What setup downloads
scripts/setup.sh calls scripts/get_bundle.sh.
get_bundle.sh populates:
| Source | What happens |
|---|---|
| Existing local bundle | If CtakesBun-bundle/.../apache-ctakes-6.0.0 already exists, the script stops. |
| Local archive | scripts/get_bundle.sh --file /path/to/CtakesBun-bundle_*.tgz extracts that file. |
| Codeberg release | The script checks Codeberg release assets first. |
| GitHub release | The script falls back to GitHub release assets. |
| Direct URL | CTAKES_BUNDLE_URL overrides release lookup. |
scripts/get_java.sh is separate. It downloads a portable Temurin Java into:
CTAKES_JAVA_HOME is set, and run_pipeline.sh can also detect .ctakes_deps/java/current/bin/java when the folder exists.
Files people touch
Most users only touch these:| Path | Use |
|---|---|
input/<batch>/ | Put .txt note files here. |
outputs/<batch>/ | cTAKES writes results here. |
scripts/setup.sh | Downloads or verifies the cTAKES bundle. |
scripts/flight_check.sh | Checks Java, cTAKES, pipeline files, tools, sample notes, and a dry run. |
scripts/run_pipeline.sh | Runs one folder of notes. |
scripts/run_input.sh | Runs every folder under an input root. |
| Path | Use |
|---|---|
pipelines/ | .piper recipes. |
resources/dictionary_configs/ | Dictionary XML/config recipes. |
tools/ | Java helper annotators and writers compiled into build/tools. |
dictionaries/ | Small dictionary release metadata only. |
samples/mimic/ | 100-note local validation pack. |
samples/mimic_manifest_*.json.gz | Expected sample output manifests. |
Input notes
The runner reads.txt and .xmi files from the input folder.
For STAR note extraction, use the Phase I filename shape:
documents.document_id, concepts.document_id, rxnorm.document_id, cui_counts.document_id, and the CSV/XMI filenames when those writers are enabled.