Validating & Exporting

Parse and export

KCL is both a validator and a parser, so the same source drives type-checking and serialization.

Parse and print the composed data product to standard output. This is where type errors and failed constraints surface — if a required field is missing or a value violates a constraint, the build fails here rather than in production:

kcl run product.k

Export to YAML (or JSON) by choosing the output format:

kcl run product.k --format yaml > product.yaml   # YAML
kcl run product.k --format json > product.json   # JSON

The exported product.yaml is the finished ODPS document, generated from your typed KCL — and this is where apiVersion: v1.0.0 and kind: DataProduct reappear, supplied by the schema defaults you never had to write, along with every $type rendered back as a plain type: key.

The just export shortcut

In the repository, the export step is wrapped in the Justfile. Running:

just export

parses the project's root file and writes the result to enkinex-odps-tutorial/product.yaml — the same final document, produced in one step.

Wrapping up

Starting from an untyped YAML document, we rebuilt the ODPS customer data product example as a typed, modular KCL project: product metadata, ownership, management and support, and the input and output ports, each in its own small file, composed by a root that reads like a table of contents. The result is validated at build time, and its parts are reusable across products — governance kept as code in the repository, rather than copied between YAML files.

The complete project is on GitHub under enkinex-odps-tutorial