Skip to content

CLI reference

Run each command from the project folder as uv run --project graph biotope …. Commands that write a report take --json and then write one JSON document to stdout. biotope <command> --help lists a command's arguments and options.

Project and data

Command What it does
init Creates a Biotope project in the folder NAME, or in the current folder with NAME ..
add Describes data files or folders as datasets and creates the annotation file of each.
annotate set Writes KEY=VALUE pairs into the annotation file of a dataset.
annotate validate Validates a Croissant metadata file.
config set-project-metadata Sets the project metadata, such as the creator and license, that add applies to every dataset.
config show-project-metadata Shows that project metadata.
mv Moves a tracked file or folder and updates every description and source package that names it.
rm Removes a dataset or a file from Biotope, with its data unless --keep-data is given.
status Lists the dataset descriptions and marks those whose revision differs from the one their source package pins.
check-data Compares the data files with the checksums their descriptions record and reports each changed file.
mark Sets the status of a dataset: raw, processed or mapped.
queue Shows every dataset grouped by its status.
map inspect Shows the sources and fields a description declares, without reading the data.

Extraction

Command What it does
extract try Runs one extraction on the given files and shows the records with their evidence. Nothing in the project changes.
extract run Runs the listed extractions, stores their records in data/extracted/<name>.jsonl and registers each records file as a dataset. An unchanged document is not extracted again.

Sources and graph

Command What it does
source generate Creates the missing source packages for one description, and reports drift, orphans and conflicts.
source register Registers a Croissant description written by hand, for data that add does not describe.
graph scaffold Creates the graph/ workspace.
graph check Checks the definitions and Python types without running loaders or mappings.
graph quality Runs the loaders and mappings and prints measurements of the graph, without exporting it.
graph build Runs the loaders, mappings and quality checks, then exports the graph to graph/build/.
graph metagraph Writes graph/metagraph.html, a viewer of the node and edge types; --report adds the counts of a build.