cpython/Lib/test/libregrtest/runtest.py

446 lines
14 KiB
Python
Raw Normal View History

import faulthandler
import functools
import gc
import importlib
import io
import os
import sys
import time
import traceback
import unittest
from test import support
from test.support import os_helper
from test.support import threading_helper
from test.libregrtest.cmdline import Namespace
from test.libregrtest.save_env import saved_test_environment
from test.libregrtest.utils import clear_caches, format_duration, print_warning
class TestResult:
def __init__(
self,
name: str,
duration_sec: float = 0.0,
xml_data: list[str] | None = None,
) -> None:
self.name = name
self.duration_sec = duration_sec
self.xml_data = xml_data
def __str__(self) -> str:
return f"{self.name} finished"
class Passed(TestResult):
def __str__(self) -> str:
return f"{self.name} passed"
class Failed(TestResult):
def __init__(
self,
name: str,
duration_sec: float = 0.0,
xml_data: list[str] | None = None,
errors: list[tuple[str, str]] | None = None,
failures: list[tuple[str, str]] | None = None,
) -> None:
super().__init__(name, duration_sec=duration_sec, xml_data=xml_data)
self.errors = errors
self.failures = failures
def __str__(self) -> str:
if self.errors and self.failures:
le = len(self.errors)
lf = len(self.failures)
error_s = "error" + ("s" if le > 1 else "")
failure_s = "failure" + ("s" if lf > 1 else "")
return f"{self.name} failed ({le} {error_s}, {lf} {failure_s})"
if self.errors:
le = len(self.errors)
error_s = "error" + ("s" if le > 1 else "")
return f"{self.name} failed ({le} {error_s})"
if self.failures:
lf = len(self.failures)
failure_s = "failure" + ("s" if lf > 1 else "")
return f"{self.name} failed ({lf} {failure_s})"
return f"{self.name} failed"
class UncaughtException(Failed):
def __str__(self) -> str:
return f"{self.name} failed (uncaught exception)"
class EnvChanged(Failed):
def __str__(self) -> str:
return f"{self.name} failed (env changed)"
# Convert Passed to EnvChanged
@staticmethod
def from_passed(other):
return EnvChanged(other.name, other.duration_sec, other.xml_data)
class RefLeak(Failed):
def __str__(self) -> str:
return f"{self.name} failed (reference leak)"
class Skipped(TestResult):
def __str__(self) -> str:
return f"{self.name} skipped"
class ResourceDenied(Skipped):
def __str__(self) -> str:
return f"{self.name} skipped (resource denied)"
class Interrupted(TestResult):
def __str__(self) -> str:
return f"{self.name} interrupted"
class ChildError(Failed):
def __str__(self) -> str:
return f"{self.name} crashed"
class DidNotRun(TestResult):
def __str__(self) -> str:
return f"{self.name} ran no tests"
class Timeout(Failed):
def __str__(self) -> str:
return f"{self.name} timed out ({format_duration(self.duration_sec)})"
# Minimum duration of a test to display its duration or to mention that
# the test is running in background
PROGRESS_MIN_TIME = 30.0 # seconds
gh-82054: allow test runner to split test_asyncio to execute in parallel by sharding. (#103927) This runs test_asyncio sub-tests in parallel using sharding from Cinder. This suite is typically the longest-pole in runs because it is a test package with a lot of further sub-tests otherwise run serially. By breaking out the sub-tests as independent modules we can run a lot more in parallel. After porting we can see the direct impact on a multicore system. Without this change: Running make test is 5 min 26 seconds With this change: Running make test takes 3 min 39 seconds That'll vary based on system and parallelism. On a `-j 4` run similar to what CI and buildbot systems often do, it reduced the overall test suite completion latency by 10%. The drawbacks are that this implementation is hacky and due to the sorting of the tests it obscures when the asyncio tests occur and involves changing CPython test infrastructure but, the wall time saved it is worth it, especially in low-core count CI runs as it pulls a long tail. The win for productivity and reserved CI resource usage is significant. Future tests that deserve to be refactored into split up suites to benefit from are test_concurrent_futures and the way the _test_multiprocessing suite gets run for all start methods. As exposed by passing the -o flag to python -m test to get a list of the 10 longest running tests. --------- Co-authored-by: Carl Meyer <carl@oddbird.net> Co-authored-by: Gregory P. Smith <greg@krypto.org> [Google, LLC]
2023-04-29 22:26:24 -03:00
#If these test directories are encountered recurse into them and treat each
# test_ .py or dir as a separate test module. This can increase parallelism.
# Beware this can't generally be done for any directory with sub-tests as the
# __init__.py may do things which alter what tests are to be run.
SPLITTESTDIRS = {
"test_asyncio",
"test_multiprocessing_fork",
"test_multiprocessing_forkserver",
"test_multiprocessing_spawn",
gh-82054: allow test runner to split test_asyncio to execute in parallel by sharding. (#103927) This runs test_asyncio sub-tests in parallel using sharding from Cinder. This suite is typically the longest-pole in runs because it is a test package with a lot of further sub-tests otherwise run serially. By breaking out the sub-tests as independent modules we can run a lot more in parallel. After porting we can see the direct impact on a multicore system. Without this change: Running make test is 5 min 26 seconds With this change: Running make test takes 3 min 39 seconds That'll vary based on system and parallelism. On a `-j 4` run similar to what CI and buildbot systems often do, it reduced the overall test suite completion latency by 10%. The drawbacks are that this implementation is hacky and due to the sorting of the tests it obscures when the asyncio tests occur and involves changing CPython test infrastructure but, the wall time saved it is worth it, especially in low-core count CI runs as it pulls a long tail. The win for productivity and reserved CI resource usage is significant. Future tests that deserve to be refactored into split up suites to benefit from are test_concurrent_futures and the way the _test_multiprocessing suite gets run for all start methods. As exposed by passing the -o flag to python -m test to get a list of the 10 longest running tests. --------- Co-authored-by: Carl Meyer <carl@oddbird.net> Co-authored-by: Gregory P. Smith <greg@krypto.org> [Google, LLC]
2023-04-29 22:26:24 -03:00
}
# Storage of uncollectable objects
FOUND_GARBAGE = []
def is_failed(result: TestResult, ns: Namespace) -> bool:
if isinstance(result, EnvChanged):
return ns.fail_env_changed
return isinstance(result, Failed)
def findtestdir(path=None):
return path or os.path.dirname(os.path.dirname(__file__)) or os.curdir
def findtests(*, testdir=None, exclude=(),
split_test_dirs=SPLITTESTDIRS, base_mod=""):
"""Return a list of all applicable test modules."""
testdir = findtestdir(testdir)
tests = []
for name in os.listdir(testdir):
mod, ext = os.path.splitext(name)
if (not mod.startswith("test_")) or (mod in exclude):
continue
if mod in split_test_dirs:
subdir = os.path.join(testdir, mod)
mod = f"{base_mod or 'test'}.{mod}"
tests.extend(findtests(testdir=subdir, exclude=exclude,
split_test_dirs=split_test_dirs, base_mod=mod))
elif ext in (".py", ""):
tests.append(f"{base_mod}.{mod}" if base_mod else mod)
return sorted(tests)
def split_test_packages(tests, *, testdir=None, exclude=(),
split_test_dirs=SPLITTESTDIRS):
testdir = findtestdir(testdir)
splitted = []
for name in tests:
if name in split_test_dirs:
subdir = os.path.join(testdir, name)
splitted.extend(findtests(testdir=subdir, exclude=exclude,
split_test_dirs=split_test_dirs,
base_mod=name))
else:
splitted.append(name)
return splitted
def get_abs_module(ns: Namespace, test_name: str) -> str:
if test_name.startswith('test.') or ns.testdir:
return test_name
else:
# Import it from the test package
return 'test.' + test_name
def _runtest(ns: Namespace, test_name: str) -> TestResult:
# Handle faulthandler timeout, capture stdout+stderr, XML serialization
# and measure time.
output_on_failure = ns.verbose3
use_timeout = (
ns.timeout is not None and threading_helper.can_start_thread
)
if use_timeout:
faulthandler.dump_traceback_later(ns.timeout, exit=True)
start_time = time.perf_counter()
try:
support.set_match_tests(ns.match_tests, ns.ignore_tests)
support.junit_xml_list = xml_list = [] if ns.xmlpath else None
if ns.failfast:
support.failfast = True
if output_on_failure:
support.verbose = True
stream = io.StringIO()
orig_stdout = sys.stdout
orig_stderr = sys.stderr
print_warning = support.print_warning
orig_print_warnings_stderr = print_warning.orig_stderr
output = None
try:
sys.stdout = stream
sys.stderr = stream
# print_warning() writes into the temporary stream to preserve
# messages order. If support.environment_altered becomes true,
# warnings will be written to sys.stderr below.
print_warning.orig_stderr = stream
result = _runtest_inner(ns, test_name,
display_failure=False)
if not isinstance(result, Passed):
output = stream.getvalue()
finally:
sys.stdout = orig_stdout
sys.stderr = orig_stderr
print_warning.orig_stderr = orig_print_warnings_stderr
if output is not None:
sys.stderr.write(output)
sys.stderr.flush()
else:
# Tell tests to be moderately quiet
support.verbose = ns.verbose
result = _runtest_inner(ns, test_name,
display_failure=not ns.verbose)
if xml_list:
import xml.etree.ElementTree as ET
result.xml_data = [
ET.tostring(x).decode('us-ascii')
for x in xml_list
]
result.duration_sec = time.perf_counter() - start_time
return result
finally:
if use_timeout:
faulthandler.cancel_dump_traceback_later()
support.junit_xml_list = None
def runtest(ns: Namespace, test_name: str) -> TestResult:
"""Run a single test.
ns -- regrtest namespace of options
test_name -- the name of the test
Returns a TestResult sub-class depending on the kind of result received.
If ns.xmlpath is not None, xml_data is a list containing each
generated testsuite element.
"""
try:
return _runtest(ns, test_name)
except:
if not ns.pgo:
msg = traceback.format_exc()
print(f"test {test_name} crashed -- {msg}",
file=sys.stderr, flush=True)
return Failed(test_name)
def _test_module(the_module):
loader = unittest.TestLoader()
tests = loader.loadTestsFromModule(the_module)
for error in loader.errors:
print(error, file=sys.stderr)
if loader.errors:
raise Exception("errors while loading tests")
support.run_unittest(tests)
def save_env(ns: Namespace, test_name: str):
return saved_test_environment(test_name, ns.verbose, ns.quiet, pgo=ns.pgo)
def _runtest_inner2(ns: Namespace, test_name: str) -> bool:
# Load the test function, run the test function, handle huntrleaks
# to detect leaks.
abstest = get_abs_module(ns, test_name)
# remove the module from sys.module to reload it if it was already imported
try:
del sys.modules[abstest]
except KeyError:
pass
the_module = importlib.import_module(abstest)
if ns.huntrleaks:
from test.libregrtest.refleak import dash_R
# If the test has a test_main, that will run the appropriate
# tests. If not, use normal unittest test loading.
test_runner = getattr(the_module, "test_main", None)
if test_runner is None:
test_runner = functools.partial(_test_module, the_module)
try:
with save_env(ns, test_name):
if ns.huntrleaks:
# Return True if the test leaked references
refleak = dash_R(ns, test_name, test_runner)
else:
test_runner()
refleak = False
finally:
# First kill any dangling references to open files etc.
# This can also issue some ResourceWarnings which would otherwise get
# triggered during the following test run, and possibly produce
# failures.
support.gc_collect()
cleanup_test_droppings(test_name, ns.verbose)
if gc.garbage:
support.environment_altered = True
print_warning(f"{test_name} created {len(gc.garbage)} "
f"uncollectable object(s).")
# move the uncollectable objects somewhere,
# so we don't see them again
FOUND_GARBAGE.extend(gc.garbage)
gc.garbage.clear()
support.reap_children()
return refleak
def _runtest_inner(
ns: Namespace, test_name: str, display_failure: bool = True
) -> TestResult:
# Detect environment changes, handle exceptions.
# Reset the environment_altered flag to detect if a test altered
# the environment
support.environment_altered = False
if ns.pgo:
display_failure = False
try:
clear_caches()
support.gc_collect()
with save_env(ns, test_name):
refleak = _runtest_inner2(ns, test_name)
except support.ResourceDenied as msg:
if not ns.quiet and not ns.pgo:
print(f"{test_name} skipped -- {msg}", flush=True)
return ResourceDenied(test_name)
except unittest.SkipTest as msg:
if not ns.quiet and not ns.pgo:
print(f"{test_name} skipped -- {msg}", flush=True)
return Skipped(test_name)
except support.TestFailedWithDetails as exc:
msg = f"test {test_name} failed"
if display_failure:
msg = f"{msg} -- {exc}"
print(msg, file=sys.stderr, flush=True)
return Failed(test_name, errors=exc.errors, failures=exc.failures)
except support.TestFailed as exc:
msg = f"test {test_name} failed"
if display_failure:
msg = f"{msg} -- {exc}"
print(msg, file=sys.stderr, flush=True)
return Failed(test_name)
except support.TestDidNotRun:
return DidNotRun(test_name)
except KeyboardInterrupt:
print()
return Interrupted(test_name)
except:
if not ns.pgo:
msg = traceback.format_exc()
print(f"test {test_name} crashed -- {msg}",
file=sys.stderr, flush=True)
return UncaughtException(test_name)
if refleak:
return RefLeak(test_name)
if support.environment_altered:
return EnvChanged(test_name)
return Passed(test_name)
def cleanup_test_droppings(test_name: str, verbose: int) -> None:
# Try to clean up junk commonly left behind. While tests shouldn't leave
# any files or directories behind, when a test fails that can be tedious
# for it to arrange. The consequences can be especially nasty on Windows,
# since if a test leaves a file open, it cannot be deleted by name (while
# there's nothing we can do about that here either, we can display the
# name of the offending test, which is a real help).
for name in (os_helper.TESTFN,):
if not os.path.exists(name):
continue
if os.path.isdir(name):
import shutil
kind, nuker = "directory", shutil.rmtree
elif os.path.isfile(name):
kind, nuker = "file", os.unlink
else:
raise RuntimeError(f"os.path says {name!r} exists but is neither "
f"directory nor file")
if verbose:
print_warning(f"{test_name} left behind {kind} {name!r}")
support.environment_altered = True
try:
import stat
# fix possible permissions problems that might prevent cleanup
os.chmod(name, stat.S_IRWXU | stat.S_IRWXG | stat.S_IRWXO)
nuker(name)
except Exception as exc:
print_warning(f"{test_name} left behind {kind} {name!r} "
f"and it couldn't be removed: {exc}")