def get_form_headers(
self,
form_type: Optional[Union[str, List[str]]] = None,
cik: Optional[peconf.CIK_TYPE] = None,
start_datetime: Optional[str] = None,
end_datetime: Optional[str] = None,
date_mode: Optional[str] = None,
output_type: Optional[str] = "dataframes",
) -> Union[List[peconf.SERVER_RESPONSE_TYPE], pd.DataFrame]:
"""
Get form headers metadata with the following parameters.
:param form_type: Form type or list of form types.
Example: form_type=['13F-HR', '4']
:param cik: Central Index Key as integer. It could be a list of P1_CIK
or just one identifier. None means all CIKs.
:param start_datetime: Get data where filing date is >= start_date. Date
format is "YYYY-MM-DDTHH-MI-SS". None means the entire available date range.
:param end_datetime: Get data where filing date is <= end_date. Date format
is "YYYY-MM-DDTHH-MI-SS". None means the entire available date range.
:param date_mode: Define whether dates are
interpreted as publication dates or knowledge dates
:param output_type: Output format: 'dict' or 'dataframes'.
"""
peutil.check_date_mode(start_datetime, end_datetime, date_mode)
peutil.check_form_type(form_type, self.form_types)
cik = peutil.check_sorted_unique_param("cik", cik)
params: Dict[str, Any] = {}
params = self._set_optional_params(
params,
form_type=form_type,
start_datetime=start_datetime,
end_datetime=end_datetime,
date_mode=date_mode,
cik=cik,
)
url = f'{self.base_url}{self._api_routes["HEADERS"]}'
result: Union[List[peconf.SERVER_RESPONSE_TYPE], pd.DataFrame] = []
for data in self._payload_form_cik_cusip_generator(
method="GET", url=url, headers=self.headers, params=params
):
result += data
if output_type == "dataframes":
try:
result = pd.DataFrame(result)
except (KeyError, json.JSONDecodeError) as e:
raise pexcep.ParseResponseException(
"Can't transform server response to a Pandas Dataframe"
) from e
else:
phdbg.dfatal(f"Output type {output_type} is not valid.")
return result
def get_form8_payload(
self,
cik: Optional[peconf.CIK_TYPE] = None,
start_datetime: Optional[str] = None,
end_datetime: Optional[str] = None,
date_mode: Optional[str] = None,
item: Optional[str] = None,
) -> pd.DataFrame:
"""
Get payload data for a form 8 and a company.
:param cik: Central Index Key as integer. It could be a list of P1_CIK
or just one identifier. None means all CIKs.
:param start_datetime: Get data where filing date is >= start_date. Date
format is "YYYY-MM-DDTHH-MI-SS". None means the entire available date range.
:param end_datetime: Get data where filing date is <= end_date. Date format
is "YYYY-MM-DDTHH-MI-SS". None means the entire available date range.
:param date_mode: Define whether dates are
interpreted as publication dates or knowledge dates
:param item: Item to retrieve. None means all items.
:return: Pandas dataframe with payload data.
"""
peutil.check_date_mode(start_datetime, end_datetime, date_mode)
cik = peutil.check_sorted_unique_param("cik", cik)
form_name = "form8"
params: Dict[str, Any] = {}
params = self._set_optional_params(
params,
start_datetime=start_datetime,
end_datetime=end_datetime,
item=item,
cik=cik,
date_mode=date_mode,
)
url = f'{self.base_url}{self._api_routes["PAYLOAD"]}' f"/{form_name}"
payload_dataframe = pd.DataFrame()
for df in self._payload_form_cik_cusip_generator(
method="GET", url=url, headers=self.headers, params=params
):
payload_dataframe = payload_dataframe.append(df, ignore_index=True)
if (
not payload_dataframe.empty
and {
"filing_date",
"cik",
"item_name",
}.issubset(payload_dataframe.columns)
):
payload_dataframe = payload_dataframe.sort_values(
["filing_date", "cik", "item_name"]
)
return payload_dataframe.reset_index(drop=True)
def get_form10_payload(
self,
cik: Optional[peconf.CIK_TYPE] = None,
start_datetime: Optional[str] = None,
end_datetime: Optional[str] = None,
date_mode: Optional[str] = None,
) -> List[peconf.SERVER_RESPONSE_TYPE]:
"""
Get payload data for a form10, and a company.
:param cik: Central Index Key as integer. Could be list of P1_CIK or
just one identifier.
:param start_datetime: Get data where filing date is >= start_date. Date
format is "YYYY-MM-DDTHH-MI-SS". None means the entire available date range.
:param end_datetime: Get data where filing date is <= end_date. Date format
is "YYYY-MM-DDTHH-MI-SS". None means the entire available date range.
:param date_mode: Define whether dates are
interpreted as publication dates or knowledge dates
:return: List with payload data.
"""
peutil.check_date_mode(start_datetime, end_datetime, date_mode)
cik = peutil.check_sorted_unique_param("cik", cik)
form_name = "form10"
params: Dict[str, Any] = {}
params = self._set_optional_params(
params, start_datetime=start_datetime, end_datetime=end_datetime,
date_mode=date_mode
)
url = f'{self.base_url}{self._api_routes["PAYLOAD"]}/{form_name}'
cik_list: List[Union[int, None]] = [None]
compound_data = []
if cik is not None:
cik_list = [cik] if isinstance(cik, int) else cik
self.spinner.start()
with peutil.spinner_exception_handling(self.spinner):
for current_cik in tauto.tqdm(cik_list, desc="Processing CIK: "):
self._set_optional_params(params, cik=current_cik)
response = self._make_request(
"GET", url, headers=self.headers, params=params
)
self.spinner.stop()
data = response.json()["data"]
_LOG.info("%s: %s forms loaded",
current_cik or "Total",
len(data))
compound_data += response.json()["data"]
return compound_data
def get_form4_payload(
self,
cik: Optional[peconf.CIK_TYPE] = None,
start_datetime: Optional[str] = None,
end_datetime: Optional[str] = None,
date_mode: Optional[str] = None,
output_type: str = "dataframes",
) -> Dict[str, List[peconf.SERVER_RESPONSE_TYPE]]:
peutil.check_date_mode(start_datetime, end_datetime, date_mode)
cik = peutil.check_sorted_unique_param("cik", cik)
form_type = "form4"
result = self._get_form4_13_payload(
form_type,
cik=cik,
start_datetime=start_datetime,
end_datetime=end_datetime,
date_mode=date_mode,
output_type=output_type,
)
return result