Issue #1971 - Part 2: Update ICU source to 63.2.

This commit is contained in:
Job Bautista 2022-07-24 21:03:27 +08:00 committed by roytam1
commit 1e69214382
3160 changed files with 275815 additions and 234203 deletions

View file

@ -1,4 +1,4 @@
// Copyright (C) 2016 and later: Unicode, Inc. and others.
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
@ -24,9 +24,11 @@
#include "unicode/format.h"
#include "unicode/locid.h"
#include "unicode/parseerr.h"
#include "unicode/strenum.h"
#include "unicode/ures.h"
#include "uvector.h"
#include "hash.h"
#include "uassert.h"
class PluralRulesTest;
@ -38,67 +40,73 @@ class DigitInterval;
class PluralRules;
class VisibleDigits;
static const UChar DOT = ((UChar)0x002E);
static const UChar SINGLE_QUOTE = ((UChar)0x0027);
static const UChar SLASH = ((UChar)0x002F);
static const UChar BACKSLASH = ((UChar)0x005C);
static const UChar SPACE = ((UChar)0x0020);
static const UChar EXCLAMATION = ((UChar)0x0021);
static const UChar QUOTATION_MARK = ((UChar)0x0022);
static const UChar NUMBER_SIGN = ((UChar)0x0023);
static const UChar PERCENT_SIGN = ((UChar)0x0025);
static const UChar ASTERISK = ((UChar)0x002A);
static const UChar COMMA = ((UChar)0x002C);
static const UChar HYPHEN = ((UChar)0x002D);
static const UChar U_ZERO = ((UChar)0x0030);
static const UChar U_ONE = ((UChar)0x0031);
static const UChar U_TWO = ((UChar)0x0032);
static const UChar U_THREE = ((UChar)0x0033);
static const UChar U_FOUR = ((UChar)0x0034);
static const UChar U_FIVE = ((UChar)0x0035);
static const UChar U_SIX = ((UChar)0x0036);
static const UChar U_SEVEN = ((UChar)0x0037);
static const UChar U_EIGHT = ((UChar)0x0038);
static const UChar U_NINE = ((UChar)0x0039);
static const UChar COLON = ((UChar)0x003A);
static const UChar SEMI_COLON = ((UChar)0x003B);
static const UChar EQUALS = ((UChar)0x003D);
static const UChar AT = ((UChar)0x0040);
static const UChar CAP_A = ((UChar)0x0041);
static const UChar CAP_B = ((UChar)0x0042);
static const UChar CAP_R = ((UChar)0x0052);
static const UChar CAP_Z = ((UChar)0x005A);
static const UChar LOWLINE = ((UChar)0x005F);
static const UChar LEFTBRACE = ((UChar)0x007B);
static const UChar RIGHTBRACE = ((UChar)0x007D);
static const UChar TILDE = ((UChar)0x007E);
static const UChar ELLIPSIS = ((UChar)0x2026);
namespace pluralimpl {
static const UChar LOW_A = ((UChar)0x0061);
static const UChar LOW_B = ((UChar)0x0062);
static const UChar LOW_C = ((UChar)0x0063);
static const UChar LOW_D = ((UChar)0x0064);
static const UChar LOW_E = ((UChar)0x0065);
static const UChar LOW_F = ((UChar)0x0066);
static const UChar LOW_G = ((UChar)0x0067);
static const UChar LOW_H = ((UChar)0x0068);
static const UChar LOW_I = ((UChar)0x0069);
static const UChar LOW_J = ((UChar)0x006a);
static const UChar LOW_K = ((UChar)0x006B);
static const UChar LOW_L = ((UChar)0x006C);
static const UChar LOW_M = ((UChar)0x006D);
static const UChar LOW_N = ((UChar)0x006E);
static const UChar LOW_O = ((UChar)0x006F);
static const UChar LOW_P = ((UChar)0x0070);
static const UChar LOW_Q = ((UChar)0x0071);
static const UChar LOW_R = ((UChar)0x0072);
static const UChar LOW_S = ((UChar)0x0073);
static const UChar LOW_T = ((UChar)0x0074);
static const UChar LOW_U = ((UChar)0x0075);
static const UChar LOW_V = ((UChar)0x0076);
static const UChar LOW_W = ((UChar)0x0077);
static const UChar LOW_Y = ((UChar)0x0079);
static const UChar LOW_Z = ((UChar)0x007A);
// TODO: Remove this and replace with u"" literals. Was for EBCDIC compatibility.
static const UChar DOT = ((UChar) 0x002E);
static const UChar SINGLE_QUOTE = ((UChar) 0x0027);
static const UChar SLASH = ((UChar) 0x002F);
static const UChar BACKSLASH = ((UChar) 0x005C);
static const UChar SPACE = ((UChar) 0x0020);
static const UChar EXCLAMATION = ((UChar) 0x0021);
static const UChar QUOTATION_MARK = ((UChar) 0x0022);
static const UChar NUMBER_SIGN = ((UChar) 0x0023);
static const UChar PERCENT_SIGN = ((UChar) 0x0025);
static const UChar ASTERISK = ((UChar) 0x002A);
static const UChar COMMA = ((UChar) 0x002C);
static const UChar HYPHEN = ((UChar) 0x002D);
static const UChar U_ZERO = ((UChar) 0x0030);
static const UChar U_ONE = ((UChar) 0x0031);
static const UChar U_TWO = ((UChar) 0x0032);
static const UChar U_THREE = ((UChar) 0x0033);
static const UChar U_FOUR = ((UChar) 0x0034);
static const UChar U_FIVE = ((UChar) 0x0035);
static const UChar U_SIX = ((UChar) 0x0036);
static const UChar U_SEVEN = ((UChar) 0x0037);
static const UChar U_EIGHT = ((UChar) 0x0038);
static const UChar U_NINE = ((UChar) 0x0039);
static const UChar COLON = ((UChar) 0x003A);
static const UChar SEMI_COLON = ((UChar) 0x003B);
static const UChar EQUALS = ((UChar) 0x003D);
static const UChar AT = ((UChar) 0x0040);
static const UChar CAP_A = ((UChar) 0x0041);
static const UChar CAP_B = ((UChar) 0x0042);
static const UChar CAP_R = ((UChar) 0x0052);
static const UChar CAP_Z = ((UChar) 0x005A);
static const UChar LOWLINE = ((UChar) 0x005F);
static const UChar LEFTBRACE = ((UChar) 0x007B);
static const UChar RIGHTBRACE = ((UChar) 0x007D);
static const UChar TILDE = ((UChar) 0x007E);
static const UChar ELLIPSIS = ((UChar) 0x2026);
static const UChar LOW_A = ((UChar) 0x0061);
static const UChar LOW_B = ((UChar) 0x0062);
static const UChar LOW_C = ((UChar) 0x0063);
static const UChar LOW_D = ((UChar) 0x0064);
static const UChar LOW_E = ((UChar) 0x0065);
static const UChar LOW_F = ((UChar) 0x0066);
static const UChar LOW_G = ((UChar) 0x0067);
static const UChar LOW_H = ((UChar) 0x0068);
static const UChar LOW_I = ((UChar) 0x0069);
static const UChar LOW_J = ((UChar) 0x006a);
static const UChar LOW_K = ((UChar) 0x006B);
static const UChar LOW_L = ((UChar) 0x006C);
static const UChar LOW_M = ((UChar) 0x006D);
static const UChar LOW_N = ((UChar) 0x006E);
static const UChar LOW_O = ((UChar) 0x006F);
static const UChar LOW_P = ((UChar) 0x0070);
static const UChar LOW_Q = ((UChar) 0x0071);
static const UChar LOW_R = ((UChar) 0x0072);
static const UChar LOW_S = ((UChar) 0x0073);
static const UChar LOW_T = ((UChar) 0x0074);
static const UChar LOW_U = ((UChar) 0x0075);
static const UChar LOW_V = ((UChar) 0x0076);
static const UChar LOW_W = ((UChar) 0x0077);
static const UChar LOW_Y = ((UChar) 0x0079);
static const UChar LOW_Z = ((UChar) 0x007A);
}
static const int32_t PLURAL_RANGE_HIGH = 0x7fffffff;
@ -173,7 +181,77 @@ private:
kRangeList,
kSamples
};
};
enum PluralOperand {
/**
* The double value of the entire number.
*/
PLURAL_OPERAND_N,
/**
* The integer value, with the fraction digits truncated off.
*/
PLURAL_OPERAND_I,
/**
* All visible fraction digits as an integer, including trailing zeros.
*/
PLURAL_OPERAND_F,
/**
* Visible fraction digits as an integer, not including trailing zeros.
*/
PLURAL_OPERAND_T,
/**
* Number of visible fraction digits.
*/
PLURAL_OPERAND_V,
/**
* Number of visible fraction digits, not including trailing zeros.
*/
PLURAL_OPERAND_W,
/**
* THIS OPERAND IS DEPRECATED AND HAS BEEN REMOVED FROM THE SPEC.
*
* <p>Returns the integer value, but will fail if the number has fraction digits.
* That is, using "j" instead of "i" is like implicitly adding "v is 0".
*
* <p>For example, "j is 3" is equivalent to "i is 3 and v is 0": it matches
* "3" but not "3.1" or "3.0".
*/
PLURAL_OPERAND_J
};
/**
* Converts from the tokenType enum to PluralOperand. Asserts that the given
* tokenType can be mapped to a PluralOperand.
*/
PluralOperand tokenTypeToPluralOperand(tokenType tt);
/**
* An interface to FixedDecimal, allowing for other implementations.
* @internal
*/
class U_I18N_API IFixedDecimal {
public:
virtual ~IFixedDecimal();
/**
* Returns the value corresponding to the specified operand (n, i, f, t, v, or w).
* If the operand is 'n', returns a double; otherwise, returns an integer.
*/
virtual double getPluralOperand(PluralOperand operand) const = 0;
virtual bool isNaN() const = 0;
virtual bool isInfinite() const = 0;
/** Whether the number has no nonzero fraction digits. */
virtual bool hasIntegerValue() const = 0;
};
/**
@ -183,7 +261,7 @@ private:
* see DecimalFormat::getFixedDecimal()
* @internal
*/
class U_I18N_API FixedDecimal: public UMemory {
class U_I18N_API FixedDecimal: public IFixedDecimal, public UObject {
public:
/**
* @param n the number, e.g. 12.345
@ -193,12 +271,18 @@ class U_I18N_API FixedDecimal: public UMemory {
FixedDecimal(double n, int32_t v, int64_t f);
FixedDecimal(double n, int32_t);
explicit FixedDecimal(double n);
explicit FixedDecimal(const VisibleDigits &n);
FixedDecimal();
~FixedDecimal() U_OVERRIDE;
FixedDecimal(const UnicodeString &s, UErrorCode &ec);
FixedDecimal(const FixedDecimal &other);
double get(tokenType operand) const;
double getPluralOperand(PluralOperand operand) const U_OVERRIDE;
bool isNaN() const U_OVERRIDE;
bool isInfinite() const U_OVERRIDE;
bool hasIntegerValue() const U_OVERRIDE;
bool isNanOrInfinity() const; // used in decimfmtimpl.cpp
int32_t getVisibleFractionDigitCount() const;
void init(double n, int32_t v, int64_t f);
@ -214,9 +298,10 @@ class U_I18N_API FixedDecimal: public UMemory {
int64_t decimalDigits;
int64_t decimalDigitsWithoutTrailingZeros;
int64_t intValue;
UBool hasIntegerValue;
UBool _hasIntegerValue;
UBool isNegative;
UBool isNanOrInfinity;
UBool _isNaN;
UBool _isInfinite;
};
class AndConstraint : public UMemory {
@ -225,52 +310,57 @@ public:
NONE,
MOD
} RuleOp;
RuleOp op;
int32_t opNum; // for mod expressions, the right operand of the mod.
int32_t value; // valid for 'is' rules only.
UVector32 *rangeList; // for 'in', 'within' rules. Null otherwise.
UBool negated; // TRUE for negated rules.
UBool integerOnly; // TRUE for 'within' rules.
tokenType digitsType; // n | i | v | f constraint.
AndConstraint *next;
RuleOp op = AndConstraint::NONE;
int32_t opNum = -1; // for mod expressions, the right operand of the mod.
int32_t value = -1; // valid for 'is' rules only.
UVector32 *rangeList = nullptr; // for 'in', 'within' rules. Null otherwise.
UBool negated = FALSE; // TRUE for negated rules.
UBool integerOnly = FALSE; // TRUE for 'within' rules.
tokenType digitsType = none; // n | i | v | f constraint.
AndConstraint *next = nullptr;
// Internal error status, used for errors that occur during the copy constructor.
UErrorCode fInternalStatus = U_ZERO_ERROR;
AndConstraint();
AndConstraint() = default;
AndConstraint(const AndConstraint& other);
virtual ~AndConstraint();
AndConstraint* add();
AndConstraint* add(UErrorCode& status);
// UBool isFulfilled(double number);
UBool isFulfilled(const FixedDecimal &number);
UBool isFulfilled(const IFixedDecimal &number);
};
class OrConstraint : public UMemory {
public:
AndConstraint *childNode;
OrConstraint *next;
OrConstraint();
AndConstraint *childNode = nullptr;
OrConstraint *next = nullptr;
// Internal error status, used for errors that occur during the copy constructor.
UErrorCode fInternalStatus = U_ZERO_ERROR;
OrConstraint() = default;
OrConstraint(const OrConstraint& other);
virtual ~OrConstraint();
AndConstraint* add();
AndConstraint* add(UErrorCode& status);
// UBool isFulfilled(double number);
UBool isFulfilled(const FixedDecimal &number);
UBool isFulfilled(const IFixedDecimal &number);
};
class RuleChain : public UMemory {
public:
UnicodeString fKeyword;
RuleChain *fNext;
OrConstraint *ruleHeader;
RuleChain *fNext = nullptr;
OrConstraint *ruleHeader = nullptr;
UnicodeString fDecimalSamples; // Samples strings from rule source
UnicodeString fIntegerSamples; // without @decimal or @integer, otherwise unprocessed.
UBool fDecimalSamplesUnbounded;
UBool fIntegerSamplesUnbounded;
UBool fDecimalSamplesUnbounded = FALSE;
UBool fIntegerSamplesUnbounded = FALSE;
// Internal error status, used for errors that occur during the copy constructor.
UErrorCode fInternalStatus = U_ZERO_ERROR;
RuleChain();
RuleChain() = default;
RuleChain(const RuleChain& other);
virtual ~RuleChain();
UnicodeString select(const FixedDecimal &number) const;
UnicodeString select(const IFixedDecimal &number) const;
void dumpRules(UnicodeString& result);
UErrorCode getKeywords(int32_t maxArraySize, UnicodeString *keywords, int32_t& arraySize) const;
UBool isKeyword(const UnicodeString& keyword) const;
@ -300,8 +390,8 @@ class U_I18N_API PluralAvailableLocalesEnumeration: public StringEnumeration {
virtual int32_t count(UErrorCode& status) const;
private:
UErrorCode fOpenStatus;
UResourceBundle *fLocales;
UResourceBundle *fRes;
UResourceBundle *fLocales = nullptr;
UResourceBundle *fRes = nullptr;
};
U_NAMESPACE_END