1934 lines
78 KiB
C
1934 lines
78 KiB
C
/*
|
|
******************************************************************************
|
|
*
|
|
* Copyright (C) 1999-2006, International Business Machines
|
|
* Corporation and others. All Rights Reserved.
|
|
*
|
|
******************************************************************************
|
|
* file name: ubidi.c
|
|
* encoding: US-ASCII
|
|
* tab size: 8 (not used)
|
|
* indentation:4
|
|
*
|
|
* created on: 1999jul27
|
|
* created by: Markus W. Scherer
|
|
*/
|
|
|
|
/* set import/export definitions */
|
|
#ifndef U_COMMON_IMPLEMENTATION
|
|
# define U_COMMON_IMPLEMENTATION
|
|
#endif
|
|
|
|
#include "cmemory.h"
|
|
#include "unicode/utypes.h"
|
|
#include "unicode/ustring.h"
|
|
#include "unicode/uchar.h"
|
|
#include "unicode/ubidi.h"
|
|
#include "ubidi_props.h"
|
|
#include "ubidiimp.h"
|
|
|
|
/*
|
|
* General implementation notes:
|
|
*
|
|
* Throughout the implementation, there are comments like (W2) that refer to
|
|
* rules of the BiDi algorithm in its version 5, in this example to the second
|
|
* rule of the resolution of weak types.
|
|
*
|
|
* For handling surrogate pairs, where two UChar's form one "abstract" (or UTF-32)
|
|
* character according to UTF-16, the second UChar gets the directional property of
|
|
* the entire character assigned, while the first one gets a BN, a boundary
|
|
* neutral, type, which is ignored by most of the algorithm according to
|
|
* rule (X9) and the implementation suggestions of the BiDi algorithm.
|
|
*
|
|
* Later, adjustWSLevels() will set the level for each BN to that of the
|
|
* following character (UChar), which results in surrogate pairs getting the
|
|
* same level on each of their surrogates.
|
|
*
|
|
* In a UTF-8 implementation, the same thing could be done: the last byte of
|
|
* a multi-byte sequence would get the "real" property, while all previous
|
|
* bytes of that sequence would get BN.
|
|
*
|
|
* It is not possible to assign all those parts of a character the same real
|
|
* property because this would fail in the resolution of weak types with rules
|
|
* that look at immediately surrounding types.
|
|
*
|
|
* As a related topic, this implementation does not remove Boundary Neutral
|
|
* types from the input, but ignores them wherever this is relevant.
|
|
* For example, the loop for the resolution of the weak types reads
|
|
* types until it finds a non-BN.
|
|
* Also, explicit embedding codes are neither changed into BN nor removed.
|
|
* They are only treated the same way real BNs are.
|
|
* As stated before, adjustWSLevels() takes care of them at the end.
|
|
* For the purpose of conformance, the levels of all these codes
|
|
* do not matter.
|
|
*
|
|
* Note that this implementation never modifies the dirProps
|
|
* after the initial setup.
|
|
*
|
|
*
|
|
* In this implementation, the resolution of weak types (Wn),
|
|
* neutrals (Nn), and the assignment of the resolved level (In)
|
|
* are all done in one single loop, in resolveImplicitLevels().
|
|
* Changes of dirProp values are done on the fly, without writing
|
|
* them back to the dirProps array.
|
|
*
|
|
*
|
|
* This implementation contains code that allows to bypass steps of the
|
|
* algorithm that are not needed on the specific paragraph
|
|
* in order to speed up the most common cases considerably,
|
|
* like text that is entirely LTR, or RTL text without numbers.
|
|
*
|
|
* Most of this is done by setting a bit for each directional property
|
|
* in a flags variable and later checking for whether there are
|
|
* any LTR characters or any RTL characters, or both, whether
|
|
* there are any explicit embedding codes, etc.
|
|
*
|
|
* If the (Xn) steps are performed, then the flags are re-evaluated,
|
|
* because they will then not contain the embedding codes any more
|
|
* and will be adjusted for override codes, so that subsequently
|
|
* more bypassing may be possible than what the initial flags suggested.
|
|
*
|
|
* If the text is not mixed-directional, then the
|
|
* algorithm steps for the weak type resolution are not performed,
|
|
* and all levels are set to the paragraph level.
|
|
*
|
|
* If there are no explicit embedding codes, then the (Xn) steps
|
|
* are not performed.
|
|
*
|
|
* If embedding levels are supplied as a parameter, then all
|
|
* explicit embedding codes are ignored, and the (Xn) steps
|
|
* are not performed.
|
|
*
|
|
* White Space types could get the level of the run they belong to,
|
|
* and are checked with a test of (flags&MASK_EMBEDDING) to
|
|
* consider if the paragraph direction should be considered in
|
|
* the flags variable.
|
|
*
|
|
* If there are no White Space types in the paragraph, then
|
|
* (L1) is not necessary in adjustWSLevels().
|
|
*/
|
|
|
|
/* to avoid some conditional statements, use tiny constant arrays */
|
|
static const Flags flagLR[2]={ DIRPROP_FLAG(L), DIRPROP_FLAG(R) };
|
|
static const Flags flagE[2]={ DIRPROP_FLAG(LRE), DIRPROP_FLAG(RLE) };
|
|
static const Flags flagO[2]={ DIRPROP_FLAG(LRO), DIRPROP_FLAG(RLO) };
|
|
|
|
#define DIRPROP_FLAG_LR(level) flagLR[(level)&1]
|
|
#define DIRPROP_FLAG_E(level) flagE[(level)&1]
|
|
#define DIRPROP_FLAG_O(level) flagO[(level)&1]
|
|
|
|
#define uLRM 0x200E
|
|
#define uRLM 0x200F
|
|
#define uLRE 0x202A
|
|
#define BEFORE 0
|
|
#define AFTER 1
|
|
|
|
/* UBiDi object management -------------------------------------------------- */
|
|
|
|
U_CAPI UBiDi * U_EXPORT2
|
|
ubidi_open(void)
|
|
{
|
|
UErrorCode errorCode=U_ZERO_ERROR;
|
|
return ubidi_openSized(0, 0, &errorCode);
|
|
}
|
|
|
|
U_CAPI UBiDi * U_EXPORT2
|
|
ubidi_openSized(int32_t maxLength, int32_t maxRunCount, UErrorCode *pErrorCode) {
|
|
UBiDi *pBiDi;
|
|
|
|
/* check the argument values */
|
|
if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
|
|
return NULL;
|
|
} else if(maxLength<0 || maxRunCount<0) {
|
|
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
|
|
return NULL; /* invalid arguments */
|
|
}
|
|
|
|
/* allocate memory for the object */
|
|
pBiDi=(UBiDi *)uprv_malloc(sizeof(UBiDi));
|
|
if(pBiDi==NULL) {
|
|
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
return NULL;
|
|
}
|
|
|
|
/* reset the object, all pointers NULL, all flags FALSE, all sizes 0 */
|
|
uprv_memset(pBiDi, 0, sizeof(UBiDi));
|
|
|
|
/* get BiDi properties */
|
|
pBiDi->bdp=ubidi_getSingleton(pErrorCode);
|
|
if(U_FAILURE(*pErrorCode)) {
|
|
uprv_free(pBiDi);
|
|
return NULL;
|
|
}
|
|
|
|
/* allocate memory for arrays as requested */
|
|
if(maxLength>0) {
|
|
if( !getInitialDirPropsMemory(pBiDi, maxLength) ||
|
|
!getInitialLevelsMemory(pBiDi, maxLength)
|
|
) {
|
|
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
}
|
|
} else {
|
|
pBiDi->mayAllocateText=TRUE;
|
|
}
|
|
|
|
if(maxRunCount>0) {
|
|
if(maxRunCount==1) {
|
|
/* use simpleRuns[] */
|
|
pBiDi->runsSize=sizeof(Run);
|
|
} else if(!getInitialRunsMemory(pBiDi, maxRunCount)) {
|
|
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
}
|
|
} else {
|
|
pBiDi->mayAllocateRuns=TRUE;
|
|
}
|
|
|
|
if(U_SUCCESS(*pErrorCode)) {
|
|
return pBiDi;
|
|
} else {
|
|
ubidi_close(pBiDi);
|
|
return NULL;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* We are allowed to allocate memory if memory==NULL or
|
|
* mayAllocate==TRUE for each array that we need.
|
|
* We also try to grow and shrink memory as needed if we
|
|
* allocate it.
|
|
*
|
|
* Assume sizeNeeded>0.
|
|
* If *pMemory!=NULL, then assume *pSize>0.
|
|
*
|
|
* ### this realloc() may unnecessarily copy the old data,
|
|
* which we know we don't need any more;
|
|
* is this the best way to do this??
|
|
*/
|
|
U_CFUNC UBool
|
|
ubidi_getMemory(void **pMemory, int32_t *pSize, UBool mayAllocate, int32_t sizeNeeded) {
|
|
/* check for existing memory */
|
|
if(*pMemory==NULL) {
|
|
/* we need to allocate memory */
|
|
if(mayAllocate && (*pMemory=uprv_malloc(sizeNeeded))!=NULL) {
|
|
*pSize=sizeNeeded;
|
|
return TRUE;
|
|
} else {
|
|
return FALSE;
|
|
}
|
|
} else {
|
|
/* there is some memory, is it enough or too much? */
|
|
if(sizeNeeded>*pSize && !mayAllocate) {
|
|
/* not enough memory, and we must not allocate */
|
|
return FALSE;
|
|
} else if(sizeNeeded!=*pSize && mayAllocate) {
|
|
/* we may try to grow or shrink */
|
|
void *memory;
|
|
|
|
if((memory=uprv_realloc(*pMemory, sizeNeeded))!=NULL) {
|
|
*pMemory=memory;
|
|
*pSize=sizeNeeded;
|
|
return TRUE;
|
|
} else {
|
|
/* we failed to grow */
|
|
return FALSE;
|
|
}
|
|
} else {
|
|
/* we have at least enough memory and must not allocate */
|
|
return TRUE;
|
|
}
|
|
}
|
|
}
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_close(UBiDi *pBiDi) {
|
|
if(pBiDi!=NULL) {
|
|
pBiDi->pParaBiDi=NULL; /* in case one tries to reuse this block */
|
|
if(pBiDi->dirPropsMemory!=NULL) {
|
|
uprv_free(pBiDi->dirPropsMemory);
|
|
}
|
|
if(pBiDi->levelsMemory!=NULL) {
|
|
uprv_free(pBiDi->levelsMemory);
|
|
}
|
|
if(pBiDi->runsMemory!=NULL) {
|
|
uprv_free(pBiDi->runsMemory);
|
|
}
|
|
if(pBiDi->parasMemory!=NULL) {
|
|
uprv_free(pBiDi->parasMemory);
|
|
}
|
|
if(pBiDi->insertPoints.points!=NULL) {
|
|
uprv_free(pBiDi->insertPoints.points);
|
|
}
|
|
|
|
uprv_free(pBiDi);
|
|
}
|
|
}
|
|
|
|
/* set to approximate "inverse BiDi" ---------------------------------------- */
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_setInverse(UBiDi *pBiDi, UBool isInverse) {
|
|
if(pBiDi!=NULL) {
|
|
pBiDi->isInverse=isInverse;
|
|
pBiDi->reorderingMode = isInverse ? UBIDI_REORDER_INVERSE_NUMBERS_AS_L
|
|
: UBIDI_REORDER_DEFAULT;
|
|
}
|
|
}
|
|
|
|
U_CAPI UBool U_EXPORT2
|
|
ubidi_isInverse(UBiDi *pBiDi) {
|
|
if(pBiDi!=NULL) {
|
|
return pBiDi->isInverse;
|
|
} else {
|
|
return FALSE;
|
|
}
|
|
}
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_setReorderingMode(UBiDi *pBiDi, UBiDiReorderingMode reorderingMode) {
|
|
if ((pBiDi != NULL) && (reorderingMode >= UBIDI_REORDER_DEFAULT)
|
|
&& (reorderingMode < UBIDI_REORDER_COUNT)) {
|
|
pBiDi->reorderingMode = reorderingMode;
|
|
pBiDi->isInverse = reorderingMode == UBIDI_REORDER_INVERSE_NUMBERS_AS_L;
|
|
}
|
|
}
|
|
|
|
U_CAPI UBiDiReorderingMode U_EXPORT2
|
|
ubidi_getReorderingMode(UBiDi *pBiDi) {
|
|
if (pBiDi != NULL) {
|
|
return pBiDi->reorderingMode;
|
|
} else {
|
|
return UBIDI_REORDER_DEFAULT;
|
|
}
|
|
}
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_setReorderingOptions(UBiDi *pBiDi, uint32_t reorderingOptions) {
|
|
if (reorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) {
|
|
reorderingOptions&=~UBIDI_OPTION_INSERT_MARKS;
|
|
}
|
|
if (pBiDi != NULL) {
|
|
pBiDi->reorderingOptions = reorderingOptions;
|
|
}
|
|
}
|
|
|
|
U_CAPI uint32_t U_EXPORT2
|
|
ubidi_getReorderingOptions(UBiDi *pBiDi) {
|
|
if (pBiDi != NULL) {
|
|
return pBiDi->reorderingOptions;
|
|
} else {
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
/* perform (P2)..(P3) ------------------------------------------------------- */
|
|
|
|
/*
|
|
* Get the directional properties for the text,
|
|
* calculate the flags bit-set, and
|
|
* determine the paragraph level if necessary.
|
|
*/
|
|
static void
|
|
getDirProps(UBiDi *pBiDi) {
|
|
const UChar *text=pBiDi->text;
|
|
DirProp *dirProps=pBiDi->dirPropsMemory; /* pBiDi->dirProps is const */
|
|
|
|
int32_t i=0, i0, i1, length=pBiDi->originalLength;
|
|
Flags flags=0; /* collect all directionalities in the text */
|
|
UChar32 uchar;
|
|
DirProp dirProp=0, paraDirDefault=0;/* initialize to avoid compiler warnings */
|
|
UBool isDefaultLevel=IS_DEFAULT_LEVEL(pBiDi->paraLevel);
|
|
UBool paraLevelStillDefault=FALSE; /* flag for real value not set */
|
|
UBool evenPara;
|
|
int32_t countBiDiControls = 0;
|
|
UBool removeBiDiControls = pBiDi->reorderingOptions &
|
|
UBIDI_OPTION_REMOVE_CONTROLS;
|
|
|
|
typedef enum {
|
|
NOT_CONTEXTUAL, /* 0: not contextual paraLevel */
|
|
LOOKING_FOR_STRONG, /* 1: looking for first strong char */
|
|
FOUND_STRONG_CHAR /* 2: found first strong char */
|
|
} State;
|
|
State state;
|
|
int32_t paraStart=0; /* index of first char in paragraph */
|
|
DirProp paraDir; /* == CONTEXT_RTL within paragraphs
|
|
starting with strong R char */
|
|
|
|
if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) {
|
|
pBiDi->length=0;
|
|
}
|
|
if(isDefaultLevel) {
|
|
paraDirDefault=pBiDi->paraLevel&1 ? CONTEXT_RTL : 0;
|
|
state=LOOKING_FOR_STRONG;
|
|
paraStart=0;
|
|
paraDir=paraDirDefault;
|
|
pBiDi->paraLevel&=1; /* set to default */
|
|
paraLevelStillDefault=TRUE;
|
|
} else {
|
|
state=NOT_CONTEXTUAL;
|
|
paraDir=0;
|
|
evenPara=!(pBiDi->paraLevel&1);
|
|
}
|
|
/* count paragraphs and determine the paragraph level (P2..P3) */
|
|
/*
|
|
* see comment in ubidi.h:
|
|
* the DEFAULT_XXX values are designed so that
|
|
* their bit 0 alone yields the intended default
|
|
*/
|
|
for( /* i=0 above */ ; i<length; ) {
|
|
/* i is incremented by UTF_NEXT_CHAR */
|
|
i0=i; /* index of first code unit */
|
|
UTF_NEXT_CHAR(text, i, length, uchar);
|
|
i1=i-1; /* index of last code unit, gets the directional property */
|
|
flags|=DIRPROP_FLAG(dirProp=ubidi_getCustomizedClass(pBiDi, uchar));
|
|
dirProps[i1]=dirProp|paraDir;
|
|
if(i1>i0) { /* set previous code units' properties to BN */
|
|
flags|=DIRPROP_FLAG(BN);
|
|
do {
|
|
dirProps[--i1]=BN|paraDir;
|
|
} while(i1>i0);
|
|
}
|
|
if(state==LOOKING_FOR_STRONG) {
|
|
if(dirProp==L) {
|
|
evenPara=TRUE;
|
|
state=FOUND_STRONG_CHAR;
|
|
if(paraLevelStillDefault) {
|
|
paraLevelStillDefault=FALSE;
|
|
pBiDi->paraLevel=0;
|
|
}
|
|
if(paraDir) {
|
|
paraDir=0;
|
|
for(i1=paraStart; i1<=i; i1++) {
|
|
dirProps[i1]&=~CONTEXT_RTL;
|
|
}
|
|
}
|
|
continue;
|
|
}
|
|
if(dirProp==R || dirProp==AL) {
|
|
evenPara=FALSE;
|
|
state=FOUND_STRONG_CHAR;
|
|
if(paraLevelStillDefault) {
|
|
paraLevelStillDefault=FALSE;
|
|
pBiDi->paraLevel=1;
|
|
}
|
|
if(paraDir==0) {
|
|
paraDir=CONTEXT_RTL;
|
|
for(i1=paraStart; i1<=i; i1++) {
|
|
dirProps[i1]|=CONTEXT_RTL;
|
|
}
|
|
}
|
|
continue;
|
|
}
|
|
}
|
|
if((dirProp==L) && evenPara &&
|
|
(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING)) {
|
|
pBiDi->length = i; /* i is index to next character */
|
|
}
|
|
if(dirProp==B) {
|
|
if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) {
|
|
pBiDi->length = i; /* i is index to next character */
|
|
}
|
|
if(i<length) { /* B not last char in text */
|
|
if(!((uchar==CR) && (text[i]==LF))) {
|
|
pBiDi->paraCount++;
|
|
}
|
|
if(isDefaultLevel) {
|
|
state=LOOKING_FOR_STRONG;
|
|
paraStart=i; /* i is index to next character */
|
|
paraDir=paraDirDefault;
|
|
/* keep the paraLevel of the first paragraph even if it
|
|
defaulted (no strong char was found) */
|
|
paraLevelStillDefault=FALSE;
|
|
}
|
|
}
|
|
}
|
|
if(removeBiDiControls) {
|
|
if(((uint32_t)(uchar - uLRM)<2) || ((uint32_t)(uchar - uLRE)<5)) {
|
|
countBiDiControls++;
|
|
}
|
|
}
|
|
}
|
|
/* The following line does nothing new for contextual paraLevel, but is
|
|
needed for absolute paraLevel. */
|
|
flags|=DIRPROP_FLAG_LR(pBiDi->paraLevel);
|
|
|
|
if(pBiDi->orderParagraphsLTR && (flags&DIRPROP_FLAG(B))) {
|
|
flags|=DIRPROP_FLAG(L);
|
|
}
|
|
if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) {
|
|
if(pBiDi->length<pBiDi->originalLength) {
|
|
pBiDi->paraCount--;
|
|
}
|
|
}
|
|
|
|
pBiDi->countBiDiControls = countBiDiControls;
|
|
pBiDi->flags=flags;
|
|
}
|
|
|
|
/* perform (X1)..(X9) ------------------------------------------------------- */
|
|
|
|
/* determine if the text is mixed-directional or single-directional */
|
|
static UBiDiDirection
|
|
directionFromFlags(Flags flags) {
|
|
/* if the text contains AN and neutrals, then some neutrals may become RTL */
|
|
if(!(flags&MASK_RTL || ((flags&DIRPROP_FLAG(AN)) && (flags&MASK_POSSIBLE_N)))) {
|
|
return UBIDI_LTR;
|
|
} else if(!(flags&MASK_LTR)) {
|
|
return UBIDI_RTL;
|
|
} else {
|
|
return UBIDI_MIXED;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* Resolve the explicit levels as specified by explicit embedding codes.
|
|
* Recalculate the flags to have them reflect the real properties
|
|
* after taking the explicit embeddings into account.
|
|
*
|
|
* The BiDi algorithm is designed to result in the same behavior whether embedding
|
|
* levels are externally specified (from "styled text", supposedly the preferred
|
|
* method) or set by explicit embedding codes (LRx, RLx, PDF) in the plain text.
|
|
* That is why (X9) instructs to remove all explicit codes (and BN).
|
|
* However, in a real implementation, this removal of these codes and their index
|
|
* positions in the plain text is undesirable since it would result in
|
|
* reallocated, reindexed text.
|
|
* Instead, this implementation leaves the codes in there and just ignores them
|
|
* in the subsequent processing.
|
|
* In order to get the same reordering behavior, positions with a BN or an
|
|
* explicit embedding code just get the same level assigned as the last "real"
|
|
* character.
|
|
*
|
|
* Some implementations, not this one, then overwrite some of these
|
|
* directionality properties at "real" same-level-run boundaries by
|
|
* L or R codes so that the resolution of weak types can be performed on the
|
|
* entire paragraph at once instead of having to parse it once more and
|
|
* perform that resolution on same-level-runs.
|
|
* This limits the scope of the implicit rules in effectively
|
|
* the same way as the run limits.
|
|
*
|
|
* Instead, this implementation does not modify these codes.
|
|
* On one hand, the paragraph has to be scanned for same-level-runs, but
|
|
* on the other hand, this saves another loop to reset these codes,
|
|
* or saves making and modifying a copy of dirProps[].
|
|
*
|
|
*
|
|
* Note that (Pn) and (Xn) changed significantly from version 4 of the BiDi algorithm.
|
|
*
|
|
*
|
|
* Handling the stack of explicit levels (Xn):
|
|
*
|
|
* With the BiDi stack of explicit levels,
|
|
* as pushed with each LRE, RLE, LRO, and RLO and popped with each PDF,
|
|
* the explicit level must never exceed UBIDI_MAX_EXPLICIT_LEVEL==61.
|
|
*
|
|
* In order to have a correct push-pop semantics even in the case of overflows,
|
|
* there are two overflow counters:
|
|
* - countOver60 is incremented with each LRx at level 60
|
|
* - from level 60, one RLx increases the level to 61
|
|
* - countOver61 is incremented with each LRx and RLx at level 61
|
|
*
|
|
* Popping levels with PDF must work in the opposite order so that level 61
|
|
* is correct at the correct point. Underflows (too many PDFs) must be checked.
|
|
*
|
|
* This implementation assumes that UBIDI_MAX_EXPLICIT_LEVEL is odd.
|
|
*/
|
|
static UBiDiDirection
|
|
resolveExplicitLevels(UBiDi *pBiDi) {
|
|
const DirProp *dirProps=pBiDi->dirProps;
|
|
UBiDiLevel *levels=pBiDi->levels;
|
|
const UChar *text=pBiDi->text;
|
|
|
|
int32_t i=0, length=pBiDi->length;
|
|
Flags flags=pBiDi->flags; /* collect all directionalities in the text */
|
|
DirProp dirProp;
|
|
UBiDiLevel level=GET_PARALEVEL(pBiDi, 0);
|
|
|
|
UBiDiDirection direction;
|
|
int32_t paraIndex=0;
|
|
|
|
/* determine if the text is mixed-directional or single-directional */
|
|
direction=directionFromFlags(flags);
|
|
|
|
/* we may not need to resolve any explicit levels, but for multiple
|
|
paragraphs we want to loop on all chars to set the para boundaries */
|
|
if((direction!=UBIDI_MIXED) && (pBiDi->paraCount==1)) {
|
|
/* not mixed directionality: levels don't matter - trailingWSStart will be 0 */
|
|
} else if((pBiDi->paraCount==1) &&
|
|
(!(flags&MASK_EXPLICIT) ||
|
|
(pBiDi->reorderingMode > UBIDI_REORDER_LAST_LOGICAL_TO_VISUAL))) {
|
|
/* mixed, but all characters are at the same embedding level */
|
|
/* or we are in "inverse BiDi" */
|
|
/* and we don't have contextual multiple paragraphs with some B char */
|
|
/* set all levels to the paragraph level */
|
|
for(i=0; i<length; ++i) {
|
|
levels[i]=level;
|
|
}
|
|
} else {
|
|
/* continue to perform (Xn) */
|
|
|
|
/* (X1) level is set for all codes, embeddingLevel keeps track of the push/pop operations */
|
|
/* both variables may carry the UBIDI_LEVEL_OVERRIDE flag to indicate the override status */
|
|
UBiDiLevel embeddingLevel=level, newLevel, stackTop=0;
|
|
|
|
UBiDiLevel stack[UBIDI_MAX_EXPLICIT_LEVEL]; /* we never push anything >=UBIDI_MAX_EXPLICIT_LEVEL */
|
|
uint32_t countOver60=0, countOver61=0; /* count overflows of explicit levels */
|
|
|
|
/* recalculate the flags */
|
|
flags=0;
|
|
|
|
for(i=0; i<length; ++i) {
|
|
dirProp=NO_CONTEXT_RTL(dirProps[i]);
|
|
switch(dirProp) {
|
|
case LRE:
|
|
case LRO:
|
|
/* (X3, X5) */
|
|
newLevel=(UBiDiLevel)((embeddingLevel+2)&~(UBIDI_LEVEL_OVERRIDE|1)); /* least greater even level */
|
|
if(newLevel<=UBIDI_MAX_EXPLICIT_LEVEL) {
|
|
stack[stackTop]=embeddingLevel;
|
|
++stackTop;
|
|
embeddingLevel=newLevel;
|
|
if(dirProp==LRO) {
|
|
embeddingLevel|=UBIDI_LEVEL_OVERRIDE;
|
|
}
|
|
/* we don't need to set UBIDI_LEVEL_OVERRIDE off for LRE
|
|
since this has already been done for newLevel which is
|
|
the source for embeddingLevel.
|
|
*/
|
|
} else if((embeddingLevel&~UBIDI_LEVEL_OVERRIDE)==UBIDI_MAX_EXPLICIT_LEVEL) {
|
|
++countOver61;
|
|
} else /* (embeddingLevel&~UBIDI_LEVEL_OVERRIDE)==UBIDI_MAX_EXPLICIT_LEVEL-1 */ {
|
|
++countOver60;
|
|
}
|
|
flags|=DIRPROP_FLAG(BN);
|
|
break;
|
|
case RLE:
|
|
case RLO:
|
|
/* (X2, X4) */
|
|
newLevel=(UBiDiLevel)(((embeddingLevel&~UBIDI_LEVEL_OVERRIDE)+1)|1); /* least greater odd level */
|
|
if(newLevel<=UBIDI_MAX_EXPLICIT_LEVEL) {
|
|
stack[stackTop]=embeddingLevel;
|
|
++stackTop;
|
|
embeddingLevel=newLevel;
|
|
if(dirProp==RLO) {
|
|
embeddingLevel|=UBIDI_LEVEL_OVERRIDE;
|
|
}
|
|
/* we don't need to set UBIDI_LEVEL_OVERRIDE off for RLE
|
|
since this has already been done for newLevel which is
|
|
the source for embeddingLevel.
|
|
*/
|
|
} else {
|
|
++countOver61;
|
|
}
|
|
flags|=DIRPROP_FLAG(BN);
|
|
break;
|
|
case PDF:
|
|
/* (X7) */
|
|
/* handle all the overflow cases first */
|
|
if(countOver61>0) {
|
|
--countOver61;
|
|
} else if(countOver60>0 && (embeddingLevel&~UBIDI_LEVEL_OVERRIDE)!=UBIDI_MAX_EXPLICIT_LEVEL) {
|
|
/* handle LRx overflows from level 60 */
|
|
--countOver60;
|
|
} else if(stackTop>0) {
|
|
/* this is the pop operation; it also pops level 61 while countOver60>0 */
|
|
--stackTop;
|
|
embeddingLevel=stack[stackTop];
|
|
/* } else { (underflow) */
|
|
}
|
|
flags|=DIRPROP_FLAG(BN);
|
|
break;
|
|
case B:
|
|
stackTop=0;
|
|
countOver60=countOver61=0;
|
|
level=GET_PARALEVEL(pBiDi, i);
|
|
if((i+1)<length) {
|
|
embeddingLevel=GET_PARALEVEL(pBiDi, i+1);
|
|
if(!((text[i]==CR) && (text[i+1]==LF))) {
|
|
pBiDi->paras[paraIndex++]=i+1;
|
|
}
|
|
}
|
|
flags|=DIRPROP_FLAG(B);
|
|
break;
|
|
case BN:
|
|
/* BN, LRE, RLE, and PDF are supposed to be removed (X9) */
|
|
/* they will get their levels set correctly in adjustWSLevels() */
|
|
flags|=DIRPROP_FLAG(BN);
|
|
break;
|
|
default:
|
|
/* all other types get the "real" level */
|
|
if(level!=embeddingLevel) {
|
|
level=embeddingLevel;
|
|
if(level&UBIDI_LEVEL_OVERRIDE) {
|
|
flags|=DIRPROP_FLAG_O(level)|DIRPROP_FLAG_MULTI_RUNS;
|
|
} else {
|
|
flags|=DIRPROP_FLAG_E(level)|DIRPROP_FLAG_MULTI_RUNS;
|
|
}
|
|
}
|
|
if(!(level&UBIDI_LEVEL_OVERRIDE)) {
|
|
flags|=DIRPROP_FLAG(dirProp);
|
|
}
|
|
break;
|
|
}
|
|
|
|
/*
|
|
* We need to set reasonable levels even on BN codes and
|
|
* explicit codes because we will later look at same-level runs (X10).
|
|
*/
|
|
levels[i]=level;
|
|
}
|
|
if(flags&MASK_EMBEDDING) {
|
|
flags|=DIRPROP_FLAG_LR(pBiDi->paraLevel);
|
|
}
|
|
if(pBiDi->orderParagraphsLTR && (flags&DIRPROP_FLAG(B))) {
|
|
flags|=DIRPROP_FLAG(L);
|
|
}
|
|
|
|
/* subsequently, ignore the explicit codes and BN (X9) */
|
|
|
|
/* again, determine if the text is mixed-directional or single-directional */
|
|
pBiDi->flags=flags;
|
|
direction=directionFromFlags(flags);
|
|
}
|
|
|
|
return direction;
|
|
}
|
|
|
|
/*
|
|
* Use a pre-specified embedding levels array:
|
|
*
|
|
* Adjust the directional properties for overrides (->LEVEL_OVERRIDE),
|
|
* ignore all explicit codes (X9),
|
|
* and check all the preset levels.
|
|
*
|
|
* Recalculate the flags to have them reflect the real properties
|
|
* after taking the explicit embeddings into account.
|
|
*/
|
|
static UBiDiDirection
|
|
checkExplicitLevels(UBiDi *pBiDi, UErrorCode *pErrorCode) {
|
|
const DirProp *dirProps=pBiDi->dirProps;
|
|
DirProp dirProp;
|
|
UBiDiLevel *levels=pBiDi->levels;
|
|
const UChar *text=pBiDi->text;
|
|
|
|
int32_t i, length=pBiDi->length;
|
|
Flags flags=0; /* collect all directionalities in the text */
|
|
UBiDiLevel level;
|
|
uint32_t paraIndex=0;
|
|
|
|
for(i=0; i<length; ++i) {
|
|
level=levels[i];
|
|
dirProp=NO_CONTEXT_RTL(dirProps[i]);
|
|
if(level&UBIDI_LEVEL_OVERRIDE) {
|
|
/* keep the override flag in levels[i] but adjust the flags */
|
|
level&=~UBIDI_LEVEL_OVERRIDE; /* make the range check below simpler */
|
|
flags|=DIRPROP_FLAG_O(level);
|
|
} else {
|
|
/* set the flags */
|
|
flags|=DIRPROP_FLAG_E(level)|DIRPROP_FLAG(dirProp);
|
|
}
|
|
if((level<GET_PARALEVEL(pBiDi, i) &&
|
|
!((0==level)&&(dirProp==B))) ||
|
|
(UBIDI_MAX_EXPLICIT_LEVEL<level)) {
|
|
/* level out of bounds */
|
|
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
|
|
return UBIDI_LTR;
|
|
}
|
|
if((dirProp==B) && ((i+1)<length)) {
|
|
if(!((text[i]==CR) && (text[i+1]==LF))) {
|
|
pBiDi->paras[paraIndex++]=i+1;
|
|
}
|
|
}
|
|
}
|
|
if(flags&MASK_EMBEDDING) {
|
|
flags|=DIRPROP_FLAG_LR(pBiDi->paraLevel);
|
|
}
|
|
|
|
/* determine if the text is mixed-directional or single-directional */
|
|
pBiDi->flags=flags;
|
|
return directionFromFlags(flags);
|
|
}
|
|
|
|
/*********************************************************************/
|
|
/* The Properties state machine table */
|
|
/*********************************************************************/
|
|
/* */
|
|
/* All table cells are 8 bits: */
|
|
/* bits 0..4: next state */
|
|
/* bits 5..7: action to perform (if > 0) */
|
|
/* */
|
|
/* Cells may be of format "n" where n represents the next state */
|
|
/* (except for the rightmost column). */
|
|
/* Cells may also be of format "_(x,y)" where x represents an action */
|
|
/* to perform and y represents the next state. */
|
|
/* */
|
|
/*********************************************************************/
|
|
/* Definitions and type for properties state tables */
|
|
/*********************************************************************/
|
|
#define IMPTABPROPS_COLUMNS 14
|
|
#define IMPTABPROPS_RES (IMPTABPROPS_COLUMNS - 1)
|
|
#define GET_STATEPROPS(cell) ((cell)&0x1f)
|
|
#define GET_ACTIONPROPS(cell) ((cell)>>5)
|
|
#define _(action, newState) ((uint8_t)(newState+(action<<5)))
|
|
|
|
static const uint8_t groupProp[] = /* dirProp regrouped */
|
|
{
|
|
/* L R EN ES ET AN CS B S WS ON LRE LRO AL RLE RLO PDF NSM BN */
|
|
0, 1, 2, 7, 8, 3, 9, 6, 5, 4, 4, 10, 10, 12, 10, 10, 10, 11, 10
|
|
};
|
|
enum { _L=0, _R=1, _EN=2, _AN=3, _ON=4, _S=5, _B=6 }; /* reduced dirProp */
|
|
|
|
/*********************************************************************/
|
|
/* */
|
|
/* PROPERTIES STATE TABLE */
|
|
/* */
|
|
/* In table impTabProps, */
|
|
/* - the ON column regroups ON and WS */
|
|
/* - the BN column regroups BN, LRE, RLE, LRO, RLO, PDF */
|
|
/* - the Res column is the reduced property assigned to a run */
|
|
/* */
|
|
/* Action 1: process current run1, init new run1 */
|
|
/* 2: init new run2 */
|
|
/* 3: process run1, process run2, init new run1 */
|
|
/* 4: process run1, set run1=run2, init new run2 */
|
|
/* */
|
|
/* Notes: */
|
|
/* 1) This table is used in resolveImplicitLevels(). */
|
|
/* 2) This table triggers actions when there is a change in the Bidi*/
|
|
/* property of incoming characters (action 1). */
|
|
/* 3) Most such property sequences are processed immediately (in */
|
|
/* fact, passed to processPropertySeq(). */
|
|
/* 4) However, numbers are assembled as one sequence. This means */
|
|
/* that undefined situations (like CS following digits, until */
|
|
/* it is known if the next char will be a digit) are held until */
|
|
/* following chars define them. */
|
|
/* Example: digits followed by CS, then comes another CS or ON; */
|
|
/* the digits will be processed, then the CS assigned */
|
|
/* as the start of an ON sequence (action 3). */
|
|
/* 5) There are cases where more than one sequence must be */
|
|
/* processed, for instance digits followed by CS followed by L: */
|
|
/* the digits must be processed as one sequence, and the CS */
|
|
/* must be processed as an ON sequence, all this before starting */
|
|
/* assembling chars for the opening L sequence. */
|
|
/* */
|
|
/* */
|
|
static const uint8_t impTabProps[][IMPTABPROPS_COLUMNS] =
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , ES , ET , CS , BN , NSM , AL , Res */
|
|
/* 0 Init */ { 1 , 2 , 4 , 5 , 7 , 15 , 17 , 7 , 9 , 7 , 0 , 7 , 3 , _ON },
|
|
/* 1 L */ { 1 , _(1,2), _(1,4), _(1,5), _(1,7),_(1,15),_(1,17), _(1,7), _(1,9), _(1,7), 1 , 1 , _(1,3), _L },
|
|
/* 2 R */ { _(1,1), 2 , _(1,4), _(1,5), _(1,7),_(1,15),_(1,17), _(1,7), _(1,9), _(1,7), 2 , 2 , _(1,3), _R },
|
|
/* 3 AL */ { _(1,1), _(1,2), _(1,6), _(1,6), _(1,8),_(1,16),_(1,17), _(1,8), _(1,8), _(1,8), 3 , 3 , 3 , _R },
|
|
/* 4 EN */ { _(1,1), _(1,2), 4 , _(1,5), _(1,7),_(1,15),_(1,17),_(2,10), 11 ,_(2,10), 4 , 4 , _(1,3), _EN },
|
|
/* 5 AN */ { _(1,1), _(1,2), _(1,4), 5 , _(1,7),_(1,15),_(1,17), _(1,7), _(1,9),_(2,12), 5 , 5 , _(1,3), _AN },
|
|
/* 6 AL:EN/AN */ { _(1,1), _(1,2), 6 , 6 , _(1,8),_(1,16),_(1,17), _(1,8), _(1,8),_(2,13), 6 , 6 , _(1,3), _AN },
|
|
/* 7 ON */ { _(1,1), _(1,2), _(1,4), _(1,5), 7 ,_(1,15),_(1,17), 7 ,_(2,14), 7 , 7 , 7 , _(1,3), _ON },
|
|
/* 8 AL:ON */ { _(1,1), _(1,2), _(1,6), _(1,6), 8 ,_(1,16),_(1,17), 8 , 8 , 8 , 8 , 8 , _(1,3), _ON },
|
|
/* 9 ET */ { _(1,1), _(1,2), 4 , _(1,5), 7 ,_(1,15),_(1,17), 7 , 9 , 7 , 9 , 9 , _(1,3), _ON },
|
|
/*10 EN+ES/CS */ { _(3,1), _(3,2), 4 , _(3,5), _(4,7),_(3,15),_(3,17), _(4,7),_(4,14), _(4,7), 10 , _(4,7), _(3,3), _EN },
|
|
/*11 EN+ET */ { _(1,1), _(1,2), 4 , _(1,5), _(1,7),_(1,15),_(1,17), _(1,7), 11 , _(1,7), 11 , 11 , _(1,3), _EN },
|
|
/*12 AN+CS */ { _(3,1), _(3,2), _(3,4), 5 , _(4,7),_(3,15),_(3,17), _(4,7),_(4,14), _(4,7), 12 , _(4,7), _(3,3), _AN },
|
|
/*13 AL:EN/AN+CS */ { _(3,1), _(3,2), 6 , 6 , _(4,8),_(3,16),_(3,17), _(4,8), _(4,8), _(4,8), 13 , _(4,8), _(3,3), _AN },
|
|
/*14 ON+ET */ { _(1,1), _(1,2), _(4,4), _(1,5), 7 ,_(1,15),_(1,17), 7 , 14 , 7 , 14 , 14 , _(1,3), _ON },
|
|
/*15 S */ { _(1,1), _(1,2), _(1,4), _(1,5), _(1,7), 15 ,_(1,17), _(1,7), _(1,9), _(1,7), 15 , _(1,7), _(1,3), _S },
|
|
/*16 AL:S */ { _(1,1), _(1,2), _(1,6), _(1,6), _(1,8), 16 ,_(1,17), _(1,8), _(1,8), _(1,8), 16 , _(1,8), _(1,3), _S },
|
|
/*17 B */ { _(1,1), _(1,2), _(1,4), _(1,5), _(1,7),_(1,15), 17 , _(1,7), _(1,9), _(1,7), 17 , _(1,7), _(1,3), _B }
|
|
};
|
|
|
|
/* we must undef macro _ because the levels table have a different
|
|
* structure (4 bits for action and 4 bits for next state.
|
|
*/
|
|
#undef _
|
|
|
|
/*********************************************************************/
|
|
/* The levels state machine tables */
|
|
/*********************************************************************/
|
|
/* */
|
|
/* All table cells are 8 bits: */
|
|
/* bits 0..3: next state */
|
|
/* bits 4..7: action to perform (if > 0) */
|
|
/* */
|
|
/* Cells may be of format "n" where n represents the next state */
|
|
/* (except for the rightmost column). */
|
|
/* Cells may also be of format "_(x,y)" where x represents an action */
|
|
/* to perform and y represents the next state. */
|
|
/* */
|
|
/* This format limits each table to 16 states each and to 15 actions.*/
|
|
/* */
|
|
/*********************************************************************/
|
|
/* Definitions and type for levels state tables */
|
|
/*********************************************************************/
|
|
#define IMPTABLEVELS_COLUMNS (_B + 2)
|
|
#define IMPTABLEVELS_RES (IMPTABLEVELS_COLUMNS - 1)
|
|
#define GET_STATE(cell) ((cell)&0x0f)
|
|
#define GET_ACTION(cell) ((cell)>>4)
|
|
#define _(action, newState) ((uint8_t)(newState+(action<<4)))
|
|
|
|
typedef uint8_t ImpTab[][IMPTABLEVELS_COLUMNS];
|
|
typedef uint8_t ImpAct[];
|
|
|
|
typedef struct ImpTabPair {
|
|
ImpTab * pImpTab[2];
|
|
ImpAct * pImpAct[2];
|
|
} ImpTabPair;
|
|
|
|
/*********************************************************************/
|
|
/* */
|
|
/* LEVELS STATE TABLES */
|
|
/* */
|
|
/* In all levels state tables, */
|
|
/* - state 0 is the initial state */
|
|
/* - the Res column is the increment to add to the text level */
|
|
/* for this property sequence. */
|
|
/* */
|
|
/* The impAct arrays for each table of a pair map the local action */
|
|
/* numbers of the table to the total list of actions. For instance, */
|
|
/* action 2 in a given table corresponds to the action number which */
|
|
/* appears in entry [2] of the impAct array for that table. */
|
|
/* The first entry of all impAct arrays must be 0. */
|
|
/* */
|
|
/* Action 1: init conditional sequence */
|
|
/* 2: prepend conditional sequence to current sequence */
|
|
/* 3: set ON sequence to new level - 1 */
|
|
/* 4: init EN/AN/ON sequence */
|
|
/* 5: fix EN/AN/ON sequence followed by R */
|
|
/* 6: set previous level sequence to level 2 */
|
|
/* */
|
|
/* Notes: */
|
|
/* 1) These tables are used in processPropertySeq(). The input */
|
|
/* is property sequences as determined by resolveImplicitLevels. */
|
|
/* 2) Most such property sequences are processed immediately */
|
|
/* (levels are assigned). */
|
|
/* 3) However, some sequences cannot be assigned a final level till */
|
|
/* one or more following sequences are received. For instance, */
|
|
/* ON following an R sequence within an even-level paragraph. */
|
|
/* If the following sequence is R, the ON sequence will be */
|
|
/* assigned basic run level+1, and so will the R sequence. */
|
|
/* 4) S is generally handled like ON, since its level will be fixed */
|
|
/* to paragraph level in adjustWSLevels(). */
|
|
/* */
|
|
|
|
static const ImpTab impTabL_DEFAULT = /* Even paragraph level */
|
|
/* In this table, conditional sequences receive the higher possible level
|
|
until proven otherwise.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 0 , 1 , 0 , 2 , 0 , 0 , 0 , 0 },
|
|
/* 1 : R */ { 0 , 1 , 3 , 3 , _(1,4), _(1,4), 0 , 1 },
|
|
/* 2 : AN */ { 0 , 1 , 0 , 2 , _(1,5), _(1,5), 0 , 2 },
|
|
/* 3 : R+EN/AN */ { 0 , 1 , 3 , 3 , _(1,4), _(1,4), 0 , 2 },
|
|
/* 4 : R+ON */ { _(2,0), 1 , 3 , 3 , 4 , 4 , _(2,0), 1 },
|
|
/* 5 : AN+ON */ { _(2,0), 1 , _(2,0), 2 , 5 , 5 , _(2,0), 1 }
|
|
};
|
|
static const ImpTab impTabR_DEFAULT = /* Odd paragraph level */
|
|
/* In this table, conditional sequences receive the lower possible level
|
|
until proven otherwise.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 0 },
|
|
/* 1 : L */ { 1 , 0 , 1 , 3 , _(1,4), _(1,4), 0 , 1 },
|
|
/* 2 : EN/AN */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 1 },
|
|
/* 3 : L+AN */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 1 },
|
|
/* 4 : L+ON */ { _(2,1), 0 , _(2,1), 3 , 4 , 4 , 0 , 0 },
|
|
/* 5 : L+AN+ON */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 0 }
|
|
};
|
|
static const ImpAct impAct0 = {0,1,2,3,4,5,6};
|
|
static const ImpTabPair impTab_DEFAULT = {{(ImpTab*)&impTabL_DEFAULT,
|
|
(ImpTab*)&impTabR_DEFAULT},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
static const ImpTab impTabL_NUMBERS_SPECIAL = /* Even paragraph level */
|
|
/* In this table, conditional sequences receive the higher possible level
|
|
until proven otherwise.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 0 , 2 , 1 , 1 , 0 , 0 , 0 , 0 },
|
|
/* 1 : L+EN/AN */ { 0 , 2 , 1 , 1 , 0 , 0 , 0 , 2 },
|
|
/* 2 : R */ { 0 , 2 , 4 , 4 , _(1,3), 0 , 0 , 1 },
|
|
/* 3 : R+ON */ { _(2,0), 2 , 4 , 4 , 3 , 3 , _(2,0), 1 },
|
|
/* 4 : R+EN/AN */ { 0 , 2 , 4 , 4 , _(1,3), _(1,3), 0 , 2 }
|
|
};
|
|
static const ImpTabPair impTab_NUMBERS_SPECIAL = {{(ImpTab*)&impTabL_NUMBERS_SPECIAL,
|
|
(ImpTab*)&impTabR_DEFAULT},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
static const ImpTab impTabL_GROUP_NUMBERS_WITH_R =
|
|
/* In this table, EN/AN+ON sequences receive levels as if associated with R
|
|
until proven that there is L or sor/eor on both sides. AN is handled like EN.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 init */ { 0 , 3 , _(1,1), _(1,1), 0 , 0 , 0 , 0 },
|
|
/* 1 EN/AN */ { _(2,0), 3 , 1 , 1 , 2 , _(2,0), _(2,0), 2 },
|
|
/* 2 EN/AN+ON */ { _(2,0), 3 , 1 , 1 , 2 , _(2,0), _(2,0), 1 },
|
|
/* 3 R */ { 0 , 3 , 5 , 5 , _(1,4), 0 , 0 , 1 },
|
|
/* 4 R+ON */ { _(2,0), 3 , 5 , 5 , 4 , _(2,0), _(2,0), 1 },
|
|
/* 5 R+EN/AN */ { 0 , 3 , 5 , 5 , _(1,4), 0 , 0 , 2 }
|
|
};
|
|
static const ImpTab impTabR_GROUP_NUMBERS_WITH_R =
|
|
/* In this table, EN/AN+ON sequences receive levels as if associated with R
|
|
until proven that there is L on both sides. AN is handled like EN.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 init */ { 2 , 0 , 1 , 1 , 0 , 0 , 0 , 0 },
|
|
/* 1 EN/AN */ { 2 , 0 , 1 , 1 , 0 , 0 , 0 , 1 },
|
|
/* 2 L */ { 2 , 0 , _(1,4), _(1,4), _(1,3), 0 , 0 , 1 },
|
|
/* 3 L+ON */ { _(2,2), 0 , 4 , 4 , 3 , 0 , 0 , 0 },
|
|
/* 4 L+EN/AN */ { _(2,2), 0 , 4 , 4 , 3 , 0 , 0 , 1 }
|
|
};
|
|
static const ImpTabPair impTab_GROUP_NUMBERS_WITH_R = {
|
|
{(ImpTab*)&impTabL_GROUP_NUMBERS_WITH_R,
|
|
(ImpTab*)&impTabR_GROUP_NUMBERS_WITH_R},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
/* TBD !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! */
|
|
static const ImpTab impTabL_RUNS_ONLY = /* Even paragraph level */
|
|
/* In this table, conditional sequences receive the higher possible level
|
|
until proven otherwise.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 0 , 1 , 0 , 2 , 0 , 0 , 0 , 0 },
|
|
/* 1 : R */ { 0 , 1 , 3 , 3 , _(1,4), _(1,4), 0 , 1 },
|
|
/* 2 : AN */ { 0 , 1 , 0 , 2 , _(1,5), _(1,5), 0 , 2 },
|
|
/* 3 : R+EN/AN */ { 0 , 1 , 3 , 3 , _(1,4), _(1,4), 0 , 2 },
|
|
/* 4 : R+ON */ { _(2,0), 1 , 3 , 3 , 4 , 4 , _(2,0), 1 },
|
|
/* 5 : AN+ON */ { _(2,0), 1 , _(2,0), 2 , 5 , 5 , _(2,0), 1 }
|
|
};
|
|
static const ImpTab impTabR_RUNS_ONLY = /* Odd paragraph level */
|
|
/* In this table, conditional sequences receive the lower possible level
|
|
until proven otherwise.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 0 },
|
|
/* 1 : L */ { 1 , 0 , 1 , 3 , _(1,4), _(1,4), 0 , 1 },
|
|
/* 2 : EN/AN */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 1 },
|
|
/* 3 : L+AN */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 1 },
|
|
/* 4 : L+ON */ { _(2,1), 0 , _(2,1), 3 , 4 , 4 , 0 , 0 },
|
|
/* 5 : L+AN+ON */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 0 }
|
|
};
|
|
static const ImpTabPair impTab_RUNS_ONLY = {{(ImpTab*)&impTabL_RUNS_ONLY,
|
|
(ImpTab*)&impTabR_RUNS_ONLY},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
/* TBD !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! */
|
|
static const ImpTabPair impTab_RUNS_ONLY_WITH_MARKS = {
|
|
{(ImpTab*)&impTabL_RUNS_ONLY,
|
|
(ImpTab*)&impTabR_RUNS_ONLY},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
static const ImpTab impTabL_INVERSE_NUMBERS_AS_L =
|
|
/* This table is identical to the Default LTR table except that EN and AN are
|
|
handled like L.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 0 , 1 , 0 , 0 , 0 , 0 , 0 , 0 },
|
|
/* 1 : R */ { 0 , 1 , 0 , 0 , _(1,4), _(1,4), 0 , 1 },
|
|
/* 2 : AN */ { 0 , 1 , 0 , 0 , _(1,5), _(1,5), 0 , 2 },
|
|
/* 3 : R+EN/AN */ { 0 , 1 , 0 , 0 , _(1,4), _(1,4), 0 , 2 },
|
|
/* 4 : R+ON */ { _(2,0), 1 , _(2,0), _(2,0), 4 , 4 , _(2,0), 1 },
|
|
/* 5 : AN+ON */ { _(2,0), 1 , _(2,0), _(2,0), 5 , 5 , _(2,0), 1 }
|
|
};
|
|
static const ImpTab impTabR_INVERSE_NUMBERS_AS_L =
|
|
/* This table is identical to the Default RTL table except that EN and AN are
|
|
handled like L.
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 1 , 0 , 1 , 1 , 0 , 0 , 0 , 0 },
|
|
/* 1 : L */ { 1 , 0 , 1 , 1 , _(1,4), _(1,4), 0 , 1 },
|
|
/* 2 : EN/AN */ { 1 , 0 , 1 , 1 , 0 , 0 , 0 , 1 },
|
|
/* 3 : L+AN */ { 1 , 0 , 1 , 1 , 5 , 5 , 0 , 1 },
|
|
/* 4 : L+ON */ { _(2,1), 0 , _(2,1), _(2,1), 4 , 4 , 0 , 0 },
|
|
/* 5 : L+AN+ON */ { 1 , 0 , 1 , 1 , 5 , 5 , 0 , 0 }
|
|
};
|
|
static const ImpTabPair impTab_INVERSE_NUMBERS_AS_L = {
|
|
{(ImpTab*)&impTabL_INVERSE_NUMBERS_AS_L,
|
|
(ImpTab*)&impTabR_INVERSE_NUMBERS_AS_L},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
/* TBD !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! */
|
|
static const ImpTabPair impTab_INVERSE_LIKE_DIRECT = {
|
|
{(ImpTab*)&impTabL_DEFAULT,
|
|
(ImpTab*)&impTabR_DEFAULT},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
static const ImpTab impTabL_INVERSE_LIKE_DIRECT_WITH_MARKS =
|
|
/* The case handled in this table is (visually): R EN L
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 0 , _(6,3), 0 , 1 , 0 , 0 , 0 , 0 },
|
|
/* 1 : L+AN */ { 0 , _(6,3), 0 , 1 , _(1,2), _(3,0), 0 , 4 },
|
|
/* 2 : L+AN+ON */ { _(2,0), _(6,3), _(2,0), 1 , 2 , _(3,0), _(2,0), 3 },
|
|
/* 3 : R */ { 0 , _(6,3), _(5,5), _(5,6), _(1,4), _(3,0), 0 , 3 },
|
|
/* 4 : R+ON */ { _(3,0), _(4,3), _(5,5), _(5,6), 4 , _(3,0), _(3,0), 3 },
|
|
/* 5 : R+EN */ { _(3,0), _(4,3), 5 , _(5,6), _(1,4), _(3,0), _(3,0), 4 },
|
|
/* 6 : R+AN */ { _(3,0), _(4,3), _(5,5), 6 , _(1,4), _(3,0), _(3,0), 4 }
|
|
};
|
|
static const ImpTab impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS =
|
|
/* The cases handled in this table are (visually): R EN L
|
|
R L AN L
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { _(1,3), 0 , 1 , 1 , 0 , 0 , 0 , 0 },
|
|
/* 1 : R+EN/AN */ { _(2,3), 0 , 1 , 1 , 2 , _(4,0), 0 , 1 },
|
|
/* 2 : R+EN/AN+ON */ { _(2,3), 0 , 1 , 1 , 2 , _(4,0), 0 , 0 },
|
|
/* 3 : L */ { 3 , 0 , 3 , _(3,6), _(1,4), _(4,0), 0 , 1 },
|
|
/* 4 : L+ON */ { _(5,3), _(4,0), 5 , _(3,6), 4 , _(4,0), _(4,0), 0 },
|
|
/* 5 : L+ON+EN */ { _(5,3), _(4,0), 5 , _(3,6), 4 , _(4,0), _(4,0), 1 },
|
|
/* 6 : L+AN */ { _(5,3), _(4,0), 6 , 6 , 4 , _(4,0), _(4,0), 3 }
|
|
};
|
|
static const ImpAct impAct2 = {0,1,7,8,9,10};
|
|
static const ImpTabPair impTab_INVERSE_LIKE_DIRECT_WITH_MARKS = {
|
|
{(ImpTab*)&impTabL_INVERSE_LIKE_DIRECT_WITH_MARKS,
|
|
(ImpTab*)&impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct2}};
|
|
|
|
static const ImpTabPair impTab_INVERSE_FOR_NUMBERS_SPECIAL = {
|
|
{(ImpTab*)&impTabL_NUMBERS_SPECIAL,
|
|
(ImpTab*)&impTabR_DEFAULT},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct0}};
|
|
|
|
static const ImpTab impTabL_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS =
|
|
/* The case handled in this table is (visually): R EN L
|
|
*/
|
|
{
|
|
/* L , R , EN , AN , ON , S , B , Res */
|
|
/* 0 : init */ { 0 , _(6,2), 1 , 1 , 0 , 0 , 0 , 0 },
|
|
/* 1 : L+EN/AN */ { 0 , _(6,2), 1 , 1 , 0 , _(3,0), 0 , 4 },
|
|
/* 2 : R */ { 0 , _(6,2), _(5,4), _(5,4), _(1,3), _(3,0), 0 , 3 },
|
|
/* 3 : R+ON */ { _(3,0), _(4,2), _(5,4), _(5,4), 3 , _(3,0), _(3,0), 3 },
|
|
/* 4 : R+EN/AN */ { _(3,0), _(4,2), 4 , 4 , _(1,3), _(3,0), _(3,0), 4 }
|
|
};
|
|
static const ImpTabPair impTab_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS = {
|
|
{(ImpTab*)&impTabL_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS,
|
|
(ImpTab*)&impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS},
|
|
{(ImpAct*)&impAct0, (ImpAct*)&impAct2}};
|
|
|
|
#undef _
|
|
|
|
typedef struct {
|
|
ImpTab * pImpTab; /* level table pointer */
|
|
ImpAct * pImpAct; /* action map array */
|
|
int32_t startON; /* start of ON sequence */
|
|
int32_t startL2EN; /* start of level 2 sequence */
|
|
int32_t lastStrongRTL; /* index of last found R or AL */
|
|
int32_t state; /* current state */
|
|
UBiDiLevel runLevel; /* run level before implicit solving */
|
|
} LevState;
|
|
|
|
/*------------------------------------------------------------------------*/
|
|
|
|
static void addPoint(UBiDi *pBiDi, int32_t pos, char where, UChar insert)
|
|
/* param pos: position where to insert
|
|
param where: must be BEFORE or AFTER
|
|
param insert: char to insert (should be LRM or RLM, but can be any char)
|
|
*/
|
|
{
|
|
#define FIRSTALLOC 10
|
|
Point point;
|
|
InsertPoints * pInsertPoints=&(pBiDi->insertPoints);
|
|
|
|
if (pInsertPoints->capacity == 0)
|
|
{
|
|
pInsertPoints->points=uprv_malloc(sizeof(Point)*FIRSTALLOC);
|
|
if (pInsertPoints->points == NULL)
|
|
{
|
|
pInsertPoints->errorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
return;
|
|
}
|
|
pInsertPoints->capacity=FIRSTALLOC;
|
|
}
|
|
if (pInsertPoints->size >= pInsertPoints->capacity) /* no room for new point */
|
|
{
|
|
void * savePoints=pInsertPoints->points;
|
|
pInsertPoints->points=uprv_realloc(pInsertPoints->points,
|
|
pInsertPoints->capacity*2*sizeof(Point));
|
|
if (pInsertPoints->points == NULL)
|
|
{
|
|
pInsertPoints->points=savePoints;
|
|
pInsertPoints->errorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
return;
|
|
}
|
|
else pInsertPoints->capacity*=2;
|
|
}
|
|
point.pos=pos;
|
|
point.c=insert;
|
|
point.where=where;
|
|
pInsertPoints->points[pInsertPoints->size]=point;
|
|
pInsertPoints->size++;
|
|
#undef FIRSTALLOC
|
|
}
|
|
|
|
/* perform rules (Wn), (Nn), and (In) on a run of the text ------------------ */
|
|
|
|
/*
|
|
* This implementation of the (Wn) rules applies all rules in one pass.
|
|
* In order to do so, it needs a look-ahead of typically 1 character
|
|
* (except for W5: sequences of ET) and keeps track of changes
|
|
* in a rule Wp that affect a later Wq (p<q).
|
|
*
|
|
* The (Nn) and (In) rules are also performed in that same single loop,
|
|
* but effectively one iteration behind for white space.
|
|
*
|
|
* Since all implicit rules are performed in one step, it is not necessary
|
|
* to actually store the intermediate directional properties in dirProps[].
|
|
*/
|
|
|
|
static void
|
|
processPropertySeq(UBiDi *pBiDi, LevState *pLevState, uint8_t _prop,
|
|
int32_t start, int32_t limit) {
|
|
uint8_t cell, oldStateSeq, actionSeq;
|
|
ImpTab * pImpTab=pLevState->pImpTab;
|
|
ImpAct * pImpAct=pLevState->pImpAct;
|
|
UBiDiLevel * levels=pBiDi->levels;
|
|
UBiDiLevel level, addLevel;
|
|
InsertPoints * pInsertPoints;
|
|
int32_t start0, k;
|
|
|
|
start0=start; /* save original start position */
|
|
oldStateSeq=pLevState->state;
|
|
cell=(*pImpTab)[oldStateSeq][_prop];
|
|
pLevState->state=GET_STATE(cell); /* isolate the new state */
|
|
actionSeq=(*pImpAct)[GET_ACTION(cell)]; /* isolate the action */
|
|
addLevel=(*pImpTab)[pLevState->state][IMPTABLEVELS_RES];
|
|
|
|
if(actionSeq) {
|
|
switch(actionSeq) {
|
|
case 1: /* init ON seq */
|
|
pLevState->startON=start0;
|
|
break;
|
|
|
|
case 2: /* prepend ON seq to current seq */
|
|
start=pLevState->startON;
|
|
break;
|
|
|
|
case 3: /* L or S after possible relevant EN/AN */
|
|
/* check if we had EN after R/AL */
|
|
if (pLevState->startL2EN >= 0) {
|
|
addPoint(pBiDi, pLevState->startL2EN, BEFORE, uLRM);
|
|
}
|
|
pLevState->startL2EN=-1; /* not within previous if since could also be -2 */
|
|
/* check if we had any relevant EN/AN after R/AL */
|
|
pInsertPoints=&(pBiDi->insertPoints);
|
|
if ((pInsertPoints->capacity == 0) ||
|
|
(pInsertPoints->size <= pInsertPoints->confirmed))
|
|
{
|
|
/* nothing, just clean up */
|
|
pLevState->lastStrongRTL=-1;
|
|
/* check if we have a pending conditional segment */
|
|
level=(*pImpTab)[oldStateSeq][IMPTABLEVELS_RES];
|
|
if ((level & 1) && (pLevState->startON > 0)) { /* after ON */
|
|
start=pLevState->startON; /* reset to basic run level */
|
|
}
|
|
if (_prop == _S) /* add LRM before S */
|
|
{
|
|
addPoint(pBiDi, start0, BEFORE, uLRM);
|
|
pInsertPoints->confirmed=pInsertPoints->size;
|
|
}
|
|
break;
|
|
}
|
|
/* reset previous RTL cont to level for LTR text */
|
|
for (k=pLevState->lastStrongRTL+1; k<start0; k++)
|
|
{
|
|
/* reset odd level, leave runLevel+2 as is */
|
|
levels[k]=(levels[k] - 2) & ~1;
|
|
}
|
|
/* mark insert points as confirmed */
|
|
pInsertPoints->confirmed=pInsertPoints->size;
|
|
pLevState->lastStrongRTL=-1;
|
|
if (_prop == _S) /* add LRM before S */
|
|
{
|
|
addPoint(pBiDi, start0, BEFORE, uLRM);
|
|
pInsertPoints->confirmed=pInsertPoints->size;
|
|
}
|
|
break;
|
|
|
|
case 4: /* R/AL after possible relevant EN/AN */
|
|
/* just clean up */
|
|
pInsertPoints=&(pBiDi->insertPoints);
|
|
if (pInsertPoints->capacity > 0)
|
|
/* remove all non confirmed insert points */
|
|
pInsertPoints->size=pInsertPoints->confirmed;
|
|
pLevState->startON=-1;
|
|
pLevState->startL2EN=-1;
|
|
pLevState->lastStrongRTL=limit - 1;
|
|
break;
|
|
|
|
case 5: /* EN/AN after R/AL + possible cont */
|
|
/* check for real AN */
|
|
if ((_prop == _AN) && (NO_CONTEXT_RTL(pBiDi->dirProps[start0]) == AN) &&
|
|
(pBiDi->reorderingMode!=UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL))
|
|
{
|
|
/* real AN */
|
|
if (pLevState->startL2EN == -1) /* if no relevant EN already found */
|
|
{
|
|
/* just note the righmost digit as a strong RTL */
|
|
pLevState->lastStrongRTL=limit - 1;
|
|
break;
|
|
}
|
|
if (pLevState->startL2EN >= 0) /* after EN, no AN */
|
|
{
|
|
addPoint(pBiDi, pLevState->startL2EN, BEFORE, uLRM);
|
|
pLevState->startL2EN=-2;
|
|
}
|
|
/* note AN */
|
|
addPoint(pBiDi, start0, BEFORE, uLRM);
|
|
break;
|
|
}
|
|
/* if first EN/AN after R/AL */
|
|
if (pLevState->startL2EN == -1) {
|
|
pLevState->startL2EN=start0;
|
|
}
|
|
break;
|
|
|
|
case 6: /* note location of latest R/AL */
|
|
pLevState->lastStrongRTL=limit - 1;
|
|
pLevState->startON=-1;
|
|
break;
|
|
|
|
case 7: /* L after R+ON/EN/AN */
|
|
/* include possible adjacent number on the left */
|
|
for (k=start0-1; !(levels[k]&1); k--);
|
|
addPoint(pBiDi, k, BEFORE, uRLM); /* add RLM before */
|
|
pInsertPoints=&(pBiDi->insertPoints);
|
|
pInsertPoints->confirmed=pInsertPoints->size; /* confirm it */
|
|
pLevState->startON=start0;
|
|
break;
|
|
|
|
case 8: /* AN after L */
|
|
/* AN numbers between L text on both sides may be trouble. */
|
|
/* tentatively bracket with LRMs; will be confirmed if followed by L */
|
|
addPoint(pBiDi, start0, BEFORE, uLRM); /* add LRM before */
|
|
addPoint(pBiDi, start0, AFTER, uLRM); /* add LRM after */
|
|
break;
|
|
|
|
case 9: /* R after L+ON/EN/AN */
|
|
/* false alert, infirm LRMs around previous AN */
|
|
pInsertPoints=&(pBiDi->insertPoints);
|
|
pInsertPoints->size=pInsertPoints->confirmed;
|
|
if (_prop == _S) /* add RLM before S */
|
|
{
|
|
addPoint(pBiDi, start0, BEFORE, uRLM);
|
|
pInsertPoints->confirmed=pInsertPoints->size;
|
|
}
|
|
break;
|
|
|
|
case 10: /* L after L+ON/AN */
|
|
level=pLevState->runLevel + addLevel;
|
|
for(k=pLevState->startON; k<start0; k++) {
|
|
if (levels[k]<level)
|
|
levels[k]=level;
|
|
}
|
|
pInsertPoints=&(pBiDi->insertPoints);
|
|
pInsertPoints->confirmed=pInsertPoints->size; /* confirm inserts */
|
|
pLevState->startON=start0;
|
|
break;
|
|
|
|
default: /* we should never get here */
|
|
start=start0+25;
|
|
start/=(start-start0-25); /* force program crash */
|
|
break;
|
|
}
|
|
}
|
|
if((addLevel) || (start < start0)) {
|
|
level=pLevState->runLevel + addLevel;
|
|
for(k=start; k<limit; k++) {
|
|
levels[k]=level;
|
|
}
|
|
}
|
|
}
|
|
|
|
static void
|
|
resolveImplicitLevels(UBiDi *pBiDi,
|
|
int32_t start, int32_t limit,
|
|
DirProp sor, DirProp eor) {
|
|
const DirProp *dirProps=pBiDi->dirProps;
|
|
|
|
LevState levState;
|
|
int32_t i, start1, start2;
|
|
uint8_t oldStateImp, stateImp, actionImp;
|
|
uint8_t gprop, resProp, cell;
|
|
|
|
/* initialize for levels state table */
|
|
levState.startL2EN=-1; /* used for inverse3 */
|
|
levState.lastStrongRTL=-1; /* used for inverse3 */
|
|
levState.state=0;
|
|
levState.runLevel=pBiDi->levels[start];
|
|
levState.pImpTab=((pBiDi->pImpTabPair)->pImpTab)[levState.runLevel&1];
|
|
levState.pImpAct=((pBiDi->pImpTabPair)->pImpAct)[levState.runLevel&1];
|
|
processPropertySeq(pBiDi, &levState, sor, start, start);
|
|
/* initialize for property state table */
|
|
if(dirProps[start]==NSM) {
|
|
stateImp = 1 + sor;
|
|
} else {
|
|
stateImp=0;
|
|
}
|
|
start1=start;
|
|
|
|
for(i=start; i<=limit; i++) {
|
|
if(i<limit) {
|
|
gprop=groupProp[NO_CONTEXT_RTL(dirProps[i])];
|
|
} else {
|
|
gprop=eor;
|
|
}
|
|
oldStateImp=stateImp;
|
|
cell=impTabProps[oldStateImp][gprop];
|
|
stateImp=GET_STATEPROPS(cell); /* isolate the new state */
|
|
actionImp=GET_ACTIONPROPS(cell); /* isolate the action */
|
|
if((i==limit) && (actionImp==0)) {
|
|
/* there is an unprocessed sequence if its property == eor */
|
|
actionImp=1; /* process the last sequence */
|
|
}
|
|
if(actionImp) {
|
|
resProp=impTabProps[oldStateImp][IMPTABPROPS_RES];
|
|
switch(actionImp) {
|
|
case 1: /* process current seq1, init new seq1 */
|
|
processPropertySeq(pBiDi, &levState, resProp, start1, i);
|
|
start1=i;
|
|
break;
|
|
case 2: /* init new seq2 */
|
|
start2=i;
|
|
break;
|
|
case 3: /* process seq1, process seq2, init new seq1 */
|
|
processPropertySeq(pBiDi, &levState, resProp, start1, start2);
|
|
processPropertySeq(pBiDi, &levState, _ON, start2, i);
|
|
start1=i;
|
|
break;
|
|
case 4: /* process seq1, set seq1=seq2, init new seq2 */
|
|
processPropertySeq(pBiDi, &levState, resProp, start1, start2);
|
|
start1=start2;
|
|
start2=i;
|
|
break;
|
|
default: /* we should never get here */
|
|
start=start1+25;
|
|
start/=(start-start1-25); /* force program crash */
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
/* flush possible pending sequence, e.g. ON */
|
|
processPropertySeq(pBiDi, &levState, eor, limit, limit);
|
|
}
|
|
|
|
/* perform (L1) and (X9) ---------------------------------------------------- */
|
|
|
|
/*
|
|
* Reset the embedding levels for some non-graphic characters (L1).
|
|
* This function also sets appropriate levels for BN, and
|
|
* explicit embedding types that are supposed to have been removed
|
|
* from the paragraph in (X9).
|
|
*/
|
|
static void
|
|
adjustWSLevels(UBiDi *pBiDi) {
|
|
const DirProp *dirProps=pBiDi->dirProps;
|
|
UBiDiLevel *levels=pBiDi->levels;
|
|
int32_t i;
|
|
|
|
if(pBiDi->flags&MASK_WS) {
|
|
UBool orderParagraphsLTR=pBiDi->orderParagraphsLTR;
|
|
Flags flag;
|
|
|
|
i=pBiDi->trailingWSStart;
|
|
while(i>0) {
|
|
/* reset a sequence of WS/BN before eop and B/S to the paragraph paraLevel */
|
|
while(i>0 && (flag=DIRPROP_FLAG_NC(dirProps[--i]))&MASK_WS) {
|
|
if(orderParagraphsLTR&&(flag&DIRPROP_FLAG(B))) {
|
|
levels[i]=0;
|
|
} else {
|
|
levels[i]=GET_PARALEVEL(pBiDi, i);
|
|
}
|
|
}
|
|
|
|
/* reset BN to the next character's paraLevel until B/S, which restarts above loop */
|
|
/* here, i+1 is guaranteed to be <length */
|
|
while(i>0) {
|
|
flag=DIRPROP_FLAG_NC(dirProps[--i]);
|
|
if(flag&MASK_BN_EXPLICIT) {
|
|
levels[i]=levels[i+1];
|
|
} else if(orderParagraphsLTR&&(flag&DIRPROP_FLAG(B))) {
|
|
levels[i]=0;
|
|
break;
|
|
} else if(flag&MASK_B_S) {
|
|
levels[i]=GET_PARALEVEL(pBiDi, i);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
/* ubidi_setPara ------------------------------------------------------------ */
|
|
|
|
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);
|
|
}
|
|
|
|
/* initialize the UBiDi structure */
|
|
pBiDi->pParaBiDi=NULL; /* mark unfinished setPara */
|
|
pBiDi->text=text;
|
|
pBiDi->length=pBiDi->originalLength=pBiDi->resultLength=length;
|
|
pBiDi->paraLevel=paraLevel;
|
|
pBiDi->direction=UBIDI_LTR;
|
|
pBiDi->paraCount=1;
|
|
|
|
pBiDi->dirProps=NULL;
|
|
pBiDi->levels=NULL;
|
|
pBiDi->runs=NULL;
|
|
pBiDi->insertPoints.size=0; /* clean up from last call */
|
|
pBiDi->insertPoints.confirmed=0; /* clean up from last call */
|
|
|
|
/*
|
|
* Save the original paraLevel if contextual; otherwise, set to 0.
|
|
*/
|
|
if(IS_DEFAULT_LEVEL(paraLevel)) {
|
|
pBiDi->defaultParaLevel=paraLevel;
|
|
} else {
|
|
pBiDi->defaultParaLevel=0;
|
|
}
|
|
|
|
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;
|
|
pBiDi->defaultParaLevel=0;
|
|
}
|
|
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;
|
|
pBiDi->pParaBiDi=pBiDi; /* mark successful setPara */
|
|
return;
|
|
}
|
|
|
|
pBiDi->runCount=-1;
|
|
|
|
/*
|
|
* Get the directional properties,
|
|
* the flags bit-set, and
|
|
* determine the paragraph level if necessary.
|
|
*/
|
|
if(getDirPropsMemory(pBiDi, length)) {
|
|
pBiDi->dirProps=pBiDi->dirPropsMemory;
|
|
getDirProps(pBiDi);
|
|
} else {
|
|
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
return;
|
|
}
|
|
/* the processed length may have changed if UBIDI_OPTION_STREAMING */
|
|
length= pBiDi->length;
|
|
pBiDi->trailingWSStart=length; /* the levels[] will reflect the WS run */
|
|
/* allocate paras memory */
|
|
if(pBiDi->paraCount>1) {
|
|
if(getInitialParasMemory(pBiDi, pBiDi->paraCount)) {
|
|
pBiDi->paras=pBiDi->parasMemory;
|
|
pBiDi->paras[pBiDi->paraCount-1]=length;
|
|
} else {
|
|
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
return;
|
|
}
|
|
} else {
|
|
/* initialize paras for single paragraph */
|
|
pBiDi->paras=pBiDi->simpleParas;
|
|
pBiDi->simpleParas[0]=length;
|
|
}
|
|
|
|
/* are explicit levels specified? */
|
|
if(embeddingLevels==NULL) {
|
|
/* no: determine explicit levels according to the (Xn) rules */\
|
|
if(getLevelsMemory(pBiDi, length)) {
|
|
pBiDi->levels=pBiDi->levelsMemory;
|
|
direction=resolveExplicitLevels(pBiDi);
|
|
} else {
|
|
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
|
|
return;
|
|
}
|
|
} else {
|
|
/* set BN for all explicit codes, check that all levels are 0 or paraLevel..UBIDI_MAX_EXPLICIT_LEVEL */
|
|
pBiDi->levels=embeddingLevels;
|
|
direction=checkExplicitLevels(pBiDi, pErrorCode);
|
|
if(U_FAILURE(*pErrorCode)) {
|
|
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:
|
|
/*
|
|
* Choose the right implicit state table
|
|
*/
|
|
switch(pBiDi->reorderingMode) {
|
|
case UBIDI_REORDER_DEFAULT:
|
|
pBiDi->pImpTabPair=&impTab_DEFAULT;
|
|
break;
|
|
case UBIDI_REORDER_NUMBERS_SPECIAL:
|
|
pBiDi->pImpTabPair=&impTab_NUMBERS_SPECIAL;
|
|
break;
|
|
case UBIDI_REORDER_GROUP_NUMBERS_WITH_R:
|
|
pBiDi->pImpTabPair=&impTab_GROUP_NUMBERS_WITH_R;
|
|
break;
|
|
case UBIDI_REORDER_RUNS_ONLY:
|
|
if (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) {
|
|
pBiDi->pImpTabPair=&impTab_RUNS_ONLY_WITH_MARKS;
|
|
} else {
|
|
pBiDi->pImpTabPair=&impTab_RUNS_ONLY;
|
|
}
|
|
break;
|
|
case UBIDI_REORDER_INVERSE_NUMBERS_AS_L:
|
|
pBiDi->pImpTabPair=&impTab_INVERSE_NUMBERS_AS_L;
|
|
break;
|
|
case UBIDI_REORDER_INVERSE_LIKE_DIRECT:
|
|
if (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) {
|
|
pBiDi->pImpTabPair=&impTab_INVERSE_LIKE_DIRECT_WITH_MARKS;
|
|
} else {
|
|
pBiDi->pImpTabPair=&impTab_INVERSE_LIKE_DIRECT;
|
|
}
|
|
break;
|
|
case UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL:
|
|
if (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) {
|
|
pBiDi->pImpTabPair=&impTab_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS;
|
|
} else {
|
|
pBiDi->pImpTabPair=&impTab_INVERSE_FOR_NUMBERS_SPECIAL;
|
|
}
|
|
break;
|
|
default:
|
|
pBiDi->pImpTabPair=&impTab_DEFAULT;
|
|
break;
|
|
}
|
|
/*
|
|
* 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(GET_PARALEVEL(pBiDi, 0)),
|
|
GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, length-1)));
|
|
} 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=GET_PARALEVEL(pBiDi, 0);
|
|
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 */
|
|
start=limit;
|
|
level=nextLevel;
|
|
if((start>0) && (NO_CONTEXT_RTL(pBiDi->dirProps[start-1])==B)) {
|
|
/* except if this is a new paragraph, then set sor = para level */
|
|
sor=GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, start));
|
|
} else {
|
|
sor=eor;
|
|
}
|
|
|
|
/* 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=GET_PARALEVEL(pBiDi, length-1);
|
|
}
|
|
|
|
/* 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);
|
|
}
|
|
/* check if we got any memory shortage while adding insert points */
|
|
if (U_FAILURE(pBiDi->insertPoints.errorCode))
|
|
{
|
|
*pErrorCode=pBiDi->insertPoints.errorCode;
|
|
return;
|
|
}
|
|
|
|
/* 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;
|
|
}
|
|
if(pBiDi->reorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) {
|
|
pBiDi->resultLength -= pBiDi->countBiDiControls;
|
|
} else {
|
|
pBiDi->resultLength += pBiDi->insertPoints.size;
|
|
}
|
|
pBiDi->pParaBiDi=pBiDi; /* mark successful setPara */
|
|
}
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_orderParagraphsLTR(UBiDi *pBiDi, UBool orderParagraphsLTR) {
|
|
if(pBiDi!=NULL) {
|
|
pBiDi->orderParagraphsLTR=orderParagraphsLTR;
|
|
}
|
|
}
|
|
|
|
U_CAPI UBool U_EXPORT2
|
|
ubidi_isOrderParagraphsLTR(UBiDi *pBiDi) {
|
|
if(pBiDi!=NULL) {
|
|
return pBiDi->orderParagraphsLTR;
|
|
} else {
|
|
return FALSE;
|
|
}
|
|
}
|
|
|
|
U_CAPI UBiDiDirection U_EXPORT2
|
|
ubidi_getDirection(const UBiDi *pBiDi) {
|
|
if(IS_VALID_PARA_OR_LINE(pBiDi)) {
|
|
return pBiDi->direction;
|
|
} else {
|
|
return UBIDI_LTR;
|
|
}
|
|
}
|
|
|
|
U_CAPI const UChar * U_EXPORT2
|
|
ubidi_getText(const UBiDi *pBiDi) {
|
|
if(IS_VALID_PARA_OR_LINE(pBiDi)) {
|
|
return pBiDi->text;
|
|
} else {
|
|
return NULL;
|
|
}
|
|
}
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
ubidi_getLength(const UBiDi *pBiDi) {
|
|
if(IS_VALID_PARA_OR_LINE(pBiDi)) {
|
|
return pBiDi->originalLength;
|
|
} else {
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
ubidi_getProcessedLength(const UBiDi *pBiDi) {
|
|
if(IS_VALID_PARA_OR_LINE(pBiDi)) {
|
|
return pBiDi->length;
|
|
} else {
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
ubidi_getResultLength(const UBiDi *pBiDi) {
|
|
if(IS_VALID_PARA_OR_LINE(pBiDi)) {
|
|
return pBiDi->resultLength;
|
|
} else {
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
/* paragraphs API functions ------------------------------------------------- */
|
|
|
|
U_CAPI UBiDiLevel U_EXPORT2
|
|
ubidi_getParaLevel(const UBiDi *pBiDi) {
|
|
if(IS_VALID_PARA_OR_LINE(pBiDi)) {
|
|
return pBiDi->paraLevel;
|
|
} else {
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
ubidi_countParagraphs(UBiDi *pBiDi) {
|
|
if(!IS_VALID_PARA_OR_LINE(pBiDi)) {
|
|
return 0;
|
|
} else {
|
|
return pBiDi->paraCount;
|
|
}
|
|
}
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_getParagraphByIndex(const UBiDi *pBiDi, int32_t paraIndex,
|
|
int32_t *pParaStart, int32_t *pParaLimit,
|
|
UBiDiLevel *pParaLevel, UErrorCode *pErrorCode) {
|
|
int32_t paraStart;
|
|
|
|
/* check the argument values */
|
|
if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
|
|
return;
|
|
} else if( !IS_VALID_PARA_OR_LINE(pBiDi) || /* no valid setPara/setLine */
|
|
paraIndex<0 || paraIndex>=pBiDi->paraCount ) {
|
|
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
|
|
return;
|
|
}
|
|
pBiDi=pBiDi->pParaBiDi; /* get Para object if Line object */
|
|
if(paraIndex) {
|
|
paraStart=pBiDi->paras[paraIndex-1];
|
|
} else {
|
|
paraStart=0;
|
|
}
|
|
if(pParaStart!=NULL) {
|
|
*pParaStart=paraStart;
|
|
}
|
|
if(pParaLimit!=NULL) {
|
|
*pParaLimit=pBiDi->paras[paraIndex];
|
|
}
|
|
if(pParaLevel!=NULL) {
|
|
*pParaLevel=GET_PARALEVEL(pBiDi, paraStart);
|
|
}
|
|
return;
|
|
}
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
ubidi_getParagraph(const UBiDi *pBiDi, int32_t charIndex,
|
|
int32_t *pParaStart, int32_t *pParaLimit,
|
|
UBiDiLevel *pParaLevel, UErrorCode *pErrorCode) {
|
|
uint32_t paraIndex;
|
|
|
|
/* check the argument values */
|
|
/* pErrorCode will be checked by the call to ubidi_getParagraphByIndex */
|
|
if( !IS_VALID_PARA_OR_LINE(pBiDi)) {/* no valid setPara/setLine */
|
|
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
|
|
return -1;
|
|
}
|
|
pBiDi=pBiDi->pParaBiDi; /* get Para object if Line object */
|
|
if( charIndex<0 || charIndex>=pBiDi->length ) {
|
|
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
|
|
return -1;
|
|
}
|
|
for(paraIndex=0; charIndex>=pBiDi->paras[paraIndex]; paraIndex++);
|
|
ubidi_getParagraphByIndex(pBiDi, paraIndex, pParaStart, pParaLimit, pParaLevel, pErrorCode);
|
|
return paraIndex;
|
|
}
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_setClassCallback(UBiDi *pBiDi, UBiDiClassCallback *newFn,
|
|
const void *newContext, UBiDiClassCallback **oldFn,
|
|
const void **oldContext, UErrorCode *pErrorCode)
|
|
{
|
|
if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
|
|
return;
|
|
} else if(pBiDi==NULL) {
|
|
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
|
|
return;
|
|
}
|
|
if( oldFn )
|
|
{
|
|
*oldFn = pBiDi->fnClassCallback;
|
|
}
|
|
if( oldContext )
|
|
{
|
|
*oldContext = pBiDi->coClassCallback;
|
|
}
|
|
pBiDi->fnClassCallback = newFn;
|
|
pBiDi->coClassCallback = newContext;
|
|
}
|
|
|
|
U_CAPI void U_EXPORT2
|
|
ubidi_getClassCallback(UBiDi *pBiDi, UBiDiClassCallback **fn, const void **context)
|
|
{
|
|
if( fn )
|
|
{
|
|
*fn = pBiDi->fnClassCallback;
|
|
}
|
|
if( context )
|
|
{
|
|
*context = pBiDi->coClassCallback;
|
|
}
|
|
}
|
|
|
|
U_CAPI UCharDirection U_EXPORT2
|
|
ubidi_getCustomizedClass(UBiDi *pBiDi, UChar32 c)
|
|
{
|
|
UCharDirection dir;
|
|
|
|
if( pBiDi->fnClassCallback == NULL ||
|
|
(dir = (*pBiDi->fnClassCallback)(pBiDi->coClassCallback, c)) == U_BIDI_CLASS_DEFAULT )
|
|
{
|
|
return ubidi_getClass(pBiDi->bdp, c);
|
|
} else {
|
|
return dir;
|
|
}
|
|
}
|
|
|