source/common/rbbiscan.h

   1 //
   2 //  rbbiscan.h
   3 //
   4 //  Copyright (C) 2002-2008, 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 "rbbitblb.h"
  25
  26
  27
  28 U_NAMESPACE_BEGIN
  29
  30 class   RBBIRuleBuilder;
  31 class   RBBISymbolTable;
  32
  33
  34 //--------------------------------------------------------------------------------
  35 //
  36 //  class RBBIRuleScanner does the lowest level, character-at-a-time
  37 //                        scanning of break iterator rules.
  38 //
  39 //                        The output of the scanner is parse trees for
  40 //                        the rule expressions and a list of all Unicode Sets
  41 //                        encountered.
  42 //
  43 //--------------------------------------------------------------------------------
  44
  45 class RBBIRuleScanner : public UMemory {
  46 public:
  47
  48     enum {
  49         kStackSize = 100            // The size of the state stack for
  50     };                              //   rules parsing.  Corresponds roughly
  51                                     //   to the depth of parentheses nesting
  52                                     //   that is allowed in the rules.
  53
  54     struct RBBIRuleChar {
  55         UChar32             fChar;
  56         UBool               fEscaped;
  57     };
  58
  59     RBBIRuleScanner(RBBIRuleBuilder  *rb);
  60
  61
  62     virtual    ~RBBIRuleScanner();
  63
  64     void        nextChar(RBBIRuleChar &c);          // Get the next char from the input stream.
  65                                                     // Return false if at end.
  66
  67     UBool       push(const RBBIRuleChar &c);        // Push (unget) one character.
  68                                                     //   Only a single character may be pushed.
  69
  70     void        parse();                            // Parse the rules, generating two parse
  71                                                     //   trees, one each for the forward and
  72                                                     //   reverse rules,
  73                                                     //   and a list of UnicodeSets encountered.
  74
  75     /**
  76      * Return a rules string without unnecessary
  77      * characters.
  78      */
  79     static UnicodeString stripRules(const UnicodeString &rules);
  80 private:
  81
  82     UBool       doParseActions(int32_t a);
  83     void        error(UErrorCode e);                   // error reporting convenience function.
  84     void        fixOpStack(RBBINode::OpPrecedence p);
  85                                                        //   a character.
  86     void        findSetFor(const UnicodeString &s, RBBINode *node, UnicodeSet *setToAdopt = NULL);
  87
  88     UChar32     nextCharLL();
  89 #ifdef RBBI_DEBUG
  90     void        printNodeStack(const char *title);
  91 #endif
  92     RBBINode    *pushNewNode(RBBINode::NodeType  t);
  93     void        scanSet();
  94
  95
  96     RBBIRuleBuilder               *fRB;              // The rule builder that we are part of.
  97
  98     int32_t                       fScanIndex;        // Index of current character being processed
  99                                                      //   in the rule input string.
 100     int32_t                       fNextIndex;        // Index of the next character, which
 101                                                      //   is the first character not yet scanned.
 102     UBool                         fQuoteMode;        // Scan is in a 'quoted region'
 103     int32_t                       fLineNum;          // Line number in input file.
 104     int32_t                       fCharNum;          // Char position within the line.
 105     UChar32                       fLastChar;         // Previous char, needed to count CR-LF
 106                                                      //   as a single line, not two.
 107
 108     RBBIRuleChar                  fC;                // Current char for parse state machine
 109                                                      //   processing.
 110     UnicodeString                 fVarName;          // $variableName, valid when we've just
 111                                                      //   scanned one.
 112
 113     RBBIRuleTableEl               **fStateTable;     // State Transition Table for RBBI Rule
 114                                                      //   parsing.  index by p[state][char-class]
 115
 116     uint16_t                      fStack[kStackSize];  // State stack, holds state pushes
 117     int32_t                       fStackPtr;           //  and pops as specified in the state
 118                                                        //  transition rules.
 119
 120     RBBINode                      *fNodeStack[kStackSize]; // Node stack, holds nodes created
 121                                                            //  during the parse of a rule
 122     int32_t                        fNodeStackPtr;
 123
 124
 125     UBool                          fReverseRule;     // True if the rule currently being scanned
 126                                                      //  is a reverse direction rule (if it
 127                                                      //  starts with a '!')
 128
 129     UBool                          fLookAheadRule;   // True if the rule includes a '/'
 130                                                      //   somewhere within it.
 131
 132     RBBISymbolTable               *fSymbolTable;     // symbol table, holds definitions of
 133                                                      //   $variable symbols.
 134
 135     UHashtable                    *fSetTable;        // UnicocodeSet hash table, holds indexes to
 136                                                      //   the sets created while parsing rules.
 137                                                      //   The key is the string used for creating
 138                                                      //   the set.
 139
 140     UnicodeSet                     fRuleSets[10];    // Unicode Sets that are needed during
 141                                                      //  the scanning of RBBI rules.  The
 142                                                      //  indicies for these are assigned by the
 143                                                      //  perl script that builds the state tables.
 144                                                      //  See rbbirpt.h.
 145
 146     int32_t                        fRuleNum;         // Counts each rule as it is scanned.
 147
 148     int32_t                        fOptionStart;     // Input index of start of a !!option
 149                                                      //   keyword, while being scanned.
 150
 151     UnicodeSet *gRuleSet_rule_char;
 152     UnicodeSet *gRuleSet_white_space;
 153     UnicodeSet *gRuleSet_name_char;
 154     UnicodeSet *gRuleSet_name_start_char;
 155
 156     RBBIRuleScanner(const RBBIRuleScanner &other); // forbid copying of this class
 157     RBBIRuleScanner &operator=(const RBBIRuleScanner &other); // forbid copying of this class
 158 };
 159
 160 U_NAMESPACE_END
 161
 162 #endif