michael@0: /*
michael@0: ******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 1999-2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: ******************************************************************************
michael@0: *   file name:  ubidiln.c
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 1999aug06
michael@0: *   created by: Markus W. Scherer, updated by Matitiahu Allouche
michael@0: */
michael@0: 
michael@0: #include "cmemory.h"
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/ustring.h"
michael@0: #include "unicode/uchar.h"
michael@0: #include "unicode/ubidi.h"
michael@0: #include "ubidiimp.h"
michael@0: #include "uassert.h"
michael@0: 
michael@0: #ifndef U_COMMON_IMPLEMENTATION
michael@0: #error U_COMMON_IMPLEMENTATION not set - must be set for all ICU source files in common/ - see http://userguide.icu-project.org/howtouseicu
michael@0: #endif
michael@0: 
michael@0: /*
michael@0:  * General remarks about the functions in this file:
michael@0:  *
michael@0:  * These functions deal with the aspects of potentially mixed-directional
michael@0:  * text in a single paragraph or in a line of a single paragraph
michael@0:  * which has already been processed according to
michael@0:  * the Unicode 6.3 BiDi algorithm as defined in
michael@0:  * http://www.unicode.org/unicode/reports/tr9/ , version 28,
michael@0:  * also described in The Unicode Standard, Version 6.3.0 .
michael@0:  *
michael@0:  * This means that there is a UBiDi object with a levels
michael@0:  * and a dirProps array.
michael@0:  * paraLevel and direction are also set.
michael@0:  * Only if the length of the text is zero, then levels==dirProps==NULL.
michael@0:  *
michael@0:  * The overall directionality of the paragraph
michael@0:  * or line is used to bypass the reordering steps if possible.
michael@0:  * Even purely RTL text does not need reordering there because
michael@0:  * the ubidi_getLogical/VisualIndex() functions can compute the
michael@0:  * index on the fly in such a case.
michael@0:  *
michael@0:  * The implementation of the access to same-level-runs and of the reordering
michael@0:  * do attempt to provide better performance and less memory usage compared to
michael@0:  * a direct implementation of especially rule (L2) with an array of
michael@0:  * one (32-bit) integer per text character.
michael@0:  *
michael@0:  * Here, the levels array is scanned as soon as necessary, and a vector of
michael@0:  * same-level-runs is created. Reordering then is done on this vector.
michael@0:  * For each run of text positions that were resolved to the same level,
michael@0:  * only 8 bytes are stored: the first text position of the run and the visual
michael@0:  * position behind the run after reordering.
michael@0:  * One sign bit is used to hold the directionality of the run.
michael@0:  * This is inefficient if there are many very short runs. If the average run
michael@0:  * length is <2, then this uses more memory.
michael@0:  *
michael@0:  * In a further attempt to save memory, the levels array is never changed
michael@0:  * after all the resolution rules (Xn, Wn, Nn, In).
michael@0:  * Many functions have to consider the field trailingWSStart:
michael@0:  * if it is less than length, then there is an implicit trailing run
michael@0:  * at the paraLevel,
michael@0:  * which is not reflected in the levels array.
michael@0:  * This allows a line UBiDi object to use the same levels array as
michael@0:  * its paragraph parent object.
michael@0:  *
michael@0:  * When a UBiDi object is created for a line of a paragraph, then the
michael@0:  * paragraph's levels and dirProps arrays are reused by way of setting
michael@0:  * a pointer into them, not by copying. This again saves memory and forbids to
michael@0:  * change the now shared levels for (L1).
michael@0:  */
michael@0: 
michael@0: /* handle trailing WS (L1) -------------------------------------------------- */
michael@0: 
michael@0: /*
michael@0:  * setTrailingWSStart() sets the start index for a trailing
michael@0:  * run of WS in the line. This is necessary because we do not modify
michael@0:  * the paragraph's levels array that we just point into.
michael@0:  * Using trailingWSStart is another form of performing (L1).
michael@0:  *
michael@0:  * To make subsequent operations easier, we also include the run
michael@0:  * before the WS if it is at the paraLevel - we merge the two here.
michael@0:  *
michael@0:  * This function is called only from ubidi_setLine(), so pBiDi->paraLevel is
michael@0:  * set correctly for the line even when contextual multiple paragraphs.
michael@0:  */
michael@0: static void
michael@0: setTrailingWSStart(UBiDi *pBiDi) {
michael@0:     /* pBiDi->direction!=UBIDI_MIXED */
michael@0: 
michael@0:     const DirProp *dirProps=pBiDi->dirProps;
michael@0:     UBiDiLevel *levels=pBiDi->levels;
michael@0:     int32_t start=pBiDi->length;
michael@0:     UBiDiLevel paraLevel=pBiDi->paraLevel;
michael@0: 
michael@0:     /* If the line is terminated by a block separator, all preceding WS etc...
michael@0:        are already set to paragraph level.
michael@0:        Setting trailingWSStart to pBidi->length will avoid changing the
michael@0:        level of B chars from 0 to paraLevel in ubidi_getLevels when
michael@0:        orderParagraphsLTR==TRUE.
michael@0:      */
michael@0:     if(dirProps[start-1]==B) {
michael@0:         pBiDi->trailingWSStart=start;   /* currently == pBiDi->length */
michael@0:         return;
michael@0:     }
michael@0:     /* go backwards across all WS, BN, explicit codes */
michael@0:     while(start>0 && DIRPROP_FLAG(PURE_DIRPROP(dirProps[start-1]))&MASK_WS) {
michael@0:         --start;
michael@0:     }
michael@0: 
michael@0:     /* if the WS run can be merged with the previous run then do so here */
michael@0:     while(start>0 && levels[start-1]==paraLevel) {
michael@0:         --start;
michael@0:     }
michael@0: 
michael@0:     pBiDi->trailingWSStart=start;
michael@0: }
michael@0: 
michael@0: /* ubidi_setLine ------------------------------------------------------------ */
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ubidi_setLine(const UBiDi *pParaBiDi,
michael@0:               int32_t start, int32_t limit,
michael@0:               UBiDi *pLineBiDi,
michael@0:               UErrorCode *pErrorCode) {
michael@0:     int32_t length;
michael@0: 
michael@0:     /* check the argument values */
michael@0:     RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode);
michael@0:     RETURN_VOID_IF_NOT_VALID_PARA(pParaBiDi, *pErrorCode);
michael@0:     RETURN_VOID_IF_BAD_RANGE(start, 0, limit, *pErrorCode);
michael@0:     RETURN_VOID_IF_BAD_RANGE(limit, 0, pParaBiDi->length+1, *pErrorCode);
michael@0:     if(pLineBiDi==NULL) {
michael@0:         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return;
michael@0:     }
michael@0:     if(ubidi_getParagraph(pParaBiDi, start, NULL, NULL, NULL, pErrorCode) !=
michael@0:        ubidi_getParagraph(pParaBiDi, limit-1, NULL, NULL, NULL, pErrorCode)) {
michael@0:         /* the line crosses a paragraph boundary */
michael@0:         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /* set the values in pLineBiDi from its pParaBiDi parent */
michael@0:     pLineBiDi->pParaBiDi=NULL;          /* mark unfinished setLine */
michael@0:     pLineBiDi->text=pParaBiDi->text+start;
michael@0:     length=pLineBiDi->length=limit-start;
michael@0:     pLineBiDi->resultLength=pLineBiDi->originalLength=length;
michael@0:     pLineBiDi->paraLevel=GET_PARALEVEL(pParaBiDi, start);
michael@0:     pLineBiDi->paraCount=pParaBiDi->paraCount;
michael@0:     pLineBiDi->runs=NULL;
michael@0:     pLineBiDi->flags=0;
michael@0:     pLineBiDi->reorderingMode=pParaBiDi->reorderingMode;
michael@0:     pLineBiDi->reorderingOptions=pParaBiDi->reorderingOptions;
michael@0:     pLineBiDi->controlCount=0;
michael@0:     if(pParaBiDi->controlCount>0) {
michael@0:         int32_t j;
michael@0:         for(j=start; j<limit; j++) {
michael@0:             if(IS_BIDI_CONTROL_CHAR(pParaBiDi->text[j])) {
michael@0:                 pLineBiDi->controlCount++;
michael@0:             }
michael@0:         }
michael@0:         pLineBiDi->resultLength-=pLineBiDi->controlCount;
michael@0:     }
michael@0: 
michael@0:     pLineBiDi->dirProps=pParaBiDi->dirProps+start;
michael@0:     pLineBiDi->levels=pParaBiDi->levels+start;
michael@0:     pLineBiDi->runCount=-1;
michael@0: 
michael@0:     if(pParaBiDi->direction!=UBIDI_MIXED) {
michael@0:         /* the parent is already trivial */
michael@0:         pLineBiDi->direction=pParaBiDi->direction;
michael@0: 
michael@0:         /*
michael@0:          * The parent's levels are all either
michael@0:          * implicitly or explicitly ==paraLevel;
michael@0:          * do the same here.
michael@0:          */
michael@0:         if(pParaBiDi->trailingWSStart<=start) {
michael@0:             pLineBiDi->trailingWSStart=0;
michael@0:         } else if(pParaBiDi->trailingWSStart<limit) {
michael@0:             pLineBiDi->trailingWSStart=pParaBiDi->trailingWSStart-start;
michael@0:         } else {
michael@0:             pLineBiDi->trailingWSStart=length;
michael@0:         }
michael@0:     } else {
michael@0:         const UBiDiLevel *levels=pLineBiDi->levels;
michael@0:         int32_t i, trailingWSStart;
michael@0:         UBiDiLevel level;
michael@0: 
michael@0:         setTrailingWSStart(pLineBiDi);
michael@0:         trailingWSStart=pLineBiDi->trailingWSStart;
michael@0: 
michael@0:         /* recalculate pLineBiDi->direction */
michael@0:         if(trailingWSStart==0) {
michael@0:             /* all levels are at paraLevel */
michael@0:             pLineBiDi->direction=(UBiDiDirection)(pLineBiDi->paraLevel&1);
michael@0:         } else {
michael@0:             /* get the level of the first character */
michael@0:             level=(UBiDiLevel)(levels[0]&1);
michael@0: 
michael@0:             /* if there is anything of a different level, then the line is mixed */
michael@0:             if(trailingWSStart<length && (pLineBiDi->paraLevel&1)!=level) {
michael@0:                 /* the trailing WS is at paraLevel, which differs from levels[0] */
michael@0:                 pLineBiDi->direction=UBIDI_MIXED;
michael@0:             } else {
michael@0:                 /* see if levels[1..trailingWSStart-1] have the same direction as levels[0] and paraLevel */
michael@0:                 i=1;
michael@0:                 for(;;) {
michael@0:                     if(i==trailingWSStart) {
michael@0:                         /* the direction values match those in level */
michael@0:                         pLineBiDi->direction=(UBiDiDirection)level;
michael@0:                         break;
michael@0:                     } else if((levels[i]&1)!=level) {
michael@0:                         pLineBiDi->direction=UBIDI_MIXED;
michael@0:                         break;
michael@0:                     }
michael@0:                     ++i;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         switch(pLineBiDi->direction) {
michael@0:         case UBIDI_LTR:
michael@0:             /* make sure paraLevel is even */
michael@0:             pLineBiDi->paraLevel=(UBiDiLevel)((pLineBiDi->paraLevel+1)&~1);
michael@0: 
michael@0:             /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */
michael@0:             pLineBiDi->trailingWSStart=0;
michael@0:             break;
michael@0:         case UBIDI_RTL:
michael@0:             /* make sure paraLevel is odd */
michael@0:             pLineBiDi->paraLevel|=1;
michael@0: 
michael@0:             /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */
michael@0:             pLineBiDi->trailingWSStart=0;
michael@0:             break;
michael@0:         default:
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0:     pLineBiDi->pParaBiDi=pParaBiDi;     /* mark successful setLine */
michael@0:     return;
michael@0: }
michael@0: 
michael@0: U_CAPI UBiDiLevel U_EXPORT2
michael@0: ubidi_getLevelAt(const UBiDi *pBiDi, int32_t charIndex) {
michael@0:     /* return paraLevel if in the trailing WS run, otherwise the real level */
michael@0:     if(!IS_VALID_PARA_OR_LINE(pBiDi) || charIndex<0 || pBiDi->length<=charIndex) {
michael@0:         return 0;
michael@0:     } else if(pBiDi->direction!=UBIDI_MIXED || charIndex>=pBiDi->trailingWSStart) {
michael@0:         return GET_PARALEVEL(pBiDi, charIndex);
michael@0:     } else {
michael@0:         return pBiDi->levels[charIndex];
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI const UBiDiLevel * U_EXPORT2
michael@0: ubidi_getLevels(UBiDi *pBiDi, UErrorCode *pErrorCode) {
michael@0:     int32_t start, length;
michael@0: 
michael@0:     RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, NULL);
michael@0:     RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, NULL);
michael@0:     if((length=pBiDi->length)<=0) {
michael@0:         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return NULL;
michael@0:     }
michael@0:     if((start=pBiDi->trailingWSStart)==length) {
michael@0:         /* the current levels array reflects the WS run */
michael@0:         return pBiDi->levels;
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * After the previous if(), we know that the levels array
michael@0:      * has an implicit trailing WS run and therefore does not fully
michael@0:      * reflect itself all the levels.
michael@0:      * This must be a UBiDi object for a line, and
michael@0:      * we need to create a new levels array.
michael@0:      */
michael@0:     if(getLevelsMemory(pBiDi, length)) {
michael@0:         UBiDiLevel *levels=pBiDi->levelsMemory;
michael@0: 
michael@0:         if(start>0 && levels!=pBiDi->levels) {
michael@0:             uprv_memcpy(levels, pBiDi->levels, start);
michael@0:         }
michael@0:         /* pBiDi->paraLevel is ok even if contextual multiple paragraphs,
michael@0:            since pBidi is a line object                                     */
michael@0:         uprv_memset(levels+start, pBiDi->paraLevel, length-start);
michael@0: 
michael@0:         /* this new levels array is set for the line and reflects the WS run */
michael@0:         pBiDi->trailingWSStart=length;
michael@0:         return pBiDi->levels=levels;
michael@0:     } else {
michael@0:         /* out of memory */
michael@0:         *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ubidi_getLogicalRun(const UBiDi *pBiDi, int32_t logicalPosition,
michael@0:                     int32_t *pLogicalLimit, UBiDiLevel *pLevel) {
michael@0:     UErrorCode errorCode;
michael@0:     int32_t runCount, visualStart, logicalLimit, logicalFirst, i;
michael@0:     Run iRun;
michael@0: 
michael@0:     errorCode=U_ZERO_ERROR;
michael@0:     RETURN_VOID_IF_BAD_RANGE(logicalPosition, 0, pBiDi->length, errorCode);
michael@0:     /* ubidi_countRuns will check VALID_PARA_OR_LINE */
michael@0:     runCount=ubidi_countRuns((UBiDi *)pBiDi, &errorCode);
michael@0:     if(U_FAILURE(errorCode)) {
michael@0:         return;
michael@0:     }
michael@0:     /* this is done based on runs rather than on levels since levels have
michael@0:        a special interpretation when UBIDI_REORDER_RUNS_ONLY
michael@0:      */
michael@0:     visualStart=logicalLimit=0;
michael@0:     iRun=pBiDi->runs[0];
michael@0: 
michael@0:     for(i=0; i<runCount; i++) {
michael@0:         iRun = pBiDi->runs[i];
michael@0:         logicalFirst=GET_INDEX(iRun.logicalStart);
michael@0:         logicalLimit=logicalFirst+iRun.visualLimit-visualStart;
michael@0:         if((logicalPosition>=logicalFirst) &&
michael@0:            (logicalPosition<logicalLimit)) {
michael@0:             break;
michael@0:         }
michael@0:         visualStart = iRun.visualLimit;
michael@0:     }
michael@0:     if(pLogicalLimit) {
michael@0:         *pLogicalLimit=logicalLimit;
michael@0:     }
michael@0:     if(pLevel) {
michael@0:         if(pBiDi->reorderingMode==UBIDI_REORDER_RUNS_ONLY) {
michael@0:             *pLevel=(UBiDiLevel)GET_ODD_BIT(iRun.logicalStart);
michael@0:         }
michael@0:         else if(pBiDi->direction!=UBIDI_MIXED || logicalPosition>=pBiDi->trailingWSStart) {
michael@0:             *pLevel=GET_PARALEVEL(pBiDi, logicalPosition);
michael@0:         } else {
michael@0:         *pLevel=pBiDi->levels[logicalPosition];
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* runs API functions ------------------------------------------------------- */
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ubidi_countRuns(UBiDi *pBiDi, UErrorCode *pErrorCode) {
michael@0:     RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1);
michael@0:     RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1);
michael@0:     ubidi_getRuns(pBiDi, pErrorCode);
michael@0:     if(U_FAILURE(*pErrorCode)) {
michael@0:         return -1;
michael@0:     }
michael@0:     return pBiDi->runCount;
michael@0: }
michael@0: 
michael@0: U_CAPI UBiDiDirection U_EXPORT2
michael@0: ubidi_getVisualRun(UBiDi *pBiDi, int32_t runIndex,
michael@0:                    int32_t *pLogicalStart, int32_t *pLength)
michael@0: {
michael@0:     int32_t start;
michael@0:     UErrorCode errorCode = U_ZERO_ERROR;
michael@0:     RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, errorCode, UBIDI_LTR);
michael@0:     ubidi_getRuns(pBiDi, &errorCode);
michael@0:     if(U_FAILURE(errorCode)) {
michael@0:         return UBIDI_LTR;
michael@0:     }
michael@0:     RETURN_IF_BAD_RANGE(runIndex, 0, pBiDi->runCount, errorCode, UBIDI_LTR);
michael@0: 
michael@0:     start=pBiDi->runs[runIndex].logicalStart;
michael@0:     if(pLogicalStart!=NULL) {
michael@0:         *pLogicalStart=GET_INDEX(start);
michael@0:     }
michael@0:     if(pLength!=NULL) {
michael@0:         if(runIndex>0) {
michael@0:             *pLength=pBiDi->runs[runIndex].visualLimit-
michael@0:                      pBiDi->runs[runIndex-1].visualLimit;
michael@0:         } else {
michael@0:             *pLength=pBiDi->runs[0].visualLimit;
michael@0:         }
michael@0:     }
michael@0:     return (UBiDiDirection)GET_ODD_BIT(start);
michael@0: }
michael@0: 
michael@0: /* in trivial cases there is only one trivial run; called by ubidi_getRuns() */
michael@0: static void
michael@0: getSingleRun(UBiDi *pBiDi, UBiDiLevel level) {
michael@0:     /* simple, single-run case */
michael@0:     pBiDi->runs=pBiDi->simpleRuns;
michael@0:     pBiDi->runCount=1;
michael@0: 
michael@0:     /* fill and reorder the single run */
michael@0:     pBiDi->runs[0].logicalStart=MAKE_INDEX_ODD_PAIR(0, level);
michael@0:     pBiDi->runs[0].visualLimit=pBiDi->length;
michael@0:     pBiDi->runs[0].insertRemove=0;
michael@0: }
michael@0: 
michael@0: /* reorder the runs array (L2) ---------------------------------------------- */
michael@0: 
michael@0: /*
michael@0:  * Reorder the same-level runs in the runs array.
michael@0:  * Here, runCount>1 and maxLevel>=minLevel>=paraLevel.
michael@0:  * All the visualStart fields=logical start before reordering.
michael@0:  * The "odd" bits are not set yet.
michael@0:  *
michael@0:  * Reordering with this data structure lends itself to some handy shortcuts:
michael@0:  *
michael@0:  * Since each run is moved but not modified, and since at the initial maxLevel
michael@0:  * each sequence of same-level runs consists of only one run each, we
michael@0:  * don't need to do anything there and can predecrement maxLevel.
michael@0:  * In many simple cases, the reordering is thus done entirely in the
michael@0:  * index mapping.
michael@0:  * Also, reordering occurs only down to the lowest odd level that occurs,
michael@0:  * which is minLevel|1. However, if the lowest level itself is odd, then
michael@0:  * in the last reordering the sequence of the runs at this level or higher
michael@0:  * will be all runs, and we don't need the elaborate loop to search for them.
michael@0:  * This is covered by ++minLevel instead of minLevel|=1 followed
michael@0:  * by an extra reorder-all after the reorder-some loop.
michael@0:  * About a trailing WS run:
michael@0:  * Such a run would need special treatment because its level is not
michael@0:  * reflected in levels[] if this is not a paragraph object.
michael@0:  * Instead, all characters from trailingWSStart on are implicitly at
michael@0:  * paraLevel.
michael@0:  * However, for all maxLevel>paraLevel, this run will never be reordered
michael@0:  * and does not need to be taken into account. maxLevel==paraLevel is only reordered
michael@0:  * if minLevel==paraLevel is odd, which is done in the extra segment.
michael@0:  * This means that for the main reordering loop we don't need to consider
michael@0:  * this run and can --runCount. If it is later part of the all-runs
michael@0:  * reordering, then runCount is adjusted accordingly.
michael@0:  */
michael@0: static void
michael@0: reorderLine(UBiDi *pBiDi, UBiDiLevel minLevel, UBiDiLevel maxLevel) {
michael@0:     Run *runs, tempRun;
michael@0:     UBiDiLevel *levels;
michael@0:     int32_t firstRun, endRun, limitRun, runCount;
michael@0: 
michael@0:     /* nothing to do? */
michael@0:     if(maxLevel<=(minLevel|1)) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * Reorder only down to the lowest odd level
michael@0:      * and reorder at an odd minLevel in a separate, simpler loop.
michael@0:      * See comments above for why minLevel is always incremented.
michael@0:      */
michael@0:     ++minLevel;
michael@0: 
michael@0:     runs=pBiDi->runs;
michael@0:     levels=pBiDi->levels;
michael@0:     runCount=pBiDi->runCount;
michael@0: 
michael@0:     /* do not include the WS run at paraLevel<=old minLevel except in the simple loop */
michael@0:     if(pBiDi->trailingWSStart<pBiDi->length) {
michael@0:         --runCount;
michael@0:     }
michael@0: 
michael@0:     while(--maxLevel>=minLevel) {
michael@0:         firstRun=0;
michael@0: 
michael@0:         /* loop for all sequences of runs */
michael@0:         for(;;) {
michael@0:             /* look for a sequence of runs that are all at >=maxLevel */
michael@0:             /* look for the first run of such a sequence */
michael@0:             while(firstRun<runCount && levels[runs[firstRun].logicalStart]<maxLevel) {
michael@0:                 ++firstRun;
michael@0:             }
michael@0:             if(firstRun>=runCount) {
michael@0:                 break;  /* no more such runs */
michael@0:             }
michael@0: 
michael@0:             /* look for the limit run of such a sequence (the run behind it) */
michael@0:             for(limitRun=firstRun; ++limitRun<runCount && levels[runs[limitRun].logicalStart]>=maxLevel;) {}
michael@0: 
michael@0:             /* Swap the entire sequence of runs from firstRun to limitRun-1. */
michael@0:             endRun=limitRun-1;
michael@0:             while(firstRun<endRun) {
michael@0:                 tempRun = runs[firstRun];
michael@0:                 runs[firstRun]=runs[endRun];
michael@0:                 runs[endRun]=tempRun;
michael@0:                 ++firstRun;
michael@0:                 --endRun;
michael@0:             }
michael@0: 
michael@0:             if(limitRun==runCount) {
michael@0:                 break;  /* no more such runs */
michael@0:             } else {
michael@0:                 firstRun=limitRun+1;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* now do maxLevel==old minLevel (==odd!), see above */
michael@0:     if(!(minLevel&1)) {
michael@0:         firstRun=0;
michael@0: 
michael@0:         /* include the trailing WS run in this complete reordering */
michael@0:         if(pBiDi->trailingWSStart==pBiDi->length) {
michael@0:             --runCount;
michael@0:         }
michael@0: 
michael@0:         /* Swap the entire sequence of all runs. (endRun==runCount) */
michael@0:         while(firstRun<runCount) {
michael@0:             tempRun=runs[firstRun];
michael@0:             runs[firstRun]=runs[runCount];
michael@0:             runs[runCount]=tempRun;
michael@0:             ++firstRun;
michael@0:             --runCount;
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* compute the runs array --------------------------------------------------- */
michael@0: 
michael@0: static int32_t getRunFromLogicalIndex(UBiDi *pBiDi, int32_t logicalIndex, UErrorCode *pErrorCode) {
michael@0:     Run *runs=pBiDi->runs;
michael@0:     int32_t runCount=pBiDi->runCount, visualStart=0, i, length, logicalStart;
michael@0: 
michael@0:     for(i=0; i<runCount; i++) {
michael@0:         length=runs[i].visualLimit-visualStart;
michael@0:         logicalStart=GET_INDEX(runs[i].logicalStart);
michael@0:         if((logicalIndex>=logicalStart) && (logicalIndex<(logicalStart+length))) {
michael@0:             return i;
michael@0:         }
michael@0:         visualStart+=length;
michael@0:     }
michael@0:     /* we should never get here */
michael@0:     U_ASSERT(FALSE);
michael@0:     *pErrorCode = U_INVALID_STATE_ERROR;
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: /*
michael@0:  * Compute the runs array from the levels array.
michael@0:  * After ubidi_getRuns() returns TRUE, runCount is guaranteed to be >0
michael@0:  * and the runs are reordered.
michael@0:  * Odd-level runs have visualStart on their visual right edge and
michael@0:  * they progress visually to the left.
michael@0:  * If option UBIDI_OPTION_INSERT_MARKS is set, insertRemove will contain the
michael@0:  * sum of appropriate LRM/RLM_BEFORE/AFTER flags.
michael@0:  * If option UBIDI_OPTION_REMOVE_CONTROLS is set, insertRemove will contain the
michael@0:  * negative number of BiDi control characters within this run.
michael@0:  */
michael@0: U_CFUNC UBool
michael@0: ubidi_getRuns(UBiDi *pBiDi, UErrorCode *pErrorCode) {
michael@0:     /*
michael@0:      * This method returns immediately if the runs are already set. This
michael@0:      * includes the case of length==0 (handled in setPara)..
michael@0:      */
michael@0:     if (pBiDi->runCount>=0) {
michael@0:         return TRUE;
michael@0:     }
michael@0: 
michael@0:     if(pBiDi->direction!=UBIDI_MIXED) {
michael@0:         /* simple, single-run case - this covers length==0 */
michael@0:         /* pBiDi->paraLevel is ok even for contextual multiple paragraphs */
michael@0:         getSingleRun(pBiDi, pBiDi->paraLevel);
michael@0:     } else /* UBIDI_MIXED, length>0 */ {
michael@0:         /* mixed directionality */
michael@0:         int32_t length=pBiDi->length, limit;
michael@0:         UBiDiLevel *levels=pBiDi->levels;
michael@0:         int32_t i, runCount;
michael@0:         UBiDiLevel level=UBIDI_DEFAULT_LTR;   /* initialize with no valid level */
michael@0:         /*
michael@0:          * If there are WS characters at the end of the line
michael@0:          * and the run preceding them has a level different from
michael@0:          * paraLevel, then they will form their own run at paraLevel (L1).
michael@0:          * Count them separately.
michael@0:          * We need some special treatment for this in order to not
michael@0:          * modify the levels array which a line UBiDi object shares
michael@0:          * with its paragraph parent and its other line siblings.
michael@0:          * In other words, for the trailing WS, it may be
michael@0:          * levels[]!=paraLevel but we have to treat it like it were so.
michael@0:          */
michael@0:         limit=pBiDi->trailingWSStart;
michael@0:         /* count the runs, there is at least one non-WS run, and limit>0 */
michael@0:         runCount=0;
michael@0:         for(i=0; i<limit; ++i) {
michael@0:             /* increment runCount at the start of each run */
michael@0:             if(levels[i]!=level) {
michael@0:                 ++runCount;
michael@0:                 level=levels[i];
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         /*
michael@0:          * We don't need to see if the last run can be merged with a trailing
michael@0:          * WS run because setTrailingWSStart() would have done that.
michael@0:          */
michael@0:         if(runCount==1 && limit==length) {
michael@0:             /* There is only one non-WS run and no trailing WS-run. */
michael@0:             getSingleRun(pBiDi, levels[0]);
michael@0:         } else /* runCount>1 || limit<length */ {
michael@0:             /* allocate and set the runs */
michael@0:             Run *runs;
michael@0:             int32_t runIndex, start;
michael@0:             UBiDiLevel minLevel=UBIDI_MAX_EXPLICIT_LEVEL+1, maxLevel=0;
michael@0: 
michael@0:             /* now, count a (non-mergeable) WS run */
michael@0:             if(limit<length) {
michael@0:                 ++runCount;
michael@0:             }
michael@0: 
michael@0:             /* runCount>1 */
michael@0:             if(getRunsMemory(pBiDi, runCount)) {
michael@0:                 runs=pBiDi->runsMemory;
michael@0:             } else {
michael@0:                 return FALSE;
michael@0:             }
michael@0: 
michael@0:             /* set the runs */
michael@0:             /* FOOD FOR THOUGHT: this could be optimized, e.g.:
michael@0:              * 464->444, 484->444, 575->555, 595->555
michael@0:              * However, that would take longer. Check also how it would
michael@0:              * interact with BiDi control removal and inserting Marks.
michael@0:              */
michael@0:             runIndex=0;
michael@0: 
michael@0:             /* search for the run limits and initialize visualLimit values with the run lengths */
michael@0:             i=0;
michael@0:             do {
michael@0:                 /* prepare this run */
michael@0:                 start=i;
michael@0:                 level=levels[i];
michael@0:                 if(level<minLevel) {
michael@0:                     minLevel=level;
michael@0:                 }
michael@0:                 if(level>maxLevel) {
michael@0:                     maxLevel=level;
michael@0:                 }
michael@0: 
michael@0:                 /* look for the run limit */
michael@0:                 while(++i<limit && levels[i]==level) {}
michael@0: 
michael@0:                 /* i is another run limit */
michael@0:                 runs[runIndex].logicalStart=start;
michael@0:                 runs[runIndex].visualLimit=i-start;
michael@0:                 runs[runIndex].insertRemove=0;
michael@0:                 ++runIndex;
michael@0:             } while(i<limit);
michael@0: 
michael@0:             if(limit<length) {
michael@0:                 /* there is a separate WS run */
michael@0:                 runs[runIndex].logicalStart=limit;
michael@0:                 runs[runIndex].visualLimit=length-limit;
michael@0:                 /* For the trailing WS run, pBiDi->paraLevel is ok even
michael@0:                    if contextual multiple paragraphs.                   */
michael@0:                 if(pBiDi->paraLevel<minLevel) {
michael@0:                     minLevel=pBiDi->paraLevel;
michael@0:                 }
michael@0:             }
michael@0: 
michael@0:             /* set the object fields */
michael@0:             pBiDi->runs=runs;
michael@0:             pBiDi->runCount=runCount;
michael@0: 
michael@0:             reorderLine(pBiDi, minLevel, maxLevel);
michael@0: 
michael@0:             /* now add the direction flags and adjust the visualLimit's to be just that */
michael@0:             /* this loop will also handle the trailing WS run */
michael@0:             limit=0;
michael@0:             for(i=0; i<runCount; ++i) {
michael@0:                 ADD_ODD_BIT_FROM_LEVEL(runs[i].logicalStart, levels[runs[i].logicalStart]);
michael@0:                 limit+=runs[i].visualLimit;
michael@0:                 runs[i].visualLimit=limit;
michael@0:             }
michael@0: 
michael@0:             /* Set the "odd" bit for the trailing WS run. */
michael@0:             /* For a RTL paragraph, it will be the *first* run in visual order. */
michael@0:             /* For the trailing WS run, pBiDi->paraLevel is ok even if
michael@0:                contextual multiple paragraphs.                          */
michael@0:             if(runIndex<runCount) {
michael@0:                 int32_t trailingRun = ((pBiDi->paraLevel & 1) != 0)? 0 : runIndex;
michael@0: 
michael@0:                 ADD_ODD_BIT_FROM_LEVEL(runs[trailingRun].logicalStart, pBiDi->paraLevel);
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* handle insert LRM/RLM BEFORE/AFTER run */
michael@0:     if(pBiDi->insertPoints.size>0) {
michael@0:         Point *point, *start=pBiDi->insertPoints.points,
michael@0:                       *limit=start+pBiDi->insertPoints.size;
michael@0:         int32_t runIndex;
michael@0:         for(point=start; point<limit; point++) {
michael@0:             runIndex=getRunFromLogicalIndex(pBiDi, point->pos, pErrorCode);
michael@0:             pBiDi->runs[runIndex].insertRemove|=point->flag;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* handle remove BiDi control characters */
michael@0:     if(pBiDi->controlCount>0) {
michael@0:         int32_t runIndex;
michael@0:         const UChar *start=pBiDi->text, *limit=start+pBiDi->length, *pu;
michael@0:         for(pu=start; pu<limit; pu++) {
michael@0:             if(IS_BIDI_CONTROL_CHAR(*pu)) {
michael@0:                 runIndex=getRunFromLogicalIndex(pBiDi, (int32_t)(pu-start), pErrorCode);
michael@0:                 pBiDi->runs[runIndex].insertRemove--;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: static UBool
michael@0: prepareReorder(const UBiDiLevel *levels, int32_t length,
michael@0:                int32_t *indexMap,
michael@0:                UBiDiLevel *pMinLevel, UBiDiLevel *pMaxLevel) {
michael@0:     int32_t start;
michael@0:     UBiDiLevel level, minLevel, maxLevel;
michael@0: 
michael@0:     if(levels==NULL || length<=0) {
michael@0:         return FALSE;
michael@0:     }
michael@0: 
michael@0:     /* determine minLevel and maxLevel */
michael@0:     minLevel=UBIDI_MAX_EXPLICIT_LEVEL+1;
michael@0:     maxLevel=0;
michael@0:     for(start=length; start>0;) {
michael@0:         level=levels[--start];
michael@0:         if(level>UBIDI_MAX_EXPLICIT_LEVEL+1) {
michael@0:             return FALSE;
michael@0:         }
michael@0:         if(level<minLevel) {
michael@0:             minLevel=level;
michael@0:         }
michael@0:         if(level>maxLevel) {
michael@0:             maxLevel=level;
michael@0:         }
michael@0:     }
michael@0:     *pMinLevel=minLevel;
michael@0:     *pMaxLevel=maxLevel;
michael@0: 
michael@0:     /* initialize the index map */
michael@0:     for(start=length; start>0;) {
michael@0:         --start;
michael@0:         indexMap[start]=start;
michael@0:     }
michael@0: 
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: /* reorder a line based on a levels array (L2) ------------------------------ */
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ubidi_reorderLogical(const UBiDiLevel *levels, int32_t length, int32_t *indexMap) {
michael@0:     int32_t start, limit, sumOfSosEos;
michael@0:     UBiDiLevel minLevel = 0, maxLevel = 0;
michael@0: 
michael@0:     if(indexMap==NULL || !prepareReorder(levels, length, indexMap, &minLevel, &maxLevel)) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /* nothing to do? */
michael@0:     if(minLevel==maxLevel && (minLevel&1)==0) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /* reorder only down to the lowest odd level */
michael@0:     minLevel|=1;
michael@0: 
michael@0:     /* loop maxLevel..minLevel */
michael@0:     do {
michael@0:         start=0;
michael@0: 
michael@0:         /* loop for all sequences of levels to reorder at the current maxLevel */
michael@0:         for(;;) {
michael@0:             /* look for a sequence of levels that are all at >=maxLevel */
michael@0:             /* look for the first index of such a sequence */
michael@0:             while(start<length && levels[start]<maxLevel) {
michael@0:                 ++start;
michael@0:             }
michael@0:             if(start>=length) {
michael@0:                 break;  /* no more such sequences */
michael@0:             }
michael@0: 
michael@0:             /* look for the limit of such a sequence (the index behind it) */
michael@0:             for(limit=start; ++limit<length && levels[limit]>=maxLevel;) {}
michael@0: 
michael@0:             /*
michael@0:              * sos=start of sequence, eos=end of sequence
michael@0:              *
michael@0:              * The closed (inclusive) interval from sos to eos includes all the logical
michael@0:              * and visual indexes within this sequence. They are logically and
michael@0:              * visually contiguous and in the same range.
michael@0:              *
michael@0:              * For each run, the new visual index=sos+eos-old visual index;
michael@0:              * we pre-add sos+eos into sumOfSosEos ->
michael@0:              * new visual index=sumOfSosEos-old visual index;
michael@0:              */
michael@0:             sumOfSosEos=start+limit-1;
michael@0: 
michael@0:             /* reorder each index in the sequence */
michael@0:             do {
michael@0:                 indexMap[start]=sumOfSosEos-indexMap[start];
michael@0:             } while(++start<limit);
michael@0: 
michael@0:             /* start==limit */
michael@0:             if(limit==length) {
michael@0:                 break;  /* no more such sequences */
michael@0:             } else {
michael@0:                 start=limit+1;
michael@0:             }
michael@0:         }
michael@0:     } while(--maxLevel>=minLevel);
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ubidi_reorderVisual(const UBiDiLevel *levels, int32_t length, int32_t *indexMap) {
michael@0:     int32_t start, end, limit, temp;
michael@0:     UBiDiLevel minLevel = 0, maxLevel = 0;
michael@0: 
michael@0:     if(indexMap==NULL || !prepareReorder(levels, length, indexMap, &minLevel, &maxLevel)) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /* nothing to do? */
michael@0:     if(minLevel==maxLevel && (minLevel&1)==0) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /* reorder only down to the lowest odd level */
michael@0:     minLevel|=1;
michael@0: 
michael@0:     /* loop maxLevel..minLevel */
michael@0:     do {
michael@0:         start=0;
michael@0: 
michael@0:         /* loop for all sequences of levels to reorder at the current maxLevel */
michael@0:         for(;;) {
michael@0:             /* look for a sequence of levels that are all at >=maxLevel */
michael@0:             /* look for the first index of such a sequence */
michael@0:             while(start<length && levels[start]<maxLevel) {
michael@0:                 ++start;
michael@0:             }
michael@0:             if(start>=length) {
michael@0:                 break;  /* no more such runs */
michael@0:             }
michael@0: 
michael@0:             /* look for the limit of such a sequence (the index behind it) */
michael@0:             for(limit=start; ++limit<length && levels[limit]>=maxLevel;) {}
michael@0: 
michael@0:             /*
michael@0:              * Swap the entire interval of indexes from start to limit-1.
michael@0:              * We don't need to swap the levels for the purpose of this
michael@0:              * algorithm: the sequence of levels that we look at does not
michael@0:              * move anyway.
michael@0:              */
michael@0:             end=limit-1;
michael@0:             while(start<end) {
michael@0:                 temp=indexMap[start];
michael@0:                 indexMap[start]=indexMap[end];
michael@0:                 indexMap[end]=temp;
michael@0: 
michael@0:                 ++start;
michael@0:                 --end;
michael@0:             }
michael@0: 
michael@0:             if(limit==length) {
michael@0:                 break;  /* no more such sequences */
michael@0:             } else {
michael@0:                 start=limit+1;
michael@0:             }
michael@0:         }
michael@0:     } while(--maxLevel>=minLevel);
michael@0: }
michael@0: 
michael@0: /* API functions for logical<->visual mapping ------------------------------- */
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ubidi_getVisualIndex(UBiDi *pBiDi, int32_t logicalIndex, UErrorCode *pErrorCode) {
michael@0:     int32_t visualIndex=UBIDI_MAP_NOWHERE;
michael@0:     RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1);
michael@0:     RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1);
michael@0:     RETURN_IF_BAD_RANGE(logicalIndex, 0, pBiDi->length, *pErrorCode, -1);
michael@0: 
michael@0:     /* we can do the trivial cases without the runs array */
michael@0:     switch(pBiDi->direction) {
michael@0:     case UBIDI_LTR:
michael@0:         visualIndex=logicalIndex;
michael@0:         break;
michael@0:     case UBIDI_RTL:
michael@0:         visualIndex=pBiDi->length-logicalIndex-1;
michael@0:         break;
michael@0:     default:
michael@0:         if(!ubidi_getRuns(pBiDi, pErrorCode)) {
michael@0:             *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
michael@0:             return -1;
michael@0:         } else {
michael@0:             Run *runs=pBiDi->runs;
michael@0:             int32_t i, visualStart=0, offset, length;
michael@0: 
michael@0:             /* linear search for the run, search on the visual runs */
michael@0:             for(i=0; i<pBiDi->runCount; ++i) {
michael@0:                 length=runs[i].visualLimit-visualStart;
michael@0:                 offset=logicalIndex-GET_INDEX(runs[i].logicalStart);
michael@0:                 if(offset>=0 && offset<length) {
michael@0:                     if(IS_EVEN_RUN(runs[i].logicalStart)) {
michael@0:                         /* LTR */
michael@0:                         visualIndex=visualStart+offset;
michael@0:                     } else {
michael@0:                         /* RTL */
michael@0:                         visualIndex=visualStart+length-offset-1;
michael@0:                     }
michael@0:                     break;          /* exit for loop */
michael@0:                 }
michael@0:                 visualStart+=length;
michael@0:             }
michael@0:             if(i>=pBiDi->runCount) {
michael@0:                 return UBIDI_MAP_NOWHERE;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     if(pBiDi->insertPoints.size>0) {
michael@0:         /* add the number of added marks until the calculated visual index */
michael@0:         Run *runs=pBiDi->runs;
michael@0:         int32_t i, length, insertRemove;
michael@0:         int32_t visualStart=0, markFound=0;
michael@0:         for(i=0; ; i++, visualStart+=length) {
michael@0:             length=runs[i].visualLimit-visualStart;
michael@0:             insertRemove=runs[i].insertRemove;
michael@0:             if(insertRemove & (LRM_BEFORE|RLM_BEFORE)) {
michael@0:                 markFound++;
michael@0:             }
michael@0:             /* is it the run containing the visual index? */
michael@0:             if(visualIndex<runs[i].visualLimit) {
michael@0:                 return visualIndex+markFound;
michael@0:             }
michael@0:             if(insertRemove & (LRM_AFTER|RLM_AFTER)) {
michael@0:                 markFound++;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     else if(pBiDi->controlCount>0) {
michael@0:         /* subtract the number of controls until the calculated visual index */
michael@0:         Run *runs=pBiDi->runs;
michael@0:         int32_t i, j, start, limit, length, insertRemove;
michael@0:         int32_t visualStart=0, controlFound=0;
michael@0:         UChar uchar=pBiDi->text[logicalIndex];
michael@0:         /* is the logical index pointing to a control ? */
michael@0:         if(IS_BIDI_CONTROL_CHAR(uchar)) {
michael@0:             return UBIDI_MAP_NOWHERE;
michael@0:         }
michael@0:         /* loop on runs */
michael@0:         for(i=0; ; i++, visualStart+=length) {
michael@0:             length=runs[i].visualLimit-visualStart;
michael@0:             insertRemove=runs[i].insertRemove;
michael@0:             /* calculated visual index is beyond this run? */
michael@0:             if(visualIndex>=runs[i].visualLimit) {
michael@0:                 controlFound-=insertRemove;
michael@0:                 continue;
michael@0:             }
michael@0:             /* calculated visual index must be within current run */
michael@0:             if(insertRemove==0) {
michael@0:                 return visualIndex-controlFound;
michael@0:             }
michael@0:             if(IS_EVEN_RUN(runs[i].logicalStart)) {
michael@0:                 /* LTR: check from run start to logical index */
michael@0:                 start=runs[i].logicalStart;
michael@0:                 limit=logicalIndex;
michael@0:             } else {
michael@0:                 /* RTL: check from logical index to run end */
michael@0:                 start=logicalIndex+1;
michael@0:                 limit=GET_INDEX(runs[i].logicalStart)+length;
michael@0:             }
michael@0:             for(j=start; j<limit; j++) {
michael@0:                 uchar=pBiDi->text[j];
michael@0:                 if(IS_BIDI_CONTROL_CHAR(uchar)) {
michael@0:                     controlFound++;
michael@0:                 }
michael@0:             }
michael@0:             return visualIndex-controlFound;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     return visualIndex;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ubidi_getLogicalIndex(UBiDi *pBiDi, int32_t visualIndex, UErrorCode *pErrorCode) {
michael@0:     Run *runs;
michael@0:     int32_t i, runCount, start;
michael@0:     RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1);
michael@0:     RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1);
michael@0:     RETURN_IF_BAD_RANGE(visualIndex, 0, pBiDi->resultLength, *pErrorCode, -1);
michael@0:     /* we can do the trivial cases without the runs array */
michael@0:     if(pBiDi->insertPoints.size==0 && pBiDi->controlCount==0) {
michael@0:         if(pBiDi->direction==UBIDI_LTR) {
michael@0:             return visualIndex;
michael@0:         }
michael@0:         else if(pBiDi->direction==UBIDI_RTL) {
michael@0:             return pBiDi->length-visualIndex-1;
michael@0:         }
michael@0:     }
michael@0:     if(!ubidi_getRuns(pBiDi, pErrorCode)) {
michael@0:         *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
michael@0:         return -1;
michael@0:     }
michael@0: 
michael@0:     runs=pBiDi->runs;
michael@0:     runCount=pBiDi->runCount;
michael@0:     if(pBiDi->insertPoints.size>0) {
michael@0:         /* handle inserted LRM/RLM */
michael@0:         int32_t markFound=0, insertRemove;
michael@0:         int32_t visualStart=0, length;
michael@0:         runs=pBiDi->runs;
michael@0:         /* subtract number of marks until visual index */
michael@0:         for(i=0; ; i++, visualStart+=length) {
michael@0:             length=runs[i].visualLimit-visualStart;
michael@0:             insertRemove=runs[i].insertRemove;
michael@0:             if(insertRemove&(LRM_BEFORE|RLM_BEFORE)) {
michael@0:                 if(visualIndex<=(visualStart+markFound)) {
michael@0:                     return UBIDI_MAP_NOWHERE;
michael@0:                 }
michael@0:                 markFound++;
michael@0:             }
michael@0:             /* is adjusted visual index within this run? */
michael@0:             if(visualIndex<(runs[i].visualLimit+markFound)) {
michael@0:                 visualIndex-=markFound;
michael@0:                 break;
michael@0:             }
michael@0:             if(insertRemove&(LRM_AFTER|RLM_AFTER)) {
michael@0:                 if(visualIndex==(visualStart+length+markFound)) {
michael@0:                     return UBIDI_MAP_NOWHERE;
michael@0:                 }
michael@0:                 markFound++;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     else if(pBiDi->controlCount>0) {
michael@0:         /* handle removed BiDi control characters */
michael@0:         int32_t controlFound=0, insertRemove, length;
michael@0:         int32_t logicalStart, logicalEnd, visualStart=0, j, k;
michael@0:         UChar uchar;
michael@0:         UBool evenRun;
michael@0:         /* add number of controls until visual index */
michael@0:         for(i=0; ; i++, visualStart+=length) {
michael@0:             length=runs[i].visualLimit-visualStart;
michael@0:             insertRemove=runs[i].insertRemove;
michael@0:             /* is adjusted visual index beyond current run? */
michael@0:             if(visualIndex>=(runs[i].visualLimit-controlFound+insertRemove)) {
michael@0:                 controlFound-=insertRemove;
michael@0:                 continue;
michael@0:             }
michael@0:             /* adjusted visual index is within current run */
michael@0:             if(insertRemove==0) {
michael@0:                 visualIndex+=controlFound;
michael@0:                 break;
michael@0:             }
michael@0:             /* count non-control chars until visualIndex */
michael@0:             logicalStart=runs[i].logicalStart;
michael@0:             evenRun=IS_EVEN_RUN(logicalStart);
michael@0:             REMOVE_ODD_BIT(logicalStart);
michael@0:             logicalEnd=logicalStart+length-1;
michael@0:             for(j=0; j<length; j++) {
michael@0:                 k= evenRun ? logicalStart+j : logicalEnd-j;
michael@0:                 uchar=pBiDi->text[k];
michael@0:                 if(IS_BIDI_CONTROL_CHAR(uchar)) {
michael@0:                     controlFound++;
michael@0:                 }
michael@0:                 if((visualIndex+controlFound)==(visualStart+j)) {
michael@0:                     break;
michael@0:                 }
michael@0:             }
michael@0:             visualIndex+=controlFound;
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0:     /* handle all cases */
michael@0:     if(runCount<=10) {
michael@0:         /* linear search for the run */
michael@0:         for(i=0; visualIndex>=runs[i].visualLimit; ++i) {}
michael@0:     } else {
michael@0:         /* binary search for the run */
michael@0:         int32_t begin=0, limit=runCount;
michael@0: 
michael@0:         /* the middle if() is guaranteed to find the run, we don't need a loop limit */
michael@0:         for(;;) {
michael@0:             i=(begin+limit)/2;
michael@0:             if(visualIndex>=runs[i].visualLimit) {
michael@0:                 begin=i+1;
michael@0:             } else if(i==0 || visualIndex>=runs[i-1].visualLimit) {
michael@0:                 break;
michael@0:             } else {
michael@0:                 limit=i;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     start=runs[i].logicalStart;
michael@0:     if(IS_EVEN_RUN(start)) {
michael@0:         /* LTR */
michael@0:         /* the offset in runs[i] is visualIndex-runs[i-1].visualLimit */
michael@0:         if(i>0) {
michael@0:             visualIndex-=runs[i-1].visualLimit;
michael@0:         }
michael@0:         return start+visualIndex;
michael@0:     } else {
michael@0:         /* RTL */
michael@0:         return GET_INDEX(start)+runs[i].visualLimit-visualIndex-1;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ubidi_getLogicalMap(UBiDi *pBiDi, int32_t *indexMap, UErrorCode *pErrorCode) {
michael@0:     RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode);
michael@0:     /* ubidi_countRuns() checks for VALID_PARA_OR_LINE */
michael@0:     ubidi_countRuns(pBiDi, pErrorCode);
michael@0:     if(U_FAILURE(*pErrorCode)) {
michael@0:         /* no op */
michael@0:     } else if(indexMap==NULL) {
michael@0:         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:     } else {
michael@0:         /* fill a logical-to-visual index map using the runs[] */
michael@0:         int32_t visualStart, visualLimit, i, j, k;
michael@0:         int32_t logicalStart, logicalLimit;
michael@0:         Run *runs=pBiDi->runs;
michael@0:         if (pBiDi->length<=0) {
michael@0:             return;
michael@0:         }
michael@0:         if (pBiDi->length>pBiDi->resultLength) {
michael@0:             uprv_memset(indexMap, 0xFF, pBiDi->length*sizeof(int32_t));
michael@0:         }
michael@0: 
michael@0:         visualStart=0;
michael@0:         for(j=0; j<pBiDi->runCount; ++j) {
michael@0:             logicalStart=GET_INDEX(runs[j].logicalStart);
michael@0:             visualLimit=runs[j].visualLimit;
michael@0:             if(IS_EVEN_RUN(runs[j].logicalStart)) {
michael@0:                 do { /* LTR */
michael@0:                     indexMap[logicalStart++]=visualStart++;
michael@0:                 } while(visualStart<visualLimit);
michael@0:             } else {
michael@0:                 logicalStart+=visualLimit-visualStart;  /* logicalLimit */
michael@0:                 do { /* RTL */
michael@0:                     indexMap[--logicalStart]=visualStart++;
michael@0:                 } while(visualStart<visualLimit);
michael@0:             }
michael@0:             /* visualStart==visualLimit; */
michael@0:         }
michael@0: 
michael@0:         if(pBiDi->insertPoints.size>0) {
michael@0:             int32_t markFound=0, runCount=pBiDi->runCount;
michael@0:             int32_t length, insertRemove;
michael@0:             visualStart=0;
michael@0:             /* add number of marks found until each index */
michael@0:             for(i=0; i<runCount; i++, visualStart+=length) {
michael@0:                 length=runs[i].visualLimit-visualStart;
michael@0:                 insertRemove=runs[i].insertRemove;
michael@0:                 if(insertRemove&(LRM_BEFORE|RLM_BEFORE)) {
michael@0:                     markFound++;
michael@0:                 }
michael@0:                 if(markFound>0) {
michael@0:                     logicalStart=GET_INDEX(runs[i].logicalStart);
michael@0:                     logicalLimit=logicalStart+length;
michael@0:                     for(j=logicalStart; j<logicalLimit; j++) {
michael@0:                         indexMap[j]+=markFound;
michael@0:                     }
michael@0:                 }
michael@0:                 if(insertRemove&(LRM_AFTER|RLM_AFTER)) {
michael@0:                     markFound++;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:         else if(pBiDi->controlCount>0) {
michael@0:             int32_t controlFound=0, runCount=pBiDi->runCount;
michael@0:             int32_t length, insertRemove;
michael@0:             UBool evenRun;
michael@0:             UChar uchar;
michael@0:             visualStart=0;
michael@0:             /* subtract number of controls found until each index */
michael@0:             for(i=0; i<runCount; i++, visualStart+=length) {
michael@0:                 length=runs[i].visualLimit-visualStart;
michael@0:                 insertRemove=runs[i].insertRemove;
michael@0:                 /* no control found within previous runs nor within this run */
michael@0:                 if((controlFound-insertRemove)==0) {
michael@0:                     continue;
michael@0:                 }
michael@0:                 logicalStart=runs[i].logicalStart;
michael@0:                 evenRun=IS_EVEN_RUN(logicalStart);
michael@0:                 REMOVE_ODD_BIT(logicalStart);
michael@0:                 logicalLimit=logicalStart+length;
michael@0:                 /* if no control within this run */
michael@0:                 if(insertRemove==0) {
michael@0:                     for(j=logicalStart; j<logicalLimit; j++) {
michael@0:                         indexMap[j]-=controlFound;
michael@0:                     }
michael@0:                     continue;
michael@0:                 }
michael@0:                 for(j=0; j<length; j++) {
michael@0:                     k= evenRun ? logicalStart+j : logicalLimit-j-1;
michael@0:                     uchar=pBiDi->text[k];
michael@0:                     if(IS_BIDI_CONTROL_CHAR(uchar)) {
michael@0:                         controlFound++;
michael@0:                         indexMap[k]=UBIDI_MAP_NOWHERE;
michael@0:                         continue;
michael@0:                     }
michael@0:                     indexMap[k]-=controlFound;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ubidi_getVisualMap(UBiDi *pBiDi, int32_t *indexMap, UErrorCode *pErrorCode) {
michael@0:     RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode);
michael@0:     if(indexMap==NULL) {
michael@0:         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return;
michael@0:     }
michael@0:     /* ubidi_countRuns() checks for VALID_PARA_OR_LINE */
michael@0:     ubidi_countRuns(pBiDi, pErrorCode);
michael@0:     if(U_SUCCESS(*pErrorCode)) {
michael@0:         /* fill a visual-to-logical index map using the runs[] */
michael@0:         Run *runs=pBiDi->runs, *runsLimit=runs+pBiDi->runCount;
michael@0:         int32_t logicalStart, visualStart, visualLimit, *pi=indexMap;
michael@0: 
michael@0:         if (pBiDi->resultLength<=0) {
michael@0:             return;
michael@0:         }
michael@0:         visualStart=0;
michael@0:         for(; runs<runsLimit; ++runs) {
michael@0:             logicalStart=runs->logicalStart;
michael@0:             visualLimit=runs->visualLimit;
michael@0:             if(IS_EVEN_RUN(logicalStart)) {
michael@0:                 do { /* LTR */
michael@0:                     *pi++ = logicalStart++;
michael@0:                 } while(++visualStart<visualLimit);
michael@0:             } else {
michael@0:                 REMOVE_ODD_BIT(logicalStart);
michael@0:                 logicalStart+=visualLimit-visualStart;  /* logicalLimit */
michael@0:                 do { /* RTL */
michael@0:                     *pi++ = --logicalStart;
michael@0:                 } while(++visualStart<visualLimit);
michael@0:             }
michael@0:             /* visualStart==visualLimit; */
michael@0:         }
michael@0: 
michael@0:         if(pBiDi->insertPoints.size>0) {
michael@0:             int32_t markFound=0, runCount=pBiDi->runCount;
michael@0:             int32_t insertRemove, i, j, k;
michael@0:             runs=pBiDi->runs;
michael@0:             /* count all inserted marks */
michael@0:             for(i=0; i<runCount; i++) {
michael@0:                 insertRemove=runs[i].insertRemove;
michael@0:                 if(insertRemove&(LRM_BEFORE|RLM_BEFORE)) {
michael@0:                     markFound++;
michael@0:                 }
michael@0:                 if(insertRemove&(LRM_AFTER|RLM_AFTER)) {
michael@0:                     markFound++;
michael@0:                 }
michael@0:             }
michael@0:             /* move back indexes by number of preceding marks */
michael@0:             k=pBiDi->resultLength;
michael@0:             for(i=runCount-1; i>=0 && markFound>0; i--) {
michael@0:                 insertRemove=runs[i].insertRemove;
michael@0:                 if(insertRemove&(LRM_AFTER|RLM_AFTER)) {
michael@0:                     indexMap[--k]= UBIDI_MAP_NOWHERE;
michael@0:                     markFound--;
michael@0:                 }
michael@0:                 visualStart= i>0 ? runs[i-1].visualLimit : 0;
michael@0:                 for(j=runs[i].visualLimit-1; j>=visualStart && markFound>0; j--) {
michael@0:                     indexMap[--k]=indexMap[j];
michael@0:                 }
michael@0:                 if(insertRemove&(LRM_BEFORE|RLM_BEFORE)) {
michael@0:                     indexMap[--k]= UBIDI_MAP_NOWHERE;
michael@0:                     markFound--;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:         else if(pBiDi->controlCount>0) {
michael@0:             int32_t runCount=pBiDi->runCount, logicalEnd;
michael@0:             int32_t insertRemove, length, i, j, k, m;
michael@0:             UChar uchar;
michael@0:             UBool evenRun;
michael@0:             runs=pBiDi->runs;
michael@0:             visualStart=0;
michael@0:             /* move forward indexes by number of preceding controls */
michael@0:             k=0;
michael@0:             for(i=0; i<runCount; i++, visualStart+=length) {
michael@0:                 length=runs[i].visualLimit-visualStart;
michael@0:                 insertRemove=runs[i].insertRemove;
michael@0:                 /* if no control found yet, nothing to do in this run */
michael@0:                 if((insertRemove==0)&&(k==visualStart)) {
michael@0:                     k+=length;
michael@0:                     continue;
michael@0:                 }
michael@0:                 /* if no control in this run */
michael@0:                 if(insertRemove==0) {
michael@0:                     visualLimit=runs[i].visualLimit;
michael@0:                     for(j=visualStart; j<visualLimit; j++) {
michael@0:                         indexMap[k++]=indexMap[j];
michael@0:                     }
michael@0:                     continue;
michael@0:                 }
michael@0:                 logicalStart=runs[i].logicalStart;
michael@0:                 evenRun=IS_EVEN_RUN(logicalStart);
michael@0:                 REMOVE_ODD_BIT(logicalStart);
michael@0:                 logicalEnd=logicalStart+length-1;
michael@0:                 for(j=0; j<length; j++) {
michael@0:                     m= evenRun ? logicalStart+j : logicalEnd-j;
michael@0:                     uchar=pBiDi->text[m];
michael@0:                     if(!IS_BIDI_CONTROL_CHAR(uchar)) {
michael@0:                         indexMap[k++]=m;
michael@0:                     }
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ubidi_invertMap(const int32_t *srcMap, int32_t *destMap, int32_t length) {
michael@0:     if(srcMap!=NULL && destMap!=NULL && length>0) {
michael@0:         const int32_t *pi;
michael@0:         int32_t destLength=-1, count=0;
michael@0:         /* find highest value and count positive indexes in srcMap */
michael@0:         pi=srcMap+length;
michael@0:         while(pi>srcMap) {
michael@0:             if(*--pi>destLength) {
michael@0:                 destLength=*pi;
michael@0:             }
michael@0:             if(*pi>=0) {
michael@0:                 count++;
michael@0:             }
michael@0:         }
michael@0:         destLength++;           /* add 1 for origin 0 */
michael@0:         if(count<destLength) {
michael@0:             /* we must fill unmatched destMap entries with -1 */
michael@0:             uprv_memset(destMap, 0xFF, destLength*sizeof(int32_t));
michael@0:         }
michael@0:         pi=srcMap+length;
michael@0:         while(length>0) {
michael@0:             if(*--pi>=0) {
michael@0:                 destMap[*pi]=--length;
michael@0:             } else {
michael@0:                 --length;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: }