src/kits/debugger/source_language/c_family/CLanguageTokenizer.h

   1 /*
   2  * Copyright 2006-2014 Haiku, Inc. All Rights Reserved.
   3  * Distributed under the terms of the MIT License.
   4  *
   5  * Authors:
   6  *              Stephan Aßmus <superstippi@gmx.de>
   7  *              Rene Gollent <rene@gollent.com>
   8  *              John Scipione <jscipione@gmail.com>
   9  *              Ingo Weinhold <bonefish@cs.tu-berlin.de>
  10  */
  11 #ifndef C_LANGUAGE_TOKENIZER
  12 #define C_LANGUAGE_TOKENIZER
  13
  14
  15 #include <String.h>
  16
  17 #include <Variant.h>
  18
  19
  20 namespace CLanguage {
  21
  22
  23 enum {
  24         TOKEN_NONE                                      = 0,
  25         TOKEN_IDENTIFIER,
  26         TOKEN_CONSTANT,
  27         TOKEN_END_OF_LINE,
  28
  29         TOKEN_PLUS,
  30         TOKEN_MINUS,
  31
  32         TOKEN_STAR,
  33         TOKEN_SLASH,
  34         TOKEN_MODULO,
  35
  36         TOKEN_OPENING_PAREN,
  37         TOKEN_CLOSING_PAREN,
  38
  39         TOKEN_OPENING_SQUARE_BRACKET,
  40         TOKEN_CLOSING_SQUARE_BRACKET,
  41
  42         TOKEN_OPENING_CURLY_BRACE,
  43         TOKEN_CLOSING_CURLY_BRACE,
  44
  45         TOKEN_ASSIGN,
  46         TOKEN_LOGICAL_AND,
  47         TOKEN_LOGICAL_OR,
  48         TOKEN_LOGICAL_NOT,
  49         TOKEN_BITWISE_AND,
  50         TOKEN_BITWISE_OR,
  51         TOKEN_BITWISE_NOT,
  52         TOKEN_BITWISE_XOR,
  53         TOKEN_EQ,
  54         TOKEN_NE,
  55         TOKEN_GT,
  56         TOKEN_GE,
  57         TOKEN_LT,
  58         TOKEN_LE,
  59
  60         TOKEN_BACKSLASH,
  61         TOKEN_CONDITION,
  62         TOKEN_COLON,
  63         TOKEN_SEMICOLON,
  64         TOKEN_COMMA,
  65         TOKEN_PERIOD,
  66         TOKEN_POUND,
  67
  68         TOKEN_SINGLE_QUOTE,
  69         TOKEN_DOUBLE_QUOTE,
  70
  71         TOKEN_STRING_LITERAL,
  72         TOKEN_BEGIN_COMMENT_BLOCK,
  73         TOKEN_END_COMMENT_BLOCK,
  74         TOKEN_INLINE_COMMENT,
  75
  76         TOKEN_MEMBER_PTR
  77 };
  78
  79
  80 class ParseException {
  81  public:
  82         ParseException(const char* message, int32 position)
  83                 : message(message),
  84                   position(position)
  85         {
  86         }
  87
  88         ParseException(const ParseException& other)
  89                 : message(other.message),
  90                   position(other.position)
  91         {
  92         }
  93
  94         BString message;
  95         int32   position;
  96 };
  97
  98
  99 struct Token {
 100                                                                 Token();
 101                                                                 Token(const Token& other);
 102                                                                 Token(const char* string, int32 length,
 103                                                                 int32 position, int32 type);
 104                         Token&  operator=(const Token& other);
 105
 106         BString                                         string;
 107         int32                                           type;
 108         BVariant                                        value;
 109         int32                                           position;
 110 };
 111
 112
 113 class Tokenizer {
 114 public:
 115                                                                 Tokenizer();
 116
 117                         void                            SetTo(const char* string);
 118
 119                         const Token&            NextToken();
 120                         void                            RewindToken();
 121 private:
 122                         bool                            _ParseOperator();
 123                         char                            _Peek() const;
 124
 125         static  bool                            _IsHexDigit(char c);
 126
 127                         Token&                          _ParseHexOperand();
 128                         int32                           _CurrentPos() const;
 129
 130 private:
 131         BString                                         fString;
 132         const char*                                     fCurrentChar;
 133         Token                                           fCurrentToken;
 134         bool                                            fReuseToken;
 135 };
 136
 137
 138 }       // namespace CLanguage
 139
 140
 141 #endif  // C_LANGUAGE_TOKENIZER