EntsoePandasClient: query_procured_balancing_capacity does not return the right time span

[danshaman05]

import pandas as pd
from entsoe import EntsoePandasClient

api_key = 'your-api-key'

start_date = pd.Timestamp("2024-08-20T0000", tz='Europe/Prague')
end_date = start_date + pd.Timedelta(hours=23)

country_code = 'CZ'
type_marketagreement_type = 'A01'
contract_marketagreement_type = 'A01'
process_type = 'A47'  # mFRR

client = EntsoePandasClient(api_key)

df = client.query_procured_balancing_capacity(
            country_code,
            process_type=process_type,
            start=start_date,
            end=end_date,
            type_marketagreement_type=type_marketagreement_type)

print("number of rows:", len(df))
print(df.index)

output:

number of rows: 21
DatetimeIndex(['2024-08-20 03:00:00+02:00', '2024-08-20 04:00:00+02:00',
               '2024-08-20 05:00:00+02:00', '2024-08-20 06:00:00+02:00',
               '2024-08-20 07:00:00+02:00', '2024-08-20 08:00:00+02:00',
               '2024-08-20 09:00:00+02:00', '2024-08-20 10:00:00+02:00',
               '2024-08-20 11:00:00+02:00', '2024-08-20 12:00:00+02:00',
               '2024-08-20 13:00:00+02:00', '2024-08-20 14:00:00+02:00',
               '2024-08-20 15:00:00+02:00', '2024-08-20 16:00:00+02:00',
               '2024-08-20 17:00:00+02:00', '2024-08-20 18:00:00+02:00',
               '2024-08-20 19:00:00+02:00', '2024-08-20 20:00:00+02:00',
               '2024-08-20 21:00:00+02:00', '2024-08-20 22:00:00+02:00',
               '2024-08-20 23:00:00+02:00'],
              dtype='datetime64[ns, Europe/Prague]', freq='60min')

Note, that first three hours are missing. It should return 24 rows instead of 21.
Probably similar to issue #187.

[danshaman05]

I noticed that this is making a lot more API requests, than is needed. The query requested for 135 xml elements (maximum is 100 per request), so ideally it should make only 2 requests.

Maybe the problem is that API expects start and end timestamps in local timezone, but ours is 'UTC'.
When I get rid of conversions of start and end timestamps in EntsoeRawClient._base_request method (conversion is in _datetime_to_str), it works like a charm. I have less requests and right time span.

output:

number of rows: 24
DatetimeIndex(['2024-08-20 00:00:00+02:00', '2024-08-20 01:00:00+02:00',
               '2024-08-20 02:00:00+02:00', '2024-08-20 03:00:00+02:00',
               '2024-08-20 04:00:00+02:00', '2024-08-20 05:00:00+02:00',
               '2024-08-20 06:00:00+02:00', '2024-08-20 07:00:00+02:00',
               '2024-08-20 08:00:00+02:00', '2024-08-20 09:00:00+02:00',
               '2024-08-20 10:00:00+02:00', '2024-08-20 11:00:00+02:00',
               '2024-08-20 12:00:00+02:00', '2024-08-20 13:00:00+02:00',
               '2024-08-20 14:00:00+02:00', '2024-08-20 15:00:00+02:00',
               '2024-08-20 16:00:00+02:00', '2024-08-20 17:00:00+02:00',
               '2024-08-20 18:00:00+02:00', '2024-08-20 19:00:00+02:00',
               '2024-08-20 20:00:00+02:00', '2024-08-20 21:00:00+02:00',
               '2024-08-20 22:00:00+02:00', '2024-08-20 23:00:00+02:00'],
              dtype='datetime64[ns, Europe/Prague]', freq='60min')

I haven't tested yet if this will affect other functions, but this could be also a solution for #187.

[fboerman]

hi! this is very strange since all other endpoints do expect (and work with) UTC. Thus removing the conversions in the base_request will probably break a lot of things. Do you have ideas for other fixes which are more specific to this method? I dont use this endpoint myself so kinda hard to test it