def _structure_artifact_data(self, image_tag, bp, jp):
log.info('Extracting metadata for jobpair', image_tag + '.')
reproduced = Packager._is_jobpair_reproduced(jp)
repo = bp['repo']
reproduce_successes, reproduce_attempts, stability = Packager._calc_stability(jp)
failed_job = jp['failed_job']
passed_job = jp['passed_job']
builds = [bp['failed_build'], bp['passed_build']]
jobs = [failed_job, passed_job]
today = str(date.today())
status = 'Unreproducible'
if reproduce_successes == 5:
status = 'Reproducible'
elif 0 < reproduce_successes < 5:
status = 'Flaky'
current_status = {
'time_stamp': today,
'status': status
}
d = {
'tag': image_tag,
'image_tag': image_tag,
'repo': repo,
'repo_mined_version': bp['repo_mined_version'],
'pr_num': int(bp['pr_num']),
'branch': bp['branch'],
'base_branch': bp['base_branch'],
'lang': [j['language'] for j in bp['failed_build']['jobs']
if j['job_id'] == failed_job['job_id']][0].title(),
# Assume the build system and test framework is the same for both jobs. In some rare cases, this assumption
# will not hold.
'build_system': failed_job['orig_result']['tr_build_system'] if failed_job['orig_result'] else '',
'test_framework': failed_job['orig_result']['tr_log_frameworks'] if failed_job['orig_result'] else '',
'merged_at': bp['merged_at'],
'is_error_pass': bp['is_error_pass'],
'reproduced': reproduced,
'match': Packager._calc_match_over_run(jp) if reproduced else '',
'reproduce_successes': reproduce_successes,
'reproduce_attempts': reproduce_attempts,
'stability': stability,
'creation_time': int(time.time()),
# Evaluation info.
'filtered_reason': jp.get('filtered_reason', ''),
# Metrics. Empty by default and will be populated later by other components during post-processing.
'metrics': {},
'current_status': current_status
}
for i in range(2):
job_key = 'failed_job' if i == 0 else 'passed_job'
patches = {}
job = jobs[i]
# Find index of job in builds[i]['jobs'].
job_id = job['job_id']
jobs_index = next(i for i, j in enumerate(builds[i]['jobs']) if j['job_id'] == job_id)
# Add patch information if the job is Java 7 and has at least one reproduce success.
job_config = builds[i]['jobs'][jobs_index]['config']
if job_config.get('jdk') in ['oraclejdk7', 'openjdk7']:
# TODO: Collect patch names as each patch is applied. That is, do not wait until this method because the
# patches created now may not exactly match the patches applied.
patches['mvn-tls'] = today
if job.get('pip_patch'):
patches['pip-yaml-patch'] = today
patches['remove-ppa'] = today
d[job_key] = {
'base_sha': builds[i]['base_sha'],
'build_id': builds[i]['build_id'],
'build_job': [j['build_job'] for j in builds[i]['jobs'] if j['job_id'] == jobs[i]['job_id']][0],
'committed_at': builds[i]['committed_at'],
'failed_tests': jobs[i]['orig_result']['tr_log_tests_failed'] if jobs[i]['orig_result'] else '',
'job_id': job_id,
'message': builds[i]['message'],
'mismatch_attrs': jobs[i]['mismatch_attrs'],
'num_tests_failed': jobs[i]['orig_result']['tr_log_num_tests_failed'] if jobs[i]['orig_result'] else '',
'num_tests_run': jobs[i]['orig_result']['tr_log_num_tests_run'] if jobs[i]['orig_result'] else '',
'trigger_sha': builds[i]['travis_merge_sha'] if builds[i]['travis_merge_sha'] else
builds[i]['head_sha'],
'is_git_repo': Packager._artifact_is_git_repo(builds[i]),
'config': job_config,
'patches': patches,
'component_versions': {
'analyzer': Utils.get_analyzer_version(),
'reproducer': Utils.get_reproducer_version(),
},
}
return d
def _structure_artifact_data(self, image_tag, bp, jp):
log.info('Extracting metadata for jobpair', image_tag + '.')
reproduced = Packager._is_jobpair_reproduced(jp)
repo = bp['repo']
reproduce_successes, reproduce_attempts, stability = Packager._calc_stability(jp)
failed_job = jp['failed_job']
passed_job = jp['passed_job']
builds = [bp['failed_build'], bp['passed_build']]
jobs = [failed_job, passed_job]
# In the case where "No files are changed", determined by our pair-classifier.py, there would not be a
# classification key for that jobpair. We insert a blank template in case the artifact is populated to our
# website
try:
classification = jp['classification']
classification = {
'test': classification['test'],
'build': classification['build'],
'code': classification['code'],
'exceptions': classification['exceptions']
}
except KeyError:
classification = {
'test': 'NA',
'build': 'NA',
'code': 'NA',
'exceptions': []
}
today = str(date.today())
status = 'Unreproducible'
if reproduce_successes == 5:
status = 'Reproducible'
elif 0 < reproduce_successes < 5:
status = 'Flaky'
current_status = {
'time_stamp': today,
'status': status
}
# Previously mined jobpairs that are ran through the Reproducer, for example from generated pairs,
# may not contain any 'metrics' so we add add a blank template for updating
try:
metrics = jp['metrics']
except KeyError:
metrics = {}
d = {
'current_image_tag': image_tag,
'image_tag': image_tag,
'repo': repo,
'repo_mined_version': bp['repo_mined_version'],
'pr_num': int(bp['pr_num']),
'branch': bp['branch'],
'base_branch': bp['base_branch'],
'lang': [j['language'] for j in bp['failed_build']['jobs']
if j['job_id'] == failed_job['job_id']][0].title(),
# Assume the build system and test framework is the same for both jobs. In some rare cases, this assumption
# will not hold.
'build_system': failed_job['orig_result']['tr_build_system'] if failed_job['orig_result'] else '',
'test_framework': failed_job['orig_result']['tr_log_frameworks'] if failed_job['orig_result'] else '',
'merged_at': bp['merged_at'],
'is_error_pass': bp['is_error_pass'],
'reproduced': reproduced,
'match': Packager._calc_match_over_run(jp) if reproduced else '',
'reproduce_successes': reproduce_successes,
'reproduce_attempts': reproduce_attempts,
'stability': stability,
'creation_time': int(time.time()),
# Evaluation info.
'filtered_reason': jp.get('filtered_reason', ''),
'metrics': metrics,
'current_status': current_status,
'classification': classification
}
for i in range(2):
job_key = 'failed_job' if i == 0 else 'passed_job'
patches = {}
job = jobs[i]
# Find index of job in builds[i]['jobs'].
job_id = job['job_id']
jobs_index = next(i for i, j in enumerate(builds[i]['jobs']) if j['job_id'] == job_id)
# Add patch information if the job is Java 7 and has at least one reproduce success.
job_config = builds[i]['jobs'][jobs_index]['config']
if job_config.get('jdk') in ['oraclejdk7', 'openjdk7']:
# TODO: Collect patch names as each patch is applied. That is, do not wait until this method because the
# patches created now may not exactly match the patches applied.
patches['mvn-tls'] = today
if job.get('pip_patch'):
patches['pip-yaml-patch'] = today
patches['remove-ppa'] = today
d[job_key] = {
'base_sha': builds[i]['base_sha'],
'build_id': builds[i]['build_id'],
'build_job': [j['build_job'] for j in builds[i]['jobs'] if j['job_id'] == jobs[i]['job_id']][0],
'committed_at': builds[i]['committed_at'],
'failed_tests': jobs[i]['orig_result']['tr_log_tests_failed'] if jobs[i]['orig_result'] else '',
'job_id': job_id,
'message': builds[i]['message'],
'mismatch_attrs': jobs[i]['mismatch_attrs'],
'num_tests_failed': jobs[i]['orig_result']['tr_log_num_tests_failed'] if jobs[i]['orig_result'] else '',
'num_tests_run': jobs[i]['orig_result']['tr_log_num_tests_run'] if jobs[i]['orig_result'] else '',
'trigger_sha': builds[i]['travis_merge_sha'] if builds[i]['travis_merge_sha'] else
builds[i]['head_sha'],
'is_git_repo': Packager._artifact_is_git_repo(builds[i]),
'config': job_config,
'patches': patches,
'component_versions': {
'analyzer': Utils.get_analyzer_version(),
'reproducer': Utils.get_reproducer_version(),
},
}
return d
