ProgramParser.java 文件源码

public Program parse(CharStream stream) throws IOException {
    /*
    // In order to require less memory: use unbuffered streams and avoid constructing a full parse tree.
    ASPCore2Lexer lexer = new ASPCore2Lexer(new UnbufferedCharStream(is));
    lexer.setTokenFactory(new CommonTokenFactory(true));
    final ASPCore2Parser parser = new ASPCore2Parser(new UnbufferedTokenStream<>(lexer));
    parser.setBuildParseTree(false);
    */
    CommonTokenStream tokens = new CommonTokenStream(
        new ASPCore2Lexer(stream)
    );
    final ASPCore2Parser parser = new ASPCore2Parser(tokens);

    // Try SLL parsing mode (faster but may terminate incorrectly).
    parser.getInterpreter().setPredictionMode(PredictionMode.SLL);
    parser.removeErrorListeners();
    parser.setErrorHandler(new BailErrorStrategy());

    final CustomErrorListener errorListener = new CustomErrorListener(stream.getSourceName());

    ASPCore2Parser.ProgramContext programContext;
    try {
        // Parse program
        programContext = parser.program();
    } catch (ParseCancellationException e) {
        // Recognition exception may be caused simply by SLL parsing failing,
        // retry with LL parser and DefaultErrorStrategy printing errors to console.
        if (e.getCause() instanceof RecognitionException) {
            tokens.seek(0);
            parser.addErrorListener(errorListener);
            parser.setErrorHandler(new DefaultErrorStrategy());
            parser.getInterpreter().setPredictionMode(PredictionMode.LL);
            // Re-run parse.
            programContext = parser.program();
        } else {
            throw e;
        }
    }

    // If the our SwallowingErrorListener has handled some exception during parsing
    // just re-throw that exception.
    // At this time, error messages will be already printed out to standard error
    // because ANTLR by default adds an org.antlr.v4.runtime.ConsoleErrorListener
    // to every parser.
    // That ConsoleErrorListener will print useful messages, but not report back to
    // our code.
    // org.antlr.v4.runtime.BailErrorStrategy cannot be used here, because it would
    // abruptly stop parsing as soon as the first error is reached (i.e. no recovery
    // is attempted) and the user will only see the first error encountered.
    if (errorListener.getRecognitionException() != null) {
        throw errorListener.getRecognitionException();
    }

    // Abort parsing if there were some (recoverable) syntax errors.
    if (parser.getNumberOfSyntaxErrors() != 0) {
        throw new ParseCancellationException();
    }

    // Construct internal program representation.
    ParseTreeVisitor visitor = new ParseTreeVisitor(externals);
    return visitor.translate(programContext);
}