def upload_data_to_kaggle(self):
files = []
for output_folder in self.config.output_folder:
t.log_message('Uploading an output folder to the Kaggle: ' +
output_folder)
for item in os.listdir(output_folder):
path = os.path.join(output_folder, item)
if os.path.isfile(path):
token = self.upload_file_to_kaggle(path)
files.append({'token': token})
elif os.path.isdir(path) and self.kaggle_dirmode in [
'zip', 'tar'
]:
temp_dir = tempfile.mkdtemp()
try:
_, dir_name = os.path.split(path)
archive_path = shutil.make_archive(
os.path.join(temp_dir, dir_name),
self.kaggle_dirmode, path)
token = self.upload_file_to_kaggle(archive_path)
files.append({'token': token})
finally:
shutil.rmtree(temp_dir)
t.log_message(output_folder + ' - uploaded.')
dataset = self.prepare_dataset(files)
self.kaggle_api_call(resource='/datasets/create/new',
method='POST',
body=dataset)
def upload_file_to_kaggle(self, file_path: str):
file_token = None
try:
file_name = os.path.basename(file_path)
content_length = os.path.getsize(file_path)
last_modified_date_utc = int(os.path.getmtime(file_path))
post_params = [('fileName', file_name)]
kaggle_response = self.kaggle_api_call(
resource='/datasets/upload/file/' + str(content_length) + '/' +
str(last_modified_date_utc),
method='POST',
post_params=post_params)
kaggle_data = json.loads(kaggle_response.data.decode('utf8'))
create_url = kaggle_data['createUrl']
with io.open(file_path, 'rb', buffering=0) as fp:
reader = io.BufferedReader(fp)
session = requests.Session()
retries = Retry(total=10, backoff_factor=0.5)
adapter = HTTPAdapter(max_retries=retries)
session.mount('http://', adapter)
session.mount('https://', adapter)
response = session.put(create_url, data=reader)
if response.status_code == 200 or response.status_code == 201:
file_token = kaggle_data['token']
if file_token is None:
t.log_message('Upload unsuccessful: ' + file_path)
except Exception as error:
t.log_message('Upload filed: ' + file_path + '\n' + str(error))
return file_token
def start_auth_telegram(self, client_config):
if self.telegram_bot is None:
t.log_message('telegram bot is None. Telegram auth canceled.')
return
auth = GoogleAuth()
auth.LoadClientConfigFile(client_config_file=client_config)
if auth.flow is None:
auth.GetFlow()
auth.flow.redirect_uri = OOB_CALLBACK_URN
self.telegram_bot.send_message(
'Please go to the following link in your browser and send me a Google verification code. \nAuth url: '
+ auth.GetAuthUrl())
dirty = False
code = None
save_credentials = auth.settings.get('save_credentials')
if auth.credentials is None and save_credentials:
auth.LoadCredentials()
if auth.credentials is None:
code = self.telegram_bot.get_code()
dirty = True
else:
if auth.access_token_expired:
if auth.credentials.refresh_token is not None:
auth.Refresh()
else:
code = self.telegram_bot.get_code()
dirty = True
if code is not None:
auth.Auth(code)
if dirty and save_credentials:
auth.SaveCredentials()
return auth
def run(self, run_file_list: list):
for item in run_file_list:
if not os.path.exists(item):
t.log_message('ERROR. File not found: ' + item)
continue
t.log_message("Executing file: " + item)
os.system('python ' + item)
def send_output_to_gdrive(self, output_folders: list, drive_folders: list):
if self.gauth is None and self.config.gdrive_folders is not None:
t.log_message('GoogleDrive is unauthorised. Upload canceled.')
return
drive = GoogleDrive(self.gauth)
t.log_message('Uploading an output folders to the Google Drive')
for drive_folder in drive_folders:
for folder in output_folders:
self.upload_to_drive(folder, drive_folder, drive)
def download_list(self, url_list: list):
for item in url_list:
t.log_message('Downloading: ' + item)
try:
download = Download(item, retries=5)
download.download()
path = os.path.abspath(download.download_path)
_, extension = os.path.splitext(path)
if extension[1:] in dict(shutil.get_archive_formats()).keys(
) and self.config.extract_archives:
shutil.unpack_archive(path)
except Exception as e:
t.log_message("ERROR. Download: " + item + ' FAILED.\n' +
str(e))
def build_workspace(self):
if self.config is None:
return
self.gauth = self.get_gauth()
if self.config.dataset_list is not None:
self.download_list(self.config.dataset_list)
if self.config.repos is not None:
self.clone_repos(self.config.repos)
if self.config.script_files is not None:
self.run(self.config.script_files)
if self.config.gdrive_folders is not None:
self.send_output_to_gdrive(self.config.output_folder,
self.config.gdrive_folders)
if self.config.kaggle is not None:
self.upload_data_to_kaggle()
if 'dirmode' in self.config.kaggle:
self.kaggle_dirmode = self.config.kaggle['dirmode']
t.log_message('Done.')
if self.telegram_bot is not None:
self.telegram_bot.send_message('Workspace build done.')
def get_gauth(self):
gauth = None
packge_path, _ = os.path.split(__file__)
client_config = os.path.join(packge_path, 'client_secrets.json')
credentials_file = os.path.join(packge_path, 'drive_credentials')
if os.path.exists(credentials_file):
try:
gauth = GoogleAuth()
gauth.LoadClientConfigFile(client_config_file=client_config)
gauth.LoadCredentialsFile(credentials_file=credentials_file)
return gauth
except Exception as e:
t.log_message(str(e))
gauth = None
if self.config.gdrive_folders is not None and self.config.telegram_channels is not None and self.telegram_bot is not None:
try:
gauth = self.start_auth_telegram(client_config=client_config)
gauth.SaveCredentialsFile(credentials_file=credentials_file)
except Exception as e:
t.log_message(str(e))
gauth = None
elif self.config.gdrive_folders is not None and self.telegram_bot is None or self.config.telegram_channels is None:
try:
gauth = GoogleAuth()
gauth.LoadClientConfigFile(client_config_file=client_config)
gauth.CommandLineAuth()
gauth.SaveCredentialsFile(credentials_file=credentials_file)
except Exception as e:
t.log_message(str(e))
gauth = None
return gauth
def clone_repos(self, repos: dict):
for repo_name, repo_data in repos.items():
branch: str = None
if 'branch' in repo_data:
branch = repo_data['branch']
if 'url' in repo_data:
url: str = repo_data['url']
if os.path.exists(repo_name):
shutil.rmtree(repo_name)
if branch is not None:
t.log_message('Cloning repo: ' + url + ', branch: ' +
branch + ', to the folder: ' + repo_name)
Repo.clone_from(url=url, to_path=repo_name, branch=branch)
else:
t.log_message('Cloning repo: ' + url +
', to the folder: ' + repo_name)
Repo.clone_from(url=url, to_path=repo_name)
else:
t.log_message('ERROR. URL not found for a repo: ' + repo_name)
