Unable to Search for UUIDs in MongoDB Clusters

[tnaum-ms]

A customer reported an issue where they are unable to search for UUIDs in MongoDB Clusters (Version 3 or higher). The customer's unique IDs are UUID-based, and this limitation is impacting their workflows.

Steps to Reproduce:

1. Use MongoDB Cluster (Version 3 or higher).
2. Attempt to search for documents with UUID-based unique IDs in the cluster.
3. Observe that the search does not return expected results.

Expected Behavior:
Search functionality should support UUID-based identifiers to ensure all document types can be queried effectively.

Impact:

• Users relying on UUIDs as their unique identifiers cannot effectively search for their data.
• Affects workflows for customers using MongoDB Clusters with UUID-based IDs.

[tnaum-ms]

This is a larger task to complete. Currently, we have limited support for complex BSON data types (UUID, Binary, DBRef, Decimal128, RegExp, etc.).

We are able to display them in all views, but we're showing their internal (behind-the-scenes) representation, rather than their end-user form.

For example:

"known_uuid": UUID("7643ba8b-7ad6-405a-8627-986b43d4a609")

is actually stored as:

"known_uuid": {
    "$binary": {
        "base64": "dkO6i3rWQFqGJ5hrQ9SmCQ==",
        "subType": "04"
    }
}

Similarly, a Decimal128 field:

"price": Decimal128("99.99")

is stored as:

"price": {
    "$numberDecimal": "99.99"
}

Or a DBRef (a reference to another document):

"user": DBRef("users", ObjectId("507f1f77bcf86cd799439011"))

which is actually stored as:

"user": {
    "$ref": "users",
    "$id": {
        "$oid": "507f1f77bcf86cd799439011"
    }
}

The Challenge

The challenge here is to provide a set of converters that would process queries containing special BSON functions (like UUID(...) or Decimal128(...)) and construct the correct BSON objects to be used in find operations.

Possible Approaches

We have two possible solutions:

1. Internal String Handling: Detect known formats and automatically convert them into BSON objects when needed.
2. Query Interpreter: Implement an interpreter that processes input, allowing not only known functions like UUID(...) but also JavaScript expressions, enabling features like Date() to get the current timestamp in queries.

Both approaches require significant work, but finding a solution is essential, and we'll revisit this soon.