Conditional nodes

back to main results

To evaluate conditional nodes we provide two variations of the if-else statement in both Dart (in Snippets \ref{lst:ifdA} \ref{lst:ifdB}), and Kotlin (in Snippets \ref{lst:ifkA} and \ref{lst:ifkB}). These code snippets are defined to exhibit Type 1 clones in the else branch, and Type 2 clones in the if condition.

Snippet 1. Dart version A if example with Type-2 (with Snippet 2) clone in conditional variable.

 String conditional() {
  var i;

  if (i > 10) { return "Hi"; }
  else { return "Bye"; }
  
 }

Snippet 2. Dart version B if example with Type-2 (with Snippet 1) clone in conditional variable.

String conditional() {
  var j;
  
  if (j > 10) { 
    return "Hi"; 
  }
  else { 
    return "Bye"; 
  }
}

Snippet 3. Kotlin version A if example with Type-2 (with Snippet 4) clone in conditional variable

fun conditional() : String {
    val i : Int

    if (i > 10) {
        return "Hi"
    } else {
        return "Bye"
    }
}

Snippet 4. Kotlin version B if example with Type-2 (with Snippet 3) clone in conditional variable

fun conditional() : String {
    val j : Int
    if (j > 10) { return "Hi" }
    else { return "Bye" }
}

Analyzing the snippets for a single language, our algorithm is able to detect both, the Type-1 and Type-2 clones. In the case of the Kotlin snippets, a Type-2 clone was found in the first line for the variable definitions, i and j, while Type-1 clones were found across the rest of the snippet. The same exact analysis applies for the Dart snippets.

Given the nature of the examples and the similarity of the if abstraction in both programming languages, when versions A are analyzed across languages, we detect Type 1 clones in the complete snippet. The same result ie yield when analyzing versions B of the if abstraction. Finally, the outcome of comparing any A vs any B versions across languages will find the inserted Type 2 clone for the variables i and j in the condition.

Same-language evaluation

In teh case of conditionals we purpusly add two version of the same program, with the existence of two Type 2 clones (the var defition) as shown in the previous snippets. For the comparison of the same langauge experiments, we compare both version A, version B, and version A v. version B for each language.

Language	Total	Type 1	Type 2	Type 3	FP	FN	Precision	Recall
Dart A	15	0	0	0	0	0	1	1
Dart B	15	0	0	0	0	0	1	1
Dart A v. B	15	13	2	0	0	0	1	1
Kotlin A	16	0	0	0	0	0	1	1
Kotlin B	16	0	0	0	0	0	1	1
Kotlin A v. B	16	14	2	0	0	0	1	1

Cross-language evaluation

The results of the cross-language comparison take into account the A v. A, B v. B and A v. B versions of the program. The table below shows the summary of the results.

Language	Total	Type 1	Type 2	Type 3	FP	FN	Precision	Recall
Dart v. Kotlin A	5	3	2	0	1	0	0.8	1
Dart v. Kotlin B	5	3	2	0	1	0	0.8	1
Dart A v. Kotlin B	5	2	3	0	1	0	0.8	1
Dart B v. Kotlin A	5	2	3	0	1	0	0.8	1

The FP obtained in the last two cases comes from the mis classification of LITERALS AND IDENTIFIERS as Type 2 clones (they should be Type 3 clones). The algorithm correcly captures teh Type 2 clone in the variable declarations for the A v. B comparison, while detecting a Type 1 clone in the same version of the programs.