gnucparser.g

plugin for eclipse

                        } :
                    ( "typedef" | "__label__" | declaration )=> declarationList
                    | (nestedFunctionDef)=> nestedFunctionDef
                )*
                ( statementList )?
                            { popScope(); }
                RCURLY
                            { ##.setType( NCompoundStatement ); ##.setAttribute( "scopeName", scopeName ); }
        ;

nestedFunctionDef
                            { String declName; }
        :       ( "auto" )? //only for nested functions
                ( (functionDeclSpecifiers)=> ds:functionDeclSpecifiers
                )?
                declName = d:declarator[false]
                            {
                            AST d2, ds2;
                            d2 = astFactory.dupList(#d);
                            ds2 = astFactory.dupList(#ds);
                            symbolTable.add(declName, #(null, ds2, d2));
                            pushScope(declName);
                            }
                ( declaration )*
                            { popScope(); }
                compoundStatement[declName]
                            { ## = #( #[NFunctionDef], ## );}
        ;

statement
        :       SEMI                    // Empty statements
        
        |       compoundStatement[getAScopeName()]       // Group of statements

        |       expr SEMI!               { ## = #( #[NStatementExpr], ## );} // Expressions

// Iteration statements:

        |       "while"^ LPAREN! expr RPAREN! statement
        |       "do"^ statement "while"! LPAREN! expr RPAREN! SEMI!
        |!       "for"
                LPAREN ( e1:expr )? SEMI ( e2:expr )? SEMI ( e3:expr )? RPAREN
                s:statement
                                    {
                                        if ( #e1 == null) { #e1 = #[ NEmptyExpression ]; }
                                        if ( #e2 == null) { #e2 = #[ NEmptyExpression ]; }
                                        if ( #e3 == null) { #e3 = #[ NEmptyExpression ]; }
                                        ## = #( #[LITERAL_for, "for"], #e1, #e2, #e3, #s );
                                    }


// Jump statements:

        |       "goto"^ expr SEMI!
        |       "continue" SEMI!
        |       "break" SEMI!
        |       "return"^ ( expr )? SEMI!


        |       ID COLON! (options {warnWhenFollowAmbig=false;}: statement)?  { ## = #( #[NLabel], ## ); }
// GNU allows range expressions in case statements
        |       "case"^ ((constExpr VARARGS)=> rangeExpr | constExpr) COLON! ( options{warnWhenFollowAmbig=false;}:statement )?
        |       "default"^ COLON! ( options{warnWhenFollowAmbig=false;}: statement )?

// Selection statements:

        |       "if"^
                 LPAREN! expr RPAREN! statement  
                ( //standard if-else ambiguity
                        options {
                            warnWhenFollowAmbig = false;
                        } :
                "else" statement )?
        |       "switch"^ LPAREN! expr RPAREN! statement
        ;



conditionalExpr
        :       logicalOrExpr
                ( QUESTION^ (expr)? COLON conditionalExpr )?
        ;

rangeExpr   //used in initializers only
        :  constExpr VARARGS constExpr
                                { ## = #(#[NRangeExpr], ##); }
        ;

castExpr
        :       ( LPAREN typeName RPAREN )=>
                LPAREN^ typeName RPAREN ( castExpr | lcurlyInitializer )
                            { ##.setType(NCast); }

        |       unaryExpr
        ;
nonemptyAbstractDeclarator
        :   (
                pointerGroup
                (   (LPAREN  
                    (   nonemptyAbstractDeclarator
                        | parameterTypeList
                    )?
                    ( COMMA! )?
                    RPAREN)
                | (LBRACKET (expr)? RBRACKET)
                )*

            |   (   (LPAREN  
                    (   nonemptyAbstractDeclarator
                        | parameterTypeList
                    )?
                    ( COMMA! )?
                    RPAREN)
                | (LBRACKET (expr)? RBRACKET)
                )+
            )
                            {   ## = #( #[NNonemptyAbstractDeclarator], ## ); }
                                
        ;



unaryExpr
        :       postfixExpr
        |       INC^ castExpr
        |       DEC^ castExpr
        |       u:unaryOperator castExpr { ## = #( #[NUnaryExpr], ## ); }

        |       "sizeof"^
                ( ( LPAREN typeName )=> LPAREN typeName RPAREN
                | unaryExpr
                )
        |       "__alignof"^
                ( ( LPAREN typeName )=> LPAREN typeName RPAREN
                | unaryExpr
                )       
        |       gnuAsmExpr
        ;

unaryOperator
        :       BAND
        |       STAR
        |       PLUS
        |       MINUS
        |       BNOT    //also stands for complex conjugation
        |       LNOT
        |       LAND    //for label dereference (&&label)
        |       "__real"
        |       "__imag"
        ;

gnuAsmExpr
        :       "asm"^ ("volatile")? 
                LPAREN stringConst
                ( options { warnWhenFollowAmbig = false; }:
                  COLON (strOptExprPair ( COMMA strOptExprPair)* )?
                  ( options { warnWhenFollowAmbig = false; }:
                    COLON (strOptExprPair ( COMMA strOptExprPair)* )?
                  )?
                )?
                ( COLON stringConst ( COMMA stringConst)* )?
                RPAREN
                                { ##.setType(NGnuAsmExpr); }
        ;

//GCC requires the PARENs
strOptExprPair
        :  stringConst ( LPAREN expr RPAREN )?
        ;


primaryExpr
        :       ID
        |       Number
        |       charConst
        |       stringConst
// JTC:
// ID should catch the enumerator
// leaving it in gives ambiguous err
//      | enumerator
        |       (LPAREN LCURLY) => LPAREN^ compoundStatement[getAScopeName()] RPAREN
        |       LPAREN^ expr RPAREN        { ##.setType(NExpressionGroup); }
        ;
    

{
import isis.anp.common.CToken;
import isis.anp.common.LineObject;
import isis.anp.common.PreprocessorInfoChannel;

import java.io.InputStream;
import java.io.Reader;
import java.util.Hashtable;

import antlr.ANTLRHashString;
import antlr.ByteBuffer;
import antlr.CharBuffer;
import antlr.CharStreamException;
import antlr.CharStreamIOException;
import antlr.InputBuffer;
import antlr.LexerSharedInputState;
import antlr.NoViableAltForCharException;
import antlr.RecognitionException;
import antlr.Token;
import antlr.TokenStream;
import antlr.TokenStreamException;
import antlr.TokenStreamIOException;
import antlr.TokenStreamRecognitionException;
import antlr.collections.impl.BitSet;
}

class GnuCLexer extends StdCLexer;
options 
        {
        k = 3;
        importVocab = GnuC;
        testLiterals = false;
        }    
tokens {
        LITERAL___extension__ = "__extension__";
}

{
  public void initialize(String src)
  {
    setOriginalSource(src);
    initialize();
  }

  public void initialize() 
  {
    literals.put(new ANTLRHashString("__alignof__", this), new Integer(LITERAL___alignof));
    literals.put(new ANTLRHashString("__asm", this), new Integer(LITERAL_asm));
    literals.put(new ANTLRHashString("__asm__", this), new Integer(LITERAL_asm));
    literals.put(new ANTLRHashString("__attribute__", this), new Integer(LITERAL___attribute));
    literals.put(new ANTLRHashString("__complex__", this), new Integer(LITERAL___complex));
    literals.put(new ANTLRHashString("__const", this), new Integer(LITERAL_const));
    literals.put(new ANTLRHashString("__const__", this), new Integer(LITERAL_const));
    literals.put(new ANTLRHashString("__imag__", this), new Integer(LITERAL___imag));
    literals.put(new ANTLRHashString("__inline", this), new Integer(LITERAL_inline));
    literals.put(new ANTLRHashString("__inline__", this), new Integer(LITERAL_inline));
    literals.put(new ANTLRHashString("__real__", this), new Integer(LITERAL___real));
    literals.put(new ANTLRHashString("__signed", this), new Integer(LITERAL_signed));
    literals.put(new ANTLRHashString("__signed__", this), new Integer(LITERAL_signed));
    literals.put(new ANTLRHashString("__typeof", this), new Integer(LITERAL_typeof));
    literals.put(new ANTLRHashString("__typeof__", this), new Integer(LITERAL_typeof));
    literals.put(new ANTLRHashString("__volatile", this), new Integer(LITERAL_volatile));
    literals.put(new ANTLRHashString("__volatile__", this), new Integer(LITERAL_volatile));
  }


  LineObject lineObject = new LineObject();
  String originalSource = "";
  PreprocessorInfoChannel preprocessorInfoChannel = new PreprocessorInfoChannel();
  int tokenNumber = 0;
  boolean countingTokens = true;
  int deferredLineCount = 0;

  public void setCountingTokens(boolean ct) 
  {
    countingTokens = ct;
    if ( countingTokens ) {
      tokenNumber = 0;
    }
    else {
      tokenNumber = 1;
    }
  }

  public void setOriginalSource(String src) 
  {
    originalSource = src;
    lineObject.setSource(src);
  }
  public void setSource(String src) 
  {
    lineObject.setSource(src);
  }
  
  public PreprocessorInfoChannel getPreprocessorInfoChannel() 
  {
    return preprocessorInfoChannel;
  }

  public void setPreprocessingDirective(String pre)
  {
    preprocessorInfoChannel.addLineForTokenNumber( pre, new Integer(tokenNumber) );
  }
  
  protected Token makeToken(int t)
  {
    if ( t != Token.SKIP && countingTokens) {
        tokenNumber++;
    }
    CToken tok = (CToken) super.makeToken(t);
    tok.setLine(lineObject.line);
    tok.setSource(lineObject.source);
    tok.setTokenNumber(tokenNumber);

    lineObject.line += deferredLineCount;
    deferredLineCount = 0;
    return tok;
  }

    public void deferredNewline() { 
        deferredLineCount++;
    }

    public void newline() { 
    	super.newline();
        lineObject.newline();
    }






}
Whitespace
        :       ( ( ' ' | '\t' | '\014')
                | "\r\n"                { newline(); }
                | ( '\n' | '\r' )       { newline();    }
                )                       { _ttype = Token.SKIP;  }
        ;


protected
Escape
        :       '\\'
                ( options{warnWhenFollowAmbig=false;}: 
                  ~('0'..'7' | 'x')
                | ('0'..'3') ( options{warnWhenFollowAmbig=false;}: Digit )*
                | ('4'..'7') ( options{warnWhenFollowAmbig=false;}: Digit )*
                | 'x' ( options{warnWhenFollowAmbig=false;}: Digit | 'a'..'f' | 'A'..'F' )+
                )
        ;

protected IntSuffix
        :   'L'
            | 'l'
            | 'U'
            | 'u'
            | 'I'
            | 'i'
            | 'J'
            | 'j'
        ;
protected NumberSuffix
        :
            IntSuffix
            | 'F'
            | 'f'
        ;
    
Number
        :       ( ( Digit )+ ( '.' | 'e' | 'E' ) )=> ( Digit )+
                ( '.' ( Digit )* ( Exponent )?
                | Exponent
                ) 
                ( NumberSuffix
                )*

        |       ( "..." )=> "..."       { _ttype = VARARGS;     }

        |       '.'                     { _ttype = DOT; }
                ( ( Digit )+ ( Exponent )?
                                        { _ttype = Number;   }
                    ( NumberSuffix
                    )*
                )?

        |       '0' ( '0'..'7' )*       
                ( NumberSuffix
                )*

        |       '1'..'9' ( Digit )*     
                ( NumberSuffix
                )*

        |       '0' ( 'x' | 'X' ) ( 'a'..'f' | 'A'..'F' | Digit )+
                ( IntSuffix
                )*
        ;

IDMEAT
        :
                i:ID                {
                                        
                                        if ( i.getType() == LITERAL___extension__ ) {
                                                $setType(Token.SKIP);
                                        }
                                        else {
                                                $setType(i.getType());
                                        }
                                        
                                    }
        ;

protected ID
        options 
                {
                testLiterals = true; 
                }
        :       ( 'a'..'z' | 'A'..'Z' | '_' | '$')
                ( 'a'..'z' | 'A'..'Z' | '_' | '$' | '0'..'9' )*
        ;

WideCharLiteral
        :
                'L' CharLiteral
                                { $setType(CharLiteral); }
        ;



WideStringLiteral
        :
                'L' StringLiteral
                                { $setType(StringLiteral); }
        ;

StringLiteral
        :
                '"'
                ( ('\\' ~('\n'))=> Escape
                | ( '\r'        { newline(); }
                  | '\n'        {
                                newline();
                                }
                  | '\\' '\n'   {
                                newline();
                                }
                  )
                | ~( '"' | '\r' | '\n' | '\\' )
                )*
                '"'
        ;