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                         NUMBER,
  41                         STRING,
  42                         KEYWORD_EXPORT,
  43                         KEYWORD_FALSE,
  44                         KEYWORD_INCLUDE,
  45                         KEYWORD_TRUE,
  46                         IDENTIFIER,
  47                         TYPE_NAME,
  48                 };
  49
  50                 Token() : type(UNKNOWN), number(0) { }
  51                 explicit Token(Type t) : type(t), number(0) { }
  52
  53                 Type type;
  54                 std::string str;
  55                 int number;
  56
  57         };
  58
  59         class LexerError: public std::runtime_error {
  60         public:
  61                 explicit LexerError(const std::string &msg) : std::runtime_error(msg) { }
  62         };
  63
  64         bool HasMore();
  65         Token GetNext();
  66         void Putback(const Token &);
  67
  68 private:
  69         Token ReadToken();
  70
  71         Token ReadNumber();
  72         Token ReadString();
  73         Token ReadIdentifier();
  74
  75         bool CheckKeyword(Token &);
  76
  77 private:
  78         std::istream &in;
  79         Token putback;
  80         bool isPutback;
  81
  82 };
  83
  84 inline const char *TokenTypeToString(Tokenizer::Token::Type t) {
  85         switch (t) {
  86                 case Tokenizer::Token::ANGLE_BRACKET_OPEN:
  87                         return "ANGLE_BRACKET_OPEN";
  88                 case Tokenizer::Token::ANGLE_BRACKET_CLOSE:
  89                         return "ANGLE_BRACKET_CLOSE";
  90                 case Tokenizer::Token::CHEVRON_OPEN:
  91                         return "CHEVRON_OPEN";
  92                 case Tokenizer::Token::CHEVRON_CLOSE:
  93                         return "CHEVRON_CLOSE";
  94                 case Tokenizer::Token::COLON:
  95                         return "COLON";
  96                 case Tokenizer::Token::COMMA:
  97                         return "COMMA";
  98                 case Tokenizer::Token::BRACKET_OPEN:
  99                         return "BRACKET_OPEN";
 100                 case Tokenizer::Token::BRACKET_CLOSE:
 101                         return "BRACKET_CLOSE";
 102                 case Tokenizer::Token::NUMBER:
 103                         return "NUMBER";
 104                 case Tokenizer::Token::STRING:
 105                         return "STRING";
 106                 case Tokenizer::Token::KEYWORD_EXPORT:
 107                         return "KEYWORD_EXPORT";
 108                 case Tokenizer::Token::KEYWORD_FALSE:
 109                         return "KEYWORD_FALSE";
 110                 case Tokenizer::Token::KEYWORD_INCLUDE:
 111                         return "KEYWORD_INCLUDE";
 112                 case Tokenizer::Token::KEYWORD_TRUE:
 113                         return "KEYWORD_TRUE";
 114                 case Tokenizer::Token::IDENTIFIER:
 115                         return "IDENTIFIER";
 116                 case Tokenizer::Token::TYPE_NAME:
 117                         return "TYPE_NAME";
 118                 default:
 119                         return "UNKNOWN";
 120         }
 121 }
 122
 123 inline std::ostream &operator <<(std::ostream &out, Tokenizer::Token::Type t) {
 124         out << TokenTypeToString(t);
 125         return out;
 126 }
 127
 128 }
 129
 130 #endif /* LOADER_TOKENIZER_H_ */