Add CPython to CI

.circleci/config.yml

@@ -17,6 +17,14 @@ executors:
       EMSDK_NOTTY: "1"
       EMTEST_WASI_SYSROOT: "~/wasi-sdk/wasi-sysroot"
       EMTEST_DETECT_TEMPFILE_LEAKS: "1"
+  linux-python:
+    # We need Python >= 3.12.
+    docker:
+      - image: cimg/python:3.13
+    environment:
+      LANG: "C.UTF-8"
+      EMCC_CORES: "4"
+      EMSDK_NOTTY: "1"
   linux-arm64:
     machine:
       image: ubuntu-2204:current
@@ -1053,6 +1061,30 @@ jobs:
       - run:
           name: Test size benchmarker
           command: EMTEST_BENCHMARKERS=size ./test/runner benchmark.test_primes
+  test-cpython:
+    # Build CPython against this emscripten checkout and run its test suite.
+    # CPython's test suite exercises a lot of libc/syscall behaviour that isn't
+    # covered elsewhere, and emscripten releases have historically regressed
+    # CPython.
+    #
+    # CPython's Platforms/emscripten build scripts need Python >= 3.12, hence
+    # the different executor.
+    executor: linux-python
+    resource_class: large
+    steps:
+      - checkout
+      - pip-install
+      - install-emsdk
+      - run:
+          name: Sync and build CPython
+          command: |
+            ./test/cpython_test.py sync
+            ./test/cpython_test.py build
+      - run:
+          name: Test CPython
+          no_output_timeout: 30m
+          # TODO(hoodmane): Pass --junit-xml and add upload-test-results
+          command: ./test/cpython_test.py test
   test-browser-chrome:
     executor: ubuntu-lts
     environment:
@@ -1353,6 +1385,7 @@ workflows:
       - test-misc:
           requires:
             - build-linux
+      - test-cpython
       - test-modularize-instance:
           requires:
             - build-linux

test/cpython_test.py

@@ -0,0 +1,217 @@
+#!/usr/bin/env python3
+# Copyright 2026 The Emscripten Authors.  All rights reserved.
+# Emscripten is available under two separate licenses, the MIT license and the
+# University of Illinois/NCSA Open Source License.  Both these licenses can be
+# found in the LICENSE file.
+
+"""Build CPython against this emscripten checkout and run its test suite.
+
+This script is run in CI, but is also runnable locally. CPython is
+shallow-cloned into `out/cpython` and synced to a pinned revision. Requires
+Python >= 3.12.
+
+  ./test/cpython_test.py                      # same as `all`: sync, build and test
+  ./test/cpython_test.py sync                 # clone/checkout the pinned CPython revision
+  ./test/cpython_test.py build                # build an existing checkout
+  ./test/cpython_test.py test                 # run the tests on an existing build
+  ./test/cpython_test.py test test_os test_io # run only specific tests
+
+"""
+
+import argparse
+import os
+import subprocess
+import sys
+from pathlib import Path
+
+__rootpath__ = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+sys.path.insert(0, __rootpath__)
+
+from tools import config
+from tools.utils import path_from_root
+
+CPYTHON_REPO = 'https://github.com/python/cpython.git'
+
+# The CPython revision to test.  This must be recent enough to not rely on
+# `FS.filesystems`.  Once CPython 3.16.0a2 comes out we can switch to a tag.
+CPYTHON_REVISION = '069aac92c8c40e53b6931a4d31029c5c7bef0c15'
+
+DEFAULT_CPYTHON_DIR = path_from_root('out', 'cpython')
+
+# Minimum Python version required to run CPython's Platforms/emscripten build
+# scripts
+MIN_BUILD_PYTHON = (3, 12)
+
+
+IGNORED_TESTS = [
+  # Asserts that Emscripten version matches the one pinned by CPython
+  'test_libc_ver',
+]
+
+
+def run(cmd, **kwargs):
+  print('+ ' + ' '.join(str(c) for c in cmd), flush=True)
+  return subprocess.run([str(c) for c in cmd], check=True, **kwargs)
+
+
+def node_path():
+  # config.NODE_JS is always normalized to a list (see config.listify).
+  if not config.NODE_JS:
+    sys.exit('error: NODE_JS not set in emscripten config')
+  return config.NODE_JS[0]
+
+
+def git_head(cwd):
+  """Return the sha of HEAD, or None if the repo has no commits yet."""
+  proc = subprocess.run(['git', 'rev-parse', '--verify', '-q', 'HEAD'], cwd=cwd,
+                        text=True, capture_output=True)
+  return proc.stdout.strip() if proc.returncode == 0 else None
+
+
+def sync(args):
+  """Ensure a CPython checkout exists at `cpython_dir` and is at `revision`."""
+  cpython_dir = Path(args.cpython_dir)
+  revision = args.revision
+  if not (cpython_dir / '.git').exists():
+    print(f'Cloning CPython into {cpython_dir}')
+    cpython_dir.mkdir(parents=True, exist_ok=True)
+    run(['git', 'init', '-q', cpython_dir])
+    run(['git', 'remote', 'add', 'origin', CPYTHON_REPO], cwd=cpython_dir)
+  elif git_head(cpython_dir) == revision:
+    print(f'CPython checkout at {cpython_dir} is already at {revision}')
+    return 0
+
+  print(f'Syncing CPython checkout at {cpython_dir} to {revision}')
+  # Shallow fetch of a single commit rather than a full clone.  GitHub allows
+  # fetching arbitrary commits by sha.
+  run(['git', 'fetch', '-q', '--depth=1', 'origin', revision], cwd=cpython_dir)
+  run(['git', 'checkout', '-q', '--detach', 'FETCH_HEAD'], cwd=cpython_dir)
+  return 0
+
+
+def build_env():
+  """Environment for running the CPython build scripts and the built python.
+
+  Ensures that emcc/emconfigure/etc from this checkout are used.
+  """
+  env = os.environ.copy()
+  env['PATH'] = path_from_root() + os.pathsep + env.get('PATH', '')
+  return env
+
+
+def build_script_runner(args):
+  """Return (cpython_dir, step) where step(*cmd) runs CPython's Platforms/emscripten script."""
+  cpython_dir = Path(args.cpython_dir)
+  if not (cpython_dir / 'configure').exists():
+    sys.exit(f'error: no CPython checkout at {cpython_dir}; run `{sys.argv[0]} sync` first')
+  build_script = cpython_dir / 'Platforms' / 'emscripten'
+  if not build_script.exists():
+    sys.exit(f'error: {build_script} not found; CPython checkout is too old')
+  env = build_env()
+
+  def step(*cmd):
+    run([sys.executable, build_script, *cmd], cwd=cpython_dir, env=env)
+
+  return cpython_dir, step
+
+
+def build(args):
+  cpython_dir, step = build_script_runner(args)
+  print(f'Building CPython checkout at {cpython_dir} ({git_head(cpython_dir)})')
+
+  configure_args = ['--config-cache']
+  if args.pydebug:
+    configure_args.append('--with-pydebug')
+  step('configure-build-python', '--', *configure_args)
+  step('make-build-python')
+  step('make-dependencies')
+  step('configure-host', '--host-runner', node_path(), '--', '--config-cache')
+  step('make-host')
+  step('pythoninfo-host')
+  return 0
+
+
+def test(args):
+  cpython_dir, step = build_script_runner(args)
+  python_sh = cpython_dir / 'cross-build' / 'wasm32-emscripten' / 'build' / 'python' / 'python.sh'
+  if not python_sh.exists():
+    sys.exit(f'error: {python_sh} not found; run `{sys.argv[0]} build` first')
+
+  # `run --test` prepends the default test arguments from CPython's config.toml
+  test_args = []
+  for t in IGNORED_TESTS:
+    test_args += ['-i', t]
+  if args.junit_xml:
+    junit_xml = os.path.abspath(args.junit_xml)
+    os.makedirs(os.path.dirname(junit_xml), exist_ok=True)
+    test_args += ['--junit-xml', junit_xml]
+  test_args += args.test_args
+  step('run', '--test', '--', *test_args)
+  return 0
+
+
+def all_steps(args):
+  sync(args)
+  build(args)
+  test(args)
+  return 0
+
+
+def main(argv):
+  if sys.version_info < (3, 12):
+    sys.exit("Requires Python version >= 3.12.0")
+
+  parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
+  subparsers = parser.add_subparsers(dest='subcommand')
+
+  def add_common_args(p):
+    p.add_argument('--cpython-dir', default=DEFAULT_CPYTHON_DIR,
+                   help=f'CPython checkout to use (default: {DEFAULT_CPYTHON_DIR})')
+
+  def add_sync_args(p):
+    p.add_argument('--revision', default=CPYTHON_REVISION,
+                   help=f'CPython git revision to sync to (default: {CPYTHON_REVISION})')
+
+  def add_build_args(p):
+    p.add_argument('--pydebug', action='store_true', help='Build CPython with --with-pydebug')
+
+  def add_test_args(p):
+    p.add_argument('--junit-xml', help='Write test results in junit xml format to this file')
+    p.add_argument('test_args', nargs='*',
+                   help='Extra arguments for `python -m test` (use `--` before any that start with `-`)')
+
+  all_parser = subparsers.add_parser('all', help='Sync, build and test CPython (the default)')
+  sync_parser = subparsers.add_parser('sync', help='Clone/checkout the pinned CPython revision')
+  build_parser = subparsers.add_parser('build', help='Build an existing CPython checkout')
+  test_parser = subparsers.add_parser('test', help='Run the test suite of an existing CPython build')
+  for p in (all_parser, sync_parser, build_parser, test_parser):
+    add_common_args(p)
+  for p in (all_parser, sync_parser):
+    add_sync_args(p)
+  for p in (all_parser, build_parser):
+    add_build_args(p)
+  for p in (all_parser, test_parser):
+    add_test_args(p)
+
+  # Default to `all` if no subcommand is given.
+  if not argv or (argv[0] not in subparsers.choices and argv[0] not in {'-h', '--help'}):
+    argv = ['all', *argv]
+  args = parser.parse_args(argv)
+
+  dispatch = {
+    'all': all_steps,
+    'sync': sync,
+    'build': build,
+    'test': test,
+  }
+  try:
+    return dispatch[args.subcommand](args)
+  except subprocess.CalledProcessError as e:
+    # The failing command has already printed its own output; avoid a noisy
+    # traceback and just propagate its exit code.
+    print(f'error: command failed with exit code {e.returncode}: {" ".join(e.cmd)}', file=sys.stderr)
+    return e.returncode
+
+
+if __name__ == '__main__':
+  sys.exit(main(sys.argv[1:]))