Mongo not removing fields with empty strings on upsert


I’m using SimpleSchema for modeling a Collection.
I’m retrieving data from an API and have an edge case.

  1. gets set to "bar" after first request
  2. “foo” in external API gets deleted
  3. 2nd API request is made and “foo” comes over as “” (empty string)
  4. Call CollectionSchema.clean(Document)
  5. App upserts new Document instance where foo should be '' instead of bar
  6. Bug: bar is left as the value for

So issue seemed simple enough in that SimpleSchema didn’t want to insert '' into the DB and just retained its original value.

However, I noticed that clean has a removeEmptyString option set to true by default.
[](http://SimpleSchema Clean)

Therefore, should just be deleted.

Based on my debugging, this seems to work. I did a console log of after the call to clean and it shows up as undefined indicating the property was removed.

I then printed out the same thing after the upsert call and verified the original value in the DB is what was still there. So seems like issue is with the Mongo.upsert

const docUpsert= Collection.upsert( { someUniqueKey: apiDoc.someUniqueKey }, { $setOnInsert: { createdBy: Meteor.userId(), createdDate: new Date() }, $set: apiDoc });

The $set seems to be ignoring/overriding the new empty string value for foo and SimpleSchema's rules.

I’m seeing collection2 also has its own removeEmptyStrings option and that its default is also true. So this should’ve just always worked. My guess is it’s something with a mixture of using SimpleSchema to clean and then collection2 on the upsert which does its own cleaning/validating.

Even setting { removeEmptyStrings: true } in the upsert call does not work. Could this be a bug for upsert?

Does anybody know how to resolve this? Thanks!