Related Python packages

Packages that are useful next to ddi-l when you read, validate or transform DDI Lifecycle (DDI-L) content. Apart from the dependencies ddi-l already installs, none of them is required.

XML parsing and validation

• xmlschema: validates documents against the bundled DDI schemas. It is a dependency of ddi-l, so it is always installed.
• lxml: a faster XML parser and serializer. Install it with pip install 'ddi-l[full]'; ddi-l uses it automatically when it is present and falls back to the standard library otherwise.

Data processing

• pandas: turn variables, questions or code lists into tables for reports and checks.
• polars: a faster alternative to pandas for large documents or many files.
• pyarrow: write Parquet or Feather files, or move data frames between pandas and polars.

Files, HTTP and command-line tools

• requests or httpx: download DDI instances, schemas or controlled vocabularies before you validate them.
• typer or click: build your own command-line tools on top of ddi-l.
• fsspec: read and write files on local disk, S3-compatible storage or other remote file systems with the same code.

Testing

• pytest: the test runner this repository uses.
• hypothesis: property-based tests for validation rules or conversion code.
• ruff: linting and formatting.

Dependency management

• uv: this repository uses uv and its uv.lock file. Run uv sync --group dev to get the same development tools as CI.

Choosing packages

• Pin optional dependencies in a lock file, and use the same lock file in deployment, so a pipeline behaves the same everywhere.
• Validate at the boundaries of each pipeline step (doc.validate(), or ddi validate on the command line) so a malformed file is caught where it enters, not several steps later.