kszucs / tpc-queries

TPC queries

TPC query translations

All queries are ready-to-run, using the default substitution parameters as specified by the TPC-H spec.

• Each translation of each query is in its own file in the queries/ directory.
  • sqlite_tpc/h*.sql files are taken from the spec and modified as needed to be executed directly by SQLite
  • ibis_tpc/h*.py files are translated into Python using Ibis
  • dplyr_tpc/h*.R files are translated into R using dplyr, from ursacomputing/arrowbench

Usage

./runtpc [--db <tpch.sqlite>] [--output <output_dir>] [-v] <queries>

where:

• <output_dir> specifies a directory for debugging files: results, generated queries, logs (default is no debugging output)
• <tpch.sqlite> is the TPC database to use (default tpch.db, not included)
• <queries> is any number of TPC query ids: h01 .. h22

Notes

• Some queries require a TPC database with scale factor of .01 at minimum to generate results.
• Use this sqlite db with a scaling factor of .01 (10MB).

Links

• TPC-H v3.0 specification

Included scripts

write_tpc_proto.py

Uses ibis_substrait to compile Ibis expressions to binary blobs and stores them in the substrait_tpc directory.

Usage:

Generate updated blobs for all queries (for those queries that can be compiled)

python write_tpc_proto.py

Generate updated blob for query h#

python write_tpc_proto.py --query h01

`

`