Tooling to Build a Database of SMT-LIB Problems

This repository contains Python scripts that can be used to build a database from a folder that contains a copy of the SMT-LIB benchmark library. Such a database can than be used to quickly select benchmarks for experiments, or to study the benchmark library itself.

The database will also store results large-scale evaluations, such as SMT-COMP. This will allow us to track benchmark difficulty over time.

For more information on the webapp and the database scheme see the README_RELEASE.md file.

Scripts

• prepopulate.py sets up the database file and inserts static data.
• addbenchmark.py adds a benchmark to the database file.
• postprocess.py adds evaluations, and performs any other operation that requires all benchmarks to be in the database.

TODO

• Optimize for reading. See for example here: https://jacobfilipp.com/sqliteoptimize/