def get_reports_for_last_days_of_month(
self,
report_name: str,
date: datetime,
company: str = None) -> List[Tuple[datetime, pd.DataFrame]]:
path = self._report_base_path
if company is not None:
company = normalize_company(name=company)
path = f'{path}/{company}'
path = date.strftime(
f'{path}/{report_name}/{company.upper() + "_" if company else ""}{report_name}_%Y'
)
reports_set = []
for report in self.container_client.list_blobs(name_starts_with=path):
report_name = report['name']
match = re.search(pattern=r'\d{4}-\d{2}-\d{2}', string=report_name)
report_date = datetime.strptime(match.group(), '%Y-%m-%d').date()
date = datetime(year=report_date.year,
month=report_date.month,
day=monthrange(report_date.year,
report_date.month)[1])
if report_date.strftime('%Y-%m-%d') == date.strftime('%Y-%m-%d'):
reports_set.append(
(report_date,
self.read_pandas_dataframe_from_csv(path=report_name)))
return reports_set
def get_private_push_events_commits_file_path(self, date: datetime,
company: str):
path = self.BASE_PATH / self.BASE_AREA_DIR / normalize_company(
name=company).upper()
path = path / 'github' / 'events' / 'push' / date.strftime(
"%Y") / date.strftime("%m")
path.mkdir(parents=True, exist_ok=True)
return path / date.strftime('%Y-%m-%d.parquet')
def get_report_path(self,
report_name: str,
date: datetime,
company: str = None) -> str:
path = self._report_base_path
if company is not None:
company = normalize_company(name=company)
path = f'{path}/{company}'
return date.strftime(
f'{path}/{report_name}/{company.upper() + "_" if company else ""}{report_name}_%Y-%m-%d.csv'
)
def get_report_path(self,
report_name: str,
date: datetime,
company: str = None) -> Path:
path = self._report_base_path
if company is not None:
company = normalize_company(name=company)
path /= company
path = path / report_name
path.mkdir(parents=True, exist_ok=True)
filename = f'{company.upper() + "_" if company else ""}{report_name}_{date.strftime("%Y-%m-%d")}.csv'
return path / filename