CppCodeGenerator


1   package antlr;
2   
3   /* ANTLR Translator Generator
4    * Project led by Terence Parr at http://www.jGuru.com
5    * Software rights: http://www.antlr.org/RIGHTS.html
6    *
7    * $Id: //depot/code/org.antlr/main/main/antlr/CppCodeGenerator.java#26 $
8    */
9   
10  // C++ code generator by Pete Wells: pete@yamuna.demon.co.uk
11  // #line generation contributed by: Ric Klaren <klaren@cs.utwente.nl>
12  
13  import java.util.Enumeration  ;
14  import java.util.Hashtable  ;
15  import java.util.HashSet  ;
16  import antlr.collections.impl.BitSet;
17  import antlr.collections.impl.Vector;
18  import java.io.PrintWriter  ; //SAS: changed for proper text file io
19  import java.io.IOException  ;
20  import java.io.FileWriter  ;
21  
22  /**Generate MyParser.cpp, MyParser.hpp, MyLexer.cpp, MyLexer.hpp and MyParserTokenTypes.hpp */
23  public class CppCodeGenerator extends CodeGenerator {
24      // non-zero if inside syntactic predicate generation
25      protected int syntacticPredLevel = 0;
26  
27      // Are we generating ASTs (for parsers and tree parsers) right now?
28      protected boolean genAST = false;
29  
30      // Are we saving the text consumed (for lexers) right now?
31      protected boolean saveText = false;
32  
33      // Generate #line's
34      protected boolean genHashLines = true;
35      // Used to keep track of lineno in output
36      protected int outputLine;
37      protected String   outputFile;
38  
39      // Grammar parameters set up to handle different grammar classes.
40      // These are used to get instanceof tests out of code generation
41      boolean usingCustomAST = false;
42      String   labeledElementType;
43      String   labeledElementASTType; // mostly the same as labeledElementType except in parsers
44      String   labeledElementASTInit;
45      String   labeledElementInit;
46      String   commonExtraArgs;
47      String   commonExtraParams;
48      String   commonLocalVars;
49      String   lt1Value;
50      String   exceptionThrown;
51      String   throwNoViable;
52  
53      // Tracks the rule being generated.  Used for mapTreeId
54      RuleBlock currentRule;
55      // Tracks the rule or labeled subrule being generated.  Used for AST generation.
56      String   currentASTResult;
57      // Mapping between the ids used in the current alt, and the
58      // names of variables used to represent their AST values.
59      Hashtable   treeVariableMap = new Hashtable  ();
60  
61      /** Used to keep track of which AST variables have been defined in a rule
62       * (except for the #rule_name and #rule_name_in var's
63       */
64      HashSet   declaredASTVariables = new HashSet  ();
65  
66      // Count of unnamed generated variables
67      int astVarNumber = 1;
68      // Special value used to mark duplicate in treeVariableMap
69      protected static final String   NONUNIQUE = new String  ();
70  
71      public static final int caseSizeThreshold = 127; // ascii is max
72  
73      private Vector semPreds;
74  
75      // Used to keep track of which (heterogeneous AST types are used)
76      // which need to be set in the ASTFactory of the generated parser
77      private Vector astTypes;
78  
79      private static String   namespaceStd   = "ANTLR_USE_NAMESPACE(std)";
80      private static String   namespaceAntlr = "ANTLR_USE_NAMESPACE(antlr)";
81      private static NameSpace nameSpace = null;
82  
83      private static final String   preIncludeCpp  = "pre_include_cpp";
84      private static final String   preIncludeHpp  = "pre_include_hpp";
85      private static final String   postIncludeCpp = "post_include_cpp";
86      private static final String   postIncludeHpp = "post_include_hpp";
87  
88      /** Create a C++ code-generator using the given Grammar.
89       * The caller must still call setTool, setBehavior, and setAnalyzer
90       * before generating code.
91       */
92      public CppCodeGenerator() {
93          super();
94          charFormatter = new CppCharFormatter();
95      }
96      /** Adds a semantic predicate string to the sem pred vector
97          These strings will be used to build an array of sem pred names
98          when building a debugging parser.  This method should only be
99          called when the debug option is specified
100      */
101     protected int addSemPred(String   predicate) {
102         semPreds.appendElement(predicate);
103         return semPreds.size()-1;
104     }
105     public void exitIfError()
106     {
107         if (antlrTool.hasError())
108         {
109             System.out.println("Exiting due to errors.");
110             System.exit(1);
111         }
112     }
113     protected int countLines( String   s )
114     {
115         int lines = 0;
116         for( int i = 0; i < s.length(); i++ )
117         {
118             if( s.charAt(i) == '\n' )
119                 lines++;
120         }
121         return lines;
122     }
123     /** Output a String to the currentOutput stream.
124      * Ignored if string is null.
125      * @param s The string to output
126      */
127     protected void _print(String   s)
128     {
129         if (s != null)
130         {
131             outputLine += countLines(s);
132             currentOutput.print(s);
133         }
134     }
135     /** Print an action without leading tabs, attempting to
136      * preserve the current indentation level for multi-line actions
137      * Ignored if string is null.
138      * @param s The action string to output
139      */
140     protected void _printAction(String   s)
141     {
142         if (s != null)
143         {
144             outputLine += countLines(s)+1;
145             super._printAction(s);
146         }
147     }
148     /** Print an action stored in a token surrounded by #line stuff */
149     public void printAction(Token t)
150     {
151         if (t != null)
152         {
153             genLineNo(t.getLine());
154             printTabs();
155             _printAction(processActionForTreeSpecifiers(t.getText(), t.getLine(),
156                                                                 null, null) );
157             genLineNo2();
158         }
159     }
160     /** Print a header action by #line stuff also process any tree construction
161      * @param name The name of the header part
162      */
163     public void printHeaderAction(String   name)
164     {
165         Token a = (antlr.Token)behavior.headerActions.get(name);
166         if (a != null)
167         {
168             genLineNo(a.getLine());
169             println(processActionForTreeSpecifiers(a.getText(), a.getLine(),
170                                                                 null, null) );
171             genLineNo2();
172         }
173     }
174     /** Output a String followed by newline, to the currentOutput stream.
175      * Ignored if string is null.
176      * @param s The string to output
177      */
178     protected void _println(String   s) {
179         if (s != null) {
180             outputLine += countLines(s)+1;
181             currentOutput.println(s);
182         }
183     }
184     /** Output tab indent followed by a String followed by newline,
185      * to the currentOutput stream.  Ignored if string is null.
186      * @param s The string to output
187      */
188     protected void println(String   s) {
189         if (s != null) {
190             printTabs();
191             outputLine += countLines(s)+1;
192             currentOutput.println(s);
193         }
194     }
195 
196     /** Generate a #line or // line depending on options */
197     public void genLineNo(int line) {
198         if ( line == 0 ) {
199             line++;
200         }
201         if( genHashLines )
202             _println("#line "+line+" \""+antlrTool.fileMinusPath(antlrTool.grammarFile)+"\"");
203     }
204 
205     /** Generate a #line or // line depending on options */
206     public void genLineNo(GrammarElement el)
207     {
208         if( el != null )
209             genLineNo(el.getLine());
210     }
211     /** Generate a #line or // line depending on options */
212     public void genLineNo(Token t)
213     {
214         if (t != null)
215             genLineNo(t.getLine());
216     }
217     /** Generate a #line or // line depending on options */
218     public void genLineNo2()
219     {
220         if( genHashLines )
221         {
222             _println("#line "+(outputLine+1)+" \""+outputFile+"\"");
223         }
224     }
225     /**Generate the parser, lexer, treeparser, and token types in C++ */
226     public void gen() {
227         // Do the code generation
228         try {
229             // Loop over all grammars
230             Enumeration   grammarIter = behavior.grammars.elements();
231             while (grammarIter.hasMoreElements()) {
232                 Grammar g = (Grammar)grammarIter.nextElement();
233                 // Connect all the components to each other
234                 g.setGrammarAnalyzer(analyzer);
235                 g.setCodeGenerator(this);
236                 analyzer.setGrammar(g);
237                 // To get right overloading behavior across hetrogeneous grammars
238                 setupGrammarParameters(g);
239                 g.generate();
240                 exitIfError();
241             }
242 
243             // Loop over all token managers (some of which are lexers)
244             Enumeration   tmIter = behavior.tokenManagers.elements();
245             while (tmIter.hasMoreElements()) {
246                 TokenManager tm = (TokenManager)tmIter.nextElement();
247                 if (!tm.isReadOnly()) {
248                     // Write the token manager tokens as C++
249                     // this must appear before genTokenInterchange so that
250                     // labels are set on string literals
251                     genTokenTypes(tm);
252                     // Write the token manager tokens as plain text
253                     genTokenInterchange(tm);
254                 }
255                 exitIfError();
256             }
257         }
258         catch (IOException   e) {
259             System.out.println(e.getMessage());
260         }
261     }
262     /** Generate code for the given grammar element.
263      * @param blk The {...} action to generate
264      */
265     public void gen(ActionElement action) {
266         if ( DEBUG_CODE_GENERATOR ) System.out.println("genAction("+action+")");
267         if ( action.isSemPred ) {
268             genSemPred(action.actionText, action.line);
269         }
270         else {
271             if ( grammar.hasSyntacticPredicate ) {
272                 println("if ( inputState->guessing==0 ) {");
273                 tabs++;
274             }
275 
276             ActionTransInfo tInfo = new ActionTransInfo();
277             String   actionStr = processActionForTreeSpecifiers(action.actionText,
278                                                                               action.getLine(),
279                                                                               currentRule, tInfo);
280 
281             if ( tInfo.refRuleRoot!=null ) {
282                 // Somebody referenced "#rule", make sure translated var is valid
283                 // assignment to #rule is left as a ref also, meaning that assignments
284                 // with no other refs like "#rule = foo();" still forces this code to be
285                 // generated (unnecessarily).
286                 println(tInfo.refRuleRoot + " = "+labeledElementASTType+"(currentAST.root);");
287             }
288 
289             // dump the translated action
290             genLineNo(action);
291             printAction(actionStr);
292             genLineNo2();
293 
294             if ( tInfo.assignToRoot ) {
295                 // Somebody did a "#rule=", reset internal currentAST.root
296                 println("currentAST.root = "+tInfo.refRuleRoot+";");
297                 // reset the child pointer too to be last sibling in sibling list
298                 // now use if else in stead of x ? y : z to shut CC 4.2 up.
299                 println("if ( "+tInfo.refRuleRoot+"!="+labeledElementASTInit+" &&");
300                 tabs++;
301                 println(tInfo.refRuleRoot+"->getFirstChild() != "+labeledElementASTInit+" )");
302                 println("  currentAST.child = "+tInfo.refRuleRoot+"->getFirstChild();");
303                 tabs--;
304                 println("else");
305                 tabs++;
306                 println("currentAST.child = "+tInfo.refRuleRoot+";");
307                 tabs--;
308                 println("currentAST.advanceChildToEnd();");
309             }
310 
311             if ( grammar.hasSyntacticPredicate ) {
312                 tabs--;
313                 println("}");
314             }
315         }
316     }
317 
318     /** Generate code for the given grammar element.
319      * @param blk The "x|y|z|..." block to generate
320      */
321     public void gen(AlternativeBlock blk) {
322         if ( DEBUG_CODE_GENERATOR ) System.out.println("gen("+blk+")");
323         println("{");
324         genBlockPreamble(blk);
325         genBlockInitAction(blk);
326 
327         // Tell AST generation to build subrule result
328         String   saveCurrentASTResult = currentASTResult;
329         if (blk.getLabel() != null) {
330             currentASTResult = blk.getLabel();
331         }
332 
333         boolean ok = grammar.theLLkAnalyzer.deterministic(blk);
334 
335         CppBlockFinishingInfo howToFinish = genCommonBlock(blk, true);
336         genBlockFinish(howToFinish, throwNoViable);
337 
338         println("}");
339 
340         // Restore previous AST generation
341         currentASTResult = saveCurrentASTResult;
342     }
343     /** Generate code for the given grammar element.
344      * @param blk The block-end element to generate.  Block-end
345      * elements are synthesized by the grammar parser to represent
346      * the end of a block.
347      */
348     public void gen(BlockEndElement end) {
349         if ( DEBUG_CODE_GENERATOR ) System.out.println("genRuleEnd("+end+")");
350     }
351     /** Generate code for the given grammar element.
352      * @param blk The character literal reference to generate
353      */
354     public void gen(CharLiteralElement atom) {
355         if ( DEBUG_CODE_GENERATOR ) System.out.println("genChar("+atom+")");
356 
357         if ( atom.getLabel()!=null ) {
358             println(atom.getLabel() + " = " + lt1Value + ";");
359         }
360 
361         boolean oldsaveText = saveText;
362         saveText = saveText && atom.getAutoGenType()==GrammarElement.AUTO_GEN_NONE;
363         genMatch(atom);
364         saveText = oldsaveText;
365     }
366     /** Generate code for the given grammar element.
367      * @param blk The character-range reference to generate
368      */
369     public void gen(CharRangeElement r) {
370         if ( r.getLabel()!=null  && syntacticPredLevel == 0) {
371             println(r.getLabel() + " = " + lt1Value + ";");
372         }
373         // Correctly take care of saveIndex stuff...
374         boolean save = ( grammar instanceof LexerGrammar &&
375                               ( !saveText ||
376                                  r.getAutoGenType() == GrammarElement.AUTO_GEN_BANG )
377                            );
378         if (save)
379          println("_saveIndex=text.length();");
380 
381         println("matchRange("+textOrChar(r.beginText)+","+textOrChar(r.endText)+");");
382 
383         if (save)
384          println("text.setLength(_saveIndex);");
385     }
386     /** Generate the lexer C++ files */
387     public  void gen(LexerGrammar g) throws IOException   {
388         // If debugging, create a new sempred vector for this grammar
389         if (g.debuggingOutput)
390             semPreds = new Vector();
391 
392         setGrammar(g);
393         if (!(grammar instanceof LexerGrammar)) {
394             antlrTool.panic("Internal error generating lexer");
395         }
396 
397         genBody(g);
398         genInclude(g);
399     }
400     /** Generate code for the given grammar element.
401      * @param blk The (...)+ block to generate
402      */
403     public void gen(OneOrMoreBlock blk) {
404         if ( DEBUG_CODE_GENERATOR ) System.out.println("gen+("+blk+")");
405         String   label;
406         String   cnt;
407         println("{ // ( ... )+");
408         genBlockPreamble(blk);
409         if ( blk.getLabel() != null ) {
410             cnt = "_cnt_"+blk.getLabel();
411         }
412         else {
413             cnt = "_cnt" + blk.ID;
414         }
415         println("int "+cnt+"=0;");
416         if ( blk.getLabel() != null ) {
417             label = blk.getLabel();
418         }
419         else {
420             label = "_loop" + blk.ID;
421         }
422 
423         println("for (;;) {");
424         tabs++;
425         // generate the init action for ()+ ()* inside the loop
426         // this allows us to do usefull EOF checking...
427         genBlockInitAction(blk);
428 
429         // Tell AST generation to build subrule result
430         String   saveCurrentASTResult = currentASTResult;
431         if (blk.getLabel() != null) {
432             currentASTResult = blk.getLabel();
433         }
434 
435         boolean ok = grammar.theLLkAnalyzer.deterministic(blk);
436 
437         // generate exit test if greedy set to false
438         // and an alt is ambiguous with exit branch
439         // or when lookahead derived purely from end-of-file
440         // Lookahead analysis stops when end-of-file is hit,
441         // returning set {epsilon}.  Since {epsilon} is not
442         // ambig with any real tokens, no error is reported
443         // by deterministic() routines and we have to check
444         // for the case where the lookahead depth didn't get
445         // set to NONDETERMINISTIC (this only happens when the
446         // FOLLOW contains real atoms + epsilon).
447         boolean generateNonGreedyExitPath = false;
448         int nonGreedyExitDepth = grammar.maxk;
449 
450         if ( !blk.greedy &&
451              blk.exitLookaheadDepth<=grammar.maxk &&
452              blk.exitCache[blk.exitLookaheadDepth].containsEpsilon() )
453         {
454             generateNonGreedyExitPath = true;
455             nonGreedyExitDepth = blk.exitLookaheadDepth;
456         }
457         else if ( !blk.greedy &&
458                   blk.exitLookaheadDepth==LLkGrammarAnalyzer.NONDETERMINISTIC )
459         {
460             generateNonGreedyExitPath = true;
461         }
462 
463         // generate exit test if greedy set to false
464         // and an alt is ambiguous with exit branch
465         if ( generateNonGreedyExitPath ) {
466             if ( DEBUG_CODE_GENERATOR ) {
467                 System.out.println("nongreedy (...)+ loop; exit depth is "+
468                                    blk.exitLookaheadDepth);
469             }
470             String   predictExit =
471                 getLookaheadTestExpression(blk.exitCache,
472                                            nonGreedyExitDepth);
473             println("// nongreedy exit test");
474             println("if ( "+cnt+">=1 && "+predictExit+") goto "+label+";");
475         }
476 
477         CppBlockFinishingInfo howToFinish = genCommonBlock(blk, false);
478         genBlockFinish(
479             howToFinish,
480             "if ( "+cnt+">=1 ) { goto "+label+"; } else {" + throwNoViable + "}"
481         );
482 
483         println(cnt+"++;");
484         tabs--;
485         println("}");
486         println(label+":;");
487         println("}  // ( ... )+");
488 
489         // Restore previous AST generation
490         currentASTResult = saveCurrentASTResult;
491     }
492     /** Generate the parser C++ file */
493     public void gen(ParserGrammar g) throws IOException   {
494 
495         // if debugging, set up a new vector to keep track of sempred
496         //   strings for this grammar
497         if (g.debuggingOutput)
498             semPreds = new Vector();
499 
500         setGrammar(g);
501         if (!(grammar instanceof ParserGrammar)) {
502             antlrTool.panic("Internal error generating parser");
503         }
504 
505         genBody(g);
506         genInclude(g);
507     }
508     /** Generate code for the given grammar element.
509      * @param blk The rule-reference to generate
510      */
511     public void gen(RuleRefElement rr)
512     {
513         if ( DEBUG_CODE_GENERATOR ) System.out.println("genRR("+rr+")");
514         RuleSymbol rs = (RuleSymbol)grammar.getSymbol(rr.targetRule);
515         if (rs == null || !rs.isDefined())
516         {
517             // Is this redundant???
518             antlrTool.error("Rule '" + rr.targetRule + "' is not defined", grammar.getFilename(), rr.getLine(), rr.getColumn());
519             return;
520         }
521         if (!(rs instanceof RuleSymbol))
522         {
523             // Is this redundant???
524             antlrTool.error("'" + rr.targetRule + "' does not name a grammar rule", grammar.getFilename(), rr.getLine(), rr.getColumn());
525             return;
526         }
527 
528         genErrorTryForElement(rr);
529 
530         // AST value for labeled rule refs in tree walker.
531         // This is not AST construction;  it is just the input tree node value.
532         if ( grammar instanceof TreeWalkerGrammar &&
533             rr.getLabel() != null &&
534             syntacticPredLevel == 0 )
535         {
536             println(rr.getLabel() + " = (_t == ASTNULL) ? "+labeledElementASTInit+" : "+lt1Value+";");
537         }
538 
539         // if in lexer and ! on rule ref or alt or rule, save buffer index to
540         // kill later
541         if ( grammar instanceof LexerGrammar && (!saveText||rr.getAutoGenType()==GrammarElement.AUTO_GEN_BANG) )
542         {
543             println("_saveIndex = text.length();");
544         }
545 
546         // Process return value assignment if any
547         printTabs();
548         if (rr.idAssign != null)
549         {
550             // Warn if the rule has no return type
551             if (rs.block.returnAction == null)
552             {
553                 antlrTool.warning("Rule '" + rr.targetRule + "' has no return type", grammar.getFilename(), rr.getLine(), rr.getColumn());
554             }
555             _print(rr.idAssign + "=");
556         } else {
557             // Warn about return value if any, but not inside syntactic predicate
558             if ( !(grammar instanceof LexerGrammar) && syntacticPredLevel == 0 && rs.block.returnAction != null)
559             {
560                 antlrTool.warning("Rule '" + rr.targetRule + "' returns a value", grammar.getFilename(), rr.getLine(), rr.getColumn());
561             }
562         }
563 
564         // Call the rule
565         GenRuleInvocation(rr);
566 
567         // if in lexer and ! on element or alt or rule, save buffer index to kill later
568         if ( grammar instanceof LexerGrammar && (!saveText||rr.getAutoGenType()==GrammarElement.AUTO_GEN_BANG) ) {
569             println("text.erase(_saveIndex);");
570         }
571 
572         // if not in a syntactic predicate
573         if (syntacticPredLevel == 0)
574         {
575             boolean doNoGuessTest = (
576                 grammar.hasSyntacticPredicate &&
577                 (
578                     grammar.buildAST && rr.getLabel() != null ||
579                     (genAST && rr.getAutoGenType() == GrammarElement.AUTO_GEN_NONE)
580                 )
581             );
582 
583             if (doNoGuessTest) {
584                 println("if (inputState->guessing==0) {");
585                 tabs++;
586             }
587 
588             if (grammar.buildAST && rr.getLabel() != null)
589             {
590                 // always gen variable for rule return on labeled rules
591                 println(rr.getLabel() + "_AST = returnAST;");
592             }
593 
594             if (genAST)
595             {
596                 switch (rr.getAutoGenType())
597                 {
598                 case GrammarElement.AUTO_GEN_NONE:
599                     if( usingCustomAST )
600                         println("astFactory->addASTChild(currentAST, static_cast<"+namespaceAntlr+"RefAST>(returnAST));");
601                     else
602                         println("astFactory->addASTChild( currentAST, returnAST );");
603                     break;
604                 case GrammarElement.AUTO_GEN_CARET:
605                     // FIXME: RK: I'm not so sure this should be an error..
606                     // I think it might actually work and be usefull at times.
607                     antlrTool.error("Internal: encountered ^ after rule reference");
608                     break;
609                 default:
610                     break;
611                 }
612             }
613 
614             // if a lexer and labeled, Token label defined at rule level, just set it here
615             if ( grammar instanceof LexerGrammar && rr.getLabel() != null )
616             {
617                 println(rr.getLabel()+"=_returnToken;");
618             }
619 
620             if (doNoGuessTest)
621             {
622                 tabs--;
623                 println("}");
624             }
625         }
626         genErrorCatchForElement(rr);
627     }
628     /** Generate code for the given grammar element.
629      * @param blk The string-literal reference to generate
630      */
631     public void gen(StringLiteralElement atom) {
632         if ( DEBUG_CODE_GENERATOR ) System.out.println("genString("+atom+")");
633 
634         // Variable declarations for labeled elements
635         if (atom.getLabel()!=null && syntacticPredLevel == 0) {
636             println(atom.getLabel() + " = " + lt1Value + ";");
637         }
638 
639         // AST
640         genElementAST(atom);
641 
642         // is there a bang on the literal?
643         boolean oldsaveText = saveText;
644         saveText = saveText && atom.getAutoGenType()==GrammarElement.AUTO_GEN_NONE;
645 
646         // matching
647         genMatch(atom);
648 
649         saveText = oldsaveText;
650 
651         // tack on tree cursor motion if doing a tree walker
652         if (grammar instanceof TreeWalkerGrammar) {
653             println("_t = _t->getNextSibling();");
654         }
655     }
656     /** Generate code for the given grammar element.
657      * @param blk The token-range reference to generate
658      */
659     public void gen(TokenRangeElement r) {
660         genErrorTryForElement(r);
661         if ( r.getLabel()!=null  && syntacticPredLevel == 0) {
662             println(r.getLabel() + " = " + lt1Value + ";");
663         }
664 
665         // AST
666         genElementAST(r);
667 
668         // match
669         println("matchRange("+r.beginText+","+r.endText+");");
670         genErrorCatchForElement(r);
671     }
672     /** Generate code for the given grammar element.
673      * @param blk The token-reference to generate
674      */
675     public void gen(TokenRefElement atom) {
676         if ( DEBUG_CODE_GENERATOR ) System.out.println("genTokenRef("+atom+")");
677         if ( grammar instanceof LexerGrammar ) {
678             antlrTool.panic("Token reference found in lexer");
679         }
680         genErrorTryForElement(atom);
681         // Assign Token value to token label variable
682         if ( atom.getLabel()!=null && syntacticPredLevel == 0) {
683             println(atom.getLabel() + " = " + lt1Value + ";");
684         }
685 
686         // AST
687         genElementAST(atom);
688         // matching
689         genMatch(atom);
690         genErrorCatchForElement(atom);
691 
692         // tack on tree cursor motion if doing a tree walker
693         if (grammar instanceof TreeWalkerGrammar) {
694             println("_t = _t->getNextSibling();");
695         }
696     }
697     public void gen(TreeElement t) {
698         // save AST cursor
699         println(labeledElementType+" __t" + t.ID + " = _t;");
700 
701         // If there is a label on the root, then assign that to the variable
702         if (t.root.getLabel() != null) {
703             println(t.root.getLabel() + " = (_t == ASTNULL) ? "+labeledElementASTInit+" : _t;");
704         }
705 
706         // check for invalid modifiers ! and ^ on tree element roots
707         if ( t.root.getAutoGenType() == GrammarElement.AUTO_GEN_BANG ) {
708             antlrTool.error("Suffixing a root node with '!' is not implemented",
709                           grammar.getFilename(), t.getLine(), t.getColumn());
710             t.root.setAutoGenType(GrammarElement.AUTO_GEN_NONE);
711         }
712         if ( t.root.getAutoGenType() == GrammarElement.AUTO_GEN_CARET ) {
713             antlrTool.warning("Suffixing a root node with '^' is redundant; already a root",
714                              grammar.getFilename(), t.getLine(), t.getColumn());
715             t.root.setAutoGenType(GrammarElement.AUTO_GEN_NONE);
716         }
717 
718         // Generate AST variables
719         genElementAST(t.root);
720         if (grammar.buildAST) {
721             // Save the AST construction state
722             println(namespaceAntlr+"ASTPair __currentAST" + t.ID + " = currentAST;");
723             // Make the next item added a child of the TreeElement root
724             println("currentAST.root = currentAST.child;");
725             println("currentAST.child = "+labeledElementASTInit+";");
726         }
727 
728         // match root
729         if ( t.root instanceof WildcardElement ) {
730             println("if ( _t == ASTNULL ) throw MismatchedTokenException();");
731         }
732         else {
733             genMatch(t.root);
734         }
735         // move to list of children
736         println("_t = _t->getFirstChild();");
737 
738         // walk list of children, generating code for each
739         for (int i=0; i<t.getAlternatives().size(); i++) {
740             Alternative a = t.getAlternativeAt(i);
741             AlternativeElement e = a.head;
742             while ( e != null ) {
743                 e.generate();
744                 e = e.next;
745             }
746         }
747 
748         if (grammar.buildAST) {
749             // restore the AST construction state to that just after the
750             // tree root was added
751             println("currentAST = __currentAST" + t.ID + ";");
752         }
753         // restore AST cursor
754         println("_t = __t" + t.ID + ";");
755         // move cursor to sibling of tree just parsed
756         println("_t = _t->getNextSibling();");
757     }
758     /** Generate the tree-parser C++ files */
759     public void gen(TreeWalkerGrammar g) throws IOException   {
760         setGrammar(g);
761         if (!(grammar instanceof TreeWalkerGrammar)) {
762             antlrTool.panic("Internal error generating tree-walker");
763         }
764 
765         genBody(g);
766         genInclude(g);
767     }
768     /** Generate code for the given grammar element.
769      * @param wc The wildcard element to generate
770      */
771     public void gen(WildcardElement wc) {
772         // Variable assignment for labeled elements
773         if (wc.getLabel()!=null && syntacticPredLevel == 0) {
774             println(wc.getLabel() + " = " + lt1Value + ";");
775         }
776 
777         // AST
778         genElementAST(wc);
779         // Match anything but EOF
780         if (grammar instanceof TreeWalkerGrammar) {
781             println("if ( _t == "+labeledElementASTInit+" ) throw "+namespaceAntlr+"MismatchedTokenException();");
782         }
783         else if (grammar instanceof LexerGrammar) {
784             if ( grammar instanceof LexerGrammar &&
785                     (!saveText||wc.getAutoGenType()==GrammarElement.AUTO_GEN_BANG) ) {
786                 println("_saveIndex = text.length();");
787             }
788             println("matchNot(EOF/*_CHAR*/);");
789             if ( grammar instanceof LexerGrammar &&
790                     (!saveText||wc.getAutoGenType()==GrammarElement.AUTO_GEN_BANG) ) {
791                 println("text.erase(_saveIndex);");      // kill text atom put in buffer
792             }
793         }
794         else {
795             println("matchNot(" + getValueString(Token.EOF_TYPE) + ");");
796         }
797 
798         // tack on tree cursor motion if doing a tree walker
799         if (grammar instanceof TreeWalkerGrammar) {
800             println("_t = _t->getNextSibling();");
801         }
802     }
803     /** Generate code for the given grammar element.
804      * @param blk The (...)* block to generate
805      */
806     public void gen(ZeroOrMoreBlock blk) {
807         if ( DEBUG_CODE_GENERATOR ) System.out.println("gen*("+blk+")");
808         println("{ // ( ... )*");
809         genBlockPreamble(blk);
810         String   label;
811         if ( blk.getLabel() != null ) {
812             label = blk.getLabel();
813         }
814         else {
815             label = "_loop" + blk.ID;
816         }
817         println("for (;;) {");
818         tabs++;
819         // generate the init action for ()+ ()* inside the loop
820         // this allows us to do usefull EOF checking...
821         genBlockInitAction(blk);
822 
823         // Tell AST generation to build subrule result
824         String   saveCurrentASTResult = currentASTResult;
825         if (blk.getLabel() != null) {
826             currentASTResult = blk.getLabel();
827         }
828 
829         boolean ok = grammar.theLLkAnalyzer.deterministic(blk);
830 
831         // generate exit test if greedy set to false
832         // and an alt is ambiguous with exit branch
833         // or when lookahead derived purely from end-of-file
834         // Lookahead analysis stops when end-of-file is hit,
835         // returning set {epsilon}.  Since {epsilon} is not
836         // ambig with any real tokens, no error is reported
837         // by deterministic() routines and we have to check
838         // for the case where the lookahead depth didn't get
839         // set to NONDETERMINISTIC (this only happens when the
840         // FOLLOW contains real atoms + epsilon).
841         boolean generateNonGreedyExitPath = false;
842         int nonGreedyExitDepth = grammar.maxk;
843 
844         if ( !blk.greedy &&
845              blk.exitLookaheadDepth<=grammar.maxk &&
846              blk.exitCache[blk.exitLookaheadDepth].containsEpsilon() )
847         {
848             generateNonGreedyExitPath = true;
849             nonGreedyExitDepth = blk.exitLookaheadDepth;
850         }
851         else if ( !blk.greedy &&
852                   blk.exitLookaheadDepth==LLkGrammarAnalyzer.NONDETERMINISTIC )
853         {
854             generateNonGreedyExitPath = true;
855         }
856         if ( generateNonGreedyExitPath ) {
857             if ( DEBUG_CODE_GENERATOR ) {
858                 System.out.println("nongreedy (...)* loop; exit depth is "+
859                                    blk.exitLookaheadDepth);
860             }
861             String   predictExit =
862                 getLookaheadTestExpression(blk.exitCache,
863                                            nonGreedyExitDepth);
864             println("// nongreedy exit test");
865             println("if ("+predictExit+") goto "+label+";");
866         }
867 
868         CppBlockFinishingInfo howToFinish = genCommonBlock(blk, false);
869         genBlockFinish(howToFinish, "goto " + label + ";");
870 
871         tabs--;
872         println("}");
873         println(label+":;");
874         println("} // ( ... )*");
875 
876         // Restore previous AST generation
877         currentASTResult = saveCurrentASTResult;
878     }
879     /** Generate an alternative.
880       * @param alt  The alternative to generate
881       * @param blk The block to which the alternative belongs
882       */
883     protected void genAlt(Alternative alt, AlternativeBlock blk)
884     {
885         // Save the AST generation state, and set it to that of the alt
886         boolean savegenAST = genAST;
887         genAST = genAST && alt.getAutoGen();
888 
889         boolean oldsaveTest = saveText;
890         saveText = saveText && alt.getAutoGen();
891 
892         // Reset the variable name map for the alternative
893         Hashtable   saveMap = treeVariableMap;
894         treeVariableMap = new Hashtable  ();
895 
896         // Generate try block around the alt for  error handling
897         if (alt.exceptionSpec != null) {
898             println("try {      // for error handling");
899             tabs++;
900         }
901 
902         AlternativeElement elem = alt.head;
903         while ( !(elem instanceof BlockEndElement) ) {
904             elem.generate(); // alt can begin with anything. Ask target to gen.
905             elem = elem.next;
906         }
907 
908         if ( genAST)
909         {
910             if (blk instanceof RuleBlock)
911             {
912                 // Set the AST return value for the rule
913                 RuleBlock rblk = (RuleBlock)blk;
914                 if( usingCustomAST )
915                     println(rblk.getRuleName() + "_AST = static_cast<"+labeledElementASTType+">(currentAST.root);");
916                 else
917                     println(rblk.getRuleName() + "_AST = currentAST.root;");
918             }
919             else if (blk.getLabel() != null) {
920                 // ### future: also set AST value for labeled subrules.
921                 // println(blk.getLabel() + "_AST = "+labeledElementASTType+"(currentAST.root);");
922                 antlrTool.warning("Labeled subrules are not implemented", grammar.getFilename(), blk.getLine(), blk.getColumn());
923             }
924         }
925 
926         if (alt.exceptionSpec != null)
927         {
928             // close try block
929             tabs--;
930             println("}");
931             genErrorHandler(alt.exceptionSpec);
932         }
933 
934         genAST = savegenAST;
935         saveText = oldsaveTest;
936 
937         treeVariableMap = saveMap;
938     }
939     /** Generate all the bitsets to be used in the parser or lexer
940      * Generate the raw bitset data like "long _tokenSet1_data[] = {...};"
941      * and the BitSet object declarations like "BitSet _tokenSet1 = new BitSet(_tokenSet1_data);"
942      * Note that most languages do not support object initialization inside a
943      * class definition, so other code-generators may have to separate the
944      * bitset declarations from the initializations (e.g., put the initializations
945      * in the generated constructor instead).
946      * @param bitsetList The list of bitsets to generate.
947      * @param maxVocabulary Ensure that each generated bitset can contain at least this value.
948      * @param dumpSets Dump out the token definitions of the contents of the bitset
949      * only for grammars/parsers.
950      */
951     protected void genBitsets(
952         Vector bitsetList,
953         int maxVocabulary,
954         String   prefix
955     )
956     {
957         TokenManager tm = grammar.tokenManager;
958 
959         println("");
960 
961         for (int i = 0; i < bitsetList.size(); i++)
962         {
963             BitSet p = (BitSet)bitsetList.elementAt(i);
964             // Ensure that generated BitSet is large enough for vocabulary
965             p.growToInclude(maxVocabulary);
966 
967             // initialization data
968             println(
969                 "const unsigned long " + prefix + getBitsetName(i) + "_data_" + "[] = { " +
970                 p.toStringOfHalfWords() +
971                 " };"
972             );
973 
974             // Dump the contents of the bitset in readable format...
975             String   t = "// ";
976             for( int j = 0; j < tm.getVocabulary().size(); j++ )
977             {
978                 if ( p.member( j ) )
979                 {
980                     if ( (grammar instanceof LexerGrammar) )
981                         t += tm.getVocabulary().elementAt(j)+" ";
982                     else
983                         t += tm.getTokenStringAt(j)+" ";
984 
985                     if( t.length() > 70 )
986                     {
987                         println(t);
988                         t = "// ";
989                     }
990                 }
991             }
992             if ( t != "// " )
993                 println(t);
994 
995             // BitSet object
996             println(
997                 "const "+namespaceAntlr+"BitSet " + prefix + getBitsetName(i) + "(" +
998                 getBitsetName(i) + "_data_," + p.size()/32 +
999                 ");"
1000            );
1001        }
1002    }
1003    protected void genBitsetsHeader(
1004        Vector bitsetList,
1005        int maxVocabulary
1006    ) {
1007        println("");
1008        for (int i = 0; i < bitsetList.size(); i++)
1009        {
1010            BitSet p = (BitSet)bitsetList.elementAt(i);
1011            // Ensure that generated BitSet is large enough for vocabulary
1012            p.growToInclude(maxVocabulary);
1013            // initialization data
1014            println("static const unsigned long " + getBitsetName(i) + "_data_" + "[];");
1015            // BitSet object
1016            println("static const "+namespaceAntlr+"BitSet " + getBitsetName(i) + ";");
1017        }
1018    }
1019    /** Generate the finish of a block, using a combination of the info
1020     * returned from genCommonBlock() and the action to perform when
1021     * no alts were taken
1022     * @param howToFinish The return of genCommonBlock()
1023     * @param noViableAction What to generate when no alt is taken
1024     */
1025    private void genBlockFinish(CppBlockFinishingInfo howToFinish, String   noViableAction)
1026    {
1027        if (howToFinish.needAnErrorClause &&
1028             (howToFinish.generatedAnIf || howToFinish.generatedSwitch)) {
1029            if ( howToFinish.generatedAnIf ) {
1030                println("else {");
1031            }
1032            else {
1033                println("{");
1034            }
1035            tabs++;
1036            println(noViableAction);
1037            tabs--;
1038            println("}");
1039        }
1040
1041        if ( howToFinish.postscript!=null ) {
1042            println(howToFinish.postscript);
1043        }
1044    }
1045    /** Generate the initaction for a block, which may be a RuleBlock or a
1046     * plain AlternativeBLock.
1047     * @blk The block for which the preamble is to be generated.
1048     */
1049    protected void genBlockInitAction( AlternativeBlock blk )
1050    {
1051        // dump out init action
1052        if ( blk.initAction!=null ) {
1053            genLineNo(blk);
1054            printAction(processActionForTreeSpecifiers(blk.initAction, blk.line,
1055                                                                     currentRule, null) );
1056            genLineNo2();
1057        }
1058    }
1059    /** Generate the header for a block, which may be a RuleBlock or a
1060     * plain AlternativeBLock. This generates any variable declarations
1061     * and syntactic-predicate-testing variables.
1062     * @blk The block for which the preamble is to be generated.
1063     */
1064    protected void genBlockPreamble(AlternativeBlock blk) {
1065        // define labels for rule blocks.
1066        if ( blk instanceof RuleBlock ) {
1067            RuleBlock rblk = (RuleBlock)blk;
1068            if ( rblk.labeledElements!=null ) {
1069                for (int i=0; i<rblk.labeledElements.size(); i++) {
1070
1071                    AlternativeElement a = (AlternativeElement)rblk.labeledElements.elementAt(i);
1072                    //System.out.println("looking at labeled element: "+a);
1073                    // Variables for labeled rule refs and subrules are different than
1074                    // variables for grammar atoms.  This test is a little tricky because
1075                    // we want to get all rule refs and ebnf, but not rule blocks or
1076                    // syntactic predicates
1077                    if (
1078                        a instanceof RuleRefElement ||
1079                        a instanceof AlternativeBlock &&
1080                        !(a instanceof RuleBlock) &&
1081                        !(a instanceof SynPredBlock)
1082                    ) {
1083
1084                        if (
1085                            !(a instanceof RuleRefElement) &&
1086                            ((AlternativeBlock)a).not &&
1087                            analyzer.subruleCanBeInverted(((AlternativeBlock)a), grammar instanceof LexerGrammar)
1088                        ) {
1089                            // Special case for inverted subrules that will be
1090                            // inlined. Treat these like token or char literal
1091                            // references
1092                            println(labeledElementType + " " + a.getLabel() + " = " + labeledElementInit + ";");
1093                            if (grammar.buildAST) {
1094                                genASTDeclaration( a );
1095                            }
1096                        }
1097                        else {
1098                            if (grammar.buildAST) {
1099                                // Always gen AST variables for labeled elements, even if the
1100                                // element itself is marked with !
1101                                genASTDeclaration( a );
1102                            }
1103                            if ( grammar instanceof LexerGrammar ) {
1104                                println(namespaceAntlr+"RefToken "+a.getLabel()+";");
1105                            }
1106                            if (grammar instanceof TreeWalkerGrammar) {
1107                                // always generate rule-ref variables for tree walker
1108                                println(labeledElementType + " " + a.getLabel() + " = " + labeledElementInit + ";");
1109                            }
1110                        }
1111                    }
1112                    else {
1113                        // It is a token or literal reference.  Generate the
1114                        // correct variable type for this grammar
1115                        println(labeledElementType + " " + a.getLabel() + " = " + labeledElementInit + ";");
1116                        // In addition, generate *_AST variables if building ASTs
1117                        if (grammar.buildAST) {
1118//                          println(labeledElementASTType+" " + a.getLabel() + "_AST = "+labeledElementASTInit+";");
1119//  RK: Hmm this is strange.
1120//  The java codegen contained this check for nodetype while we only had the
1121//  println right above here..
1122                            if (a instanceof GrammarAtom &&
1123                                 ((GrammarAtom)a).getASTNodeType()!=null ) {
1124                                GrammarAtom ga = (GrammarAtom)a;
1125                                genASTDeclaration( a, ga.getASTNodeType() );
1126                            }
1127                            else {
1128                                genASTDeclaration( a );
1129                            }
1130                        }
1131                    }
1132                }
1133            }
1134        }
1135    }
1136    public void genBody(LexerGrammar g) throws IOException  
1137    {
1138        outputFile = grammar.getClassName() + ".cpp";
1139        outputLine = 1;
1140        currentOutput = antlrTool.openOutputFile(outputFile);
1141        //SAS: changed for proper text file io
1142
1143        genAST = false; // no way to gen trees.
1144        saveText = true;    // save consumed characters.
1145
1146        tabs=0;
1147
1148        // Generate header common to all C++ output files
1149        genHeader(outputFile);
1150
1151        printHeaderAction(preIncludeCpp);
1152        // Generate header specific to lexer C++ file
1153        println("#include \"" + grammar.getClassName() + ".hpp\"");
1154        println("#include <antlr/CharBuffer.hpp>");
1155        println("#include <antlr/TokenStreamException.hpp>");
1156        println("#include <antlr/TokenStreamIOException.hpp>");
1157        println("#include <antlr/TokenStreamRecognitionException.hpp>");
1158        println("#include <antlr/CharStreamException.hpp>");
1159        println("#include <antlr/CharStreamIOException.hpp>");
1160        println("#include <antlr/NoViableAltForCharException.hpp>");
1161        if (grammar.debuggingOutput)
1162            println("#include <antlr/DebuggingInputBuffer.hpp>");
1163        println("");
1164        printHeaderAction(postIncludeCpp);
1165
1166        if (nameSpace != null)
1167            nameSpace.emitDeclarations(currentOutput);
1168
1169        // Generate user-defined lexer file preamble
1170        printAction(grammar.preambleAction);
1171
1172        // Generate lexer class definition
1173        String   sup=null;
1174        if ( grammar.superClass!=null ) {
1175            sup = grammar.superClass;
1176        }
1177        else {
1178            sup = grammar.getSuperClass();
1179            if (sup.lastIndexOf('.') != -1)
1180                sup = sup.substring(sup.lastIndexOf('.')+1);
1181            sup = namespaceAntlr + sup;
1182        }
1183
1184        //
1185        // Generate the constructor from InputStream
1186        //
1187        println(grammar.getClassName() + "::" + grammar.getClassName() + "(" + namespaceStd + "istream& in)");
1188        tabs++;
1189        // if debugging, wrap the input buffer in a debugger
1190        if (grammar.debuggingOutput)
1191            println(": " + sup + "(new "+namespaceAntlr+"DebuggingInputBuffer(new "+namespaceAntlr+"CharBuffer(in)))");
1192        else
1193            println(": " + sup + "(new "+namespaceAntlr+"CharBuffer(in))");
1194        tabs--;
1195        println("{");
1196        tabs++;
1197
1198        // if debugging, set up array variables and call user-overridable
1199        //   debugging setup method
1200        if ( grammar.debuggingOutput ) {
1201            println("setRuleNames(_ruleNames);");
1202            println("setSemPredNames(_semPredNames);");
1203            println("setupDebugging();");
1204        }
1205
1206        println("setCaseSensitive("+g.caseSensitive+");");
1207        println("initLiterals();");
1208        tabs--;
1209        println("}");
1210        println("");
1211
1212        // Generate the constructor from InputBuffer
1213        println(grammar.getClassName() + "::" + grammar.getClassName() + "("+namespaceAntlr+"InputBuffer& ib)");
1214        tabs++;
1215        // if debugging, wrap the input buffer in a debugger
1216        if (grammar.debuggingOutput)
1217            println(": " + sup + "(new "+namespaceAntlr+"DebuggingInputBuffer(ib))");
1218        else
1219            println(": " + sup + "(ib)");
1220        tabs--;
1221        println("{");
1222        tabs++;
1223
1224        // if debugging, set up array variables and call user-overridable
1225        //   debugging setup method
1226        if ( grammar.debuggingOutput ) {
1227            println("setRuleNames(_ruleNames);");
1228            println("setSemPredNames(_semPredNames);");
1229            println("setupDebugging();");
1230        }
1231
1232        println("setCaseSensitive("+g.caseSensitive+");");
1233        println("initLiterals();");
1234        tabs--;
1235        println("}");
1236        println("");
1237
1238        // Generate the constructor from LexerSharedInputState
1239        println(grammar.getClassName() + "::" + grammar.getClassName() + "(const "+namespaceAntlr+"LexerSharedInputState& state)");
1240        tabs++;
1241        println(": " + sup + "(state)");
1242        tabs--;
1243        println("{");
1244        tabs++;
1245
1246        // if debugging, set up array variables and call user-overridable
1247        //   debugging setup method
1248        if ( grammar.debuggingOutput ) {
1249            println("setRuleNames(_ruleNames);");
1250            println("setSemPredNames(_semPredNames);");
1251            println("setupDebugging();");
1252        }
1253
1254        println("setCaseSensitive("+g.caseSensitive+");");
1255        println("initLiterals();");
1256        tabs--;
1257        println("}");
1258        println("");
1259
1260        println("void " + grammar.getClassName() + "::initLiterals()");
1261        println("{");
1262        tabs++;
1263        // Generate the initialization of the map
1264        // containing the string literals used in the lexer
1265        // The literals variable itself is in CharScanner
1266        /* TJP: get keys now and check to make sure it's a literal not
1267         * a label to a literal; was dup'ing string literals before
1268         * change.
1269         *
1270        Enumeration ids = grammar.tokenManager.getTokenSymbolElements();
1271        while ( ids.hasMoreElements() ) {
1272            TokenSymbol sym = (TokenSymbol)ids.nextElement();
1273            if ( sym instanceof StringLiteralSymbol ) {
1274                StringLiteralSymbol s = (StringLiteralSymbol)sym;
1275                println("literals["+s.getId()+"] = "+s.getTokenType()+";");
1276            }
1277        }
1278        */
1279        // TJP changed it to following loop.
1280        Enumeration   keys = grammar.tokenManager.getTokenSymbolKeys();
1281        while ( keys.hasMoreElements() ) {
1282            String   key = (String  )keys.nextElement();
1283            if ( key.charAt(0) != '"' ) {
1284                continue;
1285            }
1286            TokenSymbol sym = grammar.tokenManager.getTokenSymbol(key);
1287            if ( sym instanceof StringLiteralSymbol ) {
1288                StringLiteralSymbol s = (StringLiteralSymbol)sym;
1289                println("literals["+s.getId()+"] = "+s.getTokenType()+";");
1290            }
1291        }
1292
1293        // Generate the setting of various generated options.
1294        tabs--;
1295        println("}");
1296
1297        // Generate getCaseSensitiveLiterals() method
1298        println("bool " + grammar.getClassName() + "::getCaseSensitiveLiterals() const");
1299        println("{");
1300        tabs++;
1301        println("return "+g.caseSensitiveLiterals + ";");
1302        tabs--;
1303        println("}");
1304
1305        Enumeration   ids;
1306        // generate the rule name array for debugging
1307        if (grammar.debuggingOutput) {
1308            println("const char* "+grammar.getClassName()+"::_ruleNames[] = {");
1309            tabs++;
1310
1311            ids = grammar.rules.elements();
1312            int ruleNum=0;
1313            while ( ids.hasMoreElements() ) {
1314                GrammarSymbol sym = (GrammarSymbol) ids.nextElement();
1315                if ( sym instanceof RuleSymbol)
1316                    println("\""+((RuleSymbol)sym).getId()+"\",");
1317            }
1318            println("0");
1319            tabs--;
1320            println("};");
1321        }
1322
1323        // Generate nextToken() rule.
1324        // nextToken() is a synthetic lexer rule that is the implicit OR of all
1325        // user-defined lexer rules.
1326        genNextToken();
1327
1328        // Generate code for each rule in the lexer
1329        ids = grammar.rules.elements();
1330        int ruleNum=0;
1331        while ( ids.hasMoreElements() ) {
1332            RuleSymbol sym = (RuleSymbol) ids.nextElement();
1333            // Don't generate the synthetic rules
1334            if (!sym.getId().equals("mnextToken")) {
1335                genRule(sym, false, ruleNum++, grammar.getClassName() + "::");
1336            }
1337            exitIfError();
1338        }
1339
1340        // Generate the semantic predicate map for debugging
1341        if (grammar.debuggingOutput)
1342            genSemPredMap(grammar.getClassName() + "::");
1343
1344        // Generate the bitsets used throughout the lexer
1345        genBitsets(bitsetsUsed, ((LexerGrammar)grammar).charVocabulary.size(), grammar.getClassName() + "::" );
1346
1347        println("");
1348        if (nameSpace != null)
1349            nameSpace.emitClosures(currentOutput);
1350
1351        // Close the lexer output stream
1352        currentOutput.close();
1353        currentOutput = null;
1354    }
1355    public void genBody(ParserGrammar g) throws IOException  
1356    {
1357        // Open the output stream for the parser and set the currentOutput
1358        outputFile = grammar.getClassName() + ".cpp";
1359        outputLine = 1;
1360        currentOutput = antlrTool.openOutputFile(outputFile);
1361
1362        genAST = grammar.buildAST;
1363
1364        tabs = 0;
1365
1366        // Generate the header common to all output files.
1367        genHeader(outputFile);
1368
1369        printHeaderAction(preIncludeCpp);
1370
1371        // Generate header for the parser
1372        println("#include \"" + grammar.getClassName() + ".hpp\"");
1373        println("#include <antlr/NoViableAltException.hpp>");
1374        println("#include <antlr/SemanticException.hpp>");
1375        println("#include <antlr/ASTFactory.hpp>");
1376
1377        printHeaderAction(postIncludeCpp);
1378
1379        if (nameSpace != null)
1380            nameSpace.emitDeclarations(currentOutput);
1381
1382        // Output the user-defined parser preamble
1383        printAction(grammar.preambleAction);
1384
1385        String   sup=null;
1386        if ( grammar.superClass!=null )
1387            sup = grammar.superClass;
1388        else {
1389            sup = grammar.getSuperClass();
1390            if (sup.lastIndexOf('.') != -1)
1391                sup = sup.substring(sup.lastIndexOf('.')+1);
1392            sup = namespaceAntlr + sup;
1393        }
1394
1395        // set up an array of all the rule names so the debugger can
1396        // keep track of them only by number -- less to store in tree...
1397        if (grammar.debuggingOutput) {
1398            println("const char* "+grammar.getClassName()+"::_ruleNames[] = {");
1399            tabs++;
1400
1401            Enumeration   ids = grammar.rules.elements();
1402            int ruleNum=0;
1403            while ( ids.hasMoreElements() ) {
1404                GrammarSymbol sym = (GrammarSymbol) ids.nextElement();
1405                if ( sym instanceof RuleSymbol)
1406                    println("\""+((RuleSymbol)sym).getId()+"\",");
1407            }
1408            println("0");
1409            tabs--;
1410            println("};");
1411        }
1412
1413        // Generate _initialize function
1414        println("void " + grammar.getClassName() + "::_initialize(void)");
1415        println("{");
1416        tabs++;
1417
1418//      println("setTokenNames(_tokenNames);");
1419
1420        if( grammar.buildAST )
1421            println("initializeFactory();");
1422
1423        // if debugging, set up arrays and call the user-overridable
1424        //   debugging setup method
1425        if ( grammar.debuggingOutput ) {
1426            println("setRuleNames(_ruleNames);");
1427            println("setSemPredNames(_semPredNames);");
1428            println("setupDebugging();");
1429        }
1430        tabs--;
1431        println("}");
1432
1433        // Generate parser class constructor from TokenBuffer
1434        print(grammar.getClassName() + "::" + grammar.getClassName());
1435        println("("+namespaceAntlr+"TokenBuffer& tokenBuf, int k)");
1436        println(": " + sup + "(tokenBuf,k)");
1437        println("{");
1438        tabs++;
1439        println("_initialize();");
1440        tabs--;
1441        println("}");
1442        println("");
1443
1444        print(grammar.getClassName() + "::" + grammar.getClassName());
1445        println("("+namespaceAntlr+"TokenBuffer& tokenBuf)");
1446        println(": " + sup + "(tokenBuf," + grammar.maxk + ")");
1447        println("{");
1448        tabs++;
1449        println("_initialize();");
1450        tabs--;
1451        println("}");
1452        println("");
1453
1454        // Generate parser class constructor from TokenStream
1455        print(grammar.getClassName() + "::" + grammar.getClassName());
1456        println("("+namespaceAntlr+"TokenStream& lexer, int k)");
1457        println(": " + sup + "(lexer,k)");
1458        println("{");
1459        tabs++;
1460        println("_initialize();");
1461        tabs--;
1462        println("}");
1463        println("");
1464
1465        print(grammar.getClassName() + "::" + grammar.getClassName());
1466        println("("+namespaceAntlr+"TokenStream& lexer)");
1467        println(": " + sup + "(lexer," + grammar.maxk + ")");
1468        println("{");
1469        tabs++;
1470        println("_initialize();");
1471        tabs--;
1472        println("}");
1473        println("");
1474
1475        print(grammar.getClassName() + "::" + grammar.getClassName());
1476        println("(const "+namespaceAntlr+"ParserSharedInputState& state)");
1477        println(": " + sup + "(state," + grammar.maxk + ")");
1478        println("{");
1479        tabs++;
1480        println("_initialize();");
1481        tabs--;
1482        println("}");
1483        println("");
1484
1485        astTypes = new Vector();
1486
1487        // Generate code for each rule in the grammar
1488        Enumeration   ids = grammar.rules.elements();
1489        int ruleNum=0;
1490        while ( ids.hasMoreElements() ) {
1491            GrammarSymbol sym = (GrammarSymbol) ids.nextElement();
1492            if ( sym instanceof RuleSymbol) {
1493                RuleSymbol rs = (RuleSymbol)sym;
1494                genRule(rs, rs.references.size()==0, ruleNum++, grammar.getClassName() + "::");
1495            }
1496            exitIfError();
1497        }
1498        if ( usingCustomAST )
1499        {
1500//          println("void "+grammar.getClassName()+"::setASTNodeFactory("+labeledElementASTType+" (*factory)() )");
1501//          println("{");
1502//          println("}");
1503//          println("");
1504
1505            // when we are using a custom ast override Parser::getAST to return the
1506            // custom AST type
1507            println(labeledElementASTType+" "+grammar.getClassName()+"::getAST()");
1508            println("{");
1509            println("\treturn returnAST;");
1510            println("}");
1511            println("");
1512        }
1513
1514        // Generate the method that initializes the ASTFactory when we're
1515        // building AST's
1516        println("void "+ grammar.getClassName() + "::initializeFactory( void )");
1517        println("{");
1518        tabs++;
1519        if( grammar.buildAST )
1520        {
1521            println("if( ! astFactory )");
1522            tabs++;
1523            if( usingCustomAST )
1524            {
1525                // Hmmm don't like this... kick of Ref Prefix...
1526                String   nodetype = labeledElementASTType.substring(3, labeledElementASTType.length());
1527                println("astFactory = new "+namespaceAntlr+"ASTFactory(\""+nodetype+"\", "+nodetype+"::factory);");
1528            }
1529            else
1530                println("astFactory = new "+namespaceAntlr+"ASTFactory();");
1531            tabs--;
1532            Enumeration   e = astTypes.elements();
1533            while( e.hasMoreElements() )
1534                println((String  )e.nextElement());
1535            println("astFactory->setMaxNodeType("+grammar.tokenManager.maxTokenType()+");");
1536        }
1537        tabs--;
1538        println("}");
1539
1540        // Generate the token names
1541        genTokenStrings(grammar.getClassName() + "::");
1542
1543        // Generate the bitsets used throughout the grammar
1544        genBitsets(bitsetsUsed, grammar.tokenManager.maxTokenType(), grammar.getClassName() + "::" );
1545
1546        // Generate the semantic predicate map for debugging
1547        if (grammar.debuggingOutput)
1548            genSemPredMap(grammar.getClassName() + "::");
1549
1550        // Close class definition
1551        println("");
1552        println("");
1553        if (nameSpace != null)
1554            nameSpace.emitClosures(currentOutput);
1555
1556        // Close the parser output stream
1557        currentOutput.close();
1558        currentOutput = null;
1559    }
1560    public void genBody(TreeWalkerGrammar g) throws IOException  
1561    {
1562        // Open the output stream for the parser and set the currentOutput
1563        outputFile = grammar.getClassName() + ".cpp";
1564        outputLine = 1;
1565        currentOutput = antlrTool.openOutputFile(outputFile);
1566        //SAS: changed for proper text file io
1567
1568        genAST = grammar.buildAST;
1569        tabs = 0;
1570
1571        // Generate the header common to all output files.
1572        genHeader(outputFile);
1573
1574        printHeaderAction(preIncludeCpp);
1575
1576        // Generate header for the parser
1577        println("#include \"" + grammar.getClassName() + ".hpp\"");
1578        println("#include <antlr/Token.hpp>");
1579        println("#include <antlr/AST.hpp>");
1580        println("#include <antlr/NoViableAltException.hpp>");
1581        println("#include <antlr/MismatchedTokenException.hpp>");
1582        println("#include <antlr/SemanticException.hpp>");
1583        println("#include <antlr/BitSet.hpp>");
1584
1585        printHeaderAction(postIncludeCpp);
1586
1587        if (nameSpace != null)
1588            nameSpace.emitDeclarations(currentOutput);
1589
1590        // Output the user-defined parser premamble
1591        printAction(grammar.preambleAction);
1592
1593        // Generate parser class definition
1594        String   sup = null;
1595        if ( grammar.superClass!=null ) {
1596            sup = grammar.superClass;
1597        }
1598        else {
1599            sup = grammar.getSuperClass();
1600            if (sup.lastIndexOf('.') != -1)
1601                sup = sup.substring(sup.lastIndexOf('.')+1);
1602            sup = namespaceAntlr + sup;
1603        }
1604
1605        // Generate default parser class constructor
1606        println(grammar.getClassName() + "::" + grammar.getClassName() + "()");
1607        println("\t: "+namespaceAntlr+"TreeParser() {");
1608        tabs++;
1609//      println("setTokenNames(_tokenNames);");
1610        tabs--;
1611        println("}");
1612        println("");
1613
1614        // Generate code for each rule in the grammar
1615        Enumeration   ids = grammar.rules.elements();
1616        int ruleNum=0;
1617        String   ruleNameInits = "";
1618        while ( ids.hasMoreElements() ) {
1619            GrammarSymbol sym = (GrammarSymbol) ids.nextElement();
1620            if ( sym instanceof RuleSymbol) {
1621                RuleSymbol rs = (RuleSymbol)sym;
1622                genRule(rs, rs.references.size()==0, ruleNum++, grammar.getClassName() + "::");
1623            }
1624            exitIfError();
1625        }
1626
1627        if ( usingCustomAST )
1628        {
1629            // when we are using a custom ast override Parser::getAST to return the
1630            // custom AST type
1631            println(labeledElementASTType+" "+grammar.getClassName()+"::getAST()");
1632            println("{");
1633            println("\treturn returnAST;");
1634            println("}");
1635            println("");
1636        }
1637        // Generate the token names
1638        genTokenStrings(grammar.getClassName() + "::");
1639
1640        // Generate the bitsets used throughout the grammar
1641        genBitsets(bitsetsUsed, grammar.tokenManager.maxTokenType(), grammar.getClassName() + "::" );
1642
1643        // Close class definition
1644        println("");
1645        println("");
1646
1647        if (nameSpace != null)
1648            nameSpace.emitClosures(currentOutput);
1649
1650        // Close the parser output stream
1651        currentOutput.close();
1652        currentOutput = null;
1653    }
1654    /** Generate a series of case statements that implement a BitSet test.
1655     * @param p The Bitset for which cases are to be generated
1656     */
1657    protected void genCases(BitSet p) {
1658        if ( DEBUG_CODE_GENERATOR ) System.out.println("genCases("+p+")");
1659        int[] elems;
1660
1661        elems = p.toArray();
1662        // Wrap cases four-per-line for lexer, one-per-line for parser
1663        int wrap = 1; //(grammar instanceof LexerGrammar) ? 4 : 1;
1664        int j=1;
1665        boolean startOfLine = true;
1666        for (int i = 0; i < elems.length; i++) {
1667            if (j==1) {
1668                print("");
1669            } else {
1670                _print("  ");
1671            }
1672            _print("case " + getValueString(elems[i]) + ":");
1673
1674            if (j==wrap) {
1675                _println("");
1676                startOfLine = true;
1677                j=1;
1678            }
1679            else {
1680                j++;
1681                startOfLine = false;
1682            }
1683        }
1684        if (!startOfLine) {
1685            _println("");
1686        }
1687    }
1688    /** Generate common code for a block of alternatives; return a postscript
1689     * that needs to be generated at the end of the block.  Other routines
1690     * may append else-clauses and such for error checking before the postfix
1691     * is generated.
1692     * If the grammar is a lexer, then generate alternatives in an order where
1693     * alternatives requiring deeper lookahead are generated first, and
1694     * EOF in the lookahead set reduces the depth of the lookahead.
1695     * @param blk The block to generate
1696     * @param noTestForSingle If true, then it does not generate a test for a single alternative.
1697     */
1698    public CppBlockFinishingInfo genCommonBlock(
1699        AlternativeBlock blk,
1700        boolean noTestForSingle )
1701    {
1702        int nIF=0;
1703        boolean createdLL1Switch = false;
1704        int closingBracesOfIFSequence = 0;
1705        CppBlockFinishingInfo finishingInfo = new CppBlockFinishingInfo();
1706        if ( DEBUG_CODE_GENERATOR ) System.out.println("genCommonBlk("+blk+")");
1707
1708        // Save the AST generation state, and set it to that of the block
1709        boolean savegenAST = genAST;
1710        genAST = genAST && blk.getAutoGen();
1711
1712        boolean oldsaveTest = saveText;
1713        saveText = saveText && blk.getAutoGen();
1714
1715        // Is this block inverted?  If so, generate special-case code
1716        if ( blk.not &&
1717            analyzer.subruleCanBeInverted(blk, grammar instanceof LexerGrammar) )
1718        {
1719            Lookahead p = analyzer.look(1, blk);
1720            // Variable assignment for labeled elements
1721            if (blk.getLabel() != null && syntacticPredLevel == 0) {
1722                println(blk.getLabel() + " = " + lt1Value + ";");
1723            }
1724
1725            // AST
1726            genElementAST(blk);
1727
1728            String   astArgs="";
1729            if (grammar instanceof TreeWalkerGrammar) {
1730                if( usingCustomAST )
1731                    astArgs="static_cast<"+namespaceAntlr+"RefAST"+">(_t),";
1732                else
1733                    astArgs="_t,";
1734            }
1735
1736            // match the bitset for the alternative
1737            println("match(" + astArgs + getBitsetName(markBitsetForGen(p.fset)) + ");");
1738
1739            // tack on tree cursor motion if doing a tree walker
1740            if (grammar instanceof TreeWalkerGrammar)
1741            {
1742                println("_t = _t->getNextSibling();");
1743            }
1744            return finishingInfo;
1745        }
1746
1747        // Special handling for single alt
1748        if (blk.getAlternatives().size() == 1)
1749        {
1750            Alternative alt = blk.getAlternativeAt(0);
1751            // Generate a warning if there is a synPred for single alt.
1752            if (alt.synPred != null)
1753            {
1754                antlrTool.warning(
1755                                 "Syntactic predicate superfluous for single alternative",
1756                                 grammar.getFilename(),
1757                                 blk.getAlternativeAt(0).synPred.getLine(),
1758                                 blk.getAlternativeAt(0).synPred.getColumn()
1759                );
1760            }
1761            if (noTestForSingle)
1762            {
1763                if (alt.semPred != null)
1764                {
1765                    // Generate validating predicate
1766                    genSemPred(alt.semPred, blk.line);
1767                }
1768                genAlt(alt, blk);
1769                return finishingInfo;
1770            }
1771        }
1772
1773        // count number of simple LL(1) cases; only do switch for
1774        // many LL(1) cases (no preds, no end of token refs)
1775        // We don't care about exit paths for (...)*, (...)+
1776        // because we don't explicitly have a test for them
1777        // as an alt in the loop.
1778        //
1779        // Also, we now count how many unicode lookahead sets
1780        // there are--they must be moved to DEFAULT or ELSE
1781        // clause.
1782
1783        int nLL1 = 0;
1784        for (int i=0; i<blk.getAlternatives().size(); i++)
1785        {
1786            Alternative a = blk.getAlternativeAt(i);
1787            if ( suitableForCaseExpression(a) )
1788                nLL1++;
1789        }
1790
1791        // do LL(1) cases
1792        if ( nLL1 >= makeSwitchThreshold )
1793        {
1794            // Determine the name of the item to be compared
1795            String   testExpr = lookaheadString(1);
1796            createdLL1Switch = true;
1797            // when parsing trees, convert null to valid tree node with NULL lookahead
1798            if ( grammar instanceof TreeWalkerGrammar )
1799            {
1800                println("if (_t == "+labeledElementASTInit+" )");
1801                tabs++;
1802                println("_t = ASTNULL;");
1803                tabs--;
1804            }
1805            println("switch ( "+testExpr+") {");
1806            for (int i=0; i<blk.alternatives.size(); i++)
1807            {
1808                Alternative alt = blk.getAlternativeAt(i);
1809                // ignore any non-LL(1) alts, predicated alts or end-of-token alts
1810                // or end-of-token alts for case expressions
1811                if ( !suitableForCaseExpression(alt) )
1812                {
1813                    continue;
1814                }
1815                Lookahead p = alt.cache[1];
1816                if (p.fset.degree() == 0 && !p.containsEpsilon())
1817                {
1818                    antlrTool.warning("Alternate omitted due to empty prediction set",
1819                        grammar.getFilename(),
1820                        alt.head.getLine(), alt.head.getColumn());
1821                }
1822                else
1823                {
1824                    genCases(p.fset);
1825                    println("{");
1826                    tabs++;
1827                    genAlt(alt, blk);
1828                    println("break;");
1829                    tabs--;
1830                    println("}");
1831                }
1832            }
1833            println("default:");
1834            tabs++;
1835        }
1836
1837        // do non-LL(1) and nondeterministic cases
1838        // This is tricky in the lexer, because of cases like:
1839        //     STAR : '*' ;
1840        //     ASSIGN_STAR : "*=";
1841        // Since nextToken is generated without a loop, then the STAR will
1842        // have end-of-token as it's lookahead set for LA(2).  So, we must generate the
1843        // alternatives containing trailing end-of-token in their lookahead sets *after*
1844        // the alternatives without end-of-token.  This implements the usual
1845        // lexer convention that longer matches come before shorter ones, e.g.
1846        // "*=" matches ASSIGN_STAR not STAR
1847        //
1848        // For non-lexer grammars, this does not sort the alternates by depth
1849        // Note that alts whose lookahead is purely end-of-token at k=1 end up
1850        // as default or else clauses.
1851        int startDepth = (grammar instanceof LexerGrammar) ? grammar.maxk : 0;
1852        for (int altDepth = startDepth; altDepth >= 0; altDepth--) {
1853            if ( DEBUG_CODE_GENERATOR ) System.out.println("checking depth "+altDepth);
1854            for (int i=0; i<blk.alternatives.size(); i++) {
1855                Alternative alt = blk.getAlternativeAt(i);
1856                if ( DEBUG_CODE_GENERATOR ) System.out.println("genAlt: "+i);
1857                // if we made a switch above, ignore what we already took care
1858                // of.  Specifically, LL(1) alts with no preds
1859                // that do not have end-of-token in their prediction set
1860                if ( createdLL1Switch &&
1861                     suitableForCaseExpression(alt) )
1862                {
1863                    if ( DEBUG_CODE_GENERATOR )
1864                        System.out.println("ignoring alt because it was in the switch");
1865                    continue;
1866                }
1867                String   e;
1868
1869                boolean unpredicted = false;
1870
1871                if (grammar instanceof LexerGrammar) {
1872                    // Calculate the "effective depth" of the alt, which is the max
1873                    // depth at which cache[depth]!=end-of-token
1874                    int effectiveDepth = alt.lookaheadDepth;
1875                    if (effectiveDepth == GrammarAnalyzer.NONDETERMINISTIC)
1876                    {
1877                        // use maximum lookahead
1878                        effectiveDepth = grammar.maxk;
1879                    }
1880                    while ( effectiveDepth >= 1 &&
1881                             alt.cache[effectiveDepth].containsEpsilon() )
1882                    {
1883                        effectiveDepth--;
1884                    }
1885                    // Ignore alts whose effective depth is other than the ones we
1886                    // are generating for this iteration.
1887                    if (effectiveDepth != altDepth)
1888                    {
1889                        if ( DEBUG_CODE_GENERATOR )
1890                            System.out.println("ignoring alt because effectiveDepth!=altDepth;"+effectiveDepth+"!="+altDepth);
1891                        continue;
1892                    }
1893                    unpredicted = lookaheadIsEmpty(alt, effectiveDepth);
1894                    e = getLookaheadTestExpression(alt, effectiveDepth);
1895                }
1896                else
1897                {
1898                    unpredicted = lookaheadIsEmpty(alt, grammar.maxk);
1899                    e = getLookaheadTestExpression(alt, grammar.maxk);
1900                }
1901
1902                // Was it a big unicode range that forced unsuitability
1903                // for a case expression?
1904                if ( alt.cache[1].fset.degree() > caseSizeThreshold &&
1905                      suitableForCaseExpression(alt))
1906                {
1907                    if ( nIF==0 )
1908                    {
1909                        // generate this only for the first if the elseif's
1910                        // are covered by this one
1911                        if ( grammar instanceof TreeWalkerGrammar ) {
1912                            println("if (_t == "+labeledElementASTInit+" )");
1913                            tabs++;
1914                            println("_t = ASTNULL;");
1915                            tabs--;
1916                        }
1917                        println("if " + e + " {");
1918                    }
1919                    else
1920                        println("else if " + e + " {");
1921                }
1922                else if (unpredicted &&
1923                            alt.semPred==null &&
1924                            alt.synPred==null)
1925                {
1926                    // The alt has empty prediction set and no
1927                    // predicate to help out.  if we have not
1928                    // generated a previous if, just put {...} around
1929                    // the end-of-token clause
1930                    if ( nIF==0 ) {
1931                        println("{");
1932                    }
1933                    else {
1934                        println("else {");
1935                    }
1936                    finishingInfo.needAnErrorClause = false;
1937                }
1938                else
1939                {
1940                    // check for sem and syn preds
1941                    // Add any semantic predicate expression to the lookahead test
1942                    if ( alt.semPred != null ) {
1943                        // if debugging, wrap the evaluation of the predicate in a method
1944                        //
1945                        // translate $ and # references
1946                        ActionTransInfo tInfo = new ActionTransInfo();
1947                        String   actionStr = processActionForTreeSpecifiers(alt.semPred,
1948                                                                          blk.line,
1949                                                                          currentRule,
1950                                                                          tInfo);
1951                        // ignore translation info...we don't need to do anything with it.
1952
1953                        // call that will inform SemanticPredicateListeners of the result
1954                        if (((grammar instanceof ParserGrammar) || (grammar instanceof LexerGrammar)) && grammar.debuggingOutput)
1955                            e = "("+e+"&& fireSemanticPredicateEvaluated(antlr.debug.SemanticPredicateEvent.PREDICTING,"+ //FIXME
1956                                    addSemPred(charFormatter.escapeString(actionStr))+","+actionStr+"))";
1957                        else
1958                            e = "("+e+"&&("+actionStr +"))";
1959                    }
1960
1961                    // Generate any syntactic predicates
1962                    if ( nIF>0 ) {
1963                        if ( alt.synPred != null ) {
1964                            println("else {");
1965                            tabs++;
1966                            genSynPred( alt.synPred, e );
1967                            closingBracesOfIFSequence++;
1968                        }
1969                        else {
1970                            println("else if " + e + " {");
1971                        }
1972                    }
1973                    else {
1974                        if ( alt.synPred != null ) {
1975                            genSynPred( alt.synPred, e );
1976                        }
1977                        else {
1978                            // when parsing trees, convert null to valid tree node
1979                            // with NULL lookahead.
1980                            if ( grammar instanceof TreeWalkerGrammar ) {
1981                                println("if (_t == "+labeledElementASTInit+" )");
1982                                tabs++;
1983                                println("_t = ASTNULL;");
1984                                tabs--;
1985                            }
1986                            println("if " + e + " {");
1987                        }
1988                    }
1989
1990                }
1991
1992                nIF++;
1993                tabs++;
1994                genAlt(alt, blk);
1995                tabs--;
1996                println("}");
1997            }
1998        }
1999        String   ps = "";
2000        for (int i=1; i<=closingBracesOfIFSequence; i++) {
2001            tabs--; // does JavaCodeGenerator need this?
2002            ps+="}";
2003        }
2004
2005        // Restore the AST generation state
2006        genAST = savegenAST;
2007
2008        // restore save text state
2009        saveText=oldsaveTest;
2010
2011        // Return the finishing info.
2012        if ( createdLL1Switch ) {
2013            tabs--;
2014            finishingInfo.postscript = ps+"}";
2015            finishingInfo.generatedSwitch = true;
2016            finishingInfo.generatedAnIf = nIF>0;
2017            //return new CppBlockFinishingInfo(ps+"}",true,nIF>0); // close up switch statement
2018
2019        }
2020        else {
2021            finishingInfo.postscript = ps;
2022            finishingInfo.generatedSwitch = false;
2023            finishingInfo.generatedAnIf = nIF>0;
2024            //return new CppBlockFinishingInfo(ps, false,nIF>0);
2025        }
2026        return finishingInfo;
2027    }
2028
2029    private static boolean suitableForCaseExpression(Alternative a) {
2030        return a.lookaheadDepth == 1 &&
2031            a.semPred == null &&
2032            !a.cache[1].containsEpsilon() &&
2033            a.cache[1].fset.degree()<=caseSizeThreshold;
2034    }
2035
2036    /** Generate code to link an element reference into the AST
2037     */
2038    private void genElementAST(AlternativeElement el) {
2039
2040        // handle case where you're not building trees, but are in tree walker.
2041        // Just need to get labels set up.
2042        if ( grammar instanceof TreeWalkerGrammar && !grammar.buildAST )
2043        {
2044            String   elementRef;
2045            String   astName;
2046
2047            // Generate names and declarations of the AST variable(s)
2048            if (el.getLabel() == null)
2049            {
2050                elementRef = lt1Value;
2051                // Generate AST variables for unlabeled stuff
2052                astName = "tmp" + astVarNumber + "_AST";
2053                astVarNumber++;
2054                // Map the generated AST variable in the alternate
2055                mapTreeVariable(el, astName);
2056                // Generate an "input" AST variable also
2057                println(labeledElementASTType+" "+astName+"_in = "+elementRef+";");
2058            }
2059            return;
2060        }
2061
2062        if (grammar.buildAST && syntacticPredLevel == 0)
2063        {
2064            boolean needASTDecl =
2065                ( genAST && (el.getLabel() != null ||
2066                  el.getAutoGenType() != GrammarElement.AUTO_GEN_BANG ));
2067
2068            // RK: if we have a grammar element always generate the decl
2069            // since some guy can access it from an action and we can't
2070            // peek ahead (well not without making a mess).
2071            // I'd prefer taking this out.
2072            if( el.getAutoGenType() != GrammarElement.AUTO_GEN_BANG &&
2073                 (el instanceof TokenRefElement) )
2074                needASTDecl = true;
2075
2076            boolean doNoGuessTest =
2077                ( grammar.hasSyntacticPredicate && needASTDecl );
2078
2079            String   elementRef;
2080            String   astNameBase;
2081
2082            // Generate names and declarations of the AST variable(s)
2083            if (el.getLabel() != null)
2084            {
2085                // if the element is labeled use that name...
2086                elementRef = el.getLabel();
2087                astNameBase = el.getLabel();
2088            }
2089            else
2090            {
2091                // else generate a temporary name...
2092                elementRef = lt1Value;
2093                // Generate AST variables for unlabeled stuff
2094                astNameBase = "tmp" + astVarNumber;
2095                astVarNumber++;
2096            }
2097
2098            // Generate the declaration if required.
2099            if ( needASTDecl )
2100            {
2101                if ( el instanceof GrammarAtom )
2102                {
2103                    GrammarAtom ga = (GrammarAtom)el;
2104                    if ( ga.getASTNodeType()!=null )
2105                    {
2106                        genASTDeclaration( el, astNameBase, "Ref"+ga.getASTNodeType() );
2107//                      println("Ref"+ga.getASTNodeType()+" " + astName + ";");
2108                    }
2109                    else
2110                    {
2111                        genASTDeclaration( el, astNameBase, labeledElementASTType );
2112//                      println(labeledElementASTType+" " + astName + " = "+labeledElementASTInit+";");
2113                    }
2114                }
2115                else
2116                {
2117                    genASTDeclaration( el, astNameBase, labeledElementASTType );
2118//                  println(labeledElementASTType+" " + astName + " = "+labeledElementASTInit+";");
2119                }
2120            }
2121
2122            // for convenience..
2123            String   astName = astNameBase + "_AST";
2124
2125            // Map the generated AST variable in the alternate
2126            mapTreeVariable(el, astName);
2127            if (grammar instanceof TreeWalkerGrammar)
2128            {
2129                // Generate an "input" AST variable also
2130                println(labeledElementASTType+" " + astName + "_in = "+labeledElementASTInit+";");
2131            }
2132
2133            // Enclose actions with !guessing
2134            if (doNoGuessTest) {
2135                println("if ( inputState->guessing == 0 ) {");
2136                tabs++;
2137            }
2138
2139            // if something has a label assume it will be used
2140            // so we must initialize the RefAST
2141            if (el.getLabel() != null)
2142            {
2143                if ( el instanceof GrammarAtom )
2144                {
2145                    println(astName + " = "+
2146                              getASTCreateString((GrammarAtom)el,elementRef) + ";");
2147                }
2148                else
2149                {
2150                    println(astName + " = "+
2151                              getASTCreateString(elementRef) + ";");
2152                }
2153            }
2154
2155            // if it has no label but a declaration exists initialize it.
2156            if( el.getLabel() == null && needASTDecl )
2157            {
2158                elementRef = lt1Value;
2159                if ( el instanceof GrammarAtom )
2160                {
2161                    println(astName + " = "+
2162                              getASTCreateString((GrammarAtom)el,elementRef) + ";");
2163                }
2164                else
2165                {
2166                    println(astName + " = "+
2167                              getASTCreateString(elementRef) + ";");
2168                }
2169                // Map the generated AST variable in the alternate
2170                if (grammar instanceof TreeWalkerGrammar)
2171                {
2172                    // set "input" AST variable also
2173                    println(astName + "_in = " + elementRef + ";");
2174                }
2175            }
2176
2177            if (genAST)
2178            {
2179                switch (el.getAutoGenType())
2180                {
2181                case GrammarElement.AUTO_GEN_NONE:
2182                    if( usingCustomAST ||
2183                         (el instanceof GrammarAtom &&
2184                          ((GrammarAtom)el).getASTNodeType() != null) )
2185                        println("astFactory->addASTChild(currentAST, static_cast<"+namespaceAntlr+"RefAST>("+ astName + "));");
2186                    else
2187                        println("astFactory->addASTChild(currentAST, "+ astName + ");");
2188                    //                      println("astFactory.addASTChild(currentAST, "+namespaceAntlr+"RefAST(" + astName + "));");
2189                    break;
2190                case GrammarElement.AUTO_GEN_CARET:
2191                    if( usingCustomAST ||
2192                         (el instanceof GrammarAtom &&
2193                         ((GrammarAtom)el).getASTNodeType() != null) )
2194                        println("astFactory->makeASTRoot(currentAST, static_cast<"+namespaceAntlr+"RefAST>(" + astName + "));");
2195                    else
2196                        println("astFactory->makeASTRoot(currentAST, " + astName + ");");
2197                    break;
2198                default:
2199                    break;
2200                }
2201            }
2202            if (doNoGuessTest)
2203            {
2204                tabs--;
2205                println("}");
2206            }
2207        }
2208    }
2209    /** Close the try block and generate catch phrases
2210     * if the element has a labeled handler in the rule
2211     */
2212    private void genErrorCatchForElement(AlternativeElement el) {
2213        if (el.getLabel() == null) return;
2214        String   r = el.enclosingRuleName;
2215        if ( grammar instanceof LexerGrammar ) {
2216            r = CodeGenerator.encodeLexerRuleName(el.enclosingRuleName);
2217        }
2218        RuleSymbol rs = (RuleSymbol)grammar.getSymbol(r);
2219        if (rs == null) {
2220            antlrTool.panic("Enclosing rule not found!");
2221        }
2222        ExceptionSpec ex = rs.block.findExceptionSpec(el.getLabel());
2223        if (ex != null) {
2224            tabs--;
2225            println("}");
2226            genErrorHandler(ex);
2227        }
2228    }
2229    /** Generate the catch phrases for a user-specified error handler */
2230    private void genErrorHandler(ExceptionSpec ex)
2231    {
2232        // Each ExceptionHandler in the ExceptionSpec is a separate catch
2233        for (int i = 0; i < ex.handlers.size(); i++)
2234        {
2235            ExceptionHandler handler = (ExceptionHandler)ex.handlers.elementAt(i);
2236            // Generate catch phrase
2237            println("catch (" + handler.exceptionTypeAndName.getText() + ") {");
2238            tabs++;
2239            if (grammar.hasSyntacticPredicate) {
2240                println("if (inputState->guessing==0) {");
2241                tabs++;
2242            }
2243
2244            // get the name of the followSet for the current rule so that we
2245            // can replace $lookaheadSet in the .g file.
2246            ActionTransInfo tInfo=null;
2247            if (currentRule != null && (grammar instanceof ParserGrammar) ) // only supply follow set if we're parsing
2248            {
2249                tInfo = new ActionTransInfo();
2250                Lookahead follow = grammar.theLLkAnalyzer.FOLLOW(1, currentRule.endNode);
2251                String   followSetName = getBitsetName(markBitsetForGen(follow.fset));
2252                tInfo.lookaheadSetName = followSetName;
2253            }
2254
2255            // When not guessing, execute user handler action
2256            genLineNo(handler.action);
2257            printAction(
2258                processActionForTreeSpecifiers( handler.action.getText(),
2259                                                         handler.action.getLine(),
2260                                                         currentRule, tInfo )
2261            );
2262            genLineNo2();
2263
2264            if (grammar.hasSyntacticPredicate)
2265            {
2266                tabs--;
2267                println("} else {");
2268                tabs++;
2269                // When guessing, rethrow exception
2270                println("throw "+extractIdOfAction(handler.exceptionTypeAndName)+";");
2271                tabs--;
2272                println("}");
2273            }
2274            // Close catch phrase
2275            tabs--;
2276            println("}");
2277        }
2278    }
2279    /** Generate a try { opening if the element has a labeled handler in the rule */
2280    private void genErrorTryForElement(AlternativeElement el) {
2281        if (el.getLabel() == null) return;
2282        String   r = el.enclosingRuleName;
2283        if ( grammar instanceof LexerGrammar ) {
2284            r = CodeGenerator.encodeLexerRuleName(el.enclosingRuleName);
2285        }
2286        RuleSymbol rs = (RuleSymbol)grammar.getSymbol(r);
2287        if (rs == null) {
2288            antlrTool.panic("Enclosing rule not found!");
2289        }
2290        ExceptionSpec ex = rs.block.findExceptionSpec(el.getLabel());
2291        if (ex != null) {
2292            println("try { // for error handling");
2293            tabs++;
2294        }
2295    }
2296    /** Generate a header that is common to all C++ files */
2297    protected void genHeader(String   fileName)
2298    {
2299        println("/* $ANTLR "+antlrTool.version+": "+
2300                "\""+antlrTool.fileMinusPath(antlrTool.grammarFile)+"\""+
2301                " -> "+
2302                "\""+fileName+"\"$ */");
2303    }
2304
2305    // these are unique to C++ mode
2306    public void genInclude(LexerGrammar g) throws IOException  
2307    {
2308        outputFile = grammar.getClassName() + ".hpp";
2309        outputLine = 1;
2310        currentOutput = antlrTool.openOutputFile(outputFile);
2311        //SAS: changed for proper text file io
2312
2313        genAST = false; // no way to gen trees.
2314        saveText = true;    // save consumed characters.
2315
2316        tabs=0;
2317
2318        // Generate a guard wrapper
2319        println("#ifndef INC_"+grammar.getClassName()+"_hpp_");
2320        println("#define INC_"+grammar.getClassName()+"_hpp_");
2321        println("");
2322
2323        printHeaderAction(preIncludeHpp);
2324
2325        println("#include <antlr/config.hpp>");
2326
2327        // Generate header common to all C++ output files
2328        genHeader(outputFile);
2329
2330        // Generate header specific to lexer header file
2331        println("#include <antlr/CommonToken.hpp>");
2332        println("#include <antlr/InputBuffer.hpp>");
2333        println("#include <antlr/BitSet.hpp>");
2334        println("#include \"" + grammar.tokenManager.getName() + TokenTypesFileSuffix+".hpp\"");
2335
2336        // Find the name of the super class
2337        String   sup=null;
2338        if ( grammar.superClass!=null ) {
2339            sup = grammar.superClass;
2340            println("#include \""+sup+".hpp\"");
2341        }
2342        else {
2343            sup = grammar.getSuperClass();
2344            if (sup.lastIndexOf('.') != -1)
2345                sup = sup.substring(sup.lastIndexOf('.')+1);
2346            println("#include <antlr/"+sup+".hpp>");
2347            sup = namespaceAntlr + sup;
2348        }
2349
2350        // Do not use printAction because we assume tabs==0
2351        printHeaderAction(postIncludeHpp);
2352
2353        if (nameSpace != null)
2354               nameSpace.emitDeclarations(currentOutput);
2355
2356        printHeaderAction("");
2357
2358        // print javadoc comment if any
2359        if ( grammar.comment!=null ) {
2360            _println(grammar.comment);
2361        }
2362
2363        // Generate lexer class definition
2364        print("class " + grammar.getClassName() + " : public " + sup);
2365        println(", public " + grammar.tokenManager.getName() + TokenTypesFileSuffix);
2366
2367        Token tsuffix = (Token)grammar.options.get("classHeaderSuffix");
2368        if ( tsuffix != null ) {
2369            String   suffix = StringUtils.stripFrontBack(tsuffix.getText(),"\"","\"");
2370            if ( suffix != null ) {
2371                print(", "+suffix);  // must be an interface name for Java
2372            }
2373        }
2374        println("{");
2375
2376        // Generate user-defined lexer class members
2377        if (grammar.classMemberAction != null) {
2378            genLineNo(grammar.classMemberAction);
2379            print(
2380                processActionForTreeSpecifiers(grammar.classMemberAction.getText(),
2381                                                         grammar.classMemberAction.getLine(),
2382                                                         currentRule, null)
2383            );
2384            genLineNo2();
2385        }
2386
2387        // Generate initLiterals() method
2388        tabs=0;
2389        println("private:");
2390        tabs=1;
2391        println("void initLiterals();");
2392
2393        // Generate getCaseSensitiveLiterals() method
2394        tabs=0;
2395        println("public:");
2396        tabs=1;
2397        println("bool getCaseSensitiveLiterals() const;");
2398
2399        // Make constructors public
2400        tabs=0;
2401        println("public:");
2402        tabs=1;
2403
2404        // Generate the constructor from std::istream
2405        println(grammar.getClassName() + "(" + namespaceStd + "istream& in);");
2406
2407        // Generate the constructor from InputBuffer
2408        println(grammar.getClassName() + "("+namespaceAntlr+"InputBuffer& ib);");
2409
2410        println(grammar.getClassName() + "(const "+namespaceAntlr+"LexerSharedInputState& state);");
2411
2412        // Generate nextToken() rule.
2413        // nextToken() is a synthetic lexer rule that is the implicit OR of all
2414        // user-defined lexer rules.
2415        println(namespaceAntlr+"RefToken nextToken();");
2416
2417        // Generate code for each rule in the lexer
2418        Enumeration   ids = grammar.rules.elements();
2419        while ( ids.hasMoreElements() ) {
2420            RuleSymbol sym = (RuleSymbol) ids.nextElement();
2421            // Don't generate the synthetic rules
2422            if (!sym.getId().equals("mnextToken")) {
2423                genRuleHeader(sym, false);
2424            }
2425            exitIfError();
2426        }
2427
2428        // Make the rest private
2429        tabs=0;
2430        println("private:");
2431        tabs=1;
2432
2433        // generate the rule name array for debugging
2434        if ( grammar.debuggingOutput ) {
2435            println("static const char* _ruleNames[];");
2436        }
2437
2438        // Generate the semantic predicate map for debugging
2439        if (grammar.debuggingOutput)
2440            println("static const char* _semPredNames[];");
2441
2442        // Generate the bitsets used throughout the lexer
2443        genBitsetsHeader(bitsetsUsed, ((LexerGrammar)grammar).charVocabulary.size());
2444
2445        tabs=0;
2446        println("};");
2447        println("");
2448        if (nameSpace != null)
2449            nameSpace.emitClosures(currentOutput);
2450
2451        // Generate a guard wrapper
2452        println("#endif /*INC_"+grammar.getClassName()+"_hpp_*/");
2453
2454        // Close the lexer output stream
2455        currentOutput.close();
2456        currentOutput = null;
2457    }
2458    public void genInclude(ParserGrammar g) throws IOException  
2459    {
2460        // Open the output stream for the parser and set the currentOutput
2461        outputFile = grammar.getClassName() + ".hpp";
2462        outputLine = 1;
2463        currentOutput = antlrTool.openOutputFile(outputFile);
2464        //SAS: changed for proper text file io
2465
2466        genAST = grammar.buildAST;
2467
2468        tabs = 0;
2469
2470        // Generate a guard wrapper
2471        println("#ifndef INC_"+grammar.getClassName()+"_hpp_");
2472        println("#define INC_"+grammar.getClassName()+"_hpp_");
2473        println("");
2474        printHeaderAction(preIncludeHpp);
2475        println("#include <antlr/config.hpp>");
2476
2477        // Generate the header common to all output files.
2478        genHeader(outputFile);
2479
2480        // Generate header for the parser
2481        println("#include <antlr/TokenStream.hpp>");
2482        println("#include <antlr/TokenBuffer.hpp>");
2483        println("#include \"" + grammar.tokenManager.getName() + TokenTypesFileSuffix+".hpp\"");
2484
2485        // Generate parser class definition
2486        String   sup=null;
2487        if ( grammar.superClass!=null ) {
2488            sup = grammar.superClass;
2489            println("#include \""+sup+".hpp\"");
2490        }
2491        else {
2492            sup = grammar.getSuperClass();
2493            if (sup.lastIndexOf('.') != -1)
2494                sup = sup.substring(sup.lastIndexOf('.')+1);
2495            println("#include <antlr/"+sup+".hpp>");
2496            sup = namespaceAntlr + sup;
2497        }
2498        println("");
2499
2500        // Do not use printAction because we assume tabs==0
2501        printHeaderAction(postIncludeHpp);
2502
2503        if (nameSpace != null)
2504            nameSpace.emitDeclarations(currentOutput);
2505
2506        printHeaderAction("");
2507
2508        // print javadoc comment if any
2509        if ( grammar.comment!=null ) {
2510            _println(grammar.comment);
2511        }
2512
2513        // generate the actual class definition
2514        print("class " + grammar.getClassName() + " : public " + sup);
2515        println(", public " + grammar.tokenManager.getName() + TokenTypesFileSuffix);
2516
2517        Token tsuffix = (Token)grammar.options.get("classHeaderSuffix");
2518        if ( tsuffix != null ) {
2519            String   suffix = StringUtils.stripFrontBack(tsuffix.getText(),"\"","\"");
2520            if ( suffix != null )
2521                print(", "+suffix);  // must be an interface name for Java
2522        }
2523        println("{");
2524
2525        // set up an array of all the rule names so the debugger can
2526        // keep track of them only by number -- less to store in tree...
2527        if (grammar.debuggingOutput) {
2528            println("public: static const char* _ruleNames[];");
2529        }
2530        // Generate user-defined parser class members
2531        if (grammar.classMemberAction != null) {
2532            genLineNo(grammar.classMemberAction.getLine());
2533            print(
2534                processActionForTreeSpecifiers(grammar.classMemberAction.getText(),
2535                                                         grammar.classMemberAction.getLine(),
2536                                                         currentRule, null)
2537            );
2538            genLineNo2();
2539        }
2540        println("public:");
2541        tabs = 1;
2542        println("void initializeFactory( void );");
2543        println("// called from constructors");
2544        println("void _initialize( void );");
2545
2546        // Generate parser class constructor from TokenBuffer
2547        tabs=0;
2548        println("protected:");
2549        tabs=1;
2550        println(grammar.getClassName() + "("+namespaceAntlr+"TokenBuffer& tokenBuf, int k);");
2551        tabs=0;
2552        println("public:");
2553        tabs=1;
2554        println(grammar.getClassName() + "("+namespaceAntlr+"TokenBuffer& tokenBuf);");
2555
2556        // Generate parser class constructor from TokenStream
2557        tabs=0;
2558        println("protected:");
2559        tabs=1;
2560        println(grammar.getClassName()+"("+namespaceAntlr+"TokenStream& lexer, int k);");
2561        tabs=0;
2562        println("public:");
2563        tabs=1;
2564        println(grammar.getClassName()+"("+namespaceAntlr+"TokenStream& lexer);");
2565
2566        println(grammar.getClassName()+"(const "+namespaceAntlr+"ParserSharedInputState& state);");
2567
2568        println("int getNumTokens() const");
2569        println("{"); tabs++;
2570        println("return "+grammar.getClassName()+"::NUM_TOKENS;");
2571        tabs--; println("}");
2572        println("const char* getTokenName( int type ) const");
2573        println("{"); tabs++;
2574        println("if( type > getNumTokens() ) return 0;");
2575        println("return "+grammar.getClassName()+"::tokenNames[type];");
2576        tabs--; println("}");
2577        println("const char* const* getTokenNames() const");
2578        println("{"); tabs++;
2579        println("return "+grammar.getClassName()+"::tokenNames;");
2580        tabs--; println("}");
2581
2582        // Generate code for each rule in the grammar
2583        Enumeration   ids = grammar.rules.elements();
2584        while ( ids.hasMoreElements() ) {
2585            GrammarSymbol sym = (GrammarSymbol) ids.nextElement();
2586            if ( sym instanceof RuleSymbol) {
2587                RuleSymbol rs = (RuleSymbol)sym;
2588                genRuleHeader(rs, rs.references.size()==0);
2589            }
2590            exitIfError();
2591        }
2592
2593        if ( usingCustomAST )
2594        {
2595            // when we are using a custom ast override Parser::getAST to return the
2596            // custom AST type
2597            tabs=0; println("public:"); tabs=1;
2598            println(labeledElementASTType+" getAST();");
2599            println("");
2600            tabs=0; println("protected:"); tabs=1;
2601            println(labeledElementASTType+" returnAST;");
2602        }
2603        // Make the rest private
2604        tabs=0;
2605        println("private:");
2606        tabs=1;
2607
2608        // Generate the token names
2609        println("static const char* tokenNames[];");
2610        // and how many there are of them
2611        _println("#ifndef NO_STATIC_CONSTS");
2612        println("static const int NUM_TOKENS = "+grammar.tokenManager.getVocabulary().size()+";");
2613        _println("#else");
2614        println("enum {");
2615        println("\tNUM_TOKENS = "+grammar.tokenManager.getVocabulary().size());
2616        println("};");
2617        _println("#endif");
2618
2619        // Generate the bitsets used throughout the grammar
2620        genBitsetsHeader(bitsetsUsed, grammar.tokenManager.maxTokenType());
2621
2622        // Generate the semantic predicate map for debugging
2623        if (grammar.debuggingOutput)
2624            println("static const char* _semPredNames[];");
2625
2626        // Close class definition
2627        tabs=0;
2628        println("};");
2629        println("");
2630        if (nameSpace != null)
2631            nameSpace.emitClosures(currentOutput);
2632
2633        // Generate a guard wrapper
2634        println("#endif /*INC_"+grammar.getClassName()+"_hpp_*/");
2635
2636        // Close the parser output stream
2637        currentOutput.close();
2638        currentOutput = null;
2639    }
2640    public void genInclude(TreeWalkerGrammar g) throws IOException  
2641    {
2642        // Open the output stream for the parser and set the currentOutput
2643        outputFile = grammar.getClassName() + ".hpp";
2644        outputLine = 1;
2645        currentOutput = antlrTool.openOutputFile(outputFile);
2646        //SAS: changed for proper text file io
2647
2648        genAST = grammar.buildAST;
2649        tabs = 0;
2650
2651        // Generate a guard wrapper
2652        println("#ifndef INC_"+grammar.getClassName()+"_hpp_");
2653        println("#define INC_"+grammar.getClassName()+"_hpp_");
2654        println("");
2655        printHeaderAction(preIncludeHpp);
2656        println("#include <antlr/config.hpp>");
2657        println("#include \"" + grammar.tokenManager.getName() + TokenTypesFileSuffix+".hpp\"");
2658
2659        // Generate the header common to all output files.
2660        genHeader(outputFile);
2661
2662        // Find the name of the super class
2663        String   sup=null;
2664        if ( grammar.superClass!=null ) {
2665            sup = grammar.superClass;
2666            println("#include \""+sup+".hpp\"");
2667        }
2668        else {
2669            sup = grammar.getSuperClass();
2670            if (sup.lastIndexOf('.') != -1)
2671                sup = sup.substring(sup.lastIndexOf('.')+1);
2672            println("#include <antlr/"+sup+".hpp>");
2673            sup = namespaceAntlr + sup;
2674        }
2675        println("");
2676
2677        // Generate header for the parser
2678        //
2679        // Do not use printAction because we assume tabs==0
2680        printHeaderAction(postIncludeHpp);
2681
2682        if (nameSpace != null)
2683            nameSpace.emitDeclarations(currentOutput);
2684
2685        printHeaderAction("");
2686
2687        // print javadoc comment if any
2688        if ( grammar.comment!=null ) {
2689            _println(grammar.comment);
2690        }
2691
2692        // Generate parser class definition
2693        print("class " + grammar.getClassName() + " : public "+sup);
2694        println(", public " + grammar.tokenManager.getName() + TokenTypesFileSuffix);
2695
2696        Token tsuffix = (Token)grammar.options.get("classHeaderSuffix");
2697        if ( tsuffix != null ) {
2698            String   suffix = StringUtils.stripFrontBack(tsuffix.getText(),"\"","\"");
2699            if ( suffix != null ) {
2700                print(", "+suffix);  // must be an interface name for Java
2701            }
2702        }
2703        println("{");
2704
2705        // Generate user-defined parser class members
2706        if (grammar.classMemberAction != null) {
2707            genLineNo(grammar.classMemberAction.getLine());
2708            print(
2709                    processActionForTreeSpecifiers(grammar.classMemberAction.getText(),
2710                                                             grammar.classMemberAction.getLine(),
2711                                                             currentRule, null)
2712                    );
2713            genLineNo2();
2714        }
2715
2716        // Generate default parser class constructor
2717        tabs=0;
2718        println("public:");
2719        tabs=1;
2720        println(grammar.getClassName() + "();");
2721
2722        println("int getNumTokens() const");
2723        println("{"); tabs++;
2724        println("return "+grammar.getClassName()+"::NUM_TOKENS;");
2725        tabs--; println("}");
2726        println("const char* getTokenName( int type ) const");
2727        println("{"); tabs++;
2728        println("if( type > getNumTokens() ) return 0;");
2729        println("return "+grammar.getClassName()+"::tokenNames[type];");
2730        tabs--; println("}");
2731
2732        // Generate code for each rule in the grammar
2733        Enumeration   ids = grammar.rules.elements();
2734        String   ruleNameInits = "";
2735        while ( ids.hasMoreElements() ) {
2736            GrammarSymbol sym = (GrammarSymbol) ids.nextElement();
2737            if ( sym instanceof RuleSymbol) {
2738                RuleSymbol rs = (RuleSymbol)sym;
2739                genRuleHeader(rs, rs.references.size()==0);
2740            }
2741            exitIfError();
2742        }
2743        if ( usingCustomAST )
2744        {
2745            // when we are using a custom ast override TreeParser::getAST to return the
2746            // custom AST type
2747            tabs=0; println("public:"); tabs=1;
2748            println(labeledElementASTType+" getAST();");
2749            println("");
2750            tabs=0; println("protected:"); tabs=1;
2751            println(labeledElementASTType+" returnAST;");
2752            println(labeledElementASTType+" _retTree;");
2753        }
2754
2755        // Make the rest private
2756        tabs=0;
2757        println("private:");
2758        tabs=1;
2759
2760        // Generate the token names
2761        println("static const char* tokenNames[];");
2762        // and how many there are of them
2763        _println("#ifndef NO_STATIC_CONSTS");
2764        println("static const int NUM_TOKENS = "+grammar.tokenManager.getVocabulary().size()+";");
2765        _println("#else");
2766        println("enum {");
2767        println("\tNUM_TOKENS = "+grammar.tokenManager.getVocabulary().size());
2768        println("};");
2769        _println("#endif");
2770
2771        // Generate the bitsets used throughout the grammar
2772        genBitsetsHeader(bitsetsUsed, grammar.tokenManager.maxTokenType());
2773
2774        // Close class definition
2775        tabs=0;
2776        println("};");
2777        println("");
2778        if (nameSpace != null)
2779            nameSpace.emitClosures(currentOutput);
2780
2781        // Generate a guard wrapper
2782        println("#endif /*INC_"+grammar.getClassName()+"_hpp_*/");
2783
2784        // Close the parser output stream
2785        currentOutput.close();
2786        currentOutput = null;
2787    }
2788    /// for convenience
2789    protected void genASTDeclaration( AlternativeElement el ) {
2790        genASTDeclaration( el, labeledElementASTType );
2791    }
2792    /// for convenience
2793    protected void genASTDeclaration( AlternativeElement el, String   node_type ) {
2794        genASTDeclaration( el, el.getLabel(), node_type );
2795    }
2796    /// Generate (if not already done) a declaration for the AST for el.
2797    protected void genASTDeclaration( AlternativeElement el, String   var_name, String   node_type ) {
2798        // already declared?
2799        if( declaredASTVariables.contains(el) )
2800            return;
2801
2802        String   init = labeledElementASTInit;
2803
2804        if (el instanceof GrammarAtom &&
2805             ((GrammarAtom)el).getASTNodeType() != null )
2806            init = "static_cast<Ref"+((GrammarAtom)el).getASTNodeType()+">("+labeledElementASTInit+")";
2807
2808        // emit code
2809        println(node_type+" " + var_name + "_AST = "+init+";");
2810
2811        // mark as declared
2812        declaredASTVariables.add( el );
2813    }
2814    private void genLiteralsTest() {
2815        println("_ttype = testLiteralsTable(_ttype);");
2816    }
2817    private void genLiteralsTestForPartialToken() {
2818        println("_ttype = testLiteralsTable(text.substr(_begin, text.length()-_begin),_ttype);");
2819    }
2820    protected void genMatch(BitSet b) {
2821    }
2822    protected void genMatch(GrammarAtom atom) {
2823        if ( atom instanceof StringLiteralElement ) {
2824            if ( grammar instanceof LexerGrammar ) {
2825                genMatchUsingAtomText(atom);
2826            }
2827            else {
2828                genMatchUsingAtomTokenType(atom);
2829            }
2830        }
2831        else if ( atom instanceof CharLiteralElement ) {
2832            if ( grammar instanceof LexerGrammar ) {
2833                genMatchUsingAtomText(atom);
2834            }
2835            else {
2836                antlrTool.error("cannot ref character literals in grammar: "+atom);
2837            }
2838        }
2839        else if ( atom instanceof TokenRefElement ) {
2840            genMatchUsingAtomText(atom);
2841        } else if (atom instanceof WildcardElement) {
2842            gen((WildcardElement)atom);
2843        }
2844    }
2845    protected void genMatchUsingAtomText(GrammarAtom atom) {
2846        // match() for trees needs the _t cursor
2847        String   astArgs="";
2848        if (grammar instanceof TreeWalkerGrammar) {
2849            if( usingCustomAST )
2850                astArgs="static_cast<"+namespaceAntlr+"RefAST"+">(_t),";
2851            else
2852                astArgs="_t,";
2853        }
2854
2855        // if in lexer and ! on element, save buffer index to kill later
2856        if ( grammar instanceof LexerGrammar && (!saveText||atom.getAutoGenType()==GrammarElement.AUTO_GEN_BANG) ) {
2857            println("_saveIndex=text.length();");
2858        }
2859
2860        print(atom.not ? "matchNot(" : "match(");
2861        _print(astArgs);
2862
2863        // print out what to match
2864        if (atom.atomText.equals("EOF")) {
2865            // horrible hack to handle EOF case
2866            _print(namespaceAntlr+"Token::EOF_TYPE");
2867        }
2868        else {
2869            _print(textOrChar(atom.atomText));
2870        }
2871        _println(");");
2872
2873        if ( grammar instanceof LexerGrammar && (!saveText||atom.getAutoGenType()==GrammarElement.AUTO_GEN_BANG) ) {
2874            println("text.erase(_saveIndex);");      // kill text atom put in buffer
2875        }
2876    }
2877    protected void genMatchUsingAtomTokenType(GrammarAtom atom) {
2878        // match() for trees needs the _t cursor
2879        String   astArgs="";
2880        if (grammar instanceof TreeWalkerGrammar) {
2881            if( usingCustomAST )
2882                astArgs="static_cast<"+namespaceAntlr+"RefAST"+">(_t),";
2883            else
2884                astArgs="_t,";
2885        }
2886
2887        // If the literal can be mangled, generate the symbolic constant instead
2888        String   mangledName = null;
2889        String   s = astArgs + getValueString(atom.getType());
2890
2891        // matching
2892        println( (atom.not ? "matchNot(" : "match(") + s + ");");
2893    }
2894    /** Generate the nextToken() rule.
2895     * nextToken() is a synthetic lexer rule that is the implicit OR of all
2896     * user-defined lexer rules.
2897     * @param RuleBlock
2898     */
2899    public void genNextToken() {
2900        // Are there any public rules?  If not, then just generate a
2901        // fake nextToken().
2902        boolean hasPublicRules = false;
2903        for (int i = 0; i < grammar.rules.size(); i++) {
2904            RuleSymbol rs = (RuleSymbol)grammar.rules.elementAt(i);
2905            if ( rs.isDefined() && rs.access.equals("public") ) {
2906                hasPublicRules = true;
2907                break;
2908            }
2909        }
2910        if (!hasPublicRules) {
2911            println("");
2912            println(namespaceAntlr+"RefToken "+grammar.getClassName()+"::nextToken() { return "+namespaceAntlr+"RefToken(new "+namespaceAntlr+"CommonToken("+namespaceAntlr+"Token::EOF_TYPE, \"\")); }");
2913            println("");
2914            return;
2915        }
2916
2917        // Create the synthesized nextToken() rule
2918        RuleBlock nextTokenBlk = MakeGrammar.createNextTokenRule(grammar, grammar.rules, "nextToken");
2919        // Define the nextToken rule symbol
2920        RuleSymbol nextTokenRs = new RuleSymbol("mnextToken");
2921        nextTokenRs.setDefined();
2922        nextTokenRs.setBlock(nextTokenBlk);
2923        nextTokenRs.access = "private";
2924        grammar.define(nextTokenRs);
2925        // Analyze the nextToken rule
2926        boolean ok = grammar.theLLkAnalyzer.deterministic(nextTokenBlk);
2927
2928        // Generate the next token rule
2929        String   filterRule=null;
2930        if ( ((LexerGrammar)grammar).filterMode ) {
2931            filterRule = ((LexerGrammar)grammar).filterRule;
2932        }
2933
2934        println("");
2935        println(namespaceAntlr+"RefToken "+grammar.getClassName()+"::nextToken()");
2936        println("{");
2937        tabs++;
2938        println(namespaceAntlr+"RefToken theRetToken;");
2939        println("for (;;) {");
2940        tabs++;
2941        println(namespaceAntlr+"RefToken theRetToken;");
2942        println("int _ttype = "+namespaceAntlr+"Token::INVALID_TYPE;");
2943        if ( ((LexerGrammar)grammar).filterMode ) {
2944            println("setCommitToPath(false);");
2945            if ( filterRule!=null ) {
2946                // Here's a good place to ensure that the filter rule actually exists
2947                if ( !grammar.isDefined(CodeGenerator.encodeLexerRuleName(filterRule)) ) {
2948                    grammar.antlrTool.error("Filter rule "+filterRule+" does not exist in this lexer");
2949                }
2950                else {
2951                    RuleSymbol rs = (RuleSymbol)grammar.getSymbol(CodeGenerator.encodeLexerRuleName(filterRule));
2952                    if ( !rs.isDefined() ) {
2953                        grammar.antlrTool.error("Filter rule "+filterRule+" does not exist in this lexer");
2954                    }
2955                    else if ( rs.access.equals("public") ) {
2956                        grammar.antlrTool.error("Filter rule "+filterRule+" must be protected");
2957                    }
2958                }
2959                println("int _m;");
2960                println("_m = mark();");
2961            }
2962        }
2963        println("resetText();");
2964
2965        // Generate try around whole thing to trap scanner errors
2966        println("try {   // for lexical and char stream error handling");
2967        tabs++;
2968
2969        // Test for public lexical rules with empty paths
2970        for (int i=0; i<nextTokenBlk.getAlternatives().size(); i++) {
2971            Alternative a = nextTokenBlk.getAlternativeAt(i);
2972            if ( a.cache[1].containsEpsilon() ) {
2973                antlrTool.warning("found optional path in nextToken()");
2974            }
2975        }
2976
2977        // Generate the block
2978        String   newline = System.getProperty("line.separator");
2979        CppBlockFinishingInfo howToFinish = genCommonBlock(nextTokenBlk, false);
2980        String   errFinish = "if (LA(1)==EOF_CHAR)"+newline+
2981            "\t\t\t\t{"+newline+"\t\t\t\t\tuponEOF();"+newline+
2982            "\t\t\t\t\t_returnToken = makeToken("+namespaceAntlr+"Token::EOF_TYPE);"+
2983            newline+"\t\t\t\t}";
2984        errFinish += newline+"\t\t\t\t";
2985        if ( ((LexerGrammar)grammar).filterMode ) {
2986            if ( filterRule==null ) {
2987                errFinish += "else {consume(); goto tryAgain;}";
2988            }
2989            else {
2990                errFinish += "else {"+newline+
2991                        "\t\t\t\t\tcommit();"+newline+
2992                        "\t\t\t\t\ttry {m"+filterRule+"(false);}"+newline+
2993                        "\t\t\t\t\tcatch("+namespaceAntlr+"RecognitionException& e) {"+newline+
2994                        "\t\t\t\t\t // catastrophic failure"+newline+
2995                        "\t\t\t\t\t reportError(e);"+newline+
2996                        "\t\t\t\t\t consume();"+newline+
2997                        "\t\t\t\t\t}"+newline+
2998                        "\t\t\t\t\tgoto tryAgain;"+newline+
2999                        "\t\t\t\t}";
3000            }
3001        }
3002        else {
3003            errFinish += "else {"+throwNoViable+"}";
3004        }
3005        genBlockFinish(howToFinish, errFinish);
3006
3007        // at this point a valid token has been matched, undo "mark" that was done
3008        if ( ((LexerGrammar)grammar).filterMode && filterRule!=null ) {
3009            println("commit();");
3010        }
3011
3012        // Generate literals test if desired
3013        // make sure _ttype is set first; note _returnToken must be
3014        // non-null as the rule was required to create it.
3015        println("if ( !_returnToken )"+newline+
3016                  "\t\t\t\tgoto tryAgain; // found SKIP token"+newline);
3017        println("_ttype = _returnToken->getType();");
3018        if ( ((LexerGrammar)grammar).getTestLiterals()) {
3019            genLiteralsTest();
3020        }
3021
3022        // return token created by rule reference in switch
3023        println("_returnToken->setType(_ttype);");
3024        println("return _returnToken;");
3025
3026        // Close try block
3027        tabs--;
3028        println("}");
3029        println("catch ("+namespaceAntlr+"RecognitionException& e) {");
3030        tabs++;
3031        if ( ((LexerGrammar)grammar).filterMode ) {
3032            if ( filterRule==null ) {
3033                println("if ( !getCommitToPath() ) {");
3034                tabs++;
3035                println("consume();");
3036                println("goto tryAgain;");
3037                tabs--;
3038                println("}");
3039            }
3040            else {
3041                println("if ( !getCommitToPath() ) {");
3042                tabs++;
3043                println("rewind(_m);");
3044                println("resetText();");
3045                println("try {m"+filterRule+"(false);}");
3046                println("catch("+namespaceAntlr+"RecognitionException& ee) {");
3047                println("   // horrendous failure: error in filter rule");
3048                println("   reportError(ee);");
3049                println("   consume();");
3050                println("}");
3051                println("goto tryAgain;");
3052                tabs--;
3053                println("}");
3054            }
3055        }
3056        if ( nextTokenBlk.getDefaultErrorHandler() ) {
3057            println("reportError(e);");
3058            println("consume();");
3059        }
3060        else {
3061            // pass on to invoking routine
3062            println("throw "+namespaceAntlr+"TokenStreamRecognitionException(e);");
3063        }
3064
3065        // close CharStreamException try
3066        tabs--;
3067        println("}");
3068        println("catch ("+namespaceAntlr+"CharStreamIOException& csie) {");
3069        println("\tthrow "+namespaceAntlr+"TokenStreamIOException(csie.io);");
3070        println("}");
3071        println("catch ("+namespaceAntlr+"CharStreamException& cse) {");
3072        println("\tthrow "+namespaceAntlr+"TokenStreamException(cse.getMessage());");
3073        println("}");
3074
3075        // close for-loop
3076        _println("tryAgain:;");
3077        tabs--;
3078        println("}");
3079
3080        // close method nextToken
3081        tabs--;
3082        println("}");
3083        println("");
3084    }
3085    /** Gen a named rule block.
3086     * ASTs are generated for each element of an alternative unless
3087     * the rule or the alternative have a '!' modifier.
3088     *
3089     * If an alternative defeats the default tree construction, it
3090     * must set <rule>_AST to the root of the returned AST.
3091     *
3092     * Each alternative that does automatic tree construction, builds
3093     * up root and child list pointers in an ASTPair structure.
3094     *
3095     * A rule finishes by setting the returnAST variable from the
3096     * ASTPair.
3097     *
3098     * @param rule The name of the rule to generate
3099     * @param startSymbol true if the rule is a start symbol (i.e., not referenced elsewhere)
3100    */
3101    public void genRule(RuleSymbol s, boolean startSymbol, int ruleNum, String   prefix) {
3102//      tabs=1; // JavaCodeGenerator needs this
3103        if ( DEBUG_CODE_GENERATOR ) System.out.println("genRule("+ s.getId() +")");
3104        if ( !s.isDefined() ) {
3105            antlrTool.error("undefined rule: "+ s.getId());
3106            return;
3107        }
3108
3109        // Generate rule return type, name, arguments
3110        RuleBlock rblk = s.getBlock();
3111
3112        currentRule = rblk;
3113        currentASTResult = s.getId();
3114
3115        // clear list of declared ast variables..
3116        declaredASTVariables.clear();
3117
3118        // Save the AST generation state, and set it to that of the rule
3119        boolean savegenAST = genAST;
3120        genAST = genAST && rblk.getAutoGen();
3121
3122        // boolean oldsaveTest = saveText;
3123        saveText = rblk.getAutoGen();
3124
3125        // print javadoc comment if any
3126        if ( s.comment!=null ) {
3127            _println(s.comment);
3128        }
3129
3130        // Gen method return type (note lexer return action set at rule creation)
3131        if (rblk.returnAction != null)
3132        {
3133            // Has specified return value
3134            _print(extractTypeOfAction(rblk.returnAction, rblk.getLine(), rblk.getColumn()) + " ");
3135        } else {
3136            // No specified return value
3137            _print("void ");
3138        }
3139
3140        // Gen method name
3141        _print(prefix + s.getId() + "(");
3142
3143        // Additional rule parameters common to all rules for this grammar
3144        _print(commonExtraParams);
3145        if (commonExtraParams.length() != 0 && rblk.argAction != null ) {
3146            _print(",");
3147        }
3148
3149        // Gen arguments
3150        if (rblk.argAction != null)
3151        {
3152            // Has specified arguments
3153            _println("");
3154// FIXME: make argAction also a token? Hmmmmm
3155//          genLineNo(rblk);
3156            tabs++;
3157            println(rblk.argAction);
3158            tabs--;
3159            print(") ");
3160//          genLineNo2();   // gcc gives error on the brace... hope it works for the others too
3161        } else {
3162            // No specified arguments
3163            _print(") ");
3164        }
3165
3166        // Gen throws clause and open curly
3167//      _print(" throws " + exceptionThrown);
3168//      if ( !(grammar instanceof TreeWalkerGrammar) ) {
3169//          _print(", IOException");
3170//      }
3171        _println("{");
3172        tabs++;
3173
3174        if (grammar.traceRules) {
3175            if ( grammar instanceof TreeWalkerGrammar ) {
3176                if ( usingCustomAST )
3177                    println("Tracer traceInOut(this,\""+ s.getId() +"\",static_cast<"+namespaceAntlr+"RefAST"+">(_t));");
3178                else
3179                    println("Tracer traceInOut(this,\""+ s.getId() +"\",_t);");
3180            }
3181            else {
3182                println("Tracer traceInOut(this, \""+ s.getId() +"\");");
3183            }
3184        }
3185
3186        // Convert return action to variable declaration
3187        if (rblk.returnAction != null)
3188        {
3189            genLineNo(rblk);
3190            println(rblk.returnAction + ";");
3191            genLineNo2();
3192        }
3193
3194        // print out definitions needed by rules for various grammar types
3195        if (!commonLocalVars.equals(""))
3196            println(commonLocalVars);
3197
3198        if ( grammar instanceof LexerGrammar ) {
3199            // RK: why is this here? It seems not supported in the rest of the
3200            // tool.
3201            // lexer rule default return value is the rule's token name
3202            // This is a horrible hack to support the built-in EOF lexer rule.
3203            if (s.getId().equals("mEOF"))
3204                println("_ttype = "+namespaceAntlr+"Token::EOF_TYPE;");
3205            else
3206                println("_ttype = "+ s.getId().substring(1)+";");
3207            println("int _saveIndex;");     // used for element! (so we can kill text matched for element)
3208/*
3209            println("boolean old_saveConsumedInput=saveConsumedInput;");
3210            if ( !rblk.getAutoGen() ) {      // turn off "save input" if ! on rule
3211                println("saveConsumedInput=false;");
3212            }
3213*/
3214        }
3215
3216        // if debugging, write code to mark entry to the rule
3217        if ( grammar.debuggingOutput)
3218            if (grammar instanceof ParserGrammar)
3219                println("fireEnterRule(" + ruleNum + ",0);");
3220            else if (grammar instanceof LexerGrammar)
3221                println("fireEnterRule(" + ruleNum + ",_ttype);");
3222
3223        // Generate trace code if desired
3224//      if ( grammar.debuggingOutput || grammar.traceRules) {
3225//          println("try { // debugging");
3226//          tabs++;
3227//      }
3228
3229        // Initialize AST variables
3230        if (grammar instanceof TreeWalkerGrammar) {
3231            // "Input" value for rule
3232//          println(labeledElementASTType+" " + s.getId() + "_AST_in = "+labeledElementASTType+"(_t);");
3233            println(labeledElementASTType+" " + s.getId() + "_AST_in = _t;");
3234        }
3235        if (grammar.buildAST) {
3236            // Parser member used to pass AST returns from rule invocations
3237            println("returnAST = "+labeledElementASTInit+";");
3238            // Tracks AST construction
3239            println(namespaceAntlr+"ASTPair currentAST;"); // = new ASTPair();");
3240            // User-settable return value for rule.
3241            println(labeledElementASTType+" " + s.getId() + "_AST = "+labeledElementASTInit+";");
3242        }
3243
3244        genBlockPreamble(rblk);
3245        genBlockInitAction(rblk);
3246        println("");
3247
3248        // Search for an unlabeled exception specification attached to the rule
3249        ExceptionSpec unlabeledUserSpec = rblk.findExceptionSpec("");
3250
3251        // Generate try block around the entire rule for  error handling
3252        if (unlabeledUserSpec != null || rblk.getDefaultErrorHandler() ) {
3253            println("try {      // for error handling");
3254            tabs++;
3255        }
3256
3257        // Generate the alternatives
3258        if ( rblk.alternatives.size()==1 )
3259        {
3260            // One alternative -- use simple form
3261            Alternative alt = rblk.getAlternativeAt(0);
3262            String   pred = alt.semPred;
3263            if ( pred!=null )
3264                genSemPred(pred, currentRule.line);
3265            if (alt.synPred != null) {
3266                antlrTool.warning(
3267                    "Syntactic predicate ignored for single alternative",
3268                    grammar.getFilename(),
3269                    alt.synPred.getLine(),
3270                    alt.synPred.getColumn()
3271                );
3272            }
3273            genAlt(alt, rblk);
3274        }
3275        else
3276        {
3277            // Multiple alternatives -- generate complex form
3278            boolean ok = grammar.theLLkAnalyzer.deterministic(rblk);
3279
3280            CppBlockFinishingInfo howToFinish = genCommonBlock(rblk, false);
3281            genBlockFinish(howToFinish, throwNoViable);
3282        }
3283
3284        // Generate catch phrase for error handling
3285        if (unlabeledUserSpec != null || rblk.getDefaultErrorHandler() ) {
3286            // Close the try block
3287            tabs--;
3288            println("}");
3289        }
3290
3291        // Generate user-defined or default catch phrases
3292        if (unlabeledUserSpec != null)
3293        {
3294            genErrorHandler(unlabeledUserSpec);
3295        }
3296        else if (rblk.getDefaultErrorHandler())
3297        {
3298            // Generate default catch phrase
3299            println("catch (" + exceptionThrown + "& ex) {");
3300            tabs++;
3301            // Generate code to handle error if not guessing
3302            if (grammar.hasSyntacticPredicate) {
3303                println("if( inputState->guessing == 0 ) {");
3304                tabs++;
3305            }
3306            println("reportError(ex);");
3307            if ( !(grammar instanceof TreeWalkerGrammar) )
3308            {
3309                // Generate code to consume until token in k==1 follow set
3310                Lookahead follow = grammar.theLLkAnalyzer.FOLLOW(1, rblk.endNode);
3311                String   followSetName = getBitsetName(markBitsetForGen(follow.fset));
3312                println("consume();");
3313                println("consumeUntil(" + followSetName + ");");
3314            }
3315            else
3316            {
3317                // Just consume one token
3318                println("if ( _t != "+labeledElementASTInit+" )");
3319                tabs++;
3320                println("_t = _t->getNextSibling();");
3321                tabs--;
3322            }
3323            if (grammar.hasSyntacticPredicate)
3324            {
3325                tabs--;
3326                // When guessing, rethrow exception
3327                println("} else {");
3328                tabs++;
3329                println("throw ex;");
3330                tabs--;
3331                println("}");
3332            }
3333            // Close catch phrase
3334            tabs--;
3335            println("}");
3336        }
3337
3338        // Squirrel away the AST "return" value
3339        if (grammar.buildAST) {
3340            println("returnAST = " + s.getId() + "_AST;");
3341        }
3342
3343        // Set return tree value for tree walkers
3344        if ( grammar instanceof TreeWalkerGrammar ) {
3345            println("_retTree = _t;");
3346        }
3347
3348        // Generate literals test for lexer rules so marked
3349        if (rblk.getTestLiterals()) {
3350            if ( s.access.equals("protected") ) {
3351                genLiteralsTestForPartialToken();
3352            }
3353            else {
3354                genLiteralsTest();
3355            }
3356        }
3357
3358        // if doing a lexer rule, dump code to create token if necessary
3359        if ( grammar instanceof LexerGrammar ) {
3360            println("if ( _createToken && _token=="+namespaceAntlr+"nullToken && _ttype!="+namespaceAntlr+"Token::SKIP ) {");
3361            println("   _token = makeToken(_ttype);");
3362            println("   _token->setText(text.substr(_begin, text.length()-_begin));");
3363            println("}");
3364            println("_returnToken = _token;");
3365            // It should be easy for an optimizing compiler to realize this does nothing
3366            // but it avoids the warning about the variable being unused.
3367            println("_saveIndex=0;");
3368        }
3369
3370        // Gen the return statement if there is one (lexer has hard-wired return action)
3371        if (rblk.returnAction != null) {
3372            println("return " + extractIdOfAction(rblk.returnAction, rblk.getLine(), rblk.getColumn()) + ";");
3373        }
3374
3375//      if ( grammar.debuggingOutput || grammar.traceRules) {
3376////            tabs--;
3377////            println("} finally { // debugging");
3378////            tabs++;
3379//
3380//          // Generate trace code if desired
3381//          if ( grammar.debuggingOutput)
3382//              if (grammar instanceof ParserGrammar)
3383//                  println("fireExitRule(" + ruleNum + ",0);");
3384//              else if (grammar instanceof LexerGrammar)
3385//                  println("fireExitRule(" + ruleNum + ",_ttype);");
3386//
3387////            if (grammar.traceRules) {
3388////                if ( grammar instanceof TreeWalkerGrammar ) {
3389////                    println("traceOut(\""+ s.getId() +"\",_t);");
3390////                }
3391////                else {
3392////                    println("traceOut(\""+ s.getId() +"\");");
3393////                }
3394////            }
3395////
3396////            tabs--;
3397////            println("}");
3398//      }
3399
3400        tabs--;
3401        println("}");
3402        println("");
3403
3404        // Restore the AST generation state
3405        genAST = savegenAST;
3406
3407        // restore char save state
3408        // saveText = oldsaveTest;
3409    }
3410    public void genRuleHeader(RuleSymbol s, boolean startSymbol) {
3411        tabs=1;
3412        if ( DEBUG_CODE_GENERATOR ) System.out.println("genRuleHeader("+ s.getId() +")");
3413        if ( !s.isDefined() ) {
3414            antlrTool.error("undefined rule: "+ s.getId());
3415            return;
3416        }
3417
3418        // Generate rule return type, name, arguments
3419        RuleBlock rblk = s.getBlock();
3420        currentRule = rblk;
3421        currentASTResult = s.getId();
3422
3423        // Save the AST generation state, and set it to that of the rule
3424        boolean savegenAST = genAST;
3425        genAST = genAST && rblk.getAutoGen();
3426
3427        // boolean oldsaveTest = saveText;
3428        saveText = rblk.getAutoGen();
3429
3430        // Gen method access
3431        print(s.access + ": ");
3432
3433        // Gen method return type (note lexer return action set at rule creation)
3434        if (rblk.returnAction != null)
3435        {
3436            // Has specified return value
3437            _print(extractTypeOfAction(rblk.returnAction, rblk.getLine(), rblk.getColumn()) + " ");
3438        } else {
3439            // No specified return value
3440            _print("void ");
3441        }
3442
3443        // Gen method name
3444        _print(s.getId() + "(");
3445
3446        // Additional rule parameters common to all rules for this grammar
3447        _print(commonExtraParams);
3448        if (commonExtraParams.length() != 0 && rblk.argAction != null ) {
3449            _print(",");
3450        }
3451
3452        // Gen arguments
3453        if (rblk.argAction != null)
3454        {
3455            // Has specified arguments
3456            _println("");
3457            tabs++;
3458            println(rblk.argAction);
3459            tabs--;
3460            print(")");
3461        } else {
3462            // No specified arguments
3463            _print(")");
3464        }
3465        _println(";");
3466
3467        tabs--;
3468
3469        // Restore the AST generation state
3470        genAST = savegenAST;
3471
3472        // restore char save state
3473        // saveText = oldsaveTest;
3474    }
3475    private void GenRuleInvocation(RuleRefElement rr) {
3476        // dump rule name
3477        _print(rr.targetRule + "(");
3478
3479        // lexers must tell rule if it should set _returnToken
3480        if ( grammar instanceof LexerGrammar ) {
3481            // if labeled, could access Token, so tell rule to create
3482            if ( rr.getLabel() != null ) {
3483                _print("true");
3484            }
3485            else {
3486                _print("false");
3487            }
3488            if (commonExtraArgs.length() != 0 || rr.args!=null ) {
3489                _print(",");
3490            }
3491        }
3492
3493        // Extra arguments common to all rules for this grammar
3494        _print(commonExtraArgs);
3495        if (commonExtraArgs.length() != 0 && rr.args!=null ) {
3496            _print(",");
3497        }
3498
3499        // Process arguments to method, if any
3500        RuleSymbol rs = (RuleSymbol)grammar.getSymbol(rr.targetRule);
3501        if (rr.args != null)
3502        {
3503            // When not guessing, execute user arg action
3504            ActionTransInfo tInfo = new ActionTransInfo();
3505            // FIXME: fix line number passed to processActionForTreeSpecifiers here..
3506            // this one might be a bit off..
3507            String   args = processActionForTreeSpecifiers(rr.args, rr.line,
3508                                                                        currentRule, tInfo);
3509            if ( tInfo.assignToRoot || tInfo.refRuleRoot!=null )
3510            {
3511                antlrTool.error("Arguments of rule reference '" + rr.targetRule + "' cannot set or ref #"+
3512                    currentRule.getRuleName()+" on line "+rr.getLine());
3513            }
3514            _print(args);
3515
3516            // Warn if the rule accepts no arguments
3517            if (rs.block.argAction == null)
3518            {
3519                antlrTool.warning("Rule '" + rr.targetRule + "' accepts no arguments",
3520                    grammar.getFilename(),
3521                    rr.getLine(), rr.getColumn());
3522            }
3523        }
3524        else
3525        {
3526            // For C++, no warning if rule has parameters, because there may be default
3527            // values for all of the parameters
3528            //if (rs.block.argAction != null) {
3529            //  tool.warning("Missing parameters on reference to rule "+rr.targetRule, rr.getLine());
3530            //}
3531        }
3532        _println(");");
3533
3534        // move down to the first child while parsing
3535        if ( grammar instanceof TreeWalkerGrammar ) {
3536            println("_t = _retTree;");
3537        }
3538    }
3539    protected void genSemPred(String   pred, int line) {
3540        // translate $ and # references
3541        ActionTransInfo tInfo = new ActionTransInfo();
3542        pred = processActionForTreeSpecifiers(pred, line, currentRule, tInfo);
3543        // ignore translation info...we don't need to do anything with it.
3544        String   escapedPred = charFormatter.escapeString(pred);
3545
3546        // if debugging, wrap the semantic predicate evaluation in a method
3547        // that can tell SemanticPredicateListeners the result
3548        if (grammar.debuggingOutput && ((grammar instanceof ParserGrammar) ||
3549              (grammar instanceof LexerGrammar)))
3550            pred = "fireSemanticPredicateEvaluated(antlr.debug.SemanticPredicateEvent.VALIDATING," //FIXME
3551                + addSemPred(escapedPred) + "," + pred + ")";
3552        println("if (!(" + pred + "))");
3553        tabs++;
3554        println("throw "+namespaceAntlr+"SemanticException(\"" + escapedPred + "\");");
3555        tabs--;
3556    }
3557    /** Write an array of Strings which are the semantic predicate
3558     *  expressions.  The debugger will reference them by number only
3559     */
3560    protected void genSemPredMap(String   prefix) {
3561        Enumeration   e = semPreds.elements();
3562        println("const char* " + prefix + "_semPredNames[] = {");
3563        tabs++;
3564        while(e.hasMoreElements())
3565            println("\""+e.nextElement()+"\",");
3566        println("0");
3567        tabs--;
3568        println("};");
3569    }
3570    protected void genSynPred(SynPredBlock blk, String   lookaheadExpr) {
3571        if ( DEBUG_CODE_GENERATOR ) System.out.println("gen=>("+blk+")");
3572
3573        // Dump synpred result variable
3574        println("bool synPredMatched" + blk.ID + " = false;");
3575        // Gen normal lookahead test
3576        println("if (" + lookaheadExpr + ") {");
3577        tabs++;
3578
3579        // Save input state
3580        if ( grammar instanceof TreeWalkerGrammar ) {
3581            println(labeledElementType + " __t" + blk.ID + " = _t;");
3582        }
3583        else {
3584            println("int _m" + blk.ID + " = mark();");
3585        }
3586
3587        // Once inside the try, assume synpred works unless exception caught
3588        println("synPredMatched" + blk.ID + " = true;");
3589        println("inputState->guessing++;");
3590
3591        // if debugging, tell listeners that a synpred has started
3592        if (grammar.debuggingOutput && ((grammar instanceof ParserGrammar) ||
3593             (grammar instanceof LexerGrammar))) {
3594            println("fireSyntacticPredicateStarted();");
3595        }
3596
3597        syntacticPredLevel++;
3598        println("try {");
3599        tabs++;
3600        gen((AlternativeBlock)blk);     // gen code to test predicate
3601        tabs--;
3602        //println("System.out.println(\"pred "+blk+" succeeded\");");
3603        println("}");
3604        println("catch (" + exceptionThrown + "& pe) {");
3605        tabs++;
3606        println("synPredMatched"+blk.ID+" = false;");
3607        //println("System.out.println(\"pred "+blk+" failed\");");
3608        tabs--;
3609        println("}");
3610
3611        // Restore input state
3612        if ( grammar instanceof TreeWalkerGrammar ) {
3613            println("_t = __t"+blk.ID+";");
3614        }
3615        else {
3616            println("rewind(_m"+blk.ID+");");
3617        }
3618
3619        println("inputState->guessing--;");
3620
3621        // if debugging, tell listeners how the synpred turned out
3622        if (grammar.debuggingOutput && ((grammar instanceof ParserGrammar) ||
3623             (grammar instanceof LexerGrammar))) {
3624            println("if (synPredMatched" + blk.ID +")");
3625            println("  fireSyntacticPredicateSucceeded();");
3626            println("else");
3627            println("  fireSyntacticPredicateFailed();");
3628        }
3629
3630        syntacticPredLevel--;
3631        tabs--;
3632
3633        // Close lookahead test
3634        println("}");
3635
3636        // Test synpred result
3637        println("if ( synPredMatched"+blk.ID+" ) {");
3638    }
3639    /** Generate a static array containing the names of the tokens,
3640     * indexed by the token type values.  This static array is used
3641     * to format error messages so that the token identifers or literal
3642     * strings are displayed instead of the token numbers.
3643     *
3644     * If a lexical rule has a paraphrase, use it rather than the
3645     * token label.
3646     */
3647    public void genTokenStrings(String   prefix) {
3648        // Generate a string for each token.  This creates a static
3649        // array of Strings indexed by token type.
3650//      println("");
3651        println("const char* " + prefix + "tokenNames[] = {");
3652        tabs++;
3653
3654        // Walk the token vocabulary and generate a Vector of strings
3655        // from the tokens.
3656        Vector v = grammar.tokenManager.getVocabulary();
3657        for (int i = 0; i < v.size(); i++)
3658        {
3659            String   s = (String  )v.elementAt(i);
3660            if (s == null)
3661            {
3662                s = "<"+String.valueOf(i)+">";
3663            }
3664            if ( !s.startsWith("\"") && !s.startsWith("<") ) {
3665                TokenSymbol ts = (TokenSymbol)grammar.tokenManager.getTokenSymbol(s);
3666                if ( ts!=null && ts.getParaphrase()!=null ) {
3667                    s = StringUtils.stripFrontBack(ts.getParaphrase(), "\"", "\"");
3668                }
3669            }
3670            print(charFormatter.literalString(s));
3671            _println(",");
3672        }
3673        println("0");
3674
3675        // Close the string array initailizer
3676        tabs--;
3677        println("};");
3678    }
3679    /** Generate the token types C++ file */
3680    protected void genTokenTypes(TokenManager tm) throws IOException   {
3681        // Open the token output header file and set the currentOutput stream
3682        outputFile = tm.getName() + TokenTypesFileSuffix+".hpp";
3683        outputLine = 1;
3684        currentOutput = antlrTool.openOutputFile(outputFile);
3685        //SAS: changed for proper text file io
3686
3687        tabs = 0;
3688
3689        // Generate a guard wrapper
3690        println("#ifndef INC_"+tm.getName()+TokenTypesFileSuffix+"_hpp_");
3691        println("#define INC_"+tm.getName()+TokenTypesFileSuffix+"_hpp_");
3692        println("");
3693
3694        if (nameSpace != null)
3695            nameSpace.emitDeclarations(currentOutput);
3696
3697        // Generate the header common to all C++ files
3698        genHeader(outputFile);
3699
3700        // Encapsulate the definitions in an interface.  This can be done
3701        // because they are all constants.
3702        println("struct " + tm.getName() + TokenTypesFileSuffix+" {");
3703        tabs++;
3704        println("enum {");
3705        tabs++;
3706
3707        // Generate a definition for each token type
3708        Vector v = tm.getVocabulary();
3709
3710        // Do special tokens manually
3711        println("EOF_ = " + Token.EOF_TYPE + ",");
3712
3713        // Move the other special token to the end, so we can solve
3714        // the superfluous comma problem easily
3715
3716        for (int i = Token.MIN_USER_TYPE; i < v.size(); i++) {
3717            String   s = (String  )v.elementAt(i);
3718            if (s != null) {
3719                if ( s.startsWith("\"") ) {
3720                    // a string literal
3721                    StringLiteralSymbol sl = (StringLiteralSymbol)tm.getTokenSymbol(s);
3722                    if ( sl==null ) {
3723                        antlrTool.panic("String literal "+s+" not in symbol table");
3724                    }
3725                    else if ( sl.label != null ) {
3726                        println(sl.label + " = " + i + ",");
3727                    }
3728                    else {
3729                        String   mangledName = mangleLiteral(s);
3730                        if (mangledName != null) {
3731                            // We were able to create a meaningful mangled token name
3732                            println(mangledName + " = " + i + ",");
3733                            // if no label specified, make the label equal to the mangled name
3734                            sl.label = mangledName;
3735                        }
3736                        else {
3737                            println("// " + s + " = " + i);
3738                        }
3739                    }
3740                }
3741                else if ( !s.startsWith("<") ) {
3742                    println(s + " = " + i + ",");
3743                }
3744            }
3745        }
3746
3747        // Moved from above
3748        println("NULL_TREE_LOOKAHEAD = " + Token.NULL_TREE_LOOKAHEAD);
3749
3750        // Close the enum
3751        tabs--;
3752        println("};");
3753
3754        // Close the interface
3755        tabs--;
3756        println("};");
3757
3758        if (nameSpace != null)
3759            nameSpace.emitClosures(currentOutput);
3760
3761        // Generate a guard wrapper
3762        println("#endif /*INC_"+tm.getName()+TokenTypesFileSuffix+"_hpp_*/");
3763
3764        // Close the tokens output file
3765        currentOutput.close();
3766        currentOutput = null;
3767        exitIfError();
3768    }
3769    /** Process a string for an simple expression for use in xx/action.g
3770     * it is used to cast simple tokens/references to the right type for
3771     * the generated language. Basically called for every element in
3772     * the vector to getASTCreateString(vector V)
3773     * @param str A String.
3774     */
3775    public String   processStringForASTConstructor( String   str )
3776    {
3777        if( usingCustomAST &&
3778            ((grammar instanceof TreeWalkerGrammar) ||
3779             (grammar instanceof ParserGrammar))  &&
3780            !(grammar.tokenManager.tokenDefined(str) ) )
3781        {
3782//          System.out.println("processStringForASTConstructor: "+str+" with cast");
3783            return "static_cast<"+namespaceAntlr+"RefAST>("+str+")";
3784        }
3785        else
3786        {
3787//          System.out.println("processStringForASTConstructor: "+str);
3788            return str;
3789        }
3790    }
3791    /** Get a string for an expression to generate creation of an AST subtree.
3792      * @param v A Vector of String, where each element is an expression
3793      * in the target language yielding an AST node.
3794      */
3795    public String   getASTCreateString(Vector v) {
3796        if (v.size() == 0) {
3797            return "";
3798        }
3799        StringBuffer   buf = new StringBuffer  ();
3800        // the labeledElementASTType here can probably be a cast or nothing
3801        // in the case of ! usingCustomAST
3802        buf.append(labeledElementASTType+
3803                    "(astFactory->make((new "+namespaceAntlr+
3804                      "ASTArray("+v.size()+"))");
3805        for (int i = 0; i < v.size(); i++) {
3806            buf.append("->add("+ v.elementAt(i) + ")");
3807        }
3808        buf.append("))");
3809        return buf.toString();
3810    }
3811    /** Get a string for an expression to generate creating of an AST node
3812     * @param str The arguments to the AST constructor
3813     */
3814    public String   getASTCreateString(GrammarAtom atom, String   str) {
3815        if ( atom!=null && atom.getASTNodeType() != null ) {
3816
3817            // this atom is using a heterogeneous AST type.
3818            // make note of the factory needed to generate it..
3819            // later this is inserted into the initializeFactory method.
3820            astTypes.appendElement("astFactory->registerFactory("+
3821                                      atom.getType() + ", \""+atom.getASTNodeType()+
3822                                      "\", "+atom.getASTNodeType()+"::factory);");
3823
3824            // after above init the factory knows what to generate...
3825            // maybe add apropriate cast here..
3826//          return "astFactory->create("+atom.getType()+", "+str+")";
3827//          // The string itself should be enough...
3828            return "astFactory->create("+str+")";
3829//          return "Ref"+atom.getASTNodeType()+"(new "+atom.getASTNodeType()+"("+str+"))";
3830//          return "Ref"+atom.getASTNodeType()+"(new "+atom.getASTNodeType()+"("+str+"))";
3831        }
3832        else
3833        {
3834            // FIXME: This is *SO* ugly! but it will have to do for now...
3835            // 2.7.2 will have better I hope
3836            // this is due to the usage of getASTCreateString from inside
3837            // actions/cpp/action.g
3838            boolean is_constructor = false;
3839            if( str.indexOf(',') != -1 )
3840                is_constructor = grammar.tokenManager.tokenDefined(str.substring(0,str.indexOf(',')));
3841
3842//          System.out.println("getAstCreateString(as): "+str+" "+grammar.tokenManager.tokenDefined(str));
3843            if( usingCustomAST &&
3844               (grammar instanceof TreeWalkerGrammar) &&
3845                !(grammar.tokenManager.tokenDefined(str) ) &&
3846                ! is_constructor )
3847                return "astFactory->create(static_cast<"+namespaceAntlr+"RefAST>("+str+"))";
3848            else
3849                return "astFactory->create("+str+")";
3850        }
3851    }
3852
3853    /** Get a string for an expression to generate creating of an AST node
3854     * @param str The arguments to the AST constructor
3855     */
3856    public String   getASTCreateString(String   str) {
3857//      System.out.println("getAstCreateString(str): "+str+" "+grammar.tokenManager.tokenDefined(str));
3858        if( usingCustomAST )
3859            return "static_cast<"+labeledElementASTType+">(astFactory->create(static_cast<"+namespaceAntlr+"RefAST>("+str+")))";
3860        else
3861            return "astFactory->create("+str+")";
3862    }
3863
3864    protected String   getLookaheadTestExpression(Lookahead[] look, int k) {
3865        StringBuffer   e = new StringBuffer  (100);
3866        boolean first = true;
3867
3868        e.append("(");
3869        for (int i = 1; i <= k; i++) {
3870            BitSet p = look[i].fset;
3871            if (!first) {
3872                e.append(") && (");
3873            }
3874            first = false;
3875
3876            // Syn preds can yield <end-of-syn-pred> (epsilon) lookahead.
3877            // There is no way to predict what that token would be.  Just
3878            // allow anything instead.
3879            if (look[i].containsEpsilon()) {
3880                e.append("true");
3881            } else {
3882                e.append(getLookaheadTestTerm(i, p));
3883            }
3884        }
3885        e.append(")");
3886
3887        return e.toString();
3888    }
3889    /** Generate a lookahead test expression for an alternate.  This
3890     * will be a series of tests joined by '&&' and enclosed by '()',
3891     * the number of such tests being determined by the depth of the lookahead.
3892     */
3893    protected String   getLookaheadTestExpression(Alternative alt, int maxDepth) {
3894        int depth = alt.lookaheadDepth;
3895        if ( depth == GrammarAnalyzer.NONDETERMINISTIC ) {
3896            // if the decision is nondeterministic, do the best we can: LL(k)
3897            // any predicates that are around will be generated later.
3898            depth = grammar.maxk;
3899        }
3900
3901        if ( maxDepth==0 ) {
3902            // empty lookahead can result from alt with sem pred
3903            // that can see end of token.  E.g., A : {pred}? ('a')? ;
3904            return "true";
3905        }
3906
3907/*
3908boolean first = true;
3909        for (int i=1; i<=depth && i<=maxDepth; i++) {
3910            BitSet p = alt.cache[i].fset;
3911            if (!first) {
3912                e.append(") && (");
3913            }
3914            first = false;
3915
3916            // Syn preds can yield <end-of-syn-pred> (epsilon) lookahead.
3917            // There is no way to predict what that token would be.  Just
3918            // allow anything instead.
3919            if ( alt.cache[i].containsEpsilon() ) {
3920                e.append("true");
3921            }
3922            else {
3923                e.append(getLookaheadTestTerm(i, p));
3924            }
3925        }
3926
3927        e.append(")");
3928*/
3929
3930        return "(" + getLookaheadTestExpression(alt.cache,depth) + ")";
3931    }
3932    /**Generate a depth==1 lookahead test expression given the BitSet.
3933     * This may be one of:
3934     * 1) a series of 'x==X||' tests
3935     * 2) a range test using >= && <= where possible,
3936     * 3) a bitset membership test for complex comparisons
3937     * @param k The lookahead level
3938     * @param p The lookahead set for level k
3939     */
3940    protected String   getLookaheadTestTerm(int k, BitSet p) {
3941        // Determine the name of the item to be compared
3942        String   ts = lookaheadString(k);
3943
3944        // Generate a range expression if possible
3945        int[] elems = p.toArray();
3946        if (elementsAreRange(elems)) {
3947            return getRangeExpression(k, elems);
3948        }
3949
3950        // Generate a bitset membership test if possible
3951        StringBuffer   e;
3952        int degree = p.degree();
3953        if ( degree == 0 ) {
3954            return "true";
3955        }
3956
3957        if (degree >= bitsetTestThreshold) {
3958            int bitsetIdx = markBitsetForGen(p);
3959            return getBitsetName(bitsetIdx) + ".member(" + ts + ")";
3960        }
3961
3962        // Otherwise, generate the long-winded series of "x==X||" tests
3963        e = new StringBuffer  ();
3964        for (int i = 0; i < elems.length; i++) {
3965            // Get the compared-to item (token or character value)
3966            String   cs = getValueString(elems[i]);
3967
3968            // Generate the element comparison
3969            if( i > 0 ) e.append(" || ");
3970            e.append(ts);
3971            e.append(" == ");
3972            e.append(cs);
3973        }
3974        return e.toString();
3975    }
3976    /** Return an expression for testing a contiguous renage of elements
3977     * @param k The lookahead level
3978     * @param elems The elements representing the set, usually from BitSet.toArray().
3979     * @return String containing test expression.
3980     */
3981    public String   getRangeExpression(int k, int[] elems) {
3982        if (!elementsAreRange(elems)) {
3983            antlrTool.panic("getRangeExpression called with non-range");
3984        }
3985        int begin = elems[0];
3986        int end = elems[elems.length-1];
3987        return
3988            "(" + lookaheadString(k) + " >= " + getValueString(begin) + " && " +
3989              lookaheadString(k) + " <= " + getValueString(end) + ")";
3990    }
3991    /** getValueString: get a string representation of a token or char value
3992     * @param value The token or char value
3993     */
3994    private String   getValueString(int value) {
3995        String   cs;
3996        if ( grammar instanceof LexerGrammar ) {
3997            cs = charFormatter.literalChar(value);
3998        }
3999        else
4000        {
4001            TokenSymbol ts = grammar.tokenManager.getTokenSymbolAt(value);
4002            if ( ts == null ) {
4003                return ""+value; // return token type as string
4004                // tool.panic("vocabulary for token type " + value + " is null");
4005            }
4006            String   tId = ts.getId();
4007            if ( ts instanceof StringLiteralSymbol ) {
4008                // if string literal, use predefined label if any
4009                // if no predefined, try to mangle into LITERAL_xxx.
4010                // if can't mangle, use int value as last resort
4011                StringLiteralSymbol sl = (StringLiteralSymbol)ts;
4012                String   label = sl.getLabel();
4013                if ( label!=null ) {
4014                    cs = label;
4015                }
4016                else {
4017                    cs = mangleLiteral(tId);
4018                    if (cs == null) {
4019                        cs = String.valueOf(value);
4020                    }
4021                }
4022            }
4023            else {
4024                if ( tId.equals("EOF") )
4025                    cs = namespaceAntlr+"Token::EOF_TYPE";
4026                else
4027                    cs = tId;
4028            }
4029        }
4030        return cs;
4031    }
4032    /**Is the lookahead for this alt empty? */
4033    protected boolean lookaheadIsEmpty(Alternative alt, int maxDepth) {
4034        int depth = alt.lookaheadDepth;
4035        if ( depth == GrammarAnalyzer.NONDETERMINISTIC ) {
4036            depth = grammar.maxk;
4037        }
4038        for (int i=1; i<=depth && i<=maxDepth; i++) {
4039            BitSet p = alt.cache[i].fset;
4040            if (p.degree() != 0) {
4041                return false;
4042            }
4043        }
4044        return true;
4045    }
4046    private String   lookaheadString(int k) {
4047        if (grammar instanceof TreeWalkerGrammar) {
4048            return "_t->getType()";
4049        }
4050        return "LA(" + k + ")";
4051    }
4052    /** Mangle a string literal into a meaningful token name.  This is
4053      * only possible for literals that are all characters.  The resulting
4054      * mangled literal name is literalsPrefix with the text of the literal
4055      * appended.
4056      * @return A string representing the mangled literal, or null if not possible.
4057      */
4058    private String   mangleLiteral(String   s) {
4059        String   mangled = antlrTool.literalsPrefix;
4060        for (int i = 1; i < s.length()-1; i++) {
4061            if (!Character.isLetter(s.charAt(i)) &&
4062                 s.charAt(i) != '_') {
4063                return null;
4064            }
4065            mangled += s.charAt(i);
4066        }
4067        if ( antlrTool.upperCaseMangledLiterals ) {
4068            mangled = mangled.toUpperCase();
4069        }
4070        return mangled;
4071    }
4072    /** Map an identifier to it's corresponding tree-node variable.
4073      * This is context-sensitive, depending on the rule and alternative
4074      * being generated
4075      * @param idParam The identifier name to map
4076      * @return The mapped id (which may be the same as the input), or null if the mapping is invalid due to duplicates
4077      */
4078    public String   mapTreeId(String   idParam, ActionTransInfo transInfo) {
4079        // if not in an action of a rule, nothing to map.
4080        if ( currentRule==null ) return idParam;
4081//      System.out.print("mapTreeId: "+idParam+" "+currentRule.getRuleName()+" ");
4082
4083        boolean in_var = false;
4084        String   id = idParam;
4085        if (grammar instanceof TreeWalkerGrammar)
4086        {
4087//          RK: hmmm this seems odd. If buildAST is false it translates
4088//          #rulename_in to 'rulename_in' else to 'rulename_AST_in' which indeed
4089//          exists. disabling for now.. and hope it doesn't blow up somewhere.
4090            if ( !grammar.buildAST )
4091            {
4092                in_var = true;
4093//              System.out.println("in_var1");
4094            }
4095            // If the id ends with "_in", then map it to the input variable
4096//          else
4097            if (id.length() > 3 && id.lastIndexOf("_in") == id.length()-3)
4098            {
4099                // Strip off the "_in"
4100                id = id.substring(0, id.length()-3);
4101                in_var = true;
4102//              System.out.println("in_var2");
4103            }
4104        }
4105//      System.out.print(in_var+"\t");
4106
4107        // Check the rule labels.  If id is a label, then the output
4108        // variable is label_AST, and the input variable is plain label.
4109        for (int i = 0; i < currentRule.labeledElements.size(); i++)
4110        {
4111            AlternativeElement elt = (AlternativeElement)currentRule.labeledElements.elementAt(i);
4112            if (elt.getLabel().equals(id))
4113            {
4114//              if( in_var )
4115//                  System.out.println("returning (vec) "+(in_var ? id : id + "_AST"));
4116                return in_var ? id : id + "_AST";
4117            }
4118        }
4119
4120        // Failing that, check the id-to-variable map for the alternative.
4121        // If the id is in the map, then output variable is the name in the
4122        // map, and input variable is name_in
4123        String   s = (String  )treeVariableMap.get(id);
4124        if (s != null)
4125        {
4126            if (s == NONUNIQUE)
4127            {
4128//              if( in_var )
4129//                  System.out.println("returning null (nonunique)");
4130                // There is more than one element with this id
4131                return null;
4132            }
4133            else if (s.equals(currentRule.getRuleName()))
4134            {
4135                // a recursive call to the enclosing rule is
4136                // ambiguous with the rule itself.
4137//              if( in_var )
4138//                  System.out.println("returning null (rulename)");
4139                return null;
4140            }
4141            else
4142            {
4143//              if( in_var )
4144//              System.out.println("returning "+(in_var?s+"_in":s));
4145                return in_var ? s + "_in" : s;
4146            }
4147        }
4148
4149//      System.out.println("Last check: "+id+" == "+currentRule.getRuleName());
4150        // Failing that, check the rule name itself.  Output variable
4151        // is rule_AST; input variable is rule_AST_in (treeparsers).
4152        if( id.equals(currentRule.getRuleName()) )
4153        {
4154            String   r = in_var ? id + "_AST_in" : id + "_AST";
4155            if ( transInfo!=null ) {
4156                if ( !in_var ) {
4157                    transInfo.refRuleRoot = r;
4158                }
4159            }
4160//          if( in_var )
4161//              System.out.println("returning (r) "+r);
4162            return r;
4163        }
4164        else
4165        {
4166//          if( in_var )
4167//          System.out.println("returning (last) "+id);
4168            // id does not map to anything -- return itself.
4169            return id;
4170        }
4171    }
4172    /** Given an element and the name of an associated AST variable,
4173      * create a mapping between the element "name" and the variable name.
4174      */
4175    private void mapTreeVariable(AlternativeElement e, String   name)
4176    {
4177        // For tree elements, defer to the root
4178        if (e instanceof TreeElement) {
4179            mapTreeVariable( ((TreeElement)e).root, name);
4180            return;
4181        }
4182
4183        // Determine the name of the element, if any, for mapping purposes
4184        String   elName = null;
4185
4186        // Don't map labeled items
4187        if (e.getLabel() == null) {
4188            if (e instanceof TokenRefElement) {
4189                // use the token id
4190                elName = ((TokenRefElement)e).atomText;
4191            }
4192            else if (e instanceof RuleRefElement) {
4193                // use the rule name
4194                elName = ((RuleRefElement)e).targetRule;
4195            }
4196        }
4197        // Add the element to the tree variable map if it has a name
4198        if (elName != null) {
4199            if (treeVariableMap.get(elName) != null) {
4200                // Name is already in the map -- mark it as duplicate
4201                treeVariableMap.remove(elName);
4202                treeVariableMap.put(elName, NONUNIQUE);
4203            }
4204            else {
4205                treeVariableMap.put(elName, name);
4206            }
4207        }
4208    }
4209
4210    /** Lexically process tree-specifiers in the action.
4211     * This will replace #id and #(...) with the appropriate
4212     * function calls and/or variables.
4213     */
4214    protected String   processActionForTreeSpecifiers(String   actionStr,
4215                                                                    int line,
4216                                                                    RuleBlock currentRule,
4217                                                                    ActionTransInfo tInfo)
4218    {
4219        if ( actionStr==null || actionStr.length()==0 )
4220            return null;
4221
4222        // The action trans info tells us (at the moment) whether an
4223        // assignment was done to the rule's tree root.
4224        if (grammar==null)
4225            return actionStr;
4226
4227        if ((grammar.buildAST && actionStr.indexOf('#') != -1) ||
4228             grammar instanceof TreeWalkerGrammar ||
4229             ((grammar instanceof LexerGrammar ||
4230                grammar instanceof ParserGrammar)
4231                && actionStr.indexOf('$') != -1) )
4232        {
4233            // Create a lexer to read an action and return the translated version
4234            antlr.actions.cpp.ActionLexer lexer =
4235                new antlr.actions.cpp.ActionLexer(actionStr, currentRule, this, tInfo);
4236            lexer.setLineOffset(line);
4237            lexer.setFilename(grammar.getFilename());
4238            lexer.setTool(antlrTool);
4239
4240            try {
4241                lexer.mACTION(true);
4242                actionStr = lexer.getTokenObject().getText();
4243                // System.out.println("action translated: "+actionStr);
4244                // System.out.println("trans info is "+tInfo);
4245            }
4246            catch (RecognitionException ex) {
4247                lexer.reportError(ex);
4248                return actionStr;
4249            }
4250            catch (TokenStreamException tex) {
4251                antlrTool.panic("Error reading action:"+actionStr);
4252                return actionStr;
4253            }
4254            catch (CharStreamException io) {
4255                antlrTool.panic("Error reading action:"+actionStr);
4256                return actionStr;
4257            }
4258        }
4259        return actionStr;
4260    }
4261
4262    private String   fixNameSpaceOption( String   ns )
4263    {
4264        ns = StringUtils.stripFrontBack(ns,"\"","\"");
4265        if( ns.length() > 2 &&
4266             !ns.substring(ns.length()-2, ns.length()).equals("::") )
4267        ns += "::";
4268        return ns;
4269    }
4270
4271    private void setupGrammarParameters(Grammar g) {
4272        if (g instanceof ParserGrammar ||
4273             g instanceof LexerGrammar  ||
4274             g instanceof TreeWalkerGrammar
4275            )
4276        {
4277            /* RK: options also have to be added to Grammar.java and for options
4278             * on the file level entries have to be defined in
4279             * DefineGrammarSymbols.java and passed around via 'globals' in antlrTool.java
4280             */
4281            if( antlrTool.nameSpace != null )
4282                nameSpace = antlrTool.nameSpace;
4283
4284            if( antlrTool.namespaceStd != null )
4285                namespaceStd = fixNameSpaceOption(antlrTool.namespaceStd);
4286
4287            if( antlrTool.namespaceAntlr != null )
4288                namespaceAntlr = fixNameSpaceOption(antlrTool.namespaceAntlr);
4289
4290            genHashLines = antlrTool.genHashLines;
4291
4292            /* let grammar level options override filelevel ones...
4293             */
4294            if( g.hasOption("namespace") ) {
4295                Token t = g.getOption("namespace");
4296                if( t != null ) {
4297                    nameSpace = new NameSpace(t.getText());
4298                }
4299            }
4300            if( g.hasOption("namespaceAntlr") ) {
4301                Token t = g.getOption("namespaceAntlr");
4302                if( t != null ) {
4303                    String   ns = StringUtils.stripFrontBack(t.getText(),"\"","\"");
4304                    if ( ns != null ) {
4305                        if( ns.length() > 2 &&
4306                             !ns.substring(ns.length()-2, ns.length()).equals("::") )
4307                            ns += "::";
4308                        namespaceAntlr = ns;
4309                    }
4310                }
4311            }
4312            if( g.hasOption("namespaceStd") ) {
4313                Token t = g.getOption("namespaceStd");
4314                if( t != null ) {
4315                    String   ns = StringUtils.stripFrontBack(t.getText(),"\"","\"");
4316                    if ( ns != null ) {
4317                        if( ns.length() > 2 &&
4318                             !ns.substring(ns.length()-2, ns.length()).equals("::") )
4319                            ns += "::";
4320                        namespaceStd = ns;
4321                    }
4322                }
4323            }
4324            if( g.hasOption("genHashLines") ) {
4325                Token t = g.getOption("genHashLines");
4326                if( t != null ) {
4327                    String   val = StringUtils.stripFrontBack(t.getText(),"\"","\"");
4328                    genHashLines = val.equals("true");
4329                }
4330            }
4331        }
4332        if (g instanceof ParserGrammar) {
4333            labeledElementASTType = namespaceAntlr+"RefAST";
4334            labeledElementASTInit = namespaceAntlr+"nullAST";
4335            if ( g.hasOption("ASTLabelType") ) {
4336                Token tsuffix = g.getOption("ASTLabelType");
4337                if ( tsuffix != null ) {
4338                    String   suffix = StringUtils.stripFrontBack(tsuffix.getText(),"\"","\"");
4339                    if ( suffix != null ) {
4340                        usingCustomAST = true;
4341                        labeledElementASTType = suffix;
4342                        labeledElementASTInit = "static_cast<"+suffix+">("+namespaceAntlr+"nullAST)";
4343                    }
4344                }
4345            }
4346            labeledElementType = namespaceAntlr+"RefToken ";
4347            labeledElementInit = namespaceAntlr+"nullToken";
4348            commonExtraArgs = "";
4349            commonExtraParams = "";
4350            commonLocalVars = "";
4351            lt1Value = "LT(1)";
4352            exceptionThrown = namespaceAntlr+"RecognitionException";
4353            throwNoViable = "throw "+namespaceAntlr+"NoViableAltException(LT(1), getFilename());";
4354        }
4355        else if (g instanceof LexerGrammar) {
4356            labeledElementType = "char ";
4357            labeledElementInit = "'\\0'";
4358            commonExtraArgs = "";
4359            commonExtraParams = "bool _createToken";
4360            commonLocalVars = "int _ttype; "+namespaceAntlr+"RefToken _token; int _begin=text.length();";
4361            lt1Value = "LA(1)";
4362            exceptionThrown = namespaceAntlr+"RecognitionException";
4363            throwNoViable = "throw "+namespaceAntlr+"NoViableAltForCharException(LA(1), getFilename(), getLine(), getColumn());";
4364        }
4365        else if (g instanceof TreeWalkerGrammar) {
4366            labeledElementInit = namespaceAntlr+"nullAST";
4367            labeledElementASTInit = namespaceAntlr+"nullAST";
4368            labeledElementASTType = namespaceAntlr+"RefAST";
4369            labeledElementType = namespaceAntlr+"RefAST";
4370            commonExtraParams = namespaceAntlr+"RefAST _t";
4371            throwNoViable = "throw "+namespaceAntlr+"NoViableAltException(_t);";
4372            lt1Value = "_t";
4373            if ( g.hasOption("ASTLabelType") ) {
4374                Token tsuffix = g.getOption("ASTLabelType");
4375                if ( tsuffix != null ) {
4376                    String   suffix = StringUtils.stripFrontBack(tsuffix.getText(),"\"","\"");
4377                    if ( suffix != null ) {
4378                        usingCustomAST = true;
4379                        labeledElementASTType = suffix;
4380                        labeledElementType = suffix;
4381                        labeledElementInit = "static_cast<"+suffix+">("+namespaceAntlr+"nullAST)";
4382                        labeledElementASTInit = labeledElementInit;
4383                        commonExtraParams = suffix+" _t";
4384                        throwNoViable = "throw "+namespaceAntlr+"NoViableAltException(static_cast<"+namespaceAntlr+"RefAST>(_t));";
4385                        lt1Value = "_t";
4386                    }
4387                }
4388            }
4389            if ( !g.hasOption("ASTLabelType") ) {
4390                g.setOption("ASTLabelType", new Token(ANTLRTokenTypes.STRING_LITERAL,namespaceAntlr+"RefAST"));
4391            }
4392            commonExtraArgs = "_t";
4393            commonLocalVars = "";
4394            exceptionThrown = namespaceAntlr+"RecognitionException";
4395        }
4396        else {
4397            antlrTool.panic("Unknown grammar type");
4398        }
4399    }
4400    private String   textOrChar(String   text) {
4401        // check to see if the text is a single character
4402        if (text.startsWith("'")) {
4403            // assume it also ends with '
4404            return charFormatter.literalChar(ANTLRLexer.tokenTypeForCharLiteral(text));
4405        }
4406        else
4407            return text;
4408    }
4409}
4410
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Free Books Free Magazines
Popular Tags