StructField Metadata Dictionary - What are the possible keys?

[chaosBEE]

I have a Delta Live Table which is being deposited to Unity Catalog. In the Python notebook, I am defining the schema with a series of StructFields, for example:

StructField(

    "columnName",

    StringType(),

    True,

    metadata = {

        'comment': "This is a comment"

    }

)

This works, however now I am trying to add tags, I have tried the following, but it doesn't seem to work:

metadata = {

        'comment': "This is a comment"

        'tags': {

            'tag1': 'x',

            'tag2': 'y',

        }

    }

Does anyone know to add tags to columns programmatically using the Python API?

Also, does anyone know how to add tags to the specific table, I am using 

dlt.create_streaming_table(), but there doesn't seem to be an option here: Delta Live Tables Python language reference | Databricks on AWS

Any help would be greatly appreciated!

Thanks,

Riqo

[chaosBEE]

Thank you Kaniz!

A few follow-up questions:

Could you give an example of each?

For 1. you said "define your schema with metadata", how do you this, what is the syntax for defining metadata with tags, like I said before, the following does not work:

StructField(

    "columnName",

    StringType(),

    True,

    metadata = {

        'comment': "This is a comment",

        'tags': {

            'tag1': 'x',

            'tag2': 'y',

        }

    }

)

For 2., again, what is the syntax? Could you provide an example? In the link you have shown, there doesn't seem to be a tags option as shown below.

@Dlt.table( name="<name>", comment="<comment>", spark_conf={"<key>" : "<value>", "<key>" : "<value>"}, table_properties={"<key>" : "<value>", "<key>" : "<value>"}, path="<storage-location-path>", partition_cols=["<partition-column>", "<partition-column>"], schema="schema-definition", temporary=False)

Also, is this available with @Dlt.append_flow too?

Thanks

[ChrisLawford_n1]

Did you ever get an answer to this ?

[Panda]

@chaosBEE Have you try the below

Use following SQL command to add tags programmatically

spark.sql("""ALTER TABLE your_catalog.your_schema.your_table SET TAGS ('key1' = 'value1');""")

[ChrisLawford_n1]

Hey @Panda ,

That will work but when you want to do this for each of the columns in your table it becomes very unclean in comparison to using something like the StructField Metadata attribute.

At the moment you would end up doing somethings like:

    def _create_tags_script(self, column_name, field_name, value):
        return f"ALTER COLUMN {column_name} SET TAGS ('{field_name}' = '{value}')"

    def get_sql_script(self, catalog_name, schema_name, table_name):
        alter_table_script = f"ALTER TABLE {catalog_name}.{schema_name}.{table_name}"
        output_sql_script = []
        for schema_field in self.schema_fields:
            if schema_field.source_field:
                output_sql_script.append(
                    f"{alter_table_script} {self._create_tags_script(schema_field.column_name, 'source_field', schema_field.source_field)}"
                )
            if schema_field.legacy_field:
                output_sql_script.append(
                    f"{alter_table_script} {self._create_tags_script(schema_field.column_name, 'legacy_field', schema_field.legacy_field)}"
                )
            if schema_field.comment:
                output_sql_script.append(
                    f"{alter_table_script} ALTER COLUMN {schema_field.column_name} COMMENT '{schema_field.comment}'"
                )
        return output_sql_script

which is just a bit more messy than 

StructType(
[
StructField(
    "columnName1",
    StringType(),
    True,
    metadata = {
        'comment': "This is a comment"
    }
),
StructField(
    "columnName2",
    StringType(),
    True,
    metadata = {
        'comment': "This is a comment"
    }
),
StructField(
    "columnName3",
    StringType(),
    True,
    metadata = {
        'comment': "This is a comment"
    }
)
]
)