Handle try-catch-finally blocks

[palashborhanuddin]

public static void TryCatchTestMethod() {
        Scanner keys = new Scanner(System.in);
        for (;;) {
            String nameOfFile = keys.nextLine().trim();
            if (nameOfFile.equalsIgnoreCase("quit")) {
                break;
            }
            File f = new File(nameOfFile);
            if (f.exists()) {
                if (f.isFile() && f.canRead()) {
                    Scanner input = null;
                    try {
                        input = new Scanner(f);
                        while (input.hasNextLine()) {
                            String contents = input.nextLine();
                            System.out.println(contents);
                        }
                    } catch (FileNotFoundException e) {
                        e.printStackTrace();
                    } finally {
                        if (input != null) {
                            input.close();
                        }
                    }
                } else if (f.isDirectory()) {
                    try {
                        System.out.println("File "
                                + f.getCanonicalPath()
                                + " is a directory");
                    } catch (IOException e) {
                        e.printStackTrace();
                    }
                }
            }
        }
    }

PDG generated using JDeodorant,

Question, why there is no PDG Node for the statements inside catch and finally blocks?

[tsantalis]

@palashborhanuddin JDeodorant models the entire try-catch structure as a single PDG Control Predicate node. For the sake of Groum generation, we don't care so much about how exactly the exception handling is done (in other words, what exactly happens inside the catch blocks). We mainly care about what exception types are handled. So, we can model try-catch as a single node in the Groum representation, and keep as metadata information the exception types handled in the catch blocks.

[palashborhanuddin]

Thanks, what about the finally block? For example, from the above code snippet, shouldn't we have the GROUM nodes for the possible ACTION/CONTROL nodes inside the finally block?

[tsantalis]

@palashborhanuddin The same principle holds for finally blocks. Typically, in the finally block developers handle some resources used within the try block (e.g., close streams, connections). With the new try-with-resources this handling is automatically done by Java. Therefore, in order to abstract the try blocks, we will omit the implementation details inside finally block. We can keep as metadata, only the fact that exists a finally block, in the same way we just kept the exception types being handled in the catch blocks.

[palashborhanuddin]

Thank you.

[palashborhanuddin]

nodes=[1 Scanner.<init>, 2 FOR, 4 Scanner.nextLine, 3 String.trim, 6 String.equalsIgnoreCase, 5 IF, 7 File.<init>, 9 File.exists, 8 IF, 12 File.canRead, 10 IF, 11 File.isFile, 13 Scanner.<init>, 15 Scanner.hasNextLine, 14 WHILE, 16 Scanner.nextLine, 19 File.isDirectory, 18 IF, 21 File.getCanonicalPath],
edges=[
, 1 Scanner.<init>-->2 FOR
, 2 FOR-->4 Scanner.nextLine
, 4 Scanner.nextLine-->3 String.trim
, 3 String.trim-->6 String.equalsIgnoreCase
, 6 String.equalsIgnoreCase-->5 IF
, 5 IF-->7 File.<init>
, 7 File.<init>-->9 File.exists
, 9 File.exists-->8 IF
, 8 IF-->12 File.canRead
, 8 IF-->11 File.isFile
, 12 File.canRead-->10 IF
, 11 File.isFile-->10 IF
, 10 IF-->13 Scanner.<init>
, 13 Scanner.<init>-->15 Scanner.hasNextLine
, 15 Scanner.hasNextLine-->14 WHILE
, 14 WHILE-->16 Scanner.nextLine
, 10 IF-->19 File.isDirectory
, 19 File.isDirectory-->18 IF
, 18 IF-->21 File.getCanonicalPath
, 1 Scanner.<init>-->3 String.trim
, 1 Scanner.<init>-->4 Scanner.nextLine
, 3 String.trim-->7 File.<init>
, 6 String.equalsIgnoreCase-->7 File.<init>
, 7 File.<init>-->12 File.canRead
, 2 FOR-->8 IF
, 7 File.<init>-->11 File.isFile
, 9 File.exists-->12 File.canRead
, 9 File.exists-->11 File.isFile
, 7 File.<init>-->13 Scanner.<init>
, 2 FOR-->10 IF
, 9 File.exists-->13 Scanner.<init>
, 12 File.canRead-->13 Scanner.<init>
, 8 IF-->10 IF
, 11 File.isFile-->13 Scanner.<init>
, 7 File.<init>-->19 File.isDirectory
, 9 File.exists-->19 File.isDirectory
, 12 File.canRead-->19 File.isDirectory
, 11 File.isFile-->19 File.isDirectory
, 7 File.<init>-->21 File.getCanonicalPath
, 2 FOR-->18 IF
, 19 File.isDirectory-->21 File.getCanonicalPath
, 10 IF-->18 IF
, 9 File.exists-->21 File.getCanonicalPath
, 12 File.canRead-->21 File.getCanonicalPath
, 8 IF-->18 IF
, 11 File.isFile-->21 File.getCanonicalPath
, 13 Scanner.<init>-->16 Scanner.nextLine
, 10 IF-->14 WHILE
, 15 Scanner.hasNextLine-->16 Scanner.nextLine

Node and edges for the above test code.

Note that, there is no node TRY, CATCH or FINALLY statement (also statements inside CATCH and FINALLY) as there is no PDG edge for these!