icuSources/common/rbbisetb57.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 **********************************************************************
   5 *   Copyright (c) 2001-2005, International Business Machines
   6 *   Corporation and others.  All Rights Reserved.
   7 **********************************************************************
   8 *
   9 **********************************************************************
  10 *   Legacy version of RangeDescriptor and RBBISetBuilder from ICU 57,
  11 *   only for use by Apple RuleBasedTokenizer
  12 **********************************************************************
  13 */
  14
  15 #ifndef RBBISETB57_H
  16 #define RBBISETB57_H
  17
  18 #include "unicode/utypes.h"
  19 #include "unicode/uobject.h"
  20 #include "rbbirb57.h"
  21 #include "rbbinode.h"
  22 #include "uvector.h"
  23
  24 struct  UNewTrie;
  25
  26 U_NAMESPACE_BEGIN
  27
  28 class   RBBIRuleBuilder57;
  29
  30 //
  31 //  RBBISetBuilder57   Derives the character categories used by the runtime RBBI engine
  32 //                   from the Unicode Sets appearing in the source  RBBI rules, and
  33 //                   creates the TRIE table used to map from Unicode to the
  34 //                   character categories.
  35 //
  36
  37
  38 //
  39 //  RangeDescriptor57
  40 //
  41 //     Each of the non-overlapping character ranges gets one of these descriptors.
  42 //     All of them are strung together in a linked list, which is kept in order
  43 //     (by character)
  44 //
  45 class RangeDescriptor57 : public UMemory {
  46 public:
  47     UChar32            fStartChar;      // Start of range, unicode 32 bit value.
  48     UChar32            fEndChar;        // End of range, unicode 32 bit value.
  49     int32_t            fNum;            // runtime-mapped input value for this range.
  50     UVector           *fIncludesSets;   // vector of the the original
  51                                         //   Unicode sets that include this range.
  52                                         //    (Contains ptrs to uset nodes)
  53     RangeDescriptor57 *fNext;           // Next RangeDescriptor57 in the linked list.
  54
  55     RangeDescriptor57(UErrorCode &status);
  56     RangeDescriptor57(const RangeDescriptor57 &other, UErrorCode &status);
  57     ~RangeDescriptor57();
  58     void split(UChar32 where, UErrorCode &status);   // Spit this range in two at "where", with
  59                                         //   where appearing in the second (higher) part.
  60     void setDictionaryFlag();           // Check whether this range appears as part of
  61                                         //   the Unicode set named "dictionary"
  62
  63 private:
  64     RangeDescriptor57(const RangeDescriptor57 &other); // forbid copying of this class
  65     RangeDescriptor57 &operator=(const RangeDescriptor57 &other); // forbid copying of this class
  66 };
  67
  68
  69 //
  70 //  RBBISetBuilder57   Handles processing of Unicode Sets from RBBI rules.
  71 //
  72 //      Starting with the rules parse tree from the scanner,
  73 //
  74 //                   -  Enumerate the set of UnicodeSets that are referenced
  75 //                      by the RBBI rules.
  76 //                   -  compute a derived set of non-overlapping UnicodeSets
  77 //                      that will correspond to columns in the state table for
  78 //                      the RBBI execution engine.
  79 //                   -  construct the trie table that maps input characters
  80 //                      to set numbers in the non-overlapping set of sets.
  81 //
  82
  83
  84 class RBBISetBuilder57 : public UMemory {
  85 public:
  86     RBBISetBuilder57(RBBIRuleBuilder57 *rb);
  87     ~RBBISetBuilder57();
  88
  89     void     build();
  90     void     addValToSets(UVector *sets,      uint32_t val);
  91     void     addValToSet (RBBINode *usetNode, uint32_t val);
  92     int32_t  getNumCharCategories() const;   // CharCategories are the same as input symbol set to the
  93                                              //    runtime state machine, which are the same as
  94                                              //    columns in the DFA state table
  95     int32_t  getTrieSize() /*const*/;        // Size in bytes of the serialized Trie.
  96     void     serializeTrie(uint8_t *where);  // write out the serialized Trie.
  97     UChar32  getFirstChar(int32_t  val) const;
  98     UBool    sawBOF() const;                 // Indicate whether any references to the {bof} pseudo
  99                                              //   character were encountered.
 100 #ifdef RBBI_DEBUG
 101     void     printSets();
 102     void     printRanges();
 103     void     printRangeGroups();
 104 #else
 105     #define printSets()
 106     #define printRanges()
 107     #define printRangeGroups()
 108 #endif
 109
 110 private:
 111     void           numberSets();
 112
 113     RBBIRuleBuilder57     *fRB;             // The RBBI Rule Compiler that owns us.
 114     UErrorCode            *fStatus;
 115
 116     RangeDescriptor57     *fRangeList;      // Head of the linked list of RangeDescriptors
 117
 118     UNewTrie              *fTrie;           // The mapping TRIE that is the end result of processing
 119     uint32_t               fTrieSize;       //  the Unicode Sets.
 120
 121     // Groups correspond to character categories -
 122     //       groups of ranges that are in the same original UnicodeSets.
 123     //       fGroupCount is the index of the last used group.
 124     //       fGroupCount+1 is also the number of columns in the RBBI state table being compiled.
 125     //       State table column 0 is not used.  Column 1 is for end-of-input.
 126     //       column 2 is for group 0.  Funny counting.
 127     int32_t               fGroupCount;
 128
 129     UBool                 fSawBOF;
 130
 131     RBBISetBuilder57(const RBBISetBuilder57 &other); // forbid copying of this class
 132     RBBISetBuilder57 &operator=(const RBBISetBuilder57 &other); // forbid copying of this class
 133 };
 134
 135
 136
 137 U_NAMESPACE_END
 138 #endif