michael@0: /*
michael@0: *******************************************************************************
michael@0: * Copyright (C) 2007-2013, International Business Machines Corporation and
michael@0: * others. All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *
michael@0: * File PLURRULE_IMPL.H
michael@0: *
michael@0: *******************************************************************************
michael@0: */
michael@0: 
michael@0: 
michael@0: #ifndef PLURRULE_IMPLE
michael@0: #define PLURRULE_IMPLE
michael@0: 
michael@0: // Internal definitions for the PluralRules implementation.
michael@0: 
michael@0: #if !UCONFIG_NO_FORMATTING
michael@0: 
michael@0: #include "unicode/format.h"
michael@0: #include "unicode/locid.h"
michael@0: #include "unicode/parseerr.h"
michael@0: #include "unicode/ures.h"
michael@0: #include "unicode/utypes.h"
michael@0: #include "uvector.h"
michael@0: #include "hash.h"
michael@0: 
michael@0: class PluralRulesTest;
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: class AndConstraint;
michael@0: class RuleChain;
michael@0: 
michael@0: static const UChar DOT             = ((UChar)0x002E);
michael@0: static const UChar SINGLE_QUOTE    = ((UChar)0x0027);
michael@0: static const UChar SLASH           = ((UChar)0x002F);
michael@0: static const UChar BACKSLASH       = ((UChar)0x005C);
michael@0: static const UChar SPACE           = ((UChar)0x0020);
michael@0: static const UChar EXCLAMATION     = ((UChar)0x0021);
michael@0: static const UChar QUOTATION_MARK  = ((UChar)0x0022);
michael@0: static const UChar NUMBER_SIGN     = ((UChar)0x0023);
michael@0: static const UChar PERCENT_SIGN    = ((UChar)0x0025);
michael@0: static const UChar ASTERISK        = ((UChar)0x002A);
michael@0: static const UChar COMMA           = ((UChar)0x002C);
michael@0: static const UChar HYPHEN          = ((UChar)0x002D);
michael@0: static const UChar U_ZERO          = ((UChar)0x0030);
michael@0: static const UChar U_ONE           = ((UChar)0x0031);
michael@0: static const UChar U_TWO           = ((UChar)0x0032);
michael@0: static const UChar U_THREE         = ((UChar)0x0033);
michael@0: static const UChar U_FOUR          = ((UChar)0x0034);
michael@0: static const UChar U_FIVE          = ((UChar)0x0035);
michael@0: static const UChar U_SIX           = ((UChar)0x0036);
michael@0: static const UChar U_SEVEN         = ((UChar)0x0037);
michael@0: static const UChar U_EIGHT         = ((UChar)0x0038);
michael@0: static const UChar U_NINE          = ((UChar)0x0039);
michael@0: static const UChar COLON           = ((UChar)0x003A);
michael@0: static const UChar SEMI_COLON      = ((UChar)0x003B);
michael@0: static const UChar EQUALS          = ((UChar)0x003D);
michael@0: static const UChar AT              = ((UChar)0x0040);
michael@0: static const UChar CAP_A           = ((UChar)0x0041);
michael@0: static const UChar CAP_B           = ((UChar)0x0042);
michael@0: static const UChar CAP_R           = ((UChar)0x0052);
michael@0: static const UChar CAP_Z           = ((UChar)0x005A);
michael@0: static const UChar LOWLINE         = ((UChar)0x005F);
michael@0: static const UChar LEFTBRACE       = ((UChar)0x007B);
michael@0: static const UChar RIGHTBRACE      = ((UChar)0x007D);
michael@0: static const UChar TILDE           = ((UChar)0x007E);
michael@0: static const UChar ELLIPSIS        = ((UChar)0x2026);
michael@0: 
michael@0: static const UChar LOW_A           = ((UChar)0x0061);
michael@0: static const UChar LOW_B           = ((UChar)0x0062);
michael@0: static const UChar LOW_C           = ((UChar)0x0063);
michael@0: static const UChar LOW_D           = ((UChar)0x0064);
michael@0: static const UChar LOW_E           = ((UChar)0x0065);
michael@0: static const UChar LOW_F           = ((UChar)0x0066);
michael@0: static const UChar LOW_G           = ((UChar)0x0067);
michael@0: static const UChar LOW_H           = ((UChar)0x0068);
michael@0: static const UChar LOW_I           = ((UChar)0x0069);
michael@0: static const UChar LOW_J           = ((UChar)0x006a);
michael@0: static const UChar LOW_K           = ((UChar)0x006B);
michael@0: static const UChar LOW_L           = ((UChar)0x006C);
michael@0: static const UChar LOW_M           = ((UChar)0x006D);
michael@0: static const UChar LOW_N           = ((UChar)0x006E);
michael@0: static const UChar LOW_O           = ((UChar)0x006F);
michael@0: static const UChar LOW_P           = ((UChar)0x0070);
michael@0: static const UChar LOW_Q           = ((UChar)0x0071);
michael@0: static const UChar LOW_R           = ((UChar)0x0072);
michael@0: static const UChar LOW_S           = ((UChar)0x0073);
michael@0: static const UChar LOW_T           = ((UChar)0x0074);
michael@0: static const UChar LOW_U           = ((UChar)0x0075);
michael@0: static const UChar LOW_V           = ((UChar)0x0076);
michael@0: static const UChar LOW_W           = ((UChar)0x0077);
michael@0: static const UChar LOW_Y           = ((UChar)0x0079);
michael@0: static const UChar LOW_Z           = ((UChar)0x007A);
michael@0: 
michael@0: 
michael@0: static const int32_t PLURAL_RANGE_HIGH = 0x7fffffff;
michael@0: 
michael@0: enum tokenType {
michael@0:   none,
michael@0:   tNumber,
michael@0:   tComma,
michael@0:   tSemiColon,
michael@0:   tSpace,
michael@0:   tColon,
michael@0:   tAt,           // '@'
michael@0:   tDot,
michael@0:   tDot2,
michael@0:   tEllipsis,
michael@0:   tKeyword,
michael@0:   tAnd,
michael@0:   tOr,
michael@0:   tMod,          // 'mod' or '%'
michael@0:   tNot,          //  'not' only.
michael@0:   tIn,           //  'in'  only.
michael@0:   tEqual,        //  '='   only.
michael@0:   tNotEqual,     //  '!='
michael@0:   tTilde,
michael@0:   tWithin,
michael@0:   tIs,
michael@0:   tVariableN,
michael@0:   tVariableI,
michael@0:   tVariableF,
michael@0:   tVariableV,
michael@0:   tVariableT,
michael@0:   tDecimal,
michael@0:   tInteger,
michael@0:   tEOF
michael@0: };
michael@0: 
michael@0: 
michael@0: class PluralRuleParser: public UMemory {
michael@0: public:
michael@0:     PluralRuleParser();
michael@0:     virtual ~PluralRuleParser();
michael@0: 
michael@0:     void parse(const UnicodeString &rules, PluralRules *dest, UErrorCode &status);
michael@0:     void getNextToken(UErrorCode &status);
michael@0:     void checkSyntax(UErrorCode &status);
michael@0:     static int32_t getNumberValue(const UnicodeString &token);
michael@0: 
michael@0: private:
michael@0:     static tokenType getKeyType(const UnicodeString& token, tokenType type);
michael@0:     static tokenType charType(UChar ch);
michael@0:     static UBool isValidKeyword(const UnicodeString& token);
michael@0: 
michael@0:     const UnicodeString  *ruleSrc;  // The rules string.
michael@0:     int32_t        ruleIndex;       // String index in the input rules, the current parse position.
michael@0:     UnicodeString  token;           // Token most recently scanned.
michael@0:     tokenType      type;
michael@0:     tokenType      prevType;
michael@0: 
michael@0:                                     // The items currently being parsed & built.
michael@0:                                     // Note: currentChain may not be the last RuleChain in the
michael@0:                                     //       list because the "other" chain is forced to the end.
michael@0:     AndConstraint *curAndConstraint;
michael@0:     RuleChain     *currentChain;
michael@0: 
michael@0:     int32_t        rangeLowIdx;     // Indices in the UVector of ranges of the
michael@0:     int32_t        rangeHiIdx;      //    low and hi values currently being parsed.
michael@0: 
michael@0:     enum EParseState {
michael@0:        kKeyword,
michael@0:        kExpr,
michael@0:        kValue,
michael@0:        kRangeList,
michael@0:        kSamples
michael@0:     };
michael@0: 
michael@0: };
michael@0: 
michael@0: /**
michael@0:  * class FixedDecimal serves to communicate the properties
michael@0:  * of a formatted number from a decimal formatter to PluralRules::select()
michael@0:  *
michael@0:  * see DecimalFormat::getFixedDecimal()
michael@0:  * @internal
michael@0:  */
michael@0: class U_I18N_API FixedDecimal: public UMemory {
michael@0:   public:
michael@0:     /**
michael@0:       * @param n   the number, e.g. 12.345
michael@0:       * @param v   The number of visible fraction digits, e.g. 3
michael@0:       * @param f   The fraction digits, e.g. 345
michael@0:       */
michael@0:     FixedDecimal(double  n, int32_t v, int64_t f);
michael@0:     FixedDecimal(double n, int32_t);
michael@0:     explicit FixedDecimal(double n);
michael@0:     FixedDecimal();
michael@0:     FixedDecimal(const UnicodeString &s, UErrorCode &ec);
michael@0:     FixedDecimal(const FixedDecimal &other);
michael@0: 
michael@0:     double get(tokenType operand) const;
michael@0:     int32_t getVisibleFractionDigitCount() const;
michael@0: 
michael@0:     void init(double n, int32_t v, int64_t f);
michael@0:     void init(double n);
michael@0:     UBool quickInit(double n);  // Try a fast-path only initialization,
michael@0:                                 //    return TRUE if successful.
michael@0:     void adjustForMinFractionDigits(int32_t min);
michael@0:     static int64_t getFractionalDigits(double n, int32_t v);
michael@0:     static int32_t decimals(double n);
michael@0: 
michael@0:     double      source;
michael@0:     int32_t     visibleDecimalDigitCount;
michael@0:     int64_t     decimalDigits;
michael@0:     int64_t     decimalDigitsWithoutTrailingZeros;
michael@0:     int64_t     intValue;
michael@0:     UBool       hasIntegerValue;
michael@0:     UBool       isNegative;
michael@0:     UBool       isNanOrInfinity;
michael@0: };
michael@0: 
michael@0: class AndConstraint : public UMemory  {
michael@0: public:
michael@0:     typedef enum RuleOp {
michael@0:         NONE,
michael@0:         MOD
michael@0:     } RuleOp;
michael@0:     RuleOp  op;
michael@0:     int32_t opNum;           // for mod expressions, the right operand of the mod.
michael@0:     int32_t     value;       // valid for 'is' rules only.
michael@0:     UVector32   *rangeList;  // for 'in', 'within' rules. Null otherwise.
michael@0:     UBool   negated;           // TRUE for negated rules.
michael@0:     UBool   integerOnly;     // TRUE for 'within' rules.
michael@0:     tokenType digitsType;    // n | i | v | f constraint.
michael@0:     AndConstraint *next;
michael@0: 
michael@0:     AndConstraint();
michael@0:     AndConstraint(const AndConstraint& other);
michael@0:     virtual ~AndConstraint();
michael@0:     AndConstraint* add();
michael@0:     // UBool isFulfilled(double number);
michael@0:     UBool isFulfilled(const FixedDecimal &number);
michael@0: };
michael@0: 
michael@0: class OrConstraint : public UMemory  {
michael@0: public:
michael@0:     AndConstraint *childNode;
michael@0:     OrConstraint *next;
michael@0:     OrConstraint();
michael@0: 
michael@0:     OrConstraint(const OrConstraint& other);
michael@0:     virtual ~OrConstraint();
michael@0:     AndConstraint* add();
michael@0:     // UBool isFulfilled(double number);
michael@0:     UBool isFulfilled(const FixedDecimal &number);
michael@0: };
michael@0: 
michael@0: class RuleChain : public UMemory  {
michael@0: public:
michael@0:     UnicodeString   fKeyword;
michael@0:     RuleChain      *fNext;
michael@0:     OrConstraint   *ruleHeader;
michael@0:     UnicodeString   fDecimalSamples;  // Samples strings from rule source
michael@0:     UnicodeString   fIntegerSamples;  //   without @decimal or @integer, otherwise unprocessed.
michael@0:     UBool           fDecimalSamplesUnbounded;
michael@0:     UBool           fIntegerSamplesUnbounded;
michael@0: 
michael@0: 
michael@0:     RuleChain();
michael@0:     RuleChain(const RuleChain& other);
michael@0:     virtual ~RuleChain();
michael@0: 
michael@0:     UnicodeString select(const FixedDecimal &number) const;
michael@0:     void          dumpRules(UnicodeString& result);
michael@0:     UErrorCode    getKeywords(int32_t maxArraySize, UnicodeString *keywords, int32_t& arraySize) const;
michael@0:     UBool         isKeyword(const UnicodeString& keyword) const;
michael@0: };
michael@0: 
michael@0: class PluralKeywordEnumeration : public StringEnumeration {
michael@0: public:
michael@0:     PluralKeywordEnumeration(RuleChain *header, UErrorCode& status);
michael@0:     virtual ~PluralKeywordEnumeration();
michael@0:     static UClassID U_EXPORT2 getStaticClassID(void);
michael@0:     virtual UClassID getDynamicClassID(void) const;
michael@0:     virtual const UnicodeString* snext(UErrorCode& status);
michael@0:     virtual void reset(UErrorCode& status);
michael@0:     virtual int32_t count(UErrorCode& status) const;
michael@0: private:
michael@0:     int32_t         pos;
michael@0:     UVector         fKeywordNames;
michael@0: };
michael@0: 
michael@0: 
michael@0: class U_I18N_API PluralAvailableLocalesEnumeration: public StringEnumeration {
michael@0:   public:
michael@0:     PluralAvailableLocalesEnumeration(UErrorCode &status);
michael@0:     virtual ~PluralAvailableLocalesEnumeration();
michael@0:     virtual const char* next(int32_t *resultLength, UErrorCode& status);
michael@0:     virtual void reset(UErrorCode& status);
michael@0:     virtual int32_t count(UErrorCode& status) const;
michael@0:   private:
michael@0:     UErrorCode      fOpenStatus;
michael@0:     UResourceBundle *fLocales;
michael@0:     UResourceBundle *fRes;
michael@0: };
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0: #endif /* #if !UCONFIG_NO_FORMATTING */
michael@0: 
michael@0: #endif // _PLURRULE_IMPL
michael@0: //eof