Output Formats #

TesterKit produces test data through a fixed pipeline — four on-disk stores plus two CLI commands for post-hoc rendering and conversion.

What's always on #

test execution
    └→ events/   (typed event log)
    └→ runs/     (sealed per-run Parquet — analysis-ready)
    └→ channels/ (time-series instrument samples)
    └→ files/    (captured artifacts — images, video, vendor files)

These four stores are the platform. They're populated automatically by every test run; there's no configuration knob to disable them. See Data stores and Event Log for the data model.

Reading runs back #

Two CLI commands cover the post-hoc rendering and format-conversion paths:

`testerkit show <run_id> -f <format> [-o <path>]` #

Renders a stored run for human consumption. Reads the run's parquet, formats output, writes to disk.

testerkit show abc123                 # terminal pretty-print
testerkit show abc123 -f html         # HTML report
testerkit show abc123 -f pdf -o out/  # PDF (requires `testerkit[pdf]`)
testerkit show abc123 -f csv          # CSV
testerkit show abc123 -f json         # JSON

`testerkit export <run_id> -f <format> [-o <dir>]` #

Converts a stored run to industry data formats.

testerkit export abc123 -f csv                            # CSV (built-in)
testerkit export abc123 -f json                           # JSON (built-in)
testerkit export abc123 -f stdf -o exports/stdf/          # STDF (requires [stdf])
testerkit export abc123 -f hdf5 -o exports/hdf5/          # HDF5 (requires [hdf5])
testerkit export abc123 -f tdms -o exports/tdms/          # TDMS (requires [tdms])
testerkit export abc123 -f mdf4 -o exports/mdf4/          # MDF4 (requires [mdf4])

Available formats #

Format	Command	Library	Install	Industry
html	`testerkit show`	jinja2	(built-in)	Universal
pdf	`testerkit show`	weasyprint	`pip install testerkit[pdf]`	Universal
csv	both	stdlib	(built-in)	Universal
json	both	stdlib	(built-in)	Universal
stdf	`testerkit export`	Semi-ATE-STDF	`pip install testerkit[stdf]`	Semiconductor
hdf5	`testerkit export`	h5py	`pip install testerkit[hdf5]`	Scientific
tdms	`testerkit export`	npTDMS	`pip install testerkit[tdms]`	NI/LabVIEW
mdf4	`testerkit export`	asammdf	`pip install testerkit[mdf4]`	Automotive

Bundle: pip install testerkit[all-exporters] installs stdf, hdf5, tdms, mdf4 together.

Cloud destinations (S3, Snowflake, lakehouse) #

For shipping data to cloud destinations or lakehouse formats (Snowflake, Delta, Iceberg), TesterKit does not ship a built-in transport in the current release. The parquet files in runs/ are the contract. Consumers run their own pipeline:

DuckDB / Polars / Pandas: read directly from data/runs/{date}/*.parquet. Rows are typed by record_type (run / step / vector); measurements are nested under the vector rows, so UNNEST(measurements) to flatten them.
Snowflake / Databricks / Trino-Iceberg: copy parquets to your storage layer and ingest with a record_type-keyed split (run / step / vector), unnesting the vector rows' measurements list into your measurement fact table.

Canonical recipes — see Lakehouse Import.

Adding a format #

The set of export formats is fixed by the package; there is no third-party plugin hook to register new ones. To produce a format TesterKit doesn't ship, read the parquet in runs/ and convert it yourself (see the cloud-destinations recipe above).