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)
  22         : in(in), line(1), isPutback(false), skipComments(true) { }
  23         ~Tokenizer() { }
  24 private:
  25         Tokenizer(const Tokenizer &);
  26         Tokenizer &operator =(const Tokenizer &);
  27
  28 public:
  29         struct Token {
  30
  31                 enum Type {
  32                         UNKNOWN = 0,
  33                         ANGLE_BRACKET_OPEN = '{',
  34                         ANGLE_BRACKET_CLOSE = '}',
  35                         CHEVRON_OPEN = '<',
  36                         CHEVRON_CLOSE = '>',
  37                         COLON = ':',
  38                         COMMA = ',',
  39                         BRACKET_OPEN = '[',
  40                         BRACKET_CLOSE = ']',
  41                         PARENTHESIS_OPEN = '(',
  42                         PARENTHESIS_CLOSE = ')',
  43                         NUMBER = '0',
  44                         STRING = '"',
  45                         KEYWORD_EXPORT = 'e',
  46                         KEYWORD_FALSE = 'f',
  47                         KEYWORD_INCLUDE = 'i',
  48                         KEYWORD_TRUE = 't',
  49                         IDENTIFIER = 'x',
  50                         TYPE_NAME = 'n',
  51                         COMMENT = 'c'
  52                 };
  53
  54                 Token() : type(UNKNOWN), number(0) { }
  55                 explicit Token(Type t) : type(t), number(0) { }
  56
  57                 Type type;
  58                 std::string str;
  59                 int number;
  60
  61         };
  62
  63         class LexerError: public std::runtime_error {
  64         public:
  65                 LexerError(int line, const std::string &msg)
  66                 : std::runtime_error(msg), line(line) { }
  67                 int Line() const { return line; }
  68         private:
  69                 int line;
  70         };
  71
  72         bool HasMore();
  73         Token GetNext();
  74         const Token &Peek();
  75         void Putback(const Token &);
  76         int Line() const { return line; }
  77
  78 private:
  79         void ScanSpace();
  80         Token ReadToken();
  81
  82         Token ReadNumber();
  83         Token ReadString();
  84         Token ReadIdentifier();
  85
  86         Token ReadComment();
  87         Token ReadMultilineComment();
  88
  89         bool CheckKeyword(Token &);
  90
  91 private:
  92         std::istream &in;
  93         Token putback;
  94         int line;
  95         bool isPutback;
  96         bool skipComments;
  97
  98 };
  99
 100 inline const char *TokenTypeToString(Tokenizer::Token::Type t) {
 101         switch (t) {
 102                 case Tokenizer::Token::ANGLE_BRACKET_OPEN:
 103                         return "ANGLE_BRACKET_OPEN";
 104                 case Tokenizer::Token::ANGLE_BRACKET_CLOSE:
 105                         return "ANGLE_BRACKET_CLOSE";
 106                 case Tokenizer::Token::CHEVRON_OPEN:
 107                         return "CHEVRON_OPEN";
 108                 case Tokenizer::Token::CHEVRON_CLOSE:
 109                         return "CHEVRON_CLOSE";
 110                 case Tokenizer::Token::COLON:
 111                         return "COLON";
 112                 case Tokenizer::Token::COMMA:
 113                         return "COMMA";
 114                 case Tokenizer::Token::BRACKET_OPEN:
 115                         return "BRACKET_OPEN";
 116                 case Tokenizer::Token::BRACKET_CLOSE:
 117                         return "BRACKET_CLOSE";
 118                 case Tokenizer::Token::PARENTHESIS_OPEN:
 119                         return "PARENTHESIS_OPEN";
 120                 case Tokenizer::Token::PARENTHESIS_CLOSE:
 121                         return "PARENTHESIS_CLOSE";
 122                 case Tokenizer::Token::NUMBER:
 123                         return "NUMBER";
 124                 case Tokenizer::Token::STRING:
 125                         return "STRING";
 126                 case Tokenizer::Token::KEYWORD_EXPORT:
 127                         return "KEYWORD_EXPORT";
 128                 case Tokenizer::Token::KEYWORD_FALSE:
 129                         return "KEYWORD_FALSE";
 130                 case Tokenizer::Token::KEYWORD_INCLUDE:
 131                         return "KEYWORD_INCLUDE";
 132                 case Tokenizer::Token::KEYWORD_TRUE:
 133                         return "KEYWORD_TRUE";
 134                 case Tokenizer::Token::IDENTIFIER:
 135                         return "IDENTIFIER";
 136                 case Tokenizer::Token::TYPE_NAME:
 137                         return "TYPE_NAME";
 138                 default:
 139                         return "UNKNOWN";
 140         }
 141 }
 142
 143 inline std::ostream &operator <<(std::ostream &out, Tokenizer::Token::Type t) {
 144         out << TokenTypeToString(t);
 145         return out;
 146 }
 147
 148 }
 149
 150 #endif /* LOADER_TOKENIZER_H_ */