icuSources/common/rbbidata.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 1999-2014 International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  rbbidata.h
  11 *   encoding:   UTF-8
  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 */
  33
  34 #ifndef __RBBIDATA_H__
  35 #define __RBBIDATA_H__
  36
  37 #include "unicode/utypes.h"
  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
  50 #ifdef __cplusplus
  51
  52 #include "unicode/uobject.h"
  53 #include "unicode/unistr.h"
  54 #include "umutex.h"
  55 #include "utrie.h"
  56
  57 U_NAMESPACE_BEGIN
  58
  59 /*
  60  *   The following structs map exactly onto the raw data from ICU common data file.
  61  */
  62 struct RBBIDataHeader {
  63     uint32_t         fMagic;           /*  == 0xbla0                                               */
  64     uint8_t          fFormatVersion[4]; /* Data Format.  Same as the value in struct UDataInfo      */
  65                                        /*   if there is one associated with this data.             */
  66                                        /*     (version originates in rbbi, is copied to UDataInfo) */
  67                                        /*   For ICU 3.2 and earlier, this field was                */
  68                                        /*       uint32_t  fVersion                                 */
  69                                        /*   with a value of 1.                                     */
  70     uint32_t         fLength;          /*  Total length in bytes of this RBBI Data,                */
  71                                        /*      including all sections, not just the header.        */
  72     uint32_t         fCatCount;        /*  Number of character categories.                         */
  73
  74     /*                                                                        */
  75     /*  Offsets and sizes of each of the subsections within the RBBI data.    */
  76     /*  All offsets are bytes from the start of the RBBIDataHeader.           */
  77     /*  All sizes are in bytes.                                               */
  78     /*                                                                        */
  79     uint32_t         fFTable;         /*  forward state transition table. */
  80     uint32_t         fFTableLen;
  81     uint32_t         fRTable;         /*  Offset to the reverse state transition table. */
  82     uint32_t         fRTableLen;
  83     uint32_t         fSFTable;        /*  safe point forward transition table */
  84     uint32_t         fSFTableLen;
  85     uint32_t         fSRTable;        /*  safe point reverse transition table */
  86     uint32_t         fSRTableLen;
  87     uint32_t         fTrie;           /*  Offset to Trie data for character categories */
  88     uint32_t         fTrieLen;
  89     uint32_t         fRuleSource;     /*  Offset to the source for for the break */
  90     uint32_t         fRuleSourceLen;  /*    rules.  Stored UChar *. */
  91     uint32_t         fStatusTable;    /* Offset to the table of rule status values */
  92     uint32_t         fStatusTableLen;
  93
  94     uint32_t         fReserved[6];    /*  Reserved for expansion */
  95
  96 };
  97
  98
  99
 100 struct  RBBIStateTableRow {
 101     int16_t          fAccepting;    /*  Non-zero if this row is for an accepting state.   */
 102                                     /*  Value 0: not an accepting state.                  */
 103                                     /*       -1: Unconditional Accepting state.           */
 104                                     /*    positive:  Look-ahead match has completed.      */
 105                                     /*           Actual boundary position happened earlier */
 106                                     /*           Value here == fLookAhead in earlier      */
 107                                     /*              state, at actual boundary pos.        */
 108     int16_t          fLookAhead;    /*  Non-zero if this row is for a state that          */
 109                                     /*    corresponds to a '/' in the rule source.        */
 110                                     /*    Value is the same as the fAccepting             */
 111                                     /*      value for the rule (which will appear         */
 112                                     /*      in a different state.                         */
 113     int16_t          fTagIdx;       /*  Non-zero if this row covers a {tagged} position   */
 114                                     /*     from a rule.  Value is the index in the        */
 115                                     /*     StatusTable of the set of matching             */
 116                                     /*     tags (rule status values)                      */
 117     int16_t          fReserved;
 118     uint16_t         fNextState[2]; /*  Next State, indexed by char category.             */
 119                                     /*  This array does not have two elements             */
 120                                     /*    Array Size is actually fData->fHeader->fCatCount         */
 121                                     /*    CAUTION:  see RBBITableBuilder::getTableSize()  */
 122                                     /*              before changing anything here.        */
 123 };
 124
 125
 126 struct RBBIStateTable {
 127     uint32_t         fNumStates;    /*  Number of states.                                 */
 128     uint32_t         fRowLen;       /*  Length of a state table row, in bytes.            */
 129     uint32_t         fFlags;        /*  Option Flags for this state table                 */
 130     uint32_t         fReserved;     /*  reserved                                          */
 131     char             fTableData[4]; /*  First RBBIStateTableRow begins here.              */
 132                                     /*    (making it char[] simplifies ugly address       */
 133                                     /*     arithmetic for indexing variable length rows.) */
 134 };
 135
 136 typedef enum {
 137     RBBI_LOOKAHEAD_HARD_BREAK = 1,
 138     RBBI_BOF_REQUIRED = 2
 139 } RBBIStateTableFlags;
 140
 141
 142 /*                                        */
 143 /*   The reference counting wrapper class */
 144 /*                                        */
 145 class RBBIDataWrapper : public UMemory {
 146 public:
 147     enum EDontAdopt {
 148         kDontAdopt
 149     };
 150     RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status);
 151     RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt dontAdopt, UErrorCode &status);
 152     RBBIDataWrapper(UDataMemory* udm, UErrorCode &status);
 153     ~RBBIDataWrapper();
 154
 155     void                  init0();
 156     void                  init(const RBBIDataHeader *data, UErrorCode &status);
 157     RBBIDataWrapper      *addReference();
 158     void                  removeReference();
 159     UBool                 operator ==(const RBBIDataWrapper &other) const;
 160     int32_t               hashCode();
 161     const UnicodeString  &getRuleSourceString() const;
 162 #ifdef RBBI_DEBUG
 163     void                  printData();
 164     void                  printTable(const char *heading, const RBBIStateTable *table);
 165 #else
 166     #define printData()
 167     #define printTable(heading, table)
 168 #endif
 169
 170     /*                                     */
 171     /*   Pointers to items within the data */
 172     /*                                     */
 173     const RBBIDataHeader     *fHeader;
 174     const RBBIStateTable     *fForwardTable;
 175     const RBBIStateTable     *fReverseTable;
 176     const RBBIStateTable     *fSafeFwdTable;
 177     const RBBIStateTable     *fSafeRevTable;
 178     const UChar              *fRuleSource;
 179     const int32_t            *fRuleStatusTable;
 180
 181     /* number of int32_t values in the rule status table.   Used to sanity check indexing */
 182     int32_t             fStatusMaxIdx;
 183
 184     UTrie               fTrie;
 185
 186 private:
 187     u_atomic_int32_t    fRefCount;
 188     UDataMemory  *fUDataMem;
 189     UnicodeString       fRuleString;
 190     UBool               fDontFreeData;
 191
 192     RBBIDataWrapper(const RBBIDataWrapper &other); /*  forbid copying of this class */
 193     RBBIDataWrapper &operator=(const RBBIDataWrapper &other); /*  forbid copying of this class */
 194 };
 195
 196
 197
 198 U_NAMESPACE_END
 199
 200 #endif /* C++ */
 201
 202 #endif