def main():
parser = argparse.ArgumentParser()
parser.add_argument(
'directory', help='Path to the repository. Example usage: run.sh path/to/directory')
parser.add_argument('--output', default='./repo_data.json', dest='output',
help='Path to the JSON file that will contain the result')
parser.add_argument('--skip_obfuscation', default=False, dest='skip_obfuscation',
help='If true it won\'t obfuscate the sensitive data such as emails and file names. Mostly for testing purpuse')
parser.add_argument('--parse_libraries', default=False,
dest='parse_libraries', help='If true, used libraries will be parsed')
args = parser.parse_args()
repo = git.Repo(args.directory)
ar = AnalyzeRepo(repo)
q = Questions()
print('Initialization...')
for branch in repo.branches:
ar.create_commits_entity_from_branch(branch.name)
ar.flag_duplicated_commits()
ar.get_commit_stats()
r = ar.create_repo_entity(args.directory)
# Ask the user if we cannot find remote URL
if r.primary_remote_url == '':
answer = q.ask_primary_remote_url(r)
authors = [(c['name'], c['email']) for _, c in r.contributors.items()]
identities_err = None
identities = q.ask_user_identity(authors, identities_err)
MAX_LIMIT = 50
while len(identities['user_identity']) == 0 or len(identities['user_identity']) > MAX_LIMIT:
if len(identities['user_identity']) == 0:
identities_err = 'Please select at least one author'
if len(identities['user_identity']) > MAX_LIMIT:
identities_err = 'You cannot select more than', MAX_LIMIT
identities = q.ask_user_identity(authors, identities_err)
r.local_usernames = identities['user_identity']
if args.parse_libraries:
# build authors from the selection
al = AnalyzeLibraries(r.commits, authors, repo.working_tree_dir)
libs = al.get_libraries()
# combine repo stats with libs used
for i in range(len(r.commits)):
c = r.commits[i]
if c.hash in libs.keys():
r.commits[i].libraries = libs[c.hash]
if not args.skip_obfuscation:
r = obfuscate(r)
er = ExportResult(r)
er.export_to_json(args.output)
def main():
parser = argparse.ArgumentParser()
parser.add_argument(
'directory',
help='Path to the repository. Example usage: run.sh path/to/directory')
parser.add_argument(
'--output',
default='./repo_data.json',
dest='output',
help='Path to the JSON file that will contain the result')
parser.add_argument(
'--skip_obfuscation',
default=False,
dest='skip_obfuscation',
help=
'If true it won\'t obfuscate the sensitive data such as emails and file names. Mostly for testing purpuse'
)
args = parser.parse_args()
repo = git.Repo(args.directory)
ar = AnalyzeRepo(repo, args.skip_obfuscation)
q = Questions()
print('Initialization...')
for branch in repo.branches:
ar.create_commits_entity_from_branch(branch.name)
ar.flag_duplicated_commits()
ar.get_commit_stats()
r = ar.create_repo_entity(args.directory)
# Ask the user if we cannot find remote URL
if r.primary_remote_url == '':
answer = q.ask_primary_remote_url(r)
identities = q.ask_user_identity(r)
MAX_LIMIT = 50
while len(identities['user_identity']) == 0 or len(
identities['user_identity']) > MAX_LIMIT:
if len(identities['user_identity']) == 0:
print('Please select at least one.')
if len(identities['user_identity']) > MAX_LIMIT:
print('You cannot select more than', MAX_LIMIT)
identities = q.ask_user_identity(r)
r.local_usernames = identities['user_identity']
er = ExportResult(r)
er.export_to_json(args.output)
def initialize(directory, skip_obfuscation, output, parse_libraries, email,
skip_upload, debug_mode, skip, commit_size_limit,
file_size_limit):
# Initialize logger
logger = logging.getLogger("main")
if debug_mode:
logger.setLevel(logging.DEBUG)
fh = logging.FileHandler('extractor_debug_info.log')
fh.setLevel(logging.DEBUG)
formatter = logging.Formatter(
'%(asctime)s - %(name)s - %(levelname)s - %(message)s')
fh.setFormatter(formatter)
logger.addHandler(fh)
else:
logger.setLevel(logging.WARNING)
logger.debug("Initialized main logger.")
repo = git.Repo(directory)
ar = AnalyzeRepo(repo)
q = Questions()
print('Analyzing repo under %s ...' % (directory))
try:
# Stop parsing if there are no branches
if not repo.branches:
print('No branches detected, will ignore this repo')
return
for branch in repo.branches:
ar.create_commits_entity_from_branch(branch.name)
ar.flag_duplicated_commits()
ar.get_commit_stats()
r = ar.create_repo_entity(directory)
# Stop parsing if there are no remotes
if not r.original_remotes:
print('No remotes detected, will ignore this repo')
return
# Ask the user if we cannot find remote URL
if r.primary_remote_url == '':
answer = q.ask_primary_remote_url(r)
if not r.contributors.items():
print('No authors detected, will ignore this repo')
return
authors = [(c['name'], c['email']) for _, c in r.contributors.items()]
identities = {}
identities['user_identity'] = []
# Stop parsing if there are no authors
if len(authors) == 0:
print('No authors detected, will ignore this repo')
return
identities_err = None
identities = q.ask_user_identity(authors, identities_err, email)
MAX_LIMIT = 50
while len(identities['user_identity']) == 0 or len(
identities['user_identity']) > MAX_LIMIT:
if len(identities['user_identity']) == 0:
identities_err = 'Please select at least one author'
if len(identities['user_identity']) > MAX_LIMIT:
identities_err = 'You cannot select more than', MAX_LIMIT
identities = q.ask_user_identity(authors, identities_err)
r.local_usernames = identities['user_identity']
if parse_libraries:
# build authors from the selection
# extract email from name -> email list
author_emails = [i.split(' -> ', 1)[1] for i in r.local_usernames]
if author_emails:
al = AnalyzeLibraries(r.commits, author_emails,
repo.working_tree_dir, skip,
commit_size_limit, file_size_limit)
libs = al.get_libraries()
# combine repo stats with libs used
for i in range(len(r.commits)):
c = r.commits[i]
if c.hash in libs.keys():
r.commits[i].libraries = libs[c.hash]
if not skip_obfuscation:
r = obfuscate(r)
er = ExportResult(r)
er.export_to_json_interactive(output, skip_upload)
except KeyboardInterrupt:
print("Cancelled by user")
return