1 /* 2 ******************************************************************************* 3 * 4 * Copyright (C) 1999-2014 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 __cplusplus 49 50 #include "unicode/uobject.h" 51 #include "unicode/unistr.h" 52 #include "umutex.h" 53 #include "utrie.h" 54 55 U_NAMESPACE_BEGIN 56 57 /* 58 * The following structs map exactly onto the raw data from ICU common data file. 59 */ 60 struct RBBIDataHeader { 61 uint32_t fMagic; /* == 0xbla0 */ 62 uint8_t fFormatVersion[4]; /* Data Format. Same as the value in struct UDataInfo */ 63 /* if there is one associated with this data. */ 64 /* (version originates in rbbi, is copied to UDataInfo) */ 65 /* For ICU 3.2 and earlier, this field was */ 66 /* uint32_t fVersion */ 67 /* with a value of 1. */ 68 uint32_t fLength; /* Total length in bytes of this RBBI Data, */ 69 /* including all sections, not just the header. */ 70 uint32_t fCatCount; /* Number of character categories. */ 71 72 /* */ 73 /* Offsets and sizes of each of the subsections within the RBBI data. */ 74 /* All offsets are bytes from the start of the RBBIDataHeader. */ 75 /* All sizes are in bytes. */ 76 /* */ 77 uint32_t fFTable; /* forward state transition table. */ 78 uint32_t fFTableLen; 79 uint32_t fRTable; /* Offset to the reverse state transition table. */ 80 uint32_t fRTableLen; 81 uint32_t fSFTable; /* safe point forward transition table */ 82 uint32_t fSFTableLen; 83 uint32_t fSRTable; /* safe point reverse transition table */ 84 uint32_t fSRTableLen; 85 uint32_t fTrie; /* Offset to Trie data for character categories */ 86 uint32_t fTrieLen; 87 uint32_t fRuleSource; /* Offset to the source for for the break */ 88 uint32_t fRuleSourceLen; /* rules. Stored UChar *. */ 89 uint32_t fStatusTable; /* Offset to the table of rule status values */ 90 uint32_t fStatusTableLen; 91 92 uint32_t fReserved[6]; /* Reserved for expansion */ 93 94 }; 95 96 97 98 struct RBBIStateTableRow { 99 int16_t fAccepting; /* Non-zero if this row is for an accepting state. */ 100 /* Value 0: not an accepting state. */ 101 /* -1: Unconditional Accepting state. */ 102 /* positive: Look-ahead match has completed. */ 103 /* Actual boundary position happened earlier */ 104 /* Value here == fLookAhead in earlier */ 105 /* state, at actual boundary pos. */ 106 int16_t fLookAhead; /* Non-zero if this row is for a state that */ 107 /* corresponds to a '/' in the rule source. */ 108 /* Value is the same as the fAccepting */ 109 /* value for the rule (which will appear */ 110 /* in a different state. */ 111 int16_t fTagIdx; /* Non-zero if this row covers a {tagged} position */ 112 /* from a rule. Value is the index in the */ 113 /* StatusTable of the set of matching */ 114 /* tags (rule status values) */ 115 int16_t fReserved; 116 uint16_t fNextState[2]; /* Next State, indexed by char category. */ 117 /* This array does not have two elements */ 118 /* Array Size is actually fData->fHeader->fCatCount */ 119 /* CAUTION: see RBBITableBuilder::getTableSize() */ 120 /* before changing anything here. */ 121 }; 122 123 124 struct RBBIStateTable { 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 */ 129 char fTableData[4]; /* First RBBIStateTableRow begins here. */ 130 /* (making it char[] simplifies ugly address */ 131 /* arithmetic for indexing variable length rows.) */ 132 }; 133 134 typedef enum { 135 RBBI_LOOKAHEAD_HARD_BREAK = 1, 136 RBBI_BOF_REQUIRED = 2 137 } RBBIStateTableFlags; 138 139 140 /* */ 141 /* The reference counting wrapper class */ 142 /* */ 143 class RBBIDataWrapper : public UMemory { 144 public: 145 enum EDontAdopt { 146 kDontAdopt 147 }; 148 RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status); 149 RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt dontAdopt, UErrorCode &status); 150 RBBIDataWrapper(UDataMemory* udm, UErrorCode &status); 151 ~RBBIDataWrapper(); 152 153 void init0(); 154 void init(const RBBIDataHeader *data, UErrorCode &status); 155 RBBIDataWrapper *addReference(); 156 void removeReference(); 157 UBool operator ==(const RBBIDataWrapper &other) const; 158 int32_t hashCode(); 159 const UnicodeString &getRuleSourceString() const; 160 #ifdef RBBI_DEBUG 161 void printData(); 162 void printTable(const char *heading, const RBBIStateTable *table); 163 #else 164 #define printData() 165 #define printTable(heading, table) 166 #endif 167 168 /* */ 169 /* Pointers to items within the data */ 170 /* */ 171 const RBBIDataHeader *fHeader; 172 const RBBIStateTable *fForwardTable; 173 const RBBIStateTable *fReverseTable; 174 const RBBIStateTable *fSafeFwdTable; 175 const RBBIStateTable *fSafeRevTable; 176 const UChar *fRuleSource; 177 const int32_t *fRuleStatusTable; 178 179 /* number of int32_t values in the rule status table. Used to sanity check indexing */ 180 int32_t fStatusMaxIdx; 181 182 UTrie fTrie; 183 184 private: 185 u_atomic_int32_t fRefCount; 186 UDataMemory *fUDataMem; 187 UnicodeString fRuleString; 188 UBool fDontFreeData; 189 190 RBBIDataWrapper(const RBBIDataWrapper &other); /* forbid copying of this class */ 191 RBBIDataWrapper &operator=(const RBBIDataWrapper &other); /* forbid copying of this class */ 192 }; 193 194 195 196 U_NAMESPACE_END 197 198 #endif /* C++ */ 199 200 #endif 201