def collect_cpp_tests(max_tests, cpp_test_program_filter,
cpp_test_program_re_str):
"""
Collect C++ test programs to run.
@param max_tests: maximum number of tests to run. Used in debugging.
@param cpp_test_program_filter: a collection of C++ test program names to be used as a filter
@param cpp_test_program_re_str: a regular expression string to be used as a filter for the set
of C++ test programs.
"""
global_conf = yb_dist_tests.global_conf
logging.info("Collecting the list of C++ test programs")
start_time_sec = time.time()
ctest_cmd_result = command_util.run_program([
'/bin/bash', '-c',
'cd "{}" && YB_LIST_CTEST_TESTS_ONLY=1 ctest -j8 --verbose'.format(
global_conf.build_root)
])
test_programs = []
for line in ctest_cmd_result.stdout.split("\n"):
re_match = CTEST_TEST_PROGRAM_RE.match(line)
if re_match:
rel_ctest_prog_path = os.path.relpath(re_match.group(1),
global_conf.build_root)
test_programs.append(rel_ctest_prog_path)
test_programs = sorted(set(test_programs))
elapsed_time_sec = time.time() - start_time_sec
logging.info("Collected %d test programs in %.2f sec" %
(len(test_programs), elapsed_time_sec))
if cpp_test_program_re_str:
cpp_test_program_re = re.compile(cpp_test_program_re_str)
test_programs = [
test_program for test_program in test_programs
if cpp_test_program_re.search(test_program)
]
logging.info(
"Filtered down to %d test programs using regular expression '%s'" %
(len(test_programs), cpp_test_program_re_str))
if cpp_test_program_filter:
cpp_test_program_filter = set(cpp_test_program_filter)
unfiltered_test_programs = test_programs
# test_program contains test paths relative to the root directory (including directory
# names), and cpp_test_program_filter contains basenames only.
test_programs = sorted(
set([
test_program for test_program in test_programs
if os.path.basename(test_program) in cpp_test_program_filter
]))
logging.info(
"Filtered down to %d test programs using the list from test conf file"
% len(test_programs))
if unfiltered_test_programs and not test_programs:
# This means we've filtered the list of C++ test programs down to an empty set.
logging.info((
"NO MATCHING C++ TEST PROGRAMS FOUND! Test programs from conf file: {}, "
"collected from ctest before filtering: {}").format(
set_to_comma_sep_str(cpp_test_program_filter),
set_to_comma_sep_str(unfiltered_test_programs)))
if max_tests and len(test_programs) > max_tests:
logging.info(
"Randomly selecting {} test programs out of {} possible".format(
max_tests, len(test_programs)))
random.shuffle(test_programs)
test_programs = test_programs[:max_tests]
if not test_programs:
logging.info("Found no test programs")
return []
fine_granularity_gtest_programs = []
one_shot_test_programs = []
for test_program in test_programs:
if is_one_shot_test(test_program):
one_shot_test_programs.append(test_program)
else:
fine_granularity_gtest_programs.append(test_program)
logging.info(
("Found {} gtest test programs where tests will be run separately, "
"{} test programs to be run on one shot").format(
len(fine_granularity_gtest_programs),
len(one_shot_test_programs)))
test_programs = fine_granularity_gtest_programs
logging.info(
"Collecting gtest tests for {} test programs where tests will be run separately"
.format(len(test_programs)))
start_time_sec = time.time()
all_test_programs = fine_granularity_gtest_programs + one_shot_test_programs
if len(all_test_programs) <= 5:
app_name_details = [
'test programs: [{}]'.format(', '.join(all_test_programs))
]
else:
app_name_details = ['{} test programs'.format(len(all_test_programs))]
init_spark_context(app_name_details)
set_global_conf_for_spark_jobs()
# Use fewer "slices" (tasks) than there are test programs, in hope to get some batching.
num_slices = (len(test_programs) + 1) / 2
all_test_descriptor_lists = run_spark_action(
lambda: spark_context.parallelize(test_programs, numSlices=num_slices).
map(parallel_list_test_descriptors).collect())
elapsed_time_sec = time.time() - start_time_sec
test_descriptor_strs = one_shot_test_programs + [
test_descriptor_str
for test_descriptor_str_list in all_test_descriptor_lists
for test_descriptor_str in test_descriptor_str_list
]
logging.info("Collected the list of %d gtest tests in %.2f sec" %
(len(test_descriptor_strs), elapsed_time_sec))
return [yb_dist_tests.TestDescriptor(s) for s in test_descriptor_strs]
def collect_cpp_tests(max_tests, cpp_test_program_filter, cpp_test_program_re_str):
"""
Collect C++ test programs to run.
@param max_tests: maximum number of tests to run. Used in debugging.
@param cpp_test_program_filter: a collection of C++ test program names to be used as a filter
@param cpp_test_program_re_str: a regular expression string to be used as a filter for the set
of C++ test programs.
"""
global_conf = yb_dist_tests.global_conf
logging.info("Collecting the list of C++ test programs")
start_time_sec = time.time()
ctest_cmd_result = command_util.run_program(
['/bin/bash',
'-c',
'cd "{}" && YB_LIST_CTEST_TESTS_ONLY=1 ctest -j8 --verbose'.format(
global_conf.build_root)])
test_programs = []
for line in ctest_cmd_result.stdout.split("\n"):
re_match = CTEST_TEST_PROGRAM_RE.match(line)
if re_match:
rel_ctest_prog_path = os.path.relpath(re_match.group(1), global_conf.build_root)
test_programs.append(rel_ctest_prog_path)
test_programs = sorted(set(test_programs))
elapsed_time_sec = time.time() - start_time_sec
logging.info("Collected %d test programs in %.2f sec" % (
len(test_programs), elapsed_time_sec))
if cpp_test_program_re_str:
cpp_test_program_re = re.compile(cpp_test_program_re_str)
test_programs = [test_program for test_program in test_programs
if cpp_test_program_re.search(test_program)]
logging.info("Filtered down to %d test programs using regular expression '%s'" %
(len(test_programs), cpp_test_program_re_str))
if cpp_test_program_filter:
cpp_test_program_filter = set(cpp_test_program_filter)
unfiltered_test_programs = test_programs
# test_program contains test paths relative to the root directory (including directory
# names), and cpp_test_program_filter contains basenames only.
test_programs = sorted(set([
test_program for test_program in test_programs
if os.path.basename(test_program) in cpp_test_program_filter
]))
logging.info("Filtered down to %d test programs using the list from test conf file" %
len(test_programs))
if unfiltered_test_programs and not test_programs:
# This means we've filtered the list of C++ test programs down to an empty set.
logging.info(
("NO MATCHING C++ TEST PROGRAMS FOUND! Test programs from conf file: {}, "
"collected from ctest before filtering: {}").format(
set_to_comma_sep_str(cpp_test_program_filter),
set_to_comma_sep_str(unfiltered_test_programs)))
if max_tests and len(test_programs) > max_tests:
logging.info("Randomly selecting {} test programs out of {} possible".format(
max_tests, len(test_programs)))
random.shuffle(test_programs)
test_programs = test_programs[:max_tests]
if not test_programs:
logging.info("Found no test programs")
return []
fine_granularity_gtest_programs = []
one_shot_test_programs = []
for test_program in test_programs:
if is_one_shot_test(test_program):
one_shot_test_programs.append(test_program)
else:
fine_granularity_gtest_programs.append(test_program)
logging.info(("Found {} gtest test programs where tests will be run separately, "
"{} test programs to be run on one shot").format(
len(fine_granularity_gtest_programs),
len(one_shot_test_programs)))
test_programs = fine_granularity_gtest_programs
logging.info(
"Collecting gtest tests for {} test programs where tests will be run separately".format(
len(test_programs)))
start_time_sec = time.time()
all_test_programs = fine_granularity_gtest_programs + one_shot_test_programs
if len(all_test_programs) <= 5:
app_name_details = ['test programs: [{}]'.format(', '.join(all_test_programs))]
else:
app_name_details = ['{} test programs'.format(len(all_test_programs))]
init_spark_context(app_name_details)
set_global_conf_for_spark_jobs()
# Use fewer "slices" (tasks) than there are test programs, in hope to get some batching.
num_slices = (len(test_programs) + 1) / 2
all_test_descriptor_lists = run_spark_action(
lambda: spark_context.parallelize(
test_programs, numSlices=num_slices).map(parallel_list_test_descriptors).collect()
)
elapsed_time_sec = time.time() - start_time_sec
test_descriptor_strs = one_shot_test_programs + [
test_descriptor_str
for test_descriptor_str_list in all_test_descriptor_lists
for test_descriptor_str in test_descriptor_str_list]
logging.info("Collected the list of %d gtest tests in %.2f sec" % (
len(test_descriptor_strs), elapsed_time_sec))
return [yb_dist_tests.TestDescriptor(s) for s in test_descriptor_strs]