Skip to content

Convert CSV/TSV to AUCX

Goal: turn a delimited experiment into a checksummed archive.

CLI

uv run openauc convert examples/data/demo_experiment demo.aucx

wrote demo.aucx (18244 bytes)

The output must end in .aucx. An existing file is refused with exit code 3 unless --overwrite is given:

uv run openauc convert examples/data/demo_experiment demo.aucx --overwrite
uv run openauc convert examples/data/demo_experiment demo.aucx --json

Python

import openauc

experiment = openauc.load("examples/data/demo_experiment")
experiment.export("demo.aucx")

Archive in, archive out

convert also accepts an archive, which is how you verify and rewrite one:

uv run openauc convert old.aucx new.aucx

The input is fully verified on read, so a successful rewrite proves the original was intact.

What is preserved

Everything the canonical model holds: both radius modes, masks, dtypes, array shapes, scan order, declared units, provenance, and all four value statuses.

restored = openauc.load("demo.aucx")
assert restored.to_dict() == experiment.to_dict()

Nothing is interpolated, resampled, sorted or unit-converted.

Going the other way

There is no AUCX → CSV command. It would need a documented lossiness policy, since a delimited file cannot carry per-scan axes or the unknown/missing distinction.

Batch conversion

#!/usr/bin/env bash
set -euo pipefail
mkdir -p archives
for dir in data/*/; do
    name=$(basename "$dir")
    if uv run openauc validate "$dir" >/dev/null 2>&1; then
        uv run openauc convert "$dir" "archives/${name}.aucx" --overwrite
        echo "ok: $name"
    else
        echo "SKIPPED (validation failed): $name" >&2
    fi
done

Common failure modes

Symptom Exit Fix
output must be an .aucx archive 2 Give the output a .aucx suffix
already exists; pass --overwrite 3 Add --overwrite
Input error before writing 2 openauc inspect the input for the message

Next step