fairsoft-core

Core FAIRsoft evaluation engine for research software metadata.

fairsoft-core provides the reusable logic used to compute FAIRsoft indicators, FAIR scores, and evaluation feedback from normalized research software metadata.

Installation

Install the latest release from PyPI:

pip install fairsoft-core

Install a specific version:

pip install fairsoft-core==0.1.1

For local development:

git clone https://github.com/inab/fairsoft-core.git
cd fairsoft-core
pip install -e ".[dev]"

Development versions can also be installed directly from GitHub:

pip install "fairsoft-core @ git+https://github.com/inab/fairsoft-core.git@main"

What this package does

• Computes FAIRsoft indicators.
• Computes FAIR scores.
• Generates evaluation feedback.
• Provides a command-line interface for FAIRsoft evaluation workflows.
• Provides reusable Python functions for integration into other services.

Input expectations

The evaluator expects metadata that has already been normalized to the internal FAIRsoft input model.

This means that fairsoft-core does not prepare, clean, enrich, or disambiguate raw metadata. The input JSON must already match the structure expected by the core Instance model.

Command-line usage

Full FAIRsoft evaluation

Run a full FAIRsoft evaluation from a JSON file:

fairsoft evaluate path/to/metadata.json

Write the output to a file:

fairsoft evaluate path/to/metadata.json --output result.json

Return only part of the evaluation output:

fairsoft evaluate path/to/metadata.json --select result
fairsoft evaluate path/to/metadata.json --select logs
fairsoft evaluate path/to/metadata.json --select feedback

Choose compact JSON output:

fairsoft evaluate path/to/metadata.json --select result --format json-compact

Single-indicator evaluation

List the indicators currently available through the CLI:

fairsoft indicator --list

Describe one indicator:

fairsoft indicator --describe F3_1
fairsoft indicator --describe searchability_in_registries

Run indicator F3_1 by ID:

fairsoft indicator F3_1 path/to/metadata.json

Run the same indicator by name:

fairsoft indicator searchability_in_registries path/to/metadata.json

Return only the result or logs:

fairsoft indicator F3_1 path/to/metadata.json --select result
fairsoft indicator F3_1 path/to/metadata.json --select logs

Write compact output to a file:

fairsoft indicator F3_1 path/to/metadata.json --select logs --format json-compact --output f3_1_logs.json

CLI output options

--select

--select controls which part of the output is returned.

Available values:

• full
• result
• logs
• feedback

For full evaluation, the returned structure contains:

• result
• logs
• feedback

For single-indicator commands, the CLI keeps the same output contract, although some indicators may return an empty feedback field.

--format

--format controls serialization.

Available values:

• json
• json-compact

Examples:

fairsoft evaluate path/to/metadata.json --select full --format json
fairsoft evaluate path/to/metadata.json --select result --format json-compact
fairsoft indicator F3_1 path/to/metadata.json --select logs --format json

Python usage

Run a full FAIRsoft evaluation programmatically:

from fairsoft_core.evaluation import run_fairsoft_evaluation

result = run_fairsoft_evaluation(tool_metadata)

tool_metadata must be a dictionary or object compatible with the FAIRsoft input model used by the evaluator.

Indicator metadata requirements

Note. The “Equivalent CodeMeta fields” column reflects the current minimal adapter implemented in fairsoft-core, not the full set of CodeMeta fields that could theoretically be mapped in future versions. Entries marked as “partially” indicate indirect or limited support.

Indicator	Name	Fields used	Minimum metadata	Equivalent CodeMeta fields	Not applicable when
A1_1	existence_of_api_or_web	webpage, type	webpage	url, homepage	super_type == 'no_web'
A1_2	downloadable_and_buildable_working_version	download, src, type	download or src	downloadUrl	super_type == 'web'
A1_3	installation_instructions	inst_instr, documentation, source, type	inst_instr or documentation or source	readme, softwareHelp, documentation	super_type == 'web'
A1_4	test_data	test, documentation, type	test or documentation	no direct minimal mapping	super_type == 'web'
A1_5	source_code	src, type	src	no direct minimal mapping (codeRepository is mapped to repository, not src)	super_type == 'web'
A3_1	registration_not_compulsory	registration_not_mandatory	registration_not_mandatory	no direct minimal mapping	—
A3_2	version_for_free_os	os, type	os	operatingSystem	super_type == 'web'
A3_3	availability_for_several_os	os, type	os	operatingSystem	super_type == 'web'
A3_4	availability_on_free_e_infrastructures	e_infrastructures, webpage, source, type	e_infrastructures or webpage or source	partially url, homepage, sameAs, relatedLink	super_type == 'web'
A3_5	availability_on_several_e_infrastructures	e_infrastructures, webpage, source, type	e_infrastructures or webpage or source	partially url, homepage, sameAs, relatedLink	super_type == 'web'
F1_2	semantic_versioning	version	version	softwareVersion, version	—
F2_1	structured_metadata	source	source	no direct minimal mapping	—
F2_2	metadata_with_controlled_vocabularies	topics, operations	topics or operations	partially applicationCategory	—
F3_1	searchability_in_registries	source, registries	source or registries	no direct minimal mapping	—
F3_2	searchability_in_software_repositories	repository	repository	codeRepository	—
F3_3	searchability_in_literature	publication	publication	citation, referencePublication	—
I1_1	standard_data_formats	input, output	input or output	no direct minimal mapping	—
I1_2	api_standard_specification	documentation, type	documentation	documentation, softwareHelp, readme	super_type == 'no_web'
I1_3	verifiable_data_formats	input, output	input or output	no direct minimal mapping	—
I1_4	flexibility_of_supported_data_formats	input, output	input and output	no direct minimal mapping	—
I2_1	api_or_library_version	type	type	no direct minimal mapping	—
I2_2	e_infrastructure_compatibility	e_infrastructures, webpage, source	e_infrastructures or webpage or source	partially url, homepage, sameAs, relatedLink	—
I3_1	dependencies_statement	dependencies	dependencies	no direct minimal mapping	—
I3_2	dependencies_provided	source, registries, links	source or registries or links	partially sameAs, relatedLink	—
I3_3	dependency_aware_system	source, registries, links	source or registries or links	partially sameAs, relatedLink	—
R1_1	usage_guides	documentation	documentation	readme, softwareHelp, documentation	—
R2_1	license	license, documentation	license or documentation	license, partially documentation	—
R2_2	technical_conditions_of_use	license, documentation	license or documentation	license, partially documentation	—
R3_1	contribution_policy	documentation	documentation	documentation, readme, softwareHelp	—
R3_2	credit	authors	authors	author, contributor, maintainer	—
R4_1	public_version_control	version_control	version_control	no direct minimal mapping (codeRepository suggests it, but the adapter does not infer version_control)	—
R4_2	release_policy	documentation	documentation	documentation, readme, softwareHelp	—

Development

Install the package with development dependencies:

pip install -e ".[dev]"

A common local workflow is:

ruff check . --fix
ruff format .
pytest

Keep pyproject.toml as the source of truth for the package version.

The repository includes a pre-push hook that runs the test suite before code is pushed.