IntelliJ IDEA 2021.3 Help

Analyze data flow

When working with large codebases, it is sometimes difficult to figure out how data is processed and how the workflows could be improved to make the code more performant and readable. To facilitate this, IntelliJ IDEA dataflow analysis enables you to trace all the possible data transformations without running the program. The information can be used to improve the design of the app and diagnose bugs before they manifest themselves.

Data flow analysis provides the information on:

  • What happens to the data downstream of a method or expression: what are the consumers and what possible values can be produced.

  • All the possible input values a method can have and where particular values are coming from.

  • Whether a variable can possibly be null. Using this information you can prevent unexpected NullPointerExceptions and optimize your workflows by removing redundant null checks and @Nullable annotations.

View analysis results

  1. Place the caret at an identifier that represents the data you want to analyze. You can choose to analyze the symbol at a declaration, in a statement, in the parameters of a method, and so on.

    Dfa caret
  2. From the main menu select Code | Analyze Code | Data Flow to Here to analyze data upstream (producers) or Code | Analyze Code | Data Flow from Here to analyze data downstream (consumers).

  3. Specify the scope of the analysis. If you want to exclude tests, clear the Include test sources checkbox.

    Also, if you are interested in a particular value or expression result, you can specify it in the Filter field to only show the relevant results (available in Dataflow to here).

    Filter examples

    null/non-null values

    null !null

    String

    "Hello"

    enum

    SPRING SUMMER FALL WINTER

    boolean

    true false

    int/long

    0 >0 <=100 !=9

    Dfa scope

A tool window opens containing the results of the analysis. They are organized in nodes, each representing a data flow step.

Dfa results

In the example:

  • The getComplete() method returns the value of the complete variable.

  • The complete variable can be assigned null during initialization or get any value in the setComplete method.

  • The setComplete() method is called at lines 17 and 48 and assign the values false and true respectively.

Refresh results

  • If the code has changed, and you want to analyze the same expression again, click Refresh in the Analyze tool window.

Analyze possible values

When viewing Data flow to here, you can group the nodes by value to get the summary on the possible values or analyze their origin.

  • To get the information about specific values, click Group by leaf expression in the left part of the Analyze tool window.

  • To get the information about null/non-null values, click Group by leaf expression nullness in the left part of the Analyze tool window.

Export to file

If you want to share the results of the analysis in text format, use the Export option.

  1. Select the analysis tab you want to export.

  2. Click Export to text file Icons toolbar decorator export in the left part of the Analyze tool window.

  3. If you want to copy the results to clipboard, click Copy. To export the results to a file, select the file in the Export to file field and click Save.

Analyze stack traces

When your program crashes with an exception, you can use the stack trace as the input for data flow analysis. This helps you track where inappropriate values may come from.

public class ArrayTest { static int[] ints = new int[5]; public static void main(String[] args) { ArrayTest arrayTest = new ArrayTest(); int a = arrayTest.getRandomElement(ints); System.out.println(a); } private int getRandomElement(int[] array) { int index = new java.util.Random().nextInt(20); return array[index]; } }

The code above creates a fixed-size array and tries to access a random element from it. Sometimes it throws an ArrayOutOfBoundsException because the index may be greater than the length of the array.

  1. In the stack trace, click the source reference of the frame that threw the exception.

    Dfa stacktrace

    The editor takes you to the corresponding line in the source.

  2. Click the popup to find the source of the value that caused the exception.

    Dfa stacktrace 2

Dataflow to here opens with the filters applied.

Last modified: 01 August 2022