[apple/icu.git] / icuSources / i18n / dtptngen_impl.h

/*
*******************************************************************************
* Copyright (C) 2007-2016, International Business Machines Corporation and
* others. All Rights Reserved.
*******************************************************************************
*
* File DTPTNGEN.H
*
*******************************************************************************
*/

#ifndef __DTPTNGEN_IMPL_H__
#define __DTPTNGEN_IMPL_H__

#include "unicode/udatpg.h"
#include "uvector.h"

// TODO(claireho): Split off Builder class.
// TODO(claireho): If splitting off Builder class: As subclass or independent?

#define MAX_PATTERN_ENTRIES 52
#define MAX_CLDR_FIELD_LEN  60
#define MAX_DT_TOKEN        50
#define MAX_RESOURCE_FIELD  12
#define MAX_AVAILABLE_FORMATS  12
#define NONE          0
#define EXTRA_FIELD   0x10000
#define MISSING_FIELD  0x1000
#define MAX_STRING_ENUMERATION  200
#define SINGLE_QUOTE      ((UChar)0x0027)
#define FORWARDSLASH      ((UChar)0x002F)
#define BACKSLASH         ((UChar)0x005C)
#define SPACE             ((UChar)0x0020)
#define QUOTATION_MARK    ((UChar)0x0022)
#define ASTERISK          ((UChar)0x002A)
#define PLUSSITN          ((UChar)0x002B)
#define COMMA             ((UChar)0x002C)
#define HYPHEN            ((UChar)0x002D)
#define DOT               ((UChar)0x002E)
#define COLON             ((UChar)0x003A)
#define CAP_A             ((UChar)0x0041)
#define CAP_B             ((UChar)0x0042)
#define CAP_C             ((UChar)0x0043)
#define CAP_D             ((UChar)0x0044)
#define CAP_E             ((UChar)0x0045)
#define CAP_F             ((UChar)0x0046)
#define CAP_G             ((UChar)0x0047)
#define CAP_H             ((UChar)0x0048)
#define CAP_J             ((UChar)0x004A)
#define CAP_K             ((UChar)0x004B)
#define CAP_L             ((UChar)0x004C)
#define CAP_M             ((UChar)0x004D)
#define CAP_O             ((UChar)0x004F)
#define CAP_Q             ((UChar)0x0051)
#define CAP_S             ((UChar)0x0053)
#define CAP_T             ((UChar)0x0054)
#define CAP_U             ((UChar)0x0055)
#define CAP_V             ((UChar)0x0056)
#define CAP_W             ((UChar)0x0057)
#define CAP_X             ((UChar)0x0058)
#define CAP_Y             ((UChar)0x0059)
#define CAP_Z             ((UChar)0x005A)
#define LOWLINE           ((UChar)0x005F)
#define LOW_A             ((UChar)0x0061)
#define LOW_B             ((UChar)0x0062)
#define LOW_C             ((UChar)0x0063)
#define LOW_D             ((UChar)0x0064)
#define LOW_E             ((UChar)0x0065)
#define LOW_F             ((UChar)0x0066)
#define LOW_G             ((UChar)0x0067)
#define LOW_H             ((UChar)0x0068)
#define LOW_I             ((UChar)0x0069)
#define LOW_J             ((UChar)0x006A)
#define LOW_K             ((UChar)0x006B)
#define LOW_L             ((UChar)0x006C)
#define LOW_M             ((UChar)0x006D)
#define LOW_N             ((UChar)0x006E)
#define LOW_O             ((UChar)0x006F)
#define LOW_P             ((UChar)0x0070)
#define LOW_Q             ((UChar)0x0071)
#define LOW_R             ((UChar)0x0072)
#define LOW_S             ((UChar)0x0073)
#define LOW_T             ((UChar)0x0074)
#define LOW_U             ((UChar)0x0075)
#define LOW_V             ((UChar)0x0076)
#define LOW_W             ((UChar)0x0077)
#define LOW_X             ((UChar)0x0078)
#define LOW_Y             ((UChar)0x0079)
#define LOW_Z             ((UChar)0x007A)
#define DT_SHORT          -0x102
#define DT_LONG           -0x103
#define DT_NUMERIC         0x100
#define DT_NARROW         -0x101
#define DT_DELTA           0x10

U_NAMESPACE_BEGIN

const int32_t UDATPG_FRACTIONAL_MASK = 1<<UDATPG_FRACTIONAL_SECOND_FIELD;
const int32_t UDATPG_SECOND_AND_FRACTIONAL_MASK = (1<<UDATPG_SECOND_FIELD) | (1<<UDATPG_FRACTIONAL_SECOND_FIELD);

typedef enum dtStrEnum {
    DT_BASESKELETON,
    DT_SKELETON,
    DT_PATTERN
}dtStrEnum;

typedef struct dtTypeElem {
    UChar                  patternChar;
    UDateTimePatternField  field;
    int16_t                type;
    int16_t                minLen;
    int16_t                weight;
}dtTypeElem;

class PtnSkeleton : public UMemory {
public:
    int32_t type[UDATPG_FIELD_COUNT];
    UnicodeString original[UDATPG_FIELD_COUNT];
    UnicodeString baseOriginal[UDATPG_FIELD_COUNT];

    PtnSkeleton();
    PtnSkeleton(const PtnSkeleton& other);
    UBool equals(const PtnSkeleton& other);
    UnicodeString getSkeleton();
    UnicodeString getBaseSkeleton();
    virtual ~PtnSkeleton();
};


class PtnElem : public UMemory {
public:
    UnicodeString basePattern;
    PtnSkeleton   *skeleton;
    UnicodeString pattern;
    UBool         skeletonWasSpecified; // if specified in availableFormats, not derived
    PtnElem       *next;

    PtnElem(const UnicodeString &basePattern, const UnicodeString &pattern);
    virtual ~PtnElem();

};

class FormatParser : public UMemory {
public:
    UnicodeString items[MAX_DT_TOKEN];
    int32_t  itemNumber;

    FormatParser();
    virtual ~FormatParser();
    void set(const UnicodeString& patternString);
    void getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex);
    UBool isPatternSeparator(UnicodeString& field);
    static UBool isQuoteLiteral(const UnicodeString& s);
    static int32_t getCanonicalIndex(const UnicodeString& s) { return getCanonicalIndex(s, TRUE); }
    static int32_t getCanonicalIndex(const UnicodeString& s, UBool strict);

private:
   typedef enum TokenStatus {
       START,
       ADD_TOKEN,
       SYNTAX_ERROR,
       DONE
   } ToeknStatus;

   TokenStatus status;
   virtual TokenStatus setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len);
};

class DistanceInfo : public UMemory {
public:
    int32_t missingFieldMask;
    int32_t extraFieldMask;

    DistanceInfo() {}
    virtual ~DistanceInfo();
    void clear() { missingFieldMask = extraFieldMask = 0; }
    void setTo(DistanceInfo& other);
    void addMissing(int32_t field) { missingFieldMask |= (1<<field); }
    void addExtra(int32_t field) { extraFieldMask |= (1<<field); }
};

class DateTimeMatcher: public UMemory {
public:
    PtnSkeleton skeleton;

    void getBasePattern(UnicodeString &basePattern);
    UnicodeString getPattern();
    void set(const UnicodeString& pattern, FormatParser* fp);
    void set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeleton);
    void copyFrom(const PtnSkeleton& skeleton);
    void copyFrom();
    PtnSkeleton* getSkeletonPtr();
    UBool equals(const DateTimeMatcher* other) const;
    int32_t getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo);
    DateTimeMatcher();
    DateTimeMatcher(const DateTimeMatcher& other);
    virtual ~DateTimeMatcher();
    int32_t getFieldMask();
};

class PatternMap : public UMemory {
public:
    PtnElem *boot[MAX_PATTERN_ENTRIES];
    PatternMap();
    virtual  ~PatternMap();
    void  add(const UnicodeString& basePattern, const PtnSkeleton& skeleton, const UnicodeString& value, UBool skeletonWasSpecified, UErrorCode& status);
    const UnicodeString* getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeletonWasSpecified);
    const UnicodeString* getPatternFromSkeleton(PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr = 0);
    void copyFrom(const PatternMap& other, UErrorCode& status);
    PtnElem* getHeader(UChar baseChar);
    UBool equals(const PatternMap& other);
private:
    UBool isDupAllowed;
    PtnElem*  getDuplicateElem(const UnicodeString &basePattern, const PtnSkeleton& skeleton, PtnElem *baseElem);
}; // end  PatternMap

class PatternMapIterator : public UMemory {
public:
    PatternMapIterator();
    virtual ~PatternMapIterator();
    void set(PatternMap& patternMap);
    PtnSkeleton* getSkeleton();
    UBool hasNext();
    DateTimeMatcher& next();
private:
    int32_t bootIndex;
    PtnElem *nodePtr;
    DateTimeMatcher *matcher;
    PatternMap *patternMap;
};

class DTSkeletonEnumeration : public StringEnumeration {
public:
    DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum type, UErrorCode& status);
    virtual ~DTSkeletonEnumeration();
    static UClassID U_EXPORT2 getStaticClassID(void);
    virtual UClassID getDynamicClassID(void) const;
    virtual const UnicodeString* snext(UErrorCode& status);
    virtual void reset(UErrorCode& status);
    virtual int32_t count(UErrorCode& status) const;
private:
    int32_t pos;
    UBool isCanonicalItem(const UnicodeString& item);
    UVector *fSkeletons;
};

class DTRedundantEnumeration : public StringEnumeration {
public:
    DTRedundantEnumeration();
    virtual ~DTRedundantEnumeration();
    static UClassID U_EXPORT2 getStaticClassID(void);
    virtual UClassID getDynamicClassID(void) const;
    virtual const UnicodeString* snext(UErrorCode& status);
    virtual void reset(UErrorCode& status);
    virtual int32_t count(UErrorCode& status) const;
    void add(const UnicodeString &pattern, UErrorCode& status);
private:
    int32_t pos;
    UBool isCanonicalItem(const UnicodeString& item);
    UVector *fPatterns;
};

U_NAMESPACE_END

#endif
Commit	Line	Data
46f4442e A	1	/*
46f4442e A	2	*******************************************************************************
2ca993e8 A	3	* Copyright (C) 2007-2016, International Business Machines Corporation and
2ca993e8 A	4	* others. All Rights Reserved.
46f4442e A	5	*******************************************************************************
	6	*
	7	* File DTPTNGEN.H
	8	*
	9	*******************************************************************************
	10	*/
	11
46f4442e A	12	#ifndef __DTPTNGEN_IMPL_H__
	13	#define __DTPTNGEN_IMPL_H__
	14
2ca993e8 A	15	#include "unicode/udatpg.h"
	16	#include "uvector.h"
	17
46f4442e A	18	// TODO(claireho): Split off Builder class.
	19	// TODO(claireho): If splitting off Builder class: As subclass or independent?
	20
	21	#define MAX_PATTERN_ENTRIES 52
	22	#define MAX_CLDR_FIELD_LEN 60
	23	#define MAX_DT_TOKEN 50
	24	#define MAX_RESOURCE_FIELD 12
	25	#define MAX_AVAILABLE_FORMATS 12
	26	#define NONE 0
	27	#define EXTRA_FIELD 0x10000
	28	#define MISSING_FIELD 0x1000
	29	#define MAX_STRING_ENUMERATION 200
	30	#define SINGLE_QUOTE ((UChar)0x0027)
	31	#define FORWARDSLASH ((UChar)0x002F)
	32	#define BACKSLASH ((UChar)0x005C)
	33	#define SPACE ((UChar)0x0020)
	34	#define QUOTATION_MARK ((UChar)0x0022)
	35	#define ASTERISK ((UChar)0x002A)
	36	#define PLUSSITN ((UChar)0x002B)
	37	#define COMMA ((UChar)0x002C)
	38	#define HYPHEN ((UChar)0x002D)
	39	#define DOT ((UChar)0x002E)
	40	#define COLON ((UChar)0x003A)
	41	#define CAP_A ((UChar)0x0041)
2ca993e8	42	#define CAP_B ((UChar)0x0042)
46f4442e A	43	#define CAP_C ((UChar)0x0043)
	44	#define CAP_D ((UChar)0x0044)
	45	#define CAP_E ((UChar)0x0045)
	46	#define CAP_F ((UChar)0x0046)
	47	#define CAP_G ((UChar)0x0047)
	48	#define CAP_H ((UChar)0x0048)
57a6839d	49	#define CAP_J ((UChar)0x004A)
46f4442e A	50	#define CAP_K ((UChar)0x004B)
	51	#define CAP_L ((UChar)0x004C)
	52	#define CAP_M ((UChar)0x004D)
	53	#define CAP_O ((UChar)0x004F)
	54	#define CAP_Q ((UChar)0x0051)
	55	#define CAP_S ((UChar)0x0053)
	56	#define CAP_T ((UChar)0x0054)
4388f060	57	#define CAP_U ((UChar)0x0055)
46f4442e A	58	#define CAP_V ((UChar)0x0056)
46f4442e A	59	#define CAP_W ((UChar)0x0057)
57a6839d	60	#define CAP_X ((UChar)0x0058)
46f4442e A	61	#define CAP_Y ((UChar)0x0059)
	62	#define CAP_Z ((UChar)0x005A)
	63	#define LOWLINE ((UChar)0x005F)
	64	#define LOW_A ((UChar)0x0061)
2ca993e8	65	#define LOW_B ((UChar)0x0062)
46f4442e A	66	#define LOW_C ((UChar)0x0063)
	67	#define LOW_D ((UChar)0x0064)
	68	#define LOW_E ((UChar)0x0065)
	69	#define LOW_F ((UChar)0x0066)
	70	#define LOW_G ((UChar)0x0067)
	71	#define LOW_H ((UChar)0x0068)
	72	#define LOW_I ((UChar)0x0069)
	73	#define LOW_J ((UChar)0x006A)
	74	#define LOW_K ((UChar)0x006B)
	75	#define LOW_L ((UChar)0x006C)
	76	#define LOW_M ((UChar)0x006D)
	77	#define LOW_N ((UChar)0x006E)
	78	#define LOW_O ((UChar)0x006F)
	79	#define LOW_P ((UChar)0x0070)
	80	#define LOW_Q ((UChar)0x0071)
	81	#define LOW_R ((UChar)0x0072)
	82	#define LOW_S ((UChar)0x0073)
	83	#define LOW_T ((UChar)0x0074)
	84	#define LOW_U ((UChar)0x0075)
	85	#define LOW_V ((UChar)0x0076)
	86	#define LOW_W ((UChar)0x0077)
57a6839d	87	#define LOW_X ((UChar)0x0078)
46f4442e A	88	#define LOW_Y ((UChar)0x0079)
	89	#define LOW_Z ((UChar)0x007A)
	90	#define DT_SHORT -0x102
	91	#define DT_LONG -0x103
	92	#define DT_NUMERIC 0x100
	93	#define DT_NARROW -0x101
	94	#define DT_DELTA 0x10
	95
	96	U_NAMESPACE_BEGIN
	97
	98	const int32_t UDATPG_FRACTIONAL_MASK = 1<<UDATPG_FRACTIONAL_SECOND_FIELD;
	99	const int32_t UDATPG_SECOND_AND_FRACTIONAL_MASK = (1<<UDATPG_SECOND_FIELD) \| (1<<UDATPG_FRACTIONAL_SECOND_FIELD);
	100
	101	typedef enum dtStrEnum {
	102	DT_BASESKELETON,
	103	DT_SKELETON,
	104	DT_PATTERN
	105	}dtStrEnum;
	106
	107	typedef struct dtTypeElem {
	108	UChar patternChar;
	109	UDateTimePatternField field;
	110	int16_t type;
	111	int16_t minLen;
	112	int16_t weight;
	113	}dtTypeElem;
	114
	115	class PtnSkeleton : public UMemory {
	116	public:
	117	int32_t type[UDATPG_FIELD_COUNT];
	118	UnicodeString original[UDATPG_FIELD_COUNT];
	119	UnicodeString baseOriginal[UDATPG_FIELD_COUNT];
	120
	121	PtnSkeleton();
	122	PtnSkeleton(const PtnSkeleton& other);
	123	UBool equals(const PtnSkeleton& other);
	124	UnicodeString getSkeleton();
	125	UnicodeString getBaseSkeleton();
	126	virtual ~PtnSkeleton();
	127	};
	128
	129
	130	class PtnElem : public UMemory {
	131	public:
	132	UnicodeString basePattern;
	133	PtnSkeleton *skeleton;
	134	UnicodeString pattern;
	135	UBool skeletonWasSpecified; // if specified in availableFormats, not derived
	136	PtnElem *next;
	137
	138	PtnElem(const UnicodeString &basePattern, const UnicodeString &pattern);
	139	virtual ~PtnElem();
	140
	141	};
	142
	143	class FormatParser : public UMemory {
	144	public:
	145	UnicodeString items[MAX_DT_TOKEN];
	146	int32_t itemNumber;
	147
	148	FormatParser();
	149	virtual ~FormatParser();
	150	void set(const UnicodeString& patternString);
46f4442e	151	void getQuoteLiteral(UnicodeString& quote, int32_t *itemIndex);
46f4442e	152	UBool isPatternSeparator(UnicodeString& field);
2ca993e8 A	153	static UBool isQuoteLiteral(const UnicodeString& s);
	154	static int32_t getCanonicalIndex(const UnicodeString& s) { return getCanonicalIndex(s, TRUE); }
	155	static int32_t getCanonicalIndex(const UnicodeString& s, UBool strict);
46f4442e A	156
	157	private:
	158	typedef enum TokenStatus {
	159	START,
	160	ADD_TOKEN,
	161	SYNTAX_ERROR,
	162	DONE
	163	} ToeknStatus;
	164
	165	TokenStatus status;
	166	virtual TokenStatus setTokens(const UnicodeString& pattern, int32_t startPos, int32_t *len);
	167	};
	168
	169	class DistanceInfo : public UMemory {
	170	public:
	171	int32_t missingFieldMask;
	172	int32_t extraFieldMask;
	173
4388f060 A	174	DistanceInfo() {}
	175	virtual ~DistanceInfo();
	176	void clear() { missingFieldMask = extraFieldMask = 0; }
46f4442e	177	void setTo(DistanceInfo& other);
4388f060 A	178	void addMissing(int32_t field) { missingFieldMask \|= (1<<field); }
4388f060 A	179	void addExtra(int32_t field) { extraFieldMask \|= (1<<field); }
46f4442e A	180	};
	181
	182	class DateTimeMatcher: public UMemory {
	183	public:
	184	PtnSkeleton skeleton;
	185
	186	void getBasePattern(UnicodeString &basePattern);
	187	UnicodeString getPattern();
	188	void set(const UnicodeString& pattern, FormatParser* fp);
	189	void set(const UnicodeString& pattern, FormatParser* fp, PtnSkeleton& skeleton);
	190	void copyFrom(const PtnSkeleton& skeleton);
	191	void copyFrom();
	192	PtnSkeleton* getSkeletonPtr();
	193	UBool equals(const DateTimeMatcher* other) const;
	194	int32_t getDistance(const DateTimeMatcher& other, int32_t includeMask, DistanceInfo& distanceInfo);
	195	DateTimeMatcher();
	196	DateTimeMatcher(const DateTimeMatcher& other);
4388f060	197	virtual ~DateTimeMatcher();
46f4442e A	198	int32_t getFieldMask();
	199	};
	200
	201	class PatternMap : public UMemory {
	202	public:
	203	PtnElem *boot[MAX_PATTERN_ENTRIES];
	204	PatternMap();
	205	virtual ~PatternMap();
	206	void add(const UnicodeString& basePattern, const PtnSkeleton& skeleton, const UnicodeString& value, UBool skeletonWasSpecified, UErrorCode& status);
	207	const UnicodeString* getPatternFromBasePattern(UnicodeString& basePattern, UBool& skeletonWasSpecified);
	208	const UnicodeString* getPatternFromSkeleton(PtnSkeleton& skeleton, const PtnSkeleton** specifiedSkeletonPtr = 0);
	209	void copyFrom(const PatternMap& other, UErrorCode& status);
	210	PtnElem* getHeader(UChar baseChar);
	211	UBool equals(const PatternMap& other);
	212	private:
	213	UBool isDupAllowed;
	214	PtnElem* getDuplicateElem(const UnicodeString &basePattern, const PtnSkeleton& skeleton, PtnElem *baseElem);
	215	}; // end PatternMap
	216
	217	class PatternMapIterator : public UMemory {
	218	public:
	219	PatternMapIterator();
	220	virtual ~PatternMapIterator();
	221	void set(PatternMap& patternMap);
	222	PtnSkeleton* getSkeleton();
	223	UBool hasNext();
	224	DateTimeMatcher& next();
	225	private:
	226	int32_t bootIndex;
	227	PtnElem *nodePtr;
	228	DateTimeMatcher *matcher;
	229	PatternMap *patternMap;
	230	};
	231
	232	class DTSkeletonEnumeration : public StringEnumeration {
	233	public:
	234	DTSkeletonEnumeration(PatternMap &patternMap, dtStrEnum type, UErrorCode& status);
	235	virtual ~DTSkeletonEnumeration();
	236	static UClassID U_EXPORT2 getStaticClassID(void);
	237	virtual UClassID getDynamicClassID(void) const;
	238	virtual const UnicodeString* snext(UErrorCode& status);
	239	virtual void reset(UErrorCode& status);
	240	virtual int32_t count(UErrorCode& status) const;
	241	private:
	242	int32_t pos;
	243	UBool isCanonicalItem(const UnicodeString& item);
	244	UVector *fSkeletons;
	245	};
	246
	247	class DTRedundantEnumeration : public StringEnumeration {
	248	public:
	249	DTRedundantEnumeration();
	250	virtual ~DTRedundantEnumeration();
	251	static UClassID U_EXPORT2 getStaticClassID(void);
	252	virtual UClassID getDynamicClassID(void) const;
	253	virtual const UnicodeString* snext(UErrorCode& status);
	254	virtual void reset(UErrorCode& status);
	255	virtual int32_t count(UErrorCode& status) const;
	256	void add(const UnicodeString &pattern, UErrorCode& status);
	257	private:
	258	int32_t pos;
	259	UBool isCanonicalItem(const UnicodeString& item);
	260	UVector *fPatterns;
	261	};
262
263	U_NAMESPACE_END
264
265	#endif