annotate trunk/src/Token.d @ 107:722c05bbd5eb

- Implemented parseEnumDeclaration() and added class EnumDeclaration. - Added 'alias srcText identifier' to Token for Identifiers. - Replaced all instances of token.srcText with token.identifier where applicable.
author aziz
date Sun, 08 Jul 2007 15:22:03 +0000
parents 1a83e5070a84
children ce636f3981cc
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
1 /++
8ba2570de175 Initial import.
aziz
parents:
diff changeset
2 Author: Aziz Köksal
8ba2570de175 Initial import.
aziz
parents:
diff changeset
3 License: GPL2
8ba2570de175 Initial import.
aziz
parents:
diff changeset
4 +/
8ba2570de175 Initial import.
aziz
parents:
diff changeset
5 module Token;
8ba2570de175 Initial import.
aziz
parents:
diff changeset
6
8ba2570de175 Initial import.
aziz
parents:
diff changeset
7 struct Position
8ba2570de175 Initial import.
aziz
parents:
diff changeset
8 {
8ba2570de175 Initial import.
aziz
parents:
diff changeset
9 size_t loc;
8ba2570de175 Initial import.
aziz
parents:
diff changeset
10 size_t col;
8ba2570de175 Initial import.
aziz
parents:
diff changeset
11 }
8ba2570de175 Initial import.
aziz
parents:
diff changeset
12
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 0
diff changeset
13 enum TOK
4bbce78bfb1e - Added TOK enum.
aziz
parents: 0
diff changeset
14 {
95
0eb4c8a5b32b - Added TOK.Invalid.
aziz
parents: 84
diff changeset
15 Invalid,
0eb4c8a5b32b - Added TOK.Invalid.
aziz
parents: 84
diff changeset
16
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 0
diff changeset
17 Identifier,
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
18 Comment,
9
5d6968cc751e - Parsing string and character literals now (rudimentary implementation.)
aziz
parents: 4
diff changeset
19 String,
82
fc645fb2fe72 - scanEscapeSequences() doesn't return 0xFFFF as an error value anymore, because it is a valid codepoint usable by the user.
aziz
parents: 71
diff changeset
20 CharLiteral, WCharLiteral, DCharLiteral,
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 56
diff changeset
21
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 56
diff changeset
22 // Numbers
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
23 Number,
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 56
diff changeset
24 Int32, Int64, Uint32, Uint64,
97
1a83e5070a84 - Added code for parsing IntNumber- and RealNumberExpressions.
aziz
parents: 95
diff changeset
25 // Floating point number scanner relies on this order. (FloatXY + 3 == ImaginaryXY)
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
26 Float32, Float64, Float80,
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
27 Imaginary32, Imaginary64, Imaginary80,
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 56
diff changeset
28
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
29
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
30 // Brackets
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 18
diff changeset
31 LParen,
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 18
diff changeset
32 RParen,
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 18
diff changeset
33 LBracket,
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 18
diff changeset
34 RBracket,
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 18
diff changeset
35 LBrace,
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 18
diff changeset
36 RBrace,
21
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
37
22
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
38 Dot, Slice, Ellipses,
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
39
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
40 // Floating point operators
35
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
41 Unordered,
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
42 UorE,
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
43 UorG,
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
44 UorGorE,
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
45 UorL,
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
46 UorLorE,
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
47 LorEorG,
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
48 LorG,
35
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
49
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
50 // Normal operators
35
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 32
diff changeset
51 Assign, Equal, NotEqual, Not,
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 36
diff changeset
52 LessEqual, Less,
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 36
diff changeset
53 GreaterEqual, Greater,
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
54 LShiftAssign, LShift,
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 36
diff changeset
55 RShiftAssign,RShift,
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 36
diff changeset
56 URShiftAssign, URShift,
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
57 OrAssign, OrLogical, OrBinary,
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
58 AndAssign, AndLogical, AndBinary,
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
59 PlusAssign, PlusPlus, Plus,
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
60 MinusAssign, MinusMinus, Minus,
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
61 DivAssign, Div,
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
62 MulAssign, Mul,
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
63 ModAssign, Mod,
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
64 XorAssign, Xor,
27
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
65 CatAssign, Catenate,
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
66 Tilde,
71
b3777cca323c - Added Identity and NotIdentity tokens.
aziz
parents: 65
diff changeset
67 Identity, NotIdentity,
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
68
21
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
69 Colon,
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
70 Semicolon,
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
71 Question,
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
72 Comma,
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
73 Dollar,
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
74
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
75 /* Keywords:
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
76 NB.: Token.isKeyword() depends on this list being contiguous.
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
77 */
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
78 Abstract,Alias,Align,Asm,Assert,Auto,Body,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
79 Bool,Break,Byte,Case,Cast,Catch,Cdouble,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
80 Cent,Cfloat,Char,Class,Const,Continue,Creal,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
81 Dchar,Debug,Default,Delegate,Delete,Deprecated,Do,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
82 Double,Else,Enum,Export,Extern,False,Final,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
83 Finally,Float,For,Foreach,Foreach_reverse,Function,Goto,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
84 Idouble,If,Ifloat,Import,In,Inout,Int,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
85 Interface,Invariant,Ireal,Is,Lazy,Long,Macro,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
86 Mixin,Module,New,Null,Out,Override,Package,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
87 Pragma,Private,Protected,Public,Real,Ref,Return,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
88 Scope,Short,Static,Struct,Super,Switch,Synchronized,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
89 Template,This,Throw,True,Try,Typedef,Typeid,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
90 Typeof,Ubyte,Ucent,Uint,Ulong,Union,Unittest,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
91 Ushort,Version,Void,Volatile,Wchar,While,With,
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
92
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
93 EOF
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 0
diff changeset
94 }
4bbce78bfb1e - Added TOK enum.
aziz
parents: 0
diff changeset
95
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
96 alias TOK.Abstract KeywordsBegin;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
97 alias TOK.With KeywordsEnd;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
98
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
99 struct Token
8ba2570de175 Initial import.
aziz
parents:
diff changeset
100 {
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 0
diff changeset
101 TOK type;
84
ac8d961d10d1 - Added code for parsing This-,Super-,Null-,Bool-,Dollar-,CharLiteral- and StringLiteralExpression.
aziz
parents: 82
diff changeset
102 // Position pos;
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
103
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
104 char* start;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
105 char* end;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
106
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
107 union
8ba2570de175 Initial import.
aziz
parents:
diff changeset
108 {
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
109 struct
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
110 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
111 string str;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
112 char pf;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
113 }
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
114 dchar dchar_;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
115 long long_;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
116 ulong ulong_;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
117 int int_;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
118 uint uint_;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
119 float float_;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
120 double double_;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 58
diff changeset
121 real real_;
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
122 }
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
123
107
722c05bbd5eb - Implemented parseEnumDeclaration() and added class EnumDeclaration.
aziz
parents: 97
diff changeset
124 alias srcText identifier;
722c05bbd5eb - Implemented parseEnumDeclaration() and added class EnumDeclaration.
aziz
parents: 97
diff changeset
125
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 62
diff changeset
126 string srcText()
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
127 {
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
128 assert(start && end);
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
129 return start[0 .. end - start];
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
130 }
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
131
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
132 bool isKeyword()
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
133 {
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
134 if (KeywordsBegin <= type && type <= KeywordsEnd)
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
135 return true;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
136 return false;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
137 }
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
138 }