During development I sometimes find it handy to put debug logging statements in my code that output whole object structures as JSON strings to see what an object looks like at certain points in the code execution. Of course I could use the debugger to inspect those objects, but at times I find using the console for debug output more convenient. Also, using the debugger is not possible if it's necessary to reproduce a problem that only occurs in an environment different from your debugger equipped IDE, like inside a docker container.

So, in order to make outputting an object as easy as possible, I create a method that uses the Jackson library to generate a JSON representation for any object that it's given:

package com.anothernode.example;

import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.ObjectWriter;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

public final class LoggingHelper {

  private static Logger logger =

  public static String getJsonRepresentation(Object object) {
    ObjectWriter writer = new ObjectMapper()
    String representation = "JSON representation of ["
        + object.getClass().getSimpleName()
        + "] object:\n";
    try {
      representation += writer.writeValueAsString(object);
    } catch (JsonProcessingException e) {
      logger.error("Creating JSON representation for an object failed");
    return representation;

Not very exciting, I guess. The problem with this is though, that there might be data in the object that is not suitable for printing it out to a console, like in my case up to a few megabyte large files represented as Base64 encoded strings that would render the output virtually useless by cluttering the console or log file with huge amounts of garbage in terms of human readability.

In order to still make use of the helper method, what's needed is a way to exclude fields of objects from getting included in the JSON string representation. The easiest, least invasive and most elegant way I found to do this is using JSON Views and that's what I found interesting enough to take a note about here.

In the simplest case, where you are not using JSON Views already in your project, you only have to take a few easy steps to exclude fields from the generated JSON representations. First, you need to add a marker class like this one to your project:

package com.anothernode.example.json;

public class View {

  private View() {}

  public static interface ExcludedFromLogging {}


With the marker in place you can then mark fields with it in any domain classes you might want to output:

package com.anothernode.example.domain;

class Foo {

  private String bar;

  private String baz;


Now all that's left to do is changing the ObjectWriter creation in the LoggingHelper.getJsonRepresentation(Object object) method like this:

public static String getJsonRepresentation(Object object) {
  ObjectWriter writer = new ObjectMapper()

Henceforth the field Foo.bar will not show up in any JSON representation of Foo generated with the LoggingHelper.getJsonRepresentation(Object object) method.

This apporach exploits the fact that the JSON Views API has a default view that includes everything that is not part of any explicitly defined view. We use that default view by giving Object.class as an parameter into the writerWithView() method.

So by adding the field Foo.bar to the view ExcludedFromLogging, we effectively blacklist it from the default view. If you are using JSON Views already for something different in the project, you possibly might not be able to use the default view as described here. That's because every field that is part of any view would be excluded from the default view, which might not be what you want.

But of course in this situation it would still be possible to flip this around and use an IncludedInLogging view instead. This IncludedInLogging.class Would then have to be passed to writerWithView() instead of Object.class. This means probably a bit more work though because every field that should be included in the JSON representation would then have to be whitelisted by marking it with the IncludedInLogging view.

JSON Views are surely also very useful for many other use cases and they seem to be well supported by Spring, too.