]> git.localhorst.tv Git - l2e.git/blob - src/loader/Tokenizer.h
added peek function to tokenizer
[l2e.git] / 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 = 0,
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 = '0',
43                         STRING = '"',
44                         KEYWORD_EXPORT = 'e',
45                         KEYWORD_FALSE = 'f',
46                         KEYWORD_INCLUDE = 'i',
47                         KEYWORD_TRUE = 't',
48                         IDENTIFIER = 'x',
49                         TYPE_NAME = 'n',
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         const Token &Peek();
69         void Putback(const Token &);
70
71 private:
72         Token ReadToken();
73
74         Token ReadNumber();
75         Token ReadString();
76         Token ReadIdentifier();
77
78         bool CheckKeyword(Token &);
79
80 private:
81         std::istream &in;
82         Token putback;
83         bool isPutback;
84
85 };
86
87 inline const char *TokenTypeToString(Tokenizer::Token::Type t) {
88         switch (t) {
89                 case Tokenizer::Token::ANGLE_BRACKET_OPEN:
90                         return "ANGLE_BRACKET_OPEN";
91                 case Tokenizer::Token::ANGLE_BRACKET_CLOSE:
92                         return "ANGLE_BRACKET_CLOSE";
93                 case Tokenizer::Token::CHEVRON_OPEN:
94                         return "CHEVRON_OPEN";
95                 case Tokenizer::Token::CHEVRON_CLOSE:
96                         return "CHEVRON_CLOSE";
97                 case Tokenizer::Token::COLON:
98                         return "COLON";
99                 case Tokenizer::Token::COMMA:
100                         return "COMMA";
101                 case Tokenizer::Token::BRACKET_OPEN:
102                         return "BRACKET_OPEN";
103                 case Tokenizer::Token::BRACKET_CLOSE:
104                         return "BRACKET_CLOSE";
105                 case Tokenizer::Token::PARENTHESIS_OPEN:
106                         return "PARENTHESIS_OPEN";
107                 case Tokenizer::Token::PARENTHESIS_CLOSE:
108                         return "PARENTHESIS_CLOSE";
109                 case Tokenizer::Token::NUMBER:
110                         return "NUMBER";
111                 case Tokenizer::Token::STRING:
112                         return "STRING";
113                 case Tokenizer::Token::KEYWORD_EXPORT:
114                         return "KEYWORD_EXPORT";
115                 case Tokenizer::Token::KEYWORD_FALSE:
116                         return "KEYWORD_FALSE";
117                 case Tokenizer::Token::KEYWORD_INCLUDE:
118                         return "KEYWORD_INCLUDE";
119                 case Tokenizer::Token::KEYWORD_TRUE:
120                         return "KEYWORD_TRUE";
121                 case Tokenizer::Token::IDENTIFIER:
122                         return "IDENTIFIER";
123                 case Tokenizer::Token::TYPE_NAME:
124                         return "TYPE_NAME";
125                 default:
126                         return "UNKNOWN";
127         }
128 }
129
130 inline std::ostream &operator <<(std::ostream &out, Tokenizer::Token::Type t) {
131         out << TokenTypeToString(t);
132         return out;
133 }
134
135 }
136
137 #endif /* LOADER_TOKENIZER_H_ */