qt5base-lts/tests/auto/testlib/selftests/generate_expected_output.py
Edward Welbourne 35ad157d88 Add some testing of QTestEventLoop
Fairly minimal for now, just enough to verify a bug and serve as the
sign of success when it's fixed. Tests fail in ways they shouldn't,
for now; see expected_eventloop.* for details, notably "Earlier test
failed to clean up" messages.

Pick-to: 6.4 6.3
Task-number: QTBUG-104441
Change-Id: I59be4aa5f21fed23b19a0593a8c2f6c9956507df
Reviewed-by: Jason McDonald <macadder1@gmail.com>
2022-08-08 20:49:03 +02:00

374 lines
16 KiB
Python
Executable File

#!/usr/bin/env python3
# Copyright (C) 2022 The Qt Company Ltd.
# SPDX-License-Identifier: LicenseRef-Qt-Commercial OR GPL-3.0-only WITH Qt-GPL-exception-1.0
from argparse import ArgumentParser, RawTextHelpFormatter
import os
import subprocess
import re
import sys
USAGE = """
Regenerate all tests' output.
Usage: cd to the build directory containing the directories with
the subtest binaries, invoke this script; optionally pass the names of sub-dirs
and formats to limit which tests to regenerate expected_* files for.
The saved test output is used by ./tst_selftests.cpp, which compares
it to the output of each test, ignoring various boring changes.
This script canonicalises the parts that would exhibit those boring
changes, so as to avoid noise in git (and conflicts in merges) for
the saved copies of the output.
"""
DEFAULT_FORMATS = ['xml', 'txt', 'junitxml', 'lightxml', 'teamcity', 'tap', 'csv']
TESTS = ['assert', 'badxml', 'benchlibcallgrind', 'benchlibcounting',
'benchlibeventcounter', 'benchliboptions', 'benchlibtickcounter',
'benchlibwalltime', 'blacklisted', 'cmptest', 'commandlinedata',
'counting', 'crashes', 'datatable', 'datetime', 'deleteLater',
'deleteLater_noApp', 'differentexec', 'eventloop', 'exceptionthrow',
'expectfail', "extendedcompare", 'failcleanup', 'failcleanuptestcase',
'faildatatype', 'failfetchtype', 'failinit', 'failinitdata',
'fetchbogus', 'findtestdata', 'float', 'globaldata', 'longstring',
'maxwarnings', 'mouse', 'multiexec', 'pairdiagnostics', 'pass',
'printdatatags', 'printdatatagswithglobaltags', 'qexecstringlist',
'signaldumper', 'silent', 'singleskip', 'skip', 'skipcleanup',
'skipcleanuptestcase', 'skipinit', 'skipinitdata', 'sleep', 'strcmp',
'subtest', 'testlib', 'tuplediagnostics', 'verbose1', 'verbose2',
'verifyexceptionthrown', 'warnings', 'watchdog', 'junit', 'keyboard']
class Fail (Exception): pass
class Cleaner (object):
"""Tool to clean up test output to make diff-ing runs useful.
We care about whether tests pass or fail - if that changes,
something that matters has happened - and we care about some
changes to what they say when they do fail; but we don't care
exactly what line of what file the failing line of code now
occupies, nor do we care how many milliseconds each test took to
run; and changes to the Qt version number mean nothing to us.
Create one singleton instance; it'll do mildly expensive things
once and you can use its .clean() method to tidy up your test
output."""
def __init__(self):
"""Set up the details we need for later cleaning.
Saves the directory of this
script as self.sourceDir, so client can find tst_selftests.cpp
there. Checks here does look as expected in a build tree -
raising Fail() if not - then retrieves the Qt
version (saved as .version for the benefit of clients) and
prepares the sequence of (regex, replace) pairs that .clean()
needs to do its job."""
self.version, self.sourceDir, self.__replace = self.__getPatterns()
@staticmethod
def _read_qt_version(qtbase_dir):
cmake_conf_file = os.path.join(qtbase_dir, '.cmake.conf')
with open(cmake_conf_file) as f:
for line in f:
# set(QT_REPO_MODULE_VERSION "6.1.0")
if 'set(QT_REPO_MODULE_VERSION' in line:
return line.strip().split('"')[1]
raise RuntimeError("Someone broke .cmake.conf formatting again")
@staticmethod
def __getPatterns(patterns = (
# Timings:
(r'( *<Duration msecs=)"[\d\.]+"/>', r'\1"0"/>'), # xml, lightxml
(r'(Totals:.*,) *[0-9.]+ms', r'\1 0ms'), # txt
(r'(<testsuite .*? timestamp=")[^"]+(".*>)', r'\1@TEST_START_TIME@\2'), # junit
(r'(<(testsuite|testcase) .*? time=")[^"]+(".*>)', r'\1@TEST_DURATION@\3'), # junit
# Benchmarks:
(r'[0-9,.]+( (?:CPU ticks|msecs) per iteration \(total:) [0-9,.]+ ', r'0\1 0, '), # txt
(r'(<BenchmarkResult metric="(?:CPUTicks|WalltimeMilliseconds)".*\bvalue=)"[^"]+"', r'\1"0"'), # xml, lightxml
# Build details:
(r'(Config: Using QtTest library).*', r'\1'), # txt
(r'( *<QtBuild)>[^<]+</QtBuild>', r'\1/>'), # xml, lightxml
(r'(<property name="QtBuild" value=")[^"]+"', r'\1"'), # junitxml
(r'(<testsuite .*? hostname=")[^"]+(".*>)', r'\1@HOSTNAME@\2'), # junit
# Line numbers in source files:
(r'(ASSERT: ("|&quot;).*("|&quot;) in file .*, line) \d+', r'\1 0'), # lightxml
(r'(Loc: \[[^[\]()]+)\(\d+\)', r'\1(0)'), # txt
(r'(\[Loc: [^[\]()]+)\(\d+\)', r'\1(0)'), # teamcity
(r'(<(?:Incident|Message)\b.*\bfile=.*\bline=)"\d+"', r'\1"0"'), # lightxml, xml
(r'(at: .*?):\d+\)', r'\1:0)'), # tap
(r'(line:) \d+', r'\1 0'), # tap
# Pointers printed by signal dumper:
(r'\(\b[a-f0-9]{8,}\b\)', r'(_POINTER_)'),
# Example/for reference:
# ((QString&)@55f5fbb8dd40)
# ((const QVector<int>*)7ffd671d4558)
(r'\((\((?:const )?\w+(?:<[^>]+>)?[*&]*\)@?)\b[a-f\d]{8,}\b\)', r'(\1_POINTER_)'),
# For xml output there is no '<', '>' or '&', so we need an alternate version for that:
# ((QVector&lt;int&gt;&amp;)@5608b455e640)
(r'\((\((?:const )?\w+(?:&lt;(?:[^&]|&(?!gt;))*&gt;)?(?:\*|&amp;)?\)@?)[a-z\d]+\b\)', r'(\1_POINTER_)'),
# QEventDispatcher{Glib,Win32,etc.}
(r'\bQEventDispatcher\w+\b', r'QEventDispatcherPlatform'),
),
precook = re.compile):
"""Private implementation details of __init__()."""
# Are we being run from the right place ?
scriptPath = os.path.dirname(os.path.abspath(__file__))
hereNames, depth = scriptPath.split(os.path.sep), 5
hereNames = hereNames[-depth:] # path components from qtbase down
assert hereNames[0] == 'qtbase', ('Script moved: please correct depth', hereNames)
qtbase_dir = os.path.realpath(os.path.join(scriptPath, '..', '..', '..', '..'))
qtver = Cleaner._read_qt_version(qtbase_dir)
hereNames = tuple(hereNames)
# Add path to specific sources and to tst_*.cpp if missing (for in-source builds):
patterns += ((r'(^|[^/])\b(qtestcase.cpp)\b', r'\1qtbase/src/testlib/\2'),
# Add more special cases here, if they show up !
(r'([\[" ])\.\./(counting/tst_counting.cpp)\b',
r'\1' + os.path.sep.join(hereNames + (r'\2',))),
# The common pattern:
(r'(^|[^/])\b(tst_)?([a-z]+\d*)\.cpp\b',
r'\1' + os.path.sep.join(hereNames + (r'\3', r'\2\3.cpp'))))
sentinel = os.path.sep + hereNames[0] + os.path.sep # '/qtbase/'
# Identify the path prefix of our qtbase ancestor directory
# (source, build and $PWD, when different); trim such prefixes
# off all paths we see.
roots = tuple(r[:r.find(sentinel) + 1].encode('unicode-escape').decode('utf-8')
for r in set((os.getcwd(), scriptPath, os.environ.get('PWD', '')))
if sentinel in r)
patterns += tuple((root, r'') for root in roots) + (
(r'\.'.join(qtver.split('.')), r'@INSERT_QT_VERSION_HERE@'),)
if any('-' in r for r in roots):
# Our xml formats replace hyphens with a character entity:
patterns += tuple((root.replace('-', '&#x0*2D;'), r'')
for root in roots if '-' in root)
return qtver, scriptPath, tuple((precook(p), r) for p, r in patterns)
def clean(self, data):
"""Remove volatile details from test output.
Takes the full test output as a single (possibly huge)
multi-line string; iterates over cleaned lines of output."""
for line in data.split('\n'):
# Replace all occurrences of each regex:
for searchRe, replaceExp in self.__replace:
line = searchRe.sub(replaceExp, line)
yield line
class Scanner (object):
"""Knows which subdirectories to generate output for.
Tell its constructor the name of this source directory (see
Cleaner's .sourceDir) and it'll scan tst_selftests.cpp for the
list. Its .subdirs() can then filter a user-supplied list of
subdirs or generate the full list, when the user supplied
none."""
def __init__(self):
pass
def subdirs(self, given, skip_callgrind=False):
if given:
for d in given:
if not os.path.isdir(d):
print('No such directory:', d, '- skipped')
elif skip_callgrind and d == 'benchlibcallgrind':
pass # Skip this test, as requeted.
elif d in TESTS:
yield d
else:
print(f'Directory {d} is not in the list of tests')
else:
tests = TESTS
if skip_callgrind:
tests.remove('benchlibcallgrind')
missing = 0
for d in tests:
if os.path.isdir(d):
yield d
else:
missing += 1
print(f"directory {d} doesn't exist, was it removed?")
if missing == len(tests):
print(USAGE)
del re
# Keep in sync with tst_selftests.cpp's processEnvironment():
def baseEnv(platname=None,
keep=('PATH', 'QT_QPA_PLATFORM'),
posix=('HOME', 'USER', 'QEMU_SET_ENV', 'QEMU_LD_PREFIX'),
nonapple=('DISPLAY', 'XAUTHORITY', 'XAUTHLOCALHOSTNAME'), # and XDG_*
# Don't actually know how to test for QNX, so this is ignored:
qnx=('GRAPHICS_ROOT', 'TZ'),
# Probably not actually relevant
preserveLib=('QT_PLUGIN_PATH', 'LD_LIBRARY_PATH'),
# Shall be modified on first call (a *copy* is returned):
cached={}):
"""Lazily-evaluated standard environment for sub-tests to run in.
This prunes the parent process environment, selecting a only those
variables we chose to keep. The platname passed to the first call
helps select which variables to keep. The environment computed
then is cached: a copy of this is returned on that call and each
subsequent call.\n"""
if not cached:
xdg = False
# The platform module may be more apt for the platform tests here.
if os.name == 'posix':
keep += posix
if platname != 'darwin':
keep += nonapple
xdg = True
if 'QT_PRESERVE_TESTLIB_PATH' in os.environ:
keep += preserveLib
cached = dict(
LC_ALL = 'en-US.UTF-8', # Use standard locale
# Avoid interference from any qtlogging.ini files, e.g. in
# /etc/xdg/QtProject/, (must match tst_selftests.cpp's
# processEnvironment()'s value):
QT_LOGGING_RULES = '*.debug=true;qt.*=false')
for k, v in os.environ.items():
if k in keep or (xdg and k.startswith('XDG_')):
cached[k] = v
return cached.copy()
def testEnv(testname,
# Make sure this matches tst_Selftests::doRunSubTest():
extraEnv = {
"crashers": { "QTEST_DISABLE_CORE_DUMP": "1",
"QTEST_DISABLE_STACK_DUMP": "1" },
"watchdog": { "QTEST_FUNCTION_TIMEOUT": "100" },
},
# Must match tst_Selftests::runSubTest_data():
crashers = ("assert", "blacklisted", "crashes", "crashedterminate",
"exceptionthrow", "faildatatype", "failfetchtype",
"fetchbogus", "silent", "watchdog")):
"""Determine the environment in which to run a test."""
data = baseEnv()
if testname in crashers:
data.update(extraEnv["crashers"])
if testname in extraEnv:
data.update(extraEnv[testname])
return data
def shouldIgnoreTest(testname, format):
"""Test whether to exclude a test/format combination.
See TestLogger::shouldIgnoreTest() in tst_selftests.cpp; it starts
with various exclusions for opt-in tests, platform dependencies
and tool availability; we ignore those, as we need the test data
to be present when those exclusions aren't in effect.
In the remainder, exclude what it always excludes.
"""
if format != 'txt':
if testname in ("differentexec",
"multiexec",
"qexecstringlist",
"benchliboptions",
"printdatatags",
"printdatatagswithglobaltags",
"silent",
"crashes",
"benchlibcallgrind",
"float",
"sleep"):
return True
if testname == "badxml" and not format.endswith('xml'):
return True
# Skip benchlib* for teamcity, and everything else for csv:
if format == ('teamcity' if testname.startswith('benchlib') else 'csv'):
return True
if testname == "junit" and format != "junitxml":
return True
return False
def generateTestData(test_path, expected_path, clean, formats):
"""Run one test and save its cleaned results.
Required arguments are the path to test directory (the binary
it contains is expected to have the same name), a function
that'll clean a test-run's output; see Cleaner.clean() and a list of
formats.
"""
# MS-Win: shall need to add .exe to this
testname = os.path.basename(test_path)
path = os.path.join(test_path, testname)
if not os.path.isfile(path):
print("Warning: directory", testname, "contains no test executable")
return
# Prepare environment in which to run tests:
env = testEnv(testname)
for format in formats:
if shouldIgnoreTest(testname, format):
continue
print(f' running {testname}/{format}')
cmd = [path, f'-{format}']
expected_file = f'expected_{testname}.{format}'
data = subprocess.Popen(cmd, stdout=subprocess.PIPE, env=env,
universal_newlines=True).communicate()[0]
with open(os.path.join(expected_path, expected_file), 'w') as out:
out.write('\n'.join(clean(data))) # write() appends a newline, too
def main(argv):
"""Argument parsing and driver for the real work"""
argument_parser = ArgumentParser(description=USAGE, formatter_class=RawTextHelpFormatter)
argument_parser.add_argument('--formats', '-f',
help='Comma-separated list of formats')
argument_parser.add_argument('--skip-callgrind', '-s', action='store_true',
help='Skip the (no longer expensive) benchlib callgrind test')
argument_parser.add_argument('subtests', help='subtests to regenerate',
nargs='*', type=str)
options = argument_parser.parse_args(argv[1:])
formats = options.formats.split(',') if options.formats else DEFAULT_FORMATS
cleaner = Cleaner()
src_dir = cleaner.sourceDir
if not options.skip_callgrind:
# Skip it, even if not requested, when valgrind isn't available:
try:
probe = subprocess.Popen(['valgrind', '--version'], stdout=subprocess.PIPE,
env=testEnv('benchlibcallgrind'), universal_newlines=True)
except FileNotFoundError:
options.skip_callgrind = True
print("Failed to find valgrind, skipping benchlibcallgrind test")
tests = tuple(Scanner().subdirs(options.subtests, options.skip_callgrind))
print("Generating", len(tests), "test results for", cleaner.version, "in:", src_dir)
for path in tests:
generateTestData(path, src_dir, cleaner.clean, formats)
if __name__ == '__main__':
# Executed when script is run, not when imported (e.g. to debug)
baseEnv(sys.platform) # initializes its cache
if sys.platform.startswith('win'):
print("This script does not work on Windows.")
exit()
try:
main(sys.argv)
except Fail as what:
sys.stderr.write('Failed: ' + ' '.join(what.args) + '\n')
exit(1)