Export results #

Litmus has two CLI surfaces that take a run and produce a file:

• litmus show <run_id> -f <fmt> — generates a report (HTML for browsers, PDF for distribution, JSON / CSV for downstream tools). Driven by report templates.
• litmus export <id> -f <fmt> — replays the event stream for a run or session into a target format. The supported formats are the test-and-measurement interchange ones: CSV, JSON, STDF, HDF5, TDMS, MDF4, ATML.

Pick the command by what the receiver wants. A QA engineer wants PDF reports → litmus show -f pdf. A semiconductor vendor wants STDF → litmus export -f stdf.

Prerequisites #

• At least one completed run on disk
• A run ID or session ID (prefix match works for both — get one from litmus runs or the operator UI's Results list)

Reports — litmus show -f #

litmus show <run_id>              # text summary to the terminal
litmus show <run_id> -f html      # one HTML file
litmus show <run_id> -f pdf -o reports/   # PDF into reports/
litmus show <run_id> -f json      # one JSON file with the run's structured data
litmus show <run_id> -f csv       # tabular CSV (one row per measurement)

The HTML and PDF formats render via Jinja2 templates (default template is default; switch with -t <name>). JSON and CSV are fixed-format writers — the -t option doesn't apply to them. Output path defaults to the current directory; override with -o.

Reports run against the denormalized parquet for the run. They include the run summary, steps, measurements, and (when captured) environment snapshot.

Interchange exports — litmus export -f #

litmus export <id> -f csv                 # default output dir: exports/csv/
litmus export <id> -f json -o /tmp/out/   # explicit output dir
litmus export <id> -f stdf                # STDF v4 for semiconductor test floors
litmus export <id> -f atml                # ATML / IEEE 1671 instances
litmus export <id> -f hdf5                # HDF5 (scientific computing)
litmus export <id> -f tdms                # NI TDMS (LabVIEW ecosystem)
litmus export <id> -f mdf4                # ASAM MDF4 (automotive measurement data)

<id> accepts a run id OR a session id — the CLI auto-detects by prefix-matching the events file. Output directory defaults to exports/<fmt>/ when -o isn't given.

Mechanics differ from reports: litmus export reads the run's events from the Arrow IPC store and replays them through a subscriber registered for the requested format. This is what makes the exporter set extensible — new formats register a subscriber class with the format_name field and they show up in the -f choices.

Discoverability — what's installed #

litmus export <run_id> -f bogus
# No subscriber registered for format 'bogus'.
# Available: atml, csv, hdf5, json, mdf4, stdf, tdms

Asking for an unknown format prints the current list of installed subscribers. The format set is fixed by the Litmus package — new formats need to land as built-in subscribers in src/litmus/data/exporters/; there is no plugin / entry-point extension surface today.

Common tasks #

• Daily PDF reports for the production line — wrap litmus show $RUN -f pdf -o daily-reports/ in your build / shift-end script.
• Feed runs into a semi vendor's STDF analyzer — litmus export $RUN -f stdf -o stdf-out/, ship the file.
• Build a custom downstream tool that reads CSV — litmus export -f csv to a known path, your tool picks it up.
• Bulk-export an entire session — pass the session ID instead of a run ID; the CLI replays every run in the session.

See also #

• Lakehouse import — when the receiver wants parquet rows directly, skip export entirely
• Grafana — when the receiver wants live dashboards instead of files
• litmus show — CLI reference for the report path
• litmus export — CLI reference for the interchange path
• Three stores — where the parquet and event data live on disk