Add SPARQL Conformance Testing Framework

[ricoandris]

Summary

This branch is built on top of the multi-engine work by Tanmay from
qlever-control#190,
which introduced the basic commands for operating QLever and six additional
SPARQL engines. This PR adapts and extends those commands so they can be used
by the separate sparql-conformance test harness.

This PR is the qlever-control side of
ad-freiburg/sparql-conformance#61.
To test the integration, check out and install both PRs as described below.

What changed

• Build on the qblazegraph, qgraphdb, qjena, qmdb, qoxigraph, and
  qvirtuoso command-line tools introduced in
  qlever-control#190,
  adjusting their configuration, indexing, startup, query, reset, logging,
  status, and shutdown behavior where needed for conformance testing.
• Add a non-interactive conformance mode to engine operations. It suppresses
  terminal followers and prompts, prepares missing engine configuration where
  needed, and supports fast test-data resets.
• Discover command packages through Python imports. This works for commands in
  qlever-control as well as commands supplied by separately installed
  packages.
• Keep qlever-control independent of the test harness: it neither vendors nor
  depends on sparql-conformance.
• Retain the existing QLever CLI improvements that were developed alongside
  the multi-engine work, including the evaluation viewer, memory monitoring,
  and expanded index statistics.

Package boundary

sparql-conformance
owns the test runner, W3C manifest handling, engine adapters, result
comparison, Qleverfile templates, and the sparql_conformance executable.
qlever-control owns the reusable commands that configure and operate the
engines. The harness imports qlever-control only when a built-in engine is
selected.

This keeps a normal qlever installation free of the conformance harness and
avoids maintaining two copies of the test code.

Quick test setup

Prerequisites

• Python 3.10 or newer
• Git
• A running Docker daemon (or the container runtime configured in the
  generated Qleverfile)
• A GraphDB license when testing GraphDB

1. Check out both PRs

Run these commands from a new parent directory:

git clone --branch sparql-conformance-command-all-engines \
  https://github.com/SIRDNARch/qlever-control.git

git clone --branch main \
  https://github.com/SIRDNARch/sparql-conformance.git

The result should be two sibling checkouts:

.
├── qlever-control/
└── sparql-conformance/

2. Install both packages in one environment

Install qlever-control first and sparql-conformance second:

python3 -m venv .venv
source .venv/bin/activate

python -m pip install -e ./qlever-control
python -m pip install -e ./sparql-conformance

The installation order matters when upgrading from an older version that also
installed a sparql_conformance executable. Installing
sparql-conformance last ensures that the executable has a single owner.

3. Set up and run an engine

Use a separate working directory for each engine because setup writes an
engine-specific Qleverfile in the current directory:

mkdir -p conformance-runs/qlever
cd conformance-runs/qlever

sparql_conformance setup qlever
sparql_conformance test --report summary

setup creates the Qleverfile and downloads the W3C SPARQL 1.0 and 1.1 test
suites. test then configures and indexes the test data, starts the selected
engine, runs the tests, resets it between tests when possible, and stops it
again. No separate engine-start command is required.

Results are written below ./results.

4. Test another engine

Create another working directory and change only the engine name:

cd ../..
mkdir -p conformance-runs/jena
cd conformance-runs/jena

sparql_conformance setup jena
sparql_conformance test --report summary

Supported setup targets are:

qlever, blazegraph, graphdb, jena, mdb, oxigraph, virtuoso

For example, replace jena in the commands above with blazegraph,
graphdb, mdb, oxigraph, or virtuoso. Keep each engine in its own
directory.

Useful commands while testing:

# Run a small test group and print every result.
sparql_conformance test --include aggregates --report line

# Load one test and leave the engine available for manual debugging.
sparql_conformance analyze "COUNT 1"

# Compare with an earlier result.
sparql_conformance test \
  --compare-to results/previous.json.bz2 \
  --report summary

# View result files in the web UI.
sparql_conformance visualize --result-directory ./results

See docs/sparql-conformance.md for more
details.