1 | package antlr; |
---|
2 | |
---|
3 | /* ANTLR Translator Generator |
---|
4 | * Project led by Terence Parr at http://www.cs.usfca.edu |
---|
5 | * Software rights: http://www.antlr.org/license.html |
---|
6 | * |
---|
7 | * $Id: //depot/code/org.antlr/release/antlr-2.7.7/antlr/JavaCodeGenerator.java#2 $ |
---|
8 | */ |
---|
9 | |
---|
10 | import java.io.IOException; |
---|
11 | import java.util.Enumeration; |
---|
12 | import java.util.Hashtable; |
---|
13 | |
---|
14 | import antlr.collections.impl.BitSet; |
---|
15 | import antlr.collections.impl.Vector; |
---|
16 | |
---|
17 | /**Generate MyParser.java, MyLexer.java and MyParserTokenTypes.java */ |
---|
18 | public class JavaCodeGenerator extends CodeGenerator { |
---|
19 | // non-zero if inside syntactic predicate generation |
---|
20 | /** Constant to indicate that we shouldn't generate a mapping entry */ |
---|
21 | public static final int NO_MAPPING = -999; |
---|
22 | /** |
---|
23 | * Constant to indicate that we should continue mapping based on the |
---|
24 | * last mapping seen by the SMAP generator (there may have been |
---|
25 | * intermediate mappings generated by nested elements) |
---|
26 | */ |
---|
27 | public static final int CONTINUE_LAST_MAPPING = -888; |
---|
28 | |
---|
29 | private JavaCodeGeneratorPrintWriterManager printWriterManager; |
---|
30 | private int defaultLine = NO_MAPPING; |
---|
31 | |
---|
32 | protected int syntacticPredLevel = 0; |
---|
33 | |
---|
34 | // Are we generating ASTs (for parsers and tree parsers) right now? |
---|
35 | protected boolean genAST = false; |
---|
36 | |
---|
37 | // Are we saving the text consumed (for lexers) right now? |
---|
38 | protected boolean saveText = false; |
---|
39 | |
---|
40 | // Grammar parameters set up to handle different grammar classes. |
---|
41 | // These are used to get instanceof tests out of code generation |
---|
42 | String labeledElementType; |
---|
43 | String labeledElementASTType; |
---|
44 | String labeledElementInit; |
---|
45 | String commonExtraArgs; |
---|
46 | String commonExtraParams; |
---|
47 | String commonLocalVars; |
---|
48 | String lt1Value; |
---|
49 | String exceptionThrown; |
---|
50 | String throwNoViable; |
---|
51 | |
---|
52 | /** Tracks the rule being generated. Used for mapTreeId */ |
---|
53 | RuleBlock currentRule; |
---|
54 | |
---|
55 | /** Tracks the rule or labeled subrule being generated. Used for |
---|
56 | AST generation. */ |
---|
57 | String currentASTResult; |
---|
58 | |
---|
59 | /** Mapping between the ids used in the current alt, and the |
---|
60 | * names of variables used to represent their AST values. |
---|
61 | */ |
---|
62 | Hashtable treeVariableMap = new Hashtable(); |
---|
63 | |
---|
64 | /** Used to keep track of which AST variables have been defined in a rule |
---|
65 | * (except for the #rule_name and #rule_name_in var's |
---|
66 | */ |
---|
67 | Hashtable declaredASTVariables = new Hashtable(); |
---|
68 | |
---|
69 | /* Count of unnamed generated variables */ |
---|
70 | int astVarNumber = 1; |
---|
71 | |
---|
72 | /** Special value used to mark duplicate in treeVariableMap */ |
---|
73 | protected static final String NONUNIQUE = new String(); |
---|
74 | |
---|
75 | public static final int caseSizeThreshold = 127; // ascii is max |
---|
76 | |
---|
77 | private Vector semPreds; |
---|
78 | |
---|
79 | /** Create a Java code-generator using the given Grammar. |
---|
80 | * The caller must still call setTool, setBehavior, and setAnalyzer |
---|
81 | * before generating code. |
---|
82 | */ |
---|
83 | public JavaCodeGenerator() { |
---|
84 | super(); |
---|
85 | charFormatter = new JavaCharFormatter(); |
---|
86 | } |
---|
87 | |
---|
88 | protected void printAction(String code) { |
---|
89 | printAction(code, defaultLine); |
---|
90 | } |
---|
91 | protected void printAction(String code, int sourceStartLine) { |
---|
92 | getPrintWriterManager().startMapping(sourceStartLine); |
---|
93 | super.printAction(code); |
---|
94 | getPrintWriterManager().endMapping(); |
---|
95 | } |
---|
96 | |
---|
97 | public void println(String code) { |
---|
98 | println(code, defaultLine); |
---|
99 | } |
---|
100 | public void println(String code, int sourceStartLine) { |
---|
101 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
102 | getPrintWriterManager().startSingleSourceLineMapping(sourceStartLine); |
---|
103 | super.println(code); |
---|
104 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
105 | getPrintWriterManager().endMapping(); |
---|
106 | } |
---|
107 | |
---|
108 | protected void print(String code) { |
---|
109 | print(code, defaultLine); |
---|
110 | } |
---|
111 | protected void print(String code, int sourceStartLine) { |
---|
112 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
113 | getPrintWriterManager().startMapping(sourceStartLine); |
---|
114 | super.print(code); |
---|
115 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
116 | getPrintWriterManager().endMapping(); |
---|
117 | } |
---|
118 | |
---|
119 | protected void _print(String code) { |
---|
120 | _print(code, defaultLine); |
---|
121 | } |
---|
122 | protected void _print(String code, int sourceStartLine) { |
---|
123 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
124 | getPrintWriterManager().startMapping(sourceStartLine); |
---|
125 | super._print(code); |
---|
126 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
127 | getPrintWriterManager().endMapping(); |
---|
128 | } |
---|
129 | |
---|
130 | protected void _println(String code) { |
---|
131 | _println(code, defaultLine); |
---|
132 | } |
---|
133 | protected void _println(String code, int sourceStartLine) { |
---|
134 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
135 | getPrintWriterManager().startMapping(sourceStartLine); |
---|
136 | super._println(code); |
---|
137 | if (sourceStartLine > 0 || sourceStartLine == CONTINUE_LAST_MAPPING) |
---|
138 | getPrintWriterManager().endMapping(); |
---|
139 | } |
---|
140 | |
---|
141 | /** Adds a semantic predicate string to the sem pred vector |
---|
142 | These strings will be used to build an array of sem pred names |
---|
143 | when building a debugging parser. This method should only be |
---|
144 | called when the debug option is specified |
---|
145 | */ |
---|
146 | protected int addSemPred(String predicate) { |
---|
147 | semPreds.appendElement(predicate); |
---|
148 | return semPreds.size() - 1; |
---|
149 | } |
---|
150 | |
---|
151 | public void exitIfError() { |
---|
152 | if (antlrTool.hasError()) { |
---|
153 | antlrTool.fatalError("Exiting due to errors."); |
---|
154 | } |
---|
155 | } |
---|
156 | |
---|
157 | /**Generate the parser, lexer, treeparser, and token types in Java */ |
---|
158 | public void gen() { |
---|
159 | // Do the code generation |
---|
160 | try { |
---|
161 | // Loop over all grammars |
---|
162 | Enumeration grammarIter = behavior.grammars.elements(); |
---|
163 | while (grammarIter.hasMoreElements()) { |
---|
164 | Grammar g = (Grammar)grammarIter.nextElement(); |
---|
165 | // Connect all the components to each other |
---|
166 | g.setGrammarAnalyzer(analyzer); |
---|
167 | g.setCodeGenerator(this); |
---|
168 | analyzer.setGrammar(g); |
---|
169 | // To get right overloading behavior across hetrogeneous grammars |
---|
170 | setupGrammarParameters(g); |
---|
171 | g.generate(); |
---|
172 | // print out the grammar with lookahead sets (and FOLLOWs) |
---|
173 | // System.out.print(g.toString()); |
---|
174 | exitIfError(); |
---|
175 | } |
---|
176 | |
---|
177 | // Loop over all token managers (some of which are lexers) |
---|
178 | Enumeration tmIter = behavior.tokenManagers.elements(); |
---|
179 | while (tmIter.hasMoreElements()) { |
---|
180 | TokenManager tm = (TokenManager)tmIter.nextElement(); |
---|
181 | if (!tm.isReadOnly()) { |
---|
182 | // Write the token manager tokens as Java |
---|
183 | // this must appear before genTokenInterchange so that |
---|
184 | // labels are set on string literals |
---|
185 | genTokenTypes(tm); |
---|
186 | // Write the token manager tokens as plain text |
---|
187 | genTokenInterchange(tm); |
---|
188 | } |
---|
189 | exitIfError(); |
---|
190 | } |
---|
191 | } |
---|
192 | catch (IOException e) { |
---|
193 | antlrTool.reportException(e, null); |
---|
194 | } |
---|
195 | } |
---|
196 | |
---|
197 | /** Generate code for the given grammar element. |
---|
198 | * @param blk The {...} action to generate |
---|
199 | */ |
---|
200 | public void gen(ActionElement action) { |
---|
201 | int oldDefaultLine = defaultLine; |
---|
202 | try { |
---|
203 | defaultLine = action.getLine(); |
---|
204 | if (DEBUG_CODE_GENERATOR) System.out.println("genAction(" + action + ")"); |
---|
205 | if (action.isSemPred) { |
---|
206 | genSemPred(action.actionText, action.line); |
---|
207 | } |
---|
208 | else { |
---|
209 | if (grammar.hasSyntacticPredicate) { |
---|
210 | println("if ( inputState.guessing==0 ) {"); |
---|
211 | tabs++; |
---|
212 | } |
---|
213 | |
---|
214 | // get the name of the followSet for the current rule so that we |
---|
215 | // can replace $FOLLOW in the .g file. |
---|
216 | ActionTransInfo tInfo = new ActionTransInfo(); |
---|
217 | String actionStr = processActionForSpecialSymbols(action.actionText, |
---|
218 | action.getLine(), |
---|
219 | currentRule, |
---|
220 | tInfo); |
---|
221 | |
---|
222 | if (tInfo.refRuleRoot != null) { |
---|
223 | // Somebody referenced "#rule", make sure translated var is valid |
---|
224 | // assignment to #rule is left as a ref also, meaning that assignments |
---|
225 | // with no other refs like "#rule = foo();" still forces this code to be |
---|
226 | // generated (unnecessarily). |
---|
227 | println(tInfo.refRuleRoot + " = (" + labeledElementASTType + ")currentAST.root;"); |
---|
228 | } |
---|
229 | |
---|
230 | // dump the translated action |
---|
231 | printAction(actionStr); |
---|
232 | |
---|
233 | if (tInfo.assignToRoot) { |
---|
234 | // Somebody did a "#rule=", reset internal currentAST.root |
---|
235 | println("currentAST.root = " + tInfo.refRuleRoot + ";"); |
---|
236 | // reset the child pointer too to be last sibling in sibling list |
---|
237 | println("currentAST.child = " + tInfo.refRuleRoot + "!=null &&" + tInfo.refRuleRoot + ".getFirstChild()!=null ?", NO_MAPPING); |
---|
238 | tabs++; |
---|
239 | println(tInfo.refRuleRoot + ".getFirstChild() : " + tInfo.refRuleRoot + ";"); |
---|
240 | tabs--; |
---|
241 | println("currentAST.advanceChildToEnd();"); |
---|
242 | } |
---|
243 | |
---|
244 | if (grammar.hasSyntacticPredicate) { |
---|
245 | tabs--; |
---|
246 | println("}", NO_MAPPING); |
---|
247 | } |
---|
248 | } |
---|
249 | } finally { |
---|
250 | defaultLine = oldDefaultLine; |
---|
251 | } |
---|
252 | } |
---|
253 | |
---|
254 | /** Generate code for the given grammar element. |
---|
255 | * @param blk The "x|y|z|..." block to generate |
---|
256 | */ |
---|
257 | public void gen(AlternativeBlock blk) { |
---|
258 | if (DEBUG_CODE_GENERATOR) System.out.println("gen(" + blk + ")"); |
---|
259 | println("{", NO_MAPPING); |
---|
260 | genBlockPreamble(blk); |
---|
261 | genBlockInitAction(blk); |
---|
262 | |
---|
263 | // Tell AST generation to build subrule result |
---|
264 | String saveCurrentASTResult = currentASTResult; |
---|
265 | if (blk.getLabel() != null) { |
---|
266 | currentASTResult = blk.getLabel(); |
---|
267 | } |
---|
268 | |
---|
269 | boolean ok = grammar.theLLkAnalyzer.deterministic(blk); |
---|
270 | |
---|
271 | JavaBlockFinishingInfo howToFinish = genCommonBlock(blk, true); |
---|
272 | genBlockFinish(howToFinish, throwNoViable, blk.getLine()); |
---|
273 | |
---|
274 | println("}", NO_MAPPING); |
---|
275 | |
---|
276 | // Restore previous AST generation |
---|
277 | currentASTResult = saveCurrentASTResult; |
---|
278 | } |
---|
279 | |
---|
280 | /** Generate code for the given grammar element. |
---|
281 | * @param blk The block-end element to generate. Block-end |
---|
282 | * elements are synthesized by the grammar parser to represent |
---|
283 | * the end of a block. |
---|
284 | */ |
---|
285 | public void gen(BlockEndElement end) { |
---|
286 | if (DEBUG_CODE_GENERATOR) System.out.println("genRuleEnd(" + end + ")"); |
---|
287 | } |
---|
288 | |
---|
289 | /** Generate code for the given grammar element. |
---|
290 | * @param blk The character literal reference to generate |
---|
291 | */ |
---|
292 | public void gen(CharLiteralElement atom) { |
---|
293 | if (DEBUG_CODE_GENERATOR) System.out.println("genChar(" + atom + ")"); |
---|
294 | |
---|
295 | if (atom.getLabel() != null) { |
---|
296 | println(atom.getLabel() + " = " + lt1Value + ";", atom.getLine()); |
---|
297 | } |
---|
298 | |
---|
299 | boolean oldsaveText = saveText; |
---|
300 | saveText = saveText && atom.getAutoGenType() == GrammarElement.AUTO_GEN_NONE; |
---|
301 | genMatch(atom); |
---|
302 | saveText = oldsaveText; |
---|
303 | } |
---|
304 | |
---|
305 | /** Generate code for the given grammar element. |
---|
306 | * @param blk The character-range reference to generate |
---|
307 | */ |
---|
308 | public void gen(CharRangeElement r) { |
---|
309 | int oldDefaultLine = defaultLine; |
---|
310 | try { |
---|
311 | defaultLine = r.getLine(); |
---|
312 | if (r.getLabel() != null && syntacticPredLevel == 0) { |
---|
313 | println(r.getLabel() + " = " + lt1Value + ";"); |
---|
314 | } |
---|
315 | boolean flag = ( grammar instanceof LexerGrammar && |
---|
316 | ( !saveText || |
---|
317 | r.getAutoGenType() == |
---|
318 | GrammarElement.AUTO_GEN_BANG ) ); |
---|
319 | if (flag) { |
---|
320 | println("_saveIndex=text.length();"); |
---|
321 | } |
---|
322 | |
---|
323 | println("matchRange(" + r.beginText + "," + r.endText + ");"); |
---|
324 | |
---|
325 | if (flag) { |
---|
326 | println("text.setLength(_saveIndex);"); |
---|
327 | } |
---|
328 | } finally { |
---|
329 | defaultLine = oldDefaultLine; |
---|
330 | } |
---|
331 | } |
---|
332 | |
---|
333 | /** Generate the lexer Java file */ |
---|
334 | public void gen(LexerGrammar g) throws IOException { |
---|
335 | int oldDefaultLine = defaultLine; |
---|
336 | try { |
---|
337 | defaultLine = NO_MAPPING; |
---|
338 | // If debugging, create a new sempred vector for this grammar |
---|
339 | if (g.debuggingOutput) |
---|
340 | semPreds = new Vector(); |
---|
341 | |
---|
342 | setGrammar(g); |
---|
343 | if (!(grammar instanceof LexerGrammar)) { |
---|
344 | antlrTool.panic("Internal error generating lexer"); |
---|
345 | } |
---|
346 | |
---|
347 | // SAS: moved output creation to method so a subclass can change |
---|
348 | // how the output is generated (for VAJ interface) |
---|
349 | currentOutput = getPrintWriterManager().setupOutput(antlrTool, grammar); |
---|
350 | |
---|
351 | genAST = false; // no way to gen trees. |
---|
352 | saveText = true; // save consumed characters. |
---|
353 | |
---|
354 | tabs = 0; |
---|
355 | |
---|
356 | // Generate header common to all Java output files |
---|
357 | genHeader(); |
---|
358 | // Do not use printAction because we assume tabs==0 |
---|
359 | |
---|
360 | try { |
---|
361 | defaultLine = behavior.getHeaderActionLine(""); |
---|
362 | println(behavior.getHeaderAction("")); |
---|
363 | } finally { |
---|
364 | defaultLine = NO_MAPPING; |
---|
365 | } |
---|
366 | |
---|
367 | // Generate header specific to lexer Java file |
---|
368 | // println("import java.io.FileInputStream;"); |
---|
369 | println("import java.io.InputStream;"); |
---|
370 | println("import antlr.TokenStreamException;"); |
---|
371 | println("import antlr.TokenStreamIOException;"); |
---|
372 | println("import antlr.TokenStreamRecognitionException;"); |
---|
373 | println("import antlr.CharStreamException;"); |
---|
374 | println("import antlr.CharStreamIOException;"); |
---|
375 | println("import antlr.ANTLRException;"); |
---|
376 | println("import java.io.Reader;"); |
---|
377 | println("import java.util.Hashtable;"); |
---|
378 | println("import antlr." + grammar.getSuperClass() + ";"); |
---|
379 | println("import antlr.InputBuffer;"); |
---|
380 | println("import antlr.ByteBuffer;"); |
---|
381 | println("import antlr.CharBuffer;"); |
---|
382 | println("import antlr.Token;"); |
---|
383 | println("import antlr.CommonToken;"); |
---|
384 | println("import antlr.RecognitionException;"); |
---|
385 | println("import antlr.NoViableAltForCharException;"); |
---|
386 | println("import antlr.MismatchedCharException;"); |
---|
387 | println("import antlr.TokenStream;"); |
---|
388 | println("import antlr.ANTLRHashString;"); |
---|
389 | println("import antlr.LexerSharedInputState;"); |
---|
390 | println("import antlr.collections.impl.BitSet;"); |
---|
391 | println("import antlr.SemanticException;"); |
---|
392 | |
---|
393 | // Generate user-defined lexer file preamble |
---|
394 | println(grammar.preambleAction.getText()); |
---|
395 | |
---|
396 | // Generate lexer class definition |
---|
397 | String sup = null; |
---|
398 | if (grammar.superClass != null) { |
---|
399 | sup = grammar.superClass; |
---|
400 | } |
---|
401 | else { |
---|
402 | sup = "antlr." + grammar.getSuperClass(); |
---|
403 | } |
---|
404 | |
---|
405 | // print javadoc comment if any |
---|
406 | if (grammar.comment != null) { |
---|
407 | _println(grammar.comment); |
---|
408 | } |
---|
409 | |
---|
410 | // get prefix (replaces "public" and lets user specify) |
---|
411 | String prefix = "public"; |
---|
412 | Token tprefix = (Token)grammar.options.get("classHeaderPrefix"); |
---|
413 | if (tprefix != null) { |
---|
414 | String p = StringUtils.stripFrontBack(tprefix.getText(), "\"", "\""); |
---|
415 | if (p != null) { |
---|
416 | prefix = p; |
---|
417 | } |
---|
418 | } |
---|
419 | |
---|
420 | print(prefix+" "); |
---|
421 | print("class " + grammar.getClassName() + " extends " + sup); |
---|
422 | println(" implements " + grammar.tokenManager.getName() + TokenTypesFileSuffix + ", TokenStream"); |
---|
423 | Token tsuffix = (Token)grammar.options.get("classHeaderSuffix"); |
---|
424 | if (tsuffix != null) { |
---|
425 | String suffix = StringUtils.stripFrontBack(tsuffix.getText(), "\"", "\""); |
---|
426 | if (suffix != null) { |
---|
427 | print(", " + suffix); // must be an interface name for Java |
---|
428 | } |
---|
429 | } |
---|
430 | println(" {"); |
---|
431 | |
---|
432 | // Generate user-defined lexer class members |
---|
433 | print( |
---|
434 | processActionForSpecialSymbols(grammar.classMemberAction.getText(), grammar.classMemberAction.getLine(), currentRule, null), |
---|
435 | grammar.classMemberAction.getLine() |
---|
436 | ); |
---|
437 | |
---|
438 | // |
---|
439 | // Generate the constructor from InputStream, which in turn |
---|
440 | // calls the ByteBuffer constructor |
---|
441 | // |
---|
442 | println("public " + grammar.getClassName() + "(InputStream in) {"); |
---|
443 | tabs++; |
---|
444 | println("this(new ByteBuffer(in));"); |
---|
445 | tabs--; |
---|
446 | println("}"); |
---|
447 | |
---|
448 | // |
---|
449 | // Generate the constructor from Reader, which in turn |
---|
450 | // calls the CharBuffer constructor |
---|
451 | // |
---|
452 | println("public " + grammar.getClassName() + "(Reader in) {"); |
---|
453 | tabs++; |
---|
454 | println("this(new CharBuffer(in));"); |
---|
455 | tabs--; |
---|
456 | println("}"); |
---|
457 | |
---|
458 | println("public " + grammar.getClassName() + "(InputBuffer ib) {"); |
---|
459 | tabs++; |
---|
460 | // if debugging, wrap the input buffer in a debugger |
---|
461 | if (grammar.debuggingOutput) |
---|
462 | println("this(new LexerSharedInputState(new antlr.debug.DebuggingInputBuffer(ib)));"); |
---|
463 | else |
---|
464 | println("this(new LexerSharedInputState(ib));"); |
---|
465 | tabs--; |
---|
466 | println("}"); |
---|
467 | |
---|
468 | // |
---|
469 | // Generate the constructor from InputBuffer (char or byte) |
---|
470 | // |
---|
471 | println("public " + grammar.getClassName() + "(LexerSharedInputState state) {"); |
---|
472 | tabs++; |
---|
473 | |
---|
474 | println("super(state);"); |
---|
475 | // if debugging, set up array variables and call user-overridable |
---|
476 | // debugging setup method |
---|
477 | if (grammar.debuggingOutput) { |
---|
478 | println(" ruleNames = _ruleNames;"); |
---|
479 | println(" semPredNames = _semPredNames;"); |
---|
480 | println(" setupDebugging();"); |
---|
481 | } |
---|
482 | |
---|
483 | // Generate the setting of various generated options. |
---|
484 | // These need to be before the literals since ANTLRHashString depends on |
---|
485 | // the casesensitive stuff. |
---|
486 | println("caseSensitiveLiterals = " + g.caseSensitiveLiterals + ";"); |
---|
487 | println("setCaseSensitive(" + g.caseSensitive + ");"); |
---|
488 | |
---|
489 | // Generate the initialization of a hashtable |
---|
490 | // containing the string literals used in the lexer |
---|
491 | // The literals variable itself is in CharScanner |
---|
492 | println("literals = new Hashtable();"); |
---|
493 | Enumeration keys = grammar.tokenManager.getTokenSymbolKeys(); |
---|
494 | while (keys.hasMoreElements()) { |
---|
495 | String key = (String)keys.nextElement(); |
---|
496 | if (key.charAt(0) != '"') { |
---|
497 | continue; |
---|
498 | } |
---|
499 | TokenSymbol sym = grammar.tokenManager.getTokenSymbol(key); |
---|
500 | if (sym instanceof StringLiteralSymbol) { |
---|
501 | StringLiteralSymbol s = (StringLiteralSymbol)sym; |
---|
502 | println("literals.put(new ANTLRHashString(" + s.getId() + ", this), new Integer(" + s.getTokenType() + "));"); |
---|
503 | } |
---|
504 | } |
---|
505 | tabs--; |
---|
506 | |
---|
507 | Enumeration ids; |
---|
508 | println("}"); |
---|
509 | |
---|
510 | // generate the rule name array for debugging |
---|
511 | if (grammar.debuggingOutput) { |
---|
512 | println("private static final String _ruleNames[] = {"); |
---|
513 | |
---|
514 | ids = grammar.rules.elements(); |
---|
515 | int ruleNum = 0; |
---|
516 | while (ids.hasMoreElements()) { |
---|
517 | GrammarSymbol sym = (GrammarSymbol)ids.nextElement(); |
---|
518 | if (sym instanceof RuleSymbol) |
---|
519 | println(" \"" + ((RuleSymbol)sym).getId() + "\","); |
---|
520 | } |
---|
521 | println("};"); |
---|
522 | } |
---|
523 | |
---|
524 | // Generate nextToken() rule. |
---|
525 | // nextToken() is a synthetic lexer rule that is the implicit OR of all |
---|
526 | // user-defined lexer rules. |
---|
527 | genNextToken(); |
---|
528 | |
---|
529 | // Generate code for each rule in the lexer |
---|
530 | ids = grammar.rules.elements(); |
---|
531 | int ruleNum = 0; |
---|
532 | while (ids.hasMoreElements()) { |
---|
533 | RuleSymbol sym = (RuleSymbol)ids.nextElement(); |
---|
534 | // Don't generate the synthetic rules |
---|
535 | if (!sym.getId().equals("mnextToken")) { |
---|
536 | genRule(sym, false, ruleNum++); |
---|
537 | } |
---|
538 | exitIfError(); |
---|
539 | } |
---|
540 | |
---|
541 | // Generate the semantic predicate map for debugging |
---|
542 | if (grammar.debuggingOutput) |
---|
543 | genSemPredMap(); |
---|
544 | |
---|
545 | // Generate the bitsets used throughout the lexer |
---|
546 | genBitsets(bitsetsUsed, ((LexerGrammar)grammar).charVocabulary.size()); |
---|
547 | |
---|
548 | println(""); |
---|
549 | println("}"); |
---|
550 | |
---|
551 | // Close the lexer output stream |
---|
552 | getPrintWriterManager().finishOutput(); |
---|
553 | } finally { |
---|
554 | defaultLine = oldDefaultLine; |
---|
555 | } |
---|
556 | } |
---|
557 | |
---|
558 | /** Generate code for the given grammar element. |
---|
559 | * @param blk The (...)+ block to generate |
---|
560 | */ |
---|
561 | public void gen(OneOrMoreBlock blk) { |
---|
562 | int oldDefaultLine = defaultLine; |
---|
563 | try { |
---|
564 | defaultLine = blk.getLine(); |
---|
565 | if (DEBUG_CODE_GENERATOR) System.out.println("gen+(" + blk + ")"); |
---|
566 | String label; |
---|
567 | String cnt; |
---|
568 | println("{", NO_MAPPING); |
---|
569 | genBlockPreamble(blk); |
---|
570 | if (blk.getLabel() != null) { |
---|
571 | cnt = "_cnt_" + blk.getLabel(); |
---|
572 | } |
---|
573 | else { |
---|
574 | cnt = "_cnt" + blk.ID; |
---|
575 | } |
---|
576 | println("int " + cnt + "=0;"); |
---|
577 | if (blk.getLabel() != null) { |
---|
578 | label = blk.getLabel(); |
---|
579 | } |
---|
580 | else { |
---|
581 | label = "_loop" + blk.ID; |
---|
582 | } |
---|
583 | println(label + ":"); |
---|
584 | println("do {"); |
---|
585 | tabs++; |
---|
586 | // generate the init action for ()+ ()* inside the loop |
---|
587 | // this allows us to do usefull EOF checking... |
---|
588 | genBlockInitAction(blk); |
---|
589 | |
---|
590 | // Tell AST generation to build subrule result |
---|
591 | String saveCurrentASTResult = currentASTResult; |
---|
592 | if (blk.getLabel() != null) { |
---|
593 | currentASTResult = blk.getLabel(); |
---|
594 | } |
---|
595 | |
---|
596 | boolean ok = grammar.theLLkAnalyzer.deterministic(blk); |
---|
597 | |
---|
598 | // generate exit test if greedy set to false |
---|
599 | // and an alt is ambiguous with exit branch |
---|
600 | // or when lookahead derived purely from end-of-file |
---|
601 | // Lookahead analysis stops when end-of-file is hit, |
---|
602 | // returning set {epsilon}. Since {epsilon} is not |
---|
603 | // ambig with any real tokens, no error is reported |
---|
604 | // by deterministic() routines and we have to check |
---|
605 | // for the case where the lookahead depth didn't get |
---|
606 | // set to NONDETERMINISTIC (this only happens when the |
---|
607 | // FOLLOW contains real atoms + epsilon). |
---|
608 | boolean generateNonGreedyExitPath = false; |
---|
609 | int nonGreedyExitDepth = grammar.maxk; |
---|
610 | |
---|
611 | if (!blk.greedy && |
---|
612 | blk.exitLookaheadDepth <= grammar.maxk && |
---|
613 | blk.exitCache[blk.exitLookaheadDepth].containsEpsilon()) { |
---|
614 | generateNonGreedyExitPath = true; |
---|
615 | nonGreedyExitDepth = blk.exitLookaheadDepth; |
---|
616 | } |
---|
617 | else if (!blk.greedy && |
---|
618 | blk.exitLookaheadDepth == LLkGrammarAnalyzer.NONDETERMINISTIC) { |
---|
619 | generateNonGreedyExitPath = true; |
---|
620 | } |
---|
621 | |
---|
622 | // generate exit test if greedy set to false |
---|
623 | // and an alt is ambiguous with exit branch |
---|
624 | if (generateNonGreedyExitPath) { |
---|
625 | if (DEBUG_CODE_GENERATOR) { |
---|
626 | System.out.println("nongreedy (...)+ loop; exit depth is " + |
---|
627 | blk.exitLookaheadDepth); |
---|
628 | } |
---|
629 | String predictExit = |
---|
630 | getLookaheadTestExpression(blk.exitCache, |
---|
631 | nonGreedyExitDepth); |
---|
632 | println("// nongreedy exit test", NO_MAPPING); |
---|
633 | println("if ( " + cnt + ">=1 && " + predictExit + ") break " + label + ";", CONTINUE_LAST_MAPPING); |
---|
634 | } |
---|
635 | |
---|
636 | JavaBlockFinishingInfo howToFinish = genCommonBlock(blk, false); |
---|
637 | genBlockFinish( |
---|
638 | howToFinish, |
---|
639 | "if ( " + cnt + ">=1 ) { break " + label + "; } else {" + throwNoViable + "}", |
---|
640 | blk.getLine() |
---|
641 | ); |
---|
642 | |
---|
643 | println(cnt + "++;"); |
---|
644 | tabs--; |
---|
645 | println("} while (true);"); |
---|
646 | println("}"); |
---|
647 | |
---|
648 | // Restore previous AST generation |
---|
649 | currentASTResult = saveCurrentASTResult; |
---|
650 | } finally { |
---|
651 | defaultLine = oldDefaultLine; |
---|
652 | } |
---|
653 | } |
---|
654 | |
---|
655 | /** Generate the parser Java file */ |
---|
656 | public void gen(ParserGrammar g) throws IOException { |
---|
657 | int oldDefaultLine = defaultLine; |
---|
658 | try { |
---|
659 | defaultLine = NO_MAPPING; |
---|
660 | // if debugging, set up a new vector to keep track of sempred |
---|
661 | // strings for this grammar |
---|
662 | if (g.debuggingOutput) |
---|
663 | semPreds = new Vector(); |
---|
664 | |
---|
665 | setGrammar(g); |
---|
666 | if (!(grammar instanceof ParserGrammar)) { |
---|
667 | antlrTool.panic("Internal error generating parser"); |
---|
668 | } |
---|
669 | |
---|
670 | // Open the output stream for the parser and set the currentOutput |
---|
671 | // SAS: moved file setup so subclass could do it (for VAJ interface) |
---|
672 | currentOutput = getPrintWriterManager().setupOutput(antlrTool,grammar); |
---|
673 | |
---|
674 | genAST = grammar.buildAST; |
---|
675 | |
---|
676 | tabs = 0; |
---|
677 | |
---|
678 | // Generate the header common to all output files. |
---|
679 | genHeader(); |
---|
680 | // Do not use printAction because we assume tabs==0 |
---|
681 | try { |
---|
682 | defaultLine = behavior.getHeaderActionLine(""); |
---|
683 | println(behavior.getHeaderAction("")); |
---|
684 | } finally { |
---|
685 | defaultLine = NO_MAPPING; |
---|
686 | } |
---|
687 | |
---|
688 | // Generate header for the parser |
---|
689 | println("import antlr.TokenBuffer;"); |
---|
690 | println("import antlr.TokenStreamException;"); |
---|
691 | println("import antlr.TokenStreamIOException;"); |
---|
692 | println("import antlr.ANTLRException;"); |
---|
693 | println("import antlr." + grammar.getSuperClass() + ";"); |
---|
694 | println("import antlr.Token;"); |
---|
695 | println("import antlr.TokenStream;"); |
---|
696 | println("import antlr.RecognitionException;"); |
---|
697 | println("import antlr.NoViableAltException;"); |
---|
698 | println("import antlr.MismatchedTokenException;"); |
---|
699 | println("import antlr.SemanticException;"); |
---|
700 | println("import antlr.ParserSharedInputState;"); |
---|
701 | println("import antlr.collections.impl.BitSet;"); |
---|
702 | if ( genAST ) { |
---|
703 | println("import antlr.collections.AST;"); |
---|
704 | println("import java.util.Hashtable;"); |
---|
705 | println("import antlr.ASTFactory;"); |
---|
706 | println("import antlr.ASTPair;"); |
---|
707 | println("import antlr.collections.impl.ASTArray;"); |
---|
708 | } |
---|
709 | |
---|
710 | // Output the user-defined parser preamble |
---|
711 | println(grammar.preambleAction.getText()); |
---|
712 | |
---|
713 | // Generate parser class definition |
---|
714 | String sup = null; |
---|
715 | if (grammar.superClass != null) |
---|
716 | sup = grammar.superClass; |
---|
717 | else |
---|
718 | sup = "antlr." + grammar.getSuperClass(); |
---|
719 | |
---|
720 | // print javadoc comment if any |
---|
721 | if (grammar.comment != null) { |
---|
722 | _println(grammar.comment); |
---|
723 | } |
---|
724 | |
---|
725 | // get prefix (replaces "public" and lets user specify) |
---|
726 | String prefix = "public"; |
---|
727 | Token tprefix = (Token)grammar.options.get("classHeaderPrefix"); |
---|
728 | if (tprefix != null) { |
---|
729 | String p = StringUtils.stripFrontBack(tprefix.getText(), "\"", "\""); |
---|
730 | if (p != null) { |
---|
731 | prefix = p; |
---|
732 | } |
---|
733 | } |
---|
734 | |
---|
735 | print(prefix+" "); |
---|
736 | print("class " + grammar.getClassName() + " extends " + sup); |
---|
737 | println(" implements " + grammar.tokenManager.getName() + TokenTypesFileSuffix); |
---|
738 | |
---|
739 | Token tsuffix = (Token)grammar.options.get("classHeaderSuffix"); |
---|
740 | if (tsuffix != null) { |
---|
741 | String suffix = StringUtils.stripFrontBack(tsuffix.getText(), "\"", "\""); |
---|
742 | if (suffix != null) |
---|
743 | print(", " + suffix); // must be an interface name for Java |
---|
744 | } |
---|
745 | println(" {"); |
---|
746 | |
---|
747 | // set up an array of all the rule names so the debugger can |
---|
748 | // keep track of them only by number -- less to store in tree... |
---|
749 | if (grammar.debuggingOutput) { |
---|
750 | println("private static final String _ruleNames[] = {"); |
---|
751 | |
---|
752 | Enumeration ids = grammar.rules.elements(); |
---|
753 | int ruleNum = 0; |
---|
754 | while (ids.hasMoreElements()) { |
---|
755 | GrammarSymbol sym = (GrammarSymbol)ids.nextElement(); |
---|
756 | if (sym instanceof RuleSymbol) |
---|
757 | println(" \"" + ((RuleSymbol)sym).getId() + "\","); |
---|
758 | } |
---|
759 | println("};"); |
---|
760 | } |
---|
761 | |
---|
762 | // Generate user-defined parser class members |
---|
763 | print( |
---|
764 | processActionForSpecialSymbols(grammar.classMemberAction.getText(), grammar.classMemberAction.getLine(), currentRule, null), |
---|
765 | grammar.classMemberAction.getLine() |
---|
766 | ); |
---|
767 | |
---|
768 | // Generate parser class constructor from TokenBuffer |
---|
769 | println(""); |
---|
770 | println("protected " + grammar.getClassName() + "(TokenBuffer tokenBuf, int k) {"); |
---|
771 | println(" super(tokenBuf,k);"); |
---|
772 | println(" tokenNames = _tokenNames;"); |
---|
773 | // if debugging, set up arrays and call the user-overridable |
---|
774 | // debugging setup method |
---|
775 | if (grammar.debuggingOutput) { |
---|
776 | println(" ruleNames = _ruleNames;"); |
---|
777 | println(" semPredNames = _semPredNames;"); |
---|
778 | println(" setupDebugging(tokenBuf);"); |
---|
779 | } |
---|
780 | if ( grammar.buildAST ) { |
---|
781 | println(" buildTokenTypeASTClassMap();"); |
---|
782 | println(" astFactory = new ASTFactory(getTokenTypeToASTClassMap());"); |
---|
783 | } |
---|
784 | println("}"); |
---|
785 | println(""); |
---|
786 | |
---|
787 | println("public " + grammar.getClassName() + "(TokenBuffer tokenBuf) {"); |
---|
788 | println(" this(tokenBuf," + grammar.maxk + ");"); |
---|
789 | println("}"); |
---|
790 | println(""); |
---|
791 | |
---|
792 | // Generate parser class constructor from TokenStream |
---|
793 | println("protected " + grammar.getClassName() + "(TokenStream lexer, int k) {"); |
---|
794 | println(" super(lexer,k);"); |
---|
795 | println(" tokenNames = _tokenNames;"); |
---|
796 | |
---|
797 | // if debugging, set up arrays and call the user-overridable |
---|
798 | // debugging setup method |
---|
799 | if (grammar.debuggingOutput) { |
---|
800 | println(" ruleNames = _ruleNames;"); |
---|
801 | println(" semPredNames = _semPredNames;"); |
---|
802 | println(" setupDebugging(lexer);"); |
---|
803 | } |
---|
804 | if ( grammar.buildAST ) { |
---|
805 | println(" buildTokenTypeASTClassMap();"); |
---|
806 | println(" astFactory = new ASTFactory(getTokenTypeToASTClassMap());"); |
---|
807 | } |
---|
808 | println("}"); |
---|
809 | println(""); |
---|
810 | |
---|
811 | println("public " + grammar.getClassName() + "(TokenStream lexer) {"); |
---|
812 | println(" this(lexer," + grammar.maxk + ");"); |
---|
813 | println("}"); |
---|
814 | println(""); |
---|
815 | |
---|
816 | println("public " + grammar.getClassName() + "(ParserSharedInputState state) {"); |
---|
817 | println(" super(state," + grammar.maxk + ");"); |
---|
818 | println(" tokenNames = _tokenNames;"); |
---|
819 | if ( grammar.buildAST ) { |
---|
820 | println(" buildTokenTypeASTClassMap();"); |
---|
821 | println(" astFactory = new ASTFactory(getTokenTypeToASTClassMap());"); |
---|
822 | } |
---|
823 | println("}"); |
---|
824 | println(""); |
---|
825 | |
---|
826 | // Generate code for each rule in the grammar |
---|
827 | Enumeration ids = grammar.rules.elements(); |
---|
828 | int ruleNum = 0; |
---|
829 | while (ids.hasMoreElements()) { |
---|
830 | GrammarSymbol sym = (GrammarSymbol)ids.nextElement(); |
---|
831 | if (sym instanceof RuleSymbol) { |
---|
832 | RuleSymbol rs = (RuleSymbol)sym; |
---|
833 | genRule(rs, rs.references.size() == 0, ruleNum++); |
---|
834 | } |
---|
835 | exitIfError(); |
---|
836 | } |
---|
837 | |
---|
838 | // Generate the token names |
---|
839 | genTokenStrings(); |
---|
840 | |
---|
841 | if ( grammar.buildAST ) { |
---|
842 | genTokenASTNodeMap(); |
---|
843 | } |
---|
844 | |
---|
845 | // Generate the bitsets used throughout the grammar |
---|
846 | genBitsets(bitsetsUsed, grammar.tokenManager.maxTokenType()); |
---|
847 | |
---|
848 | // Generate the semantic predicate map for debugging |
---|
849 | if (grammar.debuggingOutput) |
---|
850 | genSemPredMap(); |
---|
851 | |
---|
852 | // Close class definition |
---|
853 | println(""); |
---|
854 | println("}"); |
---|
855 | |
---|
856 | // Close the parser output stream |
---|
857 | getPrintWriterManager().finishOutput(); |
---|
858 | } finally { |
---|
859 | defaultLine = oldDefaultLine; |
---|
860 | } |
---|
861 | } |
---|
862 | |
---|
863 | /** Generate code for the given grammar element. |
---|
864 | * @param blk The rule-reference to generate |
---|
865 | */ |
---|
866 | public void gen(RuleRefElement rr) { |
---|
867 | int oldDefaultLine = defaultLine; |
---|
868 | try { |
---|
869 | defaultLine = rr.getLine(); |
---|
870 | if (DEBUG_CODE_GENERATOR) System.out.println("genRR(" + rr + ")"); |
---|
871 | RuleSymbol rs = (RuleSymbol)grammar.getSymbol(rr.targetRule); |
---|
872 | if (rs == null || !rs.isDefined()) { |
---|
873 | // Is this redundant??? |
---|
874 | antlrTool.error("Rule '" + rr.targetRule + "' is not defined", grammar.getFilename(), rr.getLine(), rr.getColumn()); |
---|
875 | return; |
---|
876 | } |
---|
877 | if (!(rs instanceof RuleSymbol)) { |
---|
878 | // Is this redundant??? |
---|
879 | antlrTool.error("'" + rr.targetRule + "' does not name a grammar rule", grammar.getFilename(), rr.getLine(), rr.getColumn()); |
---|
880 | return; |
---|
881 | } |
---|
882 | |
---|
883 | genErrorTryForElement(rr); |
---|
884 | |
---|
885 | // AST value for labeled rule refs in tree walker. |
---|
886 | // This is not AST construction; it is just the input tree node value. |
---|
887 | if (grammar instanceof TreeWalkerGrammar && |
---|
888 | rr.getLabel() != null && |
---|
889 | syntacticPredLevel == 0) { |
---|
890 | println(rr.getLabel() + " = _t==ASTNULL ? null : " + lt1Value + ";"); |
---|
891 | } |
---|
892 | |
---|
893 | // if in lexer and ! on rule ref or alt or rule, save buffer index to kill later |
---|
894 | if (grammar instanceof LexerGrammar && (!saveText || rr.getAutoGenType() == GrammarElement.AUTO_GEN_BANG)) { |
---|
895 | println("_saveIndex=text.length();"); |
---|
896 | } |
---|
897 | |
---|
898 | // Process return value assignment if any |
---|
899 | printTabs(); |
---|
900 | if (rr.idAssign != null) { |
---|
901 | // Warn if the rule has no return type |
---|
902 | if (rs.block.returnAction == null) { |
---|
903 | antlrTool.warning("Rule '" + rr.targetRule + "' has no return type", grammar.getFilename(), rr.getLine(), rr.getColumn()); |
---|
904 | } |
---|
905 | _print(rr.idAssign + "="); |
---|
906 | } |
---|
907 | else { |
---|
908 | // Warn about return value if any, but not inside syntactic predicate |
---|
909 | if (!(grammar instanceof LexerGrammar) && syntacticPredLevel == 0 && rs.block.returnAction != null) { |
---|
910 | antlrTool.warning("Rule '" + rr.targetRule + "' returns a value", grammar.getFilename(), rr.getLine(), rr.getColumn()); |
---|
911 | } |
---|
912 | } |
---|
913 | |
---|
914 | // Call the rule |
---|
915 | GenRuleInvocation(rr); |
---|
916 | |
---|
917 | // if in lexer and ! on element or alt or rule, save buffer index to kill later |
---|
918 | if (grammar instanceof LexerGrammar && (!saveText || rr.getAutoGenType() == GrammarElement.AUTO_GEN_BANG)) { |
---|
919 | println("text.setLength(_saveIndex);"); |
---|
920 | } |
---|
921 | |
---|
922 | // if not in a syntactic predicate |
---|
923 | if (syntacticPredLevel == 0) { |
---|
924 | boolean doNoGuessTest = ( |
---|
925 | grammar.hasSyntacticPredicate && |
---|
926 | ( |
---|
927 | grammar.buildAST && rr.getLabel() != null || |
---|
928 | (genAST && rr.getAutoGenType() == GrammarElement.AUTO_GEN_NONE) |
---|
929 | ) |
---|
930 | ); |
---|
931 | if (doNoGuessTest) { |
---|
932 | // println("if (inputState.guessing==0) {"); |
---|
933 | // tabs++; |
---|
934 | } |
---|
935 | |
---|
936 | if (grammar.buildAST && rr.getLabel() != null) { |
---|
937 | // always gen variable for rule return on labeled rules |
---|
938 | println(rr.getLabel() + "_AST = (" + labeledElementASTType + ")returnAST;"); |
---|
939 | } |
---|
940 | if (genAST) { |
---|
941 | switch (rr.getAutoGenType()) { |
---|
942 | case GrammarElement.AUTO_GEN_NONE: |
---|
943 | // println("theASTFactory.addASTChild(currentAST, returnAST);"); |
---|
944 | println("astFactory.addASTChild(currentAST, returnAST);"); |
---|
945 | break; |
---|
946 | case GrammarElement.AUTO_GEN_CARET: |
---|
947 | antlrTool.error("Internal: encountered ^ after rule reference"); |
---|
948 | break; |
---|
949 | default: |
---|
950 | break; |
---|
951 | } |
---|
952 | } |
---|
953 | |
---|
954 | // if a lexer and labeled, Token label defined at rule level, just set it here |
---|
955 | if (grammar instanceof LexerGrammar && rr.getLabel() != null) { |
---|
956 | println(rr.getLabel() + "=_returnToken;"); |
---|
957 | } |
---|
958 | |
---|
959 | if (doNoGuessTest) { |
---|
960 | // tabs--; |
---|
961 | // println("}"); |
---|
962 | } |
---|
963 | } |
---|
964 | genErrorCatchForElement(rr); |
---|
965 | } finally { |
---|
966 | defaultLine = oldDefaultLine; |
---|
967 | } |
---|
968 | } |
---|
969 | |
---|
970 | /** Generate code for the given grammar element. |
---|
971 | * @param blk The string-literal reference to generate |
---|
972 | */ |
---|
973 | public void gen(StringLiteralElement atom) { |
---|
974 | if (DEBUG_CODE_GENERATOR) System.out.println("genString(" + atom + ")"); |
---|
975 | |
---|
976 | // Variable declarations for labeled elements |
---|
977 | if (atom.getLabel() != null && syntacticPredLevel == 0) { |
---|
978 | println(atom.getLabel() + " = " + lt1Value + ";", atom.getLine()); |
---|
979 | } |
---|
980 | |
---|
981 | // AST |
---|
982 | genElementAST(atom); |
---|
983 | |
---|
984 | // is there a bang on the literal? |
---|
985 | boolean oldsaveText = saveText; |
---|
986 | saveText = saveText && atom.getAutoGenType() == GrammarElement.AUTO_GEN_NONE; |
---|
987 | |
---|
988 | // matching |
---|
989 | genMatch(atom); |
---|
990 | |
---|
991 | saveText = oldsaveText; |
---|
992 | |
---|
993 | // tack on tree cursor motion if doing a tree walker |
---|
994 | if (grammar instanceof TreeWalkerGrammar) { |
---|
995 | println("_t = _t.getNextSibling();", atom.getLine()); |
---|
996 | } |
---|
997 | } |
---|
998 | |
---|
999 | /** Generate code for the given grammar element. |
---|
1000 | * @param r The token-range reference to generate |
---|
1001 | */ |
---|
1002 | public void gen(TokenRangeElement r) { |
---|
1003 | genErrorTryForElement(r); |
---|
1004 | if (r.getLabel() != null && syntacticPredLevel == 0) { |
---|
1005 | println(r.getLabel() + " = " + lt1Value + ";", r.getLine()); |
---|
1006 | } |
---|
1007 | |
---|
1008 | // AST |
---|
1009 | genElementAST(r); |
---|
1010 | |
---|
1011 | // match |
---|
1012 | println("matchRange(" + r.beginText + "," + r.endText + ");", r.getLine()); |
---|
1013 | genErrorCatchForElement(r); |
---|
1014 | } |
---|
1015 | |
---|
1016 | /** Generate code for the given grammar element. |
---|
1017 | * @param blk The token-reference to generate |
---|
1018 | */ |
---|
1019 | public void gen(TokenRefElement atom) { |
---|
1020 | if (DEBUG_CODE_GENERATOR) System.out.println("genTokenRef(" + atom + ")"); |
---|
1021 | if (grammar instanceof LexerGrammar) { |
---|
1022 | antlrTool.panic("Token reference found in lexer"); |
---|
1023 | } |
---|
1024 | genErrorTryForElement(atom); |
---|
1025 | // Assign Token value to token label variable |
---|
1026 | if (atom.getLabel() != null && syntacticPredLevel == 0) { |
---|
1027 | println(atom.getLabel() + " = " + lt1Value + ";", atom.getLine()); |
---|
1028 | } |
---|
1029 | |
---|
1030 | // AST |
---|
1031 | genElementAST(atom); |
---|
1032 | // matching |
---|
1033 | genMatch(atom); |
---|
1034 | genErrorCatchForElement(atom); |
---|
1035 | |
---|
1036 | // tack on tree cursor motion if doing a tree walker |
---|
1037 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1038 | println("_t = _t.getNextSibling();", atom.getLine()); |
---|
1039 | } |
---|
1040 | } |
---|
1041 | |
---|
1042 | public void gen(TreeElement t) { |
---|
1043 | int oldDefaultLine = defaultLine; |
---|
1044 | try { |
---|
1045 | defaultLine = t.getLine(); |
---|
1046 | // save AST cursor |
---|
1047 | println("AST __t" + t.ID + " = _t;"); |
---|
1048 | |
---|
1049 | // If there is a label on the root, then assign that to the variable |
---|
1050 | if (t.root.getLabel() != null) { |
---|
1051 | println(t.root.getLabel() + " = _t==ASTNULL ? null :(" + labeledElementASTType + ")_t;", t.root.getLine()); |
---|
1052 | } |
---|
1053 | |
---|
1054 | // check for invalid modifiers ! and ^ on tree element roots |
---|
1055 | if ( t.root.getAutoGenType() == GrammarElement.AUTO_GEN_BANG ) { |
---|
1056 | antlrTool.error("Suffixing a root node with '!' is not implemented", |
---|
1057 | grammar.getFilename(), t.getLine(), t.getColumn()); |
---|
1058 | t.root.setAutoGenType(GrammarElement.AUTO_GEN_NONE); |
---|
1059 | } |
---|
1060 | if ( t.root.getAutoGenType() == GrammarElement.AUTO_GEN_CARET ) { |
---|
1061 | antlrTool.warning("Suffixing a root node with '^' is redundant; already a root", |
---|
1062 | grammar.getFilename(), t.getLine(), t.getColumn()); |
---|
1063 | t.root.setAutoGenType(GrammarElement.AUTO_GEN_NONE); |
---|
1064 | } |
---|
1065 | |
---|
1066 | // Generate AST variables |
---|
1067 | genElementAST(t.root); |
---|
1068 | if (grammar.buildAST) { |
---|
1069 | // Save the AST construction state |
---|
1070 | println("ASTPair __currentAST" + t.ID + " = currentAST.copy();"); |
---|
1071 | // Make the next item added a child of the TreeElement root |
---|
1072 | println("currentAST.root = currentAST.child;"); |
---|
1073 | println("currentAST.child = null;"); |
---|
1074 | } |
---|
1075 | |
---|
1076 | // match root |
---|
1077 | if ( t.root instanceof WildcardElement ) { |
---|
1078 | println("if ( _t==null ) throw new MismatchedTokenException();", t.root.getLine()); |
---|
1079 | } |
---|
1080 | else { |
---|
1081 | genMatch(t.root); |
---|
1082 | } |
---|
1083 | // move to list of children |
---|
1084 | println("_t = _t.getFirstChild();"); |
---|
1085 | |
---|
1086 | // walk list of children, generating code for each |
---|
1087 | for (int i = 0; i < t.getAlternatives().size(); i++) { |
---|
1088 | Alternative a = t.getAlternativeAt(i); |
---|
1089 | AlternativeElement e = a.head; |
---|
1090 | while (e != null) { |
---|
1091 | e.generate(); |
---|
1092 | e = e.next; |
---|
1093 | } |
---|
1094 | } |
---|
1095 | |
---|
1096 | if (grammar.buildAST) { |
---|
1097 | // restore the AST construction state to that just after the |
---|
1098 | // tree root was added |
---|
1099 | println("currentAST = __currentAST" + t.ID + ";"); |
---|
1100 | } |
---|
1101 | // restore AST cursor |
---|
1102 | println("_t = __t" + t.ID + ";"); |
---|
1103 | // move cursor to sibling of tree just parsed |
---|
1104 | println("_t = _t.getNextSibling();"); |
---|
1105 | } finally { |
---|
1106 | defaultLine = oldDefaultLine; |
---|
1107 | } |
---|
1108 | } |
---|
1109 | |
---|
1110 | /** Generate the tree-parser Java file */ |
---|
1111 | public void gen(TreeWalkerGrammar g) throws IOException { |
---|
1112 | int oldDefaultLine = defaultLine; |
---|
1113 | try { |
---|
1114 | defaultLine = NO_MAPPING; |
---|
1115 | // SAS: debugging stuff removed for now... |
---|
1116 | setGrammar(g); |
---|
1117 | if (!(grammar instanceof TreeWalkerGrammar)) { |
---|
1118 | antlrTool.panic("Internal error generating tree-walker"); |
---|
1119 | } |
---|
1120 | // Open the output stream for the parser and set the currentOutput |
---|
1121 | // SAS: move file open to method so subclass can override it |
---|
1122 | // (mainly for VAJ interface) |
---|
1123 | currentOutput = getPrintWriterManager().setupOutput(antlrTool,grammar); |
---|
1124 | |
---|
1125 | genAST = grammar.buildAST; |
---|
1126 | tabs = 0; |
---|
1127 | |
---|
1128 | // Generate the header common to all output files. |
---|
1129 | genHeader(); |
---|
1130 | // Do not use printAction because we assume tabs==0 |
---|
1131 | try { |
---|
1132 | defaultLine = behavior.getHeaderActionLine(""); |
---|
1133 | println(behavior.getHeaderAction("")); |
---|
1134 | } finally { |
---|
1135 | defaultLine = NO_MAPPING; |
---|
1136 | } |
---|
1137 | |
---|
1138 | // Generate header for the parser |
---|
1139 | println("import antlr." + grammar.getSuperClass() + ";"); |
---|
1140 | println("import antlr.Token;"); |
---|
1141 | println("import antlr.collections.AST;"); |
---|
1142 | println("import antlr.RecognitionException;"); |
---|
1143 | println("import antlr.ANTLRException;"); |
---|
1144 | println("import antlr.NoViableAltException;"); |
---|
1145 | println("import antlr.MismatchedTokenException;"); |
---|
1146 | println("import antlr.SemanticException;"); |
---|
1147 | println("import antlr.collections.impl.BitSet;"); |
---|
1148 | println("import antlr.ASTPair;"); |
---|
1149 | println("import antlr.collections.impl.ASTArray;"); |
---|
1150 | |
---|
1151 | // Output the user-defined parser premamble |
---|
1152 | println(grammar.preambleAction.getText()); |
---|
1153 | |
---|
1154 | // Generate parser class definition |
---|
1155 | String sup = null; |
---|
1156 | if (grammar.superClass != null) { |
---|
1157 | sup = grammar.superClass; |
---|
1158 | } |
---|
1159 | else { |
---|
1160 | sup = "antlr." + grammar.getSuperClass(); |
---|
1161 | } |
---|
1162 | println(""); |
---|
1163 | |
---|
1164 | // print javadoc comment if any |
---|
1165 | if (grammar.comment != null) { |
---|
1166 | _println(grammar.comment); |
---|
1167 | } |
---|
1168 | |
---|
1169 | // get prefix (replaces "public" and lets user specify) |
---|
1170 | String prefix = "public"; |
---|
1171 | Token tprefix = (Token)grammar.options.get("classHeaderPrefix"); |
---|
1172 | if (tprefix != null) { |
---|
1173 | String p = StringUtils.stripFrontBack(tprefix.getText(), "\"", "\""); |
---|
1174 | if (p != null) { |
---|
1175 | prefix = p; |
---|
1176 | } |
---|
1177 | } |
---|
1178 | |
---|
1179 | print(prefix+" "); |
---|
1180 | print("class " + grammar.getClassName() + " extends " + sup); |
---|
1181 | println(" implements " + grammar.tokenManager.getName() + TokenTypesFileSuffix); |
---|
1182 | Token tsuffix = (Token)grammar.options.get("classHeaderSuffix"); |
---|
1183 | if (tsuffix != null) { |
---|
1184 | String suffix = StringUtils.stripFrontBack(tsuffix.getText(), "\"", "\""); |
---|
1185 | if (suffix != null) { |
---|
1186 | print(", " + suffix); // must be an interface name for Java |
---|
1187 | } |
---|
1188 | } |
---|
1189 | println(" {"); |
---|
1190 | |
---|
1191 | // Generate user-defined parser class members |
---|
1192 | print( |
---|
1193 | processActionForSpecialSymbols(grammar.classMemberAction.getText(), grammar.classMemberAction.getLine(), currentRule, null), |
---|
1194 | grammar.classMemberAction.getLine() |
---|
1195 | ); |
---|
1196 | |
---|
1197 | // Generate default parser class constructor |
---|
1198 | println("public " + grammar.getClassName() + "() {"); |
---|
1199 | tabs++; |
---|
1200 | println("tokenNames = _tokenNames;"); |
---|
1201 | tabs--; |
---|
1202 | println("}"); |
---|
1203 | println(""); |
---|
1204 | |
---|
1205 | // Generate code for each rule in the grammar |
---|
1206 | Enumeration ids = grammar.rules.elements(); |
---|
1207 | int ruleNum = 0; |
---|
1208 | String ruleNameInits = ""; |
---|
1209 | while (ids.hasMoreElements()) { |
---|
1210 | GrammarSymbol sym = (GrammarSymbol)ids.nextElement(); |
---|
1211 | if (sym instanceof RuleSymbol) { |
---|
1212 | RuleSymbol rs = (RuleSymbol)sym; |
---|
1213 | genRule(rs, rs.references.size() == 0, ruleNum++); |
---|
1214 | } |
---|
1215 | exitIfError(); |
---|
1216 | } |
---|
1217 | |
---|
1218 | // Generate the token names |
---|
1219 | genTokenStrings(); |
---|
1220 | |
---|
1221 | // Generate the bitsets used throughout the grammar |
---|
1222 | genBitsets(bitsetsUsed, grammar.tokenManager.maxTokenType()); |
---|
1223 | |
---|
1224 | // Close class definition |
---|
1225 | println("}"); |
---|
1226 | println(""); |
---|
1227 | |
---|
1228 | // Close the parser output stream |
---|
1229 | getPrintWriterManager().finishOutput(); |
---|
1230 | } finally { |
---|
1231 | defaultLine = oldDefaultLine; |
---|
1232 | } |
---|
1233 | } |
---|
1234 | |
---|
1235 | /** Generate code for the given grammar element. |
---|
1236 | * @param wc The wildcard element to generate |
---|
1237 | */ |
---|
1238 | public void gen(WildcardElement wc) { |
---|
1239 | int oldDefaultLine = defaultLine; |
---|
1240 | try { |
---|
1241 | defaultLine = wc.getLine(); |
---|
1242 | // Variable assignment for labeled elements |
---|
1243 | if (wc.getLabel() != null && syntacticPredLevel == 0) { |
---|
1244 | println(wc.getLabel() + " = " + lt1Value + ";"); |
---|
1245 | } |
---|
1246 | |
---|
1247 | // AST |
---|
1248 | genElementAST(wc); |
---|
1249 | // Match anything but EOF |
---|
1250 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1251 | println("if ( _t==null ) throw new MismatchedTokenException();"); |
---|
1252 | } |
---|
1253 | else if (grammar instanceof LexerGrammar) { |
---|
1254 | if (grammar instanceof LexerGrammar && |
---|
1255 | (!saveText || wc.getAutoGenType() == GrammarElement.AUTO_GEN_BANG)) { |
---|
1256 | println("_saveIndex=text.length();"); |
---|
1257 | } |
---|
1258 | println("matchNot(EOF_CHAR);"); |
---|
1259 | if (grammar instanceof LexerGrammar && |
---|
1260 | (!saveText || wc.getAutoGenType() == GrammarElement.AUTO_GEN_BANG)) { |
---|
1261 | println("text.setLength(_saveIndex);"); // kill text atom put in buffer |
---|
1262 | } |
---|
1263 | } |
---|
1264 | else { |
---|
1265 | println("matchNot(" + getValueString(Token.EOF_TYPE) + ");"); |
---|
1266 | } |
---|
1267 | |
---|
1268 | // tack on tree cursor motion if doing a tree walker |
---|
1269 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1270 | println("_t = _t.getNextSibling();"); |
---|
1271 | } |
---|
1272 | } finally { |
---|
1273 | defaultLine = oldDefaultLine; |
---|
1274 | } |
---|
1275 | } |
---|
1276 | |
---|
1277 | /** Generate code for the given grammar element. |
---|
1278 | * @param blk The (...)* block to generate |
---|
1279 | */ |
---|
1280 | public void gen(ZeroOrMoreBlock blk) { |
---|
1281 | int oldDefaultLine = defaultLine; |
---|
1282 | try { |
---|
1283 | defaultLine = blk.getLine(); |
---|
1284 | if (DEBUG_CODE_GENERATOR) System.out.println("gen*(" + blk + ")"); |
---|
1285 | println("{"); |
---|
1286 | genBlockPreamble(blk); |
---|
1287 | String label; |
---|
1288 | if (blk.getLabel() != null) { |
---|
1289 | label = blk.getLabel(); |
---|
1290 | } |
---|
1291 | else { |
---|
1292 | label = "_loop" + blk.ID; |
---|
1293 | } |
---|
1294 | println(label + ":"); |
---|
1295 | println("do {"); |
---|
1296 | tabs++; |
---|
1297 | // generate the init action for ()* inside the loop |
---|
1298 | // this allows us to do usefull EOF checking... |
---|
1299 | genBlockInitAction(blk); |
---|
1300 | |
---|
1301 | // Tell AST generation to build subrule result |
---|
1302 | String saveCurrentASTResult = currentASTResult; |
---|
1303 | if (blk.getLabel() != null) { |
---|
1304 | currentASTResult = blk.getLabel(); |
---|
1305 | } |
---|
1306 | |
---|
1307 | boolean ok = grammar.theLLkAnalyzer.deterministic(blk); |
---|
1308 | |
---|
1309 | // generate exit test if greedy set to false |
---|
1310 | // and an alt is ambiguous with exit branch |
---|
1311 | // or when lookahead derived purely from end-of-file |
---|
1312 | // Lookahead analysis stops when end-of-file is hit, |
---|
1313 | // returning set {epsilon}. Since {epsilon} is not |
---|
1314 | // ambig with any real tokens, no error is reported |
---|
1315 | // by deterministic() routines and we have to check |
---|
1316 | // for the case where the lookahead depth didn't get |
---|
1317 | // set to NONDETERMINISTIC (this only happens when the |
---|
1318 | // FOLLOW contains real atoms + epsilon). |
---|
1319 | boolean generateNonGreedyExitPath = false; |
---|
1320 | int nonGreedyExitDepth = grammar.maxk; |
---|
1321 | |
---|
1322 | if (!blk.greedy && |
---|
1323 | blk.exitLookaheadDepth <= grammar.maxk && |
---|
1324 | blk.exitCache[blk.exitLookaheadDepth].containsEpsilon()) { |
---|
1325 | generateNonGreedyExitPath = true; |
---|
1326 | nonGreedyExitDepth = blk.exitLookaheadDepth; |
---|
1327 | } |
---|
1328 | else if (!blk.greedy && |
---|
1329 | blk.exitLookaheadDepth == LLkGrammarAnalyzer.NONDETERMINISTIC) { |
---|
1330 | generateNonGreedyExitPath = true; |
---|
1331 | } |
---|
1332 | if (generateNonGreedyExitPath) { |
---|
1333 | if (DEBUG_CODE_GENERATOR) { |
---|
1334 | System.out.println("nongreedy (...)* loop; exit depth is " + |
---|
1335 | blk.exitLookaheadDepth); |
---|
1336 | } |
---|
1337 | String predictExit = |
---|
1338 | getLookaheadTestExpression(blk.exitCache, |
---|
1339 | nonGreedyExitDepth); |
---|
1340 | println("// nongreedy exit test"); |
---|
1341 | println("if (" + predictExit + ") break " + label + ";"); |
---|
1342 | } |
---|
1343 | |
---|
1344 | JavaBlockFinishingInfo howToFinish = genCommonBlock(blk, false); |
---|
1345 | genBlockFinish(howToFinish, "break " + label + ";", blk.getLine()); |
---|
1346 | |
---|
1347 | tabs--; |
---|
1348 | println("} while (true);"); |
---|
1349 | println("}"); |
---|
1350 | |
---|
1351 | // Restore previous AST generation |
---|
1352 | currentASTResult = saveCurrentASTResult; |
---|
1353 | } finally { |
---|
1354 | defaultLine = oldDefaultLine; |
---|
1355 | } |
---|
1356 | } |
---|
1357 | |
---|
1358 | /** Generate an alternative. |
---|
1359 | * @param alt The alternative to generate |
---|
1360 | * @param blk The block to which the alternative belongs |
---|
1361 | */ |
---|
1362 | protected void genAlt(Alternative alt, AlternativeBlock blk) { |
---|
1363 | // Save the AST generation state, and set it to that of the alt |
---|
1364 | boolean savegenAST = genAST; |
---|
1365 | genAST = genAST && alt.getAutoGen(); |
---|
1366 | |
---|
1367 | boolean oldsaveTest = saveText; |
---|
1368 | saveText = saveText && alt.getAutoGen(); |
---|
1369 | |
---|
1370 | // Reset the variable name map for the alternative |
---|
1371 | Hashtable saveMap = treeVariableMap; |
---|
1372 | treeVariableMap = new Hashtable(); |
---|
1373 | |
---|
1374 | // Generate try block around the alt for error handling |
---|
1375 | if (alt.exceptionSpec != null) { |
---|
1376 | println("try { // for error handling", alt.head.getLine()); |
---|
1377 | tabs++; |
---|
1378 | } |
---|
1379 | |
---|
1380 | AlternativeElement elem = alt.head; |
---|
1381 | while (!(elem instanceof BlockEndElement)) { |
---|
1382 | elem.generate(); // alt can begin with anything. Ask target to gen. |
---|
1383 | elem = elem.next; |
---|
1384 | } |
---|
1385 | |
---|
1386 | if (genAST) { |
---|
1387 | if (blk instanceof RuleBlock) { |
---|
1388 | // Set the AST return value for the rule |
---|
1389 | RuleBlock rblk = (RuleBlock)blk; |
---|
1390 | if (grammar.hasSyntacticPredicate) { |
---|
1391 | // println("if ( inputState.guessing==0 ) {"); |
---|
1392 | // tabs++; |
---|
1393 | } |
---|
1394 | println(rblk.getRuleName() + "_AST = (" + labeledElementASTType + ")currentAST.root;", CONTINUE_LAST_MAPPING); |
---|
1395 | if (grammar.hasSyntacticPredicate) { |
---|
1396 | // --tabs; |
---|
1397 | // println("}"); |
---|
1398 | } |
---|
1399 | } |
---|
1400 | else if (blk.getLabel() != null) { |
---|
1401 | // ### future: also set AST value for labeled subrules. |
---|
1402 | // println(blk.getLabel() + "_AST = ("+labeledElementASTType+")currentAST.root;"); |
---|
1403 | antlrTool.warning("Labeled subrules not yet supported", grammar.getFilename(), blk.getLine(), blk.getColumn()); |
---|
1404 | } |
---|
1405 | } |
---|
1406 | |
---|
1407 | if (alt.exceptionSpec != null) { |
---|
1408 | // close try block |
---|
1409 | tabs--; |
---|
1410 | println("}", NO_MAPPING); |
---|
1411 | genErrorHandler(alt.exceptionSpec); |
---|
1412 | } |
---|
1413 | |
---|
1414 | genAST = savegenAST; |
---|
1415 | saveText = oldsaveTest; |
---|
1416 | |
---|
1417 | treeVariableMap = saveMap; |
---|
1418 | } |
---|
1419 | |
---|
1420 | /** Generate all the bitsets to be used in the parser or lexer |
---|
1421 | * Generate the raw bitset data like "long _tokenSet1_data[] = {...};" |
---|
1422 | * and the BitSet object declarations like "BitSet _tokenSet1 = new BitSet(_tokenSet1_data);" |
---|
1423 | * Note that most languages do not support object initialization inside a |
---|
1424 | * class definition, so other code-generators may have to separate the |
---|
1425 | * bitset declarations from the initializations (e.g., put the initializations |
---|
1426 | * in the generated constructor instead). |
---|
1427 | * @param bitsetList The list of bitsets to generate. |
---|
1428 | * @param maxVocabulary Ensure that each generated bitset can contain at least this value. |
---|
1429 | */ |
---|
1430 | protected void genBitsets(Vector bitsetList, |
---|
1431 | int maxVocabulary |
---|
1432 | ) { |
---|
1433 | println("", NO_MAPPING); |
---|
1434 | for (int i = 0; i < bitsetList.size(); i++) { |
---|
1435 | BitSet p = (BitSet)bitsetList.elementAt(i); |
---|
1436 | // Ensure that generated BitSet is large enough for vocabulary |
---|
1437 | p.growToInclude(maxVocabulary); |
---|
1438 | genBitSet(p, i); |
---|
1439 | } |
---|
1440 | } |
---|
1441 | |
---|
1442 | /** Do something simple like: |
---|
1443 | * private static final long[] mk_tokenSet_0() { |
---|
1444 | * long[] data = { -2305839160922996736L, 63L, 16777216L, 0L, 0L, 0L }; |
---|
1445 | * return data; |
---|
1446 | * } |
---|
1447 | * public static final BitSet _tokenSet_0 = new BitSet(mk_tokenSet_0()); |
---|
1448 | * |
---|
1449 | * Or, for large bitsets, optimize init so ranges are collapsed into loops. |
---|
1450 | * This is most useful for lexers using unicode. |
---|
1451 | */ |
---|
1452 | private void genBitSet(BitSet p, int id) { |
---|
1453 | int oldDefaultLine = defaultLine; |
---|
1454 | try { |
---|
1455 | defaultLine = NO_MAPPING; |
---|
1456 | // initialization data |
---|
1457 | println( |
---|
1458 | "private static final long[] mk" + getBitsetName(id) + "() {" |
---|
1459 | ); |
---|
1460 | int n = p.lengthInLongWords(); |
---|
1461 | if ( n<BITSET_OPTIMIZE_INIT_THRESHOLD ) { |
---|
1462 | println("\tlong[] data = { " + p.toStringOfWords() + "};"); |
---|
1463 | } |
---|
1464 | else { |
---|
1465 | // will init manually, allocate space then set values |
---|
1466 | println("\tlong[] data = new long["+n+"];"); |
---|
1467 | long[] elems = p.toPackedArray(); |
---|
1468 | for (int i = 0; i < elems.length;) { |
---|
1469 | if ( elems[i]==0 ) { |
---|
1470 | // done automatically by Java, don't waste time/code |
---|
1471 | i++; |
---|
1472 | continue; |
---|
1473 | } |
---|
1474 | if ( (i+1)==elems.length || elems[i]!=elems[i+1] ) { |
---|
1475 | // last number or no run of numbers, just dump assignment |
---|
1476 | println("\tdata["+i+"]="+elems[i]+"L;"); |
---|
1477 | i++; |
---|
1478 | } |
---|
1479 | else { |
---|
1480 | // scan to find end of run |
---|
1481 | int j; |
---|
1482 | for (j = i + 1; |
---|
1483 | j < elems.length && elems[j]==elems[i]; |
---|
1484 | j++) |
---|
1485 | { |
---|
1486 | } |
---|
1487 | // j-1 is last member of run |
---|
1488 | println("\tfor (int i = "+i+"; i<="+(j-1)+"; i++) { data[i]="+ |
---|
1489 | elems[i]+"L; }"); |
---|
1490 | i = j; |
---|
1491 | } |
---|
1492 | } |
---|
1493 | } |
---|
1494 | |
---|
1495 | println("\treturn data;"); |
---|
1496 | println("}"); |
---|
1497 | // BitSet object |
---|
1498 | println( |
---|
1499 | "public static final BitSet " + getBitsetName(id) + " = new BitSet(" + |
---|
1500 | "mk" + getBitsetName(id) + "()" + |
---|
1501 | ");" |
---|
1502 | ); |
---|
1503 | } finally { |
---|
1504 | defaultLine = oldDefaultLine; |
---|
1505 | } |
---|
1506 | } |
---|
1507 | |
---|
1508 | /** Generate the finish of a block, using a combination of the info |
---|
1509 | * returned from genCommonBlock() and the action to perform when |
---|
1510 | * no alts were taken |
---|
1511 | * @param howToFinish The return of genCommonBlock() |
---|
1512 | * @param noViableAction What to generate when no alt is taken |
---|
1513 | */ |
---|
1514 | private void genBlockFinish(JavaBlockFinishingInfo howToFinish, String noViableAction, int line) { |
---|
1515 | int oldDefaultLine = defaultLine; |
---|
1516 | try { |
---|
1517 | defaultLine = line; |
---|
1518 | if (howToFinish.needAnErrorClause && |
---|
1519 | (howToFinish.generatedAnIf || howToFinish.generatedSwitch)) { |
---|
1520 | if (howToFinish.generatedAnIf) { |
---|
1521 | println("else {"); |
---|
1522 | } |
---|
1523 | else { |
---|
1524 | println("{"); |
---|
1525 | } |
---|
1526 | tabs++; |
---|
1527 | println(noViableAction); |
---|
1528 | tabs--; |
---|
1529 | println("}"); |
---|
1530 | } |
---|
1531 | |
---|
1532 | if (howToFinish.postscript != null) { |
---|
1533 | println(howToFinish.postscript); |
---|
1534 | } |
---|
1535 | } finally { |
---|
1536 | defaultLine = oldDefaultLine; |
---|
1537 | } |
---|
1538 | } |
---|
1539 | |
---|
1540 | /** Generate the init action for a block, which may be a RuleBlock or a |
---|
1541 | * plain AlternativeBLock. |
---|
1542 | * @blk The block for which the preamble is to be generated. |
---|
1543 | */ |
---|
1544 | protected void genBlockInitAction(AlternativeBlock blk) { |
---|
1545 | // dump out init action |
---|
1546 | if (blk.initAction != null) { |
---|
1547 | printAction(processActionForSpecialSymbols(blk.initAction, blk.getLine(), currentRule, null),blk.getLine()); |
---|
1548 | } |
---|
1549 | } |
---|
1550 | |
---|
1551 | /** Generate the header for a block, which may be a RuleBlock or a |
---|
1552 | * plain AlternativeBLock. This generates any variable declarations |
---|
1553 | * and syntactic-predicate-testing variables. |
---|
1554 | * @blk The block for which the preamble is to be generated. |
---|
1555 | */ |
---|
1556 | protected void genBlockPreamble(AlternativeBlock blk) { |
---|
1557 | // define labels for rule blocks. |
---|
1558 | if (blk instanceof RuleBlock) { |
---|
1559 | RuleBlock rblk = (RuleBlock)blk; |
---|
1560 | if (rblk.labeledElements != null) { |
---|
1561 | for (int i = 0; i < rblk.labeledElements.size(); i++) { |
---|
1562 | AlternativeElement a = (AlternativeElement)rblk.labeledElements.elementAt(i); |
---|
1563 | int oldDefaultLine = defaultLine; |
---|
1564 | try { |
---|
1565 | defaultLine = a.getLine(); |
---|
1566 | // System.out.println("looking at labeled element: "+a); |
---|
1567 | // Variables for labeled rule refs and |
---|
1568 | // subrules are different than variables for |
---|
1569 | // grammar atoms. This test is a little tricky |
---|
1570 | // because we want to get all rule refs and ebnf, |
---|
1571 | // but not rule blocks or syntactic predicates |
---|
1572 | if ( |
---|
1573 | a instanceof RuleRefElement || |
---|
1574 | a instanceof AlternativeBlock && |
---|
1575 | !(a instanceof RuleBlock) && |
---|
1576 | !(a instanceof SynPredBlock) |
---|
1577 | ) { |
---|
1578 | |
---|
1579 | if ( |
---|
1580 | !(a instanceof RuleRefElement) && |
---|
1581 | ((AlternativeBlock)a).not && |
---|
1582 | analyzer.subruleCanBeInverted(((AlternativeBlock)a), grammar instanceof LexerGrammar) |
---|
1583 | ) { |
---|
1584 | // Special case for inverted subrules that |
---|
1585 | // will be inlined. Treat these like |
---|
1586 | // token or char literal references |
---|
1587 | println(labeledElementType + " " + a.getLabel() + " = " + labeledElementInit + ";"); |
---|
1588 | if (grammar.buildAST) { |
---|
1589 | genASTDeclaration(a); |
---|
1590 | } |
---|
1591 | } |
---|
1592 | else { |
---|
1593 | if (grammar.buildAST) { |
---|
1594 | // Always gen AST variables for |
---|
1595 | // labeled elements, even if the |
---|
1596 | // element itself is marked with ! |
---|
1597 | genASTDeclaration(a); |
---|
1598 | } |
---|
1599 | if (grammar instanceof LexerGrammar) { |
---|
1600 | println("Token " + a.getLabel() + "=null;"); |
---|
1601 | } |
---|
1602 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1603 | // always generate rule-ref variables |
---|
1604 | // for tree walker |
---|
1605 | println(labeledElementType + " " + a.getLabel() + " = " + labeledElementInit + ";"); |
---|
1606 | } |
---|
1607 | } |
---|
1608 | } |
---|
1609 | else { |
---|
1610 | // It is a token or literal reference. Generate the |
---|
1611 | // correct variable type for this grammar |
---|
1612 | println(labeledElementType + " " + a.getLabel() + " = " + labeledElementInit + ";"); |
---|
1613 | |
---|
1614 | // In addition, generate *_AST variables if |
---|
1615 | // building ASTs |
---|
1616 | if (grammar.buildAST) { |
---|
1617 | if (a instanceof GrammarAtom && |
---|
1618 | ((GrammarAtom)a).getASTNodeType() != null) { |
---|
1619 | GrammarAtom ga = (GrammarAtom)a; |
---|
1620 | genASTDeclaration(a, ga.getASTNodeType()); |
---|
1621 | } |
---|
1622 | else { |
---|
1623 | genASTDeclaration(a); |
---|
1624 | } |
---|
1625 | } |
---|
1626 | } |
---|
1627 | } finally { |
---|
1628 | defaultLine = oldDefaultLine; |
---|
1629 | } |
---|
1630 | } |
---|
1631 | } |
---|
1632 | } |
---|
1633 | } |
---|
1634 | |
---|
1635 | /** Generate a series of case statements that implement a BitSet test. |
---|
1636 | * @param p The Bitset for which cases are to be generated |
---|
1637 | */ |
---|
1638 | protected void genCases(BitSet p, int line) { |
---|
1639 | int oldDefaultLine = defaultLine; |
---|
1640 | try { |
---|
1641 | defaultLine = line; |
---|
1642 | if (DEBUG_CODE_GENERATOR) System.out.println("genCases(" + p + ")"); |
---|
1643 | int[] elems; |
---|
1644 | |
---|
1645 | elems = p.toArray(); |
---|
1646 | // Wrap cases four-per-line for lexer, one-per-line for parser |
---|
1647 | int wrap = (grammar instanceof LexerGrammar) ? 4 : 1; |
---|
1648 | int j = 1; |
---|
1649 | boolean startOfLine = true; |
---|
1650 | for (int i = 0; i < elems.length; i++) { |
---|
1651 | if (j == 1) { |
---|
1652 | print(""); |
---|
1653 | } |
---|
1654 | else { |
---|
1655 | _print(" "); |
---|
1656 | } |
---|
1657 | _print("case " + getValueString(elems[i]) + ":"); |
---|
1658 | |
---|
1659 | if (j == wrap) { |
---|
1660 | _println(""); |
---|
1661 | startOfLine = true; |
---|
1662 | j = 1; |
---|
1663 | } |
---|
1664 | else { |
---|
1665 | j++; |
---|
1666 | startOfLine = false; |
---|
1667 | } |
---|
1668 | } |
---|
1669 | if (!startOfLine) { |
---|
1670 | _println(""); |
---|
1671 | } |
---|
1672 | } finally { |
---|
1673 | defaultLine = oldDefaultLine; |
---|
1674 | } |
---|
1675 | } |
---|
1676 | |
---|
1677 | /**Generate common code for a block of alternatives; return a |
---|
1678 | * postscript that needs to be generated at the end of the |
---|
1679 | * block. Other routines may append else-clauses and such for |
---|
1680 | * error checking before the postfix is generated. If the |
---|
1681 | * grammar is a lexer, then generate alternatives in an order |
---|
1682 | * where alternatives requiring deeper lookahead are generated |
---|
1683 | * first, and EOF in the lookahead set reduces the depth of |
---|
1684 | * the lookahead. @param blk The block to generate @param |
---|
1685 | * noTestForSingle If true, then it does not generate a test |
---|
1686 | * for a single alternative. |
---|
1687 | */ |
---|
1688 | public JavaBlockFinishingInfo genCommonBlock(AlternativeBlock blk, |
---|
1689 | boolean noTestForSingle) { |
---|
1690 | int oldDefaultLine = defaultLine; |
---|
1691 | try { |
---|
1692 | defaultLine = blk.getLine(); |
---|
1693 | int nIF = 0; |
---|
1694 | boolean createdLL1Switch = false; |
---|
1695 | int closingBracesOfIFSequence = 0; |
---|
1696 | JavaBlockFinishingInfo finishingInfo = new JavaBlockFinishingInfo(); |
---|
1697 | if (DEBUG_CODE_GENERATOR) System.out.println("genCommonBlock(" + blk + ")"); |
---|
1698 | |
---|
1699 | // Save the AST generation state, and set it to that of the block |
---|
1700 | boolean savegenAST = genAST; |
---|
1701 | genAST = genAST && blk.getAutoGen(); |
---|
1702 | |
---|
1703 | boolean oldsaveTest = saveText; |
---|
1704 | saveText = saveText && blk.getAutoGen(); |
---|
1705 | |
---|
1706 | // Is this block inverted? If so, generate special-case code |
---|
1707 | if ( |
---|
1708 | blk.not && |
---|
1709 | analyzer.subruleCanBeInverted(blk, grammar instanceof LexerGrammar) |
---|
1710 | ) { |
---|
1711 | if (DEBUG_CODE_GENERATOR) System.out.println("special case: ~(subrule)"); |
---|
1712 | Lookahead p = analyzer.look(1, blk); |
---|
1713 | // Variable assignment for labeled elements |
---|
1714 | if (blk.getLabel() != null && syntacticPredLevel == 0) { |
---|
1715 | println(blk.getLabel() + " = " + lt1Value + ";"); |
---|
1716 | } |
---|
1717 | |
---|
1718 | // AST |
---|
1719 | genElementAST(blk); |
---|
1720 | |
---|
1721 | String astArgs = ""; |
---|
1722 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1723 | astArgs = "_t,"; |
---|
1724 | } |
---|
1725 | |
---|
1726 | // match the bitset for the alternative |
---|
1727 | println("match(" + astArgs + getBitsetName(markBitsetForGen(p.fset)) + ");"); |
---|
1728 | |
---|
1729 | // tack on tree cursor motion if doing a tree walker |
---|
1730 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1731 | println("_t = _t.getNextSibling();"); |
---|
1732 | } |
---|
1733 | return finishingInfo; |
---|
1734 | } |
---|
1735 | |
---|
1736 | // Special handling for single alt |
---|
1737 | if (blk.getAlternatives().size() == 1) { |
---|
1738 | Alternative alt = blk.getAlternativeAt(0); |
---|
1739 | // Generate a warning if there is a synPred for single alt. |
---|
1740 | if (alt.synPred != null) { |
---|
1741 | antlrTool.warning( |
---|
1742 | "Syntactic predicate superfluous for single alternative", |
---|
1743 | grammar.getFilename(), |
---|
1744 | blk.getAlternativeAt(0).synPred.getLine(), |
---|
1745 | blk.getAlternativeAt(0).synPred.getColumn() |
---|
1746 | ); |
---|
1747 | } |
---|
1748 | if (noTestForSingle) { |
---|
1749 | if (alt.semPred != null) { |
---|
1750 | // Generate validating predicate |
---|
1751 | genSemPred(alt.semPred, blk.line); |
---|
1752 | } |
---|
1753 | genAlt(alt, blk); |
---|
1754 | return finishingInfo; |
---|
1755 | } |
---|
1756 | } |
---|
1757 | |
---|
1758 | // count number of simple LL(1) cases; only do switch for |
---|
1759 | // many LL(1) cases (no preds, no end of token refs) |
---|
1760 | // We don't care about exit paths for (...)*, (...)+ |
---|
1761 | // because we don't explicitly have a test for them |
---|
1762 | // as an alt in the loop. |
---|
1763 | // |
---|
1764 | // Also, we now count how many unicode lookahead sets |
---|
1765 | // there are--they must be moved to DEFAULT or ELSE |
---|
1766 | // clause. |
---|
1767 | int nLL1 = 0; |
---|
1768 | for (int i = 0; i < blk.getAlternatives().size(); i++) { |
---|
1769 | Alternative a = blk.getAlternativeAt(i); |
---|
1770 | if (suitableForCaseExpression(a)) { |
---|
1771 | nLL1++; |
---|
1772 | } |
---|
1773 | } |
---|
1774 | |
---|
1775 | // do LL(1) cases |
---|
1776 | if (nLL1 >= makeSwitchThreshold) { |
---|
1777 | // Determine the name of the item to be compared |
---|
1778 | String testExpr = lookaheadString(1); |
---|
1779 | createdLL1Switch = true; |
---|
1780 | // when parsing trees, convert null to valid tree node with NULL lookahead |
---|
1781 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1782 | println("if (_t==null) _t=ASTNULL;"); |
---|
1783 | } |
---|
1784 | println("switch ( " + testExpr + ") {"); |
---|
1785 | for (int i = 0; i < blk.alternatives.size(); i++) { |
---|
1786 | Alternative alt = blk.getAlternativeAt(i); |
---|
1787 | // ignore any non-LL(1) alts, predicated alts, |
---|
1788 | // or end-of-token alts for case expressions |
---|
1789 | if (!suitableForCaseExpression(alt)) { |
---|
1790 | continue; |
---|
1791 | } |
---|
1792 | Lookahead p = alt.cache[1]; |
---|
1793 | if (p.fset.degree() == 0 && !p.containsEpsilon()) { |
---|
1794 | antlrTool.warning("Alternate omitted due to empty prediction set", |
---|
1795 | grammar.getFilename(), |
---|
1796 | alt.head.getLine(), alt.head.getColumn()); |
---|
1797 | } |
---|
1798 | else { |
---|
1799 | genCases(p.fset, alt.head.getLine()); |
---|
1800 | println("{", alt.head.getLine()); |
---|
1801 | tabs++; |
---|
1802 | genAlt(alt, blk); |
---|
1803 | println("break;", NO_MAPPING); |
---|
1804 | tabs--; |
---|
1805 | println("}", NO_MAPPING); |
---|
1806 | } |
---|
1807 | } |
---|
1808 | println("default:"); |
---|
1809 | tabs++; |
---|
1810 | } |
---|
1811 | |
---|
1812 | // do non-LL(1) and nondeterministic cases This is tricky in |
---|
1813 | // the lexer, because of cases like: STAR : '*' ; ASSIGN_STAR |
---|
1814 | // : "*="; Since nextToken is generated without a loop, then |
---|
1815 | // the STAR will have end-of-token as it's lookahead set for |
---|
1816 | // LA(2). So, we must generate the alternatives containing |
---|
1817 | // trailing end-of-token in their lookahead sets *after* the |
---|
1818 | // alternatives without end-of-token. This implements the |
---|
1819 | // usual lexer convention that longer matches come before |
---|
1820 | // shorter ones, e.g. "*=" matches ASSIGN_STAR not STAR |
---|
1821 | // |
---|
1822 | // For non-lexer grammars, this does not sort the alternates |
---|
1823 | // by depth Note that alts whose lookahead is purely |
---|
1824 | // end-of-token at k=1 end up as default or else clauses. |
---|
1825 | int startDepth = (grammar instanceof LexerGrammar) ? grammar.maxk : 0; |
---|
1826 | for (int altDepth = startDepth; altDepth >= 0; altDepth--) { |
---|
1827 | if (DEBUG_CODE_GENERATOR) System.out.println("checking depth " + altDepth); |
---|
1828 | for (int i = 0; i < blk.alternatives.size(); i++) { |
---|
1829 | Alternative alt = blk.getAlternativeAt(i); |
---|
1830 | if (DEBUG_CODE_GENERATOR) System.out.println("genAlt: " + i); |
---|
1831 | // if we made a switch above, ignore what we already took care |
---|
1832 | // of. Specifically, LL(1) alts with no preds |
---|
1833 | // that do not have end-of-token in their prediction set |
---|
1834 | // and that are not giant unicode sets. |
---|
1835 | if (createdLL1Switch && suitableForCaseExpression(alt)) { |
---|
1836 | if (DEBUG_CODE_GENERATOR) System.out.println("ignoring alt because it was in the switch"); |
---|
1837 | continue; |
---|
1838 | } |
---|
1839 | String e; |
---|
1840 | |
---|
1841 | boolean unpredicted = false; |
---|
1842 | |
---|
1843 | if (grammar instanceof LexerGrammar) { |
---|
1844 | // Calculate the "effective depth" of the alt, |
---|
1845 | // which is the max depth at which |
---|
1846 | // cache[depth]!=end-of-token |
---|
1847 | int effectiveDepth = alt.lookaheadDepth; |
---|
1848 | if (effectiveDepth == GrammarAnalyzer.NONDETERMINISTIC) { |
---|
1849 | // use maximum lookahead |
---|
1850 | effectiveDepth = grammar.maxk; |
---|
1851 | } |
---|
1852 | while (effectiveDepth >= 1 && |
---|
1853 | alt.cache[effectiveDepth].containsEpsilon()) { |
---|
1854 | effectiveDepth--; |
---|
1855 | } |
---|
1856 | // Ignore alts whose effective depth is other than |
---|
1857 | // the ones we are generating for this iteration. |
---|
1858 | if (effectiveDepth != altDepth) { |
---|
1859 | if (DEBUG_CODE_GENERATOR) |
---|
1860 | System.out.println("ignoring alt because effectiveDepth!=altDepth;" + effectiveDepth + "!=" + altDepth); |
---|
1861 | continue; |
---|
1862 | } |
---|
1863 | unpredicted = lookaheadIsEmpty(alt, effectiveDepth); |
---|
1864 | e = getLookaheadTestExpression(alt, effectiveDepth); |
---|
1865 | } |
---|
1866 | else { |
---|
1867 | unpredicted = lookaheadIsEmpty(alt, grammar.maxk); |
---|
1868 | e = getLookaheadTestExpression(alt, grammar.maxk); |
---|
1869 | } |
---|
1870 | |
---|
1871 | int oldDefaultLine2 = defaultLine; |
---|
1872 | try { |
---|
1873 | defaultLine = alt.head.getLine(); |
---|
1874 | // Was it a big unicode range that forced unsuitability |
---|
1875 | // for a case expression? |
---|
1876 | if (alt.cache[1].fset.degree() > caseSizeThreshold && |
---|
1877 | suitableForCaseExpression(alt)) { |
---|
1878 | if (nIF == 0) { |
---|
1879 | println("if " + e + " {"); |
---|
1880 | } |
---|
1881 | else { |
---|
1882 | println("else if " + e + " {"); |
---|
1883 | } |
---|
1884 | } |
---|
1885 | else if (unpredicted && |
---|
1886 | alt.semPred == null && |
---|
1887 | alt.synPred == null) { |
---|
1888 | // The alt has empty prediction set and no |
---|
1889 | // predicate to help out. if we have not |
---|
1890 | // generated a previous if, just put {...} around |
---|
1891 | // the end-of-token clause |
---|
1892 | if (nIF == 0) { |
---|
1893 | println("{"); |
---|
1894 | } |
---|
1895 | else { |
---|
1896 | println("else {"); |
---|
1897 | } |
---|
1898 | finishingInfo.needAnErrorClause = false; |
---|
1899 | } |
---|
1900 | else { // check for sem and syn preds |
---|
1901 | |
---|
1902 | // Add any semantic predicate expression to the |
---|
1903 | // lookahead test |
---|
1904 | if (alt.semPred != null) { |
---|
1905 | // if debugging, wrap the evaluation of the |
---|
1906 | // predicate in a method translate $ and # |
---|
1907 | // references |
---|
1908 | ActionTransInfo tInfo = new ActionTransInfo(); |
---|
1909 | String actionStr = |
---|
1910 | processActionForSpecialSymbols(alt.semPred, |
---|
1911 | blk.line, |
---|
1912 | currentRule, |
---|
1913 | tInfo); |
---|
1914 | // ignore translation info...we don't need to |
---|
1915 | // do anything with it. call that will inform |
---|
1916 | // SemanticPredicateListeners of the result |
---|
1917 | if (((grammar instanceof ParserGrammar) || |
---|
1918 | (grammar instanceof LexerGrammar)) && |
---|
1919 | grammar.debuggingOutput) { |
---|
1920 | e = "(" + e + "&& fireSemanticPredicateEvaluated(antlr.debug.SemanticPredicateEvent.PREDICTING," + |
---|
1921 | addSemPred(charFormatter.escapeString(actionStr)) + "," + actionStr + "))"; |
---|
1922 | } |
---|
1923 | else { |
---|
1924 | e = "(" + e + "&&(" + actionStr + "))"; |
---|
1925 | } |
---|
1926 | } |
---|
1927 | |
---|
1928 | // Generate any syntactic predicates |
---|
1929 | if (nIF > 0) { |
---|
1930 | if (alt.synPred != null) { |
---|
1931 | println("else {", alt.synPred.getLine()); |
---|
1932 | tabs++; |
---|
1933 | genSynPred(alt.synPred, e); |
---|
1934 | closingBracesOfIFSequence++; |
---|
1935 | } |
---|
1936 | else { |
---|
1937 | println("else if " + e + " {"); |
---|
1938 | } |
---|
1939 | } |
---|
1940 | else { |
---|
1941 | if (alt.synPred != null) { |
---|
1942 | genSynPred(alt.synPred, e); |
---|
1943 | } |
---|
1944 | else { |
---|
1945 | // when parsing trees, convert null to |
---|
1946 | // valid tree node with NULL lookahead. |
---|
1947 | if (grammar instanceof TreeWalkerGrammar) { |
---|
1948 | println("if (_t==null) _t=ASTNULL;"); |
---|
1949 | } |
---|
1950 | println("if " + e + " {"); |
---|
1951 | } |
---|
1952 | } |
---|
1953 | |
---|
1954 | } |
---|
1955 | } finally { |
---|
1956 | defaultLine = oldDefaultLine2; |
---|
1957 | } |
---|
1958 | |
---|
1959 | nIF++; |
---|
1960 | tabs++; |
---|
1961 | genAlt(alt, blk); |
---|
1962 | tabs--; |
---|
1963 | println("}"); |
---|
1964 | } |
---|
1965 | } |
---|
1966 | String ps = ""; |
---|
1967 | for (int i = 1; i <= closingBracesOfIFSequence; i++) { |
---|
1968 | ps += "}"; |
---|
1969 | } |
---|
1970 | |
---|
1971 | // Restore the AST generation state |
---|
1972 | genAST = savegenAST; |
---|
1973 | |
---|
1974 | // restore save text state |
---|
1975 | saveText = oldsaveTest; |
---|
1976 | |
---|
1977 | // Return the finishing info. |
---|
1978 | if (createdLL1Switch) { |
---|
1979 | tabs--; |
---|
1980 | finishingInfo.postscript = ps + "}"; |
---|
1981 | finishingInfo.generatedSwitch = true; |
---|
1982 | finishingInfo.generatedAnIf = nIF > 0; |
---|
1983 | //return new JavaBlockFinishingInfo(ps+"}",true,nIF>0); // close up switch statement |
---|
1984 | |
---|
1985 | } |
---|
1986 | else { |
---|
1987 | finishingInfo.postscript = ps; |
---|
1988 | finishingInfo.generatedSwitch = false; |
---|
1989 | finishingInfo.generatedAnIf = nIF > 0; |
---|
1990 | // return new JavaBlockFinishingInfo(ps, false,nIF>0); |
---|
1991 | } |
---|
1992 | return finishingInfo; |
---|
1993 | } finally { |
---|
1994 | defaultLine = oldDefaultLine; |
---|
1995 | } |
---|
1996 | } |
---|
1997 | |
---|
1998 | private static boolean suitableForCaseExpression(Alternative a) { |
---|
1999 | return |
---|
2000 | a.lookaheadDepth == 1 && |
---|
2001 | a.semPred == null && |
---|
2002 | !a.cache[1].containsEpsilon() && |
---|
2003 | a.cache[1].fset.degree() <= caseSizeThreshold; |
---|
2004 | } |
---|
2005 | |
---|
2006 | /** Generate code to link an element reference into the AST */ |
---|
2007 | private void genElementAST(AlternativeElement el) { |
---|
2008 | int oldDefaultLine = defaultLine; |
---|
2009 | try { |
---|
2010 | defaultLine = el.getLine(); |
---|
2011 | // handle case where you're not building trees, but are in tree walker. |
---|
2012 | // Just need to get labels set up. |
---|
2013 | if (grammar instanceof TreeWalkerGrammar && !grammar.buildAST) { |
---|
2014 | String elementRef; |
---|
2015 | String astName; |
---|
2016 | |
---|
2017 | // Generate names and declarations of the AST variable(s) |
---|
2018 | if (el.getLabel() == null) { |
---|
2019 | elementRef = lt1Value; |
---|
2020 | // Generate AST variables for unlabeled stuff |
---|
2021 | astName = "tmp" + astVarNumber + "_AST"; |
---|
2022 | astVarNumber++; |
---|
2023 | // Map the generated AST variable in the alternate |
---|
2024 | mapTreeVariable(el, astName); |
---|
2025 | // Generate an "input" AST variable also |
---|
2026 | println(labeledElementASTType + " " + astName + "_in = " + elementRef + ";"); |
---|
2027 | } |
---|
2028 | return; |
---|
2029 | } |
---|
2030 | |
---|
2031 | if (grammar.buildAST && syntacticPredLevel == 0) { |
---|
2032 | boolean needASTDecl = |
---|
2033 | (genAST && |
---|
2034 | (el.getLabel() != null || |
---|
2035 | el.getAutoGenType() != GrammarElement.AUTO_GEN_BANG |
---|
2036 | ) |
---|
2037 | ); |
---|
2038 | |
---|
2039 | // RK: if we have a grammar element always generate the decl |
---|
2040 | // since some guy can access it from an action and we can't |
---|
2041 | // peek ahead (well not without making a mess). |
---|
2042 | // I'd prefer taking this out. |
---|
2043 | if (el.getAutoGenType() != GrammarElement.AUTO_GEN_BANG && |
---|
2044 | (el instanceof TokenRefElement)) |
---|
2045 | { |
---|
2046 | needASTDecl = true; |
---|
2047 | } |
---|
2048 | |
---|
2049 | boolean doNoGuessTest = |
---|
2050 | (grammar.hasSyntacticPredicate && needASTDecl); |
---|
2051 | |
---|
2052 | String elementRef; |
---|
2053 | String astNameBase; |
---|
2054 | |
---|
2055 | // Generate names and declarations of the AST variable(s) |
---|
2056 | if (el.getLabel() != null) { |
---|
2057 | elementRef = el.getLabel(); |
---|
2058 | astNameBase = el.getLabel(); |
---|
2059 | } |
---|
2060 | else { |
---|
2061 | elementRef = lt1Value; |
---|
2062 | // Generate AST variables for unlabeled stuff |
---|
2063 | astNameBase = "tmp" + astVarNumber; |
---|
2064 | ; |
---|
2065 | astVarNumber++; |
---|
2066 | } |
---|
2067 | |
---|
2068 | // Generate the declaration if required. |
---|
2069 | if (needASTDecl) { |
---|
2070 | // Generate the declaration |
---|
2071 | if (el instanceof GrammarAtom) { |
---|
2072 | GrammarAtom ga = (GrammarAtom)el; |
---|
2073 | if (ga.getASTNodeType() != null) { |
---|
2074 | genASTDeclaration(el, astNameBase, ga.getASTNodeType()); |
---|
2075 | // println(ga.getASTNodeType()+" " + astName+" = null;"); |
---|
2076 | } |
---|
2077 | else { |
---|
2078 | genASTDeclaration(el, astNameBase, labeledElementASTType); |
---|
2079 | // println(labeledElementASTType+" " + astName + " = null;"); |
---|
2080 | } |
---|
2081 | } |
---|
2082 | else { |
---|
2083 | genASTDeclaration(el, astNameBase, labeledElementASTType); |
---|
2084 | // println(labeledElementASTType+" " + astName + " = null;"); |
---|
2085 | } |
---|
2086 | } |
---|
2087 | |
---|
2088 | // for convenience.. |
---|
2089 | String astName = astNameBase + "_AST"; |
---|
2090 | |
---|
2091 | // Map the generated AST variable in the alternate |
---|
2092 | mapTreeVariable(el, astName); |
---|
2093 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2094 | // Generate an "input" AST variable also |
---|
2095 | println(labeledElementASTType + " " + astName + "_in = null;"); |
---|
2096 | } |
---|
2097 | |
---|
2098 | // Enclose actions with !guessing |
---|
2099 | if (doNoGuessTest) { |
---|
2100 | // println("if (inputState.guessing==0) {"); |
---|
2101 | // tabs++; |
---|
2102 | } |
---|
2103 | |
---|
2104 | // if something has a label assume it will be used |
---|
2105 | // so we must initialize the RefAST |
---|
2106 | if (el.getLabel() != null) { |
---|
2107 | if (el instanceof GrammarAtom) { |
---|
2108 | println(astName + " = " + getASTCreateString((GrammarAtom)el, elementRef) + ";"); |
---|
2109 | } |
---|
2110 | else { |
---|
2111 | println(astName + " = " + getASTCreateString(elementRef) + ";"); |
---|
2112 | } |
---|
2113 | } |
---|
2114 | |
---|
2115 | // if it has no label but a declaration exists initialize it. |
---|
2116 | if (el.getLabel() == null && needASTDecl) { |
---|
2117 | elementRef = lt1Value; |
---|
2118 | if (el instanceof GrammarAtom) { |
---|
2119 | println(astName + " = " + getASTCreateString((GrammarAtom)el, elementRef) + ";"); |
---|
2120 | } |
---|
2121 | else { |
---|
2122 | println(astName + " = " + getASTCreateString(elementRef) + ";"); |
---|
2123 | } |
---|
2124 | // Map the generated AST variable in the alternate |
---|
2125 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2126 | // set "input" AST variable also |
---|
2127 | println(astName + "_in = " + elementRef + ";"); |
---|
2128 | } |
---|
2129 | } |
---|
2130 | |
---|
2131 | if (genAST) { |
---|
2132 | switch (el.getAutoGenType()) { |
---|
2133 | case GrammarElement.AUTO_GEN_NONE: |
---|
2134 | println("astFactory.addASTChild(currentAST, " + astName + ");"); |
---|
2135 | break; |
---|
2136 | case GrammarElement.AUTO_GEN_CARET: |
---|
2137 | println("astFactory.makeASTRoot(currentAST, " + astName + ");"); |
---|
2138 | break; |
---|
2139 | default: |
---|
2140 | break; |
---|
2141 | } |
---|
2142 | } |
---|
2143 | if (doNoGuessTest) { |
---|
2144 | // tabs--; |
---|
2145 | // println("}"); |
---|
2146 | } |
---|
2147 | } |
---|
2148 | } finally { |
---|
2149 | defaultLine = oldDefaultLine; |
---|
2150 | } |
---|
2151 | } |
---|
2152 | |
---|
2153 | /** Close the try block and generate catch phrases |
---|
2154 | * if the element has a labeled handler in the rule |
---|
2155 | */ |
---|
2156 | private void genErrorCatchForElement(AlternativeElement el) { |
---|
2157 | if (el.getLabel() == null) return; |
---|
2158 | String r = el.enclosingRuleName; |
---|
2159 | if (grammar instanceof LexerGrammar) { |
---|
2160 | r = CodeGenerator.encodeLexerRuleName(el.enclosingRuleName); |
---|
2161 | } |
---|
2162 | RuleSymbol rs = (RuleSymbol)grammar.getSymbol(r); |
---|
2163 | if (rs == null) { |
---|
2164 | antlrTool.panic("Enclosing rule not found!"); |
---|
2165 | } |
---|
2166 | ExceptionSpec ex = rs.block.findExceptionSpec(el.getLabel()); |
---|
2167 | if (ex != null) { |
---|
2168 | tabs--; |
---|
2169 | println("}", el.getLine()); |
---|
2170 | genErrorHandler(ex); |
---|
2171 | } |
---|
2172 | } |
---|
2173 | |
---|
2174 | /** Generate the catch phrases for a user-specified error handler */ |
---|
2175 | private void genErrorHandler(ExceptionSpec ex) { |
---|
2176 | // Each ExceptionHandler in the ExceptionSpec is a separate catch |
---|
2177 | for (int i = 0; i < ex.handlers.size(); i++) { |
---|
2178 | ExceptionHandler handler = (ExceptionHandler)ex.handlers.elementAt(i); |
---|
2179 | int oldDefaultLine = defaultLine; |
---|
2180 | try { |
---|
2181 | defaultLine = handler.action.getLine(); |
---|
2182 | // Generate catch phrase |
---|
2183 | println("catch (" + handler.exceptionTypeAndName.getText() + ") {", handler.exceptionTypeAndName.getLine()); |
---|
2184 | tabs++; |
---|
2185 | if (grammar.hasSyntacticPredicate) { |
---|
2186 | println("if (inputState.guessing==0) {"); |
---|
2187 | tabs++; |
---|
2188 | } |
---|
2189 | |
---|
2190 | // When not guessing, execute user handler action |
---|
2191 | ActionTransInfo tInfo = new ActionTransInfo(); |
---|
2192 | printAction( |
---|
2193 | processActionForSpecialSymbols(handler.action.getText(), |
---|
2194 | handler.action.getLine(), |
---|
2195 | currentRule, tInfo) |
---|
2196 | ); |
---|
2197 | |
---|
2198 | if (grammar.hasSyntacticPredicate) { |
---|
2199 | tabs--; |
---|
2200 | println("} else {"); |
---|
2201 | tabs++; |
---|
2202 | // When guessing, rethrow exception |
---|
2203 | println( |
---|
2204 | "throw " + |
---|
2205 | extractIdOfAction(handler.exceptionTypeAndName) + |
---|
2206 | ";" |
---|
2207 | ); |
---|
2208 | tabs--; |
---|
2209 | println("}"); |
---|
2210 | } |
---|
2211 | // Close catch phrase |
---|
2212 | tabs--; |
---|
2213 | println("}"); |
---|
2214 | } finally { |
---|
2215 | defaultLine = oldDefaultLine; |
---|
2216 | } |
---|
2217 | } |
---|
2218 | } |
---|
2219 | |
---|
2220 | /** Generate a try { opening if the element has a labeled handler in the rule */ |
---|
2221 | private void genErrorTryForElement(AlternativeElement el) { |
---|
2222 | if (el.getLabel() == null) return; |
---|
2223 | String r = el.enclosingRuleName; |
---|
2224 | if (grammar instanceof LexerGrammar) { |
---|
2225 | r = CodeGenerator.encodeLexerRuleName(el.enclosingRuleName); |
---|
2226 | } |
---|
2227 | RuleSymbol rs = (RuleSymbol)grammar.getSymbol(r); |
---|
2228 | if (rs == null) { |
---|
2229 | antlrTool.panic("Enclosing rule not found!"); |
---|
2230 | } |
---|
2231 | ExceptionSpec ex = rs.block.findExceptionSpec(el.getLabel()); |
---|
2232 | if (ex != null) { |
---|
2233 | println("try { // for error handling", el.getLine()); |
---|
2234 | tabs++; |
---|
2235 | } |
---|
2236 | } |
---|
2237 | |
---|
2238 | protected void genASTDeclaration(AlternativeElement el) { |
---|
2239 | genASTDeclaration(el, labeledElementASTType); |
---|
2240 | } |
---|
2241 | |
---|
2242 | protected void genASTDeclaration(AlternativeElement el, String node_type) { |
---|
2243 | genASTDeclaration(el, el.getLabel(), node_type); |
---|
2244 | } |
---|
2245 | |
---|
2246 | protected void genASTDeclaration(AlternativeElement el, String var_name, String node_type) { |
---|
2247 | // already declared? |
---|
2248 | if (declaredASTVariables.contains(el)) |
---|
2249 | return; |
---|
2250 | |
---|
2251 | // emit code |
---|
2252 | println(node_type + " " + var_name + "_AST = null;"); |
---|
2253 | |
---|
2254 | // mark as declared |
---|
2255 | declaredASTVariables.put(el,el); |
---|
2256 | } |
---|
2257 | |
---|
2258 | /** Generate a header that is common to all Java files */ |
---|
2259 | protected void genHeader() { |
---|
2260 | println("// $ANTLR " + Tool.version + ": " + |
---|
2261 | "\"" + antlrTool.fileMinusPath(antlrTool.grammarFile) + "\"" + |
---|
2262 | " -> " + |
---|
2263 | "\"" + grammar.getClassName() + ".java\"$", NO_MAPPING); |
---|
2264 | } |
---|
2265 | |
---|
2266 | private void genLiteralsTest() { |
---|
2267 | println("_ttype = testLiteralsTable(_ttype);"); |
---|
2268 | } |
---|
2269 | |
---|
2270 | private void genLiteralsTestForPartialToken() { |
---|
2271 | println("_ttype = testLiteralsTable(new String(text.getBuffer(),_begin,text.length()-_begin),_ttype);"); |
---|
2272 | } |
---|
2273 | |
---|
2274 | protected void genMatch(BitSet b) { |
---|
2275 | } |
---|
2276 | |
---|
2277 | protected void genMatch(GrammarAtom atom) { |
---|
2278 | if (atom instanceof StringLiteralElement) { |
---|
2279 | if (grammar instanceof LexerGrammar) { |
---|
2280 | genMatchUsingAtomText(atom); |
---|
2281 | } |
---|
2282 | else { |
---|
2283 | genMatchUsingAtomTokenType(atom); |
---|
2284 | } |
---|
2285 | } |
---|
2286 | else if (atom instanceof CharLiteralElement) { |
---|
2287 | if (grammar instanceof LexerGrammar) { |
---|
2288 | genMatchUsingAtomText(atom); |
---|
2289 | } |
---|
2290 | else { |
---|
2291 | antlrTool.error("cannot ref character literals in grammar: " + atom); |
---|
2292 | } |
---|
2293 | } |
---|
2294 | else if (atom instanceof TokenRefElement) { |
---|
2295 | genMatchUsingAtomText(atom); |
---|
2296 | } |
---|
2297 | else if (atom instanceof WildcardElement) { |
---|
2298 | gen((WildcardElement)atom); |
---|
2299 | } |
---|
2300 | } |
---|
2301 | |
---|
2302 | protected void genMatchUsingAtomText(GrammarAtom atom) { |
---|
2303 | int oldDefaultLine = defaultLine; |
---|
2304 | try { |
---|
2305 | defaultLine = atom.getLine(); |
---|
2306 | // match() for trees needs the _t cursor |
---|
2307 | String astArgs = ""; |
---|
2308 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2309 | astArgs = "_t,"; |
---|
2310 | } |
---|
2311 | |
---|
2312 | // if in lexer and ! on element, save buffer index to kill later |
---|
2313 | if (grammar instanceof LexerGrammar && (!saveText || atom.getAutoGenType() == GrammarElement.AUTO_GEN_BANG)) { |
---|
2314 | println("_saveIndex=text.length();"); |
---|
2315 | } |
---|
2316 | |
---|
2317 | print(atom.not ? "matchNot(" : "match("); |
---|
2318 | _print(astArgs, NO_MAPPING); |
---|
2319 | |
---|
2320 | // print out what to match |
---|
2321 | if (atom.atomText.equals("EOF")) { |
---|
2322 | // horrible hack to handle EOF case |
---|
2323 | _print("Token.EOF_TYPE"); |
---|
2324 | } |
---|
2325 | else { |
---|
2326 | _print(atom.atomText); |
---|
2327 | } |
---|
2328 | _println(");"); |
---|
2329 | |
---|
2330 | if (grammar instanceof LexerGrammar && (!saveText || atom.getAutoGenType() == GrammarElement.AUTO_GEN_BANG)) { |
---|
2331 | println("text.setLength(_saveIndex);"); // kill text atom put in buffer |
---|
2332 | } |
---|
2333 | } finally { |
---|
2334 | defaultLine = oldDefaultLine; |
---|
2335 | } |
---|
2336 | } |
---|
2337 | |
---|
2338 | protected void genMatchUsingAtomTokenType(GrammarAtom atom) { |
---|
2339 | // match() for trees needs the _t cursor |
---|
2340 | String astArgs = ""; |
---|
2341 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2342 | astArgs = "_t,"; |
---|
2343 | } |
---|
2344 | |
---|
2345 | // If the literal can be mangled, generate the symbolic constant instead |
---|
2346 | String mangledName = null; |
---|
2347 | String s = astArgs + getValueString(atom.getType()); |
---|
2348 | |
---|
2349 | // matching |
---|
2350 | println((atom.not ? "matchNot(" : "match(") + s + ");", atom.getLine()); |
---|
2351 | } |
---|
2352 | |
---|
2353 | /** Generate the nextToken() rule. nextToken() is a synthetic |
---|
2354 | * lexer rule that is the implicit OR of all user-defined |
---|
2355 | * lexer rules. |
---|
2356 | */ |
---|
2357 | public void genNextToken() { |
---|
2358 | int oldDefaultLine = defaultLine; |
---|
2359 | try { |
---|
2360 | defaultLine = NO_MAPPING; |
---|
2361 | // Are there any public rules? If not, then just generate a |
---|
2362 | // fake nextToken(). |
---|
2363 | boolean hasPublicRules = false; |
---|
2364 | for (int i = 0; i < grammar.rules.size(); i++) { |
---|
2365 | RuleSymbol rs = (RuleSymbol)grammar.rules.elementAt(i); |
---|
2366 | if (rs.isDefined() && rs.access.equals("public")) { |
---|
2367 | hasPublicRules = true; |
---|
2368 | break; |
---|
2369 | } |
---|
2370 | } |
---|
2371 | if (!hasPublicRules) { |
---|
2372 | println(""); |
---|
2373 | println("public Token nextToken() throws TokenStreamException {"); |
---|
2374 | println("\ttry {uponEOF();}"); |
---|
2375 | println("\tcatch(CharStreamIOException csioe) {"); |
---|
2376 | println("\t\tthrow new TokenStreamIOException(csioe.io);"); |
---|
2377 | println("\t}"); |
---|
2378 | println("\tcatch(CharStreamException cse) {"); |
---|
2379 | println("\t\tthrow new TokenStreamException(cse.getMessage());"); |
---|
2380 | println("\t}"); |
---|
2381 | println("\treturn new CommonToken(Token.EOF_TYPE, \"\");"); |
---|
2382 | println("}"); |
---|
2383 | println(""); |
---|
2384 | return; |
---|
2385 | } |
---|
2386 | |
---|
2387 | // Create the synthesized nextToken() rule |
---|
2388 | RuleBlock nextTokenBlk = MakeGrammar.createNextTokenRule(grammar, grammar.rules, "nextToken"); |
---|
2389 | // Define the nextToken rule symbol |
---|
2390 | RuleSymbol nextTokenRs = new RuleSymbol("mnextToken"); |
---|
2391 | nextTokenRs.setDefined(); |
---|
2392 | nextTokenRs.setBlock(nextTokenBlk); |
---|
2393 | nextTokenRs.access = "private"; |
---|
2394 | grammar.define(nextTokenRs); |
---|
2395 | // Analyze the nextToken rule |
---|
2396 | boolean ok = grammar.theLLkAnalyzer.deterministic(nextTokenBlk); |
---|
2397 | |
---|
2398 | // Generate the next token rule |
---|
2399 | String filterRule = null; |
---|
2400 | if (((LexerGrammar)grammar).filterMode) { |
---|
2401 | filterRule = ((LexerGrammar)grammar).filterRule; |
---|
2402 | } |
---|
2403 | |
---|
2404 | println(""); |
---|
2405 | println("public Token nextToken() throws TokenStreamException {"); |
---|
2406 | tabs++; |
---|
2407 | println("Token theRetToken=null;"); |
---|
2408 | _println("tryAgain:"); |
---|
2409 | println("for (;;) {"); |
---|
2410 | tabs++; |
---|
2411 | println("Token _token = null;"); |
---|
2412 | println("int _ttype = Token.INVALID_TYPE;"); |
---|
2413 | if (((LexerGrammar)grammar).filterMode) { |
---|
2414 | println("setCommitToPath(false);"); |
---|
2415 | if (filterRule != null) { |
---|
2416 | // Here's a good place to ensure that the filter rule actually exists |
---|
2417 | if (!grammar.isDefined(CodeGenerator.encodeLexerRuleName(filterRule))) { |
---|
2418 | grammar.antlrTool.error("Filter rule " + filterRule + " does not exist in this lexer"); |
---|
2419 | } |
---|
2420 | else { |
---|
2421 | RuleSymbol rs = (RuleSymbol)grammar.getSymbol(CodeGenerator.encodeLexerRuleName(filterRule)); |
---|
2422 | if (!rs.isDefined()) { |
---|
2423 | grammar.antlrTool.error("Filter rule " + filterRule + " does not exist in this lexer"); |
---|
2424 | } |
---|
2425 | else if (rs.access.equals("public")) { |
---|
2426 | grammar.antlrTool.error("Filter rule " + filterRule + " must be protected"); |
---|
2427 | } |
---|
2428 | } |
---|
2429 | println("int _m;"); |
---|
2430 | println("_m = mark();"); |
---|
2431 | } |
---|
2432 | } |
---|
2433 | println("resetText();"); |
---|
2434 | |
---|
2435 | println("try { // for char stream error handling"); |
---|
2436 | tabs++; |
---|
2437 | |
---|
2438 | // Generate try around whole thing to trap scanner errors |
---|
2439 | println("try { // for lexical error handling"); |
---|
2440 | tabs++; |
---|
2441 | |
---|
2442 | // Test for public lexical rules with empty paths |
---|
2443 | for (int i = 0; i < nextTokenBlk.getAlternatives().size(); i++) { |
---|
2444 | Alternative a = nextTokenBlk.getAlternativeAt(i); |
---|
2445 | if (a.cache[1].containsEpsilon()) { |
---|
2446 | //String r = a.head.toString(); |
---|
2447 | RuleRefElement rr = (RuleRefElement)a.head; |
---|
2448 | String r = CodeGenerator.decodeLexerRuleName(rr.targetRule); |
---|
2449 | antlrTool.warning("public lexical rule "+r+" is optional (can match \"nothing\")"); |
---|
2450 | } |
---|
2451 | } |
---|
2452 | |
---|
2453 | // Generate the block |
---|
2454 | String newline = System.getProperty("line.separator"); |
---|
2455 | JavaBlockFinishingInfo howToFinish = genCommonBlock(nextTokenBlk, false); |
---|
2456 | String errFinish = "if (LA(1)==EOF_CHAR) {uponEOF(); _returnToken = makeToken(Token.EOF_TYPE);}"; |
---|
2457 | errFinish += newline + "\t\t\t\t"; |
---|
2458 | if (((LexerGrammar)grammar).filterMode) { |
---|
2459 | if (filterRule == null) { |
---|
2460 | errFinish += "else {consume(); continue tryAgain;}"; |
---|
2461 | } |
---|
2462 | else { |
---|
2463 | errFinish += "else {" + newline + |
---|
2464 | "\t\t\t\t\tcommit();" + newline + |
---|
2465 | "\t\t\t\t\ttry {m" + filterRule + "(false);}" + newline + |
---|
2466 | "\t\t\t\t\tcatch(RecognitionException e) {" + newline + |
---|
2467 | "\t\t\t\t\t // catastrophic failure" + newline + |
---|
2468 | "\t\t\t\t\t reportError(e);" + newline + |
---|
2469 | "\t\t\t\t\t consume();" + newline + |
---|
2470 | "\t\t\t\t\t}" + newline + |
---|
2471 | "\t\t\t\t\tcontinue tryAgain;" + newline + |
---|
2472 | "\t\t\t\t}"; |
---|
2473 | } |
---|
2474 | } |
---|
2475 | else { |
---|
2476 | errFinish += "else {" + throwNoViable + "}"; |
---|
2477 | } |
---|
2478 | genBlockFinish(howToFinish, errFinish, nextTokenBlk.getLine()); |
---|
2479 | |
---|
2480 | // at this point a valid token has been matched, undo "mark" that was done |
---|
2481 | if (((LexerGrammar)grammar).filterMode && filterRule != null) { |
---|
2482 | println("commit();"); |
---|
2483 | } |
---|
2484 | |
---|
2485 | // Generate literals test if desired |
---|
2486 | // make sure _ttype is set first; note _returnToken must be |
---|
2487 | // non-null as the rule was required to create it. |
---|
2488 | println("if ( _returnToken==null ) continue tryAgain; // found SKIP token"); |
---|
2489 | println("_ttype = _returnToken.getType();"); |
---|
2490 | if (((LexerGrammar)grammar).getTestLiterals()) { |
---|
2491 | genLiteralsTest(); |
---|
2492 | } |
---|
2493 | |
---|
2494 | // return token created by rule reference in switch |
---|
2495 | println("_returnToken.setType(_ttype);"); |
---|
2496 | println("return _returnToken;"); |
---|
2497 | |
---|
2498 | // Close try block |
---|
2499 | tabs--; |
---|
2500 | println("}"); |
---|
2501 | println("catch (RecognitionException e) {"); |
---|
2502 | tabs++; |
---|
2503 | if (((LexerGrammar)grammar).filterMode) { |
---|
2504 | if (filterRule == null) { |
---|
2505 | println("if ( !getCommitToPath() ) {consume(); continue tryAgain;}"); |
---|
2506 | } |
---|
2507 | else { |
---|
2508 | println("if ( !getCommitToPath() ) {"); |
---|
2509 | tabs++; |
---|
2510 | println("rewind(_m);"); |
---|
2511 | println("resetText();"); |
---|
2512 | println("try {m" + filterRule + "(false);}"); |
---|
2513 | println("catch(RecognitionException ee) {"); |
---|
2514 | println(" // horrendous failure: error in filter rule"); |
---|
2515 | println(" reportError(ee);"); |
---|
2516 | println(" consume();"); |
---|
2517 | println("}"); |
---|
2518 | println("continue tryAgain;"); |
---|
2519 | tabs--; |
---|
2520 | println("}"); |
---|
2521 | } |
---|
2522 | } |
---|
2523 | if (nextTokenBlk.getDefaultErrorHandler()) { |
---|
2524 | println("reportError(e);"); |
---|
2525 | println("consume();"); |
---|
2526 | } |
---|
2527 | else { |
---|
2528 | // pass on to invoking routine |
---|
2529 | println("throw new TokenStreamRecognitionException(e);"); |
---|
2530 | } |
---|
2531 | tabs--; |
---|
2532 | println("}"); |
---|
2533 | |
---|
2534 | // close CharStreamException try |
---|
2535 | tabs--; |
---|
2536 | println("}"); |
---|
2537 | println("catch (CharStreamException cse) {"); |
---|
2538 | println(" if ( cse instanceof CharStreamIOException ) {"); |
---|
2539 | println(" throw new TokenStreamIOException(((CharStreamIOException)cse).io);"); |
---|
2540 | println(" }"); |
---|
2541 | println(" else {"); |
---|
2542 | println(" throw new TokenStreamException(cse.getMessage());"); |
---|
2543 | println(" }"); |
---|
2544 | println("}"); |
---|
2545 | |
---|
2546 | // close for-loop |
---|
2547 | tabs--; |
---|
2548 | println("}"); |
---|
2549 | |
---|
2550 | // close method nextToken |
---|
2551 | tabs--; |
---|
2552 | println("}"); |
---|
2553 | println(""); |
---|
2554 | } finally { |
---|
2555 | defaultLine = oldDefaultLine; |
---|
2556 | } |
---|
2557 | } |
---|
2558 | |
---|
2559 | /** Gen a named rule block. |
---|
2560 | * ASTs are generated for each element of an alternative unless |
---|
2561 | * the rule or the alternative have a '!' modifier. |
---|
2562 | * |
---|
2563 | * If an alternative defeats the default tree construction, it |
---|
2564 | * must set <rule>_AST to the root of the returned AST. |
---|
2565 | * |
---|
2566 | * Each alternative that does automatic tree construction, builds |
---|
2567 | * up root and child list pointers in an ASTPair structure. |
---|
2568 | * |
---|
2569 | * A rule finishes by setting the returnAST variable from the |
---|
2570 | * ASTPair. |
---|
2571 | * |
---|
2572 | * @param rule The name of the rule to generate |
---|
2573 | * @param startSymbol true if the rule is a start symbol (i.e., not referenced elsewhere) |
---|
2574 | */ |
---|
2575 | public void genRule(RuleSymbol s, boolean startSymbol, int ruleNum) { |
---|
2576 | tabs = 1; |
---|
2577 | |
---|
2578 | if (DEBUG_CODE_GENERATOR) System.out.println("genRule(" + s.getId() + ")"); |
---|
2579 | if (!s.isDefined()) { |
---|
2580 | antlrTool.error("undefined rule: " + s.getId()); |
---|
2581 | return; |
---|
2582 | } |
---|
2583 | |
---|
2584 | // Generate rule return type, name, arguments |
---|
2585 | RuleBlock rblk = s.getBlock(); |
---|
2586 | |
---|
2587 | int oldDefaultLine = defaultLine; |
---|
2588 | try { |
---|
2589 | defaultLine = rblk.getLine(); |
---|
2590 | currentRule = rblk; |
---|
2591 | currentASTResult = s.getId(); |
---|
2592 | |
---|
2593 | // clear list of declared ast variables.. |
---|
2594 | declaredASTVariables.clear(); |
---|
2595 | |
---|
2596 | // Save the AST generation state, and set it to that of the rule |
---|
2597 | boolean savegenAST = genAST; |
---|
2598 | genAST = genAST && rblk.getAutoGen(); |
---|
2599 | |
---|
2600 | // boolean oldsaveTest = saveText; |
---|
2601 | saveText = rblk.getAutoGen(); |
---|
2602 | |
---|
2603 | // print javadoc comment if any |
---|
2604 | if (s.comment != null) { |
---|
2605 | _println(s.comment); |
---|
2606 | } |
---|
2607 | |
---|
2608 | // Gen method access and final qualifier |
---|
2609 | print(s.access + " final "); |
---|
2610 | |
---|
2611 | // Gen method return type (note lexer return action set at rule creation) |
---|
2612 | if (rblk.returnAction != null) { |
---|
2613 | // Has specified return value |
---|
2614 | _print(extractTypeOfAction(rblk.returnAction, rblk.getLine(), rblk.getColumn()) + " "); |
---|
2615 | } |
---|
2616 | else { |
---|
2617 | // No specified return value |
---|
2618 | _print("void "); |
---|
2619 | } |
---|
2620 | |
---|
2621 | // Gen method name |
---|
2622 | _print(s.getId() + "("); |
---|
2623 | |
---|
2624 | // Additional rule parameters common to all rules for this grammar |
---|
2625 | _print(commonExtraParams); |
---|
2626 | if (commonExtraParams.length() != 0 && rblk.argAction != null) { |
---|
2627 | _print(","); |
---|
2628 | } |
---|
2629 | |
---|
2630 | // Gen arguments |
---|
2631 | if (rblk.argAction != null) { |
---|
2632 | // Has specified arguments |
---|
2633 | _println(""); |
---|
2634 | tabs++; |
---|
2635 | println(rblk.argAction); |
---|
2636 | tabs--; |
---|
2637 | print(")"); |
---|
2638 | } |
---|
2639 | else { |
---|
2640 | // No specified arguments |
---|
2641 | _print(")"); |
---|
2642 | } |
---|
2643 | |
---|
2644 | // Gen throws clause and open curly |
---|
2645 | _print(" throws " + exceptionThrown); |
---|
2646 | if (grammar instanceof ParserGrammar) { |
---|
2647 | _print(", TokenStreamException"); |
---|
2648 | } |
---|
2649 | else if (grammar instanceof LexerGrammar) { |
---|
2650 | _print(", CharStreamException, TokenStreamException"); |
---|
2651 | } |
---|
2652 | // Add user-defined exceptions unless lexer (for now) |
---|
2653 | if (rblk.throwsSpec != null) { |
---|
2654 | if (grammar instanceof LexerGrammar) { |
---|
2655 | antlrTool.error("user-defined throws spec not allowed (yet) for lexer rule " + rblk.ruleName); |
---|
2656 | } |
---|
2657 | else { |
---|
2658 | _print(", " + rblk.throwsSpec); |
---|
2659 | } |
---|
2660 | } |
---|
2661 | |
---|
2662 | _println(" {"); |
---|
2663 | tabs++; |
---|
2664 | |
---|
2665 | // Convert return action to variable declaration |
---|
2666 | if (rblk.returnAction != null) |
---|
2667 | println(rblk.returnAction + ";"); |
---|
2668 | |
---|
2669 | // print out definitions needed by rules for various grammar types |
---|
2670 | println(commonLocalVars); |
---|
2671 | |
---|
2672 | if (grammar.traceRules) { |
---|
2673 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2674 | println("traceIn(\"" + s.getId() + "\",_t);"); |
---|
2675 | } |
---|
2676 | else { |
---|
2677 | println("traceIn(\"" + s.getId() + "\");"); |
---|
2678 | } |
---|
2679 | } |
---|
2680 | |
---|
2681 | if (grammar instanceof LexerGrammar) { |
---|
2682 | // lexer rule default return value is the rule's token name |
---|
2683 | // This is a horrible hack to support the built-in EOF lexer rule. |
---|
2684 | if (s.getId().equals("mEOF")) |
---|
2685 | println("_ttype = Token.EOF_TYPE;"); |
---|
2686 | else |
---|
2687 | println("_ttype = " + s.getId().substring(1) + ";"); |
---|
2688 | println("int _saveIndex;"); // used for element! (so we can kill text matched for element) |
---|
2689 | /* |
---|
2690 | println("boolean old_saveConsumedInput=saveConsumedInput;"); |
---|
2691 | if ( !rblk.getAutoGen() ) { // turn off "save input" if ! on rule |
---|
2692 | println("saveConsumedInput=false;"); |
---|
2693 | } |
---|
2694 | */ |
---|
2695 | } |
---|
2696 | |
---|
2697 | // if debugging, write code to mark entry to the rule |
---|
2698 | if (grammar.debuggingOutput) |
---|
2699 | if (grammar instanceof ParserGrammar) |
---|
2700 | println("fireEnterRule(" + ruleNum + ",0);"); |
---|
2701 | else if (grammar instanceof LexerGrammar) |
---|
2702 | println("fireEnterRule(" + ruleNum + ",_ttype);"); |
---|
2703 | |
---|
2704 | // Generate trace code if desired |
---|
2705 | if (grammar.debuggingOutput || grammar.traceRules) { |
---|
2706 | println("try { // debugging"); |
---|
2707 | tabs++; |
---|
2708 | } |
---|
2709 | |
---|
2710 | // Initialize AST variables |
---|
2711 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2712 | // "Input" value for rule |
---|
2713 | println(labeledElementASTType + " " + s.getId() + "_AST_in = (_t == ASTNULL) ? null : (" + labeledElementASTType + ")_t;", NO_MAPPING); |
---|
2714 | } |
---|
2715 | if (grammar.buildAST) { |
---|
2716 | // Parser member used to pass AST returns from rule invocations |
---|
2717 | println("returnAST = null;"); |
---|
2718 | // Tracks AST construction |
---|
2719 | // println("ASTPair currentAST = (inputState.guessing==0) ? new ASTPair() : null;"); |
---|
2720 | println("ASTPair currentAST = new ASTPair();"); |
---|
2721 | // User-settable return value for rule. |
---|
2722 | println(labeledElementASTType + " " + s.getId() + "_AST = null;"); |
---|
2723 | } |
---|
2724 | |
---|
2725 | genBlockPreamble(rblk); |
---|
2726 | genBlockInitAction(rblk); |
---|
2727 | println(""); |
---|
2728 | |
---|
2729 | // Search for an unlabeled exception specification attached to the rule |
---|
2730 | ExceptionSpec unlabeledUserSpec = rblk.findExceptionSpec(""); |
---|
2731 | |
---|
2732 | // Generate try block around the entire rule for error handling |
---|
2733 | if (unlabeledUserSpec != null || rblk.getDefaultErrorHandler()) { |
---|
2734 | println("try { // for error handling"); |
---|
2735 | tabs++; |
---|
2736 | } |
---|
2737 | |
---|
2738 | // Generate the alternatives |
---|
2739 | if (rblk.alternatives.size() == 1) { |
---|
2740 | // One alternative -- use simple form |
---|
2741 | Alternative alt = rblk.getAlternativeAt(0); |
---|
2742 | String pred = alt.semPred; |
---|
2743 | if (pred != null) |
---|
2744 | genSemPred(pred, currentRule.line); |
---|
2745 | if (alt.synPred != null) { |
---|
2746 | antlrTool.warning( |
---|
2747 | "Syntactic predicate ignored for single alternative", |
---|
2748 | grammar.getFilename(), |
---|
2749 | alt.synPred.getLine(), |
---|
2750 | alt.synPred.getColumn() |
---|
2751 | ); |
---|
2752 | } |
---|
2753 | genAlt(alt, rblk); |
---|
2754 | } |
---|
2755 | else { |
---|
2756 | // Multiple alternatives -- generate complex form |
---|
2757 | boolean ok = grammar.theLLkAnalyzer.deterministic(rblk); |
---|
2758 | |
---|
2759 | JavaBlockFinishingInfo howToFinish = genCommonBlock(rblk, false); |
---|
2760 | genBlockFinish(howToFinish, throwNoViable, rblk.getLine()); |
---|
2761 | } |
---|
2762 | |
---|
2763 | // Generate catch phrase for error handling |
---|
2764 | if (unlabeledUserSpec != null || rblk.getDefaultErrorHandler()) { |
---|
2765 | // Close the try block |
---|
2766 | tabs--; |
---|
2767 | println("}"); |
---|
2768 | } |
---|
2769 | |
---|
2770 | // Generate user-defined or default catch phrases |
---|
2771 | if (unlabeledUserSpec != null) { |
---|
2772 | genErrorHandler(unlabeledUserSpec); |
---|
2773 | } |
---|
2774 | else if (rblk.getDefaultErrorHandler()) { |
---|
2775 | // Generate default catch phrase |
---|
2776 | println("catch (" + exceptionThrown + " ex) {"); |
---|
2777 | tabs++; |
---|
2778 | // Generate code to handle error if not guessing |
---|
2779 | if (grammar.hasSyntacticPredicate) { |
---|
2780 | println("if (inputState.guessing==0) {"); |
---|
2781 | tabs++; |
---|
2782 | } |
---|
2783 | println("reportError(ex);"); |
---|
2784 | if (!(grammar instanceof TreeWalkerGrammar)) { |
---|
2785 | // Generate code to consume until token in k==1 follow set |
---|
2786 | Lookahead follow = grammar.theLLkAnalyzer.FOLLOW(1, rblk.endNode); |
---|
2787 | String followSetName = getBitsetName(markBitsetForGen(follow.fset)); |
---|
2788 | println("recover(ex," + followSetName + ");"); |
---|
2789 | } |
---|
2790 | else { |
---|
2791 | // Just consume one token |
---|
2792 | println("if (_t!=null) {_t = _t.getNextSibling();}"); |
---|
2793 | } |
---|
2794 | if (grammar.hasSyntacticPredicate) { |
---|
2795 | tabs--; |
---|
2796 | // When guessing, rethrow exception |
---|
2797 | println("} else {"); |
---|
2798 | println(" throw ex;"); |
---|
2799 | println("}"); |
---|
2800 | } |
---|
2801 | // Close catch phrase |
---|
2802 | tabs--; |
---|
2803 | println("}"); |
---|
2804 | } |
---|
2805 | |
---|
2806 | // Squirrel away the AST "return" value |
---|
2807 | if (grammar.buildAST) { |
---|
2808 | println("returnAST = " + s.getId() + "_AST;"); |
---|
2809 | } |
---|
2810 | |
---|
2811 | // Set return tree value for tree walkers |
---|
2812 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2813 | println("_retTree = _t;"); |
---|
2814 | } |
---|
2815 | |
---|
2816 | // Generate literals test for lexer rules so marked |
---|
2817 | if (rblk.getTestLiterals()) { |
---|
2818 | if (s.access.equals("protected")) { |
---|
2819 | genLiteralsTestForPartialToken(); |
---|
2820 | } |
---|
2821 | else { |
---|
2822 | genLiteralsTest(); |
---|
2823 | } |
---|
2824 | } |
---|
2825 | |
---|
2826 | // if doing a lexer rule, dump code to create token if necessary |
---|
2827 | if (grammar instanceof LexerGrammar) { |
---|
2828 | println("if ( _createToken && _token==null && _ttype!=Token.SKIP ) {"); |
---|
2829 | println(" _token = makeToken(_ttype);"); |
---|
2830 | println(" _token.setText(new String(text.getBuffer(), _begin, text.length()-_begin));"); |
---|
2831 | println("}"); |
---|
2832 | println("_returnToken = _token;"); |
---|
2833 | } |
---|
2834 | |
---|
2835 | // Gen the return statement if there is one (lexer has hard-wired return action) |
---|
2836 | if (rblk.returnAction != null) { |
---|
2837 | println("return " + extractIdOfAction(rblk.returnAction, rblk.getLine(), rblk.getColumn()) + ";"); |
---|
2838 | } |
---|
2839 | |
---|
2840 | if (grammar.debuggingOutput || grammar.traceRules) { |
---|
2841 | tabs--; |
---|
2842 | println("} finally { // debugging"); |
---|
2843 | tabs++; |
---|
2844 | |
---|
2845 | // If debugging, generate calls to mark exit of rule |
---|
2846 | if (grammar.debuggingOutput) |
---|
2847 | if (grammar instanceof ParserGrammar) |
---|
2848 | println("fireExitRule(" + ruleNum + ",0);"); |
---|
2849 | else if (grammar instanceof LexerGrammar) |
---|
2850 | println("fireExitRule(" + ruleNum + ",_ttype);"); |
---|
2851 | |
---|
2852 | if (grammar.traceRules) { |
---|
2853 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2854 | println("traceOut(\"" + s.getId() + "\",_t);"); |
---|
2855 | } |
---|
2856 | else { |
---|
2857 | println("traceOut(\"" + s.getId() + "\");"); |
---|
2858 | } |
---|
2859 | } |
---|
2860 | |
---|
2861 | tabs--; |
---|
2862 | println("}"); |
---|
2863 | } |
---|
2864 | |
---|
2865 | tabs--; |
---|
2866 | println("}"); |
---|
2867 | println(""); |
---|
2868 | |
---|
2869 | // Restore the AST generation state |
---|
2870 | genAST = savegenAST; |
---|
2871 | |
---|
2872 | // restore char save state |
---|
2873 | // saveText = oldsaveTest; |
---|
2874 | } finally { |
---|
2875 | defaultLine = oldDefaultLine; |
---|
2876 | } |
---|
2877 | } |
---|
2878 | |
---|
2879 | private void GenRuleInvocation(RuleRefElement rr) { |
---|
2880 | int oldDefaultLine = defaultLine; |
---|
2881 | try { |
---|
2882 | defaultLine = rr.getLine(); |
---|
2883 | // dump rule name |
---|
2884 | getPrintWriterManager().startSingleSourceLineMapping(rr.getLine()); |
---|
2885 | _print(rr.targetRule + "("); |
---|
2886 | getPrintWriterManager().endMapping(); |
---|
2887 | |
---|
2888 | // lexers must tell rule if it should set _returnToken |
---|
2889 | if (grammar instanceof LexerGrammar) { |
---|
2890 | // if labeled, could access Token, so tell rule to create |
---|
2891 | if (rr.getLabel() != null) { |
---|
2892 | _print("true"); |
---|
2893 | } |
---|
2894 | else { |
---|
2895 | _print("false"); |
---|
2896 | } |
---|
2897 | if (commonExtraArgs.length() != 0 || rr.args != null) { |
---|
2898 | _print(","); |
---|
2899 | } |
---|
2900 | } |
---|
2901 | |
---|
2902 | // Extra arguments common to all rules for this grammar |
---|
2903 | _print(commonExtraArgs); |
---|
2904 | if (commonExtraArgs.length() != 0 && rr.args != null) { |
---|
2905 | _print(","); |
---|
2906 | } |
---|
2907 | |
---|
2908 | // Process arguments to method, if any |
---|
2909 | RuleSymbol rs = (RuleSymbol)grammar.getSymbol(rr.targetRule); |
---|
2910 | if (rr.args != null) { |
---|
2911 | // When not guessing, execute user arg action |
---|
2912 | ActionTransInfo tInfo = new ActionTransInfo(); |
---|
2913 | String args = processActionForSpecialSymbols(rr.args, 0, currentRule, tInfo); |
---|
2914 | if (tInfo.assignToRoot || tInfo.refRuleRoot != null) { |
---|
2915 | antlrTool.error("Arguments of rule reference '" + rr.targetRule + "' cannot set or ref #" + |
---|
2916 | currentRule.getRuleName(), grammar.getFilename(), rr.getLine(), rr.getColumn()); |
---|
2917 | } |
---|
2918 | _print(args); |
---|
2919 | |
---|
2920 | // Warn if the rule accepts no arguments |
---|
2921 | if (rs.block.argAction == null) { |
---|
2922 | antlrTool.warning("Rule '" + rr.targetRule + "' accepts no arguments", grammar.getFilename(), rr.getLine(), rr.getColumn()); |
---|
2923 | } |
---|
2924 | } |
---|
2925 | else { |
---|
2926 | // For C++, no warning if rule has parameters, because there may be default |
---|
2927 | // values for all of the parameters |
---|
2928 | if (rs.block.argAction != null) { |
---|
2929 | antlrTool.warning("Missing parameters on reference to rule " + rr.targetRule, grammar.getFilename(), rr.getLine(), rr.getColumn()); |
---|
2930 | } |
---|
2931 | } |
---|
2932 | _println(");"); |
---|
2933 | |
---|
2934 | // move down to the first child while parsing |
---|
2935 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2936 | println("_t = _retTree;"); |
---|
2937 | } |
---|
2938 | } finally { |
---|
2939 | defaultLine = oldDefaultLine; |
---|
2940 | } |
---|
2941 | } |
---|
2942 | |
---|
2943 | protected void genSemPred(String pred, int line) { |
---|
2944 | // translate $ and # references |
---|
2945 | ActionTransInfo tInfo = new ActionTransInfo(); |
---|
2946 | pred = processActionForSpecialSymbols(pred, line, currentRule, tInfo); |
---|
2947 | // ignore translation info...we don't need to do anything with it. |
---|
2948 | String escapedPred = charFormatter.escapeString(pred); |
---|
2949 | |
---|
2950 | // if debugging, wrap the semantic predicate evaluation in a method |
---|
2951 | // that can tell SemanticPredicateListeners the result |
---|
2952 | if (grammar.debuggingOutput && ((grammar instanceof ParserGrammar) || (grammar instanceof LexerGrammar))) |
---|
2953 | pred = "fireSemanticPredicateEvaluated(antlr.debug.SemanticPredicateEvent.VALIDATING," |
---|
2954 | + addSemPred(escapedPred) + "," + pred + ")"; |
---|
2955 | println("if (!(" + pred + "))", line); |
---|
2956 | println(" throw new SemanticException(\"" + escapedPred + "\");", line); |
---|
2957 | } |
---|
2958 | |
---|
2959 | /** Write an array of Strings which are the semantic predicate |
---|
2960 | * expressions. The debugger will reference them by number only |
---|
2961 | */ |
---|
2962 | protected void genSemPredMap() { |
---|
2963 | Enumeration e = semPreds.elements(); |
---|
2964 | println("private String _semPredNames[] = {", NO_MAPPING); |
---|
2965 | while (e.hasMoreElements()) |
---|
2966 | println("\"" + e.nextElement() + "\",", NO_MAPPING); |
---|
2967 | println("};", NO_MAPPING); |
---|
2968 | } |
---|
2969 | |
---|
2970 | protected void genSynPred(SynPredBlock blk, String lookaheadExpr) { |
---|
2971 | int oldDefaultLine = defaultLine; |
---|
2972 | try { |
---|
2973 | defaultLine = blk.getLine(); |
---|
2974 | if (DEBUG_CODE_GENERATOR) System.out.println("gen=>(" + blk + ")"); |
---|
2975 | |
---|
2976 | // Dump synpred result variable |
---|
2977 | println("boolean synPredMatched" + blk.ID + " = false;"); |
---|
2978 | |
---|
2979 | // inserted by Ole Kniemeyer, December 9, 2005 |
---|
2980 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2981 | println("if (_t==null) _t=ASTNULL;"); |
---|
2982 | } |
---|
2983 | |
---|
2984 | // Gen normal lookahead test |
---|
2985 | println("if (" + lookaheadExpr + ") {"); |
---|
2986 | tabs++; |
---|
2987 | |
---|
2988 | // Save input state |
---|
2989 | if (grammar instanceof TreeWalkerGrammar) { |
---|
2990 | println("AST __t" + blk.ID + " = _t;"); |
---|
2991 | } |
---|
2992 | else { |
---|
2993 | println("int _m" + blk.ID + " = mark();"); |
---|
2994 | } |
---|
2995 | |
---|
2996 | // Once inside the try, assume synpred works unless exception caught |
---|
2997 | println("synPredMatched" + blk.ID + " = true;"); |
---|
2998 | println("inputState.guessing++;"); |
---|
2999 | |
---|
3000 | // if debugging, tell listeners that a synpred has started |
---|
3001 | if (grammar.debuggingOutput && ((grammar instanceof ParserGrammar) || |
---|
3002 | (grammar instanceof LexerGrammar))) { |
---|
3003 | println("fireSyntacticPredicateStarted();"); |
---|
3004 | } |
---|
3005 | |
---|
3006 | syntacticPredLevel++; |
---|
3007 | println("try {"); |
---|
3008 | tabs++; |
---|
3009 | gen((AlternativeBlock)blk); // gen code to test predicate |
---|
3010 | tabs--; |
---|
3011 | //println("System.out.println(\"pred "+blk+" succeeded\");"); |
---|
3012 | println("}"); |
---|
3013 | println("catch (" + exceptionThrown + " pe) {"); |
---|
3014 | tabs++; |
---|
3015 | println("synPredMatched" + blk.ID + " = false;"); |
---|
3016 | //println("System.out.println(\"pred "+blk+" failed\");"); |
---|
3017 | tabs--; |
---|
3018 | println("}"); |
---|
3019 | |
---|
3020 | // Restore input state |
---|
3021 | if (grammar instanceof TreeWalkerGrammar) { |
---|
3022 | println("_t = __t" + blk.ID + ";"); |
---|
3023 | } |
---|
3024 | else { |
---|
3025 | println("rewind(_m" + blk.ID + ");"); |
---|
3026 | } |
---|
3027 | |
---|
3028 | _println("inputState.guessing--;"); |
---|
3029 | |
---|
3030 | // if debugging, tell listeners how the synpred turned out |
---|
3031 | if (grammar.debuggingOutput && ((grammar instanceof ParserGrammar) || |
---|
3032 | (grammar instanceof LexerGrammar))) { |
---|
3033 | println("if (synPredMatched" + blk.ID + ")"); |
---|
3034 | println(" fireSyntacticPredicateSucceeded();"); |
---|
3035 | println("else"); |
---|
3036 | println(" fireSyntacticPredicateFailed();"); |
---|
3037 | } |
---|
3038 | |
---|
3039 | syntacticPredLevel--; |
---|
3040 | tabs--; |
---|
3041 | |
---|
3042 | // Close lookahead test |
---|
3043 | println("}"); |
---|
3044 | |
---|
3045 | // Test synpred result |
---|
3046 | println("if ( synPredMatched" + blk.ID + " ) {"); |
---|
3047 | } finally { |
---|
3048 | defaultLine = oldDefaultLine; |
---|
3049 | } |
---|
3050 | } |
---|
3051 | |
---|
3052 | /** Generate a static array containing the names of the tokens, |
---|
3053 | * indexed by the token type values. This static array is used |
---|
3054 | * to format error messages so that the token identifers or literal |
---|
3055 | * strings are displayed instead of the token numbers. |
---|
3056 | * |
---|
3057 | * If a lexical rule has a paraphrase, use it rather than the |
---|
3058 | * token label. |
---|
3059 | */ |
---|
3060 | public void genTokenStrings() { |
---|
3061 | int oldDefaultLine = defaultLine; |
---|
3062 | try { |
---|
3063 | defaultLine = NO_MAPPING; |
---|
3064 | // Generate a string for each token. This creates a static |
---|
3065 | // array of Strings indexed by token type. |
---|
3066 | println(""); |
---|
3067 | println("public static final String[] _tokenNames = {"); |
---|
3068 | tabs++; |
---|
3069 | |
---|
3070 | // Walk the token vocabulary and generate a Vector of strings |
---|
3071 | // from the tokens. |
---|
3072 | Vector v = grammar.tokenManager.getVocabulary(); |
---|
3073 | for (int i = 0; i < v.size(); i++) { |
---|
3074 | String s = (String)v.elementAt(i); |
---|
3075 | if (s == null) { |
---|
3076 | s = "<" + String.valueOf(i) + ">"; |
---|
3077 | } |
---|
3078 | if (!s.startsWith("\"") && !s.startsWith("<")) { |
---|
3079 | TokenSymbol ts = (TokenSymbol)grammar.tokenManager.getTokenSymbol(s); |
---|
3080 | if (ts != null && ts.getParaphrase() != null) { |
---|
3081 | s = StringUtils.stripFrontBack(ts.getParaphrase(), "\"", "\""); |
---|
3082 | } |
---|
3083 | } |
---|
3084 | print(charFormatter.literalString(s)); |
---|
3085 | if (i != v.size() - 1) { |
---|
3086 | _print(","); |
---|
3087 | } |
---|
3088 | _println(""); |
---|
3089 | } |
---|
3090 | |
---|
3091 | // Close the string array initailizer |
---|
3092 | tabs--; |
---|
3093 | println("};"); |
---|
3094 | } finally { |
---|
3095 | defaultLine = oldDefaultLine; |
---|
3096 | } |
---|
3097 | } |
---|
3098 | |
---|
3099 | /** Create and set Integer token type objects that map |
---|
3100 | * to Java Class objects (which AST node to create). |
---|
3101 | */ |
---|
3102 | protected void genTokenASTNodeMap() { |
---|
3103 | int oldDefaultLine = defaultLine; |
---|
3104 | try { |
---|
3105 | defaultLine = NO_MAPPING; |
---|
3106 | println(""); |
---|
3107 | println("protected void buildTokenTypeASTClassMap() {"); |
---|
3108 | // Generate a map.put("T","TNode") for each token |
---|
3109 | // if heterogeneous node known for that token T. |
---|
3110 | tabs++; |
---|
3111 | boolean generatedNewHashtable = false; |
---|
3112 | int n = 0; |
---|
3113 | // Walk the token vocabulary and generate puts. |
---|
3114 | Vector v = grammar.tokenManager.getVocabulary(); |
---|
3115 | for (int i = 0; i < v.size(); i++) { |
---|
3116 | String s = (String)v.elementAt(i); |
---|
3117 | if (s != null) { |
---|
3118 | TokenSymbol ts = grammar.tokenManager.getTokenSymbol(s); |
---|
3119 | if (ts != null && ts.getASTNodeType() != null) { |
---|
3120 | n++; |
---|
3121 | if ( !generatedNewHashtable ) { |
---|
3122 | // only generate if we are going to add a mapping |
---|
3123 | println("tokenTypeToASTClassMap = new Hashtable();"); |
---|
3124 | generatedNewHashtable = true; |
---|
3125 | } |
---|
3126 | println("tokenTypeToASTClassMap.put(new Integer("+ts.getTokenType()+"), "+ |
---|
3127 | ts.getASTNodeType()+".class);"); |
---|
3128 | } |
---|
3129 | } |
---|
3130 | } |
---|
3131 | |
---|
3132 | if ( n==0 ) { |
---|
3133 | println("tokenTypeToASTClassMap=null;"); |
---|
3134 | } |
---|
3135 | tabs--; |
---|
3136 | println("};"); |
---|
3137 | } finally { |
---|
3138 | defaultLine = oldDefaultLine; |
---|
3139 | } |
---|
3140 | } |
---|
3141 | |
---|
3142 | /** Generate the token types Java file */ |
---|
3143 | protected void genTokenTypes(TokenManager tm) throws IOException { |
---|
3144 | int oldDefaultLine = defaultLine; |
---|
3145 | try { |
---|
3146 | defaultLine = NO_MAPPING; |
---|
3147 | // Open the token output Java file and set the currentOutput stream |
---|
3148 | // SAS: file open was moved to a method so a subclass can override |
---|
3149 | // This was mainly for the VAJ interface |
---|
3150 | currentOutput = getPrintWriterManager().setupOutput(antlrTool, tm.getName() + TokenTypesFileSuffix); |
---|
3151 | |
---|
3152 | tabs = 0; |
---|
3153 | |
---|
3154 | // Generate the header common to all Java files |
---|
3155 | genHeader(); |
---|
3156 | // Do not use printAction because we assume tabs==0 |
---|
3157 | try { |
---|
3158 | defaultLine = behavior.getHeaderActionLine(""); |
---|
3159 | println(behavior.getHeaderAction("")); |
---|
3160 | } finally { |
---|
3161 | defaultLine = NO_MAPPING; |
---|
3162 | } |
---|
3163 | |
---|
3164 | // Encapsulate the definitions in an interface. This can be done |
---|
3165 | // because they are all constants. |
---|
3166 | println("public interface " + tm.getName() + TokenTypesFileSuffix + " {"); |
---|
3167 | tabs++; |
---|
3168 | |
---|
3169 | // Generate a definition for each token type |
---|
3170 | Vector v = tm.getVocabulary(); |
---|
3171 | |
---|
3172 | // Do special tokens manually |
---|
3173 | println("int EOF = " + Token.EOF_TYPE + ";"); |
---|
3174 | println("int NULL_TREE_LOOKAHEAD = " + Token.NULL_TREE_LOOKAHEAD + ";"); |
---|
3175 | |
---|
3176 | for (int i = Token.MIN_USER_TYPE; i < v.size(); i++) { |
---|
3177 | String s = (String)v.elementAt(i); |
---|
3178 | if (s != null) { |
---|
3179 | if (s.startsWith("\"")) { |
---|
3180 | // a string literal |
---|
3181 | StringLiteralSymbol sl = (StringLiteralSymbol)tm.getTokenSymbol(s); |
---|
3182 | if (sl == null) { |
---|
3183 | antlrTool.panic("String literal " + s + " not in symbol table"); |
---|
3184 | } |
---|
3185 | else if (sl.label != null) { |
---|
3186 | println("int " + sl.label + " = " + i + ";"); |
---|
3187 | } |
---|
3188 | else { |
---|
3189 | String mangledName = mangleLiteral(s); |
---|
3190 | if (mangledName != null) { |
---|
3191 | // We were able to create a meaningful mangled token name |
---|
3192 | println("int " + mangledName + " = " + i + ";"); |
---|
3193 | // if no label specified, make the label equal to the mangled name |
---|
3194 | sl.label = mangledName; |
---|
3195 | } |
---|
3196 | else { |
---|
3197 | println("// " + s + " = " + i); |
---|
3198 | } |
---|
3199 | } |
---|
3200 | } |
---|
3201 | else if (!s.startsWith("<")) { |
---|
3202 | println("int " + s + " = " + i + ";"); |
---|
3203 | } |
---|
3204 | } |
---|
3205 | } |
---|
3206 | |
---|
3207 | // Close the interface |
---|
3208 | tabs--; |
---|
3209 | println("}"); |
---|
3210 | |
---|
3211 | // Close the tokens output file |
---|
3212 | getPrintWriterManager().finishOutput(); |
---|
3213 | exitIfError(); |
---|
3214 | } finally { |
---|
3215 | defaultLine = oldDefaultLine; |
---|
3216 | } |
---|
3217 | } |
---|
3218 | |
---|
3219 | /** Get a string for an expression to generate creation of an AST subtree. |
---|
3220 | * @param v A Vector of String, where each element is an expression in the target language yielding an AST node. |
---|
3221 | */ |
---|
3222 | public String getASTCreateString(Vector v) { |
---|
3223 | if (v.size() == 0) { |
---|
3224 | return ""; |
---|
3225 | } |
---|
3226 | StringBuffer buf = new StringBuffer(); |
---|
3227 | buf.append("(" + labeledElementASTType + |
---|
3228 | ")astFactory.make( (new ASTArray(" + v.size() + |
---|
3229 | "))"); |
---|
3230 | for (int i = 0; i < v.size(); i++) { |
---|
3231 | buf.append(".add(" + v.elementAt(i) + ")"); |
---|
3232 | } |
---|
3233 | buf.append(")"); |
---|
3234 | return buf.toString(); |
---|
3235 | } |
---|
3236 | |
---|
3237 | /** Get a string for an expression to generate creating of an AST node |
---|
3238 | * @param atom The grammar node for which you are creating the node |
---|
3239 | * @param str The arguments to the AST constructor |
---|
3240 | */ |
---|
3241 | public String getASTCreateString(GrammarAtom atom, String astCtorArgs) { |
---|
3242 | //System.out.println("getASTCreateString("+atom+","+astCtorArgs+")"); |
---|
3243 | if (atom != null && atom.getASTNodeType() != null) { |
---|
3244 | // they specified a type either on the reference or in tokens{} section |
---|
3245 | return "("+atom.getASTNodeType()+")"+ |
---|
3246 | "astFactory.create("+astCtorArgs+",\""+atom.getASTNodeType()+"\")"; |
---|
3247 | } |
---|
3248 | else { |
---|
3249 | // must be an action or something since not referencing an atom |
---|
3250 | return getASTCreateString(astCtorArgs); |
---|
3251 | } |
---|
3252 | } |
---|
3253 | |
---|
3254 | /** Get a string for an expression to generate creating of an AST node. |
---|
3255 | * Parse the first (possibly only) argument looking for the token type. |
---|
3256 | * If the token type is a valid token symbol, ask for it's AST node type |
---|
3257 | * and add to the end if only 2 arguments. The forms are #[T], #[T,"t"], |
---|
3258 | * and as of 2.7.2 #[T,"t",ASTclassname]. |
---|
3259 | * |
---|
3260 | * @param str The arguments to the AST constructor |
---|
3261 | */ |
---|
3262 | public String getASTCreateString(String astCtorArgs) { |
---|
3263 | //System.out.println("AST CTOR: "+astCtorArgs); |
---|
3264 | if ( astCtorArgs==null ) { |
---|
3265 | astCtorArgs = ""; |
---|
3266 | } |
---|
3267 | int nCommas = 0; |
---|
3268 | for (int i=0; i<astCtorArgs.length(); i++) { |
---|
3269 | if ( astCtorArgs.charAt(i)==',' ) { |
---|
3270 | nCommas++; |
---|
3271 | } |
---|
3272 | } |
---|
3273 | //System.out.println("num commas="+nCommas); |
---|
3274 | if ( nCommas<2 ) { // if 1 or 2 args |
---|
3275 | int firstComma = astCtorArgs.indexOf(','); |
---|
3276 | int lastComma = astCtorArgs.lastIndexOf(','); |
---|
3277 | String tokenName = astCtorArgs; |
---|
3278 | if ( nCommas>0 ) { |
---|
3279 | tokenName = astCtorArgs.substring(0,firstComma); |
---|
3280 | } |
---|
3281 | //System.out.println("Checking for ast node type of "+tokenName); |
---|
3282 | TokenSymbol ts = grammar.tokenManager.getTokenSymbol(tokenName); |
---|
3283 | if ( ts!=null ) { |
---|
3284 | String astNodeType = ts.getASTNodeType(); |
---|
3285 | //System.out.println("node type of "+tokenName+" is "+astNodeType); |
---|
3286 | String emptyText = ""; |
---|
3287 | if ( nCommas==0 ) { |
---|
3288 | // need to add 2nd arg of blank text for token text |
---|
3289 | emptyText = ",\"\""; |
---|
3290 | } |
---|
3291 | if ( astNodeType!=null ) { |
---|
3292 | return "("+astNodeType+")"+ |
---|
3293 | "astFactory.create("+astCtorArgs+emptyText+",\""+astNodeType+"\")"; |
---|
3294 | } |
---|
3295 | // fall through and just do a regular create with cast on front |
---|
3296 | // if necessary (it differs from default "AST"). |
---|
3297 | } |
---|
3298 | if ( labeledElementASTType.equals("AST") ) { |
---|
3299 | return "astFactory.create("+astCtorArgs+")"; |
---|
3300 | } |
---|
3301 | return "("+labeledElementASTType+")"+ |
---|
3302 | "astFactory.create("+astCtorArgs+")"; |
---|
3303 | } |
---|
3304 | // create default type or (since 2.7.2) 3rd arg is classname |
---|
3305 | return "(" + labeledElementASTType + ")astFactory.create(" + astCtorArgs + ")"; |
---|
3306 | } |
---|
3307 | |
---|
3308 | protected String getLookaheadTestExpression(Lookahead[] look, int k) { |
---|
3309 | StringBuffer e = new StringBuffer(100); |
---|
3310 | boolean first = true; |
---|
3311 | |
---|
3312 | e.append("("); |
---|
3313 | for (int i = 1; i <= k; i++) { |
---|
3314 | BitSet p = look[i].fset; |
---|
3315 | if (!first) { |
---|
3316 | e.append(") && ("); |
---|
3317 | } |
---|
3318 | first = false; |
---|
3319 | |
---|
3320 | // Syn preds can yield <end-of-syn-pred> (epsilon) lookahead. |
---|
3321 | // There is no way to predict what that token would be. Just |
---|
3322 | // allow anything instead. |
---|
3323 | if (look[i].containsEpsilon()) { |
---|
3324 | e.append("true"); |
---|
3325 | } |
---|
3326 | else { |
---|
3327 | e.append(getLookaheadTestTerm(i, p)); |
---|
3328 | } |
---|
3329 | } |
---|
3330 | e.append(")"); |
---|
3331 | |
---|
3332 | return e.toString(); |
---|
3333 | } |
---|
3334 | |
---|
3335 | /**Generate a lookahead test expression for an alternate. This |
---|
3336 | * will be a series of tests joined by '&&' and enclosed by '()', |
---|
3337 | * the number of such tests being determined by the depth of the lookahead. |
---|
3338 | */ |
---|
3339 | protected String getLookaheadTestExpression(Alternative alt, int maxDepth) { |
---|
3340 | int depth = alt.lookaheadDepth; |
---|
3341 | if (depth == GrammarAnalyzer.NONDETERMINISTIC) { |
---|
3342 | // if the decision is nondeterministic, do the best we can: LL(k) |
---|
3343 | // any predicates that are around will be generated later. |
---|
3344 | depth = grammar.maxk; |
---|
3345 | } |
---|
3346 | |
---|
3347 | if (maxDepth == 0) { |
---|
3348 | // empty lookahead can result from alt with sem pred |
---|
3349 | // that can see end of token. E.g., A : {pred}? ('a')? ; |
---|
3350 | return "( true )"; |
---|
3351 | } |
---|
3352 | |
---|
3353 | return "(" + getLookaheadTestExpression(alt.cache, depth) + ")"; |
---|
3354 | } |
---|
3355 | |
---|
3356 | /**Generate a depth==1 lookahead test expression given the BitSet. |
---|
3357 | * This may be one of: |
---|
3358 | * 1) a series of 'x==X||' tests |
---|
3359 | * 2) a range test using >= && <= where possible, |
---|
3360 | * 3) a bitset membership test for complex comparisons |
---|
3361 | * @param k The lookahead level |
---|
3362 | * @param p The lookahead set for level k |
---|
3363 | */ |
---|
3364 | protected String getLookaheadTestTerm(int k, BitSet p) { |
---|
3365 | // Determine the name of the item to be compared |
---|
3366 | String ts = lookaheadString(k); |
---|
3367 | |
---|
3368 | // Generate a range expression if possible |
---|
3369 | int[] elems = p.toArray(); |
---|
3370 | if (elementsAreRange(elems)) { |
---|
3371 | return getRangeExpression(k, elems); |
---|
3372 | } |
---|
3373 | |
---|
3374 | // Generate a bitset membership test if possible |
---|
3375 | StringBuffer e; |
---|
3376 | int degree = p.degree(); |
---|
3377 | if (degree == 0) { |
---|
3378 | return "true"; |
---|
3379 | } |
---|
3380 | |
---|
3381 | if (degree >= bitsetTestThreshold) { |
---|
3382 | int bitsetIdx = markBitsetForGen(p); |
---|
3383 | return getBitsetName(bitsetIdx) + ".member(" + ts + ")"; |
---|
3384 | } |
---|
3385 | |
---|
3386 | // Otherwise, generate the long-winded series of "x==X||" tests |
---|
3387 | e = new StringBuffer(); |
---|
3388 | for (int i = 0; i < elems.length; i++) { |
---|
3389 | // Get the compared-to item (token or character value) |
---|
3390 | String cs = getValueString(elems[i]); |
---|
3391 | |
---|
3392 | // Generate the element comparison |
---|
3393 | if (i > 0) e.append("||"); |
---|
3394 | e.append(ts); |
---|
3395 | e.append("=="); |
---|
3396 | e.append(cs); |
---|
3397 | } |
---|
3398 | return e.toString(); |
---|
3399 | } |
---|
3400 | |
---|
3401 | /** Return an expression for testing a contiguous renage of elements |
---|
3402 | * @param k The lookahead level |
---|
3403 | * @param elems The elements representing the set, usually from BitSet.toArray(). |
---|
3404 | * @return String containing test expression. |
---|
3405 | */ |
---|
3406 | public String getRangeExpression(int k, int[] elems) { |
---|
3407 | if (!elementsAreRange(elems)) { |
---|
3408 | antlrTool.panic("getRangeExpression called with non-range"); |
---|
3409 | } |
---|
3410 | int begin = elems[0]; |
---|
3411 | int end = elems[elems.length - 1]; |
---|
3412 | return |
---|
3413 | "(" + lookaheadString(k) + " >= " + getValueString(begin) + " && " + |
---|
3414 | lookaheadString(k) + " <= " + getValueString(end) + ")"; |
---|
3415 | } |
---|
3416 | |
---|
3417 | /** getValueString: get a string representation of a token or char value |
---|
3418 | * @param value The token or char value |
---|
3419 | */ |
---|
3420 | private String getValueString(int value) { |
---|
3421 | String cs; |
---|
3422 | if (grammar instanceof LexerGrammar) { |
---|
3423 | cs = charFormatter.literalChar(value); |
---|
3424 | } |
---|
3425 | else { |
---|
3426 | TokenSymbol ts = grammar.tokenManager.getTokenSymbolAt(value); |
---|
3427 | if (ts == null) { |
---|
3428 | return "" + value; // return token type as string |
---|
3429 | // tool.panic("vocabulary for token type " + value + " is null"); |
---|
3430 | } |
---|
3431 | String tId = ts.getId(); |
---|
3432 | if (ts instanceof StringLiteralSymbol) { |
---|
3433 | // if string literal, use predefined label if any |
---|
3434 | // if no predefined, try to mangle into LITERAL_xxx. |
---|
3435 | // if can't mangle, use int value as last resort |
---|
3436 | StringLiteralSymbol sl = (StringLiteralSymbol)ts; |
---|
3437 | String label = sl.getLabel(); |
---|
3438 | if (label != null) { |
---|
3439 | cs = label; |
---|
3440 | } |
---|
3441 | else { |
---|
3442 | cs = mangleLiteral(tId); |
---|
3443 | if (cs == null) { |
---|
3444 | cs = String.valueOf(value); |
---|
3445 | } |
---|
3446 | } |
---|
3447 | } |
---|
3448 | else { |
---|
3449 | cs = tId; |
---|
3450 | } |
---|
3451 | } |
---|
3452 | return cs; |
---|
3453 | } |
---|
3454 | |
---|
3455 | /**Is the lookahead for this alt empty? */ |
---|
3456 | protected boolean lookaheadIsEmpty(Alternative alt, int maxDepth) { |
---|
3457 | int depth = alt.lookaheadDepth; |
---|
3458 | if (depth == GrammarAnalyzer.NONDETERMINISTIC) { |
---|
3459 | depth = grammar.maxk; |
---|
3460 | } |
---|
3461 | for (int i = 1; i <= depth && i <= maxDepth; i++) { |
---|
3462 | BitSet p = alt.cache[i].fset; |
---|
3463 | if (p.degree() != 0) { |
---|
3464 | return false; |
---|
3465 | } |
---|
3466 | } |
---|
3467 | return true; |
---|
3468 | } |
---|
3469 | |
---|
3470 | private String lookaheadString(int k) { |
---|
3471 | if (grammar instanceof TreeWalkerGrammar) { |
---|
3472 | return "_t.getType()"; |
---|
3473 | } |
---|
3474 | return "LA(" + k + ")"; |
---|
3475 | } |
---|
3476 | |
---|
3477 | /** Mangle a string literal into a meaningful token name. This is |
---|
3478 | * only possible for literals that are all characters. The resulting |
---|
3479 | * mangled literal name is literalsPrefix with the text of the literal |
---|
3480 | * appended. |
---|
3481 | * @return A string representing the mangled literal, or null if not possible. |
---|
3482 | */ |
---|
3483 | private String mangleLiteral(String s) { |
---|
3484 | String mangled = antlrTool.literalsPrefix; |
---|
3485 | for (int i = 1; i < s.length() - 1; i++) { |
---|
3486 | if (!Character.isLetter(s.charAt(i)) && |
---|
3487 | s.charAt(i) != '_') { |
---|
3488 | return null; |
---|
3489 | } |
---|
3490 | mangled += s.charAt(i); |
---|
3491 | } |
---|
3492 | if (antlrTool.upperCaseMangledLiterals) { |
---|
3493 | mangled = mangled.toUpperCase(); |
---|
3494 | } |
---|
3495 | return mangled; |
---|
3496 | } |
---|
3497 | |
---|
3498 | /** Map an identifier to it's corresponding tree-node variable. |
---|
3499 | * This is context-sensitive, depending on the rule and alternative |
---|
3500 | * being generated |
---|
3501 | * @param idParam The identifier name to map |
---|
3502 | * @return The mapped id (which may be the same as the input), or null if the mapping is invalid due to duplicates |
---|
3503 | */ |
---|
3504 | public String mapTreeId(String idParam, ActionTransInfo transInfo) { |
---|
3505 | // if not in an action of a rule, nothing to map. |
---|
3506 | if (currentRule == null) return idParam; |
---|
3507 | |
---|
3508 | boolean in_var = false; |
---|
3509 | String id = idParam; |
---|
3510 | if (grammar instanceof TreeWalkerGrammar) { |
---|
3511 | if (!grammar.buildAST) { |
---|
3512 | in_var = true; |
---|
3513 | } |
---|
3514 | // If the id ends with "_in", then map it to the input variable |
---|
3515 | else if (id.length() > 3 && id.lastIndexOf("_in") == id.length() - 3) { |
---|
3516 | // Strip off the "_in" |
---|
3517 | id = id.substring(0, id.length() - 3); |
---|
3518 | in_var = true; |
---|
3519 | } |
---|
3520 | } |
---|
3521 | |
---|
3522 | // Check the rule labels. If id is a label, then the output |
---|
3523 | // variable is label_AST, and the input variable is plain label. |
---|
3524 | for (int i = 0; i < currentRule.labeledElements.size(); i++) { |
---|
3525 | AlternativeElement elt = (AlternativeElement)currentRule.labeledElements.elementAt(i); |
---|
3526 | if (elt.getLabel().equals(id)) { |
---|
3527 | return in_var ? id : id + "_AST"; |
---|
3528 | } |
---|
3529 | } |
---|
3530 | |
---|
3531 | // Failing that, check the id-to-variable map for the alternative. |
---|
3532 | // If the id is in the map, then output variable is the name in the |
---|
3533 | // map, and input variable is name_in |
---|
3534 | String s = (String)treeVariableMap.get(id); |
---|
3535 | if (s != null) { |
---|
3536 | if (s == NONUNIQUE) { |
---|
3537 | // There is more than one element with this id |
---|
3538 | antlrTool.error("Ambiguous reference to AST element "+id+ |
---|
3539 | " in rule "+currentRule.getRuleName()); |
---|
3540 | |
---|
3541 | return null; |
---|
3542 | } |
---|
3543 | else if (s.equals(currentRule.getRuleName())) { |
---|
3544 | // a recursive call to the enclosing rule is |
---|
3545 | // ambiguous with the rule itself. |
---|
3546 | antlrTool.error("Ambiguous reference to AST element "+id+ |
---|
3547 | " in rule "+currentRule.getRuleName()); |
---|
3548 | return null; |
---|
3549 | } |
---|
3550 | else { |
---|
3551 | return in_var ? s + "_in" : s; |
---|
3552 | } |
---|
3553 | } |
---|
3554 | |
---|
3555 | // Failing that, check the rule name itself. Output variable |
---|
3556 | // is rule_AST; input variable is rule_AST_in (treeparsers). |
---|
3557 | if (id.equals(currentRule.getRuleName())) { |
---|
3558 | String r = in_var ? id + "_AST_in" : id + "_AST"; |
---|
3559 | if (transInfo != null) { |
---|
3560 | if (!in_var) { |
---|
3561 | transInfo.refRuleRoot = r; |
---|
3562 | } |
---|
3563 | } |
---|
3564 | return r; |
---|
3565 | } |
---|
3566 | else { |
---|
3567 | // id does not map to anything -- return itself. |
---|
3568 | return id; |
---|
3569 | } |
---|
3570 | } |
---|
3571 | |
---|
3572 | /** Given an element and the name of an associated AST variable, |
---|
3573 | * create a mapping between the element "name" and the variable name. |
---|
3574 | */ |
---|
3575 | private void mapTreeVariable(AlternativeElement e, String name) { |
---|
3576 | // For tree elements, defer to the root |
---|
3577 | if (e instanceof TreeElement) { |
---|
3578 | mapTreeVariable(((TreeElement)e).root, name); |
---|
3579 | return; |
---|
3580 | } |
---|
3581 | |
---|
3582 | // Determine the name of the element, if any, for mapping purposes |
---|
3583 | String elName = null; |
---|
3584 | |
---|
3585 | // Don't map labeled items |
---|
3586 | if (e.getLabel() == null) { |
---|
3587 | if (e instanceof TokenRefElement) { |
---|
3588 | // use the token id |
---|
3589 | elName = ((TokenRefElement)e).atomText; |
---|
3590 | } |
---|
3591 | else if (e instanceof RuleRefElement) { |
---|
3592 | // use the rule name |
---|
3593 | elName = ((RuleRefElement)e).targetRule; |
---|
3594 | } |
---|
3595 | } |
---|
3596 | // Add the element to the tree variable map if it has a name |
---|
3597 | if (elName != null) { |
---|
3598 | if (treeVariableMap.get(elName) != null) { |
---|
3599 | // Name is already in the map -- mark it as duplicate |
---|
3600 | treeVariableMap.remove(elName); |
---|
3601 | treeVariableMap.put(elName, NONUNIQUE); |
---|
3602 | } |
---|
3603 | else { |
---|
3604 | treeVariableMap.put(elName, name); |
---|
3605 | } |
---|
3606 | } |
---|
3607 | } |
---|
3608 | |
---|
3609 | /** Lexically process $var and tree-specifiers in the action. |
---|
3610 | * This will replace #id and #(...) with the appropriate |
---|
3611 | * function calls and/or variables etc... |
---|
3612 | */ |
---|
3613 | protected String processActionForSpecialSymbols(String actionStr, |
---|
3614 | int line, |
---|
3615 | RuleBlock currentRule, |
---|
3616 | ActionTransInfo tInfo) { |
---|
3617 | if (actionStr == null || actionStr.length() == 0) return null; |
---|
3618 | |
---|
3619 | // The action trans info tells us (at the moment) whether an |
---|
3620 | // assignment was done to the rule's tree root. |
---|
3621 | if (grammar == null) |
---|
3622 | return actionStr; |
---|
3623 | |
---|
3624 | // see if we have anything to do... |
---|
3625 | if ((grammar.buildAST && actionStr.indexOf('#') != -1) || |
---|
3626 | grammar instanceof TreeWalkerGrammar || |
---|
3627 | ((grammar instanceof LexerGrammar || |
---|
3628 | grammar instanceof ParserGrammar) |
---|
3629 | && actionStr.indexOf('$') != -1)) { |
---|
3630 | // Create a lexer to read an action and return the translated version |
---|
3631 | antlr.actions.java.ActionLexer lexer = |
---|
3632 | new antlr.actions.java.ActionLexer(actionStr, |
---|
3633 | currentRule, |
---|
3634 | this, |
---|
3635 | tInfo); |
---|
3636 | |
---|
3637 | lexer.setLineOffset(line); |
---|
3638 | lexer.setFilename(grammar.getFilename()); |
---|
3639 | lexer.setTool(antlrTool); |
---|
3640 | |
---|
3641 | try { |
---|
3642 | lexer.mACTION(true); |
---|
3643 | actionStr = lexer.getTokenObject().getText(); |
---|
3644 | // System.out.println("action translated: "+actionStr); |
---|
3645 | // System.out.println("trans info is "+tInfo); |
---|
3646 | } |
---|
3647 | catch (RecognitionException ex) { |
---|
3648 | lexer.reportError(ex); |
---|
3649 | return actionStr; |
---|
3650 | } |
---|
3651 | catch (TokenStreamException tex) { |
---|
3652 | antlrTool.panic("Error reading action:" + actionStr); |
---|
3653 | return actionStr; |
---|
3654 | } |
---|
3655 | catch (CharStreamException io) { |
---|
3656 | antlrTool.panic("Error reading action:" + actionStr); |
---|
3657 | return actionStr; |
---|
3658 | } |
---|
3659 | } |
---|
3660 | return actionStr; |
---|
3661 | } |
---|
3662 | |
---|
3663 | private void setupGrammarParameters(Grammar g) { |
---|
3664 | if (g instanceof ParserGrammar) { |
---|
3665 | labeledElementASTType = "AST"; |
---|
3666 | if (g.hasOption("ASTLabelType")) { |
---|
3667 | Token tsuffix = g.getOption("ASTLabelType"); |
---|
3668 | if (tsuffix != null) { |
---|
3669 | String suffix = StringUtils.stripFrontBack(tsuffix.getText(), "\"", "\""); |
---|
3670 | if (suffix != null) { |
---|
3671 | labeledElementASTType = suffix; |
---|
3672 | } |
---|
3673 | } |
---|
3674 | } |
---|
3675 | labeledElementType = "Token "; |
---|
3676 | labeledElementInit = "null"; |
---|
3677 | commonExtraArgs = ""; |
---|
3678 | commonExtraParams = ""; |
---|
3679 | commonLocalVars = ""; |
---|
3680 | lt1Value = "LT(1)"; |
---|
3681 | exceptionThrown = "RecognitionException"; |
---|
3682 | throwNoViable = "throw new NoViableAltException(LT(1), getFilename());"; |
---|
3683 | } |
---|
3684 | else if (g instanceof LexerGrammar) { |
---|
3685 | labeledElementType = "char "; |
---|
3686 | labeledElementInit = "'\\0'"; |
---|
3687 | commonExtraArgs = ""; |
---|
3688 | commonExtraParams = "boolean _createToken"; |
---|
3689 | commonLocalVars = "int _ttype; Token _token=null; int _begin=text.length();"; |
---|
3690 | lt1Value = "LA(1)"; |
---|
3691 | exceptionThrown = "RecognitionException"; |
---|
3692 | throwNoViable = "throw new NoViableAltForCharException((char)LA(1), getFilename(), getLine(), getColumn());"; |
---|
3693 | } |
---|
3694 | else if (g instanceof TreeWalkerGrammar) { |
---|
3695 | labeledElementASTType = "AST"; |
---|
3696 | labeledElementType = "AST"; |
---|
3697 | if (g.hasOption("ASTLabelType")) { |
---|
3698 | Token tsuffix = g.getOption("ASTLabelType"); |
---|
3699 | if (tsuffix != null) { |
---|
3700 | String suffix = StringUtils.stripFrontBack(tsuffix.getText(), "\"", "\""); |
---|
3701 | if (suffix != null) { |
---|
3702 | labeledElementASTType = suffix; |
---|
3703 | labeledElementType = suffix; |
---|
3704 | } |
---|
3705 | } |
---|
3706 | } |
---|
3707 | if (!g.hasOption("ASTLabelType")) { |
---|
3708 | g.setOption("ASTLabelType", new Token(ANTLRTokenTypes.STRING_LITERAL, "AST")); |
---|
3709 | } |
---|
3710 | labeledElementInit = "null"; |
---|
3711 | commonExtraArgs = "_t"; |
---|
3712 | commonExtraParams = "AST _t"; |
---|
3713 | commonLocalVars = ""; |
---|
3714 | lt1Value = "(" + labeledElementASTType + ")_t"; |
---|
3715 | exceptionThrown = "RecognitionException"; |
---|
3716 | throwNoViable = "throw new NoViableAltException(_t);"; |
---|
3717 | } |
---|
3718 | else { |
---|
3719 | antlrTool.panic("Unknown grammar type"); |
---|
3720 | } |
---|
3721 | } |
---|
3722 | |
---|
3723 | /** |
---|
3724 | * Get the printwriter manager that manages output |
---|
3725 | * @return The print writer manager |
---|
3726 | */ |
---|
3727 | public JavaCodeGeneratorPrintWriterManager getPrintWriterManager() { |
---|
3728 | if (printWriterManager == null) |
---|
3729 | printWriterManager = new DefaultJavaCodeGeneratorPrintWriterManager(); |
---|
3730 | return printWriterManager; |
---|
3731 | } |
---|
3732 | |
---|
3733 | /** |
---|
3734 | * Set the print writer manager |
---|
3735 | * @param printWriterManager the new manager |
---|
3736 | */ |
---|
3737 | public void setPrintWriterManager(JavaCodeGeneratorPrintWriterManager printWriterManager) { |
---|
3738 | this.printWriterManager = printWriterManager; |
---|
3739 | } |
---|
3740 | |
---|
3741 | /** {@inheritDoc} */ |
---|
3742 | public void setTool(Tool tool) { |
---|
3743 | super.setTool(tool); |
---|
3744 | } |
---|
3745 | } |
---|