mirror of
https://github.com/mozilla/gecko-dev.git
synced 2024-11-24 13:21:05 +00:00
Bug 1305700 - pt 3 & 4 - Clean up/simplify use of ENABLE_INTL_API conditionals in nsUnicodeProperties (code rearrangement, no change in behavior). r=m_kato
This commit is contained in:
Jonathan Kew
parent
7f21325a4a
commit
594fdb205d
@ -10,11 +10,6 @@
|
||||
#include "mozilla/ArrayUtils.h"
|
||||
#include "nsCharTraits.h"
|
||||
|
||||
#if ENABLE_INTL_API
|
||||
#include "unicode/uchar.h"
|
||||
#include "unicode/uscript.h"
|
||||
#endif
|
||||
|
||||
#define UNICODE_BMP_LIMIT 0x10000
|
||||
#define UNICODE_LIMIT 0x110000
|
||||
|
||||
@ -173,121 +168,70 @@ const hb_unicode_general_category_t sICUtoHBcategory[U_CHAR_CATEGORY_COUNT] = {
|
||||
};
|
||||
#endif
|
||||
|
||||
#if !ENABLE_INTL_API
|
||||
uint8_t GetGeneralCategory(uint32_t aCh) {
|
||||
#if ENABLE_INTL_API
|
||||
return sICUtoHBcategory[u_charType(aCh)];
|
||||
#else
|
||||
return GetCharProps2(aCh).mCategory;
|
||||
#endif
|
||||
}
|
||||
|
||||
nsCharType GetBidiCat(uint32_t aCh) {
|
||||
#if ENABLE_INTL_API
|
||||
return nsCharType(u_charDirection(aCh));
|
||||
#else
|
||||
return nsCharType(GetCharProps2(aCh).mBidiCategory);
|
||||
#endif
|
||||
}
|
||||
|
||||
int8_t GetNumericValue(uint32_t aCh) {
|
||||
#if ENABLE_INTL_API
|
||||
UNumericType type =
|
||||
UNumericType(u_getIntPropertyValue(aCh, UCHAR_NUMERIC_TYPE));
|
||||
return type == U_NT_DECIMAL || type == U_NT_DIGIT
|
||||
? int8_t(u_getNumericValue(aCh))
|
||||
: -1;
|
||||
#else
|
||||
return GetCharProps2(aCh).mNumericValue;
|
||||
#endif
|
||||
}
|
||||
|
||||
uint32_t
|
||||
GetMirroredChar(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_charMirror(aCh);
|
||||
#else
|
||||
return aCh + sMirrorOffsets[GetCharProps1(aCh).mMirrorOffsetIndex];
|
||||
#endif
|
||||
}
|
||||
|
||||
bool
|
||||
HasMirroredChar(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_isMirrored(aCh);
|
||||
#else
|
||||
return GetCharProps1(aCh).mMirrorOffsetIndex != 0;
|
||||
#endif
|
||||
}
|
||||
|
||||
uint8_t
|
||||
GetCombiningClass(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_getCombiningClass(aCh);
|
||||
#else
|
||||
return GetCharProps1(aCh).mCombiningClass;
|
||||
#endif
|
||||
}
|
||||
|
||||
uint8_t
|
||||
GetLineBreakClass(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_getIntPropertyValue(aCh, UCHAR_LINE_BREAK);
|
||||
#else
|
||||
return GetCharProps2(aCh).mLineBreak;
|
||||
#endif
|
||||
}
|
||||
|
||||
Script
|
||||
GetScriptCode(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
UErrorCode err = U_ZERO_ERROR;
|
||||
return Script(uscript_getScript(aCh, &err));
|
||||
#else
|
||||
return Script(GetCharProps2(aCh).mScriptCode);
|
||||
#endif
|
||||
}
|
||||
|
||||
uint32_t
|
||||
GetScriptTagForCode(Script aScriptCode)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
const char* tag = uscript_getShortName(UScriptCode(aScriptCode));
|
||||
return HB_TAG(tag[0], tag[1], tag[2], tag[3]);
|
||||
#else
|
||||
// this will safely return 0 for negative script codes, too :)
|
||||
if (static_cast<uint32_t>(aScriptCode) > ArrayLength(sScriptCodeToTag)) {
|
||||
return 0;
|
||||
}
|
||||
return sScriptCodeToTag[static_cast<uint32_t>(aScriptCode)];
|
||||
#endif
|
||||
}
|
||||
|
||||
PairedBracketType GetPairedBracketType(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return PairedBracketType
|
||||
(u_getIntPropertyValue(aCh, UCHAR_BIDI_PAIRED_BRACKET_TYPE));
|
||||
#else
|
||||
return PairedBracketType(GetCharProps2(aCh).mPairedBracketType);
|
||||
#endif
|
||||
}
|
||||
|
||||
uint32_t GetPairedBracket(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_getBidiPairedBracket(aCh);
|
||||
#else
|
||||
return GetPairedBracketType(aCh) != PAIRED_BRACKET_TYPE_NONE
|
||||
? GetMirroredChar(aCh) : aCh;
|
||||
#endif
|
||||
}
|
||||
|
||||
#if !ENABLE_INTL_API
|
||||
static inline uint32_t
|
||||
GetCaseMapValue(uint32_t aCh)
|
||||
{
|
||||
@ -302,14 +246,10 @@ GetCaseMapValue(uint32_t aCh)
|
||||
}
|
||||
return 0;
|
||||
}
|
||||
#endif
|
||||
|
||||
uint32_t
|
||||
GetUppercase(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_toupper(aCh);
|
||||
#else
|
||||
uint32_t mapValue = GetCaseMapValue(aCh);
|
||||
if (mapValue & (kLowerToUpper | kTitleToUpper)) {
|
||||
return aCh ^ (mapValue & kCaseMapCharMask);
|
||||
@ -318,15 +258,11 @@ GetUppercase(uint32_t aCh)
|
||||
return GetUppercase(aCh ^ (mapValue & kCaseMapCharMask));
|
||||
}
|
||||
return aCh;
|
||||
#endif
|
||||
}
|
||||
|
||||
uint32_t
|
||||
GetLowercase(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_tolower(aCh);
|
||||
#else
|
||||
uint32_t mapValue = GetCaseMapValue(aCh);
|
||||
if (mapValue & kUpperToLower) {
|
||||
return aCh ^ (mapValue & kCaseMapCharMask);
|
||||
@ -335,29 +271,21 @@ GetLowercase(uint32_t aCh)
|
||||
return GetLowercase(aCh ^ (mapValue & kCaseMapCharMask));
|
||||
}
|
||||
return aCh;
|
||||
#endif
|
||||
}
|
||||
|
||||
uint32_t
|
||||
GetTitlecaseForLower(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_isULowercase(aCh) ? u_totitle(aCh) : aCh;
|
||||
#else
|
||||
uint32_t mapValue = GetCaseMapValue(aCh);
|
||||
if (mapValue & (kLowerToTitle | kLowerToUpper)) {
|
||||
return aCh ^ (mapValue & kCaseMapCharMask);
|
||||
}
|
||||
return aCh;
|
||||
#endif
|
||||
}
|
||||
|
||||
uint32_t
|
||||
GetTitlecaseForAll(uint32_t aCh)
|
||||
{
|
||||
#if ENABLE_INTL_API
|
||||
return u_totitle(aCh);
|
||||
#else
|
||||
uint32_t mapValue = GetCaseMapValue(aCh);
|
||||
if (mapValue & (kLowerToTitle | kLowerToUpper)) {
|
||||
return aCh ^ (mapValue & kCaseMapCharMask);
|
||||
@ -366,27 +294,6 @@ GetTitlecaseForAll(uint32_t aCh)
|
||||
return GetTitlecaseForLower(aCh ^ (mapValue & kCaseMapCharMask));
|
||||
}
|
||||
return aCh;
|
||||
#endif
|
||||
}
|
||||
|
||||
#if 0 // currently unused - bug 857481
|
||||
HanVariantType
|
||||
GetHanVariant(uint32_t aCh)
|
||||
{
|
||||
// In the sHanVariantValues array, data for 4 successive characters
|
||||
// (2 bits each) is packed in to each uint8_t entry, with the value
|
||||
// for the lowest character stored in the least significant bits.
|
||||
uint8_t v = 0;
|
||||
if (aCh < UNICODE_BMP_LIMIT) {
|
||||
v = sHanVariantValues[sHanVariantPages[0][aCh >> kHanVariantCharBits]]
|
||||
[(aCh & ((1 << kHanVariantCharBits) - 1)) >> 2];
|
||||
} else if (aCh < (kHanVariantMaxPlane + 1) * 0x10000) {
|
||||
v = sHanVariantValues[sHanVariantPages[sHanVariantPlanes[(aCh >> 16) - 1]]
|
||||
[(aCh & 0xffff) >> kHanVariantCharBits]]
|
||||
[(aCh & ((1 << kHanVariantCharBits) - 1)) >> 2];
|
||||
}
|
||||
// extract the appropriate 2-bit field from the value
|
||||
return HanVariantType((v >> ((aCh & 3) * 2)) & 3);
|
||||
}
|
||||
#endif
|
||||
|
||||
|
||||
@ -10,6 +10,12 @@
|
||||
#include "nsBidiUtils.h"
|
||||
#include "nsIUGenCategory.h"
|
||||
#include "nsUnicodeScriptCodes.h"
|
||||
#include "harfbuzz/hb.h"
|
||||
|
||||
#if ENABLE_INTL_API
|
||||
#include "unicode/uchar.h"
|
||||
#include "unicode/uscript.h"
|
||||
#endif
|
||||
|
||||
const nsCharProps2& GetCharProps2(uint32_t aCh);
|
||||
|
||||
@ -19,29 +25,6 @@ namespace unicode {
|
||||
|
||||
extern const nsIUGenCategory::nsUGenCategory sDetailedToGeneralCategory[];
|
||||
|
||||
// Return whether the char has a mirrored-pair counterpart.
|
||||
uint32_t GetMirroredChar(uint32_t aCh);
|
||||
|
||||
bool HasMirroredChar(uint32_t aChr);
|
||||
|
||||
uint8_t GetCombiningClass(uint32_t aCh);
|
||||
|
||||
// returns the detailed General Category in terms of HB_UNICODE_* values
|
||||
uint8_t GetGeneralCategory(uint32_t aCh);
|
||||
|
||||
// returns the simplified Gen Category as defined in nsIUGenCategory
|
||||
inline nsIUGenCategory::nsUGenCategory GetGenCategory(uint32_t aCh) {
|
||||
return sDetailedToGeneralCategory[GetGeneralCategory(aCh)];
|
||||
}
|
||||
|
||||
nsCharType GetBidiCat(uint32_t aCh);
|
||||
|
||||
uint8_t GetLineBreakClass(uint32_t aCh);
|
||||
|
||||
Script GetScriptCode(uint32_t aCh);
|
||||
|
||||
uint32_t GetScriptTagForCode(Script aScriptCode);
|
||||
|
||||
/* This MUST match the values assigned by genUnicodePropertyData.pl! */
|
||||
enum VerticalOrientation {
|
||||
VERTICAL_ORIENTATION_U = 0,
|
||||
@ -50,10 +33,6 @@ enum VerticalOrientation {
|
||||
VERTICAL_ORIENTATION_Tr = 3
|
||||
};
|
||||
|
||||
inline VerticalOrientation GetVerticalOrientation(uint32_t aCh) {
|
||||
return VerticalOrientation(GetCharProps2(aCh).mVertOrient);
|
||||
}
|
||||
|
||||
/* This MUST match the values assigned by genUnicodePropertyData.pl! */
|
||||
enum PairedBracketType {
|
||||
PAIRED_BRACKET_TYPE_NONE = 0,
|
||||
@ -61,9 +40,6 @@ enum PairedBracketType {
|
||||
PAIRED_BRACKET_TYPE_CLOSE = 2
|
||||
};
|
||||
|
||||
PairedBracketType GetPairedBracketType(uint32_t aCh);
|
||||
uint32_t GetPairedBracket(uint32_t aCh);
|
||||
|
||||
enum XidmodType {
|
||||
XIDMOD_RECOMMENDED,
|
||||
XIDMOD_INCLUSION,
|
||||
@ -80,10 +56,129 @@ enum XidmodType {
|
||||
XIDMOD_NOT_CHARS
|
||||
};
|
||||
|
||||
inline XidmodType GetIdentifierModification(uint32_t aCh) {
|
||||
return XidmodType(GetCharProps2(aCh).mXidmod);
|
||||
#if ENABLE_INTL_API // ICU is available, so simply forward to its API
|
||||
|
||||
extern const hb_unicode_general_category_t sICUtoHBcategory[];
|
||||
|
||||
inline uint32_t
|
||||
GetMirroredChar(uint32_t aCh)
|
||||
{
|
||||
return u_charMirror(aCh);
|
||||
}
|
||||
|
||||
inline bool
|
||||
HasMirroredChar(uint32_t aCh)
|
||||
{
|
||||
return u_isMirrored(aCh);
|
||||
}
|
||||
|
||||
inline uint8_t
|
||||
GetCombiningClass(uint32_t aCh)
|
||||
{
|
||||
return u_getCombiningClass(aCh);
|
||||
}
|
||||
|
||||
inline uint8_t
|
||||
GetGeneralCategory(uint32_t aCh)
|
||||
{
|
||||
return sICUtoHBcategory[u_charType(aCh)];
|
||||
}
|
||||
|
||||
inline nsCharType
|
||||
GetBidiCat(uint32_t aCh)
|
||||
{
|
||||
return nsCharType(u_charDirection(aCh));
|
||||
}
|
||||
|
||||
inline int8_t
|
||||
GetNumericValue(uint32_t aCh)
|
||||
{
|
||||
UNumericType type =
|
||||
UNumericType(u_getIntPropertyValue(aCh, UCHAR_NUMERIC_TYPE));
|
||||
return type == U_NT_DECIMAL || type == U_NT_DIGIT
|
||||
? int8_t(u_getNumericValue(aCh)) : -1;
|
||||
}
|
||||
|
||||
inline uint8_t
|
||||
GetLineBreakClass(uint32_t aCh)
|
||||
{
|
||||
return u_getIntPropertyValue(aCh, UCHAR_LINE_BREAK);
|
||||
}
|
||||
|
||||
inline Script
|
||||
GetScriptCode(uint32_t aCh)
|
||||
{
|
||||
UErrorCode err = U_ZERO_ERROR;
|
||||
return Script(uscript_getScript(aCh, &err));
|
||||
}
|
||||
|
||||
inline uint32_t
|
||||
GetScriptTagForCode(Script aScriptCode)
|
||||
{
|
||||
const char* tag = uscript_getShortName(UScriptCode(aScriptCode));
|
||||
return HB_TAG(tag[0], tag[1], tag[2], tag[3]);
|
||||
}
|
||||
|
||||
inline PairedBracketType
|
||||
GetPairedBracketType(uint32_t aCh)
|
||||
{
|
||||
return PairedBracketType
|
||||
(u_getIntPropertyValue(aCh, UCHAR_BIDI_PAIRED_BRACKET_TYPE));
|
||||
}
|
||||
|
||||
inline uint32_t
|
||||
GetPairedBracket(uint32_t aCh)
|
||||
{
|
||||
return u_getBidiPairedBracket(aCh);
|
||||
}
|
||||
|
||||
inline uint32_t
|
||||
GetUppercase(uint32_t aCh)
|
||||
{
|
||||
return u_toupper(aCh);
|
||||
}
|
||||
|
||||
inline uint32_t
|
||||
GetLowercase(uint32_t aCh)
|
||||
{
|
||||
return u_tolower(aCh);
|
||||
}
|
||||
|
||||
inline uint32_t
|
||||
GetTitlecaseForLower(uint32_t aCh) // maps LC to titlecase, UC unchanged
|
||||
{
|
||||
return u_isULowercase(aCh) ? u_totitle(aCh) : aCh;
|
||||
}
|
||||
|
||||
inline uint32_t
|
||||
GetTitlecaseForAll(uint32_t aCh) // maps both UC and LC to titlecase
|
||||
{
|
||||
return u_totitle(aCh);
|
||||
}
|
||||
|
||||
#else // not ENABLE_INTL_API
|
||||
|
||||
// Return whether the char has a mirrored-pair counterpart.
|
||||
uint32_t GetMirroredChar(uint32_t aCh);
|
||||
|
||||
bool HasMirroredChar(uint32_t aChr);
|
||||
|
||||
uint8_t GetCombiningClass(uint32_t aCh);
|
||||
|
||||
// returns the detailed General Category in terms of HB_UNICODE_* values
|
||||
uint8_t GetGeneralCategory(uint32_t aCh);
|
||||
|
||||
nsCharType GetBidiCat(uint32_t aCh);
|
||||
|
||||
uint8_t GetLineBreakClass(uint32_t aCh);
|
||||
|
||||
Script GetScriptCode(uint32_t aCh);
|
||||
|
||||
uint32_t GetScriptTagForCode(Script aScriptCode);
|
||||
|
||||
PairedBracketType GetPairedBracketType(uint32_t aCh);
|
||||
uint32_t GetPairedBracket(uint32_t aCh);
|
||||
|
||||
/**
|
||||
* Return the numeric value of the character. The value returned is the value
|
||||
* of the Numeric_Value in field 7 of the UCD, or -1 if field 7 is empty.
|
||||
@ -92,16 +187,25 @@ inline XidmodType GetIdentifierModification(uint32_t aCh) {
|
||||
*/
|
||||
int8_t GetNumericValue(uint32_t aCh);
|
||||
|
||||
#if 0 // currently unused - bug 857481
|
||||
enum HanVariantType {
|
||||
HVT_NotHan = 0x0,
|
||||
HVT_SimplifiedOnly = 0x1,
|
||||
HVT_TraditionalOnly = 0x2,
|
||||
HVT_AnyHan = 0x3
|
||||
};
|
||||
uint32_t GetUppercase(uint32_t aCh);
|
||||
uint32_t GetLowercase(uint32_t aCh);
|
||||
uint32_t GetTitlecaseForLower(uint32_t aCh); // maps LC to titlecase, UC unchanged
|
||||
uint32_t GetTitlecaseForAll(uint32_t aCh); // maps both UC and LC to titlecase
|
||||
|
||||
HanVariantType GetHanVariant(uint32_t aCh);
|
||||
#endif
|
||||
#endif // !ENABLE_INTL_API
|
||||
|
||||
// returns the simplified Gen Category as defined in nsIUGenCategory
|
||||
inline nsIUGenCategory::nsUGenCategory GetGenCategory(uint32_t aCh) {
|
||||
return sDetailedToGeneralCategory[GetGeneralCategory(aCh)];
|
||||
}
|
||||
|
||||
inline VerticalOrientation GetVerticalOrientation(uint32_t aCh) {
|
||||
return VerticalOrientation(GetCharProps2(aCh).mVertOrient);
|
||||
}
|
||||
|
||||
inline XidmodType GetIdentifierModification(uint32_t aCh) {
|
||||
return XidmodType(GetCharProps2(aCh).mXidmod);
|
||||
}
|
||||
|
||||
uint32_t GetFullWidth(uint32_t aCh);
|
||||
// This is the reverse function of GetFullWidth which guarantees that
|
||||
@ -116,14 +220,6 @@ inline bool IsClusterExtender(uint32_t aCh) {
|
||||
return IsClusterExtender(aCh, GetGeneralCategory(aCh));
|
||||
}
|
||||
|
||||
// Case mappings for the full Unicode range;
|
||||
// note that it may be worth testing for ASCII chars and taking
|
||||
// a separate fast-path before calling these, in perf-critical places
|
||||
uint32_t GetUppercase(uint32_t aCh);
|
||||
uint32_t GetLowercase(uint32_t aCh);
|
||||
uint32_t GetTitlecaseForLower(uint32_t aCh); // maps LC to titlecase, UC unchanged
|
||||
uint32_t GetTitlecaseForAll(uint32_t aCh); // maps both UC and LC to titlecase
|
||||
|
||||
// A simple iterator for a string of char16_t codepoints that advances
|
||||
// by Unicode grapheme clusters
|
||||
class ClusterIterator
|
||||
|
||||
@ -214,6 +214,7 @@ typedef uint8_t DirProp;
|
||||
#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 --------------------------------------- */
|
||||
@ -338,6 +339,8 @@ typedef uint8_t DirProp;
|
||||
#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;
|
||||
|
||||
Loading…
Reference in New Issue
Block a user