mozilla
/
gecko-dev
зеркало из https://github.com/mozilla/gecko-dev.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность

Bug 924851 - patch 2 - When ENABLE_INTL_API is true, make nsBidi into a minimal wrapper around ICU's ubidi.h functions. r=xidorn 

--HG--
rename : layout/base/nsBidi.cpp => layout/base/nsBidi_noICU.cpp
rename : layout/base/nsBidi.h => layout/base/nsBidi_noICU.h
This commit is contained in:
Jonathan Kew 2016-10-06 09:49:17 +01:00
Родитель 3cc319fde5
Коммит 7e2171a18c
7 изменённых файлов: 988 добавлений и 700 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

16
layout/base/moz.build
Просмотреть файл

@ -131,7 +131,6 @@ UNIFIED_SOURCES += [
'LayoutLogging.cpp',
'MaskLayerImageCache.cpp',
'MobileViewportManager.cpp',
'nsBidi.cpp',
'nsBidiPresUtils.cpp',
'nsCaret.cpp',
'nsCounterManager.cpp',
@ -166,6 +165,21 @@ UNIFIED_SOURCES += [
'ZoomConstraintsClient.cpp',
]
if CONFIG['ENABLE_INTL_API']:
EXPORTS += [
'nsBidi_ICU.h',
]
UNIFIED_SOURCES += [
'nsBidi_ICU.cpp',
]
else:
EXPORTS += [
'nsBidi_noICU.h',
]
UNIFIED_SOURCES += [
'nsBidi_noICU.cpp',
]
# nsPresArena.cpp needs to be built separately because it uses plarena.h.
# nsRefreshDriver.cpp needs to be built separately because of name clashes in the OS X headers
SOURCES += [

703
layout/base/nsBidi.h
Просмотреть файл

@ -7,703 +7,10 @@
#ifndef nsBidi_h__
#define nsBidi_h__
#include "nsBidiUtils.h"
#include "nsIFrame.h" // for frame property declaration
// Bidi reordering engine from ICU
/*
* javadoc-style comments are intended to be transformed into HTML
* using DOC++ - see
* http://www.zib.de/Visual/software/doc++/index.html .
*
* The HTML documentation is created with
* doc++ -H nsBidi.h
*/
/**
* @mainpage BIDI algorithm for Mozilla (from ICU)
*
* <h2>BIDI algorithm for Mozilla</h2>
*
* This is an implementation of the Unicode Bidirectional algorithm.
* The algorithm is defined in the
* <a href="http://www.unicode.org/unicode/reports/tr9/">Unicode Technical Report 9</a>,
* version 5, also described in The Unicode Standard, Version 3.0 .<p>
*
* <h3>General remarks about the API:</h3>
*
* The <quote>limit</quote> of a sequence of characters is the position just after their
* last character, i.e., one more than that position.<p>
*
* Some of the API functions provide access to <quote>runs</quote>.
* Such a <quote>run</quote> is defined as a sequence of characters
* that are at the same embedding level
* after performing the BIDI algorithm.<p>
*
* @author Markus W. Scherer. Ported to Mozilla by Simon Montagu
* @version 1.0
*/
/**
* Special value which can be returned by the mapping functions when a logical
* index has no corresponding visual index or vice-versa.
* @see GetVisualIndex
* @see GetVisualMap
* @see GetLogicalIndex
* @see GetLogicalMap
*/
#define NSBIDI_MAP_NOWHERE (-1)
/* miscellaneous definitions ------------------------------------------------ */
/* helper macros for each allocated array member */
#define GETDIRPROPSMEMORY(length) nsBidi::GetMemory((void **)&mDirPropsMemory, \
&mDirPropsSize, \
(length))
#define GETLEVELSMEMORY(length) nsBidi::GetMemory((void **)&mLevelsMemory, \
&mLevelsSize, \
(length))
#define GETRUNSMEMORY(length) nsBidi::GetMemory((void **)&mRunsMemory, \
&mRunsSize, \
(length)*sizeof(Run))
#define GETISOLATESMEMORY(length) nsBidi::GetMemory((void **)&mIsolatesMemory, \
&mIsolatesSize, \
(length)*sizeof(Isolate))
#define GETOPENINGSMEMORY(length) nsBidi::GetMemory((void **)&mOpeningsMemory, \
&mOpeningsSize, \
(length)*sizeof(Opening))
/*
* 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 uint8_t 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(ENL)| \
DIRPROP_FLAG(ENR)|DIRPROP_FLAG(AN)|DIRPROP_FLAG(LRE)| \
DIRPROP_FLAG(LRO)|DIRPROP_FLAG(LRI))
#define MASK_RTL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL)|DIRPROP_FLAG(RLE)| \
DIRPROP_FLAG(RLO)|DIRPROP_FLAG(RLI))
#define MASK_R_AL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL))
/* explicit embedding codes */
#define MASK_EXPLICIT (DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO)|DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO)|DIRPROP_FLAG(PDF))
/* explicit isolate codes */
#define MASK_ISO (DIRPROP_FLAG(LRI)|DIRPROP_FLAG(RLI)|DIRPROP_FLAG(FSI)|DIRPROP_FLAG(PDI))
#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|MASK_ISO)
/* types that are neutrals or could becomes neutrals in (Wn) */
#define MASK_POSSIBLE_N (DIRPROP_FLAG(O_N)|DIRPROP_FLAG(CS)|DIRPROP_FLAG(ES)|DIRPROP_FLAG(ET)|MASK_WS)
/*
* 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)
/*
* The following bit is used for the directional isolate status.
* Stack entries corresponding to isolate sequences are greater than ISOLATE.
*/
#define ISOLATE 0x0100
/* number of isolate entries allocated initially without malloc */
#define SIMPLE_ISOLATES_SIZE 5
/* number of isolate run entries for paired brackets allocated initially without malloc */
#define SIMPLE_OPENINGS_COUNT 8
/* 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)
#if !ENABLE_INTL_API // these are provided by ICU if present in the build
#define UTF_ERROR_VALUE 0xffff
/* definitions with forward iteration --------------------------------------- */
/*
* all the macros that go forward assume that
* the initial offset is 0<=i<length;
* they update the offset
*/
/* fast versions, no error-checking */
#define UTF16_APPEND_CHAR_UNSAFE(s, i, c){ \
if((uint32_t)(c)<=0xffff) { \
(s)[(i)++]=(char16_t)(c); \
} else { \
(s)[(i)++]=(char16_t)((c)>>10)+0xd7c0; \
(s)[(i)++]=(char16_t)(c)&0x3ff|0xdc00; \
} \
}
/* safe versions with error-checking and optional regularity-checking */
#define UTF16_APPEND_CHAR_SAFE(s, i, length, c) { \
if((PRUInt32)(c)<=0xffff) { \
(s)[(i)++]=(char16_t)(c); \
} else if((PRUInt32)(c)<=0x10ffff) { \
if((i)+1<(length)) { \
(s)[(i)++]=(char16_t)((c)>>10)+0xd7c0; \
(s)[(i)++]=(char16_t)(c)&0x3ff|0xdc00; \
} else /* not enough space */ { \
(s)[(i)++]=UTF_ERROR_VALUE; \
} \
} else /* c>0x10ffff, write error value */ { \
(s)[(i)++]=UTF_ERROR_VALUE; \
} \
}
/* definitions with backward iteration -------------------------------------- */
/*
* all the macros that go backward assume that
* the valid buffer range starts at offset 0
* and that the initial offset is 0<i<=length;
* they update the offset
*/
/* fast versions, no error-checking */
/*
* Get a single code point from an offset that points behind the last
* of the code units that belong to that code point.
* Assume 0<=i<length.
*/
#define UTF16_PREV_CHAR_UNSAFE(s, i, c) { \
(c)=(s)[--(i)]; \
if(IS_SECOND_SURROGATE(c)) { \
(c)=GET_UTF_32((s)[--(i)], (c)); \
} \
}
#define UTF16_BACK_1_UNSAFE(s, i) { \
if(IS_SECOND_SURROGATE((s)[--(i)])) { \
--(i); \
} \
}
#define UTF16_BACK_N_UNSAFE(s, i, n) { \
int32_t __N=(n); \
while(__N>0) { \
UTF16_BACK_1_UNSAFE(s, i); \
--__N; \
} \
}
/* safe versions with error-checking and optional regularity-checking */
#define UTF16_PREV_CHAR_SAFE(s, start, i, c, strict) { \
(c)=(s)[--(i)]; \
if(IS_SECOND_SURROGATE(c)) { \
char16_t __c2; \
if((i)>(start) && IS_FIRST_SURROGATE(__c2=(s)[(i)-1])) { \
--(i); \
(c)=GET_UTF_32(__c2, (c)); \
/* strict: ((c)&0xfffe)==0xfffe is caught by UTF_IS_ERROR() */ \
} else if(strict) {\
/* unmatched second surrogate */ \
(c)=UTF_ERROR_VALUE; \
} \
} else if(strict && IS_FIRST_SURROGATE(c)) { \
/* unmatched first surrogate */ \
(c)=UTF_ERROR_VALUE; \
/* else strict: (c)==0xfffe is caught by UTF_IS_ERROR() */ \
} \
}
#define UTF16_BACK_1_SAFE(s, start, i) { \
if(IS_SECOND_SURROGATE((s)[--(i)]) && (i)>(start) && IS_FIRST_SURROGATE((s)[(i)-1])) { \
--(i); \
} \
}
#define UTF16_BACK_N_SAFE(s, start, i, n) { \
int32_t __N=(n); \
while(__N>0 && (i)>(start)) { \
UTF16_BACK_1_SAFE(s, start, i); \
--__N; \
} \
}
#define UTF_PREV_CHAR_UNSAFE(s, i, c) UTF16_PREV_CHAR_UNSAFE(s, i, c)
#define UTF_PREV_CHAR_SAFE(s, start, i, c, strict) UTF16_PREV_CHAR_SAFE(s, start, i, c, strict)
#define UTF_BACK_1_UNSAFE(s, i) UTF16_BACK_1_UNSAFE(s, i)
#define UTF_BACK_1_SAFE(s, start, i) UTF16_BACK_1_SAFE(s, start, i)
#define UTF_BACK_N_UNSAFE(s, i, n) UTF16_BACK_N_UNSAFE(s, i, n)
#define UTF_BACK_N_SAFE(s, start, i, n) UTF16_BACK_N_SAFE(s, start, i, n)
#define UTF_APPEND_CHAR_UNSAFE(s, i, c) UTF16_APPEND_CHAR_UNSAFE(s, i, c)
#define UTF_APPEND_CHAR_SAFE(s, i, length, c) UTF16_APPEND_CHAR_SAFE(s, i, length, c)
#define UTF_PREV_CHAR(s, start, i, c) UTF_PREV_CHAR_SAFE(s, start, i, c, false)
#define UTF_BACK_1(s, start, i) UTF_BACK_1_SAFE(s, start, i)
#define UTF_BACK_N(s, start, i, n) UTF_BACK_N_SAFE(s, start, i, n)
#define UTF_APPEND_CHAR(s, i, length, c) UTF_APPEND_CHAR_SAFE(s, i, length, c)
#endif // !ENABLE_INTL_API
struct Isolate {
int32_t start1;
int16_t stateImp;
int16_t state;
};
// For bracket matching
#define FOUND_L DIRPROP_FLAG(L)
#define FOUND_R DIRPROP_FLAG(R)
struct Opening {
int32_t position; /* position of opening bracket */
int32_t match; /* matching char or -position of closing bracket */
int32_t contextPos; /* position of last strong char found before opening */
uint16_t flags; /* bits for L or R/AL found within the pair */
DirProp contextDir; /* L or R according to last strong char before opening */
uint8_t filler; /* to complete a nice multiple of 4 chars */
};
struct IsoRun {
int32_t contextPos; /* position of char determining context */
uint16_t start; /* index of first opening entry for this run */
uint16_t limit; /* index after last opening entry for this run */
nsBidiLevel level; /* level of this run */
DirProp lastStrong; /* bidi class of last strong char found in this run */
DirProp lastBase; /* bidi class of last base char found in this run */
DirProp contextDir; /* L or R to use as context for following openings */
};
class nsBidi;
/* Run structure for reordering --------------------------------------------- */
typedef struct Run {
int32_t logicalStart; /* first character of the run; b31 indicates even/odd level */
int32_t visualLimit; /* last visual position of the run +1 */
} Run;
/* in a Run, logicalStart will get this bit set if the run level is odd */
#define INDEX_ODD_BIT (1UL<<31)
#define MAKE_INDEX_ODD_PAIR(index, level) (index|((uint32_t)level<<31))
#define ADD_ODD_BIT_FROM_LEVEL(x, level) ((x)|=((uint32_t)level<<31))
#define REMOVE_ODD_BIT(x) ((x)&=~INDEX_ODD_BIT)
#define GET_INDEX(x) ((x)&~INDEX_ODD_BIT)
#define GET_ODD_BIT(x) ((uint32_t)(x)>>31)
#define IS_ODD_RUN(x) (((x)&INDEX_ODD_BIT)!=0)
#define IS_EVEN_RUN(x) (((x)&INDEX_ODD_BIT)==0)
typedef uint32_t Flags;
enum { DirProp_L=0, DirProp_R=1, DirProp_EN=2, DirProp_AN=3, DirProp_ON=4, DirProp_S=5, DirProp_B=6 }; /* reduced dirProp */
#define IMPTABLEVELS_COLUMNS (DirProp_B + 2)
typedef const uint8_t ImpTab[][IMPTABLEVELS_COLUMNS];
typedef const uint8_t (*PImpTab)[IMPTABLEVELS_COLUMNS];
typedef const uint8_t ImpAct[];
typedef const uint8_t *PImpAct;
struct LevState {
PImpTab pImpTab; /* level table pointer */
PImpAct pImpAct; /* action map array */
int32_t startON; /* start of ON sequence */
int32_t state; /* current state */
int32_t runStart; /* start position of the run */
nsBidiLevel runLevel; /* run level before implicit solving */
};
/**
* This class holds information about a paragraph of text
* with Bidi-algorithm-related details, or about one line of
* such a paragraph.<p>
* Reordering can be done on a line, or on a paragraph which is
* then interpreted as one single line.<p>
*
* On construction, the class is initially empty. It is assigned
* the Bidi properties of a paragraph by <code>SetPara</code>
* or the Bidi properties of a line of a paragraph by
* <code>SetLine</code>.<p>
* A Bidi class can be reused for as long as it is not deallocated
* by calling its destructor.<p>
* <code>SetPara</code> 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 <code>SetPara()</code>
* or the Bidi properties of a line of a paragraph by
* <code>GetLine()</code>.<p>
* This object can be reused for as long as it is not destroyed.<p>
* <code>SetPara()</code> will allocate additional memory for
* internal structures as necessary.
*
*/
nsBidi();
/** @brief Destructor. */
virtual ~nsBidi();
/**
* Perform the Unicode Bidi algorithm. It is defined in the
* <a href="http://www.unicode.org/unicode/reports/tr9/">Unicode Technical Report 9</a>,
* version 5,
* also described in The Unicode Standard, Version 3.0 .<p>
*
* This function takes a single plain text paragraph with or without
* externally specified embedding levels from <quote>styled</quote> text
* and computes the left-right-directionality of each character.<p>
*
* 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.<br>
* 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.<p>
*
* 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).
* <strong>The text must be (at least) <code>aLength</code> long.</strong>
*
* @param aLength is the length of the text; if <code>aLength==-1</code> 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 <code>aParaLevel</code> can be set to
* either <code>NSBIDI_DEFAULT_LTR</code>
* or <code>NSBIDI_DEFAULT_RTL</code>;
* 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 <code>NSBIDI_MAX_EXPLICIT_LEVEL</code> is also valid,
* with odd levels indicating RTL.
*/
nsresult SetPara(const char16_t *aText, int32_t aLength, nsBidiLevel aParaLevel);
/**
* Get the directionality of the text.
*
* @param aDirection receives a <code>NSBIDI_XXX</code> 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 paragraph level of the text.
*
* @param aParaLevel receives a <code>NSBIDI_XXX</code> value indicating the paragraph level
*
* @see nsBidiLevel
*/
nsresult GetParaLevel(nsBidiLevel* aParaLevel);
/**
* Get a logical run.
* This function returns information about a run and is used
* to retrieve runs in logical order.<p>
* 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
* <code>aLogicalStart</code>.
* This pointer can be <code>nullptr</code> if this
* value is not necessary.
*
* @param aLevel will receive the level of the run.
* This pointer can be <code>nullptr</code> if this
* value is not necessary.
*/
nsresult GetLogicalRun(int32_t aLogicalStart, int32_t* aLogicalLimit, nsBidiLevel* aLevel);
/**
* Get the number of runs.
* This function may invoke the actual reordering on the
* <code>nsBidi</code> object, after <code>SetPara</code>
* may have resolved only the levels of the text. Therefore,
* <code>CountRuns</code> may have to allocate memory,
* and may fail doing so.
*
* @param aRunCount will receive the number of runs.
*/
nsresult CountRuns(int32_t* 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.<p>
* <code>CountRuns</code> should be called
* before the runs are retrieved.
*
* @param aRunIndex is the number of the run in visual order, in the
* range <code>[0..CountRuns-1]</code>.
*
* @param aLogicalStart is the first logical character index in the text.
* The pointer may be <code>nullptr</code> if this index is not needed.
*
* @param aLength is the number of characters (at least one) in the run.
* The pointer may be <code>nullptr</code> if this is not needed.
*
* @param aDirection will receive the directionality of the run,
* <code>NSBIDI_LTR==0</code> or <code>NSBIDI_RTL==1</code>,
* never <code>NSBIDI_MIXED</code>.
*
* @see CountRuns<p>
*
* Example:
* @code
* int32_t i, count, logicalStart, visualIndex=0, length;
* nsBidiDirection dir;
* pBidi->CountRuns(&count);
* for(i=0; i<count; ++i) {
* pBidi->GetVisualRun(i, &logicalStart, &length, &dir);
* if(NSBIDI_LTR==dir) {
* do { // LTR
* show_char(text[logicalStart++], visualIndex++);
* } while(--length>0);
* } else {
* logicalStart+=length; // logicalLimit
* do { // RTL
* show_char(text[--logicalStart], visualIndex++);
* } while(--length>0);
* }
* }
* @endcode
*
* Note that in right-to-left runs, code like this places
* modifier letters before base characters and second surrogates
* before first ones.
*/
nsresult GetVisualRun(int32_t aRunIndex, int32_t* aLogicalStart, int32_t* aLength, nsBidiDirection* aDirection);
/**
* 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 <code>GetVisualMap</code> on a
* <code>nsBidi</code> object.
*
* @param aLevels is an array with <code>aLength</code> 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 <code>aLength>0</code>.
*
* @param aIndexMap is a pointer to an array of <code>aLength</code>
* indexes which will reflect the reordering of the characters.
* The array does not need to be initialized.<p>
* The index map will result in <code>aIndexMap[aVisualIndex]==aLogicalIndex</code>.
*/
static nsresult ReorderVisual(const nsBidiLevel *aLevels, int32_t aLength, int32_t *aIndexMap);
/**
* 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 <code>aDestSize</code>
* and <code>aOptions</code> parameters and of the option bit flags.
*
* Since no Bidi controls are inserted here, this function will never
* write more than <code>aSrcLength</code> characters to <code>aDest</code>.
*
* @param aSrc A pointer to the RTL run text.
*
* @param aSrcLength The length of the RTL run.
* If the <code>NSBIDI_REMOVE_BIDI_CONTROLS</code> option
* is set, then the destination length may be less than
* <code>aSrcLength</code>.
* If this option is not set, then the destination length
* will be exactly <code>aSrcLength</code>.
*
* @param aDest A pointer to where the reordered text is to be copied.
* <code>aSrc[aSrcLength]</code> and <code>aDest[aSrcLength]</code>
* 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 <code>aDest</code>.
*/
nsresult WriteReverse(const char16_t *aSrc, int32_t aSrcLength, char16_t *aDest, uint16_t aOptions, int32_t *aDestSize);
protected:
friend class nsBidiPresUtils;
class BracketData {
public:
explicit BracketData(const nsBidi* aBidi);
~BracketData();
void ProcessBoundary(int32_t aLastDirControlCharPos,
nsBidiLevel aContextLevel,
nsBidiLevel aEmbeddingLevel,
const DirProp* aDirProps);
void ProcessLRI_RLI(nsBidiLevel aLevel);
void ProcessPDI();
bool AddOpening(char16_t aMatch, int32_t aPosition);
void FixN0c(int32_t aOpeningIndex, int32_t aNewPropPosition,
DirProp aNewProp, DirProp* aDirProps);
DirProp ProcessClosing(int32_t aOpenIdx, int32_t aPosition,
DirProp* aDirProps);
bool ProcessChar(int32_t aPosition, char16_t aCh, DirProp* aDirProps,
nsBidiLevel* aLevels);
private:
// array of opening entries which should be enough in most cases;
// no malloc() needed
Opening mSimpleOpenings[SIMPLE_OPENINGS_COUNT];
Opening* mOpenings; // pointer to current array of entries,
// either mSimpleOpenings or malloced array
Opening* mOpeningsMemory;
size_t mOpeningsSize;
// array of nested isolated sequence entries; can never exceed
// UBIDI_MAX_EXPLICIT_LEVEL
// + 1 for index 0
// + 1 for before the first isolated sequence
IsoRun mIsoRuns[NSBIDI_MAX_EXPLICIT_LEVEL+2];
int32_t mIsoRunLast; // index of last used entry in mIsoRuns
int32_t mOpeningsCount; // number of allocated entries in mOpenings
};
/** length of the current text */
int32_t mLength;
/** memory sizes in bytes */
size_t mDirPropsSize, mLevelsSize, mRunsSize;
size_t mIsolatesSize;
/** allocated memory */
DirProp* mDirPropsMemory;
nsBidiLevel* mLevelsMemory;
Run* mRunsMemory;
Isolate* mIsolatesMemory;
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 */
int32_t mTrailingWSStart;
/** fields for line reordering */
int32_t 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];
/* maxium of current nesting depth of isolate sequences */
/* Within ResolveExplicitLevels() and checkExpicitLevels(), this is the maximal
nesting encountered.
Within ResolveImplicitLevels(), this is the index of the current isolates
stack entry. */
int32_t mIsolateCount;
Isolate* mIsolates;
/** for simple text, have a small stack (no malloc()) */
Isolate mSimpleIsolates[SIMPLE_ISOLATES_SIZE];
private:
void Init();
static bool GetMemory(void **aMemory, size_t* aSize, size_t aSizeNeeded);
void Free();
void GetDirProps(const char16_t *aText);
void ResolveExplicitLevels(nsBidiDirection *aDirection, const char16_t *aText);
nsBidiDirection DirectionFromFlags(Flags aFlags);
void ProcessPropertySeq(LevState *pLevState, uint8_t _prop, int32_t start, int32_t limit);
void ResolveImplicitLevels(int32_t aStart, int32_t aLimit, DirProp aSOR, DirProp aEOR);
void AdjustWSLevels();
void SetTrailingWSStart();
bool GetRuns();
void GetSingleRun(nsBidiLevel aLevel);
void ReorderLine(nsBidiLevel aMinLevel, nsBidiLevel aMaxLevel);
static bool PrepareReorder(const nsBidiLevel *aLevels, int32_t aLength, int32_t *aIndexMap, nsBidiLevel *aMinLevel, nsBidiLevel *aMaxLevel);
};
#if ENABLE_INTL_API
#include "nsBidi_ICU.h"
#else
#include "nsBidi_noICU.h"
#endif
#endif // _nsBidi_h_

2
layout/base/nsBidiPresUtils.cpp
Просмотреть файл

@ -1926,7 +1926,7 @@ nsBidiPresUtils::CalculateCharType(nsBidi* aBidiEngine,
ch = SURROGATE_TO_UCS4(ch, aText[offset + 1]);
charLen = 2;
}
charType = GetBidiCat(ch);
charType = unicode::GetBidiCat(ch);
}
if (!CHARTYPE_IS_WEAK(charType) ) {

68
layout/base/nsBidi_ICU.cpp Normal file
Просмотреть файл

@ -0,0 +1,68 @@
/* -*- Mode: C; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-
*
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#include "nsBidi_ICU.h"
#include "ICUUtils.h"
nsBidi::nsBidi()
{
mBiDi = ubidi_open();
}
nsBidi::~nsBidi()
{
ubidi_close(mBiDi);
}
nsresult nsBidi::SetPara(const char16_t *aText, int32_t aLength,
nsBidiLevel aParaLevel)
{
UErrorCode error = U_ZERO_ERROR;
ubidi_setPara(mBiDi, reinterpret_cast<const UChar*>(aText), aLength,
aParaLevel, nullptr, &error);
return ICUUtils::UErrorToNsResult(error);
}
nsresult nsBidi::GetDirection(nsBidiDirection* aDirection)
{
*aDirection = nsBidiDirection(ubidi_getDirection(mBiDi));
return NS_OK;
}
nsresult nsBidi::GetParaLevel(nsBidiLevel* aParaLevel)
{
*aParaLevel = ubidi_getParaLevel(mBiDi);
return NS_OK;
}
nsresult nsBidi::GetLogicalRun(int32_t aLogicalStart, int32_t* aLogicalLimit,
nsBidiLevel* aLevel)
{
ubidi_getLogicalRun(mBiDi, aLogicalStart, aLogicalLimit, aLevel);
return NS_OK;
}
nsresult nsBidi::CountRuns(int32_t* aRunCount)
{
UErrorCode errorCode = U_ZERO_ERROR;
*aRunCount = ubidi_countRuns(mBiDi, &errorCode);
return ICUUtils::UErrorToNsResult(errorCode);
}
nsresult nsBidi::GetVisualRun(int32_t aRunIndex, int32_t* aLogicalStart,
int32_t* aLength, nsBidiDirection* aDirection)
{
*aDirection = nsBidiDirection(ubidi_getVisualRun(mBiDi, aRunIndex,
aLogicalStart, aLength));
return NS_OK;
}
nsresult nsBidi::ReorderVisual(const nsBidiLevel* aLevels, int32_t aLength,
int32_t* aIndexMap)
{
ubidi_reorderVisual(aLevels, aLength, aIndexMap);
return NS_OK;
}

190
layout/base/nsBidi_ICU.h Normal file
Просмотреть файл

@ -0,0 +1,190 @@
/* -*- Mode: C; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-
*
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#ifndef nsBidi_ICU_h__
#define nsBidi_ICU_h__
#include "unicode/ubidi.h"
#include "nsIFrame.h" // for nsBidiLevel/nsBidiDirection declarations
// nsBidi implemented as a simple wrapper around the bidi reordering engine
// from ICU.
// We could eliminate this and let callers use the ICU functions directly
// once we no longer care about building without ICU available.
class nsBidi
{
public:
/** @brief Default constructor.
*
* The nsBidi object is initially empty. It is assigned
* the Bidi properties of a paragraph by <code>SetPara()</code>.
*/
explicit nsBidi();
/** @brief Destructor. */
virtual ~nsBidi();
/**
* Perform the Unicode Bidi algorithm.
*
* @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).
* <strong>The text must be (at least) <code>aLength</code> long.</strong>
*
* @param aLength is the length of the text; if <code>aLength==-1</code> 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 <code>aParaLevel</code> can be set to
* either <code>NSBIDI_DEFAULT_LTR</code>
* or <code>NSBIDI_DEFAULT_RTL</code>;
* 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 <code>NSBIDI_MAX_EXPLICIT_LEVEL</code>
* is also valid, with odd levels indicating RTL.
*/
nsresult SetPara(const char16_t *aText, int32_t aLength,
nsBidiLevel aParaLevel);
/**
* Get the directionality of the text.
*
* @param aDirection receives a <code>NSBIDI_XXX</code> 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 paragraph level of the text.
*
* @param aParaLevel receives a <code>NSBIDI_XXX</code> value indicating
* the paragraph level
*
* @see nsBidiLevel
*/
nsresult GetParaLevel(nsBidiLevel* aParaLevel);
/**
* Get a logical run.
* This function returns information about a run and is used
* to retrieve runs in logical order.<p>
* 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
* <code>aLogicalStart</code>.
* This pointer can be <code>nullptr</code> if this
* value is not necessary.
*
* @param aLevel will receive the level of the run.
* This pointer can be <code>nullptr</code> if this
* value is not necessary.
*/
nsresult GetLogicalRun(int32_t aLogicalStart, int32_t* aLogicalLimit,
nsBidiLevel* aLevel);
/**
* Get the number of runs.
* This function may invoke the actual reordering on the
* <code>nsBidi</code> object, after <code>SetPara</code>
* may have resolved only the levels of the text. Therefore,
* <code>CountRuns</code> may have to allocate memory,
* and may fail doing so.
*
* @param aRunCount will receive the number of runs.
*/
nsresult CountRuns(int32_t* 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.<p>
* <code>CountRuns</code> should be called
* before the runs are retrieved.
*
* @param aRunIndex is the number of the run in visual order, in the
* range <code>[0..CountRuns-1]</code>.
*
* @param aLogicalStart is the first logical character index in the text.
* The pointer may be <code>nullptr</code> if this index is not needed.
*
* @param aLength is the number of characters (at least one) in the run.
* The pointer may be <code>nullptr</code> if this is not needed.
*
* @param aDirection will receive the directionality of the run,
* <code>NSBIDI_LTR==0</code> or <code>NSBIDI_RTL==1</code>,
* never <code>NSBIDI_MIXED</code>.
*
* @see CountRuns<p>
*
* Example:
* @code
* int32_t i, count, logicalStart, visualIndex=0, length;
* nsBidiDirection dir;
* pBidi->CountRuns(&count);
* for(i=0; i<count; ++i) {
* pBidi->GetVisualRun(i, &logicalStart, &length, &dir);
* if(NSBIDI_LTR==dir) {
* do { // LTR
* show_char(text[logicalStart++], visualIndex++);
* } while(--length>0);
* } else {
* logicalStart+=length; // logicalLimit
* do { // RTL
* show_char(text[--logicalStart], visualIndex++);
* } while(--length>0);
* }
* }
* @endcode
*
* Note that in right-to-left runs, code like this places
* modifier letters before base characters and second surrogates
* before first ones.
*/
nsresult GetVisualRun(int32_t aRunIndex, int32_t* aLogicalStart,
int32_t* aLength, nsBidiDirection* aDirection);
/**
* 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 <code>GetVisualMap</code> on a
* <code>nsBidi</code> object.
*
* @param aLevels is an array with <code>aLength</code> 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 <code>aLength>0</code>.
*
* @param aIndexMap is a pointer to an array of <code>aLength</code>
* indexes which will reflect the reordering of the characters.
* The array does not need to be initialized.<p>
* The index map will result in
* <code>aIndexMap[aVisualIndex]==aLogicalIndex</code>.
*/
static nsresult ReorderVisual(const nsBidiLevel* aLevels, int32_t aLength,
int32_t* aIndexMap);
protected:
UBiDi* mBiDi;
};
#endif // _nsBidi_ICU_h_

0
layout/base/nsBidi.cpp → layout/base/nsBidi_noICU.cpp
Просмотреть файл

709
layout/base/nsBidi_noICU.h Normal file
Просмотреть файл

@ -0,0 +1,709 @@
/* -*- Mode: C; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-
*
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#ifndef nsBidi_noICU_h__
#define nsBidi_noICU_h__
#include "nsBidiUtils.h"
#include "nsIFrame.h" // for frame property declaration
// Bidi reordering engine from ICU
/*
* javadoc-style comments are intended to be transformed into HTML
* using DOC++ - see
* http://www.zib.de/Visual/software/doc++/index.html .
*
* The HTML documentation is created with
* doc++ -H nsBidi.h
*/
/**
* @mainpage BIDI algorithm for Mozilla (from ICU)
*
* <h2>BIDI algorithm for Mozilla</h2>
*
* This is an implementation of the Unicode Bidirectional algorithm.
* The algorithm is defined in the
* <a href="http://www.unicode.org/unicode/reports/tr9/">Unicode Technical Report 9</a>,
* version 5, also described in The Unicode Standard, Version 3.0 .<p>
*
* <h3>General remarks about the API:</h3>
*
* The <quote>limit</quote> of a sequence of characters is the position just after their
* last character, i.e., one more than that position.<p>
*
* Some of the API functions provide access to <quote>runs</quote>.
* Such a <quote>run</quote> is defined as a sequence of characters
* that are at the same embedding level
* after performing the BIDI algorithm.<p>
*
* @author Markus W. Scherer. Ported to Mozilla by Simon Montagu
* @version 1.0
*/
/**
* Special value which can be returned by the mapping functions when a logical
* index has no corresponding visual index or vice-versa.
* @see GetVisualIndex
* @see GetVisualMap
* @see GetLogicalIndex
* @see GetLogicalMap
*/
#define NSBIDI_MAP_NOWHERE (-1)
/* miscellaneous definitions ------------------------------------------------ */
/* helper macros for each allocated array member */
#define GETDIRPROPSMEMORY(length) nsBidi::GetMemory((void **)&mDirPropsMemory, \
&mDirPropsSize, \
(length))
#define GETLEVELSMEMORY(length) nsBidi::GetMemory((void **)&mLevelsMemory, \
&mLevelsSize, \
(length))
#define GETRUNSMEMORY(length) nsBidi::GetMemory((void **)&mRunsMemory, \
&mRunsSize, \
(length)*sizeof(Run))
#define GETISOLATESMEMORY(length) nsBidi::GetMemory((void **)&mIsolatesMemory, \
&mIsolatesSize, \
(length)*sizeof(Isolate))
#define GETOPENINGSMEMORY(length) nsBidi::GetMemory((void **)&mOpeningsMemory, \
&mOpeningsSize, \
(length)*sizeof(Opening))
/*
* 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 uint8_t 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(ENL)| \
DIRPROP_FLAG(ENR)|DIRPROP_FLAG(AN)|DIRPROP_FLAG(LRE)| \
DIRPROP_FLAG(LRO)|DIRPROP_FLAG(LRI))
#define MASK_RTL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL)|DIRPROP_FLAG(RLE)| \
DIRPROP_FLAG(RLO)|DIRPROP_FLAG(RLI))
#define MASK_R_AL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL))
/* explicit embedding codes */
#define MASK_EXPLICIT (DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO)|DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO)|DIRPROP_FLAG(PDF))
/* explicit isolate codes */
#define MASK_ISO (DIRPROP_FLAG(LRI)|DIRPROP_FLAG(RLI)|DIRPROP_FLAG(FSI)|DIRPROP_FLAG(PDI))
#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|MASK_ISO)
/* types that are neutrals or could becomes neutrals in (Wn) */
#define MASK_POSSIBLE_N (DIRPROP_FLAG(O_N)|DIRPROP_FLAG(CS)|DIRPROP_FLAG(ES)|DIRPROP_FLAG(ET)|MASK_WS)
/*
* 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)
/*
* The following bit is used for the directional isolate status.
* Stack entries corresponding to isolate sequences are greater than ISOLATE.
*/
#define ISOLATE 0x0100
/* number of isolate entries allocated initially without malloc */
#define SIMPLE_ISOLATES_SIZE 5
/* number of isolate run entries for paired brackets allocated initially without malloc */
#define SIMPLE_OPENINGS_COUNT 8
/* 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)
#if !ENABLE_INTL_API // these are provided by ICU if present in the build
#define UTF_ERROR_VALUE 0xffff
/* definitions with forward iteration --------------------------------------- */
/*
* all the macros that go forward assume that
* the initial offset is 0<=i<length;
* they update the offset
*/
/* fast versions, no error-checking */
#define UTF16_APPEND_CHAR_UNSAFE(s, i, c){ \
if((uint32_t)(c)<=0xffff) { \
(s)[(i)++]=(char16_t)(c); \
} else { \
(s)[(i)++]=(char16_t)((c)>>10)+0xd7c0; \
(s)[(i)++]=(char16_t)(c)&0x3ff|0xdc00; \
} \
}
/* safe versions with error-checking and optional regularity-checking */
#define UTF16_APPEND_CHAR_SAFE(s, i, length, c) { \
if((PRUInt32)(c)<=0xffff) { \
(s)[(i)++]=(char16_t)(c); \
} else if((PRUInt32)(c)<=0x10ffff) { \
if((i)+1<(length)) { \
(s)[(i)++]=(char16_t)((c)>>10)+0xd7c0; \
(s)[(i)++]=(char16_t)(c)&0x3ff|0xdc00; \
} else /* not enough space */ { \
(s)[(i)++]=UTF_ERROR_VALUE; \
} \
} else /* c>0x10ffff, write error value */ { \
(s)[(i)++]=UTF_ERROR_VALUE; \
} \
}
/* definitions with backward iteration -------------------------------------- */
/*
* all the macros that go backward assume that
* the valid buffer range starts at offset 0
* and that the initial offset is 0<i<=length;
* they update the offset
*/
/* fast versions, no error-checking */
/*
* Get a single code point from an offset that points behind the last
* of the code units that belong to that code point.
* Assume 0<=i<length.
*/
#define UTF16_PREV_CHAR_UNSAFE(s, i, c) { \
(c)=(s)[--(i)]; \
if(IS_SECOND_SURROGATE(c)) { \
(c)=GET_UTF_32((s)[--(i)], (c)); \
} \
}
#define UTF16_BACK_1_UNSAFE(s, i) { \
if(IS_SECOND_SURROGATE((s)[--(i)])) { \
--(i); \
} \
}
#define UTF16_BACK_N_UNSAFE(s, i, n) { \
int32_t __N=(n); \
while(__N>0) { \
UTF16_BACK_1_UNSAFE(s, i); \
--__N; \
} \
}
/* safe versions with error-checking and optional regularity-checking */
#define UTF16_PREV_CHAR_SAFE(s, start, i, c, strict) { \
(c)=(s)[--(i)]; \
if(IS_SECOND_SURROGATE(c)) { \
char16_t __c2; \
if((i)>(start) && IS_FIRST_SURROGATE(__c2=(s)[(i)-1])) { \
--(i); \
(c)=GET_UTF_32(__c2, (c)); \
/* strict: ((c)&0xfffe)==0xfffe is caught by UTF_IS_ERROR() */ \
} else if(strict) {\
/* unmatched second surrogate */ \
(c)=UTF_ERROR_VALUE; \
} \
} else if(strict && IS_FIRST_SURROGATE(c)) { \
/* unmatched first surrogate */ \
(c)=UTF_ERROR_VALUE; \
/* else strict: (c)==0xfffe is caught by UTF_IS_ERROR() */ \
} \
}
#define UTF16_BACK_1_SAFE(s, start, i) { \
if(IS_SECOND_SURROGATE((s)[--(i)]) && (i)>(start) && IS_FIRST_SURROGATE((s)[(i)-1])) { \
--(i); \
} \
}
#define UTF16_BACK_N_SAFE(s, start, i, n) { \
int32_t __N=(n); \
while(__N>0 && (i)>(start)) { \
UTF16_BACK_1_SAFE(s, start, i); \
--__N; \
} \
}
#define UTF_PREV_CHAR_UNSAFE(s, i, c) UTF16_PREV_CHAR_UNSAFE(s, i, c)
#define UTF_PREV_CHAR_SAFE(s, start, i, c, strict) UTF16_PREV_CHAR_SAFE(s, start, i, c, strict)
#define UTF_BACK_1_UNSAFE(s, i) UTF16_BACK_1_UNSAFE(s, i)
#define UTF_BACK_1_SAFE(s, start, i) UTF16_BACK_1_SAFE(s, start, i)
#define UTF_BACK_N_UNSAFE(s, i, n) UTF16_BACK_N_UNSAFE(s, i, n)
#define UTF_BACK_N_SAFE(s, start, i, n) UTF16_BACK_N_SAFE(s, start, i, n)
#define UTF_APPEND_CHAR_UNSAFE(s, i, c) UTF16_APPEND_CHAR_UNSAFE(s, i, c)
#define UTF_APPEND_CHAR_SAFE(s, i, length, c) UTF16_APPEND_CHAR_SAFE(s, i, length, c)
#define UTF_PREV_CHAR(s, start, i, c) UTF_PREV_CHAR_SAFE(s, start, i, c, false)
#define UTF_BACK_1(s, start, i) UTF_BACK_1_SAFE(s, start, i)
#define UTF_BACK_N(s, start, i, n) UTF_BACK_N_SAFE(s, start, i, n)
#define UTF_APPEND_CHAR(s, i, length, c) UTF_APPEND_CHAR_SAFE(s, i, length, c)
#endif // !ENABLE_INTL_API
struct Isolate {
int32_t start1;
int16_t stateImp;
int16_t state;
};
// For bracket matching
#define FOUND_L DIRPROP_FLAG(L)
#define FOUND_R DIRPROP_FLAG(R)
struct Opening {
int32_t position; /* position of opening bracket */
int32_t match; /* matching char or -position of closing bracket */
int32_t contextPos; /* position of last strong char found before opening */
uint16_t flags; /* bits for L or R/AL found within the pair */
DirProp contextDir; /* L or R according to last strong char before opening */
uint8_t filler; /* to complete a nice multiple of 4 chars */
};
struct IsoRun {
int32_t contextPos; /* position of char determining context */
uint16_t start; /* index of first opening entry for this run */
uint16_t limit; /* index after last opening entry for this run */
nsBidiLevel level; /* level of this run */
DirProp lastStrong; /* bidi class of last strong char found in this run */
DirProp lastBase; /* bidi class of last base char found in this run */
DirProp contextDir; /* L or R to use as context for following openings */
};
class nsBidi;
/* Run structure for reordering --------------------------------------------- */
typedef struct Run {
int32_t logicalStart; /* first character of the run; b31 indicates even/odd level */
int32_t visualLimit; /* last visual position of the run +1 */
} Run;
/* in a Run, logicalStart will get this bit set if the run level is odd */
#define INDEX_ODD_BIT (1UL<<31)
#define MAKE_INDEX_ODD_PAIR(index, level) (index|((uint32_t)level<<31))
#define ADD_ODD_BIT_FROM_LEVEL(x, level) ((x)|=((uint32_t)level<<31))
#define REMOVE_ODD_BIT(x) ((x)&=~INDEX_ODD_BIT)
#define GET_INDEX(x) ((x)&~INDEX_ODD_BIT)
#define GET_ODD_BIT(x) ((uint32_t)(x)>>31)
#define IS_ODD_RUN(x) (((x)&INDEX_ODD_BIT)!=0)
#define IS_EVEN_RUN(x) (((x)&INDEX_ODD_BIT)==0)
typedef uint32_t Flags;
enum { DirProp_L=0, DirProp_R=1, DirProp_EN=2, DirProp_AN=3, DirProp_ON=4, DirProp_S=5, DirProp_B=6 }; /* reduced dirProp */
#define IMPTABLEVELS_COLUMNS (DirProp_B + 2)
typedef const uint8_t ImpTab[][IMPTABLEVELS_COLUMNS];
typedef const uint8_t (*PImpTab)[IMPTABLEVELS_COLUMNS];
typedef const uint8_t ImpAct[];
typedef const uint8_t *PImpAct;
struct LevState {
PImpTab pImpTab; /* level table pointer */
PImpAct pImpAct; /* action map array */
int32_t startON; /* start of ON sequence */
int32_t state; /* current state */
int32_t runStart; /* start position of the run */
nsBidiLevel runLevel; /* run level before implicit solving */
};
/**
* This class holds information about a paragraph of text
* with Bidi-algorithm-related details, or about one line of
* such a paragraph.<p>
* Reordering can be done on a line, or on a paragraph which is
* then interpreted as one single line.<p>
*
* On construction, the class is initially empty. It is assigned
* the Bidi properties of a paragraph by <code>SetPara</code>
* or the Bidi properties of a line of a paragraph by
* <code>SetLine</code>.<p>
* A Bidi class can be reused for as long as it is not deallocated
* by calling its destructor.<p>
* <code>SetPara</code> 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 <code>SetPara()</code>
* or the Bidi properties of a line of a paragraph by
* <code>GetLine()</code>.<p>
* This object can be reused for as long as it is not destroyed.<p>
* <code>SetPara()</code> will allocate additional memory for
* internal structures as necessary.
*
*/
nsBidi();
/** @brief Destructor. */
virtual ~nsBidi();
/**
* Perform the Unicode Bidi algorithm. It is defined in the
* <a href="http://www.unicode.org/unicode/reports/tr9/">Unicode Technical Report 9</a>,
* version 5,
* also described in The Unicode Standard, Version 3.0 .<p>
*
* This function takes a single plain text paragraph with or without
* externally specified embedding levels from <quote>styled</quote> text
* and computes the left-right-directionality of each character.<p>
*
* 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.<br>
* 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.<p>
*
* 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).
* <strong>The text must be (at least) <code>aLength</code> long.</strong>
*
* @param aLength is the length of the text; if <code>aLength==-1</code> 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 <code>aParaLevel</code> can be set to
* either <code>NSBIDI_DEFAULT_LTR</code>
* or <code>NSBIDI_DEFAULT_RTL</code>;
* 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 <code>NSBIDI_MAX_EXPLICIT_LEVEL</code> is also valid,
* with odd levels indicating RTL.
*/
nsresult SetPara(const char16_t *aText, int32_t aLength, nsBidiLevel aParaLevel);
/**
* Get the directionality of the text.
*
* @param aDirection receives a <code>NSBIDI_XXX</code> 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 paragraph level of the text.
*
* @param aParaLevel receives a <code>NSBIDI_XXX</code> value indicating the paragraph level
*
* @see nsBidiLevel
*/
nsresult GetParaLevel(nsBidiLevel* aParaLevel);
/**
* Get a logical run.
* This function returns information about a run and is used
* to retrieve runs in logical order.<p>
* 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
* <code>aLogicalStart</code>.
* This pointer can be <code>nullptr</code> if this
* value is not necessary.
*
* @param aLevel will receive the level of the run.
* This pointer can be <code>nullptr</code> if this
* value is not necessary.
*/
nsresult GetLogicalRun(int32_t aLogicalStart, int32_t* aLogicalLimit, nsBidiLevel* aLevel);
/**
* Get the number of runs.
* This function may invoke the actual reordering on the
* <code>nsBidi</code> object, after <code>SetPara</code>
* may have resolved only the levels of the text. Therefore,
* <code>CountRuns</code> may have to allocate memory,
* and may fail doing so.
*
* @param aRunCount will receive the number of runs.
*/
nsresult CountRuns(int32_t* 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.<p>
* <code>CountRuns</code> should be called
* before the runs are retrieved.
*
* @param aRunIndex is the number of the run in visual order, in the
* range <code>[0..CountRuns-1]</code>.
*
* @param aLogicalStart is the first logical character index in the text.
* The pointer may be <code>nullptr</code> if this index is not needed.
*
* @param aLength is the number of characters (at least one) in the run.
* The pointer may be <code>nullptr</code> if this is not needed.
*
* @param aDirection will receive the directionality of the run,
* <code>NSBIDI_LTR==0</code> or <code>NSBIDI_RTL==1</code>,
* never <code>NSBIDI_MIXED</code>.
*
* @see CountRuns<p>
*
* Example:
* @code
* int32_t i, count, logicalStart, visualIndex=0, length;
* nsBidiDirection dir;
* pBidi->CountRuns(&count);
* for(i=0; i<count; ++i) {
* pBidi->GetVisualRun(i, &logicalStart, &length, &dir);
* if(NSBIDI_LTR==dir) {
* do { // LTR
* show_char(text[logicalStart++], visualIndex++);
* } while(--length>0);
* } else {
* logicalStart+=length; // logicalLimit
* do { // RTL
* show_char(text[--logicalStart], visualIndex++);
* } while(--length>0);
* }
* }
* @endcode
*
* Note that in right-to-left runs, code like this places
* modifier letters before base characters and second surrogates
* before first ones.
*/
nsresult GetVisualRun(int32_t aRunIndex, int32_t* aLogicalStart, int32_t* aLength, nsBidiDirection* aDirection);
/**
* 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 <code>GetVisualMap</code> on a
* <code>nsBidi</code> object.
*
* @param aLevels is an array with <code>aLength</code> 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 <code>aLength>0</code>.
*
* @param aIndexMap is a pointer to an array of <code>aLength</code>
* indexes which will reflect the reordering of the characters.
* The array does not need to be initialized.<p>
* The index map will result in <code>aIndexMap[aVisualIndex]==aLogicalIndex</code>.
*/
static nsresult ReorderVisual(const nsBidiLevel *aLevels, int32_t aLength, int32_t *aIndexMap);
/**
* 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 <code>aDestSize</code>
* and <code>aOptions</code> parameters and of the option bit flags.
*
* Since no Bidi controls are inserted here, this function will never
* write more than <code>aSrcLength</code> characters to <code>aDest</code>.
*
* @param aSrc A pointer to the RTL run text.
*
* @param aSrcLength The length of the RTL run.
* If the <code>NSBIDI_REMOVE_BIDI_CONTROLS</code> option
* is set, then the destination length may be less than
* <code>aSrcLength</code>.
* If this option is not set, then the destination length
* will be exactly <code>aSrcLength</code>.
*
* @param aDest A pointer to where the reordered text is to be copied.
* <code>aSrc[aSrcLength]</code> and <code>aDest[aSrcLength]</code>
* 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 <code>aDest</code>.
*/
nsresult WriteReverse(const char16_t *aSrc, int32_t aSrcLength, char16_t *aDest, uint16_t aOptions, int32_t *aDestSize);
protected:
friend class nsBidiPresUtils;
class BracketData {
public:
explicit BracketData(const nsBidi* aBidi);
~BracketData();
void ProcessBoundary(int32_t aLastDirControlCharPos,
nsBidiLevel aContextLevel,
nsBidiLevel aEmbeddingLevel,
const DirProp* aDirProps);
void ProcessLRI_RLI(nsBidiLevel aLevel);
void ProcessPDI();
bool AddOpening(char16_t aMatch, int32_t aPosition);
void FixN0c(int32_t aOpeningIndex, int32_t aNewPropPosition,
DirProp aNewProp, DirProp* aDirProps);
DirProp ProcessClosing(int32_t aOpenIdx, int32_t aPosition,
DirProp* aDirProps);
bool ProcessChar(int32_t aPosition, char16_t aCh, DirProp* aDirProps,
nsBidiLevel* aLevels);
private:
// array of opening entries which should be enough in most cases;
// no malloc() needed
Opening mSimpleOpenings[SIMPLE_OPENINGS_COUNT];
Opening* mOpenings; // pointer to current array of entries,
// either mSimpleOpenings or malloced array
Opening* mOpeningsMemory;
size_t mOpeningsSize;
// array of nested isolated sequence entries; can never exceed
// UBIDI_MAX_EXPLICIT_LEVEL
// + 1 for index 0
// + 1 for before the first isolated sequence
IsoRun mIsoRuns[NSBIDI_MAX_EXPLICIT_LEVEL+2];
int32_t mIsoRunLast; // index of last used entry in mIsoRuns
int32_t mOpeningsCount; // number of allocated entries in mOpenings
};
/** length of the current text */
int32_t mLength;
/** memory sizes in bytes */
size_t mDirPropsSize, mLevelsSize, mRunsSize;
size_t mIsolatesSize;
/** allocated memory */
DirProp* mDirPropsMemory;
nsBidiLevel* mLevelsMemory;
Run* mRunsMemory;
Isolate* mIsolatesMemory;
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 */
int32_t mTrailingWSStart;
/** fields for line reordering */
int32_t 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];
/* maxium of current nesting depth of isolate sequences */
/* Within ResolveExplicitLevels() and checkExpicitLevels(), this is the maximal
nesting encountered.
Within ResolveImplicitLevels(), this is the index of the current isolates
stack entry. */
int32_t mIsolateCount;
Isolate* mIsolates;
/** for simple text, have a small stack (no malloc()) */
Isolate mSimpleIsolates[SIMPLE_ISOLATES_SIZE];
private:
void Init();
static bool GetMemory(void **aMemory, size_t* aSize, size_t aSizeNeeded);
void Free();
void GetDirProps(const char16_t *aText);
void ResolveExplicitLevels(nsBidiDirection *aDirection, const char16_t *aText);
nsBidiDirection DirectionFromFlags(Flags aFlags);
void ProcessPropertySeq(LevState *pLevState, uint8_t _prop, int32_t start, int32_t limit);
void ResolveImplicitLevels(int32_t aStart, int32_t aLimit, DirProp aSOR, DirProp aEOR);
void AdjustWSLevels();
void SetTrailingWSStart();
bool GetRuns();
void GetSingleRun(nsBidiLevel aLevel);
void ReorderLine(nsBidiLevel aMinLevel, nsBidiLevel aMaxLevel);
static bool PrepareReorder(const nsBidiLevel *aLevels, int32_t aLength, int32_t *aIndexMap, nsBidiLevel *aMinLevel, nsBidiLevel *aMaxLevel);
};
#endif // _nsBidi_noICU_h_