source/common/rbbiscan.h

   1 //
   2 //  rbbiscan.h
   3 //
   4 //  Copyright (C) 2002-2016, International Business Machines Corporation and others.
   5 //  All Rights Reserved.
   6 //
   7 //  This file contains declarations for class RBBIRuleScanner
   8 //
   9
  10
  11 #ifndef RBBISCAN_H
  12 #define RBBISCAN_H
  13
  14 #include "unicode/utypes.h"
  15 #include "unicode/uobject.h"
  16 #include "unicode/rbbi.h"
  17 #include "unicode/uniset.h"
  18 #include "unicode/parseerr.h"
  19 #include "uhash.h"
  20 #include "uvector.h"
  21 #include "unicode/symtable.h"// For UnicodeSet parsing, is the interface that
  22                           //    looks up references to $variables within a set.
  23 #include "rbbinode.h"
  24 #include "rbbirpt.h"
  25
  26 U_NAMESPACE_BEGIN
  27
  28 class   RBBIRuleBuilder;
  29 class   RBBISymbolTable;
  30
  31
  32 //--------------------------------------------------------------------------------
  33 //
  34 //  class RBBIRuleScanner does the lowest level, character-at-a-time
  35 //                        scanning of break iterator rules.
  36 //
  37 //                        The output of the scanner is parse trees for
  38 //                        the rule expressions and a list of all Unicode Sets
  39 //                        encountered.
  40 //
  41 //--------------------------------------------------------------------------------
  42
  43 class RBBIRuleScanner : public UMemory {
  44 public:
  45
  46     enum {
  47         kStackSize = 100            // The size of the state stack for
  48     };                              //   rules parsing.  Corresponds roughly
  49                                     //   to the depth of parentheses nesting
  50                                     //   that is allowed in the rules.
  51
  52     struct RBBIRuleChar {
  53         UChar32             fChar;
  54         UBool               fEscaped;
  55         RBBIRuleChar() : fChar(0), fEscaped(FALSE) {};
  56     };
  57
  58     RBBIRuleScanner(RBBIRuleBuilder  *rb);
  59
  60
  61     virtual    ~RBBIRuleScanner();
  62
  63     void        nextChar(RBBIRuleChar &c);          // Get the next char from the input stream.
  64                                                     // Return false if at end.
  65
  66     UBool       push(const RBBIRuleChar &c);        // Push (unget) one character.
  67                                                     //   Only a single character may be pushed.
  68
  69     void        parse();                            // Parse the rules, generating two parse
  70                                                     //   trees, one each for the forward and
  71                                                     //   reverse rules,
  72                                                     //   and a list of UnicodeSets encountered.
  73
  74     /**
  75      * Return a rules string without unnecessary
  76      * characters.
  77      */
  78     static UnicodeString stripRules(const UnicodeString &rules);
  79 private:
  80
  81     UBool       doParseActions(int32_t a);
  82     void        error(UErrorCode e);                   // error reporting convenience function.
  83     void        fixOpStack(RBBINode::OpPrecedence p);
  84                                                        //   a character.
  85     void        findSetFor(const UnicodeString &s, RBBINode *node, UnicodeSet *setToAdopt = NULL);
  86
  87     UChar32     nextCharLL();
  88 #ifdef RBBI_DEBUG
  89     void        printNodeStack(const char *title);
  90 #endif
  91     RBBINode    *pushNewNode(RBBINode::NodeType  t);
  92     void        scanSet();
  93
  94
  95     RBBIRuleBuilder               *fRB;              // The rule builder that we are part of.
  96
  97     int32_t                       fScanIndex;        // Index of current character being processed
  98                                                      //   in the rule input string.
  99     int32_t                       fNextIndex;        // Index of the next character, which
 100                                                      //   is the first character not yet scanned.
 101     UBool                         fQuoteMode;        // Scan is in a 'quoted region'
 102     int32_t                       fLineNum;          // Line number in input file.
 103     int32_t                       fCharNum;          // Char position within the line.
 104     UChar32                       fLastChar;         // Previous char, needed to count CR-LF
 105                                                      //   as a single line, not two.
 106
 107     RBBIRuleChar                  fC;                // Current char for parse state machine
 108                                                      //   processing.
 109     UnicodeString                 fVarName;          // $variableName, valid when we've just
 110                                                      //   scanned one.
 111
 112     RBBIRuleTableEl               **fStateTable;     // State Transition Table for RBBI Rule
 113                                                      //   parsing.  index by p[state][char-class]
 114
 115     uint16_t                      fStack[kStackSize];  // State stack, holds state pushes
 116     int32_t                       fStackPtr;           //  and pops as specified in the state
 117                                                        //  transition rules.
 118
 119     RBBINode                      *fNodeStack[kStackSize]; // Node stack, holds nodes created
 120                                                            //  during the parse of a rule
 121     int32_t                        fNodeStackPtr;
 122
 123
 124     UBool                          fReverseRule;     // True if the rule currently being scanned
 125                                                      //  is a reverse direction rule (if it
 126                                                      //  starts with a '!')
 127
 128     UBool                          fLookAheadRule;   // True if the rule includes a '/'
 129                                                      //   somewhere within it.
 130
 131     UBool                          fNoChainInRule;   // True if the current rule starts with a '^'.
 132
 133     RBBISymbolTable               *fSymbolTable;     // symbol table, holds definitions of
 134                                                      //   $variable symbols.
 135
 136     UHashtable                    *fSetTable;        // UnicocodeSet hash table, holds indexes to
 137                                                      //   the sets created while parsing rules.
 138                                                      //   The key is the string used for creating
 139                                                      //   the set.
 140
 141     UnicodeSet                     fRuleSets[10];    // Unicode Sets that are needed during
 142                                                      //  the scanning of RBBI rules.  The
 143                                                      //  indicies for these are assigned by the
 144                                                      //  perl script that builds the state tables.
 145                                                      //  See rbbirpt.h.
 146
 147     int32_t                        fRuleNum;         // Counts each rule as it is scanned.
 148
 149     int32_t                        fOptionStart;     // Input index of start of a !!option
 150                                                      //   keyword, while being scanned.
 151
 152     UnicodeSet *gRuleSet_rule_char;
 153     UnicodeSet *gRuleSet_white_space;
 154     UnicodeSet *gRuleSet_name_char;
 155     UnicodeSet *gRuleSet_name_start_char;
 156
 157     RBBIRuleScanner(const RBBIRuleScanner &other); // forbid copying of this class
 158     RBBIRuleScanner &operator=(const RBBIRuleScanner &other); // forbid copying of this class
 159 };
 160
 161 U_NAMESPACE_END
 162
 163 #endif