collection.document() to assign a random ID to the document reference about fake_cloud_firestore

[suztomo]

But sometimes there isn't a meaningful ID for the document, and it's more convenient to let Cloud Firestore auto-generate an ID for you. You can do this by calling add()

https://firebase.google.com/docs/firestore/manage-data/add-data

[atn832]

You're right, a semi-random id is created in one of two cases:

1. collectionRef.add(): already roughly supported at https://github.com/atn832/cloud_firestore_mocks/blob/878187ec0d94c26f921196b4f53317305db781f4/lib/src/mock_collection_reference.dart#L43
2. collectionRef.document(): not supported yet

https://pub.dev/documentation/cloud_firestore/latest/cloud_firestore/CollectionReference/document.html:

DocumentReference document (
[String path]
)
If no path is provided, an auto-generated ID is used.

To fix this issue, we should make it so that mockCollectionRef.document() generates some ID, then returns a MockDocumentRef for that ID, perhaps something like:

  @override
  DocumentReference document([String path]) {
    if (path == null) {
      path = // make some id
    }
    return MockDocumentReference(path, getSubpath(root, path), root,
        getSubpath(snapshotStreamControllerRoot, path));
  }

Do you have such a use case in your projects? I have never run into it because I always use add().

[suztomo]

The test case reflects my use case. I don’t call “add” when setData is enough.

While it’s feasible to change my code, I prefer cloud_firestore_mocks behaves the same as real firestore as much as possible.

Memo:

The random key (e.g., 0XZfFoJICLrYBtIQqCHP) contains 20-characters. Number, a-z, A-Z

[suztomo]

In real Firestore, document('nonexistent key').get returns a future of documentSnapshot but the document is not in Firebase Database.

    final x = await _personsCollection.document('nonexistentkey').get();

Without specifying documentId, document().get() returns the same. Note that in both cases, 'exists' fields are false.