1)) && (isAlefChar(source[i+1]))) || (isTashkeelCharFE(source[i])) ) {
646 if ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_UNSHAPE){
647 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE){
648 for(i=0;i<sourceLength;i++) {
649 if(isLamAlefChar(source[i]))
659 *Name : handleTashkeelWithTatweel
660 *Function : Replaces Tashkeel as following:
661 * Case 1 :if the Tashkeel on tatweel, replace it with Tatweel.
662 * Case 2 :if the Tashkeel aggregated with Shadda on Tatweel, replace
663 * it with Shadda on Tatweel.
664 * Case 3: if the Tashkeel is isolated replace it with Space.
668 handleTashkeelWithTatweel(UChar *dest, int32_t sourceLength,
669 int32_t /*destSize*/, uint32_t /*options*/,
670 UErrorCode * /*pErrorCode*/) {
672 for(i = 0; i < sourceLength; i++){
673 if((isTashkeelOnTatweelChar(dest[i]) == 1)){
674 dest[i] = TATWEEL_CHAR;
675 }else if((isTashkeelOnTatweelChar(dest[i]) == 2)){
676 dest[i] = SHADDA_TATWEEL_CHAR;
677 }else if(isIsolatedTashkeelChar(dest[i]) && dest[i] != SHADDA_CHAR){
678 dest[i] = SPACE_CHAR;
687 *Name : handleGeneratedSpaces
688 *Function : The shapeUnicode function converts Lam + Alef into LamAlef + space,
689 * and Tashkeel to space.
690 * handleGeneratedSpaces function puts these generated spaces
691 * according to the options the user specifies. LamAlef and Tashkeel
692 * spaces can be replaced at begin, at end, at near or decrease the
695 * There is also Auto option for LamAlef and tashkeel, which will put
696 * the spaces at end of the buffer (or end of text if the user used
697 * the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END).
699 * If the text type was visual_LTR and the option
700 * U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END was selected the END
701 * option will place the space at the beginning of the buffer and
702 * BEGIN will place the space at the end of the buffer.
706 handleGeneratedSpaces(UChar *dest, int32_t sourceLength,
709 UErrorCode *pErrorCode,struct uShapeVariables shapeVars ) {
711 int32_t i = 0, j = 0;
713 UChar *tempbuffer=NULL;
715 int lamAlefOption = 0;
716 int tashkeelOption = 0;
717 int shapingMode = SHAPE_MODE;
719 if (shapingMode == 0){
720 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE ){
723 if ( (options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_RESIZE ){
728 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
730 if(tempbuffer == NULL) {
731 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
736 if (lamAlefOption || tashkeelOption){
737 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
739 i = j = 0; count = 0;
740 while(i < sourceLength) {
741 if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
742 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
746 tempbuffer[j] = dest[i];
753 tempbuffer[i] = 0x0000;
758 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
759 destSize = u_strlen(dest);
764 if (shapingMode == 0){
765 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_NEAR ){
771 /* Lam+Alef is already shaped into LamAlef + FFFF */
773 while(i < sourceLength) {
774 if(lamAlefOption&&dest[i] == LAMALEF_SPACE_SUB){
775 dest[i] = SPACE_CHAR;
779 destSize = sourceLength;
784 if (shapingMode == 0) {
785 if ( ((options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefBegin) ||
786 (((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO )
787 && (shapeVars.spacesRelativeToTextBeginEnd==1)) ) {
790 if ( (options&U_SHAPE_TASHKEEL_MASK) == shapeVars.uShapeTashkeelBegin ) {
795 if(lamAlefOption || tashkeelOption){
796 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
798 i = j = sourceLength; count = 0;
801 if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
802 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
806 tempbuffer[j] = dest[i];
812 for(i=0 ;i < count; i++){
813 tempbuffer[i] = SPACE_CHAR;
816 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
817 destSize = sourceLength;
825 if (shapingMode == 0) {
826 if ( ((options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefEnd) ||
827 (((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO )
828 && (shapeVars.spacesRelativeToTextBeginEnd==0)) ) {
831 if ( (options&U_SHAPE_TASHKEEL_MASK) == shapeVars.uShapeTashkeelEnd ){
836 if(lamAlefOption || tashkeelOption){
837 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
839 i = j = 0; count = 0;
840 while(i < sourceLength) {
841 if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
842 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
846 tempbuffer[j] = dest[i];
853 tempbuffer[i] = SPACE_CHAR;
858 uprv_memcpy(dest,tempbuffer, sourceLength*U_SIZEOF_UCHAR);
859 destSize = sourceLength;
864 uprv_free(tempbuffer);
871 *Name :expandCompositCharAtBegin
872 *Function :Expands the LamAlef character to Lam and Alef consuming the required
873 * space from beginning of the buffer. If the text type was visual_LTR
874 * and the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END was selected
875 * the spaces will be located at end of buffer.
876 * If there are no spaces to expand the LamAlef, an error
877 * will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
881 expandCompositCharAtBegin(UChar *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode) {
884 UChar *tempbuffer=NULL;
886 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
889 if(tempbuffer == NULL) {
890 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
894 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
897 while(dest[i] == SPACE_CHAR) {
902 i = j = sourceLength-1;
904 while(i >= 0 && j >= 0) {
905 if( countl>0 && isLamAlefChar(dest[i])) {
906 tempbuffer[j] = LAM_CHAR;
907 /* to ensure the array index is within the range */
908 U_ASSERT(dest[i] >= 0xFEF5u
909 && dest[i]-0xFEF5u < sizeof(convertLamAlef)/sizeof(convertLamAlef[0]));
910 tempbuffer[j-1] = convertLamAlef[ dest[i] - 0xFEF5 ];
914 if( countl == 0 && isLamAlefChar(dest[i]) ) {
915 *pErrorCode=U_NO_SPACE_AVAILABLE;
917 tempbuffer[j] = dest[i];
922 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
924 uprv_free(tempbuffer);
926 destSize = sourceLength;
931 *Name : expandCompositCharAtEnd
932 *Function : Expands the LamAlef character to Lam and Alef consuming the
933 * required space from end of the buffer. If the text type was
934 * Visual LTR and the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END
935 * was used, the spaces will be consumed from begin of buffer. If
936 * there are no spaces to expand the LamAlef, an error
937 * will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
941 expandCompositCharAtEnd(UChar *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode) {
945 int32_t inpsize = sourceLength;
947 UChar *tempbuffer=NULL;
948 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
951 if(tempbuffer == NULL) {
952 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
956 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
958 while(dest[inpsize-1] == SPACE_CHAR) {
963 i = sourceLength - countr - 1;
964 j = sourceLength - 1;
966 while(i >= 0 && j >= 0) {
967 if( countr>0 && isLamAlefChar(dest[i]) ) {
968 tempbuffer[j] = LAM_CHAR;
969 tempbuffer[j-1] = convertLamAlef[ dest[i] - 0xFEF5 ];
973 if ((countr == 0) && isLamAlefChar(dest[i]) ) {
974 *pErrorCode=U_NO_SPACE_AVAILABLE;
976 tempbuffer[j] = dest[i];
983 uprv_memmove(tempbuffer, tempbuffer+countr, sourceLength*U_SIZEOF_UCHAR);
984 if(u_strlen(tempbuffer) < sourceLength) {
985 for(i=sourceLength-1;i>=sourceLength-countr;i--) {
986 tempbuffer[i] = SPACE_CHAR;
990 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
992 uprv_free(tempbuffer);
994 destSize = sourceLength;
999 *Name : expandCompositCharAtNear
1000 *Function : Expands the LamAlef character into Lam + Alef, YehHamza character
1001 * into Yeh + Hamza, SeenFamily character into SeenFamily character
1002 * + Tail, while consuming the space next to the character.
1003 * If there are no spaces next to the character, an error
1004 * will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
1008 expandCompositCharAtNear(UChar *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode,
1009 int yehHamzaOption, int seenTailOption, int lamAlefOption, struct uShapeVariables shapeVars) {
1013 UChar lamalefChar, yehhamzaChar;
1015 for(i = 0 ;i<=sourceLength-1;i++) {
1016 if (seenTailOption && isSeenTailFamilyChar(dest[i])) {
1017 if ((i>0) && (dest[i-1] == SPACE_CHAR) ) {
1018 dest[i-1] = shapeVars.tailChar;
1020 *pErrorCode=U_NO_SPACE_AVAILABLE;
1022 }else if(yehHamzaOption && (isYehHamzaChar(dest[i])) ) {
1023 if ((i>0) && (dest[i-1] == SPACE_CHAR) ) {
1024 yehhamzaChar = dest[i];
1025 dest[i] = yehHamzaToYeh[yehhamzaChar - YEH_HAMZAFE_CHAR];
1026 dest[i-1] = HAMZAFE_CHAR;
1029 *pErrorCode=U_NO_SPACE_AVAILABLE;
1031 }else if(lamAlefOption && isLamAlefChar(dest[i+1])) {
1032 if(dest[i] == SPACE_CHAR){
1033 lamalefChar = dest[i+1];
1034 dest[i+1] = LAM_CHAR;
1035 dest[i] = convertLamAlef[ lamalefChar - 0xFEF5 ];
1037 *pErrorCode=U_NO_SPACE_AVAILABLE;
1041 destSize = sourceLength;
1045 * Name : expandCompositChar
1046 * Function : LamAlef, need special handling, since it expands from one
1047 * character into two characters while shaping or deshaping.
1048 * In order to expand it, near or far spaces according to the
1049 * options user specifies. Also buffer size can be increased.
1051 * For SeenFamily characters and YehHamza only the near option is
1052 * supported, while for LamAlef we can take spaces from begin, end,
1053 * near or even increase the buffer size.
1054 * There is also the Auto option for LamAlef only, which will first
1055 * search for a space at end, begin then near, respectively.
1056 * If there are no spaces to expand these characters, an error will be set to
1057 * U_NO_SPACE_AVAILABLE as defined in utypes.h
1061 expandCompositChar(UChar *dest, int32_t sourceLength,
1062 int32_t destSize,uint32_t options,
1063 UErrorCode *pErrorCode, int shapingMode,struct uShapeVariables shapeVars) {
1065 int32_t i = 0,j = 0;
1067 UChar *tempbuffer=NULL;
1068 int yehHamzaOption = 0;
1069 int seenTailOption = 0;
1070 int lamAlefOption = 0;
1072 if (shapingMode == 1){
1073 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO){
1075 if(shapeVars.spacesRelativeToTextBeginEnd == 0) {
1076 destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
1078 if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
1079 *pErrorCode = U_ZERO_ERROR;
1080 destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
1083 destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
1085 if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
1086 *pErrorCode = U_ZERO_ERROR;
1087 destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
1091 if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
1092 *pErrorCode = U_ZERO_ERROR;
1093 destSize = expandCompositCharAtNear(dest, sourceLength, destSize, pErrorCode, yehHamzaOption,
1094 seenTailOption, 1,shapeVars);
1099 if (shapingMode == 1){
1100 if ( (options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefEnd){
1101 destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
1105 if (shapingMode == 1){
1106 if ( (options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefBegin){
1107 destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
1111 if (shapingMode == 0){
1112 if ((options&U_SHAPE_YEHHAMZA_MASK) == U_SHAPE_YEHHAMZA_TWOCELL_NEAR){
1115 if ((options&U_SHAPE_SEEN_MASK) == U_SHAPE_SEEN_TWOCELL_NEAR){
1119 if (shapingMode == 1) {
1120 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_NEAR) {
1126 if (yehHamzaOption || seenTailOption || lamAlefOption){
1127 destSize = expandCompositCharAtNear(dest, sourceLength, destSize, pErrorCode, yehHamzaOption,
1128 seenTailOption,lamAlefOption,shapeVars);
1132 if (shapingMode == 1){
1133 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE){
1134 destSize = calculateSize(dest,sourceLength,destSize,options);
1135 tempbuffer = (UChar *)uprv_malloc((destSize+1)*U_SIZEOF_UCHAR);
1138 if(tempbuffer == NULL) {
1139 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
1143 uprv_memset(tempbuffer, 0, (destSize+1)*U_SIZEOF_UCHAR);
1146 while(i < destSize && j < destSize) {
1147 if(isLamAlefChar(dest[i]) ) {
1148 tempbuffer[j] = convertLamAlef[ dest[i] - 0xFEF5 ];
1149 tempbuffer[j+1] = LAM_CHAR;
1152 tempbuffer[j] = dest[i];
1158 uprv_memcpy(dest, tempbuffer, destSize*U_SIZEOF_UCHAR);
1163 uprv_free(tempbuffer);
1169 *Name : shapeUnicode
1170 *Function : Converts an Arabic Unicode buffer in 06xx Range into a shaped
1171 * arabic Unicode buffer in FExx Range
1174 shapeUnicode(UChar *dest, int32_t sourceLength,
1175 int32_t destSize,uint32_t options,
1176 UErrorCode *pErrorCode,
1177 int tashkeelFlag, struct uShapeVariables shapeVars) {
1181 int32_t lastPos,Nx, Nw;
1183 int32_t lamalef_found = 0;
1184 int32_t seenfamFound = 0, yehhamzaFound =0, tashkeelFound = 0;
1185 UChar prevLink = 0, lastLink = 0, currLink, nextLink = 0;
1189 * Converts the input buffer from FExx Range into 06xx Range
1190 * to make sure that all characters are in the 06xx range
1191 * even the lamalef is converted to the special region in
1194 if ((options & U_SHAPE_PRESERVE_PRESENTATION_MASK) == U_SHAPE_PRESERVE_PRESENTATION_NOOP) {
1195 for (i = 0; i < sourceLength; i++) {
1196 UChar inputChar = dest[i];
1197 if ( (inputChar >= 0xFB50) && (inputChar <= 0xFBFF)) {
1198 UChar c = convertFBto06 [ (inputChar - 0xFB50) ];
1201 } else if ( (inputChar >= 0xFE70) && (inputChar <= 0xFEFC)) {
1202 dest[i] = convertFEto06 [ (inputChar - 0xFE70) ] ;
1204 dest[i] = inputChar ;
1210 /* sets the index to the end of the buffer, together with the step point to -1 */
1211 i = sourceLength - 1;
1216 * This function resolves the link between the characters .
1217 * Arabic characters have four forms :
1218 * Isolated Form, Initial Form, Middle Form and Final Form
1220 currLink = getLink(dest[i]);
1226 /* If high byte of currLink > 0 then more than one shape */
1227 if ((currLink & 0xFF00) > 0 || (getLink(dest[i]) & IRRELEVANT) != 0) {
1229 while (Nx < 0) { /* we need to know about next char */
1234 nextLink = getLink(dest[Nw]);
1235 if((nextLink & IRRELEVANT) == 0) {
1243 if ( ((currLink & ALEFTYPE) > 0) && ((lastLink & LAMTYPE) > 0) ) {
1245 wLamalef = changeLamAlef(dest[i]); /*get from 0x065C-0x065f */
1246 if ( wLamalef != 0) {
1247 dest[i] = LAMALEF_SPACE_SUB; /* The default case is to drop the Alef and replace */
1248 dest[lastPos] =wLamalef; /* it by LAMALEF_SPACE_SUB which is the last character in the */
1249 i=lastPos; /* unicode private use area, this is done to make */
1250 } /* sure that removeLamAlefSpaces() handles only the */
1251 lastLink = prevLink; /* spaces generated during lamalef generation. */
1252 currLink = getLink(wLamalef); /* LAMALEF_SPACE_SUB is added here and is replaced by spaces */
1253 } /* in removeLamAlefSpaces() */
1255 if ((i > 0) && (dest[i-1] == SPACE_CHAR)){
1256 if ( isSeenFamilyChar(dest[i])) {
1258 } else if (dest[i] == YEH_HAMZA_CHAR) {
1263 if ( isSeenFamilyChar(dest[i])){
1265 } else if (dest[i] == YEH_HAMZA_CHAR) {
1271 * get the proper shape according to link ability of neighbors
1272 * and of character; depends on the order of the shapes
1273 * (isolated, initial, middle, final) in the compatibility area
1275 Shape = shapeTable[nextLink & (LINKR + LINKL)]
1276 [lastLink & (LINKR + LINKL)]
1277 [currLink & (LINKR + LINKL)];
1279 if ((currLink & (LINKR+LINKL)) == 1) {
1281 } else if(isTashkeelChar(dest[i])) {
1282 if( (lastLink & LINKL) && (nextLink & LINKR) && (tashkeelFlag == 1) &&
1283 dest[i] != 0x064C && dest[i] != 0x064D )
1286 if( (nextLink&ALEFTYPE) == ALEFTYPE && (lastLink&LAMTYPE) == LAMTYPE ) {
1289 } else if(tashkeelFlag == 2 && dest[i] == SHADDA06_CHAR){
1295 if ((dest[i] ^ 0x0600) < 0x100) {
1296 if ( isTashkeelChar(dest[i]) ){
1297 if (tashkeelFlag == 2 && dest[i] != SHADDA06_CHAR){
1298 dest[i] = TASHKEEL_SPACE_SUB;
1301 /* to ensure the array index is within the range */
1302 U_ASSERT(dest[i] >= 0x064Bu
1303 && dest[i]-0x064Bu < sizeof(IrrelevantPos)/sizeof(IrrelevantPos[0]));
1304 dest[i] = 0xFE70 + IrrelevantPos[(dest[i] - 0x064B)] + Shape;
1306 }else if ((currLink & APRESENT) > 0) {
1307 dest[i] = (UChar)(0xFB50 + (currLink >> 8) + Shape);
1308 }else if ((currLink >> 8) > 0 && (currLink & IRRELEVANT) == 0) {
1309 dest[i] = (UChar)(0xFE70 + (currLink >> 8) + Shape);
1314 /* move one notch forward */
1315 if ((currLink & IRRELEVANT) == 0) {
1316 prevLink = lastLink;
1317 lastLink = currLink;
1323 currLink = nextLink;
1325 } else if(i != iend) {
1326 currLink = getLink(dest[i]);
1329 destSize = sourceLength;
1330 if ( (lamalef_found != 0 ) || (tashkeelFound != 0) ){
1331 destSize = handleGeneratedSpaces(dest,sourceLength,destSize,options,pErrorCode, shapeVars);
1334 if ( (seenfamFound != 0) || (yehhamzaFound != 0) ) {
1335 destSize = expandCompositChar(dest, sourceLength,destSize,options,pErrorCode, SHAPE_MODE,shapeVars);
1341 *Name : deShapeUnicode
1342 *Function : Converts an Arabic Unicode buffer in FExx Range into unshaped
1343 * arabic Unicode buffer in 06xx Range
1346 deShapeUnicode(UChar *dest, int32_t sourceLength,
1347 int32_t destSize,uint32_t options,
1348 UErrorCode *pErrorCode, struct uShapeVariables shapeVars) {
1350 int32_t lamalef_found = 0;
1351 int32_t yehHamzaComposeEnabled = 0;
1352 int32_t seenComposeEnabled = 0;
1354 yehHamzaComposeEnabled = ((options&U_SHAPE_YEHHAMZA_MASK) == U_SHAPE_YEHHAMZA_TWOCELL_NEAR) ? 1 : 0;
1355 seenComposeEnabled = ((options&U_SHAPE_SEEN_MASK) == U_SHAPE_SEEN_TWOCELL_NEAR)? 1 : 0;
1358 *This for loop changes the buffer from the Unicode FE range to
1359 *the Unicode 06 range
1362 for(i = 0; i < sourceLength; i++) {
1363 UChar inputChar = dest[i];
1364 if ( (inputChar >= 0xFB50) && (inputChar <= 0xFBFF)) { /* FBxx Arabic range */
1365 UChar c = convertFBto06 [ (inputChar - 0xFB50) ];
1368 } else if( (yehHamzaComposeEnabled == 1) && ((inputChar == HAMZA06_CHAR) || (inputChar == HAMZAFE_CHAR))
1369 && (i < (sourceLength - 1)) && isAlefMaksouraChar(dest[i+1] )) {
1370 dest[i] = SPACE_CHAR;
1371 dest[i+1] = YEH_HAMZA_CHAR;
1372 } else if ( (seenComposeEnabled == 1) && (isTailChar(inputChar)) && (i< (sourceLength - 1))
1373 && (isSeenTailFamilyChar(dest[i+1])) ) {
1374 dest[i] = SPACE_CHAR;
1375 } else if (( inputChar >= 0xFE70) && (inputChar <= 0xFEF4 )) { /* FExx Arabic range */
1376 dest[i] = convertFEto06 [ (inputChar - 0xFE70) ];
1378 dest[i] = inputChar ;
1381 if( isLamAlefChar(dest[i]) )
1385 destSize = sourceLength;
1386 if (lamalef_found != 0){
1387 destSize = expandCompositChar(dest,sourceLength,destSize,options,pErrorCode,DESHAPE_MODE, shapeVars);
1393 ****************************************
1395 ****************************************
1398 U_CAPI int32_t U_EXPORT2
1399 u_shapeArabic(const UChar *source, int32_t sourceLength,
1400 UChar *dest, int32_t destCapacity,
1402 UErrorCode *pErrorCode) {
1405 struct uShapeVariables shapeVars = { OLD_TAIL_CHAR,U_SHAPE_LAMALEF_BEGIN,U_SHAPE_LAMALEF_END,U_SHAPE_TASHKEEL_BEGIN,U_SHAPE_TASHKEEL_END,0};
1407 /* usual error checking */
1408 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1412 /* make sure that no reserved options values are used; allow dest==NULL only for preflighting */
1413 if( source==NULL || sourceLength<-1 || (dest==NULL && destCapacity!=0) || destCapacity<0 ||
1414 (((options&U_SHAPE_TASHKEEL_MASK) > 0) &&
1415 ((options&U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) == U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) ) ||
1416 (((options&U_SHAPE_TASHKEEL_MASK) > 0) &&
1417 ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_UNSHAPE)) ||
1418 (options&U_SHAPE_DIGIT_TYPE_RESERVED)==U_SHAPE_DIGIT_TYPE_RESERVED ||
1419 (options&U_SHAPE_DIGITS_MASK)==U_SHAPE_DIGITS_RESERVED ||
1420 ((options&U_SHAPE_LAMALEF_MASK) != U_SHAPE_LAMALEF_RESIZE &&
1421 (options&U_SHAPE_AGGREGATE_TASHKEEL_MASK) != 0) ||
1422 ((options&U_SHAPE_AGGREGATE_TASHKEEL_MASK) == U_SHAPE_AGGREGATE_TASHKEEL &&
1423 (options&U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) != U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED)
1426 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1429 /* Validate lamalef options */
1430 if(((options&U_SHAPE_LAMALEF_MASK) > 0)&&
1431 !(((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_BEGIN) ||
1432 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_END ) ||
1433 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_RESIZE )||
1434 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_AUTO) ||
1435 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_NEAR)))
1437 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1440 /* Validate Tashkeel options */
1441 if(((options&U_SHAPE_TASHKEEL_MASK) > 0)&&
1442 !(((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_BEGIN) ||
1443 ((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_END )
1444 ||((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_RESIZE )||
1445 ((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL)))
1447 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1450 /* determine the source length */
1451 if(sourceLength==-1) {
1452 sourceLength=u_strlen(source);
1454 if(sourceLength<=0) {
1455 return u_terminateUChars(dest, destCapacity, 0, pErrorCode);
1458 /* check that source and destination do not overlap */
1460 ((source<=dest && dest<source+sourceLength) ||
1461 (dest<=source && source<dest+destCapacity))) {
1462 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1466 /* Does Options contain the new Seen Tail Unicode code point option */
1467 if ( (options&U_SHAPE_TAIL_TYPE_MASK) == U_SHAPE_TAIL_NEW_UNICODE){
1468 shapeVars.tailChar = NEW_TAIL_CHAR;
1470 shapeVars.tailChar = OLD_TAIL_CHAR;
1473 if((options&U_SHAPE_LETTERS_MASK)!=U_SHAPE_LETTERS_NOOP) {
1475 UChar *tempbuffer, *tempsource = NULL;
1476 int32_t outputSize, spacesCountl=0, spacesCountr=0;
1478 if((options&U_SHAPE_AGGREGATE_TASHKEEL_MASK)>0) {
1479 int32_t logical_order = (options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL;
1480 int32_t aggregate_tashkeel =
1481 (options&(U_SHAPE_AGGREGATE_TASHKEEL_MASK+U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED)) ==
1482 (U_SHAPE_AGGREGATE_TASHKEEL+U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED);
1483 int step=logical_order?1:-1;
1484 int j=logical_order?-1:2*sourceLength;
1485 int i=logical_order?-1:sourceLength;
1486 int end=logical_order?sourceLength:-1;
1487 int aggregation_possible = 1;
1489 UChar prevLink, currLink = 0;
1490 int newSourceLength = 0;
1491 tempsource = (UChar *)uprv_malloc(2*sourceLength*U_SIZEOF_UCHAR);
1492 if(tempsource == NULL) {
1493 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
1497 while ((i+=step) != end) {
1498 prevLink = currLink;
1499 currLink = getLink(source[i]);
1500 if (aggregate_tashkeel && ((prevLink|currLink)&COMBINE) == COMBINE && aggregation_possible) {
1501 aggregation_possible = 0;
1502 tempsource[j] = (prev<source[i]?prev:source[i])-0x064C+0xFC5E;
1503 currLink = getLink(tempsource[j]);
1505 aggregation_possible = 1;
1506 tempsource[j+=step] = source[i];
1511 source = tempsource+(logical_order?0:j);
1512 sourceLength = newSourceLength;
1515 /* calculate destination size */
1516 /* TODO: do we ever need to do this pure preflighting? */
1517 if(((options&U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_RESIZE) ||
1518 ((options&U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_RESIZE)) {
1519 outputSize=calculateSize(source,sourceLength,destCapacity,options);
1521 outputSize=sourceLength;
1524 if(outputSize>destCapacity) {
1525 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1526 if (tempsource != NULL) uprv_free(tempsource);
1531 * need a temporary buffer of size max(outputSize, sourceLength)
1532 * because at first we copy source->temp
1534 if(sourceLength>outputSize) {
1535 outputSize=sourceLength;
1538 /* Start of Arabic letter shaping part */
1539 if(outputSize<=LENGTHOF(buffer)) {
1540 outputSize=LENGTHOF(buffer);
1543 tempbuffer = (UChar *)uprv_malloc(outputSize*U_SIZEOF_UCHAR);
1546 if(tempbuffer == NULL) {
1547 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
1548 if (tempsource != NULL) uprv_free(tempsource);
1552 uprv_memcpy(tempbuffer, source, sourceLength*U_SIZEOF_UCHAR);
1553 if (tempsource != NULL){
1554 uprv_free(tempsource);
1557 if(sourceLength<outputSize) {
1558 uprv_memset(tempbuffer+sourceLength, 0, (outputSize-sourceLength)*U_SIZEOF_UCHAR);
1561 if((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL) {
1562 countSpaces(tempbuffer,sourceLength,options,&spacesCountl,&spacesCountr);
1563 invertBuffer(tempbuffer,sourceLength,options,spacesCountl,spacesCountr);
1566 if((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_VISUAL_LTR) {
1567 if((options&U_SHAPE_SPACES_RELATIVE_TO_TEXT_MASK) == U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END) {
1568 shapeVars.spacesRelativeToTextBeginEnd = 1;
1569 shapeVars.uShapeLamalefBegin = U_SHAPE_LAMALEF_END;
1570 shapeVars.uShapeLamalefEnd = U_SHAPE_LAMALEF_BEGIN;
1571 shapeVars.uShapeTashkeelBegin = U_SHAPE_TASHKEEL_END;
1572 shapeVars.uShapeTashkeelEnd = U_SHAPE_TASHKEEL_BEGIN;
1576 switch(options&U_SHAPE_LETTERS_MASK) {
1577 case U_SHAPE_LETTERS_SHAPE :
1578 if( (options&U_SHAPE_TASHKEEL_MASK)> 0
1579 && ((options&U_SHAPE_TASHKEEL_MASK) !=U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL)) {
1580 /* Call the shaping function with tashkeel flag == 2 for removal of tashkeel */
1581 destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,2,shapeVars);
1583 /* default Call the shaping function with tashkeel flag == 1 */
1584 destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,1,shapeVars);
1586 /*After shaping text check if user wants to remove tashkeel and replace it with tatweel*/
1587 if( (options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL){
1588 destLength = handleTashkeelWithTatweel(tempbuffer,destLength,destCapacity,options,pErrorCode);
1592 case U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED :
1593 /* Call the shaping function with tashkeel flag == 0 */
1594 destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,0,shapeVars);
1597 case U_SHAPE_LETTERS_UNSHAPE :
1598 /* Call the deshaping function */
1599 destLength = deShapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,shapeVars);
1602 /* will never occur because of validity checks above */
1608 * TODO: (markus 2002aug01)
1609 * For as long as we always preflight the outputSize above
1610 * we should U_ASSERT(outputSize==destLength)
1611 * except for the adjustment above before the tempbuffer allocation
1614 if((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL) {
1615 countSpaces(tempbuffer,destLength,options,&spacesCountl,&spacesCountr);
1616 invertBuffer(tempbuffer,destLength,options,spacesCountl,spacesCountr);
1618 uprv_memcpy(dest, tempbuffer, uprv_min(destLength, destCapacity)*U_SIZEOF_UCHAR);
1620 if(tempbuffer!=buffer) {
1621 uprv_free(tempbuffer);
1624 if(destLength>destCapacity) {
1625 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1629 /* End of Arabic letter shaping part */
1632 * No letter shaping:
1633 * just make sure the destination is large enough and copy the string.
1635 if(destCapacity<sourceLength) {
1636 /* this catches preflighting, too */
1637 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1638 return sourceLength;
1640 uprv_memcpy(dest, source, sourceLength*U_SIZEOF_UCHAR);
1641 destLength=sourceLength;
1645 * Perform number shaping.
1646 * With UTF-16 or UTF-32, the length of the string is constant.
1647 * The easiest way to do this is to operate on the destination and
1648 * "shape" the digits in-place.
1650 if((options&U_SHAPE_DIGITS_MASK)!=U_SHAPE_DIGITS_NOOP) {
1654 /* select the requested digit group */
1655 switch(options&U_SHAPE_DIGIT_TYPE_MASK) {
1656 case U_SHAPE_DIGIT_TYPE_AN:
1657 digitBase=0x660; /* Unicode: "Arabic-Indic digits" */
1659 case U_SHAPE_DIGIT_TYPE_AN_EXTENDED:
1660 digitBase=0x6f0; /* Unicode: "Eastern Arabic-Indic digits (Persian and Urdu)" */
1663 /* will never occur because of validity checks above */
1668 /* perform the requested operation */
1669 switch(options&U_SHAPE_DIGITS_MASK) {
1670 case U_SHAPE_DIGITS_EN2AN:
1671 /* add (digitBase-'0') to each European (ASCII) digit code point */
1673 for(i=0; i<destLength; ++i) {
1674 if(((uint32_t)dest[i]-0x30)<10) {
1679 case U_SHAPE_DIGITS_AN2EN:
1680 /* subtract (digitBase-'0') from each Arabic digit code point */
1681 for(i=0; i<destLength; ++i) {
1682 if(((uint32_t)dest[i]-(uint32_t)digitBase)<10) {
1683 dest[i]-=digitBase-0x30;
1687 case U_SHAPE_DIGITS_ALEN2AN_INIT_LR:
1688 _shapeToArabicDigitsWithContext(dest, destLength,
1690 (UBool)((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_LOGICAL),
1693 case U_SHAPE_DIGITS_ALEN2AN_INIT_AL:
1694 _shapeToArabicDigitsWithContext(dest, destLength,
1696 (UBool)((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_LOGICAL),
1700 /* will never occur because of validity checks above */
1705 return u_terminateUChars(dest, destCapacity, destLength, pErrorCode);