def _initialize_repository(session,
src_path: str,
initialize_commands: str = None,
test_commands: str = None,
src_directory: str = None,
commit_only=False,
ignored_files=''):
'''
Given a src_path to a repository, create the repository in the database
'''
logger.info('Initializing repository at "{}"'.format(src_path))
url = get_repository_url_from_path(src_path)
repository = get(session, Repository, url=url)
name = get_repository_name_from_url(url)
logger.info('Repository name is "{}" with url "{}"'.format(name, url))
if not initialize_commands:
msg = ('Input the commands to intialize the repo (ex. '
'"source env/bin/activate"): ')
initialize_commands = input(msg)
if not test_commands:
msg = ('Input the command to run the tests for the repo: ')
test_commands = input(msg)
if not src_directory:
msg = ('Input the source directory for your project: ')
src_directory = input(msg)
# first check to see if the repository already exists
repository = get(session, Repository, url=url)
if not repository:
repository = create(session,
Repository,
name=name,
url=url,
initialize_commands=initialize_commands,
test_commands=test_commands,
src_directory=src_directory,
src_path=src_path,
ignored_files=ignored_files)
_check_repo_is_clean(repository, path=repository.original_path)
# create the mirror repository that BugBuddy primarily works on
sync_mirror_repo(repository)
# make sure the mirrored repo is on bug_buddy branch
set_bug_buddy_branch(repository)
# Initialize the repository by recording functions and creating synthetic
# diffs
if not commit_only:
snapshot(repository, allow_empty=True)
session.commit()
logger.info(
'Your repository "{}" has been successfully initialized!'.format(
repository))
return repository
def delete_command(src_path: str):
'''
Entry-point for the "bugbuddy generate" command
@param src_path: path to the repository
'''
url = get_repository_url_from_path(src_path)
with session_manager() as session:
repository = get(session, Repository, url=url)
# make sure you cannot delete the bug_buddy branch
if repository.name not in ['bug_buddy', 'BugBuddy']:
msg = ('Would you like to delete the bug_buddy branch for {}?\n'
'(y/n)\n'.format(repository))
should_delete = input(msg)
if is_affirmative(should_delete):
logger.info('Deleting bug_buddy branch')
delete_bug_buddy_branch(repository or Mock(src_path=src_path))
if repository:
logger.info('Deleting data from the database')
delete(session, repository)
else:
logger.info('No matching repo found in the database')
def get_matching_commit_for_diffs(repository, base_synthetic_diffs):
'''
Given a set of diffs, return if there is a commit that has those diffs
'''
session = Session.object_session(repository)
base_synthetic_ids = [diff.id for diff in base_synthetic_diffs]
diff_hash = get_hash_given_base_synthetic_ids(base_synthetic_ids)
return get(session, Commit, synthetic_diff_hash=diff_hash)
def _get_repository_from_src_path(session, src_path: str):
'''
Returns the repository given a src_path
'''
url = get_repository_url_from_path(src_path)
repository = get(session, Repository, url=url)
if not repository:
msg = ('This repository is not in the BudBuddy database, would you '
'like to initialize the repository? (y/n)\n'.format(src_path))
should_initialize = input(msg)
if is_affirmative(should_initialize):
repository = _initialize_repository(session, src_path)
else:
logger.info('No worries!')
if src_path != repository.original_path:
_migrate_repository_to_new_path(session, repository, src_path)
return repository
def on_any_event(self, event):
'''
Catches all events
'''
if '/.' in event.src_path:
return
updated_file = os.path.relpath(event.src_path,
self.repository.original_path)
if (not updated_file or updated_file in self.repository.ignored_files
or not updated_file.endswith('.py')):
return
# we have to recreate the repository in this thread for Sqlite
with session_manager() as session:
repository = get(session, Repository, id=self.repository.id)
# logger.info('Syncing updates')
# Copy the change over to the mirror repository
sync_mirror_repo(repository)
if not is_repo_clean(self.repository):
logger.info('Valid change event: {}'.format(event))
# make sure the repository is on the bug_buddy branch
start = time.time()
commit = snapshot(repository, commit_only=self.commit_only)
total_time = time.time() - start
logger.info(
'Completed snapshot of {commit} in {m}m {s}s'.format(
commit=commit, m=total_time / 60, s=total_time % 60))
session.commit()
if not self.commit_only:
run_all_tests(commit)
for test_failure in commit.failed_test_results:
predict_blame(test_failure)
# display the results in the cli output
# self.score_card.render(commit)
commit.summary()
else:
logger.info('Nothing was changed')
def commit_generator(repository_id: int, batch_size: int):
'''
Returns augmented commit data for training
'''
with session_manager() as session:
repository = get(session, Repository, id=repository_id)
while True:
features = []
labels = []
commits = get_commits(repository,
num_commits=batch_size,
synthetic=True)
for commit in commits:
feature = commit_to_state(commit)
set_functions_altered_noise(feature)
set_tests_not_run_noise(feature)
label = commit_to_test_failure_label(commit)
features.append(feature)
labels.append(label)
yield numpy.stack(features), numpy.stack(labels)
def get_function_from_patch(repository: Repository, patch: str, file_path: str,
first_line: int, last_line: int):
'''
Given a patch, find the corresponding function if possible
'''
session = Session.object_session(repository)
pattern = re.compile('def \w*\(')
matching_functions = pattern.findall(patch)
if len(matching_functions) != 1:
# import pdb; pdb.set_trace()
print('multiple matching_functions: ', matching_functions)
else:
function_name = matching_functions[0]
# it comes out from the regex as def xxxxxx( so we need to splice out
# just the function name
function_name = function_name[4:-1]
return get(
session,
Function,
name=function_name,
file_path=file_path,
)