1 /*
2 ******************************************************************************
3 *
4 * Copyright (C) 1999-2009, International Business Machines
5 * Corporation and others. All Rights Reserved.
6 *
7 ******************************************************************************
8 * file name: ubidi.c
9 * encoding: US-ASCII
10 * tab size: 8 (not used)
11 * indentation:4
12 *
13 * created on: 1999jul27
14 * created by: Markus W. Scherer, updated by Matitiahu Allouche
15 */
16
17 #include "cmemory.h"
18 #include "unicode/utypes.h"
19 #include "unicode/ustring.h"
20 #include "unicode/uchar.h"
21 #include "unicode/ubidi.h"
22 #include "ubidi_props.h"
23 #include "ubidiimp.h"
24 #include "uassert.h"
25
26 /*
27 * General implementation notes:
28 *
29 * Throughout the implementation, there are comments like (W2) that refer to
30 * rules of the BiDi algorithm in its version 5, in this example to the second
31 * rule of the resolution of weak types.
32 *
33 * For handling surrogate pairs, where two UChar's form one "abstract" (or UTF-32)
34 * character according to UTF-16, the second UChar gets the directional property of
35 * the entire character assigned, while the first one gets a BN, a boundary
36 * neutral, type, which is ignored by most of the algorithm according to
37 * rule (X9) and the implementation suggestions of the BiDi algorithm.
38 *
39 * Later, adjustWSLevels() will set the level for each BN to that of the
40 * following character (UChar), which results in surrogate pairs getting the
41 * same level on each of their surrogates.
42 *
43 * In a UTF-8 implementation, the same thing could be done: the last byte of
44 * a multi-byte sequence would get the "real" property, while all previous
45 * bytes of that sequence would get BN.
46 *
47 * It is not possible to assign all those parts of a character the same real
48 * property because this would fail in the resolution of weak types with rules
49 * that look at immediately surrounding types.
50 *
51 * As a related topic, this implementation does not remove Boundary Neutral
52 * types from the input, but ignores them wherever this is relevant.
53 * For example, the loop for the resolution of the weak types reads
54 * types until it finds a non-BN.
55 * Also, explicit embedding codes are neither changed into BN nor removed.
56 * They are only treated the same way real BNs are.
57 * As stated before, adjustWSLevels() takes care of them at the end.
58 * For the purpose of conformance, the levels of all these codes
59 * do not matter.
60 *
61 * Note that this implementation never modifies the dirProps
62 * after the initial setup.
63 *
64 *
65 * In this implementation, the resolution of weak types (Wn),
66 * neutrals (Nn), and the assignment of the resolved level (In)
67 * are all done in one single loop, in resolveImplicitLevels().
68 * Changes of dirProp values are done on the fly, without writing
69 * them back to the dirProps array.
70 *
71 *
72 * This implementation contains code that allows to bypass steps of the
73 * algorithm that are not needed on the specific paragraph
74 * in order to speed up the most common cases considerably,
75 * like text that is entirely LTR, or RTL text without numbers.
76 *
77 * Most of this is done by setting a bit for each directional property
78 * in a flags variable and later checking for whether there are
79 * any LTR characters or any RTL characters, or both, whether
80 * there are any explicit embedding codes, etc.
81 *
82 * If the (Xn) steps are performed, then the flags are re-evaluated,
83 * because they will then not contain the embedding codes any more
84 * and will be adjusted for override codes, so that subsequently
85 * more bypassing may be possible than what the initial flags suggested.
86 *
87 * If the text is not mixed-directional, then the
88 * algorithm steps for the weak type resolution are not performed,
89 * and all levels are set to the paragraph level.
90 *
91 * If there are no explicit embedding codes, then the (Xn) steps
92 * are not performed.
93 *
94 * If embedding levels are supplied as a parameter, then all
95 * explicit embedding codes are ignored, and the (Xn) steps
96 * are not performed.
97 *
98 * White Space types could get the level of the run they belong to,
99 * and are checked with a test of (flags&MASK_EMBEDDING) to
100 * consider if the paragraph direction should be considered in
101 * the flags variable.
102 *
103 * If there are no White Space types in the paragraph, then
104 * (L1) is not necessary in adjustWSLevels().
105 */
106
107 /* to avoid some conditional statements, use tiny constant arrays */
108 static const Flags flagLR[2]={ DIRPROP_FLAG(L), DIRPROP_FLAG(R) };
109 static const Flags flagE[2]={ DIRPROP_FLAG(LRE), DIRPROP_FLAG(RLE) };
110 static const Flags flagO[2]={ DIRPROP_FLAG(LRO), DIRPROP_FLAG(RLO) };
111
112 #define DIRPROP_FLAG_LR(level) flagLR[(level)&1]
113 #define DIRPROP_FLAG_E(level) flagE[(level)&1]
114 #define DIRPROP_FLAG_O(level) flagO[(level)&1]
115
116 /* UBiDi object management -------------------------------------------------- */
117
118 U_CAPI UBiDi * U_EXPORT2
ubidi_open(void)119 ubidi_open(void)
120 {
121 UErrorCode errorCode=U_ZERO_ERROR;
122 return ubidi_openSized(0, 0, &errorCode);
123 }
124
125 U_CAPI UBiDi * U_EXPORT2
ubidi_openSized(int32_t maxLength,int32_t maxRunCount,UErrorCode * pErrorCode)126 ubidi_openSized(int32_t maxLength, int32_t maxRunCount, UErrorCode *pErrorCode) {
127 UBiDi *pBiDi;
128
129 /* check the argument values */
130 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
131 return NULL;
132 } else if(maxLength<0 || maxRunCount<0) {
133 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
134 return NULL; /* invalid arguments */
135 }
136
137 /* allocate memory for the object */
138 pBiDi=(UBiDi *)uprv_malloc(sizeof(UBiDi));
139 if(pBiDi==NULL) {
140 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
141 return NULL;
142 }
143
144 /* reset the object, all pointers NULL, all flags FALSE, all sizes 0 */
145 uprv_memset(pBiDi, 0, sizeof(UBiDi));
146
147 /* get BiDi properties */
148 pBiDi->bdp=ubidi_getSingleton(pErrorCode);
149 if(U_FAILURE(*pErrorCode)) {
150 uprv_free(pBiDi);
151 return NULL;
152 }
153
154 /* allocate memory for arrays as requested */
155 if(maxLength>0) {
156 if( !getInitialDirPropsMemory(pBiDi, maxLength) ||
157 !getInitialLevelsMemory(pBiDi, maxLength)
158 ) {
159 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
160 }
161 } else {
162 pBiDi->mayAllocateText=TRUE;
163 }
164
165 if(maxRunCount>0) {
166 if(maxRunCount==1) {
167 /* use simpleRuns[] */
168 pBiDi->runsSize=sizeof(Run);
169 } else if(!getInitialRunsMemory(pBiDi, maxRunCount)) {
170 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
171 }
172 } else {
173 pBiDi->mayAllocateRuns=TRUE;
174 }
175
176 if(U_SUCCESS(*pErrorCode)) {
177 return pBiDi;
178 } else {
179 ubidi_close(pBiDi);
180 return NULL;
181 }
182 }
183
184 /*
185 * We are allowed to allocate memory if memory==NULL or
186 * mayAllocate==TRUE for each array that we need.
187 * We also try to grow memory as needed if we
188 * allocate it.
189 *
190 * Assume sizeNeeded>0.
191 * If *pMemory!=NULL, then assume *pSize>0.
192 *
193 * ### this realloc() may unnecessarily copy the old data,
194 * which we know we don't need any more;
195 * is this the best way to do this??
196 */
197 U_CFUNC UBool
ubidi_getMemory(BidiMemoryForAllocation * bidiMem,int32_t * pSize,UBool mayAllocate,int32_t sizeNeeded)198 ubidi_getMemory(BidiMemoryForAllocation *bidiMem, int32_t *pSize, UBool mayAllocate, int32_t sizeNeeded) {
199 void **pMemory = (void **)bidiMem;
200 /* check for existing memory */
201 if(*pMemory==NULL) {
202 /* we need to allocate memory */
203 if(mayAllocate && (*pMemory=uprv_malloc(sizeNeeded))!=NULL) {
204 *pSize=sizeNeeded;
205 return TRUE;
206 } else {
207 return FALSE;
208 }
209 } else {
210 if(sizeNeeded<=*pSize) {
211 /* there is already enough memory */
212 return TRUE;
213 }
214 else if(!mayAllocate) {
215 /* not enough memory, and we must not allocate */
216 return FALSE;
217 } else {
218 /* we try to grow */
219 void *memory;
220 /* in most cases, we do not need the copy-old-data part of
221 * realloc, but it is needed when adding runs using getRunsMemory()
222 * in setParaRunsOnly()
223 */
224 if((memory=uprv_realloc(*pMemory, sizeNeeded))!=NULL) {
225 *pMemory=memory;
226 *pSize=sizeNeeded;
227 return TRUE;
228 } else {
229 /* we failed to grow */
230 return FALSE;
231 }
232 }
233 }
234 }
235
236 U_CAPI void U_EXPORT2
ubidi_close(UBiDi * pBiDi)237 ubidi_close(UBiDi *pBiDi) {
238 if(pBiDi!=NULL) {
239 pBiDi->pParaBiDi=NULL; /* in case one tries to reuse this block */
240 if(pBiDi->dirPropsMemory!=NULL) {
241 uprv_free(pBiDi->dirPropsMemory);
242 }
243 if(pBiDi->levelsMemory!=NULL) {
244 uprv_free(pBiDi->levelsMemory);
245 }
246 if(pBiDi->runsMemory!=NULL) {
247 uprv_free(pBiDi->runsMemory);
248 }
249 if(pBiDi->parasMemory!=NULL) {
250 uprv_free(pBiDi->parasMemory);
251 }
252 if(pBiDi->insertPoints.points!=NULL) {
253 uprv_free(pBiDi->insertPoints.points);
254 }
255
256 uprv_free(pBiDi);
257 }
258 }
259
260 /* set to approximate "inverse BiDi" ---------------------------------------- */
261
262 U_CAPI void U_EXPORT2
ubidi_setInverse(UBiDi * pBiDi,UBool isInverse)263 ubidi_setInverse(UBiDi *pBiDi, UBool isInverse) {
264 if(pBiDi!=NULL) {
265 pBiDi->isInverse=isInverse;
266 pBiDi->reorderingMode = isInverse ? UBIDI_REORDER_INVERSE_NUMBERS_AS_L
267 : UBIDI_REORDER_DEFAULT;
268 }
269 }
270
271 U_CAPI UBool U_EXPORT2
ubidi_isInverse(UBiDi * pBiDi)272 ubidi_isInverse(UBiDi *pBiDi) {
273 if(pBiDi!=NULL) {
274 return pBiDi->isInverse;
275 } else {
276 return FALSE;
277 }
278 }
279
280 /* FOOD FOR THOUGHT: currently the reordering modes are a mixture of
281 * algorithm for direct BiDi, algorithm for inverse BiDi and the bizarre
282 * concept of RUNS_ONLY which is a double operation.
283 * It could be advantageous to divide this into 3 concepts:
284 * a) Operation: direct / inverse / RUNS_ONLY
285 * b) Direct algorithm: default / NUMBERS_SPECIAL / GROUP_NUMBERS_WITH_R
286 * c) Inverse algorithm: default / INVERSE_LIKE_DIRECT / NUMBERS_SPECIAL
287 * This would allow combinations not possible today like RUNS_ONLY with
288 * NUMBERS_SPECIAL.
289 * Also allow to set INSERT_MARKS for the direct step of RUNS_ONLY and
290 * REMOVE_CONTROLS for the inverse step.
291 * Not all combinations would be supported, and probably not all do make sense.
292 * This would need to document which ones are supported and what are the
293 * fallbacks for unsupported combinations.
294 */
295 U_CAPI void U_EXPORT2
ubidi_setReorderingMode(UBiDi * pBiDi,UBiDiReorderingMode reorderingMode)296 ubidi_setReorderingMode(UBiDi *pBiDi, UBiDiReorderingMode reorderingMode) {
297 if ((pBiDi!=NULL) && (reorderingMode >= UBIDI_REORDER_DEFAULT)
298 && (reorderingMode < UBIDI_REORDER_COUNT)) {
299 pBiDi->reorderingMode = reorderingMode;
300 pBiDi->isInverse = (UBool)(reorderingMode == UBIDI_REORDER_INVERSE_NUMBERS_AS_L);
301 }
302 }
303
304 U_CAPI UBiDiReorderingMode U_EXPORT2
ubidi_getReorderingMode(UBiDi * pBiDi)305 ubidi_getReorderingMode(UBiDi *pBiDi) {
306 if (pBiDi!=NULL) {
307 return pBiDi->reorderingMode;
308 } else {
309 return UBIDI_REORDER_DEFAULT;
310 }
311 }
312
313 U_CAPI void U_EXPORT2
ubidi_setReorderingOptions(UBiDi * pBiDi,uint32_t reorderingOptions)314 ubidi_setReorderingOptions(UBiDi *pBiDi, uint32_t reorderingOptions) {
315 if (reorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) {
316 reorderingOptions&=~UBIDI_OPTION_INSERT_MARKS;
317 }
318 if (pBiDi!=NULL) {
319 pBiDi->reorderingOptions=reorderingOptions;
320 }
321 }
322
323 U_CAPI uint32_t U_EXPORT2
ubidi_getReorderingOptions(UBiDi * pBiDi)324 ubidi_getReorderingOptions(UBiDi *pBiDi) {
325 if (pBiDi!=NULL) {
326 return pBiDi->reorderingOptions;
327 } else {
328 return 0;
329 }
330 }
331
332 /* perform (P2)..(P3) ------------------------------------------------------- */
333
334 /*
335 * Get the directional properties for the text,
336 * calculate the flags bit-set, and
337 * determine the paragraph level if necessary.
338 */
339 static void
getDirProps(UBiDi * pBiDi)340 getDirProps(UBiDi *pBiDi) {
341 const UChar *text=pBiDi->text;
342 DirProp *dirProps=pBiDi->dirPropsMemory; /* pBiDi->dirProps is const */
343
344 int32_t i=0, i1, length=pBiDi->originalLength;
345 Flags flags=0; /* collect all directionalities in the text */
346 UChar32 uchar;
347 DirProp dirProp=0, paraDirDefault=0;/* initialize to avoid compiler warnings */
348 UBool isDefaultLevel=IS_DEFAULT_LEVEL(pBiDi->paraLevel);
349 /* for inverse BiDi, the default para level is set to RTL if there is a
350 strong R or AL character at either end of the text */
351 UBool isDefaultLevelInverse=isDefaultLevel && (UBool)
352 (pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_LIKE_DIRECT ||
353 pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL);
354 int32_t lastArabicPos=-1;
355 int32_t controlCount=0;
356 UBool removeBiDiControls = (UBool)(pBiDi->reorderingOptions &
357 UBIDI_OPTION_REMOVE_CONTROLS);
358
359 typedef enum {
360 NOT_CONTEXTUAL, /* 0: not contextual paraLevel */
361 LOOKING_FOR_STRONG, /* 1: looking for first strong char */
362 FOUND_STRONG_CHAR /* 2: found first strong char */
363 } State;
364 State state;
365 int32_t paraStart=0; /* index of first char in paragraph */
366 DirProp paraDir; /* == CONTEXT_RTL within paragraphs
367 starting with strong R char */
368 DirProp lastStrongDir=0; /* for default level & inverse BiDi */
369 int32_t lastStrongLTR=0; /* for STREAMING option */
370
371 if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) {
372 pBiDi->length=0;
373 lastStrongLTR=0;
374 }
375 if(isDefaultLevel) {
376 paraDirDefault=pBiDi->paraLevel&1 ? CONTEXT_RTL : 0;
377 paraDir=paraDirDefault;
378 lastStrongDir=paraDirDefault;
379 state=LOOKING_FOR_STRONG;
380 } else {
381 state=NOT_CONTEXTUAL;
382 paraDir=0;
383 }
384 /* count paragraphs and determine the paragraph level (P2..P3) */
385 /*
386 * see comment in ubidi.h:
387 * the DEFAULT_XXX values are designed so that
388 * their bit 0 alone yields the intended default
389 */
390 for( /* i=0 above */ ; i<length; ) {
391 /* i is incremented by U16_NEXT */
392 U16_NEXT(text, i, length, uchar);
393 flags|=DIRPROP_FLAG(dirProp=(DirProp)ubidi_getCustomizedClass(pBiDi, uchar));
394 dirProps[i-1]=dirProp|paraDir;
395 if(uchar>0xffff) { /* set the lead surrogate's property to BN */
396 flags|=DIRPROP_FLAG(BN);
397 dirProps[i-2]=(DirProp)(BN|paraDir);
398 }
399 if(state==LOOKING_FOR_STRONG) {
400 if(dirProp==L) {
401 state=FOUND_STRONG_CHAR;
402 if(paraDir) {
403 paraDir=0;
404 for(i1=paraStart; i1<i; i1++) {
405 dirProps[i1]&=~CONTEXT_RTL;
406 }
407 }
408 continue;
409 }
410 if(dirProp==R || dirProp==AL) {
411 state=FOUND_STRONG_CHAR;
412 if(paraDir==0) {
413 paraDir=CONTEXT_RTL;
414 for(i1=paraStart; i1<i; i1++) {
415 dirProps[i1]|=CONTEXT_RTL;
416 }
417 }
418 continue;
419 }
420 }
421 if(dirProp==L) {
422 lastStrongDir=0;
423 lastStrongLTR=i; /* i is index to next character */
424 }
425 else if(dirProp==R) {
426 lastStrongDir=CONTEXT_RTL;
427 }
428 else if(dirProp==AL) {
429 lastStrongDir=CONTEXT_RTL;
430 lastArabicPos=i-1;
431 }
432 else if(dirProp==B) {
433 if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) {
434 pBiDi->length=i; /* i is index to next character */
435 }
436 if(isDefaultLevelInverse && (lastStrongDir==CONTEXT_RTL) &&(paraDir!=lastStrongDir)) {
437 for( ; paraStart<i; paraStart++) {
438 dirProps[paraStart]|=CONTEXT_RTL;
439 }
440 }
441 if(i<length) { /* B not last char in text */
442 if(!((uchar==CR) && (text[i]==LF))) {
443 pBiDi->paraCount++;
444 }
445 if(isDefaultLevel) {
446 state=LOOKING_FOR_STRONG;
447 paraStart=i; /* i is index to next character */
448 paraDir=paraDirDefault;
449 lastStrongDir=paraDirDefault;
450 }
451 }
452 }
453 if(removeBiDiControls && IS_BIDI_CONTROL_CHAR(uchar)) {
454 controlCount++;
455 }
456 }
457 if(isDefaultLevelInverse && (lastStrongDir==CONTEXT_RTL) &&(paraDir!=lastStrongDir)) {
458 for(i1=paraStart; i1<length; i1++) {
459 dirProps[i1]|=CONTEXT_RTL;
460 }
461 }
462 if(isDefaultLevel) {
463 pBiDi->paraLevel=GET_PARALEVEL(pBiDi, 0);
464 }
465 if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) {
466 if((lastStrongLTR>pBiDi->length) &&
467 (GET_PARALEVEL(pBiDi, lastStrongLTR)==0)) {
468 pBiDi->length = lastStrongLTR;
469 }
470 if(pBiDi->length<pBiDi->originalLength) {
471 pBiDi->paraCount--;
472 }
473 }
474 /* The following line does nothing new for contextual paraLevel, but is
475 needed for absolute paraLevel. */
476 flags|=DIRPROP_FLAG_LR(pBiDi->paraLevel);
477
478 if(pBiDi->orderParagraphsLTR && (flags&DIRPROP_FLAG(B))) {
479 flags|=DIRPROP_FLAG(L);
480 }
481
482 pBiDi->controlCount = controlCount;
483 pBiDi->flags=flags;
484 pBiDi->lastArabicPos=lastArabicPos;
485 }
486
487 /* perform (X1)..(X9) ------------------------------------------------------- */
488
489 /* determine if the text is mixed-directional or single-directional */
490 static UBiDiDirection
directionFromFlags(UBiDi * pBiDi)491 directionFromFlags(UBiDi *pBiDi) {
492 Flags flags=pBiDi->flags;
493 /* if the text contains AN and neutrals, then some neutrals may become RTL */
494 if(!(flags&MASK_RTL || ((flags&DIRPROP_FLAG(AN)) && (flags&MASK_POSSIBLE_N)))) {
495 return UBIDI_LTR;
496 } else if(!(flags&MASK_LTR)) {
497 return UBIDI_RTL;
498 } else {
499 return UBIDI_MIXED;
500 }
501 }
502
503 /*
504 * Resolve the explicit levels as specified by explicit embedding codes.
505 * Recalculate the flags to have them reflect the real properties
506 * after taking the explicit embeddings into account.
507 *
508 * The BiDi algorithm is designed to result in the same behavior whether embedding
509 * levels are externally specified (from "styled text", supposedly the preferred
510 * method) or set by explicit embedding codes (LRx, RLx, PDF) in the plain text.
511 * That is why (X9) instructs to remove all explicit codes (and BN).
512 * However, in a real implementation, this removal of these codes and their index
513 * positions in the plain text is undesirable since it would result in
514 * reallocated, reindexed text.
515 * Instead, this implementation leaves the codes in there and just ignores them
516 * in the subsequent processing.
517 * In order to get the same reordering behavior, positions with a BN or an
518 * explicit embedding code just get the same level assigned as the last "real"
519 * character.
520 *
521 * Some implementations, not this one, then overwrite some of these
522 * directionality properties at "real" same-level-run boundaries by
523 * L or R codes so that the resolution of weak types can be performed on the
524 * entire paragraph at once instead of having to parse it once more and
525 * perform that resolution on same-level-runs.
526 * This limits the scope of the implicit rules in effectively
527 * the same way as the run limits.
528 *
529 * Instead, this implementation does not modify these codes.
530 * On one hand, the paragraph has to be scanned for same-level-runs, but
531 * on the other hand, this saves another loop to reset these codes,
532 * or saves making and modifying a copy of dirProps[].
533 *
534 *
535 * Note that (Pn) and (Xn) changed significantly from version 4 of the BiDi algorithm.
536 *
537 *
538 * Handling the stack of explicit levels (Xn):
539 *
540 * With the BiDi stack of explicit levels,
541 * as pushed with each LRE, RLE, LRO, and RLO and popped with each PDF,
542 * the explicit level must never exceed UBIDI_MAX_EXPLICIT_LEVEL==61.
543 *
544 * In order to have a correct push-pop semantics even in the case of overflows,
545 * there are two overflow counters:
546 * - countOver60 is incremented with each LRx at level 60
547 * - from level 60, one RLx increases the level to 61
548 * - countOver61 is incremented with each LRx and RLx at level 61
549 *
550 * Popping levels with PDF must work in the opposite order so that level 61
551 * is correct at the correct point. Underflows (too many PDFs) must be checked.
552 *
553 * This implementation assumes that UBIDI_MAX_EXPLICIT_LEVEL is odd.
554 */
555 static UBiDiDirection
resolveExplicitLevels(UBiDi * pBiDi)556 resolveExplicitLevels(UBiDi *pBiDi) {
557 const DirProp *dirProps=pBiDi->dirProps;
558 UBiDiLevel *levels=pBiDi->levels;
559 const UChar *text=pBiDi->text;
560
561 int32_t i=0, length=pBiDi->length;
562 Flags flags=pBiDi->flags; /* collect all directionalities in the text */
563 DirProp dirProp;
564 UBiDiLevel level=GET_PARALEVEL(pBiDi, 0);
565
566 UBiDiDirection direction;
567 int32_t paraIndex=0;
568
569 /* determine if the text is mixed-directional or single-directional */
570 direction=directionFromFlags(pBiDi);
571
572 /* we may not need to resolve any explicit levels, but for multiple
573 paragraphs we want to loop on all chars to set the para boundaries */
574 if((direction!=UBIDI_MIXED) && (pBiDi->paraCount==1)) {
575 /* not mixed directionality: levels don't matter - trailingWSStart will be 0 */
576 } else if((pBiDi->paraCount==1) &&
577 (!(flags&MASK_EXPLICIT) ||
578 (pBiDi->reorderingMode > UBIDI_REORDER_LAST_LOGICAL_TO_VISUAL))) {
579 /* mixed, but all characters are at the same embedding level */
580 /* or we are in "inverse BiDi" */
581 /* and we don't have contextual multiple paragraphs with some B char */
582 /* set all levels to the paragraph level */
583 for(i=0; i<length; ++i) {
584 levels[i]=level;
585 }
586 } else {
587 /* continue to perform (Xn) */
588
589 /* (X1) level is set for all codes, embeddingLevel keeps track of the push/pop operations */
590 /* both variables may carry the UBIDI_LEVEL_OVERRIDE flag to indicate the override status */
591 UBiDiLevel embeddingLevel=level, newLevel, stackTop=0;
592
593 UBiDiLevel stack[UBIDI_MAX_EXPLICIT_LEVEL]; /* we never push anything >=UBIDI_MAX_EXPLICIT_LEVEL */
594 uint32_t countOver60=0, countOver61=0; /* count overflows of explicit levels */
595
596 /* recalculate the flags */
597 flags=0;
598
599 for(i=0; i<length; ++i) {
600 dirProp=NO_CONTEXT_RTL(dirProps[i]);
601 switch(dirProp) {
602 case LRE:
603 case LRO:
604 /* (X3, X5) */
605 newLevel=(UBiDiLevel)((embeddingLevel+2)&~(UBIDI_LEVEL_OVERRIDE|1)); /* least greater even level */
606 if(newLevel<=UBIDI_MAX_EXPLICIT_LEVEL) {
607 stack[stackTop]=embeddingLevel;
608 ++stackTop;
609 embeddingLevel=newLevel;
610 if(dirProp==LRO) {
611 embeddingLevel|=UBIDI_LEVEL_OVERRIDE;
612 }
613 /* we don't need to set UBIDI_LEVEL_OVERRIDE off for LRE
614 since this has already been done for newLevel which is
615 the source for embeddingLevel.
616 */
617 } else if((embeddingLevel&~UBIDI_LEVEL_OVERRIDE)==UBIDI_MAX_EXPLICIT_LEVEL) {
618 ++countOver61;
619 } else /* (embeddingLevel&~UBIDI_LEVEL_OVERRIDE)==UBIDI_MAX_EXPLICIT_LEVEL-1 */ {
620 ++countOver60;
621 }
622 flags|=DIRPROP_FLAG(BN);
623 break;
624 case RLE:
625 case RLO:
626 /* (X2, X4) */
627 newLevel=(UBiDiLevel)(((embeddingLevel&~UBIDI_LEVEL_OVERRIDE)+1)|1); /* least greater odd level */
628 if(newLevel<=UBIDI_MAX_EXPLICIT_LEVEL) {
629 stack[stackTop]=embeddingLevel;
630 ++stackTop;
631 embeddingLevel=newLevel;
632 if(dirProp==RLO) {
633 embeddingLevel|=UBIDI_LEVEL_OVERRIDE;
634 }
635 /* we don't need to set UBIDI_LEVEL_OVERRIDE off for RLE
636 since this has already been done for newLevel which is
637 the source for embeddingLevel.
638 */
639 } else {
640 ++countOver61;
641 }
642 flags|=DIRPROP_FLAG(BN);
643 break;
644 case PDF:
645 /* (X7) */
646 /* handle all the overflow cases first */
647 if(countOver61>0) {
648 --countOver61;
649 } else if(countOver60>0 && (embeddingLevel&~UBIDI_LEVEL_OVERRIDE)!=UBIDI_MAX_EXPLICIT_LEVEL) {
650 /* handle LRx overflows from level 60 */
651 --countOver60;
652 } else if(stackTop>0) {
653 /* this is the pop operation; it also pops level 61 while countOver60>0 */
654 --stackTop;
655 embeddingLevel=stack[stackTop];
656 /* } else { (underflow) */
657 }
658 flags|=DIRPROP_FLAG(BN);
659 break;
660 case B:
661 stackTop=0;
662 countOver60=countOver61=0;
663 level=GET_PARALEVEL(pBiDi, i);
664 if((i+1)<length) {
665 embeddingLevel=GET_PARALEVEL(pBiDi, i+1);
666 if(!((text[i]==CR) && (text[i+1]==LF))) {
667 pBiDi->paras[paraIndex++]=i+1;
668 }
669 }
670 flags|=DIRPROP_FLAG(B);
671 break;
672 case BN:
673 /* BN, LRE, RLE, and PDF are supposed to be removed (X9) */
674 /* they will get their levels set correctly in adjustWSLevels() */
675 flags|=DIRPROP_FLAG(BN);
676 break;
677 default:
678 /* all other types get the "real" level */
679 if(level!=embeddingLevel) {
680 level=embeddingLevel;
681 if(level&UBIDI_LEVEL_OVERRIDE) {
682 flags|=DIRPROP_FLAG_O(level)|DIRPROP_FLAG_MULTI_RUNS;
683 } else {
684 flags|=DIRPROP_FLAG_E(level)|DIRPROP_FLAG_MULTI_RUNS;
685 }
686 }
687 if(!(level&UBIDI_LEVEL_OVERRIDE)) {
688 flags|=DIRPROP_FLAG(dirProp);
689 }
690 break;
691 }
692
693 /*
694 * We need to set reasonable levels even on BN codes and
695 * explicit codes because we will later look at same-level runs (X10).
696 */
697 levels[i]=level;
698 }
699 if(flags&MASK_EMBEDDING) {
700 flags|=DIRPROP_FLAG_LR(pBiDi->paraLevel);
701 }
702 if(pBiDi->orderParagraphsLTR && (flags&DIRPROP_FLAG(B))) {
703 flags|=DIRPROP_FLAG(L);
704 }
705
706 /* subsequently, ignore the explicit codes and BN (X9) */
707
708 /* again, determine if the text is mixed-directional or single-directional */
709 pBiDi->flags=flags;
710 direction=directionFromFlags(pBiDi);
711 }
712
713 return direction;
714 }
715
716 /*
717 * Use a pre-specified embedding levels array:
718 *
719 * Adjust the directional properties for overrides (->LEVEL_OVERRIDE),
720 * ignore all explicit codes (X9),
721 * and check all the preset levels.
722 *
723 * Recalculate the flags to have them reflect the real properties
724 * after taking the explicit embeddings into account.
725 */
726 static UBiDiDirection
checkExplicitLevels(UBiDi * pBiDi,UErrorCode * pErrorCode)727 checkExplicitLevels(UBiDi *pBiDi, UErrorCode *pErrorCode) {
728 const DirProp *dirProps=pBiDi->dirProps;
729 DirProp dirProp;
730 UBiDiLevel *levels=pBiDi->levels;
731 const UChar *text=pBiDi->text;
732
733 int32_t i, length=pBiDi->length;
734 Flags flags=0; /* collect all directionalities in the text */
735 UBiDiLevel level;
736 uint32_t paraIndex=0;
737
738 for(i=0; i<length; ++i) {
739 level=levels[i];
740 dirProp=NO_CONTEXT_RTL(dirProps[i]);
741 if(level&UBIDI_LEVEL_OVERRIDE) {
742 /* keep the override flag in levels[i] but adjust the flags */
743 level&=~UBIDI_LEVEL_OVERRIDE; /* make the range check below simpler */
744 flags|=DIRPROP_FLAG_O(level);
745 } else {
746 /* set the flags */
747 flags|=DIRPROP_FLAG_E(level)|DIRPROP_FLAG(dirProp);
748 }
749 if((level<GET_PARALEVEL(pBiDi, i) &&
750 !((0==level)&&(dirProp==B))) ||
751 (UBIDI_MAX_EXPLICIT_LEVEL<level)) {
752 /* level out of bounds */
753 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
754 return UBIDI_LTR;
755 }
756 if((dirProp==B) && ((i+1)<length)) {
757 if(!((text[i]==CR) && (text[i+1]==LF))) {
758 pBiDi->paras[paraIndex++]=i+1;
759 }
760 }
761 }
762 if(flags&MASK_EMBEDDING) {
763 flags|=DIRPROP_FLAG_LR(pBiDi->paraLevel);
764 }
765
766 /* determine if the text is mixed-directional or single-directional */
767 pBiDi->flags=flags;
768 return directionFromFlags(pBiDi);
769 }
770
771 /******************************************************************
772 The Properties state machine table
773 *******************************************************************
774
775 All table cells are 8 bits:
776 bits 0..4: next state
777 bits 5..7: action to perform (if > 0)
778
779 Cells may be of format "n" where n represents the next state
780 (except for the rightmost column).
781 Cells may also be of format "s(x,y)" where x represents an action
782 to perform and y represents the next state.
783
784 *******************************************************************
785 Definitions and type for properties state table
786 *******************************************************************
787 */
788 #define IMPTABPROPS_COLUMNS 14
789 #define IMPTABPROPS_RES (IMPTABPROPS_COLUMNS - 1)
790 #define GET_STATEPROPS(cell) ((cell)&0x1f)
791 #define GET_ACTIONPROPS(cell) ((cell)>>5)
792 #define s(action, newState) ((uint8_t)(newState+(action<<5)))
793
794 static const uint8_t groupProp[] = /* dirProp regrouped */
795 {
796 /* L R EN ES ET AN CS B S WS ON LRE LRO AL RLE RLO PDF NSM BN */
797 0, 1, 2, 7, 8, 3, 9, 6, 5, 4, 4, 10, 10, 12, 10, 10, 10, 11, 10
798 };
799 enum { DirProp_L=0, DirProp_R=1, DirProp_EN=2, DirProp_AN=3, DirProp_ON=4, DirProp_S=5, DirProp_B=6 }; /* reduced dirProp */
800
801 /******************************************************************
802
803 PROPERTIES STATE TABLE
804
805 In table impTabProps,
806 - the ON column regroups ON and WS
807 - the BN column regroups BN, LRE, RLE, LRO, RLO, PDF
808 - the Res column is the reduced property assigned to a run
809
810 Action 1: process current run1, init new run1
811 2: init new run2
812 3: process run1, process run2, init new run1
813 4: process run1, set run1=run2, init new run2
814
815 Notes:
816 1) This table is used in resolveImplicitLevels().
817 2) This table triggers actions when there is a change in the Bidi
818 property of incoming characters (action 1).
819 3) Most such property sequences are processed immediately (in
820 fact, passed to processPropertySeq().
821 4) However, numbers are assembled as one sequence. This means
822 that undefined situations (like CS following digits, until
823 it is known if the next char will be a digit) are held until
824 following chars define them.
825 Example: digits followed by CS, then comes another CS or ON;
826 the digits will be processed, then the CS assigned
827 as the start of an ON sequence (action 3).
828 5) There are cases where more than one sequence must be
829 processed, for instance digits followed by CS followed by L:
830 the digits must be processed as one sequence, and the CS
831 must be processed as an ON sequence, all this before starting
832 assembling chars for the opening L sequence.
833
834
835 */
836 static const uint8_t impTabProps[][IMPTABPROPS_COLUMNS] =
837 {
838 /* L , R , EN , AN , ON , S , B , ES , ET , CS , BN , NSM , AL , Res */
839 /* 0 Init */ { 1 , 2 , 4 , 5 , 7 , 15 , 17 , 7 , 9 , 7 , 0 , 7 , 3 , DirProp_ON },
840 /* 1 L */ { 1 , s(1,2), s(1,4), s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), s(1,9), s(1,7), 1 , 1 , s(1,3), DirProp_L },
841 /* 2 R */ { s(1,1), 2 , s(1,4), s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), s(1,9), s(1,7), 2 , 2 , s(1,3), DirProp_R },
842 /* 3 AL */ { s(1,1), s(1,2), s(1,6), s(1,6), s(1,8),s(1,16),s(1,17), s(1,8), s(1,8), s(1,8), 3 , 3 , 3 , DirProp_R },
843 /* 4 EN */ { s(1,1), s(1,2), 4 , s(1,5), s(1,7),s(1,15),s(1,17),s(2,10), 11 ,s(2,10), 4 , 4 , s(1,3), DirProp_EN },
844 /* 5 AN */ { s(1,1), s(1,2), s(1,4), 5 , s(1,7),s(1,15),s(1,17), s(1,7), s(1,9),s(2,12), 5 , 5 , s(1,3), DirProp_AN },
845 /* 6 AL:EN/AN */ { s(1,1), s(1,2), 6 , 6 , s(1,8),s(1,16),s(1,17), s(1,8), s(1,8),s(2,13), 6 , 6 , s(1,3), DirProp_AN },
846 /* 7 ON */ { s(1,1), s(1,2), s(1,4), s(1,5), 7 ,s(1,15),s(1,17), 7 ,s(2,14), 7 , 7 , 7 , s(1,3), DirProp_ON },
847 /* 8 AL:ON */ { s(1,1), s(1,2), s(1,6), s(1,6), 8 ,s(1,16),s(1,17), 8 , 8 , 8 , 8 , 8 , s(1,3), DirProp_ON },
848 /* 9 ET */ { s(1,1), s(1,2), 4 , s(1,5), 7 ,s(1,15),s(1,17), 7 , 9 , 7 , 9 , 9 , s(1,3), DirProp_ON },
849 /*10 EN+ES/CS */ { s(3,1), s(3,2), 4 , s(3,5), s(4,7),s(3,15),s(3,17), s(4,7),s(4,14), s(4,7), 10 , s(4,7), s(3,3), DirProp_EN },
850 /*11 EN+ET */ { s(1,1), s(1,2), 4 , s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), 11 , s(1,7), 11 , 11 , s(1,3), DirProp_EN },
851 /*12 AN+CS */ { s(3,1), s(3,2), s(3,4), 5 , s(4,7),s(3,15),s(3,17), s(4,7),s(4,14), s(4,7), 12 , s(4,7), s(3,3), DirProp_AN },
852 /*13 AL:EN/AN+CS */ { s(3,1), s(3,2), 6 , 6 , s(4,8),s(3,16),s(3,17), s(4,8), s(4,8), s(4,8), 13 , s(4,8), s(3,3), DirProp_AN },
853 /*14 ON+ET */ { s(1,1), s(1,2), s(4,4), s(1,5), 7 ,s(1,15),s(1,17), 7 , 14 , 7 , 14 , 14 , s(1,3), DirProp_ON },
854 /*15 S */ { s(1,1), s(1,2), s(1,4), s(1,5), s(1,7), 15 ,s(1,17), s(1,7), s(1,9), s(1,7), 15 , s(1,7), s(1,3), DirProp_S },
855 /*16 AL:S */ { s(1,1), s(1,2), s(1,6), s(1,6), s(1,8), 16 ,s(1,17), s(1,8), s(1,8), s(1,8), 16 , s(1,8), s(1,3), DirProp_S },
856 /*17 B */ { s(1,1), s(1,2), s(1,4), s(1,5), s(1,7),s(1,15), 17 , s(1,7), s(1,9), s(1,7), 17 , s(1,7), s(1,3), DirProp_B }
857 };
858
859 /* we must undef macro s because the levels table have a different
860 * structure (4 bits for action and 4 bits for next state.
861 */
862 #undef s
863
864 /******************************************************************
865 The levels state machine tables
866 *******************************************************************
867
868 All table cells are 8 bits:
869 bits 0..3: next state
870 bits 4..7: action to perform (if > 0)
871
872 Cells may be of format "n" where n represents the next state
873 (except for the rightmost column).
874 Cells may also be of format "s(x,y)" where x represents an action
875 to perform and y represents the next state.
876
877 This format limits each table to 16 states each and to 15 actions.
878
879 *******************************************************************
880 Definitions and type for levels state tables
881 *******************************************************************
882 */
883 #define IMPTABLEVELS_COLUMNS (DirProp_B + 2)
884 #define IMPTABLEVELS_RES (IMPTABLEVELS_COLUMNS - 1)
885 #define GET_STATE(cell) ((cell)&0x0f)
886 #define GET_ACTION(cell) ((cell)>>4)
887 #define s(action, newState) ((uint8_t)(newState+(action<<4)))
888
889 typedef uint8_t ImpTab[][IMPTABLEVELS_COLUMNS];
890 typedef uint8_t ImpAct[];
891
892 /* FOOD FOR THOUGHT: each ImpTab should have its associated ImpAct,
893 * instead of having a pair of ImpTab and a pair of ImpAct.
894 */
895 typedef struct ImpTabPair {
896 const void * pImpTab[2];
897 const void * pImpAct[2];
898 } ImpTabPair;
899
900 /******************************************************************
901
902 LEVELS STATE TABLES
903
904 In all levels state tables,
905 - state 0 is the initial state
906 - the Res column is the increment to add to the text level
907 for this property sequence.
908
909 The impAct arrays for each table of a pair map the local action
910 numbers of the table to the total list of actions. For instance,
911 action 2 in a given table corresponds to the action number which
912 appears in entry [2] of the impAct array for that table.
913 The first entry of all impAct arrays must be 0.
914
915 Action 1: init conditional sequence
916 2: prepend conditional sequence to current sequence
917 3: set ON sequence to new level - 1
918 4: init EN/AN/ON sequence
919 5: fix EN/AN/ON sequence followed by R
920 6: set previous level sequence to level 2
921
922 Notes:
923 1) These tables are used in processPropertySeq(). The input
924 is property sequences as determined by resolveImplicitLevels.
925 2) Most such property sequences are processed immediately
926 (levels are assigned).
927 3) However, some sequences cannot be assigned a final level till
928 one or more following sequences are received. For instance,
929 ON following an R sequence within an even-level paragraph.
930 If the following sequence is R, the ON sequence will be
931 assigned basic run level+1, and so will the R sequence.
932 4) S is generally handled like ON, since its level will be fixed
933 to paragraph level in adjustWSLevels().
934
935 */
936
937 static const ImpTab impTabL_DEFAULT = /* Even paragraph level */
938 /* In this table, conditional sequences receive the higher possible level
939 until proven otherwise.
940 */
941 {
942 /* L , R , EN , AN , ON , S , B , Res */
943 /* 0 : init */ { 0 , 1 , 0 , 2 , 0 , 0 , 0 , 0 },
944 /* 1 : R */ { 0 , 1 , 3 , 3 , s(1,4), s(1,4), 0 , 1 },
945 /* 2 : AN */ { 0 , 1 , 0 , 2 , s(1,5), s(1,5), 0 , 2 },
946 /* 3 : R+EN/AN */ { 0 , 1 , 3 , 3 , s(1,4), s(1,4), 0 , 2 },
947 /* 4 : R+ON */ { s(2,0), 1 , 3 , 3 , 4 , 4 , s(2,0), 1 },
948 /* 5 : AN+ON */ { s(2,0), 1 , s(2,0), 2 , 5 , 5 , s(2,0), 1 }
949 };
950 static const ImpTab impTabR_DEFAULT = /* Odd paragraph level */
951 /* In this table, conditional sequences receive the lower possible level
952 until proven otherwise.
953 */
954 {
955 /* L , R , EN , AN , ON , S , B , Res */
956 /* 0 : init */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 0 },
957 /* 1 : L */ { 1 , 0 , 1 , 3 , s(1,4), s(1,4), 0 , 1 },
958 /* 2 : EN/AN */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 1 },
959 /* 3 : L+AN */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 1 },
960 /* 4 : L+ON */ { s(2,1), 0 , s(2,1), 3 , 4 , 4 , 0 , 0 },
961 /* 5 : L+AN+ON */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 0 }
962 };
963 static const ImpAct impAct0 = {0,1,2,3,4,5,6};
964 static const ImpTabPair impTab_DEFAULT = {{&impTabL_DEFAULT,
965 &impTabR_DEFAULT},
966 {&impAct0, &impAct0}};
967
968 static const ImpTab impTabL_NUMBERS_SPECIAL = /* Even paragraph level */
969 /* In this table, conditional sequences receive the higher possible level
970 until proven otherwise.
971 */
972 {
973 /* L , R , EN , AN , ON , S , B , Res */
974 /* 0 : init */ { 0 , 2 , 1 , 1 , 0 , 0 , 0 , 0 },
975 /* 1 : L+EN/AN */ { 0 , 2 , 1 , 1 , 0 , 0 , 0 , 2 },
976 /* 2 : R */ { 0 , 2 , 4 , 4 , s(1,3), 0 , 0 , 1 },
977 /* 3 : R+ON */ { s(2,0), 2 , 4 , 4 , 3 , 3 , s(2,0), 1 },
978 /* 4 : R+EN/AN */ { 0 , 2 , 4 , 4 , s(1,3), s(1,3), 0 , 2 }
979 };
980 static const ImpTabPair impTab_NUMBERS_SPECIAL = {{&impTabL_NUMBERS_SPECIAL,
981 &impTabR_DEFAULT},
982 {&impAct0, &impAct0}};
983
984 static const ImpTab impTabL_GROUP_NUMBERS_WITH_R =
985 /* In this table, EN/AN+ON sequences receive levels as if associated with R
986 until proven that there is L or sor/eor on both sides. AN is handled like EN.
987 */
988 {
989 /* L , R , EN , AN , ON , S , B , Res */
990 /* 0 init */ { 0 , 3 , s(1,1), s(1,1), 0 , 0 , 0 , 0 },
991 /* 1 EN/AN */ { s(2,0), 3 , 1 , 1 , 2 , s(2,0), s(2,0), 2 },
992 /* 2 EN/AN+ON */ { s(2,0), 3 , 1 , 1 , 2 , s(2,0), s(2,0), 1 },
993 /* 3 R */ { 0 , 3 , 5 , 5 , s(1,4), 0 , 0 , 1 },
994 /* 4 R+ON */ { s(2,0), 3 , 5 , 5 , 4 , s(2,0), s(2,0), 1 },
995 /* 5 R+EN/AN */ { 0 , 3 , 5 , 5 , s(1,4), 0 , 0 , 2 }
996 };
997 static const ImpTab impTabR_GROUP_NUMBERS_WITH_R =
998 /* In this table, EN/AN+ON sequences receive levels as if associated with R
999 until proven that there is L on both sides. AN is handled like EN.
1000 */
1001 {
1002 /* L , R , EN , AN , ON , S , B , Res */
1003 /* 0 init */ { 2 , 0 , 1 , 1 , 0 , 0 , 0 , 0 },
1004 /* 1 EN/AN */ { 2 , 0 , 1 , 1 , 0 , 0 , 0 , 1 },
1005 /* 2 L */ { 2 , 0 , s(1,4), s(1,4), s(1,3), 0 , 0 , 1 },
1006 /* 3 L+ON */ { s(2,2), 0 , 4 , 4 , 3 , 0 , 0 , 0 },
1007 /* 4 L+EN/AN */ { s(2,2), 0 , 4 , 4 , 3 , 0 , 0 , 1 }
1008 };
1009 static const ImpTabPair impTab_GROUP_NUMBERS_WITH_R = {
1010 {&impTabL_GROUP_NUMBERS_WITH_R,
1011 &impTabR_GROUP_NUMBERS_WITH_R},
1012 {&impAct0, &impAct0}};
1013
1014
1015 static const ImpTab impTabL_INVERSE_NUMBERS_AS_L =
1016 /* This table is identical to the Default LTR table except that EN and AN are
1017 handled like L.
1018 */
1019 {
1020 /* L , R , EN , AN , ON , S , B , Res */
1021 /* 0 : init */ { 0 , 1 , 0 , 0 , 0 , 0 , 0 , 0 },
1022 /* 1 : R */ { 0 , 1 , 0 , 0 , s(1,4), s(1,4), 0 , 1 },
1023 /* 2 : AN */ { 0 , 1 , 0 , 0 , s(1,5), s(1,5), 0 , 2 },
1024 /* 3 : R+EN/AN */ { 0 , 1 , 0 , 0 , s(1,4), s(1,4), 0 , 2 },
1025 /* 4 : R+ON */ { s(2,0), 1 , s(2,0), s(2,0), 4 , 4 , s(2,0), 1 },
1026 /* 5 : AN+ON */ { s(2,0), 1 , s(2,0), s(2,0), 5 , 5 , s(2,0), 1 }
1027 };
1028 static const ImpTab impTabR_INVERSE_NUMBERS_AS_L =
1029 /* This table is identical to the Default RTL table except that EN and AN are
1030 handled like L.
1031 */
1032 {
1033 /* L , R , EN , AN , ON , S , B , Res */
1034 /* 0 : init */ { 1 , 0 , 1 , 1 , 0 , 0 , 0 , 0 },
1035 /* 1 : L */ { 1 , 0 , 1 , 1 , s(1,4), s(1,4), 0 , 1 },
1036 /* 2 : EN/AN */ { 1 , 0 , 1 , 1 , 0 , 0 , 0 , 1 },
1037 /* 3 : L+AN */ { 1 , 0 , 1 , 1 , 5 , 5 , 0 , 1 },
1038 /* 4 : L+ON */ { s(2,1), 0 , s(2,1), s(2,1), 4 , 4 , 0 , 0 },
1039 /* 5 : L+AN+ON */ { 1 , 0 , 1 , 1 , 5 , 5 , 0 , 0 }
1040 };
1041 static const ImpTabPair impTab_INVERSE_NUMBERS_AS_L = {
1042 {&impTabL_INVERSE_NUMBERS_AS_L,
1043 &impTabR_INVERSE_NUMBERS_AS_L},
1044 {&impAct0, &impAct0}};
1045
1046 static const ImpTab impTabR_INVERSE_LIKE_DIRECT = /* Odd paragraph level */
1047 /* In this table, conditional sequences receive the lower possible level
1048 until proven otherwise.
1049 */
1050 {
1051 /* L , R , EN , AN , ON , S , B , Res */
1052 /* 0 : init */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 0 },
1053 /* 1 : L */ { 1 , 0 , 1 , 2 , s(1,3), s(1,3), 0 , 1 },
1054 /* 2 : EN/AN */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 1 },
1055 /* 3 : L+ON */ { s(2,1), s(3,0), 6 , 4 , 3 , 3 , s(3,0), 0 },
1056 /* 4 : L+ON+AN */ { s(2,1), s(3,0), 6 , 4 , 5 , 5 , s(3,0), 3 },
1057 /* 5 : L+AN+ON */ { s(2,1), s(3,0), 6 , 4 , 5 , 5 , s(3,0), 2 },
1058 /* 6 : L+ON+EN */ { s(2,1), s(3,0), 6 , 4 , 3 , 3 , s(3,0), 1 }
1059 };
1060 static const ImpAct impAct1 = {0,1,11,12};
1061 /* FOOD FOR THOUGHT: in LTR table below, check case "JKL 123abc"
1062 */
1063 static const ImpTabPair impTab_INVERSE_LIKE_DIRECT = {
1064 {&impTabL_DEFAULT,
1065 &impTabR_INVERSE_LIKE_DIRECT},
1066 {&impAct0, &impAct1}};
1067
1068 static const ImpTab impTabL_INVERSE_LIKE_DIRECT_WITH_MARKS =
1069 /* The case handled in this table is (visually): R EN L
1070 */
1071 {
1072 /* L , R , EN , AN , ON , S , B , Res */
1073 /* 0 : init */ { 0 , s(6,3), 0 , 1 , 0 , 0 , 0 , 0 },
1074 /* 1 : L+AN */ { 0 , s(6,3), 0 , 1 , s(1,2), s(3,0), 0 , 4 },
1075 /* 2 : L+AN+ON */ { s(2,0), s(6,3), s(2,0), 1 , 2 , s(3,0), s(2,0), 3 },
1076 /* 3 : R */ { 0 , s(6,3), s(5,5), s(5,6), s(1,4), s(3,0), 0 , 3 },
1077 /* 4 : R+ON */ { s(3,0), s(4,3), s(5,5), s(5,6), 4 , s(3,0), s(3,0), 3 },
1078 /* 5 : R+EN */ { s(3,0), s(4,3), 5 , s(5,6), s(1,4), s(3,0), s(3,0), 4 },
1079 /* 6 : R+AN */ { s(3,0), s(4,3), s(5,5), 6 , s(1,4), s(3,0), s(3,0), 4 }
1080 };
1081 static const ImpTab impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS =
1082 /* The cases handled in this table are (visually): R EN L
1083 R L AN L
1084 */
1085 {
1086 /* L , R , EN , AN , ON , S , B , Res */
1087 /* 0 : init */ { s(1,3), 0 , 1 , 1 , 0 , 0 , 0 , 0 },
1088 /* 1 : R+EN/AN */ { s(2,3), 0 , 1 , 1 , 2 , s(4,0), 0 , 1 },
1089 /* 2 : R+EN/AN+ON */ { s(2,3), 0 , 1 , 1 , 2 , s(4,0), 0 , 0 },
1090 /* 3 : L */ { 3 , 0 , 3 , s(3,6), s(1,4), s(4,0), 0 , 1 },
1091 /* 4 : L+ON */ { s(5,3), s(4,0), 5 , s(3,6), 4 , s(4,0), s(4,0), 0 },
1092 /* 5 : L+ON+EN */ { s(5,3), s(4,0), 5 , s(3,6), 4 , s(4,0), s(4,0), 1 },
1093 /* 6 : L+AN */ { s(5,3), s(4,0), 6 , 6 , 4 , s(4,0), s(4,0), 3 }
1094 };
1095 static const ImpAct impAct2 = {0,1,7,8,9,10};
1096 static const ImpTabPair impTab_INVERSE_LIKE_DIRECT_WITH_MARKS = {
1097 {&impTabL_INVERSE_LIKE_DIRECT_WITH_MARKS,
1098 &impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS},
1099 {&impAct0, &impAct2}};
1100
1101 static const ImpTabPair impTab_INVERSE_FOR_NUMBERS_SPECIAL = {
1102 {&impTabL_NUMBERS_SPECIAL,
1103 &impTabR_INVERSE_LIKE_DIRECT},
1104 {&impAct0, &impAct1}};
1105
1106 static const ImpTab impTabL_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS =
1107 /* The case handled in this table is (visually): R EN L
1108 */
1109 {
1110 /* L , R , EN , AN , ON , S , B , Res */
1111 /* 0 : init */ { 0 , s(6,2), 1 , 1 , 0 , 0 , 0 , 0 },
1112 /* 1 : L+EN/AN */ { 0 , s(6,2), 1 , 1 , 0 , s(3,0), 0 , 4 },
1113 /* 2 : R */ { 0 , s(6,2), s(5,4), s(5,4), s(1,3), s(3,0), 0 , 3 },
1114 /* 3 : R+ON */ { s(3,0), s(4,2), s(5,4), s(5,4), 3 , s(3,0), s(3,0), 3 },
1115 /* 4 : R+EN/AN */ { s(3,0), s(4,2), 4 , 4 , s(1,3), s(3,0), s(3,0), 4 }
1116 };
1117 static const ImpTabPair impTab_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS = {
1118 {&impTabL_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS,
1119 &impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS},
1120 {&impAct0, &impAct2}};
1121
1122 #undef s
1123
1124 typedef struct {
1125 const ImpTab * pImpTab; /* level table pointer */
1126 const ImpAct * pImpAct; /* action map array */
1127 int32_t startON; /* start of ON sequence */
1128 int32_t startL2EN; /* start of level 2 sequence */
1129 int32_t lastStrongRTL; /* index of last found R or AL */
1130 int32_t state; /* current state */
1131 UBiDiLevel runLevel; /* run level before implicit solving */
1132 } LevState;
1133
1134 /*------------------------------------------------------------------------*/
1135
1136 static void
addPoint(UBiDi * pBiDi,int32_t pos,int32_t flag)1137 addPoint(UBiDi *pBiDi, int32_t pos, int32_t flag)
1138 /* param pos: position where to insert
1139 param flag: one of LRM_BEFORE, LRM_AFTER, RLM_BEFORE, RLM_AFTER
1140 */
1141 {
1142 #define FIRSTALLOC 10
1143 Point point;
1144 InsertPoints * pInsertPoints=&(pBiDi->insertPoints);
1145
1146 if (pInsertPoints->capacity == 0)
1147 {
1148 pInsertPoints->points=uprv_malloc(sizeof(Point)*FIRSTALLOC);
1149 if (pInsertPoints->points == NULL)
1150 {
1151 pInsertPoints->errorCode=U_MEMORY_ALLOCATION_ERROR;
1152 return;
1153 }
1154 pInsertPoints->capacity=FIRSTALLOC;
1155 }
1156 if (pInsertPoints->size >= pInsertPoints->capacity) /* no room for new point */
1157 {
1158 void * savePoints=pInsertPoints->points;
1159 pInsertPoints->points=uprv_realloc(pInsertPoints->points,
1160 pInsertPoints->capacity*2*sizeof(Point));
1161 if (pInsertPoints->points == NULL)
1162 {
1163 pInsertPoints->points=savePoints;
1164 pInsertPoints->errorCode=U_MEMORY_ALLOCATION_ERROR;
1165 return;
1166 }
1167 else pInsertPoints->capacity*=2;
1168 }
1169 point.pos=pos;
1170 point.flag=flag;
1171 pInsertPoints->points[pInsertPoints->size]=point;
1172 pInsertPoints->size++;
1173 #undef FIRSTALLOC
1174 }
1175
1176 /* perform rules (Wn), (Nn), and (In) on a run of the text ------------------ */
1177
1178 /*
1179 * This implementation of the (Wn) rules applies all rules in one pass.
1180 * In order to do so, it needs a look-ahead of typically 1 character
1181 * (except for W5: sequences of ET) and keeps track of changes
1182 * in a rule Wp that affect a later Wq (p<q).
1183 *
1184 * The (Nn) and (In) rules are also performed in that same single loop,
1185 * but effectively one iteration behind for white space.
1186 *
1187 * Since all implicit rules are performed in one step, it is not necessary
1188 * to actually store the intermediate directional properties in dirProps[].
1189 */
1190
1191 static void
processPropertySeq(UBiDi * pBiDi,LevState * pLevState,uint8_t _prop,int32_t start,int32_t limit)1192 processPropertySeq(UBiDi *pBiDi, LevState *pLevState, uint8_t _prop,
1193 int32_t start, int32_t limit) {
1194 uint8_t cell, oldStateSeq, actionSeq;
1195 const ImpTab * pImpTab=pLevState->pImpTab;
1196 const ImpAct * pImpAct=pLevState->pImpAct;
1197 UBiDiLevel * levels=pBiDi->levels;
1198 UBiDiLevel level, addLevel;
1199 InsertPoints * pInsertPoints;
1200 int32_t start0, k;
1201
1202 start0=start; /* save original start position */
1203 oldStateSeq=(uint8_t)pLevState->state;
1204 cell=(*pImpTab)[oldStateSeq][_prop];
1205 pLevState->state=GET_STATE(cell); /* isolate the new state */
1206 actionSeq=(*pImpAct)[GET_ACTION(cell)]; /* isolate the action */
1207 addLevel=(*pImpTab)[pLevState->state][IMPTABLEVELS_RES];
1208
1209 if(actionSeq) {
1210 switch(actionSeq) {
1211 case 1: /* init ON seq */
1212 pLevState->startON=start0;
1213 break;
1214
1215 case 2: /* prepend ON seq to current seq */
1216 start=pLevState->startON;
1217 break;
1218
1219 case 3: /* L or S after possible relevant EN/AN */
1220 /* check if we had EN after R/AL */
1221 if (pLevState->startL2EN >= 0) {
1222 addPoint(pBiDi, pLevState->startL2EN, LRM_BEFORE);
1223 }
1224 pLevState->startL2EN=-1; /* not within previous if since could also be -2 */
1225 /* check if we had any relevant EN/AN after R/AL */
1226 pInsertPoints=&(pBiDi->insertPoints);
1227 if ((pInsertPoints->capacity == 0) ||
1228 (pInsertPoints->size <= pInsertPoints->confirmed))
1229 {
1230 /* nothing, just clean up */
1231 pLevState->lastStrongRTL=-1;
1232 /* check if we have a pending conditional segment */
1233 level=(*pImpTab)[oldStateSeq][IMPTABLEVELS_RES];
1234 if ((level & 1) && (pLevState->startON > 0)) { /* after ON */
1235 start=pLevState->startON; /* reset to basic run level */
1236 }
1237 if (_prop == DirProp_S) /* add LRM before S */
1238 {
1239 addPoint(pBiDi, start0, LRM_BEFORE);
1240 pInsertPoints->confirmed=pInsertPoints->size;
1241 }
1242 break;
1243 }
1244 /* reset previous RTL cont to level for LTR text */
1245 for (k=pLevState->lastStrongRTL+1; k<start0; k++)
1246 {
1247 /* reset odd level, leave runLevel+2 as is */
1248 levels[k]=(levels[k] - 2) & ~1;
1249 }
1250 /* mark insert points as confirmed */
1251 pInsertPoints->confirmed=pInsertPoints->size;
1252 pLevState->lastStrongRTL=-1;
1253 if (_prop == DirProp_S) /* add LRM before S */
1254 {
1255 addPoint(pBiDi, start0, LRM_BEFORE);
1256 pInsertPoints->confirmed=pInsertPoints->size;
1257 }
1258 break;
1259
1260 case 4: /* R/AL after possible relevant EN/AN */
1261 /* just clean up */
1262 pInsertPoints=&(pBiDi->insertPoints);
1263 if (pInsertPoints->capacity > 0)
1264 /* remove all non confirmed insert points */
1265 pInsertPoints->size=pInsertPoints->confirmed;
1266 pLevState->startON=-1;
1267 pLevState->startL2EN=-1;
1268 pLevState->lastStrongRTL=limit - 1;
1269 break;
1270
1271 case 5: /* EN/AN after R/AL + possible cont */
1272 /* check for real AN */
1273 if ((_prop == DirProp_AN) && (NO_CONTEXT_RTL(pBiDi->dirProps[start0]) == AN) &&
1274 (pBiDi->reorderingMode!=UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL))
1275 {
1276 /* real AN */
1277 if (pLevState->startL2EN == -1) /* if no relevant EN already found */
1278 {
1279 /* just note the righmost digit as a strong RTL */
1280 pLevState->lastStrongRTL=limit - 1;
1281 break;
1282 }
1283 if (pLevState->startL2EN >= 0) /* after EN, no AN */
1284 {
1285 addPoint(pBiDi, pLevState->startL2EN, LRM_BEFORE);
1286 pLevState->startL2EN=-2;
1287 }
1288 /* note AN */
1289 addPoint(pBiDi, start0, LRM_BEFORE);
1290 break;
1291 }
1292 /* if first EN/AN after R/AL */
1293 if (pLevState->startL2EN == -1) {
1294 pLevState->startL2EN=start0;
1295 }
1296 break;
1297
1298 case 6: /* note location of latest R/AL */
1299 pLevState->lastStrongRTL=limit - 1;
1300 pLevState->startON=-1;
1301 break;
1302
1303 case 7: /* L after R+ON/EN/AN */
1304 /* include possible adjacent number on the left */
1305 for (k=start0-1; k>=0 && !(levels[k]&1); k--);
1306 if(k>=0) {
1307 addPoint(pBiDi, k, RLM_BEFORE); /* add RLM before */
1308 pInsertPoints=&(pBiDi->insertPoints);
1309 pInsertPoints->confirmed=pInsertPoints->size; /* confirm it */
1310 }
1311 pLevState->startON=start0;
1312 break;
1313
1314 case 8: /* AN after L */
1315 /* AN numbers between L text on both sides may be trouble. */
1316 /* tentatively bracket with LRMs; will be confirmed if followed by L */
1317 addPoint(pBiDi, start0, LRM_BEFORE); /* add LRM before */
1318 addPoint(pBiDi, start0, LRM_AFTER); /* add LRM after */
1319 break;
1320
1321 case 9: /* R after L+ON/EN/AN */
1322 /* false alert, infirm LRMs around previous AN */
1323 pInsertPoints=&(pBiDi->insertPoints);
1324 pInsertPoints->size=pInsertPoints->confirmed;
1325 if (_prop == DirProp_S) /* add RLM before S */
1326 {
1327 addPoint(pBiDi, start0, RLM_BEFORE);
1328 pInsertPoints->confirmed=pInsertPoints->size;
1329 }
1330 break;
1331
1332 case 10: /* L after L+ON/AN */
1333 level=pLevState->runLevel + addLevel;
1334 for(k=pLevState->startON; k<start0; k++) {
1335 if (levels[k]<level)
1336 levels[k]=level;
1337 }
1338 pInsertPoints=&(pBiDi->insertPoints);
1339 pInsertPoints->confirmed=pInsertPoints->size; /* confirm inserts */
1340 pLevState->startON=start0;
1341 break;
1342
1343 case 11: /* L after L+ON+EN/AN/ON */
1344 level=pLevState->runLevel;
1345 for(k=start0-1; k>=pLevState->startON; k--) {
1346 if(levels[k]==level+3) {
1347 while(levels[k]==level+3) {
1348 levels[k--]-=2;
1349 }
1350 while(levels[k]==level) {
1351 k--;
1352 }
1353 }
1354 if(levels[k]==level+2) {
1355 levels[k]=level;
1356 continue;
1357 }
1358 levels[k]=level+1;
1359 }
1360 break;
1361
1362 case 12: /* R after L+ON+EN/AN/ON */
1363 level=pLevState->runLevel+1;
1364 for(k=start0-1; k>=pLevState->startON; k--) {
1365 if(levels[k]>level) {
1366 levels[k]-=2;
1367 }
1368 }
1369 break;
1370
1371 default: /* we should never get here */
1372 U_ASSERT(FALSE);
1373 break;
1374 }
1375 }
1376 if((addLevel) || (start < start0)) {
1377 level=pLevState->runLevel + addLevel;
1378 for(k=start; k<limit; k++) {
1379 levels[k]=level;
1380 }
1381 }
1382 }
1383
1384 static void
resolveImplicitLevels(UBiDi * pBiDi,int32_t start,int32_t limit,DirProp sor,DirProp eor)1385 resolveImplicitLevels(UBiDi *pBiDi,
1386 int32_t start, int32_t limit,
1387 DirProp sor, DirProp eor) {
1388 const DirProp *dirProps=pBiDi->dirProps;
1389
1390 LevState levState;
1391 int32_t i, start1, start2;
1392 uint8_t oldStateImp, stateImp, actionImp;
1393 uint8_t gprop, resProp, cell;
1394 UBool inverseRTL;
1395 DirProp nextStrongProp=R;
1396 int32_t nextStrongPos=-1;
1397
1398 levState.startON = -1; /* silence gcc flow analysis */
1399
1400 /* check for RTL inverse BiDi mode */
1401 /* FOOD FOR THOUGHT: in case of RTL inverse BiDi, it would make sense to
1402 * loop on the text characters from end to start.
1403 * This would need a different properties state table (at least different
1404 * actions) and different levels state tables (maybe very similar to the
1405 * LTR corresponding ones.
1406 */
1407 inverseRTL=(UBool)
1408 ((start<pBiDi->lastArabicPos) && (GET_PARALEVEL(pBiDi, start) & 1) &&
1409 (pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_LIKE_DIRECT ||
1410 pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL));
1411 /* initialize for levels state table */
1412 levState.startL2EN=-1; /* used for INVERSE_LIKE_DIRECT_WITH_MARKS */
1413 levState.lastStrongRTL=-1; /* used for INVERSE_LIKE_DIRECT_WITH_MARKS */
1414 levState.state=0;
1415 levState.runLevel=pBiDi->levels[start];
1416 levState.pImpTab=(const ImpTab*)((pBiDi->pImpTabPair)->pImpTab)[levState.runLevel&1];
1417 levState.pImpAct=(const ImpAct*)((pBiDi->pImpTabPair)->pImpAct)[levState.runLevel&1];
1418 processPropertySeq(pBiDi, &levState, sor, start, start);
1419 /* initialize for property state table */
1420 if(NO_CONTEXT_RTL(dirProps[start])==NSM) {
1421 stateImp = 1 + sor;
1422 } else {
1423 stateImp=0;
1424 }
1425 start1=start;
1426 start2=start;
1427
1428 for(i=start; i<=limit; i++) {
1429 if(i>=limit) {
1430 gprop=eor;
1431 } else {
1432 DirProp prop, prop1;
1433 prop=NO_CONTEXT_RTL(dirProps[i]);
1434 if(inverseRTL) {
1435 if(prop==AL) {
1436 /* AL before EN does not make it AN */
1437 prop=R;
1438 } else if(prop==EN) {
1439 if(nextStrongPos<=i) {
1440 /* look for next strong char (L/R/AL) */
1441 int32_t j;
1442 nextStrongProp=R; /* set default */
1443 nextStrongPos=limit;
1444 for(j=i+1; j<limit; j++) {
1445 prop1=NO_CONTEXT_RTL(dirProps[j]);
1446 if(prop1==L || prop1==R || prop1==AL) {
1447 nextStrongProp=prop1;
1448 nextStrongPos=j;
1449 break;
1450 }
1451 }
1452 }
1453 if(nextStrongProp==AL) {
1454 prop=AN;
1455 }
1456 }
1457 }
1458 gprop=groupProp[prop];
1459 }
1460 oldStateImp=stateImp;
1461 cell=impTabProps[oldStateImp][gprop];
1462 stateImp=GET_STATEPROPS(cell); /* isolate the new state */
1463 actionImp=GET_ACTIONPROPS(cell); /* isolate the action */
1464 if((i==limit) && (actionImp==0)) {
1465 /* there is an unprocessed sequence if its property == eor */
1466 actionImp=1; /* process the last sequence */
1467 }
1468 if(actionImp) {
1469 resProp=impTabProps[oldStateImp][IMPTABPROPS_RES];
1470 switch(actionImp) {
1471 case 1: /* process current seq1, init new seq1 */
1472 processPropertySeq(pBiDi, &levState, resProp, start1, i);
1473 start1=i;
1474 break;
1475 case 2: /* init new seq2 */
1476 start2=i;
1477 break;
1478 case 3: /* process seq1, process seq2, init new seq1 */
1479 processPropertySeq(pBiDi, &levState, resProp, start1, start2);
1480 processPropertySeq(pBiDi, &levState, DirProp_ON, start2, i);
1481 start1=i;
1482 break;
1483 case 4: /* process seq1, set seq1=seq2, init new seq2 */
1484 processPropertySeq(pBiDi, &levState, resProp, start1, start2);
1485 start1=start2;
1486 start2=i;
1487 break;
1488 default: /* we should never get here */
1489 U_ASSERT(FALSE);
1490 break;
1491 }
1492 }
1493 }
1494 /* flush possible pending sequence, e.g. ON */
1495 processPropertySeq(pBiDi, &levState, eor, limit, limit);
1496 }
1497
1498 /* perform (L1) and (X9) ---------------------------------------------------- */
1499
1500 /*
1501 * Reset the embedding levels for some non-graphic characters (L1).
1502 * This function also sets appropriate levels for BN, and
1503 * explicit embedding types that are supposed to have been removed
1504 * from the paragraph in (X9).
1505 */
1506 static void
adjustWSLevels(UBiDi * pBiDi)1507 adjustWSLevels(UBiDi *pBiDi) {
1508 const DirProp *dirProps=pBiDi->dirProps;
1509 UBiDiLevel *levels=pBiDi->levels;
1510 int32_t i;
1511
1512 if(pBiDi->flags&MASK_WS) {
1513 UBool orderParagraphsLTR=pBiDi->orderParagraphsLTR;
1514 Flags flag;
1515
1516 i=pBiDi->trailingWSStart;
1517 while(i>0) {
1518 /* reset a sequence of WS/BN before eop and B/S to the paragraph paraLevel */
1519 while(i>0 && (flag=DIRPROP_FLAG_NC(dirProps[--i]))&MASK_WS) {
1520 if(orderParagraphsLTR&&(flag&DIRPROP_FLAG(B))) {
1521 levels[i]=0;
1522 } else {
1523 levels[i]=GET_PARALEVEL(pBiDi, i);
1524 }
1525 }
1526
1527 /* reset BN to the next character's paraLevel until B/S, which restarts above loop */
1528 /* here, i+1 is guaranteed to be <length */
1529 while(i>0) {
1530 flag=DIRPROP_FLAG_NC(dirProps[--i]);
1531 if(flag&MASK_BN_EXPLICIT) {
1532 levels[i]=levels[i+1];
1533 } else if(orderParagraphsLTR&&(flag&DIRPROP_FLAG(B))) {
1534 levels[i]=0;
1535 break;
1536 } else if(flag&MASK_B_S) {
1537 levels[i]=GET_PARALEVEL(pBiDi, i);
1538 break;
1539 }
1540 }
1541 }
1542 }
1543 }
1544
1545 #define BIDI_MIN(x, y) ((x)<(y) ? (x) : (y))
1546 #define BIDI_ABS(x) ((x)>=0 ? (x) : (-(x)))
1547 static void
setParaRunsOnly(UBiDi * pBiDi,const UChar * text,int32_t length,UBiDiLevel paraLevel,UErrorCode * pErrorCode)1548 setParaRunsOnly(UBiDi *pBiDi, const UChar *text, int32_t length,
1549 UBiDiLevel paraLevel, UErrorCode *pErrorCode) {
1550 void *runsOnlyMemory;
1551 int32_t *visualMap;
1552 UChar *visualText;
1553 int32_t saveLength, saveTrailingWSStart;
1554 const UBiDiLevel *levels;
1555 UBiDiLevel *saveLevels;
1556 UBiDiDirection saveDirection;
1557 UBool saveMayAllocateText;
1558 Run *runs;
1559 int32_t visualLength, i, j, visualStart, logicalStart,
1560 runCount, runLength, addedRuns, insertRemove,
1561 start, limit, step, indexOddBit, logicalPos,
1562 index0, index1;
1563 uint32_t saveOptions;
1564
1565 pBiDi->reorderingMode=UBIDI_REORDER_DEFAULT;
1566 if(length==0) {
1567 ubidi_setPara(pBiDi, text, length, paraLevel, NULL, pErrorCode);
1568 goto cleanup3;
1569 }
1570 /* obtain memory for mapping table and visual text */
1571 runsOnlyMemory=uprv_malloc(length*(sizeof(int32_t)+sizeof(UChar)+sizeof(UBiDiLevel)));
1572 if(runsOnlyMemory==NULL) {
1573 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
1574 goto cleanup3;
1575 }
1576 visualMap=runsOnlyMemory;
1577 visualText=(UChar *)&visualMap[length];
1578 saveLevels=(UBiDiLevel *)&visualText[length];
1579 saveOptions=pBiDi->reorderingOptions;
1580 if(saveOptions & UBIDI_OPTION_INSERT_MARKS) {
1581 pBiDi->reorderingOptions&=~UBIDI_OPTION_INSERT_MARKS;
1582 pBiDi->reorderingOptions|=UBIDI_OPTION_REMOVE_CONTROLS;
1583 }
1584 paraLevel&=1; /* accept only 0 or 1 */
1585 ubidi_setPara(pBiDi, text, length, paraLevel, NULL, pErrorCode);
1586 if(U_FAILURE(*pErrorCode)) {
1587 goto cleanup3;
1588 }
1589 /* we cannot access directly pBiDi->levels since it is not yet set if
1590 * direction is not MIXED
1591 */
1592 levels=ubidi_getLevels(pBiDi, pErrorCode);
1593 uprv_memcpy(saveLevels, levels, pBiDi->length*sizeof(UBiDiLevel));
1594 saveTrailingWSStart=pBiDi->trailingWSStart;
1595 saveLength=pBiDi->length;
1596 saveDirection=pBiDi->direction;
1597
1598 /* FOOD FOR THOUGHT: instead of writing the visual text, we could use
1599 * the visual map and the dirProps array to drive the second call
1600 * to ubidi_setPara (but must make provision for possible removal of
1601 * BiDi controls. Alternatively, only use the dirProps array via
1602 * customized classifier callback.
1603 */
1604 visualLength=ubidi_writeReordered(pBiDi, visualText, length,
1605 UBIDI_DO_MIRRORING, pErrorCode);
1606 ubidi_getVisualMap(pBiDi, visualMap, pErrorCode);
1607 if(U_FAILURE(*pErrorCode)) {
1608 goto cleanup2;
1609 }
1610 pBiDi->reorderingOptions=saveOptions;
1611
1612 pBiDi->reorderingMode=UBIDI_REORDER_INVERSE_LIKE_DIRECT;
1613 paraLevel^=1;
1614 /* Because what we did with reorderingOptions, visualText may be shorter
1615 * than the original text. But we don't want the levels memory to be
1616 * reallocated shorter than the original length, since we need to restore
1617 * the levels as after the first call to ubidi_setpara() before returning.
1618 * We will force mayAllocateText to FALSE before the second call to
1619 * ubidi_setpara(), and will restore it afterwards.
1620 */
1621 saveMayAllocateText=pBiDi->mayAllocateText;
1622 pBiDi->mayAllocateText=FALSE;
1623 ubidi_setPara(pBiDi, visualText, visualLength, paraLevel, NULL, pErrorCode);
1624 pBiDi->mayAllocateText=saveMayAllocateText;
1625 ubidi_getRuns(pBiDi, pErrorCode);
1626 if(U_FAILURE(*pErrorCode)) {
1627 goto cleanup1;
1628 }
1629 /* check if some runs must be split, count how many splits */
1630 addedRuns=0;
1631 runCount=pBiDi->runCount;
1632 runs=pBiDi->runs;
1633 visualStart=0;
1634 for(i=0; i<runCount; i++, visualStart+=runLength) {
1635 runLength=runs[i].visualLimit-visualStart;
1636 if(runLength<2) {
1637 continue;
1638 }
1639 logicalStart=GET_INDEX(runs[i].logicalStart);
1640 for(j=logicalStart+1; j<logicalStart+runLength; j++) {
1641 index0=visualMap[j];
1642 index1=visualMap[j-1];
1643 if((BIDI_ABS(index0-index1)!=1) || (saveLevels[index0]!=saveLevels[index1])) {
1644 addedRuns++;
1645 }
1646 }
1647 }
1648 if(addedRuns) {
1649 if(getRunsMemory(pBiDi, runCount+addedRuns)) {
1650 if(runCount==1) {
1651 /* because we switch from UBiDi.simpleRuns to UBiDi.runs */
1652 pBiDi->runsMemory[0]=runs[0];
1653 }
1654 runs=pBiDi->runs=pBiDi->runsMemory;
1655 pBiDi->runCount+=addedRuns;
1656 } else {
1657 goto cleanup1;
1658 }
1659 }
1660 /* split runs which are not consecutive in source text */
1661 for(i=runCount-1; i>=0; i--) {
1662 runLength= i==0 ? runs[0].visualLimit :
1663 runs[i].visualLimit-runs[i-1].visualLimit;
1664 logicalStart=runs[i].logicalStart;
1665 indexOddBit=GET_ODD_BIT(logicalStart);
1666 logicalStart=GET_INDEX(logicalStart);
1667 if(runLength<2) {
1668 if(addedRuns) {
1669 runs[i+addedRuns]=runs[i];
1670 }
1671 logicalPos=visualMap[logicalStart];
1672 runs[i+addedRuns].logicalStart=MAKE_INDEX_ODD_PAIR(logicalPos,
1673 saveLevels[logicalPos]^indexOddBit);
1674 continue;
1675 }
1676 if(indexOddBit) {
1677 start=logicalStart;
1678 limit=logicalStart+runLength-1;
1679 step=1;
1680 } else {
1681 start=logicalStart+runLength-1;
1682 limit=logicalStart;
1683 step=-1;
1684 }
1685 for(j=start; j!=limit; j+=step) {
1686 index0=visualMap[j];
1687 index1=visualMap[j+step];
1688 if((BIDI_ABS(index0-index1)!=1) || (saveLevels[index0]!=saveLevels[index1])) {
1689 logicalPos=BIDI_MIN(visualMap[start], index0);
1690 runs[i+addedRuns].logicalStart=MAKE_INDEX_ODD_PAIR(logicalPos,
1691 saveLevels[logicalPos]^indexOddBit);
1692 runs[i+addedRuns].visualLimit=runs[i].visualLimit;
1693 runs[i].visualLimit-=BIDI_ABS(j-start)+1;
1694 insertRemove=runs[i].insertRemove&(LRM_AFTER|RLM_AFTER);
1695 runs[i+addedRuns].insertRemove=insertRemove;
1696 runs[i].insertRemove&=~insertRemove;
1697 start=j+step;
1698 addedRuns--;
1699 }
1700 }
1701 if(addedRuns) {
1702 runs[i+addedRuns]=runs[i];
1703 }
1704 logicalPos=BIDI_MIN(visualMap[start], visualMap[limit]);
1705 runs[i+addedRuns].logicalStart=MAKE_INDEX_ODD_PAIR(logicalPos,
1706 saveLevels[logicalPos]^indexOddBit);
1707 }
1708
1709 cleanup1:
1710 /* restore initial paraLevel */
1711 pBiDi->paraLevel^=1;
1712 cleanup2:
1713 /* restore real text */
1714 pBiDi->text=text;
1715 pBiDi->length=saveLength;
1716 pBiDi->originalLength=length;
1717 pBiDi->direction=saveDirection;
1718 /* the saved levels should never excess levelsSize, but we check anyway */
1719 if(saveLength>pBiDi->levelsSize) {
1720 saveLength=pBiDi->levelsSize;
1721 }
1722 uprv_memcpy(pBiDi->levels, saveLevels, saveLength*sizeof(UBiDiLevel));
1723 pBiDi->trailingWSStart=saveTrailingWSStart;
1724 /* free memory for mapping table and visual text */
1725 uprv_free(runsOnlyMemory);
1726 if(pBiDi->runCount>1) {
1727 pBiDi->direction=UBIDI_MIXED;
1728 }
1729 cleanup3:
1730 pBiDi->reorderingMode=UBIDI_REORDER_RUNS_ONLY;
1731 }
1732
1733 /* ubidi_setPara ------------------------------------------------------------ */
1734
1735 U_CAPI void U_EXPORT2
ubidi_setPara(UBiDi * pBiDi,const UChar * text,int32_t length,UBiDiLevel paraLevel,UBiDiLevel * embeddingLevels,UErrorCode * pErrorCode)1736 ubidi_setPara(UBiDi *pBiDi, const UChar *text, int32_t length,
1737 UBiDiLevel paraLevel, UBiDiLevel *embeddingLevels,
1738 UErrorCode *pErrorCode) {
1739 UBiDiDirection direction;
1740
1741 /* check the argument values */
1742 RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode);
1743 if(pBiDi==NULL || text==NULL || length<-1 ||
1744 (paraLevel>UBIDI_MAX_EXPLICIT_LEVEL && paraLevel<UBIDI_DEFAULT_LTR)) {
1745 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1746 return;
1747 }
1748
1749 if(length==-1) {
1750 length=u_strlen(text);
1751 }
1752
1753 /* special treatment for RUNS_ONLY mode */
1754 if(pBiDi->reorderingMode==UBIDI_REORDER_RUNS_ONLY) {
1755 setParaRunsOnly(pBiDi, text, length, paraLevel, pErrorCode);
1756 return;
1757 }
1758
1759 /* initialize the UBiDi structure */
1760 pBiDi->pParaBiDi=NULL; /* mark unfinished setPara */
1761 pBiDi->text=text;
1762 pBiDi->length=pBiDi->originalLength=pBiDi->resultLength=length;
1763 pBiDi->paraLevel=paraLevel;
1764 pBiDi->direction=UBIDI_LTR;
1765 pBiDi->paraCount=1;
1766
1767 pBiDi->dirProps=NULL;
1768 pBiDi->levels=NULL;
1769 pBiDi->runs=NULL;
1770 pBiDi->insertPoints.size=0; /* clean up from last call */
1771 pBiDi->insertPoints.confirmed=0; /* clean up from last call */
1772
1773 /*
1774 * Save the original paraLevel if contextual; otherwise, set to 0.
1775 */
1776 if(IS_DEFAULT_LEVEL(paraLevel)) {
1777 pBiDi->defaultParaLevel=paraLevel;
1778 } else {
1779 pBiDi->defaultParaLevel=0;
1780 }
1781
1782 if(length==0) {
1783 /*
1784 * For an empty paragraph, create a UBiDi object with the paraLevel and
1785 * the flags and the direction set but without allocating zero-length arrays.
1786 * There is nothing more to do.
1787 */
1788 if(IS_DEFAULT_LEVEL(paraLevel)) {
1789 pBiDi->paraLevel&=1;
1790 pBiDi->defaultParaLevel=0;
1791 }
1792 if(paraLevel&1) {
1793 pBiDi->flags=DIRPROP_FLAG(R);
1794 pBiDi->direction=UBIDI_RTL;
1795 } else {
1796 pBiDi->flags=DIRPROP_FLAG(L);
1797 pBiDi->direction=UBIDI_LTR;
1798 }
1799
1800 pBiDi->runCount=0;
1801 pBiDi->paraCount=0;
1802 pBiDi->pParaBiDi=pBiDi; /* mark successful setPara */
1803 return;
1804 }
1805
1806 pBiDi->runCount=-1;
1807
1808 /*
1809 * Get the directional properties,
1810 * the flags bit-set, and
1811 * determine the paragraph level if necessary.
1812 */
1813 if(getDirPropsMemory(pBiDi, length)) {
1814 pBiDi->dirProps=pBiDi->dirPropsMemory;
1815 getDirProps(pBiDi);
1816 } else {
1817 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
1818 return;
1819 }
1820 /* the processed length may have changed if UBIDI_OPTION_STREAMING */
1821 length= pBiDi->length;
1822 pBiDi->trailingWSStart=length; /* the levels[] will reflect the WS run */
1823 /* allocate paras memory */
1824 if(pBiDi->paraCount>1) {
1825 if(getInitialParasMemory(pBiDi, pBiDi->paraCount)) {
1826 pBiDi->paras=pBiDi->parasMemory;
1827 pBiDi->paras[pBiDi->paraCount-1]=length;
1828 } else {
1829 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
1830 return;
1831 }
1832 } else {
1833 /* initialize paras for single paragraph */
1834 pBiDi->paras=pBiDi->simpleParas;
1835 pBiDi->simpleParas[0]=length;
1836 }
1837
1838 /* are explicit levels specified? */
1839 if(embeddingLevels==NULL) {
1840 /* no: determine explicit levels according to the (Xn) rules */\
1841 if(getLevelsMemory(pBiDi, length)) {
1842 pBiDi->levels=pBiDi->levelsMemory;
1843 direction=resolveExplicitLevels(pBiDi);
1844 } else {
1845 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
1846 return;
1847 }
1848 } else {
1849 /* set BN for all explicit codes, check that all levels are 0 or paraLevel..UBIDI_MAX_EXPLICIT_LEVEL */
1850 pBiDi->levels=embeddingLevels;
1851 direction=checkExplicitLevels(pBiDi, pErrorCode);
1852 if(U_FAILURE(*pErrorCode)) {
1853 return;
1854 }
1855 }
1856
1857 /*
1858 * The steps after (X9) in the UBiDi algorithm are performed only if
1859 * the paragraph text has mixed directionality!
1860 */
1861 pBiDi->direction=direction;
1862 switch(direction) {
1863 case UBIDI_LTR:
1864 /* make sure paraLevel is even */
1865 pBiDi->paraLevel=(UBiDiLevel)((pBiDi->paraLevel+1)&~1);
1866
1867 /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */
1868 pBiDi->trailingWSStart=0;
1869 break;
1870 case UBIDI_RTL:
1871 /* make sure paraLevel is odd */
1872 pBiDi->paraLevel|=1;
1873
1874 /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */
1875 pBiDi->trailingWSStart=0;
1876 break;
1877 default:
1878 /*
1879 * Choose the right implicit state table
1880 */
1881 switch(pBiDi->reorderingMode) {
1882 case UBIDI_REORDER_DEFAULT:
1883 pBiDi->pImpTabPair=&impTab_DEFAULT;
1884 break;
1885 case UBIDI_REORDER_NUMBERS_SPECIAL:
1886 pBiDi->pImpTabPair=&impTab_NUMBERS_SPECIAL;
1887 break;
1888 case UBIDI_REORDER_GROUP_NUMBERS_WITH_R:
1889 pBiDi->pImpTabPair=&impTab_GROUP_NUMBERS_WITH_R;
1890 break;
1891 case UBIDI_REORDER_INVERSE_NUMBERS_AS_L:
1892 pBiDi->pImpTabPair=&impTab_INVERSE_NUMBERS_AS_L;
1893 break;
1894 case UBIDI_REORDER_INVERSE_LIKE_DIRECT:
1895 if (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) {
1896 pBiDi->pImpTabPair=&impTab_INVERSE_LIKE_DIRECT_WITH_MARKS;
1897 } else {
1898 pBiDi->pImpTabPair=&impTab_INVERSE_LIKE_DIRECT;
1899 }
1900 break;
1901 case UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL:
1902 if (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) {
1903 pBiDi->pImpTabPair=&impTab_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS;
1904 } else {
1905 pBiDi->pImpTabPair=&impTab_INVERSE_FOR_NUMBERS_SPECIAL;
1906 }
1907 break;
1908 default:
1909 /* we should never get here */
1910 U_ASSERT(FALSE);
1911 break;
1912 }
1913 /*
1914 * If there are no external levels specified and there
1915 * are no significant explicit level codes in the text,
1916 * then we can treat the entire paragraph as one run.
1917 * Otherwise, we need to perform the following rules on runs of
1918 * the text with the same embedding levels. (X10)
1919 * "Significant" explicit level codes are ones that actually
1920 * affect non-BN characters.
1921 * Examples for "insignificant" ones are empty embeddings
1922 * LRE-PDF, LRE-RLE-PDF-PDF, etc.
1923 */
1924 if(embeddingLevels==NULL && pBiDi->paraCount<=1 &&
1925 !(pBiDi->flags&DIRPROP_FLAG_MULTI_RUNS)) {
1926 resolveImplicitLevels(pBiDi, 0, length,
1927 GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, 0)),
1928 GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, length-1)));
1929 } else {
1930 /* sor, eor: start and end types of same-level-run */
1931 UBiDiLevel *levels=pBiDi->levels;
1932 int32_t start, limit=0;
1933 UBiDiLevel level, nextLevel;
1934 DirProp sor, eor;
1935
1936 /* determine the first sor and set eor to it because of the loop body (sor=eor there) */
1937 level=GET_PARALEVEL(pBiDi, 0);
1938 nextLevel=levels[0];
1939 if(level<nextLevel) {
1940 eor=GET_LR_FROM_LEVEL(nextLevel);
1941 } else {
1942 eor=GET_LR_FROM_LEVEL(level);
1943 }
1944
1945 do {
1946 /* determine start and limit of the run (end points just behind the run) */
1947
1948 /* the values for this run's start are the same as for the previous run's end */
1949 start=limit;
1950 level=nextLevel;
1951 if((start>0) && (NO_CONTEXT_RTL(pBiDi->dirProps[start-1])==B)) {
1952 /* except if this is a new paragraph, then set sor = para level */
1953 sor=GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, start));
1954 } else {
1955 sor=eor;
1956 }
1957
1958 /* search for the limit of this run */
1959 while(++limit<length && levels[limit]==level) {}
1960
1961 /* get the correct level of the next run */
1962 if(limit<length) {
1963 nextLevel=levels[limit];
1964 } else {
1965 nextLevel=GET_PARALEVEL(pBiDi, length-1);
1966 }
1967
1968 /* determine eor from max(level, nextLevel); sor is last run's eor */
1969 if((level&~UBIDI_LEVEL_OVERRIDE)<(nextLevel&~UBIDI_LEVEL_OVERRIDE)) {
1970 eor=GET_LR_FROM_LEVEL(nextLevel);
1971 } else {
1972 eor=GET_LR_FROM_LEVEL(level);
1973 }
1974
1975 /* if the run consists of overridden directional types, then there
1976 are no implicit types to be resolved */
1977 if(!(level&UBIDI_LEVEL_OVERRIDE)) {
1978 resolveImplicitLevels(pBiDi, start, limit, sor, eor);
1979 } else {
1980 /* remove the UBIDI_LEVEL_OVERRIDE flags */
1981 do {
1982 levels[start++]&=~UBIDI_LEVEL_OVERRIDE;
1983 } while(start<limit);
1984 }
1985 } while(limit<length);
1986 }
1987 /* check if we got any memory shortage while adding insert points */
1988 if (U_FAILURE(pBiDi->insertPoints.errorCode))
1989 {
1990 *pErrorCode=pBiDi->insertPoints.errorCode;
1991 return;
1992 }
1993 /* reset the embedding levels for some non-graphic characters (L1), (X9) */
1994 adjustWSLevels(pBiDi);
1995 break;
1996 }
1997 /* add RLM for inverse Bidi with contextual orientation resolving
1998 * to RTL which would not round-trip otherwise
1999 */
2000 if((pBiDi->defaultParaLevel>0) &&
2001 (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) &&
2002 ((pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_LIKE_DIRECT) ||
2003 (pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL))) {
2004 int32_t i, j, start, last;
2005 DirProp dirProp;
2006 for(i=0; i<pBiDi->paraCount; i++) {
2007 last=pBiDi->paras[i]-1;
2008 if((pBiDi->dirProps[last] & CONTEXT_RTL)==0) {
2009 continue; /* LTR paragraph */
2010 }
2011 start= i==0 ? 0 : pBiDi->paras[i - 1];
2012 for(j=last; j>=start; j--) {
2013 dirProp=NO_CONTEXT_RTL(pBiDi->dirProps[j]);
2014 if(dirProp==L) {
2015 if(j<last) {
2016 while(NO_CONTEXT_RTL(pBiDi->dirProps[last])==B) {
2017 last--;
2018 }
2019 }
2020 addPoint(pBiDi, last, RLM_BEFORE);
2021 break;
2022 }
2023 if(DIRPROP_FLAG(dirProp) & MASK_R_AL) {
2024 break;
2025 }
2026 }
2027 }
2028 }
2029
2030 if(pBiDi->reorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) {
2031 pBiDi->resultLength -= pBiDi->controlCount;
2032 } else {
2033 pBiDi->resultLength += pBiDi->insertPoints.size;
2034 }
2035 pBiDi->pParaBiDi=pBiDi; /* mark successful setPara */
2036 }
2037
2038 U_CAPI void U_EXPORT2
ubidi_orderParagraphsLTR(UBiDi * pBiDi,UBool orderParagraphsLTR)2039 ubidi_orderParagraphsLTR(UBiDi *pBiDi, UBool orderParagraphsLTR) {
2040 if(pBiDi!=NULL) {
2041 pBiDi->orderParagraphsLTR=orderParagraphsLTR;
2042 }
2043 }
2044
2045 U_CAPI UBool U_EXPORT2
ubidi_isOrderParagraphsLTR(UBiDi * pBiDi)2046 ubidi_isOrderParagraphsLTR(UBiDi *pBiDi) {
2047 if(pBiDi!=NULL) {
2048 return pBiDi->orderParagraphsLTR;
2049 } else {
2050 return FALSE;
2051 }
2052 }
2053
2054 U_CAPI UBiDiDirection U_EXPORT2
ubidi_getDirection(const UBiDi * pBiDi)2055 ubidi_getDirection(const UBiDi *pBiDi) {
2056 if(IS_VALID_PARA_OR_LINE(pBiDi)) {
2057 return pBiDi->direction;
2058 } else {
2059 return UBIDI_LTR;
2060 }
2061 }
2062
2063 U_CAPI const UChar * U_EXPORT2
ubidi_getText(const UBiDi * pBiDi)2064 ubidi_getText(const UBiDi *pBiDi) {
2065 if(IS_VALID_PARA_OR_LINE(pBiDi)) {
2066 return pBiDi->text;
2067 } else {
2068 return NULL;
2069 }
2070 }
2071
2072 U_CAPI int32_t U_EXPORT2
ubidi_getLength(const UBiDi * pBiDi)2073 ubidi_getLength(const UBiDi *pBiDi) {
2074 if(IS_VALID_PARA_OR_LINE(pBiDi)) {
2075 return pBiDi->originalLength;
2076 } else {
2077 return 0;
2078 }
2079 }
2080
2081 U_CAPI int32_t U_EXPORT2
ubidi_getProcessedLength(const UBiDi * pBiDi)2082 ubidi_getProcessedLength(const UBiDi *pBiDi) {
2083 if(IS_VALID_PARA_OR_LINE(pBiDi)) {
2084 return pBiDi->length;
2085 } else {
2086 return 0;
2087 }
2088 }
2089
2090 U_CAPI int32_t U_EXPORT2
ubidi_getResultLength(const UBiDi * pBiDi)2091 ubidi_getResultLength(const UBiDi *pBiDi) {
2092 if(IS_VALID_PARA_OR_LINE(pBiDi)) {
2093 return pBiDi->resultLength;
2094 } else {
2095 return 0;
2096 }
2097 }
2098
2099 /* paragraphs API functions ------------------------------------------------- */
2100
2101 U_CAPI UBiDiLevel U_EXPORT2
ubidi_getParaLevel(const UBiDi * pBiDi)2102 ubidi_getParaLevel(const UBiDi *pBiDi) {
2103 if(IS_VALID_PARA_OR_LINE(pBiDi)) {
2104 return pBiDi->paraLevel;
2105 } else {
2106 return 0;
2107 }
2108 }
2109
2110 U_CAPI int32_t U_EXPORT2
ubidi_countParagraphs(UBiDi * pBiDi)2111 ubidi_countParagraphs(UBiDi *pBiDi) {
2112 if(!IS_VALID_PARA_OR_LINE(pBiDi)) {
2113 return 0;
2114 } else {
2115 return pBiDi->paraCount;
2116 }
2117 }
2118
2119 U_CAPI void U_EXPORT2
ubidi_getParagraphByIndex(const UBiDi * pBiDi,int32_t paraIndex,int32_t * pParaStart,int32_t * pParaLimit,UBiDiLevel * pParaLevel,UErrorCode * pErrorCode)2120 ubidi_getParagraphByIndex(const UBiDi *pBiDi, int32_t paraIndex,
2121 int32_t *pParaStart, int32_t *pParaLimit,
2122 UBiDiLevel *pParaLevel, UErrorCode *pErrorCode) {
2123 int32_t paraStart;
2124
2125 /* check the argument values */
2126 RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode);
2127 RETURN_VOID_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode);
2128 RETURN_VOID_IF_BAD_RANGE(paraIndex, 0, pBiDi->paraCount, *pErrorCode);
2129
2130 pBiDi=pBiDi->pParaBiDi; /* get Para object if Line object */
2131 if(paraIndex) {
2132 paraStart=pBiDi->paras[paraIndex-1];
2133 } else {
2134 paraStart=0;
2135 }
2136 if(pParaStart!=NULL) {
2137 *pParaStart=paraStart;
2138 }
2139 if(pParaLimit!=NULL) {
2140 *pParaLimit=pBiDi->paras[paraIndex];
2141 }
2142 if(pParaLevel!=NULL) {
2143 *pParaLevel=GET_PARALEVEL(pBiDi, paraStart);
2144 }
2145 }
2146
2147 U_CAPI int32_t U_EXPORT2
ubidi_getParagraph(const UBiDi * pBiDi,int32_t charIndex,int32_t * pParaStart,int32_t * pParaLimit,UBiDiLevel * pParaLevel,UErrorCode * pErrorCode)2148 ubidi_getParagraph(const UBiDi *pBiDi, int32_t charIndex,
2149 int32_t *pParaStart, int32_t *pParaLimit,
2150 UBiDiLevel *pParaLevel, UErrorCode *pErrorCode) {
2151 uint32_t paraIndex;
2152
2153 /* check the argument values */
2154 /* pErrorCode will be checked by the call to ubidi_getParagraphByIndex */
2155 RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1);
2156 RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1);
2157 pBiDi=pBiDi->pParaBiDi; /* get Para object if Line object */
2158 RETURN_IF_BAD_RANGE(charIndex, 0, pBiDi->length, *pErrorCode, -1);
2159
2160 for(paraIndex=0; charIndex>=pBiDi->paras[paraIndex]; paraIndex++);
2161 ubidi_getParagraphByIndex(pBiDi, paraIndex, pParaStart, pParaLimit, pParaLevel, pErrorCode);
2162 return paraIndex;
2163 }
2164
2165 U_CAPI void U_EXPORT2
ubidi_setClassCallback(UBiDi * pBiDi,UBiDiClassCallback * newFn,const void * newContext,UBiDiClassCallback ** oldFn,const void ** oldContext,UErrorCode * pErrorCode)2166 ubidi_setClassCallback(UBiDi *pBiDi, UBiDiClassCallback *newFn,
2167 const void *newContext, UBiDiClassCallback **oldFn,
2168 const void **oldContext, UErrorCode *pErrorCode)
2169 {
2170 RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode);
2171 if(pBiDi==NULL) {
2172 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
2173 return;
2174 }
2175 if( oldFn )
2176 {
2177 *oldFn = pBiDi->fnClassCallback;
2178 }
2179 if( oldContext )
2180 {
2181 *oldContext = pBiDi->coClassCallback;
2182 }
2183 pBiDi->fnClassCallback = newFn;
2184 pBiDi->coClassCallback = newContext;
2185 }
2186
2187 U_CAPI void U_EXPORT2
ubidi_getClassCallback(UBiDi * pBiDi,UBiDiClassCallback ** fn,const void ** context)2188 ubidi_getClassCallback(UBiDi *pBiDi, UBiDiClassCallback **fn, const void **context)
2189 {
2190 if(pBiDi==NULL) {
2191 return;
2192 }
2193 if( fn )
2194 {
2195 *fn = pBiDi->fnClassCallback;
2196 }
2197 if( context )
2198 {
2199 *context = pBiDi->coClassCallback;
2200 }
2201 }
2202
2203 U_CAPI UCharDirection U_EXPORT2
ubidi_getCustomizedClass(UBiDi * pBiDi,UChar32 c)2204 ubidi_getCustomizedClass(UBiDi *pBiDi, UChar32 c)
2205 {
2206 UCharDirection dir;
2207
2208 if( pBiDi->fnClassCallback == NULL ||
2209 (dir = (*pBiDi->fnClassCallback)(pBiDi->coClassCallback, c)) == U_BIDI_CLASS_DEFAULT )
2210 {
2211 return ubidi_getClass(pBiDi->bdp, c);
2212 } else {
2213 return dir;
2214 }
2215 }
2216
2217