Skip to content

Latest commit

 

History

History
58 lines (40 loc) · 1.76 KB

README_run_units_on_corpus.md

File metadata and controls

58 lines (40 loc) · 1.76 KB
Raw

Build the tool

For building the entire tool chain for Units, just run below command in directory units-inference:

# cd units-inference
./dependency-setup.sh

Running Units on sci-corpus

Dependencies & prerequisites

To run driver script run-units-infer-on-corpus.py, you need first to install below packages:

  • pyYaml: pip install pyyaml

The driver script need python version higher than 2.7.10.

Regarding DLJC

We currently use a customized version of do-like- javac. In dependency- setup.sh we download this version of DLJC in the parent dir of units- inference.

You also need to install Lithium for the test case minimizer: https://github.com/pascaliUWat/do-like-javac/blob/master/README_testminimizer.md

Running the tool

To run units-inference on sci-corpus, just run following command:

# cd units-inference
python run-units-infer-on-corpus.py

This script would first fetching the projects defined in projects.yml from opprop-benchmarks into units- inference/corpus, then running Units on this corpus.

After this script finished, you can check the logs/infer.log under each project directory to see the tool running log, and logs/infer_result_*.jaif are the inference output of onology on each project.

You can also run analyze-corpus.sh to extract high level summaries for each corpus.

Regarding the sci-corpus

You may notice currently there were some projects has been commented out in projects.yml, i.e. we are running on a sub-set of sci-corpus. Most of those projects cannot compile (have basic java compile issues), with two of them we still have problems to infer results on them, and currently are working on fixing them.