Forum Discussion
Tuomas - You could simply enable the "Null-safe access" checkbox in the Mapper. With that enabled, whenever you reference a field that doesn't exist, it simply assigns it as null.
Hi koryknick ! Thanks for response!
Yes, I have the Null-safe access enabled. In this case it doesn't work though. I have "labels" as my output field, not "fi_FI" and "en_US". Thus, when "labels" in input is completely missing, then I will have null output for "labels" but also I'm missing both the properties. Sure, one way to do it would be to map each property to its own output field which I have done in the past when facing the similar problem. But I figured maybe I should find an alternative solution as I don't really need the "labels"-object to be split. Also, if I hard code the "labels"-object, I might lose something in the future if more properties are suddenly added to it.
Essentially, the missing properties were problematic when loading data to Big Query. An object with one of the properties missing happened to be the first object loaded which the connector apparently uses to determine Big Query table's schema. Big Query adds objects to table as "labels.fi_FI", "labels.en_US", therefore a property missing from first object created the table without that column and breaking the load for subsequent objects.
Hopefully this clarified the issue 🙂