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

/*
**********************************************************************
*   Copyright (C) 2001-2015 IBM and others. All rights reserved.
**********************************************************************
*   Date        Name        Description
*  08/13/2001   synwee      Creation.
**********************************************************************
*/
#ifndef USRCHIMP_H
#define USRCHIMP_H

#include "unicode/utypes.h"

#if !UCONFIG_NO_COLLATION

#include "unicode/normalizer2.h"
#include "unicode/ucol.h"
#include "unicode/ucoleitr.h"
#include "unicode/ubrk.h"

/* mask off anything but primary order */
#define UCOL_PRIMARYORDERMASK 0xffff0000
/* mask off anything but secondary order */
#define UCOL_SECONDARYORDERMASK 0x0000ff00
/* mask off anything but tertiary order */
#define UCOL_TERTIARYORDERMASK 0x000000ff
/* primary order shift */
#define UCOL_PRIMARYORDERSHIFT 16
/* secondary order shift */
#define UCOL_SECONDARYORDERSHIFT 8

#define UCOL_IGNORABLE 0

/* get weights from a CE */
#define UCOL_PRIMARYORDER(order) (((order) >> 16) & 0xffff)
#define UCOL_SECONDARYORDER(order) (((order) & UCOL_SECONDARYORDERMASK)>> UCOL_SECONDARYORDERSHIFT)
#define UCOL_TERTIARYORDER(order) ((order) & UCOL_TERTIARYORDERMASK)

#define UCOL_CONTINUATION_MARKER 0xC0

#define isContinuation(CE) (((CE) & UCOL_CONTINUATION_MARKER) == UCOL_CONTINUATION_MARKER)

/**
 * This indicates an error has occured during processing or there are no more CEs 
 * to be returned.
 */
#ifndef UCOL_PROCESSED_NULLORDER
#define UCOL_PROCESSED_NULLORDER        ((int64_t)U_INT64_MAX)
#endif

U_NAMESPACE_BEGIN

class CollationElementIterator;
class Collator;

struct PCEI
{
    uint64_t ce;
    int32_t  low;
    int32_t  high;
};

struct PCEBuffer
{
    PCEI    defaultBuffer[16];
    PCEI   *buffer;
    int32_t bufferIndex;
    int32_t bufferSize;

    PCEBuffer();
    ~PCEBuffer();

    void  reset();
    UBool isEmpty() const;
    void  put(uint64_t ce, int32_t ixLow, int32_t ixHigh, UErrorCode &errorCode);
    const PCEI *get();
};

class UCollationPCE : public UMemory {
private:
    PCEBuffer          pceBuffer;
    CollationElementIterator *cei;
    UCollationStrength strength;
    UBool              toShift;
    UBool              isShifted;
    uint32_t           variableTop;

public:
    UCollationPCE(UCollationElements *elems);
    UCollationPCE(CollationElementIterator *iter);
    ~UCollationPCE();

    void init(UCollationElements *elems);
    void init(CollationElementIterator *iter);

    /**
     * Get the processed ordering priority of the next collation element in the text.
     * A single character may contain more than one collation element.
     *
     * @param ixLow a pointer to an int32_t to receive the iterator index before fetching the CE.
     * @param ixHigh a pointer to an int32_t to receive the iterator index after fetching the CE.
     * @param status A pointer to an UErrorCode to receive any errors.
     * @return The next collation elements ordering, otherwise returns UCOL_PROCESSED_NULLORDER 
     *         if an error has occured or if the end of string has been reached
     */
    int64_t nextProcessed(int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);
    /**
     * Get the processed ordering priority of the previous collation element in the text.
     * A single character may contain more than one collation element.
     *
     * @param ixLow A pointer to an int32_t to receive the iterator index after fetching the CE
     * @param ixHigh A pointer to an int32_t to receiver the iterator index before fetching the CE
     * @param status A pointer to an UErrorCode to receive any errors. Noteably 
     *               a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
     *               buffer has been exhausted.
     * @return The previous collation elements ordering, otherwise returns 
     *         UCOL_PROCESSED_NULLORDER if an error has occured or if the start of
     *         string has been reached.
     */
    int64_t previousProcessed(int32_t *ixLow, int32_t *ixHigh, UErrorCode *status);

private:
    void init(const Collator &coll);
    uint64_t processCE(uint32_t ce);
};

U_NAMESPACE_END

#define INITIAL_ARRAY_SIZE_       256
#define MAX_TABLE_SIZE_           257

struct USearch {
    // required since collation element iterator does not have a getText API
    const UChar              *text;
          int32_t             textLength; // exact length
          UBool               isOverlap;
          UBool               isCanonicalMatch;
          int16_t             elementComparisonType;
          UBreakIterator     *internalBreakIter;  //internal character breakiterator
          UBreakIterator     *breakIter;
    // value USEARCH_DONE is the default value
    // if we are not at the start of the text or the end of the text, 
    // depending on the iteration direction and matchedIndex is USEARCH_DONE 
    // it means that we can't find any more matches in that particular direction
          int32_t             matchedIndex; 
          int32_t             matchedLength;
          UBool               isForwardSearching;
          UBool               reset;
};

struct UPattern {
    const UChar              *text;
          int32_t             textLength; // exact length
          // length required for backwards ce comparison
          int32_t             cesLength;
          int32_t            *ces;
          int32_t             cesBuffer[INITIAL_ARRAY_SIZE_];
          int32_t             pcesLength;
          int64_t            *pces;
          int64_t             pcesBuffer[INITIAL_ARRAY_SIZE_];
          UBool               hasPrefixAccents;
          UBool               hasSuffixAccents;
          int16_t             defaultShiftSize;
          int16_t             shift[MAX_TABLE_SIZE_];
          int16_t             backShift[MAX_TABLE_SIZE_];
};

struct UStringSearch {
    struct USearch            *search;
    struct UPattern            pattern;
    const  UCollator          *collator;
    const  icu::Normalizer2   *nfd;
    // positions within the collation element iterator is used to determine
    // if we are at the start of the text.
           UCollationElements *textIter;
           icu::UCollationPCE *textProcessedIter;
    // utility collation element, used throughout program for temporary 
    // iteration.
           UCollationElements *utilIter;
           UBool               ownCollator;
           UCollationStrength  strength;
           uint32_t            ceMask;
           uint32_t            variableTop;
           UBool               toShift;
           UChar               canonicalPrefixAccents[INITIAL_ARRAY_SIZE_];
           UChar               canonicalSuffixAccents[INITIAL_ARRAY_SIZE_];
};

/**
* Exact matches without checking for the ends for extra accents.
* The match after the position within the collation element iterator is to be
* found. 
* After a match is found the offset in the collation element iterator will be
* shifted to the start of the match.
* Implementation note: 
* For tertiary we can't use the collator->tertiaryMask, that is a 
* preprocessed mask that takes into account case options. since we are only 
* concerned with exact matches, we don't need that.
* Alternate handling - since only the 16 most significant digits is only used, 
* we can safely do a compare without masking if the ce is a variable, we mask 
* and get only the primary values no shifting to quartenary is required since 
* all primary values less than variabletop will need to be masked off anyway.
* If the end character is composite and the pattern ce does not match the text 
* ce, we skip it until we find a match in the end composite character or when 
* it has passed the character. This is so that we can match pattern "a" with
* the text "\u00e6" 
* @param strsrch string search data
* @param status error status if any
* @return TRUE if an exact match is found, FALSE otherwise
*/
U_CFUNC
UBool usearch_handleNextExact(UStringSearch *strsrch, UErrorCode *status);

/**
* Canonical matches.
* According to the definition, matches found here will include the whole span 
* of beginning and ending accents if it overlaps that region.
* @param strsrch string search data
* @param status error status if any
* @return TRUE if a canonical match is found, FALSE otherwise
*/
U_CFUNC
UBool usearch_handleNextCanonical(UStringSearch *strsrch, UErrorCode *status);

/**
* Gets the previous match.
* Comments follows from handleNextExact
* @param strsrch string search data
* @param status error status if any
* @return True if a exact math is found, FALSE otherwise.
*/
U_CFUNC
UBool usearch_handlePreviousExact(UStringSearch *strsrch, UErrorCode *status);

/**
* Canonical matches.
* According to the definition, matches found here will include the whole span 
* of beginning and ending accents if it overlaps that region.
* @param strsrch string search data
* @param status error status if any
* @return TRUE if a canonical match is found, FALSE otherwise
*/
U_CFUNC
UBool usearch_handlePreviousCanonical(UStringSearch *strsrch, 
                                      UErrorCode    *status);

#endif /* #if !UCONFIG_NO_COLLATION */

#endif
Commit	Line	Data
b75a7d8f A	1	/*
b75a7d8f A	2	**********************************************************************
2ca993e8	3	* Copyright (C) 2001-2015 IBM and others. All rights reserved.
b75a7d8f A	4	**********************************************************************
	5	* Date Name Description
	6	* 08/13/2001 synwee Creation.
	7	**********************************************************************
	8	*/
	9	#ifndef USRCHIMP_H
	10	#define USRCHIMP_H
	11
	12	#include "unicode/utypes.h"
	13
	14	#if !UCONFIG_NO_COLLATION
	15
729e4ab9	16	#include "unicode/normalizer2.h"
b75a7d8f A	17	#include "unicode/ucol.h"
	18	#include "unicode/ucoleitr.h"
	19	#include "unicode/ubrk.h"
	20
57a6839d A	21	/* mask off anything but primary order */
	22	#define UCOL_PRIMARYORDERMASK 0xffff0000
	23	/* mask off anything but secondary order */
	24	#define UCOL_SECONDARYORDERMASK 0x0000ff00
	25	/* mask off anything but tertiary order */
	26	#define UCOL_TERTIARYORDERMASK 0x000000ff
	27	/* primary order shift */
	28	#define UCOL_PRIMARYORDERSHIFT 16
	29	/* secondary order shift */
	30	#define UCOL_SECONDARYORDERSHIFT 8
	31
	32	#define UCOL_IGNORABLE 0
	33
	34	/* get weights from a CE */
	35	#define UCOL_PRIMARYORDER(order) (((order) >> 16) & 0xffff)
	36	#define UCOL_SECONDARYORDER(order) (((order) & UCOL_SECONDARYORDERMASK)>> UCOL_SECONDARYORDERSHIFT)
	37	#define UCOL_TERTIARYORDER(order) ((order) & UCOL_TERTIARYORDERMASK)
	38
	39	#define UCOL_CONTINUATION_MARKER 0xC0
	40
	41	#define isContinuation(CE) (((CE) & UCOL_CONTINUATION_MARKER) == UCOL_CONTINUATION_MARKER)
	42
	43	/**
	44	* This indicates an error has occured during processing or there are no more CEs
	45	* to be returned.
	46	*/
	47	#ifndef UCOL_PROCESSED_NULLORDER
	48	#define UCOL_PROCESSED_NULLORDER ((int64_t)U_INT64_MAX)
	49	#endif
	50
	51	U_NAMESPACE_BEGIN
	52
	53	class CollationElementIterator;
	54	class Collator;
	55
	56	struct PCEI
	57	{
	58	uint64_t ce;
	59	int32_t low;
	60	int32_t high;
	61	};
	62
	63	struct PCEBuffer
	64	{
	65	PCEI defaultBuffer[16];
	66	PCEI *buffer;
	67	int32_t bufferIndex;
	68	int32_t bufferSize;
	69
	70	PCEBuffer();
	71	~PCEBuffer();
	72
	73	void reset();
2ca993e8 A	74	UBool isEmpty() const;
2ca993e8 A	75	void put(uint64_t ce, int32_t ixLow, int32_t ixHigh, UErrorCode &errorCode);
57a6839d A	76	const PCEI *get();
	77	};
	78
	79	class UCollationPCE : public UMemory {
	80	private:
	81	PCEBuffer pceBuffer;
	82	CollationElementIterator *cei;
	83	UCollationStrength strength;
	84	UBool toShift;
	85	UBool isShifted;
	86	uint32_t variableTop;
	87
	88	public:
	89	UCollationPCE(UCollationElements *elems);
	90	UCollationPCE(CollationElementIterator *iter);
	91	~UCollationPCE();
	92
	93	void init(UCollationElements *elems);
	94	void init(CollationElementIterator *iter);
	95
	96	/**
	97	* Get the processed ordering priority of the next collation element in the text.
	98	* A single character may contain more than one collation element.
	99	*
	100	* @param ixLow a pointer to an int32_t to receive the iterator index before fetching the CE.
	101	* @param ixHigh a pointer to an int32_t to receive the iterator index after fetching the CE.
	102	* @param status A pointer to an UErrorCode to receive any errors.
	103	* @return The next collation elements ordering, otherwise returns UCOL_PROCESSED_NULLORDER
	104	* if an error has occured or if the end of string has been reached
	105	*/
	106	int64_t nextProcessed(int32_t ixLow, int32_t ixHigh, UErrorCode *status);
	107	/**
	108	* Get the processed ordering priority of the previous collation element in the text.
	109	* A single character may contain more than one collation element.
	110	*
	111	* @param ixLow A pointer to an int32_t to receive the iterator index after fetching the CE
	112	* @param ixHigh A pointer to an int32_t to receiver the iterator index before fetching the CE
	113	* @param status A pointer to an UErrorCode to receive any errors. Noteably
	114	* a U_BUFFER_OVERFLOW_ERROR is returned if the internal stack
	115	* buffer has been exhausted.
	116	* @return The previous collation elements ordering, otherwise returns
	117	* UCOL_PROCESSED_NULLORDER if an error has occured or if the start of
	118	* string has been reached.
	119	*/
	120	int64_t previousProcessed(int32_t ixLow, int32_t ixHigh, UErrorCode *status);
	121
	122	private:
	123	void init(const Collator &coll);
	124	uint64_t processCE(uint32_t ce);
	125	};
	126
	127	U_NAMESPACE_END
	128
b75a7d8f A	129	#define INITIAL_ARRAY_SIZE_ 256
	130	#define MAX_TABLE_SIZE_ 257
	131
	132	struct USearch {
	133	// required since collation element iterator does not have a getText API
	134	const UChar *text;
	135	int32_t textLength; // exact length
	136	UBool isOverlap;
	137	UBool isCanonicalMatch;
729e4ab9 A	138	int16_t elementComparisonType;
729e4ab9 A	139	UBreakIterator *internalBreakIter; //internal character breakiterator
b75a7d8f A	140	UBreakIterator *breakIter;
	141	// value USEARCH_DONE is the default value
	142	// if we are not at the start of the text or the end of the text,
	143	// depending on the iteration direction and matchedIndex is USEARCH_DONE
46f4442e A	144	// it means that we can't find any more matches in that particular direction
46f4442e A	145	int32_t matchedIndex;
b75a7d8f A	146	int32_t matchedLength;
	147	UBool isForwardSearching;
	148	UBool reset;
	149	};
	150
	151	struct UPattern {
	152	const UChar *text;
	153	int32_t textLength; // exact length
	154	// length required for backwards ce comparison
b331163b A	155	int32_t cesLength;
	156	int32_t *ces;
	157	int32_t cesBuffer[INITIAL_ARRAY_SIZE_];
	158	int32_t pcesLength;
	159	int64_t *pces;
	160	int64_t pcesBuffer[INITIAL_ARRAY_SIZE_];
b75a7d8f A	161	UBool hasPrefixAccents;
	162	UBool hasSuffixAccents;
	163	int16_t defaultShiftSize;
	164	int16_t shift[MAX_TABLE_SIZE_];
	165	int16_t backShift[MAX_TABLE_SIZE_];
	166	};
	167
	168	struct UStringSearch {
	169	struct USearch *search;
	170	struct UPattern pattern;
	171	const UCollator *collator;
4388f060	172	const icu::Normalizer2 *nfd;
b75a7d8f A	173	// positions within the collation element iterator is used to determine
	174	// if we are at the start of the text.
	175	UCollationElements *textIter;
57a6839d	176	icu::UCollationPCE *textProcessedIter;
b75a7d8f A	177	// utility collation element, used throughout program for temporary
	178	// iteration.
	179	UCollationElements *utilIter;
	180	UBool ownCollator;
	181	UCollationStrength strength;
	182	uint32_t ceMask;
	183	uint32_t variableTop;
	184	UBool toShift;
	185	UChar canonicalPrefixAccents[INITIAL_ARRAY_SIZE_];
	186	UChar canonicalSuffixAccents[INITIAL_ARRAY_SIZE_];
	187	};
	188
	189	/**
	190	* Exact matches without checking for the ends for extra accents.
	191	* The match after the position within the collation element iterator is to be
	192	* found.
	193	* After a match is found the offset in the collation element iterator will be
	194	* shifted to the start of the match.
	195	* Implementation note:
	196	* For tertiary we can't use the collator->tertiaryMask, that is a
	197	* preprocessed mask that takes into account case options. since we are only
	198	* concerned with exact matches, we don't need that.
	199	* Alternate handling - since only the 16 most significant digits is only used,
	200	* we can safely do a compare without masking if the ce is a variable, we mask
	201	* and get only the primary values no shifting to quartenary is required since
	202	* all primary values less than variabletop will need to be masked off anyway.
	203	* If the end character is composite and the pattern ce does not match the text
	204	* ce, we skip it until we find a match in the end composite character or when
	205	* it has passed the character. This is so that we can match pattern "a" with
	206	* the text "\u00e6"
	207	* @param strsrch string search data
	208	* @param status error status if any
	209	* @return TRUE if an exact match is found, FALSE otherwise
	210	*/
	211	U_CFUNC
	212	UBool usearch_handleNextExact(UStringSearch strsrch, UErrorCode status);
	213
	214	/**
	215	* Canonical matches.
	216	* According to the definition, matches found here will include the whole span
	217	* of beginning and ending accents if it overlaps that region.
	218	* @param strsrch string search data
	219	* @param status error status if any
	220	* @return TRUE if a canonical match is found, FALSE otherwise
	221	*/
	222	U_CFUNC
	223	UBool usearch_handleNextCanonical(UStringSearch strsrch, UErrorCode status);
	224
	225	/**
	226	* Gets the previous match.
	227	* Comments follows from handleNextExact
	228	* @param strsrch string search data
	229	* @param status error status if any
	230	* @return True if a exact math is found, FALSE otherwise.
	231	*/
	232	U_CFUNC
	233	UBool usearch_handlePreviousExact(UStringSearch strsrch, UErrorCode status);
	234
	235	/**
	236	* Canonical matches.
	237	* According to the definition, matches found here will include the whole span
	238	* of beginning and ending accents if it overlaps that region.
	239	* @param strsrch string search data
	240	* @param status error status if any
241	* @return TRUE if a canonical match is found, FALSE otherwise
242	*/
243	U_CFUNC
244	UBool usearch_handlePreviousCanonical(UStringSearch *strsrch,
245	UErrorCode *status);
246
247	#endif /* #if !UCONFIG_NO_COLLATION */
248
249	#endif