sl@0: /*
sl@0: ******************************************************************************
sl@0: *
sl@0: *   Copyright (C) 1999-2005, International Business Machines
sl@0: *   Corporation and others.  All Rights Reserved.
sl@0: *
sl@0: ******************************************************************************
sl@0: *   file name:  ubidiimp.h
sl@0: *   encoding:   US-ASCII
sl@0: *   tab size:   8 (not used)
sl@0: *   indentation:4
sl@0: *
sl@0: *   created on: 1999aug06
sl@0: *   created by: Markus W. Scherer
sl@0: */
sl@0: 
sl@0: #ifndef UBIDIIMP_H
sl@0: #define UBIDIIMP_H
sl@0: 
sl@0: /* set import/export definitions */
sl@0: #ifdef U_COMMON_IMPLEMENTATION
sl@0: 
sl@0: #include "unicode/utypes.h"
sl@0: #include "unicode/uchar.h"
sl@0: #include "ubidi_props.h"
sl@0: 
sl@0: /* miscellaneous definitions ---------------------------------------------- */
sl@0: 
sl@0: typedef uint8_t DirProp;
sl@0: typedef uint32_t Flags;
sl@0: 
sl@0: /*  Comparing the description of the BiDi algorithm with this implementation
sl@0:     is easier with the same names for the BiDi types in the code as there.
sl@0:     See UCharDirection in uchar.h .
sl@0: */
sl@0: enum {
sl@0:     L=  U_LEFT_TO_RIGHT,
sl@0:     R=  U_RIGHT_TO_LEFT,
sl@0:     EN= U_EUROPEAN_NUMBER,
sl@0:     ES= U_EUROPEAN_NUMBER_SEPARATOR,
sl@0:     ET= U_EUROPEAN_NUMBER_TERMINATOR,
sl@0:     AN= U_ARABIC_NUMBER,
sl@0:     CS= U_COMMON_NUMBER_SEPARATOR,
sl@0:     B=  U_BLOCK_SEPARATOR,
sl@0:     S=  U_SEGMENT_SEPARATOR,
sl@0:     WS= U_WHITE_SPACE_NEUTRAL,
sl@0:     ON= U_OTHER_NEUTRAL,
sl@0:     LRE=U_LEFT_TO_RIGHT_EMBEDDING,
sl@0:     LRO=U_LEFT_TO_RIGHT_OVERRIDE,
sl@0:     AL= U_RIGHT_TO_LEFT_ARABIC,
sl@0:     RLE=U_RIGHT_TO_LEFT_EMBEDDING,
sl@0:     RLO=U_RIGHT_TO_LEFT_OVERRIDE,
sl@0:     PDF=U_POP_DIRECTIONAL_FORMAT,
sl@0:     NSM=U_DIR_NON_SPACING_MARK,
sl@0:     BN= U_BOUNDARY_NEUTRAL,
sl@0:     dirPropCount
sl@0: };
sl@0: 
sl@0: /*
sl@0:  * Sometimes, bit values are more appropriate
sl@0:  * to deal with directionality properties.
sl@0:  * Abbreviations in these macro names refer to names
sl@0:  * used in the BiDi algorithm.
sl@0:  */
sl@0: #define DIRPROP_FLAG(dir) (1UL<<(dir))
sl@0: 
sl@0: /* special flag for multiple runs from explicit embedding codes */
sl@0: #define DIRPROP_FLAG_MULTI_RUNS (1UL<<31)
sl@0: 
sl@0: /* are there any characters that are LTR or RTL? */
sl@0: #define MASK_LTR (DIRPROP_FLAG(L)|DIRPROP_FLAG(EN)|DIRPROP_FLAG(AN)|DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO))
sl@0: #define MASK_RTL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL)|DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO))
sl@0: 
sl@0: /* explicit embedding codes */
sl@0: #define MASK_LRX (DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO))
sl@0: #define MASK_RLX (DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO))
sl@0: #define MASK_OVERRIDE (DIRPROP_FLAG(LRO)|DIRPROP_FLAG(RLO))
sl@0: 
sl@0: #define MASK_EXPLICIT (MASK_LRX|MASK_RLX|DIRPROP_FLAG(PDF))
sl@0: #define MASK_BN_EXPLICIT (DIRPROP_FLAG(BN)|MASK_EXPLICIT)
sl@0: 
sl@0: /* paragraph and segment separators */
sl@0: #define MASK_B_S (DIRPROP_FLAG(B)|DIRPROP_FLAG(S))
sl@0: 
sl@0: /* all types that are counted as White Space or Neutral in some steps */
sl@0: #define MASK_WS (MASK_B_S|DIRPROP_FLAG(WS)|MASK_BN_EXPLICIT)
sl@0: #define MASK_N (DIRPROP_FLAG(ON)|MASK_WS)
sl@0: 
sl@0: /* all types that are included in a sequence of European Terminators for (W5) */
sl@0: #define MASK_ET_NSM_BN (DIRPROP_FLAG(ET)|DIRPROP_FLAG(NSM)|MASK_BN_EXPLICIT)
sl@0: 
sl@0: /* types that are neutrals or could becomes neutrals in (Wn) */
sl@0: #define MASK_POSSIBLE_N (DIRPROP_FLAG(CS)|DIRPROP_FLAG(ES)|DIRPROP_FLAG(ET)|MASK_N)
sl@0: 
sl@0: /*
sl@0:  * These types may be changed to "e",
sl@0:  * the embedding type (L or R) of the run,
sl@0:  * in the BiDi algorithm (N2)
sl@0:  */
sl@0: #define MASK_EMBEDDING (DIRPROP_FLAG(NSM)|MASK_POSSIBLE_N)
sl@0: 
sl@0: /* the dirProp's L and R are defined to 0 and 1 values in UCharDirection */
sl@0: #define GET_LR_FROM_LEVEL(level) ((DirProp)((level)&1))
sl@0: 
sl@0: #define IS_DEFAULT_LEVEL(level) ((level)>=0xfe)
sl@0: 
sl@0: /*
sl@0:  * The following bit is ORed to the property of characters in paragraphs
sl@0:  * with contextual RTL direction when paraLevel is contextual.
sl@0:  */
sl@0: #define CONTEXT_RTL 0x80
sl@0: #define NO_CONTEXT_RTL(dir) ((dir)&~CONTEXT_RTL)
sl@0: /*
sl@0:  * The following is a variant of DIRPROP_FLAG which ignores the CONTEXT_RTL bit.
sl@0:  */
sl@0: #define DIRPROP_FLAG_NC(dir) (1UL<<(NO_CONTEXT_RTL(dir)))
sl@0: 
sl@0: #define GET_PARALEVEL(ubidi, index) \
sl@0:             (UBiDiLevel)((ubidi)->defaultParaLevel ? (ubidi)->dirProps[index]>>7 \
sl@0:                                                    : (ubidi)->paraLevel)
sl@0: 
sl@0: /* Paragraph type for multiple paragraph support ---   -------------------- */
sl@0: typedef int32_t Para;
sl@0: 
sl@0: #define CR  0x000D
sl@0: #define LF  0x000A
sl@0: 
sl@0: /* Run structure for reordering --------------------------------------------- */
sl@0: 
sl@0: typedef struct Run {
sl@0:     int32_t logicalStart,   /* first character of the run; b31 indicates even/odd level */
sl@0:             visualLimit;    /* last visual position of the run +1 */
sl@0: } Run;
sl@0: 
sl@0: /* in a Run, logicalStart will get this bit set if the run level is odd */
sl@0: #define INDEX_ODD_BIT (1UL<<31)
sl@0: 
sl@0: #define MAKE_INDEX_ODD_PAIR(index, level) (index|((int32_t)level<<31))
sl@0: #define ADD_ODD_BIT_FROM_LEVEL(x, level)  ((x)|=((int32_t)level<<31))
sl@0: #define REMOVE_ODD_BIT(x)                 ((x)&=~INDEX_ODD_BIT)
sl@0: 
sl@0: #define GET_INDEX(x)   (x&~INDEX_ODD_BIT)
sl@0: #define GET_ODD_BIT(x) ((uint32_t)x>>31)
sl@0: #define IS_ODD_RUN(x)  ((x&INDEX_ODD_BIT)!=0)
sl@0: #define IS_EVEN_RUN(x) ((x&INDEX_ODD_BIT)==0)
sl@0: 
sl@0: U_CFUNC UBool
sl@0: ubidi_getRuns(UBiDi *pBiDi);
sl@0: 
sl@0: /* UBiDi structure ----------------------------------------------------------- */
sl@0: 
sl@0: struct UBiDi {
sl@0:     /* pointer to parent paragraph object (pointer to self if this object is
sl@0:      * a paragraph object); set to NULL in a newly opened object; set to a
sl@0:      * real value after a successful execution of ubidi_setPara or ubidi_setLine
sl@0:      */
sl@0:     const UBiDi * pParaBiDi;
sl@0: 
sl@0:     const UBiDiProps *bdp;
sl@0: 
sl@0:     /* alias pointer to the current text */
sl@0:     const UChar *text;
sl@0: 
sl@0:     /* length of the current text */
sl@0:     int32_t length;
sl@0: 
sl@0:     /* memory sizes in bytes */
sl@0:     int32_t dirPropsSize, levelsSize, parasSize, runsSize;
sl@0: 
sl@0:     /* allocated memory */
sl@0:     DirProp *dirPropsMemory;
sl@0:     UBiDiLevel *levelsMemory;
sl@0:     Para *parasMemory;
sl@0:     Run *runsMemory;
sl@0: 
sl@0:     /* indicators for whether memory may be allocated after ubidi_open() */
sl@0:     UBool mayAllocateText, mayAllocateRuns;
sl@0: 
sl@0:     /* arrays with one value per text-character */
sl@0:     const DirProp *dirProps;
sl@0:     UBiDiLevel *levels;
sl@0: 
sl@0:     /* are we performing an approximation of the "inverse BiDi" algorithm? */
sl@0:     UBool isInverse;
sl@0:     UBool isInverse2;
sl@0: 
sl@0:     /* must block separators receive level 0? */
sl@0:     UBool orderParagraphsLTR;
sl@0: 
sl@0:     /* the paragraph level */
sl@0:     UBiDiLevel paraLevel;
sl@0:     /* original paraLevel when contextual */
sl@0:     /* must be one of UBIDI_DEFAULT_xxx or 0 if not contextual */
sl@0:     UBiDiLevel defaultParaLevel;
sl@0: 
sl@0:     /* the following is set in ubidi_setPara, used in processPropertySeq */
sl@0:     const struct ImpTabPair * pImpTabPair;  /* pointer to levels state table pair */
sl@0: 
sl@0:     /* the overall paragraph or line directionality - see UBiDiDirection */
sl@0:     UBiDiDirection direction;
sl@0: 
sl@0:     /* flags is a bit set for which directional properties are in the text */
sl@0:     Flags flags;
sl@0: 
sl@0:     /* characters after trailingWSStart are WS and are */
sl@0:     /* implicitly at the paraLevel (rule (L1)) - levels may not reflect that */
sl@0:     int32_t trailingWSStart;
sl@0: 
sl@0:     /* fields for paragraph handling */
sl@0:     int32_t paraCount;                  /* set in getDirProps() */
sl@0:     Para *paras;                        /* limits of paragraphs, filled in
sl@0:                             ResolveExplicitLevels() or CheckExplicitLevels() */
sl@0: 
sl@0:     /* for single paragraph text, we only need a tiny array of paras (no malloc()) */
sl@0:     Para simpleParas[1];
sl@0: 
sl@0:     /* fields for line reordering */
sl@0:     int32_t runCount;     /* ==-1: runs not set up yet */
sl@0:     Run *runs;
sl@0: 
sl@0:     /* for non-mixed text, we only need a tiny array of runs (no malloc()) */
sl@0:     Run simpleRuns[1];
sl@0: };
sl@0: 
sl@0: #define IS_VALID_PARA(x) ((x) && ((x)->pParaBiDi==(x)))
sl@0: #define IS_VALID_LINE(x) ((x) && ((x)->pParaBiDi) && ((x)->pParaBiDi->pParaBiDi==(x)->pParaBiDi))
sl@0: #define IS_VALID_PARA_OR_LINE(x) ((x) && ((x)->pParaBiDi==(x) || (((x)->pParaBiDi) && (x)->pParaBiDi->pParaBiDi==(x)->pParaBiDi)))
sl@0: 
sl@0: /* helper function to (re)allocate memory if allowed */
sl@0: U_CFUNC UBool
sl@0: ubidi_getMemory(void **pMemory, int32_t *pSize, UBool mayAllocate, int32_t sizeNeeded);
sl@0: 
sl@0: /* helper macros for each allocated array in UBiDi */
sl@0: #define getDirPropsMemory(pBiDi, length) \
sl@0:         ubidi_getMemory((void **)&(pBiDi)->dirPropsMemory, &(pBiDi)->dirPropsSize, \
sl@0:                         (pBiDi)->mayAllocateText, (length))
sl@0: 
sl@0: #define getLevelsMemory(pBiDi, length) \
sl@0:         ubidi_getMemory((void **)&(pBiDi)->levelsMemory, &(pBiDi)->levelsSize, \
sl@0:                         (pBiDi)->mayAllocateText, (length))
sl@0: 
sl@0: #define getRunsMemory(pBiDi, length) \
sl@0:         ubidi_getMemory((void **)&(pBiDi)->runsMemory, &(pBiDi)->runsSize, \
sl@0:                         (pBiDi)->mayAllocateRuns, (length)*sizeof(Run))
sl@0: 
sl@0: /* additional macros used by ubidi_open() - always allow allocation */
sl@0: #define getInitialDirPropsMemory(pBiDi, length) \
sl@0:         ubidi_getMemory((void **)&(pBiDi)->dirPropsMemory, &(pBiDi)->dirPropsSize, \
sl@0:                         TRUE, (length))
sl@0: 
sl@0: #define getInitialLevelsMemory(pBiDi, length) \
sl@0:         ubidi_getMemory((void **)&(pBiDi)->levelsMemory, &(pBiDi)->levelsSize, \
sl@0:                         TRUE, (length))
sl@0: 
sl@0: #define getInitialParasMemory(pBiDi, length) \
sl@0:         ubidi_getMemory((void **)&(pBiDi)->parasMemory, &(pBiDi)->parasSize, \
sl@0:                         TRUE, (length)*sizeof(Para))
sl@0: 
sl@0: #define getInitialRunsMemory(pBiDi, length) \
sl@0:         ubidi_getMemory((void **)&(pBiDi)->runsMemory, &(pBiDi)->runsSize, \
sl@0:                         TRUE, (length)*sizeof(Run))
sl@0: 
sl@0: #endif
sl@0: 
sl@0: #endif