Question

I use the grammar Java.g from the ANTLR wiki produces a lexer and parser for Java source files.Then use the following code to generate an abstract syntax tree (AST).

    ANTLRInputStream input = new ANTLRInputStream(new FileInputStream(fileName));
    JavaLexer lexer = new JavaLexer(input);     // create lexer
    // create a buffer of tokens pulled from the lexer
    CommonTokenStream tokens = new CommonTokenStream(lexer);
    JavaParser parser = new JavaParser(tokens); // create parser
    JavaParser.javaSource_return r = parser.javaSource();   // parse rule 'javaSource'
    /*RuleReturnScope result = parser.compilationUnit();
    CommonTree t = (CommonTree) result.getTree();*/
    // WALK TREE
    // get the tree from the return structure for rule prog
    CommonTree t = (CommonTree)r.getTree();

Then modify the AST. For example,replace "File file = new File(filepath, fileType);" to "S3Object _file = new S3Object(_fileName);" by modify the AST node. After this,I want to translate this AST to java source code.I modify the JavaTreeParser.g and write a stringtemplate and use the following method to get the java source code:

    FileReader groupFileR = new FileReader("src/com/googlecode/zcg/templates/JavaTemplate.stg");
    StringTemplateGroup templates = new StringTemplateGroup(groupFileR);
    groupFileR.close();
    // create a stream of tree nodes from AST built by parser
    CommonTreeNodeStream nodes = new CommonTreeNodeStream(t);
    // tell it where it can find the token objects
    nodes.setTokenStream(tokens);
    JavaTreeParser walker = new JavaTreeParser(nodes); // create the tree Walker
    walker.setTemplateLib(templates); // where to find templates
    // invoke rule prog, passing in information from parser
    JavaTreeParser.javaSource_return r2 = walker.javaSource();

    // EMIT BYTE CODES
    // get template from return values struct
    StringTemplate output = (StringTemplate)r2.getTemplate(); 
    System.out.println(output.toString()); // render full template

If I don't modify the AST,it will get the java source code correctly,but after I modify the AST,it doesn't get the right java source code(the AST was modified correctly).For example,if I input the following souce code,and translate to AST,then modify "File file = new File(filepath, fileType);" to "S3Object _file = new S3Object(_fileName);":

public void methodname(String address){
    String filepath = "file";
    int fileType = 3;       
    File file = new File(filepath, fileType);
}

the result will be the following:

public void methodname( String address)
  { 
     String filepath="file";
     int fileType=3;
     methodname (Stringaddress){Stringfilepath;//it's not  what I wanted
  }

Am I doing it wrong? Is there a more proper way for me to solve this problem?

Was it helpful?

Solution

unfortunately I cannot recommend doing source to source translation by rewriting the abstract syntax trees; try using the parse trees. If I remember ANTLR 3 can also generate those easily. Ter

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top