Fixed flaky test DataStoreConvertersTest.testCheckSameKey()

[mihirgune]

Code Change

Fixed flaxy test - https://github.com/GoogleCloudPlatform/DataflowTemplates/blob/main/v1/src/test/java/com/google/cloud/teleport/templates/common/DatastoreConvertersTest.java#L174

Error

This test is flaky because while verifying if the errorTag matches the expectedErrors, the code does a direct string comparison, in which both strings are JSON objects in string format. As the ordering of the keys inside JSON objects is non-deterministic, this string comparison is throwing an error.

This error was found by executing the NonDex tool with the following command -

mvn -pl v1 edu.illinois:nondex-maven-plugin:2.1.7:nondex -Dtest=com.google.cloud.teleport.templates.common.DatastoreConvertersTest#testCheckSameKey

Expected: iterable with items ["{\"message\":\"Duplicate Datastore Key\",\"stacktrace\":null,\"data\":\"{\\\"key\\\":{\\\"partitionId\\\":{\\\"projectId\\\":\\\"my-project\\\",\\\"namespaceId\\\":\\\"some-namespace\\\"},\\\"path\\\":[{\\\"kind\\\":\\\"monkey\\\",\\\"id\\\":\\\"1234\\\"}]},\\\"properties\\\":{\\\"someString\\\":{\\\"stringValue\\\":\\\"someValue\\\"}}}\"}", "{\"message\":\"Duplicate Datastore Key\",\"data\":\"{\\\"key\\\":{\\\"partitionId\\\":{\\\"projectId\\\":\\\"my-project\\\",\\\"namespaceId\\\":\\\"some-namespace\\\"},\\\"path\\\":[{\\\"kind\\\":\\\"monkey\\\",\\\"id\\\":\\\"1234\\\"}]},\\\"properties\\\":{\\\"SomeBSProp\\\":{\\\"stringValue\\\":\\\"Some BS Value\\\"}}}\",\"stacktrace\":null}"] in any order
     but: not matched: "{\"data\":\"{\\\"key\\\":{\\\"partitionId\\\":{\\\"projectId\\\":\\\"my-project\\\",\\\"namespaceId\\\":\\\"some-namespace\\\"},\\\"path\\\":[{\\\"kind\\\":\\\"monkey\\\",\\\"id\\\":\\\"1234\\\"}]},\\\"properties\\\":{\\\"someString\\\":{\\\"stringValue\\\":\\\"someValue\\\"}}}\",\"message\":\"Duplicate Datastore Key\",\"stacktrace\":null}

Fix

Discarded Probable Fixes

• The obvious fix for this is to convert both strings into JSON objects and do a comparison. However, a codec that will that will enable one to compare two JSONs does not exist in the project, and defining one would be a very invasive change. To combat this, I am comparing the HashCodes for both JSONs in list format.
• Another possible fix was to edit the https://github.com/GoogleCloudPlatform/DataflowTemplates/blob/587a0a85265e77529a3d8f99b56d65884de7d762/v1/src/main/java/com/google/cloud/teleport/templates/common/DatastoreConverters.java#L487 EntityJsonPrinter, which is used by the project to convert jsons into strings and ensure that the keys were sorted. However, this code is not isolated in the test logic, and would have impacted the main logic too.

Proposed Fix

Instead of comparing two jsons, I am comparing the hashcodes for those JSONs. The google.gson JsonObject which I am using uses a LinkedTreeMap to store its members, which will automatically handle sorting, thus making the hashcode reliable.