1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/plurrule_impl.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,305 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +* Copyright (C) 2007-2013, International Business Machines Corporation and
1.7 +* others. All Rights Reserved.
1.8 +*******************************************************************************
1.9 +*
1.10 +* File PLURRULE_IMPL.H
1.11 +*
1.12 +*******************************************************************************
1.13 +*/
1.14 +
1.15 +
1.16 +#ifndef PLURRULE_IMPLE
1.17 +#define PLURRULE_IMPLE
1.18 +
1.19 +// Internal definitions for the PluralRules implementation.
1.20 +
1.21 +#if !UCONFIG_NO_FORMATTING
1.22 +
1.23 +#include "unicode/format.h"
1.24 +#include "unicode/locid.h"
1.25 +#include "unicode/parseerr.h"
1.26 +#include "unicode/ures.h"
1.27 +#include "unicode/utypes.h"
1.28 +#include "uvector.h"
1.29 +#include "hash.h"
1.30 +
1.31 +class PluralRulesTest;
1.32 +
1.33 +U_NAMESPACE_BEGIN
1.34 +
1.35 +class AndConstraint;
1.36 +class RuleChain;
1.37 +
1.38 +static const UChar DOT = ((UChar)0x002E);
1.39 +static const UChar SINGLE_QUOTE = ((UChar)0x0027);
1.40 +static const UChar SLASH = ((UChar)0x002F);
1.41 +static const UChar BACKSLASH = ((UChar)0x005C);
1.42 +static const UChar SPACE = ((UChar)0x0020);
1.43 +static const UChar EXCLAMATION = ((UChar)0x0021);
1.44 +static const UChar QUOTATION_MARK = ((UChar)0x0022);
1.45 +static const UChar NUMBER_SIGN = ((UChar)0x0023);
1.46 +static const UChar PERCENT_SIGN = ((UChar)0x0025);
1.47 +static const UChar ASTERISK = ((UChar)0x002A);
1.48 +static const UChar COMMA = ((UChar)0x002C);
1.49 +static const UChar HYPHEN = ((UChar)0x002D);
1.50 +static const UChar U_ZERO = ((UChar)0x0030);
1.51 +static const UChar U_ONE = ((UChar)0x0031);
1.52 +static const UChar U_TWO = ((UChar)0x0032);
1.53 +static const UChar U_THREE = ((UChar)0x0033);
1.54 +static const UChar U_FOUR = ((UChar)0x0034);
1.55 +static const UChar U_FIVE = ((UChar)0x0035);
1.56 +static const UChar U_SIX = ((UChar)0x0036);
1.57 +static const UChar U_SEVEN = ((UChar)0x0037);
1.58 +static const UChar U_EIGHT = ((UChar)0x0038);
1.59 +static const UChar U_NINE = ((UChar)0x0039);
1.60 +static const UChar COLON = ((UChar)0x003A);
1.61 +static const UChar SEMI_COLON = ((UChar)0x003B);
1.62 +static const UChar EQUALS = ((UChar)0x003D);
1.63 +static const UChar AT = ((UChar)0x0040);
1.64 +static const UChar CAP_A = ((UChar)0x0041);
1.65 +static const UChar CAP_B = ((UChar)0x0042);
1.66 +static const UChar CAP_R = ((UChar)0x0052);
1.67 +static const UChar CAP_Z = ((UChar)0x005A);
1.68 +static const UChar LOWLINE = ((UChar)0x005F);
1.69 +static const UChar LEFTBRACE = ((UChar)0x007B);
1.70 +static const UChar RIGHTBRACE = ((UChar)0x007D);
1.71 +static const UChar TILDE = ((UChar)0x007E);
1.72 +static const UChar ELLIPSIS = ((UChar)0x2026);
1.73 +
1.74 +static const UChar LOW_A = ((UChar)0x0061);
1.75 +static const UChar LOW_B = ((UChar)0x0062);
1.76 +static const UChar LOW_C = ((UChar)0x0063);
1.77 +static const UChar LOW_D = ((UChar)0x0064);
1.78 +static const UChar LOW_E = ((UChar)0x0065);
1.79 +static const UChar LOW_F = ((UChar)0x0066);
1.80 +static const UChar LOW_G = ((UChar)0x0067);
1.81 +static const UChar LOW_H = ((UChar)0x0068);
1.82 +static const UChar LOW_I = ((UChar)0x0069);
1.83 +static const UChar LOW_J = ((UChar)0x006a);
1.84 +static const UChar LOW_K = ((UChar)0x006B);
1.85 +static const UChar LOW_L = ((UChar)0x006C);
1.86 +static const UChar LOW_M = ((UChar)0x006D);
1.87 +static const UChar LOW_N = ((UChar)0x006E);
1.88 +static const UChar LOW_O = ((UChar)0x006F);
1.89 +static const UChar LOW_P = ((UChar)0x0070);
1.90 +static const UChar LOW_Q = ((UChar)0x0071);
1.91 +static const UChar LOW_R = ((UChar)0x0072);
1.92 +static const UChar LOW_S = ((UChar)0x0073);
1.93 +static const UChar LOW_T = ((UChar)0x0074);
1.94 +static const UChar LOW_U = ((UChar)0x0075);
1.95 +static const UChar LOW_V = ((UChar)0x0076);
1.96 +static const UChar LOW_W = ((UChar)0x0077);
1.97 +static const UChar LOW_Y = ((UChar)0x0079);
1.98 +static const UChar LOW_Z = ((UChar)0x007A);
1.99 +
1.100 +
1.101 +static const int32_t PLURAL_RANGE_HIGH = 0x7fffffff;
1.102 +
1.103 +enum tokenType {
1.104 + none,
1.105 + tNumber,
1.106 + tComma,
1.107 + tSemiColon,
1.108 + tSpace,
1.109 + tColon,
1.110 + tAt, // '@'
1.111 + tDot,
1.112 + tDot2,
1.113 + tEllipsis,
1.114 + tKeyword,
1.115 + tAnd,
1.116 + tOr,
1.117 + tMod, // 'mod' or '%'
1.118 + tNot, // 'not' only.
1.119 + tIn, // 'in' only.
1.120 + tEqual, // '=' only.
1.121 + tNotEqual, // '!='
1.122 + tTilde,
1.123 + tWithin,
1.124 + tIs,
1.125 + tVariableN,
1.126 + tVariableI,
1.127 + tVariableF,
1.128 + tVariableV,
1.129 + tVariableT,
1.130 + tDecimal,
1.131 + tInteger,
1.132 + tEOF
1.133 +};
1.134 +
1.135 +
1.136 +class PluralRuleParser: public UMemory {
1.137 +public:
1.138 + PluralRuleParser();
1.139 + virtual ~PluralRuleParser();
1.140 +
1.141 + void parse(const UnicodeString &rules, PluralRules *dest, UErrorCode &status);
1.142 + void getNextToken(UErrorCode &status);
1.143 + void checkSyntax(UErrorCode &status);
1.144 + static int32_t getNumberValue(const UnicodeString &token);
1.145 +
1.146 +private:
1.147 + static tokenType getKeyType(const UnicodeString& token, tokenType type);
1.148 + static tokenType charType(UChar ch);
1.149 + static UBool isValidKeyword(const UnicodeString& token);
1.150 +
1.151 + const UnicodeString *ruleSrc; // The rules string.
1.152 + int32_t ruleIndex; // String index in the input rules, the current parse position.
1.153 + UnicodeString token; // Token most recently scanned.
1.154 + tokenType type;
1.155 + tokenType prevType;
1.156 +
1.157 + // The items currently being parsed & built.
1.158 + // Note: currentChain may not be the last RuleChain in the
1.159 + // list because the "other" chain is forced to the end.
1.160 + AndConstraint *curAndConstraint;
1.161 + RuleChain *currentChain;
1.162 +
1.163 + int32_t rangeLowIdx; // Indices in the UVector of ranges of the
1.164 + int32_t rangeHiIdx; // low and hi values currently being parsed.
1.165 +
1.166 + enum EParseState {
1.167 + kKeyword,
1.168 + kExpr,
1.169 + kValue,
1.170 + kRangeList,
1.171 + kSamples
1.172 + };
1.173 +
1.174 +};
1.175 +
1.176 +/**
1.177 + * class FixedDecimal serves to communicate the properties
1.178 + * of a formatted number from a decimal formatter to PluralRules::select()
1.179 + *
1.180 + * see DecimalFormat::getFixedDecimal()
1.181 + * @internal
1.182 + */
1.183 +class U_I18N_API FixedDecimal: public UMemory {
1.184 + public:
1.185 + /**
1.186 + * @param n the number, e.g. 12.345
1.187 + * @param v The number of visible fraction digits, e.g. 3
1.188 + * @param f The fraction digits, e.g. 345
1.189 + */
1.190 + FixedDecimal(double n, int32_t v, int64_t f);
1.191 + FixedDecimal(double n, int32_t);
1.192 + explicit FixedDecimal(double n);
1.193 + FixedDecimal();
1.194 + FixedDecimal(const UnicodeString &s, UErrorCode &ec);
1.195 + FixedDecimal(const FixedDecimal &other);
1.196 +
1.197 + double get(tokenType operand) const;
1.198 + int32_t getVisibleFractionDigitCount() const;
1.199 +
1.200 + void init(double n, int32_t v, int64_t f);
1.201 + void init(double n);
1.202 + UBool quickInit(double n); // Try a fast-path only initialization,
1.203 + // return TRUE if successful.
1.204 + void adjustForMinFractionDigits(int32_t min);
1.205 + static int64_t getFractionalDigits(double n, int32_t v);
1.206 + static int32_t decimals(double n);
1.207 +
1.208 + double source;
1.209 + int32_t visibleDecimalDigitCount;
1.210 + int64_t decimalDigits;
1.211 + int64_t decimalDigitsWithoutTrailingZeros;
1.212 + int64_t intValue;
1.213 + UBool hasIntegerValue;
1.214 + UBool isNegative;
1.215 + UBool isNanOrInfinity;
1.216 +};
1.217 +
1.218 +class AndConstraint : public UMemory {
1.219 +public:
1.220 + typedef enum RuleOp {
1.221 + NONE,
1.222 + MOD
1.223 + } RuleOp;
1.224 + RuleOp op;
1.225 + int32_t opNum; // for mod expressions, the right operand of the mod.
1.226 + int32_t value; // valid for 'is' rules only.
1.227 + UVector32 *rangeList; // for 'in', 'within' rules. Null otherwise.
1.228 + UBool negated; // TRUE for negated rules.
1.229 + UBool integerOnly; // TRUE for 'within' rules.
1.230 + tokenType digitsType; // n | i | v | f constraint.
1.231 + AndConstraint *next;
1.232 +
1.233 + AndConstraint();
1.234 + AndConstraint(const AndConstraint& other);
1.235 + virtual ~AndConstraint();
1.236 + AndConstraint* add();
1.237 + // UBool isFulfilled(double number);
1.238 + UBool isFulfilled(const FixedDecimal &number);
1.239 +};
1.240 +
1.241 +class OrConstraint : public UMemory {
1.242 +public:
1.243 + AndConstraint *childNode;
1.244 + OrConstraint *next;
1.245 + OrConstraint();
1.246 +
1.247 + OrConstraint(const OrConstraint& other);
1.248 + virtual ~OrConstraint();
1.249 + AndConstraint* add();
1.250 + // UBool isFulfilled(double number);
1.251 + UBool isFulfilled(const FixedDecimal &number);
1.252 +};
1.253 +
1.254 +class RuleChain : public UMemory {
1.255 +public:
1.256 + UnicodeString fKeyword;
1.257 + RuleChain *fNext;
1.258 + OrConstraint *ruleHeader;
1.259 + UnicodeString fDecimalSamples; // Samples strings from rule source
1.260 + UnicodeString fIntegerSamples; // without @decimal or @integer, otherwise unprocessed.
1.261 + UBool fDecimalSamplesUnbounded;
1.262 + UBool fIntegerSamplesUnbounded;
1.263 +
1.264 +
1.265 + RuleChain();
1.266 + RuleChain(const RuleChain& other);
1.267 + virtual ~RuleChain();
1.268 +
1.269 + UnicodeString select(const FixedDecimal &number) const;
1.270 + void dumpRules(UnicodeString& result);
1.271 + UErrorCode getKeywords(int32_t maxArraySize, UnicodeString *keywords, int32_t& arraySize) const;
1.272 + UBool isKeyword(const UnicodeString& keyword) const;
1.273 +};
1.274 +
1.275 +class PluralKeywordEnumeration : public StringEnumeration {
1.276 +public:
1.277 + PluralKeywordEnumeration(RuleChain *header, UErrorCode& status);
1.278 + virtual ~PluralKeywordEnumeration();
1.279 + static UClassID U_EXPORT2 getStaticClassID(void);
1.280 + virtual UClassID getDynamicClassID(void) const;
1.281 + virtual const UnicodeString* snext(UErrorCode& status);
1.282 + virtual void reset(UErrorCode& status);
1.283 + virtual int32_t count(UErrorCode& status) const;
1.284 +private:
1.285 + int32_t pos;
1.286 + UVector fKeywordNames;
1.287 +};
1.288 +
1.289 +
1.290 +class U_I18N_API PluralAvailableLocalesEnumeration: public StringEnumeration {
1.291 + public:
1.292 + PluralAvailableLocalesEnumeration(UErrorCode &status);
1.293 + virtual ~PluralAvailableLocalesEnumeration();
1.294 + virtual const char* next(int32_t *resultLength, UErrorCode& status);
1.295 + virtual void reset(UErrorCode& status);
1.296 + virtual int32_t count(UErrorCode& status) const;
1.297 + private:
1.298 + UErrorCode fOpenStatus;
1.299 + UResourceBundle *fLocales;
1.300 + UResourceBundle *fRes;
1.301 +};
1.302 +
1.303 +U_NAMESPACE_END
1.304 +
1.305 +#endif /* #if !UCONFIG_NO_FORMATTING */
1.306 +
1.307 +#endif // _PLURRULE_IMPL
1.308 +//eof
