Skip to content

How to override build's input files

By default, build reads its 9 input files from ./input/, under fixed names. This guide covers pointing it at files somewhere else instead — useful for local development (e.g. files kept in sync by OneDrive for Linux under services/onedrive/data), or for testing against a specific dataset without touching ./input/.

Option 1: point at a different directory, same filenames

If your files are all in one directory, still under build's expected names (see Getting started for the exact list), use --input-dir:

./venv/bin/imarina-load-researchers build \
  --input-dir services/onedrive/data/_Projects/imarina-load-researchers/runtime/input

Option 2: override individual files

Each input has its own option, which takes precedence over --input-dir (or the ./input/ default) for that one file:

./venv/bin/imarina-load-researchers build \
  --a3-input path/to/some_a3_dump.xlsx \
  --imarina-input path/to/some_previous_upload.xlsx \
  --countries-dict path/to/countries.xlsx \
  --jobs-dict path/to/jobs.xlsx \
  --personal-web-path path/to/personal_web.xlsx \
  --unit-group-path path/to/unit_group.xlsx \
  --entity-type-path path/to/unit_type.xlsx \
  --job-description-entity-path path/to/job_description_entity.xlsx \
  --sex-path path/to/sex.xlsx

You don't have to override every file — mix and match individual overrides with --input-dir for the rest:

./venv/bin/imarina-load-researchers build \
  --input-dir services/onedrive/data/_Projects/imarina-load-researchers/runtime/input \
  --a3-input /tmp/a_specific_a3_dump.xlsx

Here, everything except the A3 dump comes from the OneDrive-synced directory; the A3 dump comes from wherever you point --a3-input.

Where the output goes

--output-path controls where build writes the result (default: ./output/<datetime>__icl_ag_personal_12539.xlsx):

./venv/bin/imarina-load-researchers build --output-path /tmp/my-test-output.xlsx

Note that upload/publish autodetect the latest file in ./output/ by default — if you redirect the output path elsewhere, you'll need to pass --file-path explicitly to those commands too.

Skipping download entirely during development

If you keep services/onedrive/data in sync with SharePoint (see the project README's OneDrive-for-Linux setup), you can develop against production-shaped data without ever running download — just point --input-dir (or the individual file options) at that synced folder, run build, and inspect the result locally.