Mercurial > projects > dil
annotate trunk/src/Token.d @ 312:fa0b6f32c1ae
- Added Special to enum TOK.
- Added special tokens to keyword list.
- Added case T.Special for parsing SpecialTokenExpression.
author | aziz |
---|---|
date | Wed, 15 Aug 2007 19:19:00 +0000 |
parents | a416e09c08ea |
children | ebd21bbf296e |
rev | line source |
---|---|
0 | 1 /++ |
2 Author: Aziz Köksal | |
249 | 3 License: GPL3 |
0 | 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, |
312 | 22 Special, |
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
|
23 CharLiteral, WCharLiteral, DCharLiteral, |
58 | 24 |
25 // Numbers | |
26 Int32, Int64, Uint32, Uint64, | |
97
1a83e5070a84
- Added code for parsing IntNumber- and RealNumberExpressions.
aziz
parents:
95
diff
changeset
|
27 // 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
|
28 Float32, Float64, Float80, |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
29 Imaginary32, Imaginary64, Imaginary80, |
58 | 30 |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
31 |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
32 // Brackets |
20 | 33 LParen, |
34 RParen, | |
35 LBracket, | |
36 RBracket, | |
37 LBrace, | |
38 RBrace, | |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
39 |
22
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
40 Dot, Slice, Ellipses, |
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
41 |
131 | 42 // 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
|
43 Unordered, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
44 UorE, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
45 UorG, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
46 UorGorE, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
47 UorL, |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
48 UorLorE, |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
49 LorEorG, |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
50 LorG, |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
51 |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
52 // Normal operators |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
32
diff
changeset
|
53 Assign, Equal, NotEqual, Not, |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
54 LessEqual, Less, |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
55 GreaterEqual, Greater, |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
56 LShiftAssign, LShift, |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
57 RShiftAssign,RShift, |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
36
diff
changeset
|
58 URShiftAssign, URShift, |
23
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
59 OrAssign, OrLogical, OrBinary, |
24
903f91163f23
- Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents:
23
diff
changeset
|
60 AndAssign, AndLogical, AndBinary, |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
61 PlusAssign, PlusPlus, Plus, |
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
62 MinusAssign, MinusMinus, Minus, |
32
d7011daa4740
- Added missing commas after the items in the messages table.
aziz
parents:
31
diff
changeset
|
63 DivAssign, Div, |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
64 MulAssign, Mul, |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
65 ModAssign, Mod, |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
66 XorAssign, Xor, |
27 | 67 CatAssign, Catenate, |
39
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
68 Tilde, |
71 | 69 Identity, NotIdentity, |
23
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
70 |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
71 Colon, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
72 Semicolon, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
73 Question, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
74 Comma, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
75 Dollar, |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
76 |
28 | 77 /* Keywords: |
78 NB.: Token.isKeyword() depends on this list being contiguous. | |
79 */ | |
80 Abstract,Alias,Align,Asm,Assert,Auto,Body, | |
81 Bool,Break,Byte,Case,Cast,Catch,Cdouble, | |
82 Cent,Cfloat,Char,Class,Const,Continue,Creal, | |
83 Dchar,Debug,Default,Delegate,Delete,Deprecated,Do, | |
84 Double,Else,Enum,Export,Extern,False,Final, | |
85 Finally,Float,For,Foreach,Foreach_reverse,Function,Goto, | |
86 Idouble,If,Ifloat,Import,In,Inout,Int, | |
269 | 87 Interface,Invariant,Ireal,Is,Lazy,Long,Macro/+D2.0+/, |
28 | 88 Mixin,Module,New,Null,Out,Override,Package, |
269 | 89 Pragma,Private,Protected,Public,Real,Ref/+D2.0+/,Return, |
28 | 90 Scope,Short,Static,Struct,Super,Switch,Synchronized, |
269 | 91 Template,This,Throw,Traits/+D2.0+/,True,Try,Typedef,Typeid, |
28 | 92 Typeof,Ubyte,Ucent,Uint,Ulong,Union,Unittest, |
93 Ushort,Version,Void,Volatile,Wchar,While,With, | |
94 | |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
95 HEAD, // start of linked list |
4 | 96 EOF |
3 | 97 } |
98 | |
28 | 99 alias TOK.Abstract KeywordsBegin; |
100 alias TOK.With KeywordsEnd; | |
101 | |
0 | 102 struct Token |
103 { | |
3 | 104 TOK type; |
84
ac8d961d10d1
- Added code for parsing This-,Super-,Null-,Bool-,Dollar-,CharLiteral- and StringLiteralExpression.
aziz
parents:
82
diff
changeset
|
105 // Position pos; |
0 | 106 |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
107 Token* next, prev; |
131 | 108 |
4 | 109 char* start; |
110 char* end; | |
111 | |
0 | 112 union |
113 { | |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
114 struct |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
115 { |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
116 string str; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
117 char pf; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
118 } |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
119 dchar dchar_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
120 long long_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
121 ulong ulong_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
122 int int_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
123 uint uint_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
124 float float_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
125 double double_; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
58
diff
changeset
|
126 real real_; |
0 | 127 } |
28 | 128 |
107
722c05bbd5eb
- Implemented parseEnumDeclaration() and added class EnumDeclaration.
aziz
parents:
97
diff
changeset
|
129 alias srcText identifier; |
722c05bbd5eb
- Implemented parseEnumDeclaration() and added class EnumDeclaration.
aziz
parents:
97
diff
changeset
|
130 |
65 | 131 string srcText() |
28 | 132 { |
30 | 133 assert(start && end); |
28 | 134 return start[0 .. end - start]; |
135 } | |
136 | |
210 | 137 static string toString(TOK tok) |
208 | 138 { |
139 return tokToString[tok]; | |
140 } | |
141 | |
28 | 142 bool isKeyword() |
143 { | |
163
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
144 return KeywordsBegin <= type && type <= KeywordsEnd; |
28 | 145 } |
208 | 146 |
163
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
147 int opEquals(TOK type2) |
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
148 { |
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
149 return type == type2; |
f27a98bb17c7
- Fix: when parsing Declarator fails, type and ident is set to null.
aziz
parents:
131
diff
changeset
|
150 } |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
151 |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
152 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
|
153 { |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
154 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
|
155 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
|
156 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
|
157 *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
|
158 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
|
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 |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
161 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
|
162 { |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
210
diff
changeset
|
163 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
|
164 } |
208 | 165 } |
166 | |
167 string[] tokToString = [ | |
168 "Invalid", | |
169 | |
170 "Identifier", | |
171 "Comment", | |
172 "String", | |
312 | 173 "Special", |
208 | 174 "CharLiteral", "WCharLiteral", "DCharLiteral", |
175 | |
176 "Int32", "Int64", "Uint32", "Uint64", | |
177 "Float32", "Float64", "Float80", | |
178 "Imaginary32", "Imaginary64", "Imaginary80", | |
179 | |
180 "(", | |
181 ")", | |
182 "[", | |
183 "]", | |
184 "{", | |
185 "}", | |
186 | |
187 ".", "..", "...", | |
188 | |
189 "Unordered", | |
190 "UorE", | |
191 "UorG", | |
192 "UorGorE", | |
193 "UorL", | |
194 "UorLorE", | |
195 "LorEorG", | |
196 "LorG", | |
197 | |
198 "=", "==", "!=", "!", | |
199 "<=", "<", | |
200 ">=", ">", | |
201 "<<=", "<<", | |
202 ">>=",">>", | |
203 ">>>=", ">>>", | |
204 "|=", "||", "|", | |
205 "&=", "&&", "&", | |
206 "+=", "++", "+", | |
207 "-=", "--", "-", | |
208 "/=", "/", | |
209 "*=", "*", | |
210 "%=", "%", | |
211 "^=", "^", | |
212 "~=", "~", | |
213 "~", | |
214 "is", "!is", | |
215 | |
216 ":", | |
217 ";", | |
218 "?", | |
219 ",", | |
220 "$", | |
221 | |
222 "abstract","alias","align","asm","assert","auto","body", | |
223 "bool","break","byte","case","cast","catch","cdouble", | |
224 "cent","cfloat","char","class","const","continue","creal", | |
225 "dchar","debug","default","delegate","delete","deprecated","do", | |
226 "double","else","enum","export","extern","false","final", | |
227 "finally","float","for","foreach","foreach_reverse","function","goto", | |
228 "idouble","if","ifloat","import","in","inout","int", | |
229 "interface","invariant","ireal","is","lazy","long","macro", | |
230 "mixin","module","new","null","out","override","package", | |
231 "pragma","private","protected","public","real","ref","return", | |
232 "scope","short","static","struct","super","switch","synchronized", | |
233 "template","this","throw","true","try","typedef","typeid", | |
234 "typeof","ubyte","ucent","uint","ulong","union","unittest", | |
235 "ushort","version","void","volatile","wchar","while","with", | |
236 | |
237 "EOF" | |
238 ]; |