ci: give the examples job the test data it needs #689

Workflow file for this run

.github/workflows/test-complete.yml at 248d47a

	name: Complete Test Suite

	on:
	push:
	branches: [ main ]
	pull_request:
	branches: [ main ]

	env:
	CARGO_TERM_COLOR: always

	jobs:
	# Rust tests on multiple platforms
	rust-tests:
	name: Rust Tests (${{ matrix.os }})
	strategy:
	matrix:
	os: [ubuntu-latest, windows-latest, macos-latest]
	runs-on: ${{ matrix.os }}

	steps:
	- uses: actions/checkout@v4

	- name: Set up Rust
	uses: actions-rs/toolchain@v1
	with:
	toolchain: stable
	override: true
	components: rustfmt, clippy

	- name: Cache cargo registry
	uses: actions/cache@v4
	with:
	path: \|
	~/.cargo/registry
	~/.cargo/git
	key: ${{ runner.os }}-cargo-${{ hashFiles('**/Cargo.lock') }}
	restore-keys: \|
	${{ runner.os }}-cargo-

	- name: Cache cargo build
	uses: actions/cache@v4
	with:
	path: target
	key: ${{ runner.os }}-cargo-build-${{ hashFiles('**/Cargo.lock') }}
	restore-keys: \|
	${{ runner.os }}-cargo-build-

	- name: Build
	run: cargo build --release --verbose

	- name: Run Rust tests
	run: cargo test --verbose

	- name: Run clippy
	run: cargo clippy -- -W warnings

	- name: Check formatting
	run: cargo fmt -- --check

	# Python tests only on Linux (where uv is readily available)
	python-tests:
	name: Python Tests (Linux)
	runs-on: ubuntu-latest
	needs: rust-tests # Run after Rust tests to ensure binary builds

	steps:
	- uses: actions/checkout@v4

	- name: Set up Python
	id: setup-python
	uses: actions/setup-python@v5
	with:
	python-version: '3.10'

	- name: Install uv
	run: \|
	curl -LsSf https://astral.sh/uv/install.sh \| sh
	echo "$HOME/.cargo/bin" >> $GITHUB_PATH

	- name: Set up Rust (for building sql-cli)
	uses: actions-rs/toolchain@v1
	with:
	toolchain: stable
	override: true

	- name: Cache cargo build
	uses: actions/cache@v4
	with:
	path: target
	key: ${{ runner.os }}-cargo-build-python-${{ hashFiles('**/Cargo.lock') }}
	restore-keys: \|
	${{ runner.os }}-cargo-build-python-
	${{ runner.os }}-cargo-build-

	- name: Build SQL CLI
	run: cargo build --release

	- name: Cache Python dependencies
	uses: actions/cache@v4
	with:
	path: .venv
	# The interpreter version MUST be in the key. A .venv is a set of
	# symlinks into the hosted toolcache, so a venv built against
	# 3.10.<old> is broken the moment the runner image ships 3.10.<new>
	# and deletes the old directory. `python-version: '3.10'` floats across
	# patch releases, so this happens without any change on our side.
	key: ${{ runner.os }}-python-${{ steps.setup-python.outputs.python-version }}-${{ hashFiles('pyproject.toml', 'uv.lock') }}
	restore-keys: \|
	${{ runner.os }}-python-${{ steps.setup-python.outputs.python-version }}-

	- name: Install Python dependencies
	run: \|
	# Recreate unless the restored .venv has a working interpreter. Testing
	# `-d .venv` is not enough: a cache hit against a since-removed patch
	# release leaves the directory present with a dangling
	# .venv/bin/python symlink, and uv then fails with "Broken symlink"
	# rather than repairing it. `-x` follows the symlink, so it is false
	# exactly when the target is gone.
	[ -x .venv/bin/python ] \|\| uv venv --clear
	uv pip install -e .
	uv pip install pytest pytest-cov pandas numpy

	- name: Generate test data
	run: \|
	if [ ! -f "data/test_simple_math.csv" ] \|\| [ ! -f "data/test_simple_strings.csv" ]; then
	uv run python scripts/generate_simple_test.py
	fi

	- name: Run Python tests
	run: \|
	uv run pytest tests/python_tests/ -v --tb=short

	- name: Upload test results
	if: failure()
	uses: actions/upload-artifact@v4
	with:
	name: python-test-results
	path: \|
	.pytest_cache/
	*/.log

	# Examples test suite (formal and smoke tests)
	examples-tests:
	name: Examples Test Suite (Linux)
	runs-on: ubuntu-latest
	needs: rust-tests # Run after Rust tests to ensure binary builds

	steps:
	- uses: actions/checkout@v4

	- name: Set up Python
	uses: actions/setup-python@v5
	with:
	python-version: '3.10'

	- name: Install uv
	run: \|
	curl -LsSf https://astral.sh/uv/install.sh \| sh
	echo "$HOME/.cargo/bin" >> $GITHUB_PATH

	- name: Set up Rust (for building sql-cli)
	uses: actions-rs/toolchain@v1
	with:
	toolchain: stable
	override: true

	- name: Cache cargo build
	uses: actions/cache@v4
	with:
	path: target
	key: ${{ runner.os }}-cargo-build-examples-${{ hashFiles('**/Cargo.lock') }}
	restore-keys: \|
	${{ runner.os }}-cargo-build-examples-
	${{ runner.os }}-cargo-build-

	- name: Build SQL CLI
	run: cargo build --release

	# data/test_simple_*.csv are generated, not committed (.gitignore's
	# `test_*.csv` rule covers them), so several examples that pass locally
	# fail here without this. Mirrors the same step in the Python Tests job.
	- name: Generate test data
	run: \|
	if [ ! -f "data/test_simple_math.csv" ] \|\| [ ! -f "data/test_simple_strings.csv" ]; then
	uv run python scripts/generate_simple_test.py
	fi

	- name: Run examples test suite
	run: \|
	uv run python tests/integration/test_examples.py

	- name: Upload test results
	if: failure()
	uses: actions/upload-artifact@v4
	with:
	name: examples-test-results
	path: \|
	examples/expectations/*.json
	*/.log

	# SQL parity regression gate (differential test vs DuckDB)
	parity-tests:
	name: SQL Parity (vs DuckDB)
	runs-on: ubuntu-latest
	needs: rust-tests # Run after Rust tests to ensure binary builds

	steps:
	- uses: actions/checkout@v4

	- name: Set up Python
	uses: actions/setup-python@v5
	with:
	python-version: '3.12' # harness uses stdlib tomllib (3.11+)

	- name: Install uv
	run: \|
	curl -LsSf https://astral.sh/uv/install.sh \| sh
	echo "$HOME/.cargo/bin" >> $GITHUB_PATH

	- name: Set up Rust (for building sql-cli)
	uses: actions-rs/toolchain@v1
	with:
	toolchain: stable
	override: true

	- name: Cache cargo build
	uses: actions/cache@v4
	with:
	path: target
	key: ${{ runner.os }}-cargo-build-parity-${{ hashFiles('**/Cargo.lock') }}
	restore-keys: \|
	${{ runner.os }}-cargo-build-parity-
	${{ runner.os }}-cargo-build-

	- name: Build SQL CLI
	run: cargo build --release

	- name: Install Python dependencies
	# Resolves the pinned DuckDB from pyproject.toml [dependency-groups].test
	# rather than installing whatever is latest. The corpus contains cases
	# where the SQL standard leaves behaviour implementation-defined and we
	# follow the reference engine by policy, so those cases encode DuckDB's
	# current choices. An unpinned bump could redden this gate for reasons
	# unrelated to our engine — and point at the wrong place when it did.
	# Bump the pin in pyproject.toml deliberately and review any drift.
	run: uv sync --group test

	- name: Run parity check (sql-cli vs DuckDB)
	# Fails if any case drifts from its expected bucket: a regressed AGREE, a
	# silently-closed gap, or a new un-annotated non-AGREE case. See
	# docs/SQL_PARITY.md for the contract.
	run: uv run --group test python tests/comparison/runner.py --check

	- name: Upload parity report
	if: always()
	uses: actions/upload-artifact@v4
	with:
	name: parity-report
	path: tests/comparison/reports/

	# Integration test combining both
	integration-test:
	name: Integration Test (Linux)
	runs-on: ubuntu-latest
	needs: [rust-tests, python-tests, examples-tests]

	steps:
	- uses: actions/checkout@v4

	- name: Set up Rust
	uses: actions-rs/toolchain@v1
	with:
	toolchain: stable
	override: true

	- name: Build SQL CLI
	run: cargo build --release

	- name: Test CLI directly
	run: \|
	# Generate test data
	echo "id,value" > test.csv
	echo "1,10" >> test.csv
	echo "2,20" >> test.csv
	echo "3,30" >> test.csv

	# Test basic query
	./target/release/sql-cli test.csv -q "SELECT * FROM test WHERE value > 15" -o csv

	# Test with string methods
	echo "id,name" > names.csv
	echo "1, Alice " >> names.csv
	echo "2,Bob" >> names.csv

	./target/release/sql-cli names.csv -q "SELECT id, name.Trim() as trimmed FROM names" -o csv

	- name: Verify --query-plan works
	run: \|
	./target/release/sql-cli test.csv -q "SELECT id FROM test WHERE value > 15" --query-plan

	# Performance benchmarking (Linux only, non-blocking)
	performance-benchmark:
	name: Performance Benchmarks (Linux)
	runs-on: ubuntu-latest
	needs: rust-tests # Run after rust tests to get the binary
	continue-on-error: true # Don't fail the build if benchmarks have issues

	steps:
	- uses: actions/checkout@v4

	- name: Set up Python
	uses: actions/setup-python@v5
	with:
	python-version: '3.10'

	- name: Install uv
	run: \|
	curl -LsSf https://astral.sh/uv/install.sh \| sh
	echo "$HOME/.cargo/bin" >> $GITHUB_PATH

	- name: Set up Rust (for building sql-cli)
	uses: actions-rs/toolchain@v1
	with:
	toolchain: stable
	override: true

	- name: Cache cargo build
	uses: actions/cache@v4
	with:
	path: target
	key: ${{ runner.os }}-cargo-build-benchmark-${{ hashFiles('**/Cargo.lock') }}
	restore-keys: \|
	${{ runner.os }}-cargo-build-benchmark-
	${{ runner.os }}-cargo-build-

	- name: Build SQL CLI (Release with optimizations)
	run: cargo build --release

	- name: Install Python dependencies
	run: \|
	# Recreate unless an existing .venv has a working interpreter — see the
	# Python Tests job for why `-d` is the wrong test. This job caches no
	# .venv, so it cannot hit the stale-symlink case today; kept identical
	# so the two don't drift into different failure modes.
	[ -x .venv/bin/python ] \|\| uv venv --clear
	uv pip install pandas numpy

	- name: Run performance benchmarks
	run: \|
	echo "========================================="
	echo "Performance Benchmark Results"
	echo "========================================="
	uv run python scripts/benchmark_all.py \|\| echo "Benchmark completed with warnings"

	- name: Upload benchmark results
	if: always()
	uses: actions/upload-artifact@v4
	with:
	name: benchmark-results
	path: \|
	benchmarks/*.txt
	benchmarks/*.csv
	retention-days: 30

Navigation Menu

ci: give the examples job the test data it needs #689

Workflow file

ci: give the examples job the test data it needs #689

Uh oh!

Workflow file for this run