Mercurial > projects > dang

comparison lexer/Lexer.d @ 89:a49bb982a7b0 new_gen

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Using the new SourceLocation system to handle errors. Also, this is the first push for making the errors don't throw, but continue to check the source.
author Anders Johnsen <skabet@gmail.com>
date Sun, 04 May 2008 20:27:01 +0200
parents eb5b2c719a39
children 1a24e61eb104
comparison
equal deleted inserted replaced
88:eb5b2c719a39 89:a49bb982a7b0
1 module lexer.Lexer; 1 module lexer.Lexer;
2 2
3 import misc.Error, 3 import basic.Message,
4 basic.SourceManager; 4 basic.SourceManager;
5 5
6 import lexer.Token, 6 import lexer.Token,
7 lexer.Keyword; 7 lexer.Keyword;
8 8
19 public: 19 public:
20 20
21 /** 21 /**
22 Create a new Lexer. 22 Create a new Lexer.
23 */ 23 */
24 this(SourceLocation start, SourceManager src_mgr) 24
25 { 25
26 this(SourceLocation start, SourceManager src_mgr, MessageHandler messages)
27 {
28 this.messages = messages;
26 sm = src_mgr; 29 sm = src_mgr;
27 start_loc = start; 30 start_loc = start;
28 position = 0; 31 position = 0;
29 source = sm.getRawData(start_loc); 32 source = sm.getRawData(start_loc);
30 33
110 Token t = this.next; 113 Token t = this.next;
111 this.position = oldPosition; 114 this.position = oldPosition;
112 return t; 115 return t;
113 } 116 }
114 117
115 /**
116 Return all errors that occurred while tokenizing the string.
117
118 TODO: Error system not implemented yet - this is a stub!
119 */
120 public Error[] getErrors()
121 {
122 return this.errors;
123 }
124
125 private: 118 private:
126 Token eq() 119 Token eq()
127 { 120 {
128 if(source[position] == '=') 121 if(source[position] == '=')
129 return Token(Tok.Eq, Loc(position++ - 1), 2); 122 return Token(Tok.Eq, Loc(position++ - 1), 2);
229 ++position; 222 ++position;
230 if(source[position-2] == '*') 223 if(source[position-2] == '*')
231 if(source[position-1] == '/') 224 if(source[position-1] == '/')
232 return this.next; 225 return this.next;
233 } 226 }
234 throw error(__LINE__, "Unexpected end of file. Unclosed comment block"); 227 messages.report(UnexpectedEOFBlock,Loc(position));
235 228
236 case '+': 229 case '+':
237 position += 2; 230 position += 2;
238 int nesting = 1; 231 int nesting = 1;
239 while(getNextChar != CharType.EOF) 232 while(getNextChar != CharType.EOF)
254 } 247 }
255 248
256 if(nesting == 0) 249 if(nesting == 0)
257 return this.next; 250 return this.next;
258 } 251 }
259 throw error(__LINE__, "Unexpected end of file. Unclosed comment block"); 252 messages.report(UnexpectedEOFBlock,Loc(position));
260 253
261 default: 254 default:
262 return Token(Tok.Slash, Loc(position - 1), 1); 255 return Token(Tok.Slash, Loc(position - 1), 1);
263 } 256 }
264 } 257 }
285 break; 278 break;
286 case CharType.Symbol: 279 case CharType.Symbol:
287 if(this.source[position+i] == '.') 280 if(this.source[position+i] == '.')
288 { 281 {
289 if(dot) 282 if(dot)
290 throw error(__LINE__,"Only one '.' is allowed in an floating number") 283 messages.report(OnlyOneDotFloating, Loc(position + i));
291 .tok(Token(Tok.Float, Loc(position + i), 1));
292 dot = true; 284 dot = true;
293 break; 285 break;
294 } 286 }
295 end = true; 287 end = true;
296 continue; 288 continue;
299 break; 291 break;
300 if (this.source[position+i] == 'e' || 292 if (this.source[position+i] == 'e' ||
301 this.source[position+i] == 'E') 293 this.source[position+i] == 'E')
302 { 294 {
303 if (e) 295 if (e)
304 throw error(__LINE__,"Only one '"~this.source[position+i] 296 messages.report(OnlyOneEFloating, Loc(position + i));
305 ~"' is allowed in an floating number");
306 e = true; 297 e = true;
307 break; 298 break;
308 } 299 }
309 end = true; 300 end = true;
310 continue; 301 continue;
363 char current = source[position + offset]; 354 char current = source[position + offset];
364 355
365 CharType c = charTable[current]; 356 CharType c = charTable[current];
366 357
367 if(c == CharType.INVALID) 358 if(c == CharType.INVALID)
368 throw error(__LINE__, "Read invalid symbol: '%0'").arg(current); 359 messages.report(InvalidSymbol, SLoc()).arg(current);
369 360
370 return c; 361 return c;
371 362
372 }
373
374 Error error(uint line, char[] msg)
375 {
376 return (new Error(msg));//.loc(Loc(position));
377 } 363 }
378 364
379 private final SourceLocation Loc(int pos = -1) 365 private final SourceLocation Loc(int pos = -1)
380 { 366 {
381 if (pos < 0) 367 if (pos < 0)
385 371
386 SourceManager sm; 372 SourceManager sm;
387 SourceLocation start_loc; 373 SourceLocation start_loc;
388 int position; 374 int position;
389 char[] source; 375 char[] source;
390 Error[] errors; 376 MessageHandler messages;
391 CharType[] charTable; 377 CharType[] charTable;
392 Token delegate()[] symbolFunctions; 378 Token delegate()[] symbolFunctions;
393 } 379 }
394 380
395 enum CharType : ubyte 381 enum CharType : ubyte