Test

Bump actions/checkout from 4.2.2 to 6.0.3 (#69) #92

Workflow file for this run

	name: Test

	# Runs on pushes to the default branch AND on every pull request, so the
	# matrix can be used as a required merge gate. (Baseline only ran on push.)
	on:
	push:
	branches: [master]
	pull_request:
	workflow_dispatch:

	# Cancel superseded runs for the same ref to save runner minutes.
	concurrency:
	group: test-${{ github.ref }}
	cancel-in-progress: true

	permissions:
	contents: read

	jobs:
	# ---------------------------------------------------------------------------
	# Lint. rustfmt is enforced (blocking) now that the source is rustfmt-clean.
	# clippy and mypy stay advisory (continue-on-error) until their debt clears,
	# and each runs regardless of the others:
	# - clippy -> flip to blocking after the str->bytes fix (#19)
	# - mypy -> flip to blocking after the .pyi stubs type-check clean (#42)
	# ---------------------------------------------------------------------------
	lint:
	name: Lint
	runs-on: ubuntu-latest
	steps:
	- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
	- uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5
	with:
	python-version: "3.10"
	- uses: dtolnay/rust-toolchain@3c5f7ea28cd621ae0bf5283f0e981fb97b8a7af9 # master
	with:
	toolchain: "1.83"
	components: rustfmt, clippy
	- uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1

	- name: cargo fmt --check
	run: cargo fmt --all -- --check

	# clippy::cast_possible_truncation pinpoints the str->bytes corruption bug.
	- name: cargo clippy
	if: always()
	run: cargo clippy --all-targets -- -D warnings -W clippy::cast_possible_truncation
	continue-on-error: true

	- name: mypy (type stubs)
	if: always()
	run: \|
	python -m pip install --upgrade pip mypy
	mypy fast_mail_parser/
	continue-on-error: true

	# ---------------------------------------------------------------------------
	# Supply-chain audit. Blocking (no continue-on-error): the dependency stack is
	# current (PyO3 0.29) and the Cargo.lock audits clean, so a new advisory
	# against any dependency should fail the build and gate the PR rather than pass
	# silently. (It was previously advisory only because the stack was stale and a
	# fresh advisory against a pinned dep would have blocked unrelated PRs.)
	# cargo-audit needs a newer rustc than the repo's pinned 1.83, so the job runs
	# on stable via RUSTUP_TOOLCHAIN (overrides the rust-toolchain file).
	#
	# cargo-audit is `cargo install`ed from source (~3 min — previously the
	# slowest job by far). The compiled binary is cached and keyed on the pinned
	# CARGO_AUDIT_VERSION: a cache hit skips the build (restores in seconds); a
	# miss builds it once and caches it.
	#
	# Cache invalidation: to upgrade cargo-audit, bump CARGO_AUDIT_VERSION below.
	# That installs the new version AND invalidates the cache in one change — the
	# version IS the cache key, so there is no separate suffix to remember. The
	# key is intentionally NOT tied to the rustc/cargo version: the binary runs
	# fine across toolchain updates, and keying on it would force a needless
	# ~3-min rebuild on every stable bump. The RustSec advisory DB is fetched
	# fresh on every run, so findings stay current regardless of the cached
	# binary version.
	# ---------------------------------------------------------------------------
	security-audit:
	name: cargo audit
	runs-on: ubuntu-latest
	env:
	RUSTUP_TOOLCHAIN: stable
	# Bump to upgrade cargo-audit; this also invalidates the binary cache.
	CARGO_AUDIT_VERSION: "0.22.2"
	steps:
	- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
	- uses: dtolnay/rust-toolchain@3c5f7ea28cd621ae0bf5283f0e981fb97b8a7af9 # master
	with:
	toolchain: stable
	- name: Cache cargo-audit binary
	id: cache-cargo-audit
	uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
	with:
	path: ~/.cargo/bin/cargo-audit
	key: cargo-audit-${{ runner.os }}-${{ env.CARGO_AUDIT_VERSION }}
	- name: Install cargo-audit
	if: steps.cache-cargo-audit.outputs.cache-hit != 'true'
	run: cargo install cargo-audit --version ${{ env.CARGO_AUDIT_VERSION }} --locked
	- name: cargo audit
	run: cargo audit

	# ---------------------------------------------------------------------------
	# Test matrix — the real merge gate. Builds & passes on CPython 3.9–3.13 with
	# PyO3 0.29 (3.7/3.8 are EOL and unavailable on current ubuntu runners). 3.13
	# is a required cell now that PyO3 0.29 supports it (was allowed-failure under
	# the old 0.16 pin).
	# ---------------------------------------------------------------------------
	test:
	name: Test (py${{ matrix.python-version }})
	runs-on: ubuntu-latest
	strategy:
	fail-fast: false
	matrix:
	python-version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
	steps:
	- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
	- uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5
	with:
	python-version: ${{ matrix.python-version }}
	cache: pip
	- name: Install the package
	run: make install
	- name: Install test dependencies
	run: make install-test
	- name: Test
	run: make test

	# ---------------------------------------------------------------------------
	# Benchmark quality gate. The benchmark used to run inside every matrix cell
	# (6x redundant, slow). It now runs once here and acts as a performance gate:
	# fast_mail_parser must stay at least BENCH_MIN_SPEEDUP times faster than the
	# pure-Python mail-parser baseline (README reports ~8x). The ratio is measured
	# on the same runner in the same run, so it is hardware-independent and stable.
	# Getting faster always passes; only regressions below the floor fail.
	# ---------------------------------------------------------------------------
	benchmark:
	name: Benchmark quality gate
	runs-on: ubuntu-latest
	steps:
	- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
	- uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5
	with:
	python-version: "3.10"
	cache: pip
	- name: Install the package + test dependencies
	run: \|
	make install
	make install-test
	- name: Run benchmark
	run: pytest -v tests/benchmark --benchmark-min-rounds=25 --benchmark-json=benchmark.json
	- name: Enforce performance gate
	env:
	# Gate on min-time ratio. 8x matches the README headline figure and
	# requires the str-input fast path; the optimized parser measures
	# ~8.7x on CI. Headroom is tight, so a slow-relative runner can dip
	# below 8x.
	BENCH_MIN_SPEEDUP: "8.0"
	run: python .github/scripts/check_benchmark.py benchmark.json
	- name: Upload benchmark report
	if: always()
	uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4
	with:
	name: benchmark-json
	path: benchmark.json

Navigation Menu

Bump actions/checkout from 4.2.2 to 6.0.3 (#69) #92

Workflow file

Bump actions/checkout from 4.2.2 to 6.0.3 (#69) #92

Uh oh!

Workflow file for this run