The above answer is indeed correct. Here's one more option if you want to leverage some of the connectors we do have.
Azure CosmosDB offers a migration tool that will allow you to import data from a MongoDB into a CosmosDB collection of your choice. This data is accessible via the MongoDB API offered by CosmosDb (which means in theory any other application you own that needs to access this data can be reworked to reach into CosmosDB with minimal overhead).
If this is a feasible option, Azure Cognitive search indexers can then connect to the Azure CosmosDB via the MongoDB API - documentation here
This is a feature currently in gated preview, and if this interests you we highly recommend you sign up here: https://aka.ms/azure-cognitive-search/indexer-preview.
I know this solution also needs you to port over your data from Atlas into Azure, but do give it a try if it's not too much overhead. (and this will save you time to manually extract data from MongoDB into your search index via the push API)