def run_command(
self,
args,
cwd=None,
env=None,
input=None, # pylint: disable=redefined-builtin
timeout_seconds=None,
error_handler=None,
return_exit_code=False,
return_stderr=True,
ignore_stderr=False,
decode_output=True,
debug_logging=True):
self._append_call(args, cwd=cwd, input=input, env=env)
assert isinstance(args, list) or isinstance(args, tuple)
if self._should_log:
env_string = ''
if env:
env_string = ', env=%s' % env
input_string = ''
if input:
input_string = ', input=%s' % input
_log.info('MOCK run_command: %s, cwd=%s%s%s', args, cwd,
env_string, input_string)
if self._exception:
raise self._exception # pylint: disable=raising-bad-type
if self._should_throw:
raise ScriptError('MOCK ScriptError',
output=self._output,
exit_code=self._exit_code)
if self._run_command_fn:
return self._run_command_fn(args)
if return_exit_code:
return self._exit_code
if self._exit_code and error_handler:
script_error = ScriptError(
script_args=args,
exit_code=self._exit_code,
output=self._output)
error_handler(script_error)
output = self._output
if return_stderr:
output += self._stderr
if decode_output and not isinstance(output, six.text_type):
output = output.decode('utf-8')
return output
def update_expectations(self):
"""Downloads text new baselines and adds test expectations lines.
Returns:
A pair: A set of tests that are rebaselined, and a dictionary
mapping tests that couldn't be rebaselined to lists of expectation
lines written to TestExpectations.
"""
issue_number = self.get_issue_number()
if issue_number == 'None':
raise ScriptError('No issue on current branch.')
build_to_status = self.get_latest_try_jobs()
_log.debug('Latest try jobs: %r', build_to_status)
if not build_to_status:
raise ScriptError('No try job information was collected.')
# Here we build up a dict of failing test results for all platforms.
test_expectations = {}
for build, job_status in build_to_status.iteritems():
if job_status.result == 'SUCCESS':
self.configs_with_all_pass.extend(
self.get_builder_configs(build))
continue
result_dicts = self.get_failing_results_dicts(build)
for result_dict in result_dicts:
test_expectations = self.merge_dicts(
test_expectations, result_dict)
# At this point, test_expectations looks like: {
# 'test-with-failing-result': {
# config1: SimpleTestResult,
# config2: SimpleTestResult,
# config3: AnotherSimpleTestResult
# }
# }
# And then we merge results for different platforms that had the same results.
for test_name, platform_result in test_expectations.iteritems():
# platform_result is a dict mapping platforms to results.
test_expectations[test_name] = self.merge_same_valued_keys(
platform_result)
# At this point, test_expectations looks like: {
# 'test-with-failing-result': {
# (config1, config2): SimpleTestResult,
# (config3,): AnotherSimpleTestResult
# }
# }
rebaselined_tests, test_expectations = self.download_text_baselines(
test_expectations)
exp_lines_dict = self.write_to_test_expectations(test_expectations)
return rebaselined_tests, exp_lines_dict
def generate_failing_results_dict(self, build, web_test_results):
"""Makes a dict with results for one platform.
Args:
builder: Builder instance containing builder information..
web_test_results: A list of WebTestResult objects.
Returns:
A dictionary with the structure: {
'test-name': {
('full-port-name',): SimpleTestResult
}
}
"""
test_dict = {}
configs = self.get_builder_configs(build, web_test_results)
_log.debug(
'Getting failing results dictionary for %s step in latest %s build',
web_test_results.step_name(), build.builder_name)
if len(configs) > 1:
raise ScriptError('More than one configs were produced for'
' builder and web tests step combination')
if not configs:
raise ScriptError('No configuration was found for builder and web test'
' step combination ')
config = configs[0]
for result in web_test_results.didnt_run_as_expected_results():
# TODO(rmhasan) If a test fails unexpectedly then it runs multiple
# times until, it passes or a retry limit is reached. Even though
# it passed we there are still flaky failures that we are not
# creating test expectations for. Maybe we should add a mode
# which creates expectations for tests that are flaky but still
# pass in a web test step.
# Create flaky expectations for flaky tests on Android. In order to
# do this we should add 'Pass' to all tests with failing
# expectations that pass in the patchset's try job.
if result.did_pass() and not self.options.include_unexpected_pass:
continue
test_name = result.test_name()
if not self._is_wpt_test(test_name):
continue
test_dict[test_name] = {
config:
SimpleTestResult(
expected=result.expected_results(),
actual=result.actual_results(),
bug=self.UMBRELLA_BUG)
}
return test_dict
def test_message_with_tuple(self):
error = ScriptError('', ('my', 'command'), -1, 'My output.',
'/Users/username/blah')
self.assertEqual(
error.message_with_output(),
'Failed to run "(\'my\', \'command\')" exit_code: -1 cwd: /Users/username/blah\n\noutput: My output.'
)
def _remote_branch_ref(self):
# Use references so that we can avoid collisions, e.g. we don't want to operate on refs/heads/trunk if it exists.
remote_main_ref = 'refs/remotes/origin/main'
if self._branch_ref_exists(remote_main_ref):
return remote_main_ref
error_msg = "Can't find a branch to diff against. %s does not exist" % remote_main_ref
raise ScriptError(message=error_msg)
def update_expectations(self):
"""Downloads text new baselines and adds test expectations lines.
Returns:
A pair: A set of tests that are rebaselined, and a dictionary
mapping tests that couldn't be rebaselined to lists of expectation
lines written to TestExpectations.
"""
issue_number = self.get_issue_number()
if issue_number == 'None':
raise ScriptError('No issue on current branch.')
build_to_status = self.get_latest_try_jobs()
_log.debug('Latest try jobs: %r', build_to_status)
if not build_to_status:
raise ScriptError('No try job information was collected.')
# The manifest may be used below to do check which tests are reference tests.
WPTManifest.ensure_manifest(self.host)
# Here we build up a dict of failing test results for all platforms.
test_expectations = {}
for build, job_status in build_to_status.iteritems():
if job_status.result == 'SUCCESS':
self.ports_with_all_pass.add(self.port_name(build))
port_results = self.get_failing_results_dict(build)
test_expectations = self.merge_dicts(test_expectations,
port_results)
# And then we merge results for different platforms that had the same results.
for test_name, platform_result in test_expectations.iteritems():
# platform_result is a dict mapping platforms to results.
test_expectations[test_name] = self.merge_same_valued_keys(
platform_result)
# At this point, test_expectations looks like: {
# 'test-with-failing-result': {
# ('port-name1', 'port-name2'): SimpleTestResult,
# 'port-name3': SimpleTestResult
# }
# }
rebaselined_tests, test_expectations = self.download_text_baselines(
test_expectations)
test_expectation_lines = self.create_line_dict(test_expectations)
self.write_to_test_expectations(test_expectation_lines)
return rebaselined_tests, test_expectation_lines
def test_message_with_output(self):
error = ScriptError('My custom message!', '', -1)
self.assertEqual(error.message_with_output(), 'My custom message!')
error = ScriptError('My custom message!', '', -1, 'My output.')
self.assertEqual(error.message_with_output(),
'My custom message!\n\noutput: My output.')
error = ScriptError('', 'my_command!', -1, 'My output.',
'/Users/username/blah')
self.assertEqual(
error.message_with_output(),
'Failed to run "\'my_command!\'" exit_code: -1 cwd: /Users/username/blah\n\noutput: My output.'
)
error = ScriptError('', 'my_command!', -1, 'ab' + '1' * 499)
self.assertEqual(
error.message_with_output(),
'Failed to run "\'my_command!\'" exit_code: -1\n\noutput: Last 500 characters of output:\nb'
+ '1' * 499)
def process_cmds(cmd_args):
if 'token' in cmd_args:
return '00000'
elif (('weblayer_shell_wpt on '
'Ubuntu-16.04 or Ubuntu-18.04') in cmd_args):
return json.dumps(actual_mp)
elif (('chrome_public_wpt on '
'Ubuntu-16.04 or Ubuntu-18.04') in cmd_args):
raise ScriptError('Test Error')
elif 'chrome_public_wpt' in cmd_args:
return json.dumps(baseline_mp)
else:
return '{"number": 400, "id":"abcd"}'
def _merge_base(self, git_commit=None):
if git_commit:
# Rewrite UPSTREAM to the upstream branch
if 'UPSTREAM' in git_commit:
upstream = self._upstream_branch()
if not upstream:
raise ScriptError(
message='No upstream/tracking branch set.')
git_commit = git_commit.replace('UPSTREAM', upstream)
# Special-case <refname>.. to include working copy changes, e.g., 'HEAD....' shows only the diffs from HEAD.
if git_commit.endswith('....'):
return git_commit[:-4]
if '..' not in git_commit:
git_commit = git_commit + '^..' + git_commit
return git_commit
return self._remote_merge_base()
def run_command(_):
raise ScriptError('Unable to infer commit position from footers rutabaga')
def _get_marker_line_number(self, path):
for line in self._test_expectations.get_updated_lines(path):
if line.to_string() == self.MARKER_COMMENT:
return line.lineno
raise ScriptError('Marker comment does not exist in %s' % path)
def update_expectations(self):
"""Downloads text new baselines and adds test expectations lines.
Returns:
A pair: A set of tests that are rebaselined, and a dictionary
mapping tests that couldn't be rebaselined to lists of expectation
lines written to TestExpectations.
"""
# The wpt_manifest function in Port is cached by default, but may be out
# of date if this code is called during test import. An out of date
# manifest will cause us to mistreat newly added tests, as they will not
# exist in the cached manifest. To avoid this, we invalidate the cache
# here. See https://crbug.com/1154650 .
self.port.wpt_manifest.cache_clear()
issue_number = self.get_issue_number()
if issue_number == 'None':
raise ScriptError('No issue on current branch.')
build_to_status = self.get_latest_try_jobs()
_log.debug('Latest try jobs: %r', build_to_status)
if not build_to_status:
raise ScriptError('No try job information was collected.')
# Here we build up a dict of failing test results for all platforms.
test_expectations = {}
for build, job_status in build_to_status.iteritems():
if (job_status.result == 'SUCCESS' and
not self.options.include_unexpected_pass):
continue
# Temporary logging for https://crbug.com/1154650
result_dicts = self.get_failing_results_dicts(build)
_log.info('Merging failing results dicts for %s', build)
for result_dict in result_dicts:
test_expectations = self.merge_dicts(
test_expectations, result_dict)
# At this point, test_expectations looks like: {
# 'test-with-failing-result': {
# config1: SimpleTestResult,
# config2: SimpleTestResult,
# config3: AnotherSimpleTestResult
# }
# }
# And then we merge results for different platforms that had the same results.
for test_name, platform_result in test_expectations.iteritems():
# platform_result is a dict mapping platforms to results.
test_expectations[test_name] = self.merge_same_valued_keys(
platform_result)
# At this point, test_expectations looks like: {
# 'test-with-failing-result': {
# (config1, config2): SimpleTestResult,
# (config3,): AnotherSimpleTestResult
# }
# }
rebaselined_tests, test_expectations = self.download_text_baselines(
test_expectations)
exp_lines_dict = self.write_to_test_expectations(test_expectations)
return rebaselined_tests, exp_lines_dict
def _get_marker_line_number(test_expectations, path, marker_comment):
for line in test_expectations.get_updated_lines(path):
if line.to_string() == marker_comment:
return line.lineno
raise ScriptError('Marker comment does not exist in %s' % path)
def mock_run_command(args):
port.host.filesystem.write_binary_file(args[4], mock_image_diff)
raise ScriptError(exit_code=1)
def run(self, args):
self.calls.append(['git', 'cl'] + args)
arg_key = "".join(args)
if self._git_error_output and arg_key in self._git_error_output.keys():
raise ScriptError(output=self._git_error_output[arg_key])
return 'mock output'
def _run_fn(args):
if args[0] == 'git' and args[1] == 'apply':
raise ScriptError('MOCK failed applying patch')
return ''