cars: [Ford, BMW] - [Ford, Audi]
address: {street=Second Street, city=New York} - {street=Main Street, city=New York}
children: [{name=Sara, age=5}, {name=Alex, age=3}] - [{name=Peter, age=5}, {name=Cathy, age=10}]

As we can see, this shows that thecars, address, andchildren fields are different, and the differences are listed. However, this doesn’t point out which nested fields are leading to these differences. For example, it doesn’t point out that thestreet field in theaddress objects is different.

3.2. Flattening Maps

To precisely point out differences between nested Maps, we’ll flatten the Maps so each key is a path to the value. For example, thestreet key in theaddress object will be flattened toaddress.street and so on.

Let’s look at the code for this:

class FlattenUtils {
    public static Map<String, Object> flatten(Map<String, Object> map) {
        return flatten(map, null);
    }

    private static Map<String, Object> flatten(Map<String, Object> map, String prefix) {
        Map<String, Object> flatMap = new HashMap<>();
        map.forEach((key, value) -> {
            String newKey = prefix != null ? prefix + "." + key : key;
            if (value instanceof Map) {
                flatMap.putAll(flatten((Map<String, Object>) value, newKey));
            } else if (value instanceof List) {
                // check for list of primitives
                Object element = ((List<?>) value).get(0);
                if (element instanceof String || element instanceof Number || element instanceof Boolean) {
                    flatMap.put(newKey, value);
                } else {
                    // check for list of objects
                    List<Map<String, Object>> list = (List<Map<String, Object>>) value;
                    for (int i = 0; i < list.size(); i++) {
                        flatMap.putAll(flatten(list.get(i), newKey + "[" + i + "]"));
                    }
                }
            } else {
                flatMap.put(newKey, value);
            }
        });
        return flatMap;
    }
}

Here, we’re using recursion to flatten the Map. For any field, one of the following conditions will be true:

• The value could be a Map (nested JSON object). In this case, we’ll recursively call the flatten() method with the value as the parameter. For example, theaddress object will be flattened toaddress.street andaddress.city.
• Next, we can check if the value is aList (JSON array). If the list contains primitive values, we’ll add the key and value to the flattened Map.
• If the list contains objects, we’ll recursively call theflatten() method with each object as the parameter. For example, thechildren array will be flattened tochildren[0].name, children[0].age, children[1].name and children[1].age.
• If the value is neither aMap nor aList, we’ll add the key and value to the flattened Map.

This will be recursive until we reach the last level of the Map. At this point, we’ll have a flattened Map with each key as a path to the value.

3.3. Testing

Now that we’ve flattened the Maps, let’s look at how we can compare them usingMaps.difference():

@Test
void givenTwoJsonFiles_whenCompared_thenTheyAreDifferent() throws IOException {
    Map<String, Object> firstFlatMap = FlattenUtils.flatten(JsonUtils.jsonFileToMap("src/test/resources/first.json"));
    Map<String, Object> secondFlatMap = FlattenUtils.flatten(JsonUtils.jsonFileToMap("src/test/resources/second.json"));

    MapDifference<String, Object> difference = Maps.difference(firstFlatMap, secondFlatMap);
    difference.entriesDiffering().forEach((key, value) -> {
        System.out.println(key + ": " + value.leftValue() + " - " + value.rightValue());
    });
    assertThat(difference.areEqual()).isFalse();
}

Again, we’ll print the keys and values that are different. This leads to the output below:

cars: [Ford, BMW] - [Ford, Audi]
children[1].age: 3 - 10
children[1].name: Alex - Cathy
address.street: Second Street - Main Street
children[0].name: Sara - Peter

4. 結論

在本文中，我們探討了如何在Java中比較兩個JSON文檔。我們研究了將JSON文檔轉換為Map的不同方法，然後使用Guava的Maps.difference()方法進行比較。我們還研究瞭如何扁平化Map以進行嵌套Map的比較。