Testing

Overview

Code reliability in EMGFlow is enforced via an automated unittest suite run on every commit via GitHub Actions. Developers can also run these same tests locally.

Running Unit Tests

The EMGFlow package must first be installed. Test case files can then be downloaded and run. These test files are available in the tests subfolder of the GitHub repository. Alternatively, users can clone the repo. Test cases can then be run individually, or sequentially as shown below.

# 1. Install EMGFlow package
pip install EMGFlow

# 2. Clone repo to local folder
git clone git@github.com:WiIIson/EMGFlow-Python-Package.git .

# 3. Navigate to test folder
cd tests

# 5. Run all test cases
python -m unittest discover -v -s . -p 'test_*.py'

Test Coverage

EMGFlow’s tests provide broad functional coverage across modules. Access utilities are exercised via package metadata, path creation, sample-data generation, file-type reading, and directory mapping. Preprocessing is covered at both function and batch levels: PSD computation; notch filtering; bandpass filtering; full-wave rectification; artefact screening with both Hampel and Wiener methods; gap filling; four smoothing modes (boxcar, RMS, Gaussian, LOESS); a composite smooth_signals stage; and an end-to-end clean_signals pipeline. Feature tests span >20 calculators across time and frequency domains—e.g., iEMG, MAV, MMAV1/2, RMS, WL, WAMP, variance/SSI, MFL/MNF/MDF, spectral entropy and related slope/decay indices, and twitch metrics—and include a full extract_features run that produces Features.csv. The plotting layer includes a dashboard smoke test to ensure the Shiny app constructs. Assertions focus on types, successful execution, and expected file creation, giving wide “does it run and produce outputs” assurance; targeted numeric regressions, adversarial inputs (missing columns, NaNs at boundaries), and parameter-bound checks can be added to deepen correctness guarantees.