U_CAPI UBiDiDirection U_EXPORT2
ubidi_getVisualRun(UBiDi *pBiDi, int32_t runIndex,
int32_t *pLogicalStart, int32_t *pLength)
{
int32_t start;
UErrorCode errorCode = U_ZERO_ERROR;
RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, errorCode, UBIDI_LTR);
ubidi_getRuns(pBiDi, &errorCode);
if(U_FAILURE(errorCode)) {
return UBIDI_LTR;
}
RETURN_IF_BAD_RANGE(runIndex, 0, pBiDi->runCount, errorCode, UBIDI_LTR);
start=pBiDi->runs[runIndex].logicalStart;
if(pLogicalStart!=NULL) {
*pLogicalStart=GET_INDEX(start);
}
if(pLength!=NULL) {
if(runIndex>0) {
*pLength=pBiDi->runs[runIndex].visualLimit-
pBiDi->runs[runIndex-1].visualLimit;
} else {
*pLength=pBiDi->runs[0].visualLimit;
}
}
return (UBiDiDirection)GET_ODD_BIT(start);
}
U_CAPI int32_t U_EXPORT2
ubidi_countRuns(UBiDi *pBiDi, UErrorCode *pErrorCode) {
RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1);
RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1);
ubidi_getRuns(pBiDi, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
return -1;
}
return pBiDi->runCount;
}
U_CAPI int32_t U_EXPORT2
ubidi_countRuns(UBiDi *pBiDi, UErrorCode *pErrorCode) {
if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
return -1;
} else if(pBiDi==NULL || (pBiDi->runCount<0 && !ubidi_getRuns(pBiDi))) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return -1;
} else {
return pBiDi->runCount;
}
}
U_CAPI int32_t U_EXPORT2
ubidi_getVisualIndex(UBiDi *pBiDi, int32_t logicalIndex, UErrorCode *pErrorCode) {
if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
return 0;
} else if(pBiDi==NULL) {
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
return 0;
} else if(logicalIndex<0 || pBiDi->length<=logicalIndex) {
*pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
return 0;
} else {
/* we can do the trivial cases without the runs array */
switch(pBiDi->direction) {
case UBIDI_LTR:
return logicalIndex;
case UBIDI_RTL:
return pBiDi->length-logicalIndex-1;
default:
if(pBiDi->runCount<0 && !ubidi_getRuns(pBiDi)) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return 0;
} else {
Run *runs=pBiDi->runs;
int32_t i, visualStart=0, offset, length;
/* linear search for the run, search on the visual runs */
for(i=0;; ++i) {
length=runs[i].visualLimit-visualStart;
offset=logicalIndex-GET_INDEX(runs[i].logicalStart);
if(offset>=0 && offset<length) {
if(IS_EVEN_RUN(runs[i].logicalStart)) {
/* LTR */
return visualStart+offset;
} else {
/* RTL */
return visualStart+length-offset-1;
}
}
visualStart+=length;
}
}
}
}
}
U_CAPI UBiDiDirection U_EXPORT2
ubidi_getVisualRun(UBiDi *pBiDi, int32_t runIndex,
int32_t *pLogicalStart, int32_t *pLength) {
if( pBiDi==NULL || runIndex<0 ||
(pBiDi->runCount==-1 && !ubidi_getRuns(pBiDi)) ||
runIndex>=pBiDi->runCount
) {
return UBIDI_LTR;
} else {
int32_t start=pBiDi->runs[runIndex].logicalStart;
if(pLogicalStart!=NULL) {
*pLogicalStart=GET_INDEX(start);
}
if(pLength!=NULL) {
if(runIndex>0) {
*pLength=pBiDi->runs[runIndex].visualLimit-
pBiDi->runs[runIndex-1].visualLimit;
} else {
*pLength=pBiDi->runs[0].visualLimit;
}
}
return (UBiDiDirection)GET_ODD_BIT(start);
}
}
U_CAPI int32_t U_EXPORT2
ubidi_getLogicalIndex(UBiDi *pBiDi, int32_t visualIndex, UErrorCode *pErrorCode) {
Run *runs;
int32_t i, runCount, start;
RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1);
RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1);
RETURN_IF_BAD_RANGE(visualIndex, 0, pBiDi->resultLength, *pErrorCode, -1);
/* we can do the trivial cases without the runs array */
if(pBiDi->insertPoints.size==0 && pBiDi->controlCount==0) {
if(pBiDi->direction==UBIDI_LTR) {
return visualIndex;
}
else if(pBiDi->direction==UBIDI_RTL) {
return pBiDi->length-visualIndex-1;
}
}
if(!ubidi_getRuns(pBiDi, pErrorCode)) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return -1;
}
runs=pBiDi->runs;
runCount=pBiDi->runCount;
if(pBiDi->insertPoints.size>0) {
/* handle inserted LRM/RLM */
int32_t markFound=0, insertRemove;
int32_t visualStart=0, length;
runs=pBiDi->runs;
/* subtract number of marks until visual index */
for(i=0; ; i++, visualStart+=length) {
length=runs[i].visualLimit-visualStart;
insertRemove=runs[i].insertRemove;
if(insertRemove&(LRM_BEFORE|RLM_BEFORE)) {
if(visualIndex<=(visualStart+markFound)) {
return UBIDI_MAP_NOWHERE;
}
markFound++;
}
/* is adjusted visual index within this run? */
if(visualIndex<(runs[i].visualLimit+markFound)) {
visualIndex-=markFound;
break;
}
if(insertRemove&(LRM_AFTER|RLM_AFTER)) {
if(visualIndex==(visualStart+length+markFound)) {
return UBIDI_MAP_NOWHERE;
}
markFound++;
}
}
}
else if(pBiDi->controlCount>0) {
/* handle removed BiDi control characters */
int32_t controlFound=0, insertRemove, length;
int32_t logicalStart, logicalEnd, visualStart=0, j, k;
UChar uchar;
UBool evenRun;
/* add number of controls until visual index */
for(i=0; ; i++, visualStart+=length) {
length=runs[i].visualLimit-visualStart;
insertRemove=runs[i].insertRemove;
/* is adjusted visual index beyond current run? */
if(visualIndex>=(runs[i].visualLimit-controlFound+insertRemove)) {
controlFound-=insertRemove;
continue;
}
/* adjusted visual index is within current run */
if(insertRemove==0) {
visualIndex+=controlFound;
break;
}
/* count non-control chars until visualIndex */
logicalStart=runs[i].logicalStart;
evenRun=IS_EVEN_RUN(logicalStart);
REMOVE_ODD_BIT(logicalStart);
logicalEnd=logicalStart+length-1;
for(j=0; j<length; j++) {
k= evenRun ? logicalStart+j : logicalEnd-j;
uchar=pBiDi->text[k];
if(IS_BIDI_CONTROL_CHAR(uchar)) {
controlFound++;
}
if((visualIndex+controlFound)==(visualStart+j)) {
break;
}
}
visualIndex+=controlFound;
break;
}
}
/* handle all cases */
if(runCount<=10) {
/* linear search for the run */
for(i=0; visualIndex>=runs[i].visualLimit; ++i) {}
} else {
/* binary search for the run */
int32_t begin=0, limit=runCount;
/* the middle if() is guaranteed to find the run, we don't need a loop limit */
for(;;) {
i=(begin+limit)/2;
if(visualIndex>=runs[i].visualLimit) {
begin=i+1;
} else if(i==0 || visualIndex>=runs[i-1].visualLimit) {
break;
} else {
limit=i;
}
}
}
start=runs[i].logicalStart;
if(IS_EVEN_RUN(start)) {
/* LTR */
/* the offset in runs[i] is visualIndex-runs[i-1].visualLimit */
if(i>0) {
visualIndex-=runs[i-1].visualLimit;
}
return start+visualIndex;
} else {
/* RTL */
return GET_INDEX(start)+runs[i].visualLimit-visualIndex-1;
}
}
U_CAPI int32_t U_EXPORT2
ubidi_getVisualIndex(UBiDi *pBiDi, int32_t logicalIndex, UErrorCode *pErrorCode) {
int32_t visualIndex=UBIDI_MAP_NOWHERE;
RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1);
RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1);
RETURN_IF_BAD_RANGE(logicalIndex, 0, pBiDi->length, *pErrorCode, -1);
/* we can do the trivial cases without the runs array */
switch(pBiDi->direction) {
case UBIDI_LTR:
visualIndex=logicalIndex;
break;
case UBIDI_RTL:
visualIndex=pBiDi->length-logicalIndex-1;
break;
default:
if(!ubidi_getRuns(pBiDi, pErrorCode)) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return -1;
} else {
Run *runs=pBiDi->runs;
int32_t i, visualStart=0, offset, length;
/* linear search for the run, search on the visual runs */
for(i=0; i<pBiDi->runCount; ++i) {
length=runs[i].visualLimit-visualStart;
offset=logicalIndex-GET_INDEX(runs[i].logicalStart);
if(offset>=0 && offset<length) {
if(IS_EVEN_RUN(runs[i].logicalStart)) {
/* LTR */
visualIndex=visualStart+offset;
} else {
/* RTL */
visualIndex=visualStart+length-offset-1;
}
break; /* exit for loop */
}
visualStart+=length;
}
if(i>=pBiDi->runCount) {
return UBIDI_MAP_NOWHERE;
}
}
}
if(pBiDi->insertPoints.size>0) {
/* add the number of added marks until the calculated visual index */
Run *runs=pBiDi->runs;
int32_t i, length, insertRemove;
int32_t visualStart=0, markFound=0;
for(i=0; ; i++, visualStart+=length) {
length=runs[i].visualLimit-visualStart;
insertRemove=runs[i].insertRemove;
if(insertRemove & (LRM_BEFORE|RLM_BEFORE)) {
markFound++;
}
/* is it the run containing the visual index? */
if(visualIndex<runs[i].visualLimit) {
return visualIndex+markFound;
}
if(insertRemove & (LRM_AFTER|RLM_AFTER)) {
markFound++;
}
}
}
else if(pBiDi->controlCount>0) {
/* subtract the number of controls until the calculated visual index */
Run *runs=pBiDi->runs;
int32_t i, j, start, limit, length, insertRemove;
int32_t visualStart=0, controlFound=0;
UChar uchar=pBiDi->text[logicalIndex];
/* is the logical index pointing to a control ? */
if(IS_BIDI_CONTROL_CHAR(uchar)) {
return UBIDI_MAP_NOWHERE;
}
/* loop on runs */
for(i=0; ; i++, visualStart+=length) {
length=runs[i].visualLimit-visualStart;
insertRemove=runs[i].insertRemove;
/* calculated visual index is beyond this run? */
if(visualIndex>=runs[i].visualLimit) {
controlFound-=insertRemove;
continue;
}
/* calculated visual index must be within current run */
if(insertRemove==0) {
return visualIndex-controlFound;
}
if(IS_EVEN_RUN(runs[i].logicalStart)) {
/* LTR: check from run start to logical index */
start=runs[i].logicalStart;
limit=logicalIndex;
} else {
/* RTL: check from logical index to run end */
start=logicalIndex+1;
limit=GET_INDEX(runs[i].logicalStart)+length;
}
for(j=start; j<limit; j++) {
uchar=pBiDi->text[j];
if(IS_BIDI_CONTROL_CHAR(uchar)) {
controlFound++;
}
}
return visualIndex-controlFound;
}
}
return visualIndex;
}
U_CAPI int32_t U_EXPORT2
ubidi_getLogicalIndex(UBiDi *pBiDi, int32_t visualIndex, UErrorCode *pErrorCode) {
if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
return 0;
} else if(pBiDi==NULL) {
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
return 0;
} else if(visualIndex<0 || pBiDi->length<=visualIndex) {
*pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
return 0;
} else {
/* we can do the trivial cases without the runs array */
switch(pBiDi->direction) {
case UBIDI_LTR:
return visualIndex;
case UBIDI_RTL:
return pBiDi->length-visualIndex-1;
default:
if(pBiDi->runCount<0 && !ubidi_getRuns(pBiDi)) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return 0;
} else {
Run *runs=pBiDi->runs;
int32_t i, runCount=pBiDi->runCount, start;
if(runCount<=10) {
/* linear search for the run */
for(i=0; visualIndex>=runs[i].visualLimit; ++i) {}
} else {
/* binary search for the run */
int32_t begin=0, limit=runCount;
/* the middle if() will guaranteed find the run, we don't need a loop limit */
for(;;) {
i=(begin+limit)/2;
if(visualIndex>=runs[i].visualLimit) {
begin=i+1;
} else if(i==0 || visualIndex>=runs[i-1].visualLimit) {
break;
} else {
limit=i;
}
}
}
start=runs[i].logicalStart;
if(IS_EVEN_RUN(start)) {
/* LTR */
/* the offset in runs[i] is visualIndex-runs[i-1].visualLimit */
if(i>0) {
visualIndex-=runs[i-1].visualLimit;
}
return GET_INDEX(start)+visualIndex;
} else {
/* RTL */
return GET_INDEX(start)+runs[i].visualLimit-visualIndex-1;
}
}
}
}
}
U_CAPI void U_EXPORT2
ubidi_setPara(UBiDi *pBiDi, const UChar *text, int32_t length,
UBiDiLevel paraLevel, UBiDiLevel *embeddingLevels,
UErrorCode *pErrorCode) {
UBiDiDirection direction;
/* check the argument values */
if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
return;
} else if(pBiDi==NULL || text==NULL ||
((UBIDI_MAX_EXPLICIT_LEVEL<paraLevel) && !IS_DEFAULT_LEVEL(paraLevel)) ||
length<-1
) {
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
return;
}
if(length==-1) {
// length=u_strlen(text);
const UChar *p = text - 1;
while(*++p);
length = p - text;
}
/* initialize the UBiDi structure */
pBiDi->text=text;
pBiDi->length=length;
pBiDi->paraLevel=paraLevel;
pBiDi->direction=UBIDI_LTR;
pBiDi->trailingWSStart=length; /* the levels[] will reflect the WS run */
pBiDi->dirProps=NULL;
pBiDi->levels=NULL;
pBiDi->runs=NULL;
if(length==0) {
/*
* For an empty paragraph, create a UBiDi object with the paraLevel and
* the flags and the direction set but without allocating zero-length arrays.
* There is nothing more to do.
*/
if(IS_DEFAULT_LEVEL(paraLevel)) {
pBiDi->paraLevel&=1;
}
if(paraLevel&1) {
pBiDi->flags=DIRPROP_FLAG(R);
pBiDi->direction=UBIDI_RTL;
} else {
pBiDi->flags=DIRPROP_FLAG(L);
pBiDi->direction=UBIDI_LTR;
}
pBiDi->runCount=0;
return;
}
pBiDi->runCount=-1;
/*
* Get the directional properties,
* the flags bit-set, and
* determine the partagraph level if necessary.
*/
if(getDirPropsMemory(pBiDi, length)) {
pBiDi->dirProps=pBiDi->dirPropsMemory;
getDirProps(pBiDi, text);
} else {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return;
}
if (getLevelsMemory(pBiDi, length)) {
pBiDi->levels=pBiDi->levelsMemory;
/* are explicit levels specified? */
if(embeddingLevels==NULL) {
/* no: determine explicit levels according to the (Xn) rules */
direction=resolveExplicitLevels(pBiDi);
} else {
/* set BN for all explicit codes, check that all levels are paraLevel..UBIDI_MAX_EXPLICIT_LEVEL */
icu_memcpy(pBiDi->levels, embeddingLevels, length);
direction=checkExplicitLevels(pBiDi, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
return;
}
}
} else {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return;
}
/*
* The steps after (X9) in the UBiDi algorithm are performed only if
* the paragraph text has mixed directionality!
*/
pBiDi->direction=direction;
switch(direction) {
case UBIDI_LTR:
/* make sure paraLevel is even */
pBiDi->paraLevel=(UBiDiLevel)((pBiDi->paraLevel+1)&~1);
/* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */
pBiDi->trailingWSStart=0;
break;
case UBIDI_RTL:
/* make sure paraLevel is odd */
pBiDi->paraLevel|=1;
/* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */
pBiDi->trailingWSStart=0;
break;
default:
/*
* If there are no external levels specified and there
* are no significant explicit level codes in the text,
* then we can treat the entire paragraph as one run.
* Otherwise, we need to perform the following rules on runs of
* the text with the same embedding levels. (X10)
* "Significant" explicit level codes are ones that actually
* affect non-BN characters.
* Examples for "insignificant" ones are empty embeddings
* LRE-PDF, LRE-RLE-PDF-PDF, etc.
*/
if(embeddingLevels==NULL && !(pBiDi->flags&DIRPROP_FLAG_MULTI_RUNS)) {
resolveImplicitLevels(pBiDi, 0, length,
GET_LR_FROM_LEVEL(pBiDi->paraLevel),
GET_LR_FROM_LEVEL(pBiDi->paraLevel));
} else {
/* sor, eor: start and end types of same-level-run */
UBiDiLevel *levels=pBiDi->levels;
int32_t start, limit=0;
UBiDiLevel level, nextLevel;
DirProp sor, eor;
/* determine the first sor and set eor to it because of the loop body (sor=eor there) */
level=pBiDi->paraLevel;
nextLevel=levels[0];
if(level<nextLevel) {
eor=GET_LR_FROM_LEVEL(nextLevel);
} else {
eor=GET_LR_FROM_LEVEL(level);
}
do {
/* determine start and limit of the run (end points just behind the run) */
/* the values for this run's start are the same as for the previous run's end */
sor=eor;
start=limit;
level=nextLevel;
/* search for the limit of this run */
while(++limit<length && levels[limit]==level) {}
/* get the correct level of the next run */
if(limit<length) {
nextLevel=levels[limit];
} else {
nextLevel=pBiDi->paraLevel;
}
/* determine eor from max(level, nextLevel); sor is last run's eor */
if((level&~UBIDI_LEVEL_OVERRIDE)<(nextLevel&~UBIDI_LEVEL_OVERRIDE)) {
eor=GET_LR_FROM_LEVEL(nextLevel);
} else {
eor=GET_LR_FROM_LEVEL(level);
}
/* if the run consists of overridden directional types, then there
are no implicit types to be resolved */
if(!(level&UBIDI_LEVEL_OVERRIDE)) {
resolveImplicitLevels(pBiDi, start, limit, sor, eor);
} else {
/* remove the UBIDI_LEVEL_OVERRIDE flags */
do {
levels[start++]&=~UBIDI_LEVEL_OVERRIDE;
} while(start<limit);
}
} while(limit<length);
}
/* reset the embedding levels for some non-graphic characters (L1), (X9) */
adjustWSLevels(pBiDi);
/* for "inverse BiDi", ubidi_getRuns() modifies the levels of numeric runs following RTL runs */
if(pBiDi->isInverse) {
if(!ubidi_getRuns(pBiDi)) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return;
}
}
break;
}
}
