Cfrgtkky

I'm creating a configurable code formatter in Java with ANTLR4. I was able to create the parser and the lexer out of the grammar file found on github (https://github.com/antlr/grammars-v4/tree/master/plsql)

This way I also found out, why everyone said this will be a lot of work. I was a little bit overwhelmed at first about the sheer mass of functions created by ANTLR.
But after a while i found out that, (most) of the Enter and Exit-Trigger are needed for formatting PLSQL.

So I'm starting with this, to read a file:

    CharStream s = CharStreams.fromPath(Paths.get("D:\dev\sample\green3.sql"));

    CaseChangingCharStream upper = new CaseChangingCharStream(s, true);

    PlSqlLexer lexer = new PlSqlLexer(upper);

    PlSqlParser parser = new PlSqlParser(new CommonTokenStream(lexer));

    ParseTree tree = parser.sql_statement();

    //ParseTree tree = parser.sql_script();

    ParseTreeWalker walker = new ParseTreeWalker();

    PlSqlParserBaseListener listener = new PlSqlParserBaseListener();



    walker.walk(listener, tree);

    String output = listener.getOutput();

    System.out.println("[" + output +  "]");

So i need to give the ParseTree the information if it is either SQL-script or just a select-statement. Thats ok, i want to be able to do both variants, but with Implementation I'm starting with select-statements only.

e.g. SELECT and FROM have no special rule for themselves so I write "SELECT" to output:

    @Override public void enterFrom_clause(PlSqlParser.From_clauseContext ctx) {

        indentlevel--;

        output = output + " " + "nFROM" + " ";

    }

Or if i find multiple output columns in the, i need to find out if it's the first column (this gets set in a different function: enterSelect_list_elements)

    @Override public void enterExpression(PlSqlParser.ExpressionContext ctx) {

        if (firstcol) {

            output = output + " " + ctx.getText();

            indentlevel++;

            firstcol= false;

        }

        else {

        output = output + ", " + ctx.getText();

        }



    }

My PlSqlParserBaseListener feels like a state machine now. I have multiple states for when i need to indent or when specific tasks need to be done.
Is this the way to do it? I'm asking, because this is gonna be a lot of work and a lot of things and to consider. Therefore i want to do it the "right way" and create a efficient solution.
My aim is to make it possible to specify spacing, lower/upper case, operators, punctuation, list arrangements and maybe alignment of comments.

Thanks for your feedback! Greetings from Austria