u8sand/fairshake

A web interface for the scoring of biomedical digital objects by user evaluation according to the FAIR data principles: Findability, Accessibility, Interoperability, and Reusability.

FAIRshake

A web interface for the scoring of biomedical digital objects by user evaluation according to the FAIR data principles: Findability, Accessibility, Interoperability, and Reusability.

Project Layout

• app: Application primary namespace, modules are recursively imported for inversion of control registration
  • entities: Implementations of abstract interfaces. Should only ever depend on abstractions in interfaces, not directly on other entities.
  • interfaces: Abstract interfaces to be implemented. API interfaces define the swagger spec in the same place for close coupling.
  • util: Application-independent utility functions for code-reuse
• tests: Nose tests for verifying entities properly implement their interface and general app functionality.
  • test_util: Nose tests for verifying that utilities work as anticipated.

Development

Setting up environment

Python version of >= 3.5 is recommended, type annotation support is required. Install dependencies with pip install -r requirements.txt.

Run FAIRshakeWeb server

python run.py

Generate pure swagger spec from interfaces

python run.py dump

Run Application Tests

nosetests

Run Application Static Type Checks

mypy --ignore-missing-imports app

fairshake's Issues

OAuth2 integration

We should implement OAuth for user authentication.

https://pythonhosted.org/Flask-OAuth/

Finish porting from original FAIRshake

Original https://github.com/Nitrogen-DCPPC/FAIRshake

API Accession via SwaggerClient

It's important that APIs can be access programmatically, even remote ones. This should be accomplished via Swagger client injections.

Better defaults allowing anyone to run the project

Currently, OAuth are required to run the base project but it should work fine without that via a mock OAuth provider.

Runtime injection and command-line dependency injection configuration

It'd be good to allow command-driven dependency injection configuration allowing us to start APIs individually and inject dependencies as remote clients. This would allows us to split up functionality across multiple images while retaining the same code base. In the future, a Registry (e.g. SmartAPI) could be configured to obtain running APIs.

# Start FAIRshakeScore
./run.py FAIRshakeWeb --listen=0.0.0.0:8080 --FAIRshakeScore=localhost:8081 ...
./run.py FAIRshakeScore --host=0.0.0.0 --port=8081 --FAIRshakeRubric=localhost:8082 ...
./run.py FAIRshakeRubric --host=0.0.0.0 --port=8082
...

Self injection

Ensure modules can inject themselves via recursive importing.

Full API Implementation Test

1. Create and Register Test Rubric via RubricAPI
2. Create and Register Test Digital Object via RepositoryAPI
3. Perform Assessment(s) on Test Digital Object with Test Rubric via AssessmentAPI.
4. Aggregate assessment(s) with ScoreAPI.

Get Swagger-based mocks working