Hi! Is there any way of defining a table schema using StructType instead of a DDL string and defining a field that is a generated column? Right now the schema looks like this:

StructType(
    [
        StructField(
            "id",
            StringType(),
            metadata={
                "comment": "Comment for this field"
            },
        )
    ],
    # other fields
)

I was hoping there is a way of adding a column to the

metadata

dict to define a certain

StructField

as a Delta Lake generated column. Thanks for your help!

DeltaTableBuilder.addColumns

allows passing in a

StructType

.

I am feeding this to a Delta Live Table pipeline, so the schema has to be either StructType or a string

DeltaColumnBuilder

produces a

StructField

if you call

build

.

I didn't know, I'll look into it! Are you certain it carries over generated column information?

Yes, setting metadata like the comment and the expression for the generated column is the entire goal of this API, so it would be weird if it wouldn’t…

I'm checking and it looks like the

DeltaColumnBuilder

API isn't exposed to Python, so I would probably have to pass through py4j to do this... I was asking if it carries over because if calling

build

simply produces a

StructField

with some additional metadata I could try creating it directly without passing through delta (which isn't a dependency of the project right now, as everything in that department is handled by DLT)