* *
limitof a sequence of characters is the position just after their * last character, i.e., one more than that position.
*
* Some of the API functions provide access to runs
.
* Such a run
is defined as a sequence of characters
* that are at the same embedding level
* after performing the BIDI algorithm.
* * @author Markus W. Scherer. Ported to Mozilla by Simon Montagu * @version 1.0 */ /** * nsBidiLevel is the type of the level values in this * Bidi implementation. * It holds an embedding level and indicates the visual direction * by its bit 0 (even/odd value).
*
* It can also hold non-level values for the
* aParaLevel and aEmbeddingLevels
* arguments of SetPara; there:
*
aEmbeddingLevels[]
* value indicates whether the using application is
* specifying the level of a character to override whatever the
* Bidi implementation would resolve it to.aParaLevel can be set to the
* pseudo-level values NSBIDI_DEFAULT_LTR
* and NSBIDI_DEFAULT_RTL.The related constants are not real, valid level values.
* NSBIDI_DEFAULT_XXX can be used to specify
* a default for the paragraph level for
* when the SetPara function
* shall determine it but there is no
* strongly typed character in the input.
*
* Note that the value for
* Reordering can be done on a line, or on a paragraph which is
* then interpreted as one single line.
*
* On construction, the class is initially empty. It is assigned
* the Bidi properties of a paragraph by
* A Bidi class can be reused for as long as it is not deallocated
* by calling its destructor.
*
* This object can be reused for as long as it is not destroyed.
*
*
* This function takes a single plain text paragraph with or without
* externally specified embedding levels from
*
* If the entire paragraph consists of text of only one direction, then
* the function may not perform all the steps described by the algorithm,
* i.e., some levels may not be the same as if all steps were performed.
* This is not relevant for unidirectional text.
*
* The text must be externally split into separate paragraphs (rule P1).
* Paragraph separators (B) should appear at most at the very end.
*
* @param aText is a pointer to the single-paragraph text that the
* Bidi algorithm will be performed on
* (step (P1) of the algorithm is performed externally).
* The text must be (at least)
* Except for that bit, it must be
*
* Caution: A copy of this pointer, not of the levels,
* will be stored in the
* After the
* The
* In the new line object, the indexes will range from 0 to
*
* This is used after calling
*
* After line-breaking, rules (L1) and (L2) for the treatment of
* trailing WS and for reordering are performed on
* an
*
* Important: the line
*
* Note that this function may allocate memory under some
* circumstances, unlike
* This is especially useful for line-breaking on a paragraph.
*
* @param aLogicalStart is the first character of the run.
*
* @param aLogicalLimit will receive the limit of the run.
* The l-value that you point to here may be the
* same expression (variable) as the one for
*
*
*
* Example:
* @code
* PRInt32 i, count, logicalStart, visualIndex=0, length;
* nsBidiDirection dir;
* pBidi->CountRuns(&count);
* for(i=0; i
*
* Note that in right-to-left runs, this mapping places
* modifier letters before base characters and second surrogates
* before first ones.
*
* @param aLogicalIndex is the index of a character in the text.
*
* @param aVisualIndex will receive the visual position of this character.
*
* @see GetLogicalMap
* @see GetLogicalIndex
*/
nsresult GetVisualIndex(PRInt32 aLogicalIndex, PRInt32* aVisualIndex);
/**
* Get the logical text position from a visual position.
* If such a mapping is used many times on the same
*
*
* This is the inverse function to
* The index map will result in
*
* @see GetVisualMap
* @see GetVisualIndex
*/
nsresult GetLogicalMap(PRInt32 *aIndexMap);
/**
* Get a visual-to-logical index map (array) for the characters in the nsBidi
* (paragraph or line) object.
*
* @param aIndexMap is a pointer to an array of
* The index map will result in
*
* @see GetLogicalMap
* @see GetLogicalIndex
*/
nsresult GetVisualMap(PRInt32 *aIndexMap);
/**
* This is a convenience function that does not use a nsBidi object.
* It is intended to be used for when an application has determined the levels
* of objects (character sequences) and just needs to have them reordered (L2).
* This is equivalent to using
* The index map will result in
* The index map will result in NSBIDI_DEFAULT_LTR is even
* and the one for NSBIDI_DEFAULT_RTL is odd,
* just like with normal LTR and RTL level values -
* these special values are designed that way. Also, the implementation
* assumes that NSBIDI_MAX_EXPLICIT_LEVEL is odd.
*
* @see NSBIDI_DEFAULT_LTR
* @see NSBIDI_DEFAULT_RTL
* @see NSBIDI_LEVEL_OVERRIDE
* @see NSBIDI_MAX_EXPLICIT_LEVEL
*/
typedef PRUint8 nsBidiLevel;
/** Paragraph level setting.
* If there is no strong character, then set the paragraph level to 0 (left-to-right).
*/
#define NSBIDI_DEFAULT_LTR 0xfe
/** Paragraph level setting.
* If there is no strong character, then set the paragraph level to 1 (right-to-left).
*/
#define NSBIDI_DEFAULT_RTL 0xff
/**
* Maximum explicit embedding level.
* (The maximum resolved level can be up to NSBIDI_MAX_EXPLICIT_LEVEL+1).
*
*/
#define NSBIDI_MAX_EXPLICIT_LEVEL 61
/** Bit flag for level input.
* Overrides directional properties.
*/
#define NSBIDI_LEVEL_OVERRIDE 0x80
/**
* nsBidiDirection values indicate the text direction.
*/
enum nsBidiDirection {
/** All left-to-right text This is a 0 value. */
NSBIDI_LTR,
/** All right-to-left text This is a 1 value. */
NSBIDI_RTL,
/** Mixed-directional text. */
NSBIDI_MIXED
};
typedef enum nsBidiDirection nsBidiDirection;
/* miscellaneous definitions ------------------------------------------------ */
/** option flags for WriteReverse() */
/**
* option bit for WriteReverse():
* keep combining characters after their base characters in RTL runs
*
* @see WriteReverse
*/
#define NSBIDI_KEEP_BASE_COMBINING 1
/**
* option bit for WriteReverse():
* replace characters with the "mirrored" property in RTL runs
* by their mirror-image mappings
*
* @see WriteReverse
*/
#define NSBIDI_DO_MIRRORING 2
/**
* option bit for WriteReverse():
* remove Bidi control characters
*
* @see WriteReverse
*/
#define NSBIDI_REMOVE_BIDI_CONTROLS 8
/* helper macros for each allocated array member */
#define GETDIRPROPSMEMORY(length) \
GetMemory((void **)&mDirPropsMemory, &mDirPropsSize, \
mMayAllocateText, (length))
#define GETLEVELSMEMORY(length) \
GetMemory((void **)&mLevelsMemory, &mLevelsSize, \
mMayAllocateText, (length))
#define GETRUNSMEMORY(length) \
GetMemory((void **)&mRunsMemory, &mRunsSize, \
mMayAllocateRuns, (length)*sizeof(Run))
/* additional macros used by constructor - always allow allocation */
#define GETINITIALDIRPROPSMEMORY(length) \
GetMemory((void **)&mDirPropsMemory, &mDirPropsSize, \
PR_TRUE, (length))
#define GETINITIALLEVELSMEMORY(length) \
GetMemory((void **)&mLevelsMemory, &mLevelsSize, \
PR_TRUE, (length))
#define GETINITIALRUNSMEMORY(length) \
GetMemory((void **)&mRunsMemory, &mRunsSize, \
PR_TRUE, (length)*sizeof(Run))
/*
* Sometimes, bit values are more appropriate
* to deal with directionality properties.
* Abbreviations in these macro names refer to names
* used in the Bidi algorithm.
*/
typedef PRUint8 DirProp;
#define DIRPROP_FLAG(dir) (1UL<<(dir))
/* special flag for multiple runs from explicit embedding codes */
#define DIRPROP_FLAG_MULTI_RUNS (1UL<<31)
/* are there any characters that are LTR or RTL? */
#define MASK_LTR (DIRPROP_FLAG(L)|DIRPROP_FLAG(EN)|DIRPROP_FLAG(AN)|DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO))
#define MASK_RTL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL)|DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO))
/* explicit embedding codes */
#define MASK_LRX (DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO))
#define MASK_RLX (DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO))
#define MASK_OVERRIDE (DIRPROP_FLAG(LRO)|DIRPROP_FLAG(RLO))
#define MASK_EXPLICIT (MASK_LRX|MASK_RLX|DIRPROP_FLAG(PDF))
#define MASK_BN_EXPLICIT (DIRPROP_FLAG(BN)|MASK_EXPLICIT)
/* paragraph and segment separators */
#define MASK_B_S (DIRPROP_FLAG(B)|DIRPROP_FLAG(S))
/* all types that are counted as White Space or Neutral in some steps */
#define MASK_WS (MASK_B_S|DIRPROP_FLAG(WS)|MASK_BN_EXPLICIT)
#define MASK_N (DIRPROP_FLAG(O_N)|MASK_WS)
/* all types that are included in a sequence of European Terminators for (W5) */
#define MASK_ET_NSM_BN (DIRPROP_FLAG(ET)|DIRPROP_FLAG(NSM)|MASK_BN_EXPLICIT)
/* types that are neutrals or could becomes neutrals in (Wn) */
#define MASK_POSSIBLE_N (DIRPROP_FLAG(CS)|DIRPROP_FLAG(ES)|DIRPROP_FLAG(ET)|MASK_N)
/*
* These types may be changed to "e",
* the embedding type (L or R) of the run,
* in the Bidi algorithm (N2)
*/
#define MASK_EMBEDDING (DIRPROP_FLAG(NSM)|MASK_POSSIBLE_N)
/* the dirProp's L and R are defined to 0 and 1 values in nsCharType */
#define GET_LR_FROM_LEVEL(level) ((DirProp)((level)&1))
#define IS_DEFAULT_LEVEL(level) (((level)&0xfe)==0xfe)
/* handle surrogate pairs --------------------------------------------------- */
#define IS_FIRST_SURROGATE(uchar) (((uchar)&0xfc00)==0xd800)
#define IS_SECOND_SURROGATE(uchar) (((uchar)&0xfc00)==0xdc00)
/* get the UTF-32 value directly from the surrogate pseudo-characters */
#define SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000)
#define GET_UTF_32(first, second) (((first)<<10UL)+(second)-SURROGATE_OFFSET)
#define UTF_ERROR_VALUE 0xffff
/* definitions with forward iteration --------------------------------------- */
/*
* all the macros that go forward assume that
* the initial offset is 0<=iSetPara
* or the Bidi properties of a line of a paragraph by
* SetLine.SetPara will allocate additional memory for
* internal structures as necessary.
*/
class nsBidi
{
public:
/** @brief Default constructor.
*
* The nsBidi object is initially empty. It is assigned
* the Bidi properties of a paragraph by SetPara()
* or the Bidi properties of a line of a paragraph by
* GetLine().SetPara() will allocate additional memory for
* internal structures as necessary.
*
*/
nsBidi();
/** @brief Destructor. */
virtual ~nsBidi();
/**
* Perform the Unicode Bidi algorithm. It is defined in the
* Unicode Technical Report 9,
* version 5,
* also described in The Unicode Standard, Version 3.0 .styled
text
* and computes the left-right-directionality of each character.
* For example, in pure LTR text with numbers the numbers would get
* a resolved level of 2 higher than the surrounding text according to
* the algorithm. This implementation may set all resolved levels to
* the same value in such a case.aLength long.
*
* @param aLength is the length of the text; if aLength==-1 then
* the text must be zero-terminated.
*
* @param aParaLevel specifies the default level for the paragraph;
* it is typically 0 (LTR) or 1 (RTL).
* If the function shall determine the paragraph level from the text,
* then aParaLevel can be set to
* either NSBIDI_DEFAULT_LTR
* or NSBIDI_DEFAULT_RTL;
* if there is no strongly typed character, then
* the desired default is used (0 for LTR or 1 for RTL).
* Any other value between 0 and NSBIDI_MAX_EXPLICIT_LEVEL is also valid,
* with odd levels indicating RTL.
*
* @param aEmbeddingLevels (in) may be used to preset the embedding and override levels,
* ignoring characters like LRE and PDF in the text.
* A level overrides the directional property of its corresponding
* (same index) character if the level has the
* NSBIDI_LEVEL_OVERRIDE bit set.aParaLevel<=aEmbeddingLevels[]<=NSBIDI_MAX_EXPLICIT_LEVEL.nsBidi object;
* the aEmbeddingLevels array must not be
* deallocated before the nsBidi object is destroyed or reused,
* and the aEmbeddingLevels
* should not be modified to avoid unexpected results on subsequent Bidi operations.
* However, the SetPara and
* SetLine functions may modify some or all of the levels.nsBidi object is reused or destroyed, the caller
* must take care of the deallocation of the aEmbeddingLevels array.aEmbeddingLevels array must be
* at least aLength long.
*/
nsresult SetPara(const PRUnichar *aText, PRInt32 aLength, nsBidiLevel aParaLevel, nsBidiLevel *aEmbeddingLevels);
#ifdef FULL_BIDI_ENGINE
/**
* SetLine sets an nsBidi to
* contain the reordering information, especially the resolved levels,
* for all the characters in a line of text. This line of text is
* specified by referring to an nsBidi object representing
* this information for a paragraph of text, and by specifying
* a range of indexes in this paragraph.aLimit-aStart.SetPara
* for a paragraph, and after line-breaking on that paragraph.
* It is not necessary if the paragraph is treated as a single line.nsBidi object that represents a line.nsBidi object shares data with
* aParaBidi.
* You must destroy or reuse this object before aParaBidi.
* In other words, you must destroy or reuse the nsBidi object for a line
* before the object for its parent paragraph.
*
* @param aParaBidi is the parent paragraph object.
*
* @param aStart is the line's first index into the paragraph text.
*
* @param aLimit is just behind the line's last index into the paragraph text
* (its last index +1).
* It must be 0<=aStart<=aLimit<=paragraph length.
*
* @see SetPara
*/
nsresult SetLine(nsIBidi* aParaBidi, PRInt32 aStart, PRInt32 aLimit);
/**
* Get the directionality of the text.
*
* @param aDirection receives a NSBIDI_XXX value that indicates if the entire text
* represented by this object is unidirectional,
* and which direction, or if it is mixed-directional.
*
* @see nsBidiDirection
*/
nsresult GetDirection(nsBidiDirection* aDirection);
/**
* Get the length of the text.
*
* @param aLength receives the length of the text that the nsBidi object was created for.
*/
nsresult GetLength(PRInt32* aLength);
/**
* Get the paragraph level of the text.
*
* @param aParaLevel receives a NSBIDI_XXX value indicating the paragraph level
*
* @see nsBidiLevel
*/
nsresult GetParaLevel(nsBidiLevel* aParaLevel);
/**
* Get the level for one character.
*
* @param aCharIndex the index of a character.
*
* @param aLevel receives the level for the character at aCharIndex.
*
* @see nsBidiLevel
*/
nsresult GetLevelAt(PRInt32 aCharIndex, nsBidiLevel* aLevel);
/**
* Get an array of levels for each character.GetLevelAt.
*
* @param aLevels receives a pointer to the levels array for the text,
* or NULL if an error occurs.
*
* @see nsBidiLevel
*/
nsresult GetLevels(nsBidiLevel** aLevels);
#endif // FULL_BIDI_ENGINE
/**
* Get the bidirectional type for one character.
*
* @param aCharIndex the index of a character.
*
* @param aType receives the bidirectional type of the character at aCharIndex.
*/
nsresult GetCharTypeAt(PRInt32 aCharIndex, nsCharType* aType);
/**
* Get a logical run.
* This function returns information about a run and is used
* to retrieve runs in logical order.aLogicalStart.
* This pointer can be NULL if this
* value is not necessary.
*
* @param aLevel will receive the level of the run.
* This pointer can be NULL if this
* value is not necessary.
*/
nsresult GetLogicalRun(PRInt32 aLogicalStart, PRInt32* aLogicalLimit, nsBidiLevel* aLevel);
/**
* Get the number of runs.
* This function may invoke the actual reordering on the
* nsBidi object, after SetPara
* may have resolved only the levels of the text. Therefore,
* CountRuns may have to allocate memory,
* and may fail doing so.
*
* @param aRunCount will receive the number of runs.
*/
nsresult CountRuns(PRInt32* aRunCount);
/**
* Get one run's logical start, length, and directionality,
* which can be 0 for LTR or 1 for RTL.
* In an RTL run, the character at the logical start is
* visually on the right of the displayed run.
* The length is the number of characters in the run.CountRuns should be called
* before the runs are retrieved.
*
* @param aRunIndex is the number of the run in visual order, in the
* range [0..CountRuns-1].
*
* @param aLogicalStart is the first logical character index in the text.
* The pointer may be NULL if this index is not needed.
*
* @param aLength is the number of characters (at least one) in the run.
* The pointer may be NULL if this is not needed.
*
* @param aDirection will receive the directionality of the run,
* NSBIDI_LTR==0 or NSBIDI_RTL==1,
* never NSBIDI_MIXED.
*
* @see CountRunsnsBidi object, then calling
* GetLogicalMap is more efficient.nsBidi object, then calling
* GetVisualMap is more efficient.GetVisualIndex.
*
* @param aVisualIndex is the visual position of a character.
*
* @param aLogicalIndex will receive the index of this character in the text.
*
* @see GetVisualMap
* @see GetVisualIndex
*/
nsresult GetLogicalIndex(PRInt32 aVisualIndex, PRInt32* aLogicalIndex);
/**
* Get a logical-to-visual index map (array) for the characters in the nsBidi
* (paragraph or line) object.
*
* @param aIndexMap is a pointer to an array of GetLength
* indexes which will reflect the reordering of the characters.
* The array does not need to be initialized.aIndexMap[aLogicalIndex]==aVisualIndex.GetLength
* indexes which will reflect the reordering of the characters.
* The array does not need to be initialized.aIndexMap[aVisualIndex]==aLogicalIndex.GetLogicalMap on a
* nsBidi object.
*
* @param aLevels is an array with aLength levels that have been determined by
* the application.
*
* @param aLength is the number of levels in the array, or, semantically,
* the number of objects to be reordered.
* It must be aLength>0.
*
* @param aIndexMap is a pointer to an array of aLength
* indexes which will reflect the reordering of the characters.
* The array does not need to be initialized.aIndexMap[aLogicalIndex]==aVisualIndex.
*/
nsresult ReorderLogical(const nsBidiLevel *aLevels, PRInt32 aLength, PRInt32 *aIndexMap);
#endif // FULL_BIDI_ENGINE
/**
* This is a convenience function that does not use a nsBidi object.
* It is intended to be used for when an application has determined the levels
* of objects (character sequences) and just needs to have them reordered (L2).
* This is equivalent to using GetVisualMap on a
* nsBidi object.
*
* @param aLevels is an array with aLength levels that have been determined by
* the application.
*
* @param aLength is the number of levels in the array, or, semantically,
* the number of objects to be reordered.
* It must be aLength>0.
*
* @param aIndexMap is a pointer to an array of aLength
* indexes which will reflect the reordering of the characters.
* The array does not need to be initialized.aIndexMap[aVisualIndex]==aLogicalIndex.
*/
nsresult ReorderVisual(const nsBidiLevel *aLevels, PRInt32 aLength, PRInt32 *aIndexMap);
#ifdef FULL_BIDI_ENGINE
/**
* Invert an index map.
* The one-to-one index mapping of the first map is inverted and written to
* the second one.
*
* @param aSrcMap is an array with aLength indexes
* which define the original mapping.
*
* @param aDestMap is an array with aLength indexes
* which will be filled with the inverse mapping.
*
* @param aLength is the length of each array.
*/
nsresult InvertMap(const PRInt32 *aSrcMap, PRInt32 *aDestMap, PRInt32 aLength);
#endif // FULL_BIDI_ENGINE
/**
* Reverse a Right-To-Left run of Unicode text.
*
* This function preserves the integrity of characters with multiple
* code units and (optionally) modifier letters.
* Characters can be replaced by mirror-image characters
* in the destination buffer. Note that "real" mirroring has
* to be done in a rendering engine by glyph selection
* and that for many "mirrored" characters there are no
* Unicode characters as mirror-image equivalents.
* There are also options to insert or remove Bidi control
* characters; see the description of the aDestSize
* and aOptions parameters and of the option bit flags.
*
* Since no Bidi controls are inserted here, this function will never
* write more than aSrcLength characters to aDest.
*
* @param aSrc A pointer to the RTL run text.
*
* @param aSrcLength The length of the RTL run.
* If the NSBIDI_REMOVE_BIDI_CONTROLS option
* is set, then the destination length may be less than
* aSrcLength.
* If this option is not set, then the destination length
* will be exactly aSrcLength.
*
* @param aDest A pointer to where the reordered text is to be copied.
* aSrc[aSrcLength] and aDest[aSrcLength]
* must not overlap.
*
* @param aOptions A bit set of options for the reordering that control
* how the reordered text is written.
*
* @param aDestSize will receive the number of characters that were written to aDest.
*/
nsresult WriteReverse(const PRUnichar *aSrc, PRInt32 aSrcLength, PRUnichar *aDest, PRUint16 aOptions, PRInt32 *aDestSize);
protected:
friend class nsBidiPresUtils;
/** length of the current text */
PRInt32 mLength;
/** memory sizes in bytes */
PRSize mDirPropsSize, mLevelsSize, mRunsSize;
/** allocated memory */
DirProp* mDirPropsMemory;
nsBidiLevel* mLevelsMemory;
Run* mRunsMemory;
/** indicators for whether memory may be allocated after construction */
PRBool mMayAllocateText, mMayAllocateRuns;
const DirProp* mDirProps;
nsBidiLevel* mLevels;
/** the paragraph level */
nsBidiLevel mParaLevel;
/** flags is a bit set for which directional properties are in the text */
Flags mFlags;
/** the overall paragraph or line directionality - see nsBidiDirection */
nsBidiDirection mDirection;
/** characters after trailingWSStart are WS and are */
/* implicitly at the paraLevel (rule (L1)) - levels may not reflect that */
PRInt32 mTrailingWSStart;
/** fields for line reordering */
PRInt32 mRunCount; /* ==-1: runs not set up yet */
Run* mRuns;
/** for non-mixed text, we only need a tiny array of runs (no malloc()) */
Run mSimpleRuns[1];
private:
void Init();
PRBool GetMemory(void **aMemory, PRSize* aSize, PRBool aMayAllocate, PRSize aSizeNeeded);
void Free();
void GetDirProps(const PRUnichar *aText);
nsBidiDirection ResolveExplicitLevels();
nsresult CheckExplicitLevels(nsBidiDirection *aDirection);
nsBidiDirection DirectionFromFlags(Flags aFlags);
void ResolveImplicitLevels(PRInt32 aStart, PRInt32 aLimit, DirProp aSOR, DirProp aEOR);
void AdjustWSLevels();
void SetTrailingWSStart();
PRBool GetRuns();
void GetSingleRun(nsBidiLevel aLevel);
void ReorderLine(nsBidiLevel aMinLevel, nsBidiLevel aMaxLevel);
PRBool PrepareReorder(const nsBidiLevel *aLevels, PRInt32 aLength, PRInt32 *aIndexMap, nsBidiLevel *aMinLevel, nsBidiLevel *aMaxLevel);
PRInt32 doWriteReverse(const PRUnichar *src, PRInt32 srcLength,
PRUnichar *dest, PRUint16 options);
};
#endif // _nsBidi_h_