我想使用ANTLR4从Java源文件中提取所有方法名称和变量名称

Question

Basically I want to extract all variable names, irrespective of their scope. And all function/method names inside the source code.

For the given input,

     class temp{
         int a;

    public static void main(String args[]){
    int b = 0;
temp ob = new temp();
temp.printob();

    }

void printob(){
System.out.print("-");
}
        }

The output should be something like:

variables = {"a","b","ob"}

methods = {"main","printob"}

Answer 1

One way to achieve this is making small changes to the grammar specification of the said language in this case Java.

What we can do is create a global arraylist and insert all identifiers into it.

In the grammar, for each rule wherever Identifier is used for example,

methodcall : return_types IDENTIFIER LEFTPAREN params RIGHTPAREN;

we will make the following changes(where our global arraylist is called all_identifiers)

methodcall : return_types IDENTIFIER LEFTPAREN params RIGHTPAREN{

 all_identifiers.add($IDENTIFIER);

}; 

similarly adding the above code, to each rule having Identifier we will be able to extract all the methods and variable names (if you want them in different arraylists create two arraylist one for method calls and one for variable declaration)

Additional Note:

When I originally posted this question I wanted to find a way to change all method call names and all variable names to some pre-defined names to normalise the code. For example in the code I wanted to change int a,b,c; to something like int varbl,varbl,varbl; and similarly for methods I wanted to change all method names to mthd.

So the best way I found to achieve this was, 1. Identify all rules, where we want to change some Identifier 2. In each of the rule add similar code section (this step needs to be done as token object by it self is not editable so we type cast it to CommonToken object which gives us ability to set the test using setText() method.)

{
 CommonToken tkn_tmp = $IDENTIFIER;
 tkn_tmp.setText("varbl"); 
 // or if it's a method rule 
 //tkn_tmp.setText("mthd");
 }

3. Now all the tokens will be changed from their original form to the value we are setting in.

4. After this you need to parse the input code through the grammar, and the parse tree along with parser text will be updated(along with token start and token end pointers)

There is another way to achieve the same, which is by iterating through all the tokens sequentially and adding adding token in an arrylist, and checking for token type. If the token type is Identifier change the text to whatever you want to, and then as usual append it to the array list.

For segregation of method names and variable names you might need to change the grammar so that, variable Identifier and method Identifier are differentiable.