Mercurial > projects > dil
annotate trunk/src/Token.d @ 239:7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
- Tokens are stored in a doubly linked list by the Lexer now.
- Added method scanNext() which can be used by peek() and nextToken().
- Saving 'token' in Parser.try_().
author | aziz |
---|---|
date | Mon, 30 Jul 2007 12:36:04 +0000 |
parents | b7bde6583d3e |
children | 32d354584b28 |
rev | line source |
---|---|
0 | 1 /++ |
2 Author: Aziz Köksal | |
3 License: GPL2 | |
4 +/ | |
5 module Token; | |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
6 import std.c.stdlib : malloc, free; |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
7 import std.outofmemory; |
0 | 8 |
9 struct Position | |
10 { | |
11 size_t loc; | |
12 size_t col; | |
13 } | |
14 | |
3 | 15 enum TOK |
16 { | |
95 | 17 Invalid, |
18 | |
3 | 19 Identifier, |
4 | 20 Comment, |
9
5d6968cc751e
- Parsing string and character literals now (rudimentary implementation.)
aziz
parents:
4
diff
changeset
|
21 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
|
22 CharLiteral, WCharLiteral, DCharLiteral, |
58 | 23 |
24 // Numbers | |
25 Int32, Int64, Uint32, Uint64, | |
97
1a83e5070a84
- Added code for parsing IntNumber- and RealNumberExpressions.
aziz
parents:
95
diff
changeset
|
26 // 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
|
27 Float32, Float64, Float80, |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
28 Imaginary32, Imaginary64, Imaginary80, |
58 | 29 |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
30 |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
31 // Brackets |
20 | 32 LParen, |
33 RParen, | |
34 LBracket, | |
35 RBracket, | |
36 LBrace, | |
37 RBrace, | |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
38 |
22
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
39 Dot, Slice, Ellipses, |
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
40 |
131 | 41 // Floating point number operators |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
42 Unordered, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
43 UorE, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
44 UorG, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
45 UorGorE, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
46 UorL, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
47 UorLorE, |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
48 LorEorG, |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
49 LorG, |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
50 |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
51 // Normal operators |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
52 Assign, Equal, NotEqual, Not, |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
53 LessEqual, Less, |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
54 GreaterEqual, Greater, |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
55 LShiftAssign, LShift, |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
56 RShiftAssign,RShift, |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
57 URShiftAssign, URShift, |
23
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
58 OrAssign, OrLogical, OrBinary, |
24
903f91163f23
- Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents:
23
diff
changeset
|
59 AndAssign, AndLogical, AndBinary, |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
60 PlusAssign, PlusPlus, Plus, |
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
61 MinusAssign, MinusMinus, Minus, |
32
d7011daa4740
- Added missing commas after the items in the messages table.
aziz
parents:
31
diff
changeset
|
62 DivAssign, Div, |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
63 MulAssign, Mul, |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
64 ModAssign, Mod, |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
65 XorAssign, Xor, |
27 | 66 CatAssign, Catenate, |
39
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
67 Tilde, |
71 | 68 Identity, NotIdentity, |
23
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
69 |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
70 Colon, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
71 Semicolon, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
72 Question, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
73 Comma, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
74 Dollar, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
75 |
28 | 76 /* Keywords: |
77 NB.: Token.isKeyword() depends on this list being contiguous. | |
78 */ | |
79 Abstract,Alias,Align,Asm,Assert,Auto,Body, | |
80 Bool,Break,Byte,Case,Cast,Catch,Cdouble, | |
81 Cent,Cfloat,Char,Class,Const,Continue,Creal, | |
82 Dchar,Debug,Default,Delegate,Delete,Deprecated,Do, | |
83 Double,Else,Enum,Export,Extern,False,Final, | |
84 Finally,Float,For,Foreach,Foreach_reverse,Function,Goto, | |
85 Idouble,If,Ifloat,Import,In,Inout,Int, | |
86 Interface,Invariant,Ireal,Is,Lazy,Long,Macro, | |
87 Mixin,Module,New,Null,Out,Override,Package, | |
88 Pragma,Private,Protected,Public,Real,Ref,Return, | |
89 Scope,Short,Static,Struct,Super,Switch,Synchronized, | |
90 Template,This,Throw,True,Try,Typedef,Typeid, | |
91 Typeof,Ubyte,Ucent,Uint,Ulong,Union,Unittest, | |
92 Ushort,Version,Void,Volatile,Wchar,While,With, | |
93 | |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
94 HEAD, // start of linked list |
4 | 95 EOF |
3 | 96 } |
97 | |
28 | 98 alias TOK.Abstract KeywordsBegin; |
99 alias TOK.With KeywordsEnd; | |
100 | |
0 | 101 struct Token |
102 { | |
3 | 103 TOK type; |
84
ac8d961d10d1
- Added code for parsing This-,Super-,Null-,Bool-,Dollar-,CharLiteral- and StringLiteralExpression.
aziz
parents:
82
diff
changeset
|
104 // Position pos; |
0 | 105 |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
106 Token* next, prev; |
131 | 107 |
4 | 108 char* start; |
109 char* end; | |
110 | |
0 | 111 union |
112 { | |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
113 struct |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
114 { |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
115 string str; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
116 char pf; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
117 } |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
118 dchar dchar_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
119 long long_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
120 ulong ulong_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
121 int int_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
122 uint uint_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
123 float float_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
124 double double_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
125 real real_; |
0 | 126 } |
28 | 127 |
107
722c05bbd5eb
- Implemented parseEnumDeclaration() and added class EnumDeclaration.
aziz
parents:
97
diff
changeset
|
128 alias srcText identifier; |
722c05bbd5eb
- Implemented parseEnumDeclaration() and added class EnumDeclaration.
aziz
parents:
97
diff
changeset
|
129 |
65 | 130 string srcText() |
28 | 131 { |
30 | 132 assert(start && end); |
28 | 133 return start[0 .. end - start]; |
134 } | |
135 | |
210 | 136 static string toString(TOK tok) |
208 | 137 { |
138 return tokToString[tok]; | |
139 } | |
140 | |
28 | 141 bool isKeyword() |
142 { | |
163
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
143 return KeywordsBegin <= type && type <= KeywordsEnd; |
28 | 144 } |
208 | 145 |
163
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
146 int opEquals(TOK type2) |
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
147 { |
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
148 return type == type2; |
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
149 } |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
150 |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
151 new(size_t size) |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
152 { |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
153 void* p = malloc(size); |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
154 if (p is null) |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
155 throw new OutOfMemoryException(); |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
156 *cast(Token*)p = Token.init; |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
157 return p; |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
158 } |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
159 |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
160 delete(void* p) |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
161 { |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
162 free(p); |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
163 } |
208 | 164 } |
165 | |
166 string[] tokToString = [ | |
167 "Invalid", | |
168 | |
169 "Identifier", | |
170 "Comment", | |
171 "String", | |
172 "CharLiteral", "WCharLiteral", "DCharLiteral", | |
173 | |
174 "Int32", "Int64", "Uint32", "Uint64", | |
175 "Float32", "Float64", "Float80", | |
176 "Imaginary32", "Imaginary64", "Imaginary80", | |
177 | |
178 "(", | |
179 ")", | |
180 "[", | |
181 "]", | |
182 "{", | |
183 "}", | |
184 | |
185 ".", "..", "...", | |
186 | |
187 "Unordered", | |
188 "UorE", | |
189 "UorG", | |
190 "UorGorE", | |
191 "UorL", | |
192 "UorLorE", | |
193 "LorEorG", | |
194 "LorG", | |
195 | |
196 "=", "==", "!=", "!", | |
197 "<=", "<", | |
198 ">=", ">", | |
199 "<<=", "<<", | |
200 ">>=",">>", | |
201 ">>>=", ">>>", | |
202 "|=", "||", "|", | |
203 "&=", "&&", "&", | |
204 "+=", "++", "+", | |
205 "-=", "--", "-", | |
206 "/=", "/", | |
207 "*=", "*", | |
208 "%=", "%", | |
209 "^=", "^", | |
210 "~=", "~", | |
211 "~", | |
212 "is", "!is", | |
213 | |
214 ":", | |
215 ";", | |
216 "?", | |
217 ",", | |
218 "$", | |
219 | |
220 "abstract","alias","align","asm","assert","auto","body", | |
221 "bool","break","byte","case","cast","catch","cdouble", | |
222 "cent","cfloat","char","class","const","continue","creal", | |
223 "dchar","debug","default","delegate","delete","deprecated","do", | |
224 "double","else","enum","export","extern","false","final", | |
225 "finally","float","for","foreach","foreach_reverse","function","goto", | |
226 "idouble","if","ifloat","import","in","inout","int", | |
227 "interface","invariant","ireal","is","lazy","long","macro", | |
228 "mixin","module","new","null","out","override","package", | |
229 "pragma","private","protected","public","real","ref","return", | |
230 "scope","short","static","struct","super","switch","synchronized", | |
231 "template","this","throw","true","try","typedef","typeid", | |
232 "typeof","ubyte","ucent","uint","ulong","union","unittest", | |
233 "ushort","version","void","volatile","wchar","while","with", | |
234 | |
235 "EOF" | |
236 ]; |