src/loader/Tokenizer.h

   1 /*
   2  * Tokenizer.h
   3  *
   4  *  Created on: Aug 26, 2012
   5  *      Author: holy
   6  */
   7
   8 #ifndef LOADER_TOKENIZER_H_
   9 #define LOADER_TOKENIZER_H_
  10
  11 #include <iosfwd>
  12 #include <ostream>
  13 #include <stdexcept>
  14 #include <string>
  15
  16 namespace loader {
  17
  18 class Tokenizer {
  19
  20 public:
  21         explicit Tokenizer(std::istream &in) : in(in), isPutback(false) { }
  22         ~Tokenizer() { }
  23 private:
  24         Tokenizer(const Tokenizer &);
  25         Tokenizer &operator =(const Tokenizer &);
  26
  27 public:
  28         struct Token {
  29
  30                 enum Type {
  31                         UNKNOWN,
  32                         ANGLE_BRACKET_OPEN = '{',
  33                         ANGLE_BRACKET_CLOSE = '}',
  34                         CHEVRON_OPEN = '<',
  35                         CHEVRON_CLOSE = '>',
  36                         COLON = ':',
  37                         COMMA = ',',
  38                         BRACKET_OPEN = '[',
  39                         BRACKET_CLOSE = ']',
  40                         PARENTHESIS_OPEN = '(',
  41                         PARENTHESIS_CLOSE = ')',
  42                         NUMBER,
  43                         STRING,
  44                         KEYWORD_EXPORT,
  45                         KEYWORD_FALSE,
  46                         KEYWORD_INCLUDE,
  47                         KEYWORD_TRUE,
  48                         IDENTIFIER,
  49                         TYPE_NAME,
  50                 };
  51
  52                 Token() : type(UNKNOWN), number(0) { }
  53                 explicit Token(Type t) : type(t), number(0) { }
  54
  55                 Type type;
  56                 std::string str;
  57                 int number;
  58
  59         };
  60
  61         class LexerError: public std::runtime_error {
  62         public:
  63                 explicit LexerError(const std::string &msg) : std::runtime_error(msg) { }
  64         };
  65
  66         bool HasMore();
  67         Token GetNext();
  68         void Putback(const Token &);
  69
  70 private:
  71         Token ReadToken();
  72
  73         Token ReadNumber();
  74         Token ReadString();
  75         Token ReadIdentifier();
  76
  77         bool CheckKeyword(Token &);
  78
  79 private:
  80         std::istream &in;
  81         Token putback;
  82         bool isPutback;
  83
  84 };
  85
  86 inline const char *TokenTypeToString(Tokenizer::Token::Type t) {
  87         switch (t) {
  88                 case Tokenizer::Token::ANGLE_BRACKET_OPEN:
  89                         return "ANGLE_BRACKET_OPEN";
  90                 case Tokenizer::Token::ANGLE_BRACKET_CLOSE:
  91                         return "ANGLE_BRACKET_CLOSE";
  92                 case Tokenizer::Token::CHEVRON_OPEN:
  93                         return "CHEVRON_OPEN";
  94                 case Tokenizer::Token::CHEVRON_CLOSE:
  95                         return "CHEVRON_CLOSE";
  96                 case Tokenizer::Token::COLON:
  97                         return "COLON";
  98                 case Tokenizer::Token::COMMA:
  99                         return "COMMA";
 100                 case Tokenizer::Token::BRACKET_OPEN:
 101                         return "BRACKET_OPEN";
 102                 case Tokenizer::Token::BRACKET_CLOSE:
 103                         return "BRACKET_CLOSE";
 104                 case Tokenizer::Token::PARENTHESIS_OPEN:
 105                         return "PARENTHESIS_OPEN";
 106                 case Tokenizer::Token::PARENTHESIS_CLOSE:
 107                         return "PARENTHESIS_CLOSE";
 108                 case Tokenizer::Token::NUMBER:
 109                         return "NUMBER";
 110                 case Tokenizer::Token::STRING:
 111                         return "STRING";
 112                 case Tokenizer::Token::KEYWORD_EXPORT:
 113                         return "KEYWORD_EXPORT";
 114                 case Tokenizer::Token::KEYWORD_FALSE:
 115                         return "KEYWORD_FALSE";
 116                 case Tokenizer::Token::KEYWORD_INCLUDE:
 117                         return "KEYWORD_INCLUDE";
 118                 case Tokenizer::Token::KEYWORD_TRUE:
 119                         return "KEYWORD_TRUE";
 120                 case Tokenizer::Token::IDENTIFIER:
 121                         return "IDENTIFIER";
 122                 case Tokenizer::Token::TYPE_NAME:
 123                         return "TYPE_NAME";
 124                 default:
 125                         return "UNKNOWN";
 126         }
 127 }
 128
 129 inline std::ostream &operator <<(std::ostream &out, Tokenizer::Token::Type t) {
 130         out << TokenTypeToString(t);
 131         return out;
 132 }
 133
 134 }
 135
 136 #endif /* LOADER_TOKENIZER_H_ */