[apple/icu.git] / icuSources / common / rbbidata.h

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*
*   Copyright (C) 1999-2014 International Business Machines
*   Corporation and others.  All Rights Reserved.
*
*******************************************************************************
*   file name:  rbbidata.h
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   RBBI data formats  Includes
*
*                          Structs that describes the format of the Binary RBBI data,
*                          as it is stored in ICU's data file.
*
*      RBBIDataWrapper  -  Instances of this class sit between the
*                          raw data structs and the RulesBasedBreakIterator objects
*                          that are created by applications.  The wrapper class
*                          provides reference counting for the underlying data,
*                          and direct pointers to data that would not otherwise
*                          be accessible without ugly pointer arithmetic.  The
*                          wrapper does not attempt to provide any higher level
*                          abstractions for the data itself.
*
*                          There will be only one instance of RBBIDataWrapper for any
*                          set of RBBI run time data being shared by instances
*                          (clones) of RulesBasedBreakIterator.
*/

#ifndef __RBBIDATA_H__
#define __RBBIDATA_H__

#include "unicode/utypes.h"
#include "unicode/udata.h"
#include "udataswp.h"

/**
 * Swap RBBI data. See udataswp.h.
 * @internal
 */
U_CAPI int32_t U_EXPORT2
ubrk_swap(const UDataSwapper *ds,
          const void *inData, int32_t length, void *outData,
          UErrorCode *pErrorCode);

#ifdef __cplusplus

#include "unicode/uobject.h"
#include "unicode/unistr.h"
#include "unicode/uversion.h"
#include "umutex.h"
#include "utrie2.h"

U_NAMESPACE_BEGIN

// The current RBBI data format version.
static const uint8_t RBBI_DATA_FORMAT_VERSION[] = {5, 0, 0, 0};

/*  
 *   The following structs map exactly onto the raw data from ICU common data file. 
 */
struct RBBIDataHeader {
    uint32_t         fMagic;           /*  == 0xbla0                                               */
    UVersionInfo     fFormatVersion;   /* Data Format.  Same as the value in struct UDataInfo      */
                                       /*   if there is one associated with this data.             */
                                       /*     (version originates in rbbi, is copied to UDataInfo) */
    uint32_t         fLength;          /*  Total length in bytes of this RBBI Data,                */
                                       /*      including all sections, not just the header.        */
    uint32_t         fCatCount;        /*  Number of character categories.                         */

    /*                                                                        */
    /*  Offsets and sizes of each of the subsections within the RBBI data.    */
    /*  All offsets are bytes from the start of the RBBIDataHeader.           */
    /*  All sizes are in bytes.                                               */
    /*                                                                        */
    uint32_t         fFTable;         /*  forward state transition table. */
    uint32_t         fFTableLen;
    uint32_t         fRTable;         /*  Offset to the reverse state transition table. */
    uint32_t         fRTableLen;
    uint32_t         fTrie;           /*  Offset to Trie data for character categories */
    uint32_t         fTrieLen;
    uint32_t         fRuleSource;     /*  Offset to the source for for the break */
    uint32_t         fRuleSourceLen;  /*    rules.  Stored UChar *. */
    uint32_t         fStatusTable;    /* Offset to the table of rule status values */
    uint32_t         fStatusTableLen;

    uint32_t         fReserved[6];    /*  Reserved for expansion */

};


struct  RBBIStateTableRow {
    int16_t          fAccepting;    /*  Non-zero if this row is for an accepting state.   */
                                    /*  Value 0: not an accepting state.                  */
                                    /*       -1: Unconditional Accepting state.           */
                                    /*    positive:  Look-ahead match has completed.      */
                                    /*           Actual boundary position happened earlier */
                                    /*           Value here == fLookAhead in earlier      */
                                    /*              state, at actual boundary pos.        */
    int16_t          fLookAhead;    /*  Non-zero if this row is for a state that          */
                                    /*    corresponds to a '/' in the rule source.        */
                                    /*    Value is the same as the fAccepting             */
                                    /*      value for the rule (which will appear         */
                                    /*      in a different state.                         */
    int16_t          fTagIdx;       /*  Non-zero if this row covers a {tagged} position   */
                                    /*     from a rule.  Value is the index in the        */
                                    /*     StatusTable of the set of matching             */
                                    /*     tags (rule status values)                      */
    int16_t          fReserved;
    uint16_t         fNextState[1]; /*  Next State, indexed by char category.             */
                                    /*    Variable-length array declared with length 1    */
                                    /*    to disable bounds checkers.                     */
                                    /*    Array Size is actually fData->fHeader->fCatCount*/
                                    /*    CAUTION:  see RBBITableBuilder::getTableSize()  */
                                    /*              before changing anything here.        */
};


struct RBBIStateTable {
    uint32_t         fNumStates;    /*  Number of states.                                 */
    uint32_t         fRowLen;       /*  Length of a state table row, in bytes.            */
    uint32_t         fFlags;        /*  Option Flags for this state table                 */
    uint32_t         fReserved;     /*  reserved                                          */
    char             fTableData[1]; /*  First RBBIStateTableRow begins here.              */
                                    /*    Variable-length array declared with length 1    */
                                    /*    to disable bounds checkers.                     */
                                    /*    (making it char[] simplifies ugly address       */
                                    /*     arithmetic for indexing variable length rows.) */
};

typedef enum {
    RBBI_LOOKAHEAD_HARD_BREAK = 1,
    RBBI_BOF_REQUIRED = 2
} RBBIStateTableFlags;


/*                                        */
/*   The reference counting wrapper class */
/*                                        */
class RBBIDataWrapper : public UMemory {
public:
    enum EDontAdopt {
        kDontAdopt
    };
    RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status);
    RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt dontAdopt, UErrorCode &status);
    RBBIDataWrapper(UDataMemory* udm, UErrorCode &status);
    ~RBBIDataWrapper();

    static UBool          isDataVersionAcceptable(const UVersionInfo version);

    void                  init0();
    void                  init(const RBBIDataHeader *data, UErrorCode &status);
    RBBIDataWrapper      *addReference();
    void                  removeReference();
    UBool                 operator ==(const RBBIDataWrapper &other) const;
    int32_t               hashCode();
    const UnicodeString  &getRuleSourceString() const;
    void                  printData();
    void                  printTable(const char *heading, const RBBIStateTable *table);

    /*                                     */
    /*   Pointers to items within the data */
    /*                                     */
    const RBBIDataHeader     *fHeader;
    const RBBIStateTable     *fForwardTable;
    const RBBIStateTable     *fReverseTable;
    const UChar              *fRuleSource;
    const int32_t            *fRuleStatusTable; 

    /* number of int32_t values in the rule status table.   Used to sanity check indexing */
    int32_t             fStatusMaxIdx;

    UTrie2             *fTrie;

private:
    u_atomic_int32_t    fRefCount;
    UDataMemory        *fUDataMem;
    UnicodeString       fRuleString;
    UBool               fDontFreeData;

    RBBIDataWrapper(const RBBIDataWrapper &other); /*  forbid copying of this class */
    RBBIDataWrapper &operator=(const RBBIDataWrapper &other); /*  forbid copying of this class */
};


U_NAMESPACE_END

#endif /* C++ */

#endif
Commit	Line	Data
f3c0d7a5 A	1	// © 2016 and later: Unicode, Inc. and others.
f3c0d7a5 A	2	// License & terms of use: http://www.unicode.org/copyright.html
374ca955 A	3	/*
	4	*******************************************************************************
	5	*
b331163b	6	* Copyright (C) 1999-2014 International Business Machines
374ca955 A	7	* Corporation and others. All Rights Reserved.
	8	*
	9	*******************************************************************************
	10	* file name: rbbidata.h
f3c0d7a5	11	* encoding: UTF-8
374ca955 A	12	* tab size: 8 (not used)
	13	* indentation:4
	14	*
	15	* RBBI data formats Includes
	16	*
	17	* Structs that describes the format of the Binary RBBI data,
	18	* as it is stored in ICU's data file.
	19	*
	20	* RBBIDataWrapper - Instances of this class sit between the
	21	* raw data structs and the RulesBasedBreakIterator objects
	22	* that are created by applications. The wrapper class
	23	* provides reference counting for the underlying data,
	24	* and direct pointers to data that would not otherwise
	25	* be accessible without ugly pointer arithmetic. The
	26	* wrapper does not attempt to provide any higher level
	27	* abstractions for the data itself.
	28	*
	29	* There will be only one instance of RBBIDataWrapper for any
	30	* set of RBBI run time data being shared by instances
	31	* (clones) of RulesBasedBreakIterator.
	32	*/
b75a7d8f A	33
	34	#ifndef __RBBIDATA_H__
	35	#define __RBBIDATA_H__
	36
	37	#include "unicode/utypes.h"
374ca955 A	38	#include "unicode/udata.h"
	39	#include "udataswp.h"
	40
	41	/**
	42	* Swap RBBI data. See udataswp.h.
	43	* @internal
	44	*/
	45	U_CAPI int32_t U_EXPORT2
	46	ubrk_swap(const UDataSwapper *ds,
	47	const void inData, int32_t length, void outData,
	48	UErrorCode *pErrorCode);
	49
4388f060	50	#ifdef __cplusplus
374ca955	51
b75a7d8f A	52	#include "unicode/uobject.h"
b75a7d8f A	53	#include "unicode/unistr.h"
0f5d89e8	54	#include "unicode/uversion.h"
57a6839d	55	#include "umutex.h"
0f5d89e8	56	#include "utrie2.h"
b75a7d8f	57
b75a7d8f A	58	U_NAMESPACE_BEGIN
b75a7d8f A	59
0f5d89e8 A	60	// The current RBBI data format version.
	61	static const uint8_t RBBI_DATA_FORMAT_VERSION[] = {5, 0, 0, 0};
	62
374ca955 A	63	/*
	64	* The following structs map exactly onto the raw data from ICU common data file.
	65	*/
b75a7d8f	66	struct RBBIDataHeader {
73c04bcf	67	uint32_t fMagic; /* == 0xbla0 */
0f5d89e8	68	UVersionInfo fFormatVersion; /* Data Format. Same as the value in struct UDataInfo */
73c04bcf A	69	/* if there is one associated with this data. */
73c04bcf A	70	/* (version originates in rbbi, is copied to UDataInfo) */
73c04bcf A	71	uint32_t fLength; /* Total length in bytes of this RBBI Data, */
	72	/* including all sections, not just the header. */
	73	uint32_t fCatCount; /* Number of character categories. */
	74
	75	/* */
	76	/* Offsets and sizes of each of the subsections within the RBBI data. */
	77	/* All offsets are bytes from the start of the RBBIDataHeader. */
	78	/* All sizes are in bytes. */
	79	/* */
374ca955	80	uint32_t fFTable; /* forward state transition table. */
b75a7d8f	81	uint32_t fFTableLen;
374ca955	82	uint32_t fRTable; /* Offset to the reverse state transition table. */
b75a7d8f	83	uint32_t fRTableLen;
374ca955	84	uint32_t fTrie; /* Offset to Trie data for character categories */
b75a7d8f	85	uint32_t fTrieLen;
374ca955 A	86	uint32_t fRuleSource; /* Offset to the source for for the break */
	87	uint32_t fRuleSourceLen; /* rules. Stored UChar . /
	88	uint32_t fStatusTable; /* Offset to the table of rule status values */
	89	uint32_t fStatusTableLen;
b75a7d8f	90
374ca955	91	uint32_t fReserved[6]; /* Reserved for expansion */
b75a7d8f A	92
	93	};
	94
	95
	96
	97	struct RBBIStateTableRow {
374ca955 A	98	int16_t fAccepting; /* Non-zero if this row is for an accepting state. */
	99	/* Value 0: not an accepting state. */
	100	/* -1: Unconditional Accepting state. */
	101	/* positive: Look-ahead match has completed. */
	102	/* Actual boundary position happened earlier */
	103	/* Value here == fLookAhead in earlier */
	104	/* state, at actual boundary pos. */
	105	int16_t fLookAhead; /* Non-zero if this row is for a state that */
	106	/* corresponds to a '/' in the rule source. */
	107	/* Value is the same as the fAccepting */
	108	/* value for the rule (which will appear */
	109	/* in a different state. */
	110	int16_t fTagIdx; /* Non-zero if this row covers a {tagged} position */
	111	/* from a rule. Value is the index in the */
	112	/* StatusTable of the set of matching */
	113	/* tags (rule status values) */
b75a7d8f	114	int16_t fReserved;
0f5d89e8 A	115	uint16_t fNextState[1]; /* Next State, indexed by char category. */
	116	/* Variable-length array declared with length 1 */
	117	/* to disable bounds checkers. */
	118	/* Array Size is actually fData->fHeader->fCatCount*/
374ca955 A	119	/* CAUTION: see RBBITableBuilder::getTableSize() */
374ca955 A	120	/* before changing anything here. */
b75a7d8f A	121	};
	122
	123
	124	struct RBBIStateTable {
374ca955 A	125	uint32_t fNumStates; /* Number of states. */
	126	uint32_t fRowLen; /* Length of a state table row, in bytes. */
	127	uint32_t fFlags; /* Option Flags for this state table */
	128	uint32_t fReserved; /* reserved */
0f5d89e8 A	129	char fTableData[1]; /* First RBBIStateTableRow begins here. */
	130	/* Variable-length array declared with length 1 */
	131	/* to disable bounds checkers. */
374ca955 A	132	/* (making it char[] simplifies ugly address */
374ca955 A	133	/* arithmetic for indexing variable length rows.) */
b75a7d8f A	134	};
b75a7d8f A	135
374ca955	136	typedef enum {
73c04bcf A	137	RBBI_LOOKAHEAD_HARD_BREAK = 1,
73c04bcf A	138	RBBI_BOF_REQUIRED = 2
374ca955 A	139	} RBBIStateTableFlags;
374ca955 A	140
b75a7d8f	141
374ca955 A	142	/* */
	143	/* The reference counting wrapper class */
	144	/* */
b75a7d8f A	145	class RBBIDataWrapper : public UMemory {
b75a7d8f A	146	public:
46f4442e A	147	enum EDontAdopt {
	148	kDontAdopt
	149	};
b75a7d8f	150	RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status);
46f4442e	151	RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt dontAdopt, UErrorCode &status);
b75a7d8f A	152	RBBIDataWrapper(UDataMemory* udm, UErrorCode &status);
	153	~RBBIDataWrapper();
	154
0f5d89e8 A	155	static UBool isDataVersionAcceptable(const UVersionInfo version);
0f5d89e8 A	156
b331163b	157	void init0();
b75a7d8f A	158	void init(const RBBIDataHeader *data, UErrorCode &status);
	159	RBBIDataWrapper *addReference();
	160	void removeReference();
	161	UBool operator ==(const RBBIDataWrapper &other) const;
	162	int32_t hashCode();
374ca955	163	const UnicodeString &getRuleSourceString() const;
b75a7d8f	164	void printData();
374ca955	165	void printTable(const char heading, const RBBIStateTable table);
b75a7d8f	166
374ca955 A	167	/* */
	168	/* Pointers to items within the data */
	169	/* */
b75a7d8f A	170	const RBBIDataHeader *fHeader;
	171	const RBBIStateTable *fForwardTable;
	172	const RBBIStateTable *fReverseTable;
	173	const UChar *fRuleSource;
374ca955 A	174	const int32_t *fRuleStatusTable;
	175
	176	/* number of int32_t values in the rule status table. Used to sanity check indexing */
	177	int32_t fStatusMaxIdx;
b75a7d8f	178
0f5d89e8	179	UTrie2 *fTrie;
b75a7d8f A	180
b75a7d8f A	181	private:
57a6839d	182	u_atomic_int32_t fRefCount;
0f5d89e8	183	UDataMemory *fUDataMem;
b75a7d8f	184	UnicodeString fRuleString;
46f4442e	185	UBool fDontFreeData;
b75a7d8f	186
374ca955 A	187	RBBIDataWrapper(const RBBIDataWrapper &other); /* forbid copying of this class */
374ca955 A	188	RBBIDataWrapper &operator=(const RBBIDataWrapper &other); /* forbid copying of this class */
b75a7d8f A	189	};
b75a7d8f A	190
374ca955 A	191
374ca955 A	192
b75a7d8f A	193	U_NAMESPACE_END
b75a7d8f A	194
374ca955	195	#endif /* C++ */
b75a7d8f	196
374ca955	197	#endif