While trying to write mapping within a Copy shape in Azure Data Factory, I hit a stumbling block with generated JSON schemas. The problem occurs when you generate a schema containing an array, and the first instance of that array happens to be empty in your sample data.
The simplest way to generate the schema for your JSON dataset is to allow ADF to connect to the source and interpret it. However, it only seems to generate the schema from the first record in the source data, and my first record happened to contain an empty array.
The schema is then generated with an array type of “any”, which doesn’t provide you with any field names to map in the mapping UI.
To get around this, I’ve tended to just manually adjust the source schema within the mapping, adding children/nodes/arrays as necessary by clicking the “+” against another field name on the left hand side of the mapping.
So not a huge problem by any stretch, but just a bit confusing when you first encounter it.