icuSources/common/rbbidata.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1999-2005,2008 International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  rbbidata.h
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   RBBI data formats  Includes
  14 *
  15 *                          Structs that describes the format of the Binary RBBI data,
  16 *                          as it is stored in ICU's data file.
  17 *
  18 *      RBBIDataWrapper  -  Instances of this class sit between the
  19 *                          raw data structs and the RulesBasedBreakIterator objects
  20 *                          that are created by applications.  The wrapper class
  21 *                          provides reference counting for the underlying data,
  22 *                          and direct pointers to data that would not otherwise
  23 *                          be accessible without ugly pointer arithmetic.  The
  24 *                          wrapper does not attempt to provide any higher level
  25 *                          abstractions for the data itself.
  26 *
  27 *                          There will be only one instance of RBBIDataWrapper for any
  28 *                          set of RBBI run time data being shared by instances
  29 *                          (clones) of RulesBasedBreakIterator.
  30 */
  31
  32 #ifndef __RBBIDATA_H__
  33 #define __RBBIDATA_H__
  34
  35 #include "unicode/utypes.h"
  36 #include "unicode/udata.h"
  37 #include "udataswp.h"
  38
  39 /**
  40  * Swap RBBI data. See udataswp.h.
  41  * @internal
  42  */
  43 U_CAPI int32_t U_EXPORT2
  44 ubrk_swap(const UDataSwapper *ds,
  45           const void *inData, int32_t length, void *outData,
  46           UErrorCode *pErrorCode);
  47
  48 #ifdef XP_CPLUSPLUS
  49
  50 #include "unicode/uobject.h"
  51 #include "unicode/unistr.h"
  52 #include "utrie.h"
  53
  54 U_NAMESPACE_BEGIN
  55
  56 /*
  57  *   The following structs map exactly onto the raw data from ICU common data file.
  58  */
  59 struct RBBIDataHeader {
  60     uint32_t         fMagic;           /*  == 0xbla0                                               */
  61     uint8_t          fFormatVersion[4]; /* Data Format.  Same as the value in struct UDataInfo      */
  62                                        /*   if there is one associated with this data.             */
  63                                        /*     (version originates in rbbi, is copied to UDataInfo) */
  64                                        /*   For ICU 3.2 and earlier, this field was                */
  65                                        /*       uint32_t  fVersion                                 */
  66                                        /*   with a value of 1.                                     */
  67     uint32_t         fLength;          /*  Total length in bytes of this RBBI Data,                */
  68                                        /*      including all sections, not just the header.        */
  69     uint32_t         fCatCount;        /*  Number of character categories.                         */
  70
  71     /*                                                                        */
  72     /*  Offsets and sizes of each of the subsections within the RBBI data.    */
  73     /*  All offsets are bytes from the start of the RBBIDataHeader.           */
  74     /*  All sizes are in bytes.                                               */
  75     /*                                                                        */
  76     uint32_t         fFTable;         /*  forward state transition table. */
  77     uint32_t         fFTableLen;
  78     uint32_t         fRTable;         /*  Offset to the reverse state transition table. */
  79     uint32_t         fRTableLen;
  80     uint32_t         fSFTable;        /*  safe point forward transition table */
  81     uint32_t         fSFTableLen;
  82     uint32_t         fSRTable;        /*  safe point reverse transition table */
  83     uint32_t         fSRTableLen;
  84     uint32_t         fTrie;           /*  Offset to Trie data for character categories */
  85     uint32_t         fTrieLen;
  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;
  90
  91     uint32_t         fReserved[6];    /*  Reserved for expansion */
  92
  93 };
  94
  95
  96
  97 struct  RBBIStateTableRow {
  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)                      */
 114     int16_t          fReserved;
 115     uint16_t         fNextState[2]; /*  Next State, indexed by char category.             */
 116                                     /*    Array Size is fNumCols from the                 */
 117                                     /*    state table header.                             */
 118                                     /*    CAUTION:  see RBBITableBuilder::getTableSize()  */
 119                                     /*              before changing anything here.        */
 120 };
 121
 122
 123 struct RBBIStateTable {
 124     uint32_t         fNumStates;    /*  Number of states.                                 */
 125     uint32_t         fRowLen;       /*  Length of a state table row, in bytes.            */
 126     uint32_t         fFlags;        /*  Option Flags for this state table                 */
 127     uint32_t         fReserved;     /*  reserved                                          */
 128     char             fTableData[4]; /*  First RBBIStateTableRow begins here.              */
 129                                     /*    (making it char[] simplifies ugly address       */
 130                                     /*     arithmetic for indexing variable length rows.) */
 131 };
 132
 133 typedef enum {
 134     RBBI_LOOKAHEAD_HARD_BREAK = 1,
 135     RBBI_BOF_REQUIRED = 2
 136 } RBBIStateTableFlags;
 137
 138
 139 /*                                        */
 140 /*   The reference counting wrapper class */
 141 /*                                        */
 142 class RBBIDataWrapper : public UMemory {
 143 public:
 144     enum EDontAdopt {
 145         kDontAdopt
 146     };
 147     RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status);
 148     RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt dontAdopt, UErrorCode &status);
 149     RBBIDataWrapper(UDataMemory* udm, UErrorCode &status);
 150     ~RBBIDataWrapper();
 151
 152     void                  init(const RBBIDataHeader *data, UErrorCode &status);
 153     RBBIDataWrapper      *addReference();
 154     void                  removeReference();
 155     UBool                 operator ==(const RBBIDataWrapper &other) const;
 156     int32_t               hashCode();
 157     const UnicodeString  &getRuleSourceString() const;
 158 #ifdef RBBI_DEBUG
 159     void                  printData();
 160     void                  printTable(const char *heading, const RBBIStateTable *table);
 161 #else
 162     #define printData()
 163     #define printTable(heading, table)
 164 #endif
 165
 166     /*                                     */
 167     /*   Pointers to items within the data */
 168     /*                                     */
 169     const RBBIDataHeader     *fHeader;
 170     const RBBIStateTable     *fForwardTable;
 171     const RBBIStateTable     *fReverseTable;
 172     const RBBIStateTable     *fSafeFwdTable;
 173     const RBBIStateTable     *fSafeRevTable;
 174     const UChar              *fRuleSource;
 175     const int32_t            *fRuleStatusTable;
 176
 177     /* number of int32_t values in the rule status table.   Used to sanity check indexing */
 178     int32_t             fStatusMaxIdx;
 179
 180     UTrie               fTrie;
 181
 182 private:
 183     int32_t             fRefCount;
 184     UDataMemory        *fUDataMem;
 185     UnicodeString       fRuleString;
 186     UBool               fDontFreeData;
 187
 188     RBBIDataWrapper(const RBBIDataWrapper &other); /*  forbid copying of this class */
 189     RBBIDataWrapper &operator=(const RBBIDataWrapper &other); /*  forbid copying of this class */
 190 };
 191
 192
 193
 194 U_NAMESPACE_END
 195
 196 #endif /* C++ */
 197
 198 #endif