def wait_for_rate_limit_reset(self):
utc_now = pd.Timestamp.utcnow()
timedelta_to_wait = self.rate_limit_reset_time - utc_now
timedelta_to_wait += timedelta(minutes=3) # Just for safety
secs_to_wait = timedelta_to_wait.total_seconds()
retry_time_utc = utc_now + timedelta_to_wait
_print_and_log('Waiting {:.0f} seconds. Will retry at {}'.format(
secs_to_wait, retry_time_utc))
time.sleep(secs_to_wait)
def _api_query(self,
service: str,
api_params: Dict,
wait_if_rate_limit_exceeded: bool = False,
use_data_service: bool = False) -> str:
"""Send API request to PVOutput.org and return content text.
Args:
service: string, e.g. 'search' or 'getstatus'
api_params: dict
wait_if_rate_limit_exceeded: bool
use_data_service: bool
Raises:
NoStatusFound
RateLimitExceeded
"""
get_response_func = (self._get_data_service_response
if use_data_service else self._get_api_response)
try:
response = get_response_func(service, api_params)
except Exception as e:
_LOG.exception(e)
raise
try:
return self._process_api_response(response)
except RateLimitExceeded:
msg = ("PVOutput.org API rate limit exceeded!"
" Rate limit will be reset at {}".format(
self.rate_limit_reset_time))
_print_and_log(msg)
if wait_if_rate_limit_exceeded:
self.wait_for_rate_limit_reset()
return self._api_query(service,
api_params,
wait_if_rate_limit_exceeded=False)
raise RateLimitExceeded(response, msg)
def get_batch_status(self,
pv_system_id: int,
date_to: Optional[Union[str, datetime]] = None,
max_retries: Optional[int] = 1000,
**kwargs) -> Union[None, pd.DataFrame]:
"""Get batch PV system status (e.g. power generation).
The returned DataFrame will be empty if the PVOutput API
returns 'status 400: No status found'.
Data returned is limited to the last 366 days per request.
To retrieve older data, use the date_to parameter.
The PVOutput getbatchstatus API is asynchronous. When it's first
called, it replies to say 'accepted'. This function will then
wait a minute and call the API again to see if the data is ready.
Set `max_retries` to 1 if you want to return immediately, even
if data isn't ready yet (and hence this function will return None)
https://pvoutput.org/help.html#dataservice-getbatchstatus
Args:
pv_system_id: int
date_to: str in format YYYYMMDD; or datetime
(localtime of the PV system). The returned timeseries will
include 366 days of data: from YYYY-1MMDD to YYYYMMDD inclusive
max_retries: int, number of times to retry after receiving
a '202 Accepted' request. Set `max_retries` to 1 if you want
to return immediately, even if data isn't ready yet (and hence
this function will return None).
Returns:
None (if data isn't ready after retrying max_retries times) or
pd.DataFrame:
index: datetime (DatetimeIndex, localtime of the PV system)
columns: (all np.float64):
cumulative_energy_gen_Wh,
instantaneous_power_gen_W,
temperature_C,
voltage
"""
api_params = {'sid1': pv_system_id}
_set_date_param(date_to, api_params, 'dt')
for retry in range(max_retries):
try:
pv_system_status_text = self._api_query(
service='getbatchstatus',
api_params=api_params,
use_data_service=True,
**kwargs)
except NoStatusFound:
_LOG.info('system_id %d: No status found for date_to %s',
pv_system_id, date_to)
pv_system_status_text = ""
break
if 'Accepted 202' in pv_system_status_text:
if retry == 0:
_print_and_log('Request accepted.')
if retry < max_retries - 1:
_print_and_log('Sleeping for 1 minute.')
time.sleep(60)
else:
_print_and_log(
'Call get_batch_status again in a minute to see if'
' results are ready.')
else:
break
else:
return
return _process_batch_status(pv_system_status_text)
